{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 49641,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006043391551338611,
      "grad_norm": 298.2872354263952,
      "learning_rate": 4.0273862263391065e-08,
      "loss": 6.1467,
      "step": 10
    },
    {
      "epoch": 0.0012086783102677222,
      "grad_norm": 272.5003620952357,
      "learning_rate": 8.054772452678213e-08,
      "loss": 6.1276,
      "step": 20
    },
    {
      "epoch": 0.0018130174654015835,
      "grad_norm": 288.81048256119277,
      "learning_rate": 1.208215867901732e-07,
      "loss": 6.0628,
      "step": 30
    },
    {
      "epoch": 0.0024173566205354445,
      "grad_norm": 269.56952062216857,
      "learning_rate": 1.6109544905356426e-07,
      "loss": 5.7986,
      "step": 40
    },
    {
      "epoch": 0.0030216957756693057,
      "grad_norm": 206.36532951229918,
      "learning_rate": 2.013693113169553e-07,
      "loss": 5.3366,
      "step": 50
    },
    {
      "epoch": 0.003626034930803167,
      "grad_norm": 203.00258326291555,
      "learning_rate": 2.416431735803464e-07,
      "loss": 4.8985,
      "step": 60
    },
    {
      "epoch": 0.004230374085937028,
      "grad_norm": 155.13411973161922,
      "learning_rate": 2.819170358437374e-07,
      "loss": 4.4859,
      "step": 70
    },
    {
      "epoch": 0.004834713241070889,
      "grad_norm": 155.137249664987,
      "learning_rate": 3.221908981071285e-07,
      "loss": 3.847,
      "step": 80
    },
    {
      "epoch": 0.00543905239620475,
      "grad_norm": 68.62487791511931,
      "learning_rate": 3.6246476037051957e-07,
      "loss": 2.8263,
      "step": 90
    },
    {
      "epoch": 0.0060433915513386114,
      "grad_norm": 29.616720475402726,
      "learning_rate": 4.027386226339106e-07,
      "loss": 2.5251,
      "step": 100
    },
    {
      "epoch": 0.006647730706472473,
      "grad_norm": 25.716610662045905,
      "learning_rate": 4.430124848973017e-07,
      "loss": 2.3305,
      "step": 110
    },
    {
      "epoch": 0.007252069861606334,
      "grad_norm": 25.925794056215235,
      "learning_rate": 4.832863471606928e-07,
      "loss": 2.0343,
      "step": 120
    },
    {
      "epoch": 0.007856409016740194,
      "grad_norm": 16.736804626915543,
      "learning_rate": 5.235602094240838e-07,
      "loss": 1.7285,
      "step": 130
    },
    {
      "epoch": 0.008460748171874055,
      "grad_norm": 14.505748439621561,
      "learning_rate": 5.638340716874748e-07,
      "loss": 1.4641,
      "step": 140
    },
    {
      "epoch": 0.009065087327007917,
      "grad_norm": 4.088781078521003,
      "learning_rate": 6.041079339508659e-07,
      "loss": 1.2911,
      "step": 150
    },
    {
      "epoch": 0.009669426482141778,
      "grad_norm": 3.5454597385386095,
      "learning_rate": 6.44381796214257e-07,
      "loss": 1.273,
      "step": 160
    },
    {
      "epoch": 0.01027376563727564,
      "grad_norm": 3.145326385935495,
      "learning_rate": 6.84655658477648e-07,
      "loss": 1.2294,
      "step": 170
    },
    {
      "epoch": 0.0108781047924095,
      "grad_norm": 2.9493249288358885,
      "learning_rate": 7.249295207410391e-07,
      "loss": 1.2643,
      "step": 180
    },
    {
      "epoch": 0.011482443947543362,
      "grad_norm": 3.445852230847896,
      "learning_rate": 7.652033830044302e-07,
      "loss": 1.1955,
      "step": 190
    },
    {
      "epoch": 0.012086783102677223,
      "grad_norm": 2.442717320016556,
      "learning_rate": 8.054772452678212e-07,
      "loss": 1.2114,
      "step": 200
    },
    {
      "epoch": 0.012691122257811084,
      "grad_norm": 2.575139498019344,
      "learning_rate": 8.457511075312124e-07,
      "loss": 1.165,
      "step": 210
    },
    {
      "epoch": 0.013295461412944945,
      "grad_norm": 2.269127757263327,
      "learning_rate": 8.860249697946034e-07,
      "loss": 1.1835,
      "step": 220
    },
    {
      "epoch": 0.013899800568078807,
      "grad_norm": 2.7913296947023287,
      "learning_rate": 9.262988320579944e-07,
      "loss": 1.1748,
      "step": 230
    },
    {
      "epoch": 0.014504139723212668,
      "grad_norm": 2.506161432538682,
      "learning_rate": 9.665726943213855e-07,
      "loss": 1.1829,
      "step": 240
    },
    {
      "epoch": 0.015108478878346527,
      "grad_norm": 2.265386795022933,
      "learning_rate": 1.0068465565847765e-06,
      "loss": 1.1918,
      "step": 250
    },
    {
      "epoch": 0.01571281803348039,
      "grad_norm": 2.3797096473176076,
      "learning_rate": 1.0471204188481676e-06,
      "loss": 1.1839,
      "step": 260
    },
    {
      "epoch": 0.01631715718861425,
      "grad_norm": 2.2402768142876694,
      "learning_rate": 1.0873942811115588e-06,
      "loss": 1.1855,
      "step": 270
    },
    {
      "epoch": 0.01692149634374811,
      "grad_norm": 2.2249442023552852,
      "learning_rate": 1.1276681433749497e-06,
      "loss": 1.1649,
      "step": 280
    },
    {
      "epoch": 0.017525835498881974,
      "grad_norm": 2.085305674911988,
      "learning_rate": 1.1679420056383408e-06,
      "loss": 1.1806,
      "step": 290
    },
    {
      "epoch": 0.018130174654015833,
      "grad_norm": 2.0828553521779183,
      "learning_rate": 1.2082158679017318e-06,
      "loss": 1.1447,
      "step": 300
    },
    {
      "epoch": 0.018734513809149696,
      "grad_norm": 1.9032512166927018,
      "learning_rate": 1.248489730165123e-06,
      "loss": 1.1507,
      "step": 310
    },
    {
      "epoch": 0.019338852964283556,
      "grad_norm": 1.91018837067208,
      "learning_rate": 1.288763592428514e-06,
      "loss": 1.1516,
      "step": 320
    },
    {
      "epoch": 0.019943192119417415,
      "grad_norm": 1.903531796901431,
      "learning_rate": 1.329037454691905e-06,
      "loss": 1.1604,
      "step": 330
    },
    {
      "epoch": 0.02054753127455128,
      "grad_norm": 2.136992877764608,
      "learning_rate": 1.369311316955296e-06,
      "loss": 1.1458,
      "step": 340
    },
    {
      "epoch": 0.021151870429685138,
      "grad_norm": 1.74601727863523,
      "learning_rate": 1.4095851792186871e-06,
      "loss": 1.1676,
      "step": 350
    },
    {
      "epoch": 0.021756209584819,
      "grad_norm": 1.6117672740528162,
      "learning_rate": 1.4498590414820783e-06,
      "loss": 1.145,
      "step": 360
    },
    {
      "epoch": 0.02236054873995286,
      "grad_norm": 1.8149579016476314,
      "learning_rate": 1.4901329037454692e-06,
      "loss": 1.1583,
      "step": 370
    },
    {
      "epoch": 0.022964887895086723,
      "grad_norm": 1.5885652581406067,
      "learning_rate": 1.5304067660088604e-06,
      "loss": 1.1487,
      "step": 380
    },
    {
      "epoch": 0.023569227050220583,
      "grad_norm": 1.7684496469863542,
      "learning_rate": 1.5706806282722515e-06,
      "loss": 1.1128,
      "step": 390
    },
    {
      "epoch": 0.024173566205354446,
      "grad_norm": 1.7621133706989613,
      "learning_rate": 1.6109544905356425e-06,
      "loss": 1.1468,
      "step": 400
    },
    {
      "epoch": 0.024777905360488305,
      "grad_norm": 2.038485171893254,
      "learning_rate": 1.6512283527990336e-06,
      "loss": 1.1751,
      "step": 410
    },
    {
      "epoch": 0.025382244515622168,
      "grad_norm": 1.91524830783998,
      "learning_rate": 1.6915022150624248e-06,
      "loss": 1.1654,
      "step": 420
    },
    {
      "epoch": 0.025986583670756028,
      "grad_norm": 1.8236734227335434,
      "learning_rate": 1.7317760773258157e-06,
      "loss": 1.1648,
      "step": 430
    },
    {
      "epoch": 0.02659092282588989,
      "grad_norm": 2.0232671392154917,
      "learning_rate": 1.7720499395892068e-06,
      "loss": 1.1485,
      "step": 440
    },
    {
      "epoch": 0.02719526198102375,
      "grad_norm": 1.6150477738657656,
      "learning_rate": 1.812323801852598e-06,
      "loss": 1.1089,
      "step": 450
    },
    {
      "epoch": 0.027799601136157613,
      "grad_norm": 1.5572962388413956,
      "learning_rate": 1.8525976641159887e-06,
      "loss": 1.1211,
      "step": 460
    },
    {
      "epoch": 0.028403940291291473,
      "grad_norm": 1.8244931625619223,
      "learning_rate": 1.8928715263793799e-06,
      "loss": 1.1244,
      "step": 470
    },
    {
      "epoch": 0.029008279446425336,
      "grad_norm": 1.6531825283199104,
      "learning_rate": 1.933145388642771e-06,
      "loss": 1.1539,
      "step": 480
    },
    {
      "epoch": 0.029612618601559195,
      "grad_norm": 1.451041168050482,
      "learning_rate": 1.973419250906162e-06,
      "loss": 1.1212,
      "step": 490
    },
    {
      "epoch": 0.030216957756693055,
      "grad_norm": 1.957377854658712,
      "learning_rate": 2.013693113169553e-06,
      "loss": 1.144,
      "step": 500
    },
    {
      "epoch": 0.030821296911826918,
      "grad_norm": 1.886661763056281,
      "learning_rate": 2.0539669754329443e-06,
      "loss": 1.1387,
      "step": 510
    },
    {
      "epoch": 0.03142563606696078,
      "grad_norm": 1.7418883718172464,
      "learning_rate": 2.094240837696335e-06,
      "loss": 1.1398,
      "step": 520
    },
    {
      "epoch": 0.03202997522209464,
      "grad_norm": 1.7816317246389095,
      "learning_rate": 2.134514699959726e-06,
      "loss": 1.1407,
      "step": 530
    },
    {
      "epoch": 0.0326343143772285,
      "grad_norm": 1.7678275511310892,
      "learning_rate": 2.1747885622231175e-06,
      "loss": 1.1148,
      "step": 540
    },
    {
      "epoch": 0.03323865353236236,
      "grad_norm": 1.6510201333960512,
      "learning_rate": 2.2150624244865084e-06,
      "loss": 1.0989,
      "step": 550
    },
    {
      "epoch": 0.03384299268749622,
      "grad_norm": 1.528929289785965,
      "learning_rate": 2.2553362867498994e-06,
      "loss": 1.1237,
      "step": 560
    },
    {
      "epoch": 0.03444733184263008,
      "grad_norm": 1.5209366865980554,
      "learning_rate": 2.2956101490132908e-06,
      "loss": 1.1303,
      "step": 570
    },
    {
      "epoch": 0.03505167099776395,
      "grad_norm": 1.4974206580420846,
      "learning_rate": 2.3358840112766817e-06,
      "loss": 1.1397,
      "step": 580
    },
    {
      "epoch": 0.03565601015289781,
      "grad_norm": 1.690007751571185,
      "learning_rate": 2.3761578735400726e-06,
      "loss": 1.1302,
      "step": 590
    },
    {
      "epoch": 0.03626034930803167,
      "grad_norm": 1.6781638968679657,
      "learning_rate": 2.4164317358034636e-06,
      "loss": 1.1237,
      "step": 600
    },
    {
      "epoch": 0.036864688463165526,
      "grad_norm": 1.7462569548834377,
      "learning_rate": 2.4567055980668545e-06,
      "loss": 1.1316,
      "step": 610
    },
    {
      "epoch": 0.03746902761829939,
      "grad_norm": 1.755005907449779,
      "learning_rate": 2.496979460330246e-06,
      "loss": 1.1348,
      "step": 620
    },
    {
      "epoch": 0.03807336677343325,
      "grad_norm": 1.7005845203245822,
      "learning_rate": 2.537253322593637e-06,
      "loss": 1.1146,
      "step": 630
    },
    {
      "epoch": 0.03867770592856711,
      "grad_norm": 1.9761625298171615,
      "learning_rate": 2.577527184857028e-06,
      "loss": 1.112,
      "step": 640
    },
    {
      "epoch": 0.03928204508370097,
      "grad_norm": 1.8575660787738717,
      "learning_rate": 2.617801047120419e-06,
      "loss": 1.1192,
      "step": 650
    },
    {
      "epoch": 0.03988638423883483,
      "grad_norm": 1.7192840768266824,
      "learning_rate": 2.65807490938381e-06,
      "loss": 1.1197,
      "step": 660
    },
    {
      "epoch": 0.0404907233939687,
      "grad_norm": 1.764024302272521,
      "learning_rate": 2.6983487716472014e-06,
      "loss": 1.108,
      "step": 670
    },
    {
      "epoch": 0.04109506254910256,
      "grad_norm": 1.7154086975607765,
      "learning_rate": 2.738622633910592e-06,
      "loss": 1.0843,
      "step": 680
    },
    {
      "epoch": 0.041699401704236416,
      "grad_norm": 1.8445623759164267,
      "learning_rate": 2.7788964961739833e-06,
      "loss": 1.1314,
      "step": 690
    },
    {
      "epoch": 0.042303740859370276,
      "grad_norm": 1.1164464947539972,
      "learning_rate": 2.8191703584373742e-06,
      "loss": 1.1154,
      "step": 700
    },
    {
      "epoch": 0.04290808001450414,
      "grad_norm": 1.0771280141246888,
      "learning_rate": 2.8594442207007656e-06,
      "loss": 1.0927,
      "step": 710
    },
    {
      "epoch": 0.043512419169638,
      "grad_norm": 1.0823024723600558,
      "learning_rate": 2.8997180829641565e-06,
      "loss": 1.1202,
      "step": 720
    },
    {
      "epoch": 0.04411675832477186,
      "grad_norm": 1.059423786201847,
      "learning_rate": 2.939991945227548e-06,
      "loss": 1.1058,
      "step": 730
    },
    {
      "epoch": 0.04472109747990572,
      "grad_norm": 1.107829209476366,
      "learning_rate": 2.9802658074909384e-06,
      "loss": 1.131,
      "step": 740
    },
    {
      "epoch": 0.04532543663503959,
      "grad_norm": 0.9784201230186164,
      "learning_rate": 3.0205396697543298e-06,
      "loss": 1.1314,
      "step": 750
    },
    {
      "epoch": 0.04592977579017345,
      "grad_norm": 0.9044129481577555,
      "learning_rate": 3.0608135320177207e-06,
      "loss": 1.0946,
      "step": 760
    },
    {
      "epoch": 0.046534114945307306,
      "grad_norm": 0.9427655630220715,
      "learning_rate": 3.101087394281112e-06,
      "loss": 1.0971,
      "step": 770
    },
    {
      "epoch": 0.047138454100441166,
      "grad_norm": 0.9430509363329759,
      "learning_rate": 3.141361256544503e-06,
      "loss": 1.1275,
      "step": 780
    },
    {
      "epoch": 0.04774279325557503,
      "grad_norm": 0.8704066019825859,
      "learning_rate": 3.1816351188078935e-06,
      "loss": 1.1066,
      "step": 790
    },
    {
      "epoch": 0.04834713241070889,
      "grad_norm": 1.085672823483262,
      "learning_rate": 3.221908981071285e-06,
      "loss": 1.1035,
      "step": 800
    },
    {
      "epoch": 0.04895147156584275,
      "grad_norm": 1.0001828747437085,
      "learning_rate": 3.262182843334676e-06,
      "loss": 1.112,
      "step": 810
    },
    {
      "epoch": 0.04955581072097661,
      "grad_norm": 1.0092347772885002,
      "learning_rate": 3.302456705598067e-06,
      "loss": 1.1104,
      "step": 820
    },
    {
      "epoch": 0.05016014987611047,
      "grad_norm": 1.004402462110965,
      "learning_rate": 3.342730567861458e-06,
      "loss": 1.1317,
      "step": 830
    },
    {
      "epoch": 0.050764489031244336,
      "grad_norm": 0.9845202468833524,
      "learning_rate": 3.3830044301248495e-06,
      "loss": 1.1244,
      "step": 840
    },
    {
      "epoch": 0.051368828186378196,
      "grad_norm": 0.714487787028079,
      "learning_rate": 3.42327829238824e-06,
      "loss": 1.1242,
      "step": 850
    },
    {
      "epoch": 0.051973167341512055,
      "grad_norm": 0.6900310140587528,
      "learning_rate": 3.4635521546516314e-06,
      "loss": 1.0817,
      "step": 860
    },
    {
      "epoch": 0.052577506496645915,
      "grad_norm": 0.625412751358801,
      "learning_rate": 3.5038260169150223e-06,
      "loss": 1.1088,
      "step": 870
    },
    {
      "epoch": 0.05318184565177978,
      "grad_norm": 0.6783066840510502,
      "learning_rate": 3.5440998791784137e-06,
      "loss": 1.1139,
      "step": 880
    },
    {
      "epoch": 0.05378618480691364,
      "grad_norm": 0.692374586650823,
      "learning_rate": 3.5843737414418046e-06,
      "loss": 1.1108,
      "step": 890
    },
    {
      "epoch": 0.0543905239620475,
      "grad_norm": 0.678311995376911,
      "learning_rate": 3.624647603705196e-06,
      "loss": 1.0974,
      "step": 900
    },
    {
      "epoch": 0.05499486311718136,
      "grad_norm": 0.7425145958964064,
      "learning_rate": 3.6649214659685865e-06,
      "loss": 1.0763,
      "step": 910
    },
    {
      "epoch": 0.055599202272315226,
      "grad_norm": 0.6856922654960516,
      "learning_rate": 3.7051953282319774e-06,
      "loss": 1.1136,
      "step": 920
    },
    {
      "epoch": 0.056203541427449086,
      "grad_norm": 0.7064784604124972,
      "learning_rate": 3.745469190495369e-06,
      "loss": 1.1098,
      "step": 930
    },
    {
      "epoch": 0.056807880582582945,
      "grad_norm": 0.6703709229378032,
      "learning_rate": 3.7857430527587597e-06,
      "loss": 1.1197,
      "step": 940
    },
    {
      "epoch": 0.057412219737716805,
      "grad_norm": 1.2282719953549017,
      "learning_rate": 3.826016915022151e-06,
      "loss": 1.0899,
      "step": 950
    },
    {
      "epoch": 0.05801655889285067,
      "grad_norm": 1.1929340082708477,
      "learning_rate": 3.866290777285542e-06,
      "loss": 1.1288,
      "step": 960
    },
    {
      "epoch": 0.05862089804798453,
      "grad_norm": 1.2746392955027297,
      "learning_rate": 3.906564639548933e-06,
      "loss": 1.0899,
      "step": 970
    },
    {
      "epoch": 0.05922523720311839,
      "grad_norm": 1.2099210453245777,
      "learning_rate": 3.946838501812324e-06,
      "loss": 1.1243,
      "step": 980
    },
    {
      "epoch": 0.05982957635825225,
      "grad_norm": 1.2112775722352984,
      "learning_rate": 3.987112364075715e-06,
      "loss": 1.1175,
      "step": 990
    },
    {
      "epoch": 0.06043391551338611,
      "grad_norm": 0.6053511060988584,
      "learning_rate": 4.027386226339106e-06,
      "loss": 1.1148,
      "step": 1000
    },
    {
      "epoch": 0.061038254668519976,
      "grad_norm": 0.6446163720622058,
      "learning_rate": 4.067660088602497e-06,
      "loss": 1.1122,
      "step": 1010
    },
    {
      "epoch": 0.061642593823653835,
      "grad_norm": 0.5865290255556028,
      "learning_rate": 4.1079339508658885e-06,
      "loss": 1.0932,
      "step": 1020
    },
    {
      "epoch": 0.062246932978787695,
      "grad_norm": 0.5740116165642117,
      "learning_rate": 4.14820781312928e-06,
      "loss": 1.103,
      "step": 1030
    },
    {
      "epoch": 0.06285127213392155,
      "grad_norm": 0.5698314395636827,
      "learning_rate": 4.18848167539267e-06,
      "loss": 1.1156,
      "step": 1040
    },
    {
      "epoch": 0.06345561128905541,
      "grad_norm": 0.5673180485264345,
      "learning_rate": 4.228755537656062e-06,
      "loss": 1.1459,
      "step": 1050
    },
    {
      "epoch": 0.06405995044418927,
      "grad_norm": 0.6085986093961342,
      "learning_rate": 4.269029399919452e-06,
      "loss": 1.0977,
      "step": 1060
    },
    {
      "epoch": 0.06466428959932315,
      "grad_norm": 0.584955573082463,
      "learning_rate": 4.309303262182844e-06,
      "loss": 1.0929,
      "step": 1070
    },
    {
      "epoch": 0.065268628754457,
      "grad_norm": 0.6106749898890043,
      "learning_rate": 4.349577124446235e-06,
      "loss": 1.0647,
      "step": 1080
    },
    {
      "epoch": 0.06587296790959087,
      "grad_norm": 0.6164260748087341,
      "learning_rate": 4.3898509867096255e-06,
      "loss": 1.1274,
      "step": 1090
    },
    {
      "epoch": 0.06647730706472472,
      "grad_norm": 0.557833684560504,
      "learning_rate": 4.430124848973017e-06,
      "loss": 1.1089,
      "step": 1100
    },
    {
      "epoch": 0.06708164621985858,
      "grad_norm": 0.5829249216386346,
      "learning_rate": 4.470398711236407e-06,
      "loss": 1.1347,
      "step": 1110
    },
    {
      "epoch": 0.06768598537499244,
      "grad_norm": 0.6263876637695363,
      "learning_rate": 4.510672573499799e-06,
      "loss": 1.0979,
      "step": 1120
    },
    {
      "epoch": 0.0682903245301263,
      "grad_norm": 0.6196422245080135,
      "learning_rate": 4.55094643576319e-06,
      "loss": 1.1036,
      "step": 1130
    },
    {
      "epoch": 0.06889466368526016,
      "grad_norm": 0.6158195649523993,
      "learning_rate": 4.5912202980265815e-06,
      "loss": 1.1014,
      "step": 1140
    },
    {
      "epoch": 0.06949900284039402,
      "grad_norm": 0.5800059892091631,
      "learning_rate": 4.631494160289972e-06,
      "loss": 1.1179,
      "step": 1150
    },
    {
      "epoch": 0.0701033419955279,
      "grad_norm": 0.5967167581820909,
      "learning_rate": 4.671768022553363e-06,
      "loss": 1.1036,
      "step": 1160
    },
    {
      "epoch": 0.07070768115066176,
      "grad_norm": 0.5849312766034016,
      "learning_rate": 4.712041884816754e-06,
      "loss": 1.0933,
      "step": 1170
    },
    {
      "epoch": 0.07131202030579561,
      "grad_norm": 0.6111626030450139,
      "learning_rate": 4.752315747080145e-06,
      "loss": 1.088,
      "step": 1180
    },
    {
      "epoch": 0.07191635946092947,
      "grad_norm": 0.5676535264441273,
      "learning_rate": 4.792589609343537e-06,
      "loss": 1.1039,
      "step": 1190
    },
    {
      "epoch": 0.07252069861606333,
      "grad_norm": 0.6056548822582074,
      "learning_rate": 4.832863471606927e-06,
      "loss": 1.1005,
      "step": 1200
    },
    {
      "epoch": 0.0731250377711972,
      "grad_norm": 0.6232710264069062,
      "learning_rate": 4.8731373338703185e-06,
      "loss": 1.0864,
      "step": 1210
    },
    {
      "epoch": 0.07372937692633105,
      "grad_norm": 0.619023692523766,
      "learning_rate": 4.913411196133709e-06,
      "loss": 1.1474,
      "step": 1220
    },
    {
      "epoch": 0.07433371608146491,
      "grad_norm": 0.5897355405120764,
      "learning_rate": 4.9536850583971e-06,
      "loss": 1.1148,
      "step": 1230
    },
    {
      "epoch": 0.07493805523659879,
      "grad_norm": 0.6255020630314708,
      "learning_rate": 4.993958920660492e-06,
      "loss": 1.1084,
      "step": 1240
    },
    {
      "epoch": 0.07554239439173265,
      "grad_norm": 0.572696440955054,
      "learning_rate": 5.034232782923882e-06,
      "loss": 1.1045,
      "step": 1250
    },
    {
      "epoch": 0.0761467335468665,
      "grad_norm": 0.6016228313532224,
      "learning_rate": 5.074506645187274e-06,
      "loss": 1.0806,
      "step": 1260
    },
    {
      "epoch": 0.07675107270200036,
      "grad_norm": 0.6148565648976038,
      "learning_rate": 5.114780507450665e-06,
      "loss": 1.0906,
      "step": 1270
    },
    {
      "epoch": 0.07735541185713422,
      "grad_norm": 0.6250503294264665,
      "learning_rate": 5.155054369714056e-06,
      "loss": 1.1088,
      "step": 1280
    },
    {
      "epoch": 0.07795975101226808,
      "grad_norm": 0.6259358544472459,
      "learning_rate": 5.195328231977447e-06,
      "loss": 1.1065,
      "step": 1290
    },
    {
      "epoch": 0.07856409016740194,
      "grad_norm": 0.6928975213289578,
      "learning_rate": 5.235602094240838e-06,
      "loss": 1.0965,
      "step": 1300
    },
    {
      "epoch": 0.0791684293225358,
      "grad_norm": 0.7480488578586902,
      "learning_rate": 5.27587595650423e-06,
      "loss": 1.0945,
      "step": 1310
    },
    {
      "epoch": 0.07977276847766966,
      "grad_norm": 0.6970973531334974,
      "learning_rate": 5.31614981876762e-06,
      "loss": 1.0981,
      "step": 1320
    },
    {
      "epoch": 0.08037710763280353,
      "grad_norm": 0.6818063136002229,
      "learning_rate": 5.3564236810310115e-06,
      "loss": 1.0761,
      "step": 1330
    },
    {
      "epoch": 0.0809814467879374,
      "grad_norm": 0.6959641966583481,
      "learning_rate": 5.396697543294403e-06,
      "loss": 1.0968,
      "step": 1340
    },
    {
      "epoch": 0.08158578594307125,
      "grad_norm": 0.7593382806459565,
      "learning_rate": 5.436971405557794e-06,
      "loss": 1.0682,
      "step": 1350
    },
    {
      "epoch": 0.08219012509820511,
      "grad_norm": 0.846649633309298,
      "learning_rate": 5.477245267821184e-06,
      "loss": 1.1069,
      "step": 1360
    },
    {
      "epoch": 0.08279446425333897,
      "grad_norm": 0.7368100714028909,
      "learning_rate": 5.517519130084575e-06,
      "loss": 1.0768,
      "step": 1370
    },
    {
      "epoch": 0.08339880340847283,
      "grad_norm": 0.8000337335475295,
      "learning_rate": 5.557792992347967e-06,
      "loss": 1.0999,
      "step": 1380
    },
    {
      "epoch": 0.08400314256360669,
      "grad_norm": 0.8639730071659587,
      "learning_rate": 5.598066854611358e-06,
      "loss": 1.0997,
      "step": 1390
    },
    {
      "epoch": 0.08460748171874055,
      "grad_norm": 0.6671862958037743,
      "learning_rate": 5.6383407168747485e-06,
      "loss": 1.0829,
      "step": 1400
    },
    {
      "epoch": 0.08521182087387442,
      "grad_norm": 0.7100917285829133,
      "learning_rate": 5.67861457913814e-06,
      "loss": 1.0887,
      "step": 1410
    },
    {
      "epoch": 0.08581616002900828,
      "grad_norm": 0.7197083853004631,
      "learning_rate": 5.718888441401531e-06,
      "loss": 1.0744,
      "step": 1420
    },
    {
      "epoch": 0.08642049918414214,
      "grad_norm": 0.6681752999196956,
      "learning_rate": 5.7591623036649226e-06,
      "loss": 1.0782,
      "step": 1430
    },
    {
      "epoch": 0.087024838339276,
      "grad_norm": 0.6961203501826451,
      "learning_rate": 5.799436165928313e-06,
      "loss": 1.1028,
      "step": 1440
    },
    {
      "epoch": 0.08762917749440986,
      "grad_norm": 0.8093873565812489,
      "learning_rate": 5.8397100281917044e-06,
      "loss": 1.0667,
      "step": 1450
    },
    {
      "epoch": 0.08823351664954372,
      "grad_norm": 0.8993433219450102,
      "learning_rate": 5.879983890455096e-06,
      "loss": 1.1464,
      "step": 1460
    },
    {
      "epoch": 0.08883785580467758,
      "grad_norm": 0.8838967444455852,
      "learning_rate": 5.9202577527184855e-06,
      "loss": 1.0939,
      "step": 1470
    },
    {
      "epoch": 0.08944219495981144,
      "grad_norm": 0.8507589619048073,
      "learning_rate": 5.960531614981877e-06,
      "loss": 1.1251,
      "step": 1480
    },
    {
      "epoch": 0.0900465341149453,
      "grad_norm": 0.8594816568663549,
      "learning_rate": 6.000805477245268e-06,
      "loss": 1.1003,
      "step": 1490
    },
    {
      "epoch": 0.09065087327007917,
      "grad_norm": 0.8168422366319431,
      "learning_rate": 6.0410793395086596e-06,
      "loss": 1.0726,
      "step": 1500
    },
    {
      "epoch": 0.09125521242521303,
      "grad_norm": 0.8172683114812207,
      "learning_rate": 6.08135320177205e-06,
      "loss": 1.093,
      "step": 1510
    },
    {
      "epoch": 0.0918595515803469,
      "grad_norm": 0.8166330217530501,
      "learning_rate": 6.1216270640354414e-06,
      "loss": 1.0838,
      "step": 1520
    },
    {
      "epoch": 0.09246389073548075,
      "grad_norm": 0.811202383541008,
      "learning_rate": 6.161900926298833e-06,
      "loss": 1.0976,
      "step": 1530
    },
    {
      "epoch": 0.09306822989061461,
      "grad_norm": 0.8500784323743232,
      "learning_rate": 6.202174788562224e-06,
      "loss": 1.0897,
      "step": 1540
    },
    {
      "epoch": 0.09367256904574847,
      "grad_norm": 1.2505471471205927,
      "learning_rate": 6.242448650825615e-06,
      "loss": 1.1265,
      "step": 1550
    },
    {
      "epoch": 0.09427690820088233,
      "grad_norm": 1.335408088114772,
      "learning_rate": 6.282722513089006e-06,
      "loss": 1.0834,
      "step": 1560
    },
    {
      "epoch": 0.09488124735601619,
      "grad_norm": 1.3279511615791078,
      "learning_rate": 6.322996375352397e-06,
      "loss": 1.083,
      "step": 1570
    },
    {
      "epoch": 0.09548558651115006,
      "grad_norm": 1.3945538602920304,
      "learning_rate": 6.363270237615787e-06,
      "loss": 1.1037,
      "step": 1580
    },
    {
      "epoch": 0.09608992566628392,
      "grad_norm": 1.3895453761538699,
      "learning_rate": 6.4035440998791784e-06,
      "loss": 1.0854,
      "step": 1590
    },
    {
      "epoch": 0.09669426482141778,
      "grad_norm": 7.755727454888002,
      "learning_rate": 6.44381796214257e-06,
      "loss": 1.0939,
      "step": 1600
    },
    {
      "epoch": 0.09729860397655164,
      "grad_norm": 2.19733019790502,
      "learning_rate": 6.484091824405961e-06,
      "loss": 1.0602,
      "step": 1610
    },
    {
      "epoch": 0.0979029431316855,
      "grad_norm": 1.7323413551287472,
      "learning_rate": 6.524365686669352e-06,
      "loss": 1.0635,
      "step": 1620
    },
    {
      "epoch": 0.09850728228681936,
      "grad_norm": 2.5187432382415196,
      "learning_rate": 6.564639548932743e-06,
      "loss": 1.0595,
      "step": 1630
    },
    {
      "epoch": 0.09911162144195322,
      "grad_norm": 1.4119806902781733,
      "learning_rate": 6.604913411196134e-06,
      "loss": 1.0269,
      "step": 1640
    },
    {
      "epoch": 0.09971596059708708,
      "grad_norm": 2.8766116959670938,
      "learning_rate": 6.645187273459526e-06,
      "loss": 1.0537,
      "step": 1650
    },
    {
      "epoch": 0.10032029975222094,
      "grad_norm": 3.268139899080405,
      "learning_rate": 6.685461135722916e-06,
      "loss": 1.0603,
      "step": 1660
    },
    {
      "epoch": 0.10092463890735481,
      "grad_norm": 3.0981208616068776,
      "learning_rate": 6.725734997986308e-06,
      "loss": 1.0422,
      "step": 1670
    },
    {
      "epoch": 0.10152897806248867,
      "grad_norm": 2.9239848803181863,
      "learning_rate": 6.766008860249699e-06,
      "loss": 1.0532,
      "step": 1680
    },
    {
      "epoch": 0.10213331721762253,
      "grad_norm": 3.273189811884654,
      "learning_rate": 6.80628272251309e-06,
      "loss": 1.0889,
      "step": 1690
    },
    {
      "epoch": 0.10273765637275639,
      "grad_norm": 1.5966209709842396,
      "learning_rate": 6.84655658477648e-06,
      "loss": 1.0489,
      "step": 1700
    },
    {
      "epoch": 0.10334199552789025,
      "grad_norm": 1.491149466056692,
      "learning_rate": 6.886830447039871e-06,
      "loss": 1.0559,
      "step": 1710
    },
    {
      "epoch": 0.10394633468302411,
      "grad_norm": 1.320760002171467,
      "learning_rate": 6.927104309303263e-06,
      "loss": 1.0572,
      "step": 1720
    },
    {
      "epoch": 0.10455067383815797,
      "grad_norm": 1.5609458132757414,
      "learning_rate": 6.967378171566653e-06,
      "loss": 1.0589,
      "step": 1730
    },
    {
      "epoch": 0.10515501299329183,
      "grad_norm": 1.7152369715801432,
      "learning_rate": 7.007652033830045e-06,
      "loss": 1.0538,
      "step": 1740
    },
    {
      "epoch": 0.1057593521484257,
      "grad_norm": 1.3133638590103185,
      "learning_rate": 7.047925896093436e-06,
      "loss": 1.0822,
      "step": 1750
    },
    {
      "epoch": 0.10636369130355956,
      "grad_norm": 1.3315601174512322,
      "learning_rate": 7.088199758356827e-06,
      "loss": 1.0527,
      "step": 1760
    },
    {
      "epoch": 0.10696803045869342,
      "grad_norm": 1.2444197931736023,
      "learning_rate": 7.128473620620218e-06,
      "loss": 1.0621,
      "step": 1770
    },
    {
      "epoch": 0.10757236961382728,
      "grad_norm": 1.1643028023866646,
      "learning_rate": 7.168747482883609e-06,
      "loss": 1.0513,
      "step": 1780
    },
    {
      "epoch": 0.10817670876896114,
      "grad_norm": 1.0771408875481137,
      "learning_rate": 7.209021345147001e-06,
      "loss": 1.0464,
      "step": 1790
    },
    {
      "epoch": 0.108781047924095,
      "grad_norm": 1.277208641518362,
      "learning_rate": 7.249295207410392e-06,
      "loss": 1.0386,
      "step": 1800
    },
    {
      "epoch": 0.10938538707922886,
      "grad_norm": 1.118373567361937,
      "learning_rate": 7.289569069673782e-06,
      "loss": 1.0497,
      "step": 1810
    },
    {
      "epoch": 0.10998972623436272,
      "grad_norm": 1.191798426442932,
      "learning_rate": 7.329842931937173e-06,
      "loss": 1.0199,
      "step": 1820
    },
    {
      "epoch": 0.11059406538949658,
      "grad_norm": 1.2835934347441686,
      "learning_rate": 7.370116794200564e-06,
      "loss": 1.0529,
      "step": 1830
    },
    {
      "epoch": 0.11119840454463045,
      "grad_norm": 1.1854467640487083,
      "learning_rate": 7.410390656463955e-06,
      "loss": 1.0362,
      "step": 1840
    },
    {
      "epoch": 0.11180274369976431,
      "grad_norm": 1.3958257827504483,
      "learning_rate": 7.450664518727346e-06,
      "loss": 1.0269,
      "step": 1850
    },
    {
      "epoch": 0.11240708285489817,
      "grad_norm": 1.5318312249471024,
      "learning_rate": 7.490938380990738e-06,
      "loss": 1.0718,
      "step": 1860
    },
    {
      "epoch": 0.11301142201003203,
      "grad_norm": 1.2156396102319647,
      "learning_rate": 7.531212243254129e-06,
      "loss": 1.0506,
      "step": 1870
    },
    {
      "epoch": 0.11361576116516589,
      "grad_norm": 1.2622492188279455,
      "learning_rate": 7.5714861055175195e-06,
      "loss": 1.0137,
      "step": 1880
    },
    {
      "epoch": 0.11422010032029975,
      "grad_norm": 1.1776041082656958,
      "learning_rate": 7.611759967780911e-06,
      "loss": 1.0454,
      "step": 1890
    },
    {
      "epoch": 0.11482443947543361,
      "grad_norm": 1.3457164150405259,
      "learning_rate": 7.652033830044301e-06,
      "loss": 1.0562,
      "step": 1900
    },
    {
      "epoch": 0.11542877863056747,
      "grad_norm": 1.3780287006033587,
      "learning_rate": 7.692307692307694e-06,
      "loss": 1.0169,
      "step": 1910
    },
    {
      "epoch": 0.11603311778570134,
      "grad_norm": 1.3119894231887488,
      "learning_rate": 7.732581554571084e-06,
      "loss": 1.061,
      "step": 1920
    },
    {
      "epoch": 0.1166374569408352,
      "grad_norm": 1.175282317328737,
      "learning_rate": 7.772855416834475e-06,
      "loss": 1.0187,
      "step": 1930
    },
    {
      "epoch": 0.11724179609596906,
      "grad_norm": 1.2808534491075105,
      "learning_rate": 7.813129279097867e-06,
      "loss": 1.0327,
      "step": 1940
    },
    {
      "epoch": 0.11784613525110292,
      "grad_norm": 1.3145724104066527,
      "learning_rate": 7.853403141361257e-06,
      "loss": 1.0451,
      "step": 1950
    },
    {
      "epoch": 0.11845047440623678,
      "grad_norm": 1.149847233731238,
      "learning_rate": 7.893677003624648e-06,
      "loss": 1.0422,
      "step": 1960
    },
    {
      "epoch": 0.11905481356137064,
      "grad_norm": 1.1732062795371199,
      "learning_rate": 7.93395086588804e-06,
      "loss": 1.0477,
      "step": 1970
    },
    {
      "epoch": 0.1196591527165045,
      "grad_norm": 1.4198026696053145,
      "learning_rate": 7.97422472815143e-06,
      "loss": 1.0331,
      "step": 1980
    },
    {
      "epoch": 0.12026349187163836,
      "grad_norm": 1.242108882945878,
      "learning_rate": 8.014498590414821e-06,
      "loss": 1.0446,
      "step": 1990
    },
    {
      "epoch": 0.12086783102677222,
      "grad_norm": 1.3628691710522989,
      "learning_rate": 8.054772452678212e-06,
      "loss": 1.0357,
      "step": 2000
    },
    {
      "epoch": 0.12147217018190609,
      "grad_norm": 1.2834103405995956,
      "learning_rate": 8.095046314941604e-06,
      "loss": 1.0575,
      "step": 2010
    },
    {
      "epoch": 0.12207650933703995,
      "grad_norm": 1.311977552770393,
      "learning_rate": 8.135320177204994e-06,
      "loss": 1.0735,
      "step": 2020
    },
    {
      "epoch": 0.12268084849217381,
      "grad_norm": 1.2260537207160858,
      "learning_rate": 8.175594039468385e-06,
      "loss": 1.0375,
      "step": 2030
    },
    {
      "epoch": 0.12328518764730767,
      "grad_norm": 1.2377992625425653,
      "learning_rate": 8.215867901731777e-06,
      "loss": 1.0362,
      "step": 2040
    },
    {
      "epoch": 0.12388952680244153,
      "grad_norm": 1.2255633080893185,
      "learning_rate": 8.256141763995168e-06,
      "loss": 1.0185,
      "step": 2050
    },
    {
      "epoch": 0.12449386595757539,
      "grad_norm": 1.2455856325728145,
      "learning_rate": 8.29641562625856e-06,
      "loss": 1.0438,
      "step": 2060
    },
    {
      "epoch": 0.12509820511270925,
      "grad_norm": 1.184511244962389,
      "learning_rate": 8.33668948852195e-06,
      "loss": 1.0244,
      "step": 2070
    },
    {
      "epoch": 0.1257025442678431,
      "grad_norm": 1.3061707549777584,
      "learning_rate": 8.37696335078534e-06,
      "loss": 1.053,
      "step": 2080
    },
    {
      "epoch": 0.12630688342297697,
      "grad_norm": 1.1123587859483497,
      "learning_rate": 8.417237213048733e-06,
      "loss": 1.0139,
      "step": 2090
    },
    {
      "epoch": 0.12691122257811083,
      "grad_norm": 1.4726481438754204,
      "learning_rate": 8.457511075312124e-06,
      "loss": 1.0373,
      "step": 2100
    },
    {
      "epoch": 0.1275155617332447,
      "grad_norm": 1.5540649554847172,
      "learning_rate": 8.497784937575514e-06,
      "loss": 1.034,
      "step": 2110
    },
    {
      "epoch": 0.12811990088837855,
      "grad_norm": 1.3649224248440421,
      "learning_rate": 8.538058799838905e-06,
      "loss": 1.0515,
      "step": 2120
    },
    {
      "epoch": 0.1287242400435124,
      "grad_norm": 1.4468487497471012,
      "learning_rate": 8.578332662102297e-06,
      "loss": 1.0284,
      "step": 2130
    },
    {
      "epoch": 0.1293285791986463,
      "grad_norm": 1.4178686802895129,
      "learning_rate": 8.618606524365687e-06,
      "loss": 1.0316,
      "step": 2140
    },
    {
      "epoch": 0.12993291835378015,
      "grad_norm": 1.309626194559123,
      "learning_rate": 8.658880386629078e-06,
      "loss": 1.0393,
      "step": 2150
    },
    {
      "epoch": 0.130537257508914,
      "grad_norm": 1.2593406051274751,
      "learning_rate": 8.69915424889247e-06,
      "loss": 1.0316,
      "step": 2160
    },
    {
      "epoch": 0.13114159666404787,
      "grad_norm": 1.2749406125528224,
      "learning_rate": 8.73942811115586e-06,
      "loss": 1.0715,
      "step": 2170
    },
    {
      "epoch": 0.13174593581918173,
      "grad_norm": 1.3292499035112644,
      "learning_rate": 8.779701973419251e-06,
      "loss": 1.0227,
      "step": 2180
    },
    {
      "epoch": 0.1323502749743156,
      "grad_norm": 1.2617489702802869,
      "learning_rate": 8.819975835682643e-06,
      "loss": 1.0398,
      "step": 2190
    },
    {
      "epoch": 0.13295461412944945,
      "grad_norm": 1.515868586228637,
      "learning_rate": 8.860249697946034e-06,
      "loss": 1.0317,
      "step": 2200
    },
    {
      "epoch": 0.1335589532845833,
      "grad_norm": 1.5160539289633963,
      "learning_rate": 8.900523560209426e-06,
      "loss": 1.0477,
      "step": 2210
    },
    {
      "epoch": 0.13416329243971717,
      "grad_norm": 1.4076986320932554,
      "learning_rate": 8.940797422472815e-06,
      "loss": 1.0418,
      "step": 2220
    },
    {
      "epoch": 0.13476763159485103,
      "grad_norm": 1.3689604228642756,
      "learning_rate": 8.981071284736207e-06,
      "loss": 1.0393,
      "step": 2230
    },
    {
      "epoch": 0.1353719707499849,
      "grad_norm": 1.3925084507461263,
      "learning_rate": 9.021345146999598e-06,
      "loss": 1.0398,
      "step": 2240
    },
    {
      "epoch": 0.13597630990511875,
      "grad_norm": 1.6226182151963606,
      "learning_rate": 9.061619009262988e-06,
      "loss": 1.0466,
      "step": 2250
    },
    {
      "epoch": 0.1365806490602526,
      "grad_norm": 1.5785671925779194,
      "learning_rate": 9.10189287152638e-06,
      "loss": 1.0375,
      "step": 2260
    },
    {
      "epoch": 0.13718498821538647,
      "grad_norm": 1.5336500255536512,
      "learning_rate": 9.14216673378977e-06,
      "loss": 1.0175,
      "step": 2270
    },
    {
      "epoch": 0.13778932737052033,
      "grad_norm": 1.5285248771423983,
      "learning_rate": 9.182440596053163e-06,
      "loss": 1.0381,
      "step": 2280
    },
    {
      "epoch": 0.13839366652565419,
      "grad_norm": 1.5128126964514035,
      "learning_rate": 9.222714458316554e-06,
      "loss": 1.019,
      "step": 2290
    },
    {
      "epoch": 0.13899800568078804,
      "grad_norm": 1.0193583899693481,
      "learning_rate": 9.262988320579944e-06,
      "loss": 1.037,
      "step": 2300
    },
    {
      "epoch": 0.13960234483592193,
      "grad_norm": 0.8968552638464535,
      "learning_rate": 9.303262182843336e-06,
      "loss": 1.0174,
      "step": 2310
    },
    {
      "epoch": 0.1402066839910558,
      "grad_norm": 0.9955662282804074,
      "learning_rate": 9.343536045106727e-06,
      "loss": 1.0504,
      "step": 2320
    },
    {
      "epoch": 0.14081102314618965,
      "grad_norm": 0.935242322809624,
      "learning_rate": 9.383809907370117e-06,
      "loss": 1.053,
      "step": 2330
    },
    {
      "epoch": 0.1414153623013235,
      "grad_norm": 0.9136266237484135,
      "learning_rate": 9.424083769633508e-06,
      "loss": 1.0206,
      "step": 2340
    },
    {
      "epoch": 0.14201970145645737,
      "grad_norm": 0.7990622585749204,
      "learning_rate": 9.4643576318969e-06,
      "loss": 1.0361,
      "step": 2350
    },
    {
      "epoch": 0.14262404061159123,
      "grad_norm": 0.818868112843548,
      "learning_rate": 9.50463149416029e-06,
      "loss": 1.0539,
      "step": 2360
    },
    {
      "epoch": 0.1432283797667251,
      "grad_norm": 0.8291376249740307,
      "learning_rate": 9.544905356423681e-06,
      "loss": 1.0226,
      "step": 2370
    },
    {
      "epoch": 0.14383271892185895,
      "grad_norm": 0.7710790067718756,
      "learning_rate": 9.585179218687073e-06,
      "loss": 1.0188,
      "step": 2380
    },
    {
      "epoch": 0.1444370580769928,
      "grad_norm": 0.7811007599244469,
      "learning_rate": 9.625453080950464e-06,
      "loss": 1.0157,
      "step": 2390
    },
    {
      "epoch": 0.14504139723212667,
      "grad_norm": 0.8911361442899138,
      "learning_rate": 9.665726943213854e-06,
      "loss": 1.0348,
      "step": 2400
    },
    {
      "epoch": 0.14564573638726053,
      "grad_norm": 0.9358415535339627,
      "learning_rate": 9.706000805477246e-06,
      "loss": 1.0594,
      "step": 2410
    },
    {
      "epoch": 0.1462500755423944,
      "grad_norm": 0.8755710396086944,
      "learning_rate": 9.746274667740637e-06,
      "loss": 1.0426,
      "step": 2420
    },
    {
      "epoch": 0.14685441469752825,
      "grad_norm": 0.8256900857975942,
      "learning_rate": 9.78654853000403e-06,
      "loss": 1.0327,
      "step": 2430
    },
    {
      "epoch": 0.1474587538526621,
      "grad_norm": 0.8268143487314841,
      "learning_rate": 9.826822392267418e-06,
      "loss": 1.0348,
      "step": 2440
    },
    {
      "epoch": 0.14806309300779597,
      "grad_norm": 0.6426856777426775,
      "learning_rate": 9.86709625453081e-06,
      "loss": 1.0486,
      "step": 2450
    },
    {
      "epoch": 0.14866743216292982,
      "grad_norm": 0.6186159057742388,
      "learning_rate": 9.9073701167942e-06,
      "loss": 1.0212,
      "step": 2460
    },
    {
      "epoch": 0.14927177131806368,
      "grad_norm": 0.5811580952257833,
      "learning_rate": 9.947643979057593e-06,
      "loss": 1.0096,
      "step": 2470
    },
    {
      "epoch": 0.14987611047319757,
      "grad_norm": 0.5708474838279459,
      "learning_rate": 9.987917841320983e-06,
      "loss": 1.0021,
      "step": 2480
    },
    {
      "epoch": 0.15048044962833143,
      "grad_norm": 0.5654727062145292,
      "learning_rate": 9.999999456342893e-06,
      "loss": 1.0364,
      "step": 2490
    },
    {
      "epoch": 0.1510847887834653,
      "grad_norm": 0.5826878155876029,
      "learning_rate": 9.999996793532857e-06,
      "loss": 1.034,
      "step": 2500
    },
    {
      "epoch": 0.15168912793859915,
      "grad_norm": 0.613278712920909,
      "learning_rate": 9.999991911715683e-06,
      "loss": 1.0449,
      "step": 2510
    },
    {
      "epoch": 0.152293467093733,
      "grad_norm": 0.6115079312969802,
      "learning_rate": 9.99998481089354e-06,
      "loss": 1.0383,
      "step": 2520
    },
    {
      "epoch": 0.15289780624886687,
      "grad_norm": 0.6226956803391436,
      "learning_rate": 9.99997549106958e-06,
      "loss": 1.0341,
      "step": 2530
    },
    {
      "epoch": 0.15350214540400073,
      "grad_norm": 0.6528092973933503,
      "learning_rate": 9.999963952247936e-06,
      "loss": 1.05,
      "step": 2540
    },
    {
      "epoch": 0.1541064845591346,
      "grad_norm": 1.1185585768666104,
      "learning_rate": 9.999950194433732e-06,
      "loss": 1.0414,
      "step": 2550
    },
    {
      "epoch": 0.15471082371426845,
      "grad_norm": 1.0223501722353807,
      "learning_rate": 9.999934217633071e-06,
      "loss": 1.0418,
      "step": 2560
    },
    {
      "epoch": 0.1553151628694023,
      "grad_norm": 1.0729184525617614,
      "learning_rate": 9.999916021853046e-06,
      "loss": 1.0401,
      "step": 2570
    },
    {
      "epoch": 0.15591950202453617,
      "grad_norm": 0.9819539057691418,
      "learning_rate": 9.99989560710173e-06,
      "loss": 1.0469,
      "step": 2580
    },
    {
      "epoch": 0.15652384117967003,
      "grad_norm": 1.0257784624253015,
      "learning_rate": 9.999872973388185e-06,
      "loss": 1.0339,
      "step": 2590
    },
    {
      "epoch": 0.15712818033480389,
      "grad_norm": 0.5320735974056858,
      "learning_rate": 9.999848120722456e-06,
      "loss": 1.0172,
      "step": 2600
    },
    {
      "epoch": 0.15773251948993774,
      "grad_norm": 0.5348157261770901,
      "learning_rate": 9.999821049115573e-06,
      "loss": 1.0427,
      "step": 2610
    },
    {
      "epoch": 0.1583368586450716,
      "grad_norm": 0.5782711821319967,
      "learning_rate": 9.999791758579547e-06,
      "loss": 1.0294,
      "step": 2620
    },
    {
      "epoch": 0.15894119780020546,
      "grad_norm": 0.5279519393941836,
      "learning_rate": 9.99976024912738e-06,
      "loss": 1.0266,
      "step": 2630
    },
    {
      "epoch": 0.15954553695533932,
      "grad_norm": 0.5313993134461803,
      "learning_rate": 9.999726520773057e-06,
      "loss": 1.0533,
      "step": 2640
    },
    {
      "epoch": 0.1601498761104732,
      "grad_norm": 0.4983785625267852,
      "learning_rate": 9.999690573531544e-06,
      "loss": 1.0012,
      "step": 2650
    },
    {
      "epoch": 0.16075421526560707,
      "grad_norm": 0.5487759545240649,
      "learning_rate": 9.999652407418796e-06,
      "loss": 1.0237,
      "step": 2660
    },
    {
      "epoch": 0.16135855442074093,
      "grad_norm": 0.49628698674465005,
      "learning_rate": 9.99961202245175e-06,
      "loss": 1.0526,
      "step": 2670
    },
    {
      "epoch": 0.1619628935758748,
      "grad_norm": 0.5927136450856338,
      "learning_rate": 9.99956941864833e-06,
      "loss": 1.0252,
      "step": 2680
    },
    {
      "epoch": 0.16256723273100865,
      "grad_norm": 0.5363941477651917,
      "learning_rate": 9.999524596027444e-06,
      "loss": 1.0062,
      "step": 2690
    },
    {
      "epoch": 0.1631715718861425,
      "grad_norm": 0.5615419870779127,
      "learning_rate": 9.999477554608985e-06,
      "loss": 1.0504,
      "step": 2700
    },
    {
      "epoch": 0.16377591104127637,
      "grad_norm": 0.5577826362507532,
      "learning_rate": 9.999428294413827e-06,
      "loss": 1.0271,
      "step": 2710
    },
    {
      "epoch": 0.16438025019641023,
      "grad_norm": 0.5715709287082611,
      "learning_rate": 9.999376815463833e-06,
      "loss": 1.0379,
      "step": 2720
    },
    {
      "epoch": 0.1649845893515441,
      "grad_norm": 0.5213940033970969,
      "learning_rate": 9.999323117781851e-06,
      "loss": 1.0103,
      "step": 2730
    },
    {
      "epoch": 0.16558892850667795,
      "grad_norm": 0.5315095683793392,
      "learning_rate": 9.999267201391711e-06,
      "loss": 1.0162,
      "step": 2740
    },
    {
      "epoch": 0.1661932676618118,
      "grad_norm": 0.6022435742068142,
      "learning_rate": 9.99920906631823e-06,
      "loss": 1.0169,
      "step": 2750
    },
    {
      "epoch": 0.16679760681694566,
      "grad_norm": 0.5210443489858492,
      "learning_rate": 9.999148712587207e-06,
      "loss": 1.0199,
      "step": 2760
    },
    {
      "epoch": 0.16740194597207952,
      "grad_norm": 0.5729742309285049,
      "learning_rate": 9.999086140225427e-06,
      "loss": 1.0172,
      "step": 2770
    },
    {
      "epoch": 0.16800628512721338,
      "grad_norm": 0.5076629992942371,
      "learning_rate": 9.99902134926066e-06,
      "loss": 1.0148,
      "step": 2780
    },
    {
      "epoch": 0.16861062428234724,
      "grad_norm": 0.5351987826516066,
      "learning_rate": 9.99895433972166e-06,
      "loss": 1.0086,
      "step": 2790
    },
    {
      "epoch": 0.1692149634374811,
      "grad_norm": 0.5547553571639994,
      "learning_rate": 9.998885111638167e-06,
      "loss": 1.0579,
      "step": 2800
    },
    {
      "epoch": 0.16981930259261496,
      "grad_norm": 0.5692818290723988,
      "learning_rate": 9.998813665040906e-06,
      "loss": 1.0353,
      "step": 2810
    },
    {
      "epoch": 0.17042364174774885,
      "grad_norm": 0.5548982707366165,
      "learning_rate": 9.99873999996158e-06,
      "loss": 1.0596,
      "step": 2820
    },
    {
      "epoch": 0.1710279809028827,
      "grad_norm": 0.5233844420318373,
      "learning_rate": 9.998664116432886e-06,
      "loss": 0.9952,
      "step": 2830
    },
    {
      "epoch": 0.17163232005801657,
      "grad_norm": 0.5082168638296541,
      "learning_rate": 9.9985860144885e-06,
      "loss": 1.0238,
      "step": 2840
    },
    {
      "epoch": 0.17223665921315043,
      "grad_norm": 0.5578950922481182,
      "learning_rate": 9.998505694163085e-06,
      "loss": 1.0326,
      "step": 2850
    },
    {
      "epoch": 0.1728409983682843,
      "grad_norm": 0.5657283177136884,
      "learning_rate": 9.998423155492285e-06,
      "loss": 1.0091,
      "step": 2860
    },
    {
      "epoch": 0.17344533752341815,
      "grad_norm": 0.5596485344170619,
      "learning_rate": 9.998338398512733e-06,
      "loss": 1.0367,
      "step": 2870
    },
    {
      "epoch": 0.174049676678552,
      "grad_norm": 0.51696431960722,
      "learning_rate": 9.998251423262042e-06,
      "loss": 0.997,
      "step": 2880
    },
    {
      "epoch": 0.17465401583368587,
      "grad_norm": 0.5435579243065897,
      "learning_rate": 9.998162229778813e-06,
      "loss": 1.0637,
      "step": 2890
    },
    {
      "epoch": 0.17525835498881973,
      "grad_norm": 0.6325614121742273,
      "learning_rate": 9.99807081810263e-06,
      "loss": 1.0168,
      "step": 2900
    },
    {
      "epoch": 0.17586269414395359,
      "grad_norm": 0.605587728278636,
      "learning_rate": 9.997977188274064e-06,
      "loss": 1.0797,
      "step": 2910
    },
    {
      "epoch": 0.17646703329908744,
      "grad_norm": 0.6757822217558517,
      "learning_rate": 9.997881340334663e-06,
      "loss": 1.0213,
      "step": 2920
    },
    {
      "epoch": 0.1770713724542213,
      "grad_norm": 0.6176956206212408,
      "learning_rate": 9.99778327432697e-06,
      "loss": 1.0246,
      "step": 2930
    },
    {
      "epoch": 0.17767571160935516,
      "grad_norm": 0.6548094212081415,
      "learning_rate": 9.997682990294503e-06,
      "loss": 1.0146,
      "step": 2940
    },
    {
      "epoch": 0.17828005076448902,
      "grad_norm": 0.6951349743054358,
      "learning_rate": 9.997580488281769e-06,
      "loss": 1.0141,
      "step": 2950
    },
    {
      "epoch": 0.17888438991962288,
      "grad_norm": 0.772414307533063,
      "learning_rate": 9.997475768334261e-06,
      "loss": 1.05,
      "step": 2960
    },
    {
      "epoch": 0.17948872907475674,
      "grad_norm": 0.6643376685897671,
      "learning_rate": 9.99736883049845e-06,
      "loss": 1.0208,
      "step": 2970
    },
    {
      "epoch": 0.1800930682298906,
      "grad_norm": 0.765221094147346,
      "learning_rate": 9.997259674821799e-06,
      "loss": 1.0196,
      "step": 2980
    },
    {
      "epoch": 0.1806974073850245,
      "grad_norm": 0.7531802530766694,
      "learning_rate": 9.997148301352748e-06,
      "loss": 1.0172,
      "step": 2990
    },
    {
      "epoch": 0.18130174654015835,
      "grad_norm": 0.6371992091469362,
      "learning_rate": 9.997034710140728e-06,
      "loss": 1.0202,
      "step": 3000
    },
    {
      "epoch": 0.1819060856952922,
      "grad_norm": 0.5948971392823025,
      "learning_rate": 9.996918901236148e-06,
      "loss": 1.0086,
      "step": 3010
    },
    {
      "epoch": 0.18251042485042607,
      "grad_norm": 0.6334731002766227,
      "learning_rate": 9.996800874690407e-06,
      "loss": 1.0226,
      "step": 3020
    },
    {
      "epoch": 0.18311476400555993,
      "grad_norm": 0.6550854027278942,
      "learning_rate": 9.996680630555882e-06,
      "loss": 1.0084,
      "step": 3030
    },
    {
      "epoch": 0.1837191031606938,
      "grad_norm": 0.6284135437513536,
      "learning_rate": 9.996558168885942e-06,
      "loss": 1.0125,
      "step": 3040
    },
    {
      "epoch": 0.18432344231582765,
      "grad_norm": 0.7855845439023009,
      "learning_rate": 9.996433489734931e-06,
      "loss": 1.052,
      "step": 3050
    },
    {
      "epoch": 0.1849277814709615,
      "grad_norm": 0.8434267981407233,
      "learning_rate": 9.996306593158185e-06,
      "loss": 1.0281,
      "step": 3060
    },
    {
      "epoch": 0.18553212062609536,
      "grad_norm": 0.8261274066025182,
      "learning_rate": 9.996177479212019e-06,
      "loss": 1.0349,
      "step": 3070
    },
    {
      "epoch": 0.18613645978122922,
      "grad_norm": 0.7670039231998137,
      "learning_rate": 9.996046147953735e-06,
      "loss": 1.0178,
      "step": 3080
    },
    {
      "epoch": 0.18674079893636308,
      "grad_norm": 0.8340417835072003,
      "learning_rate": 9.99591259944162e-06,
      "loss": 1.0132,
      "step": 3090
    },
    {
      "epoch": 0.18734513809149694,
      "grad_norm": 0.8074205706627108,
      "learning_rate": 9.995776833734938e-06,
      "loss": 1.0344,
      "step": 3100
    },
    {
      "epoch": 0.1879494772466308,
      "grad_norm": 0.7965585543505449,
      "learning_rate": 9.995638850893947e-06,
      "loss": 1.0396,
      "step": 3110
    },
    {
      "epoch": 0.18855381640176466,
      "grad_norm": 0.7987860336651947,
      "learning_rate": 9.995498650979883e-06,
      "loss": 1.0457,
      "step": 3120
    },
    {
      "epoch": 0.18915815555689852,
      "grad_norm": 0.8341307090500774,
      "learning_rate": 9.995356234054965e-06,
      "loss": 1.0452,
      "step": 3130
    },
    {
      "epoch": 0.18976249471203238,
      "grad_norm": 0.8043702412662023,
      "learning_rate": 9.995211600182397e-06,
      "loss": 1.0462,
      "step": 3140
    },
    {
      "epoch": 0.19036683386716624,
      "grad_norm": 1.287384332451675,
      "learning_rate": 9.995064749426373e-06,
      "loss": 1.0222,
      "step": 3150
    },
    {
      "epoch": 0.19097117302230013,
      "grad_norm": 1.329146487219306,
      "learning_rate": 9.99491568185206e-06,
      "loss": 1.0179,
      "step": 3160
    },
    {
      "epoch": 0.191575512177434,
      "grad_norm": 1.3550835898336,
      "learning_rate": 9.994764397525618e-06,
      "loss": 1.0194,
      "step": 3170
    },
    {
      "epoch": 0.19217985133256785,
      "grad_norm": 1.264847416855836,
      "learning_rate": 9.994610896514186e-06,
      "loss": 1.0217,
      "step": 3180
    },
    {
      "epoch": 0.1927841904877017,
      "grad_norm": 1.2899878303672319,
      "learning_rate": 9.994455178885887e-06,
      "loss": 1.0285,
      "step": 3190
    },
    {
      "epoch": 0.19338852964283557,
      "grad_norm": 1.336794198498458,
      "learning_rate": 9.99429724470983e-06,
      "loss": 1.0146,
      "step": 3200
    },
    {
      "epoch": 0.19399286879796943,
      "grad_norm": 1.1547500078832722,
      "learning_rate": 9.994137094056106e-06,
      "loss": 1.0192,
      "step": 3210
    },
    {
      "epoch": 0.19459720795310329,
      "grad_norm": 1.348857656618809,
      "learning_rate": 9.993974726995792e-06,
      "loss": 1.001,
      "step": 3220
    },
    {
      "epoch": 0.19520154710823714,
      "grad_norm": 1.2036929661433324,
      "learning_rate": 9.993810143600944e-06,
      "loss": 1.0154,
      "step": 3230
    },
    {
      "epoch": 0.195805886263371,
      "grad_norm": 2.037536987927584,
      "learning_rate": 9.993643343944606e-06,
      "loss": 1.0206,
      "step": 3240
    },
    {
      "epoch": 0.19641022541850486,
      "grad_norm": 2.5297444195829186,
      "learning_rate": 9.993474328100803e-06,
      "loss": 1.0485,
      "step": 3250
    },
    {
      "epoch": 0.19701456457363872,
      "grad_norm": 3.589903453677699,
      "learning_rate": 9.993303096144544e-06,
      "loss": 1.014,
      "step": 3260
    },
    {
      "epoch": 0.19761890372877258,
      "grad_norm": 2.516134148310326,
      "learning_rate": 9.993129648151824e-06,
      "loss": 1.0095,
      "step": 3270
    },
    {
      "epoch": 0.19822324288390644,
      "grad_norm": 2.7676889250136862,
      "learning_rate": 9.99295398419962e-06,
      "loss": 1.0119,
      "step": 3280
    },
    {
      "epoch": 0.1988275820390403,
      "grad_norm": 2.7667612656362923,
      "learning_rate": 9.992776104365889e-06,
      "loss": 1.0407,
      "step": 3290
    },
    {
      "epoch": 0.19943192119417416,
      "grad_norm": 1.146737343870628,
      "learning_rate": 9.992596008729576e-06,
      "loss": 1.0367,
      "step": 3300
    },
    {
      "epoch": 0.20003626034930802,
      "grad_norm": 1.3733146878560363,
      "learning_rate": 9.992413697370606e-06,
      "loss": 1.055,
      "step": 3310
    },
    {
      "epoch": 0.20064059950444188,
      "grad_norm": 1.1277796533841509,
      "learning_rate": 9.992229170369893e-06,
      "loss": 1.0271,
      "step": 3320
    },
    {
      "epoch": 0.20124493865957577,
      "grad_norm": 1.1534780442251698,
      "learning_rate": 9.992042427809328e-06,
      "loss": 1.0345,
      "step": 3330
    },
    {
      "epoch": 0.20184927781470963,
      "grad_norm": 1.1475591893423531,
      "learning_rate": 9.991853469771789e-06,
      "loss": 1.032,
      "step": 3340
    },
    {
      "epoch": 0.2024536169698435,
      "grad_norm": 1.1501540349651993,
      "learning_rate": 9.991662296341132e-06,
      "loss": 1.0389,
      "step": 3350
    },
    {
      "epoch": 0.20305795612497735,
      "grad_norm": 1.323771729498539,
      "learning_rate": 9.991468907602205e-06,
      "loss": 1.0451,
      "step": 3360
    },
    {
      "epoch": 0.2036622952801112,
      "grad_norm": 1.0373182991384071,
      "learning_rate": 9.99127330364083e-06,
      "loss": 0.9728,
      "step": 3370
    },
    {
      "epoch": 0.20426663443524506,
      "grad_norm": 1.0488182248336289,
      "learning_rate": 9.99107548454382e-06,
      "loss": 1.0142,
      "step": 3380
    },
    {
      "epoch": 0.20487097359037892,
      "grad_norm": 1.0604249314096938,
      "learning_rate": 9.990875450398964e-06,
      "loss": 1.0225,
      "step": 3390
    },
    {
      "epoch": 0.20547531274551278,
      "grad_norm": 1.2090228060976917,
      "learning_rate": 9.990673201295042e-06,
      "loss": 1.0389,
      "step": 3400
    },
    {
      "epoch": 0.20607965190064664,
      "grad_norm": 1.192454200611463,
      "learning_rate": 9.990468737321809e-06,
      "loss": 1.0291,
      "step": 3410
    },
    {
      "epoch": 0.2066839910557805,
      "grad_norm": 1.0957087501776068,
      "learning_rate": 9.990262058570007e-06,
      "loss": 1.0122,
      "step": 3420
    },
    {
      "epoch": 0.20728833021091436,
      "grad_norm": 1.0568327775353616,
      "learning_rate": 9.990053165131361e-06,
      "loss": 1.0539,
      "step": 3430
    },
    {
      "epoch": 0.20789266936604822,
      "grad_norm": 1.143890601431876,
      "learning_rate": 9.989842057098578e-06,
      "loss": 1.0208,
      "step": 3440
    },
    {
      "epoch": 0.20849700852118208,
      "grad_norm": 1.2152187663565115,
      "learning_rate": 9.989628734565347e-06,
      "loss": 1.0253,
      "step": 3450
    },
    {
      "epoch": 0.20910134767631594,
      "grad_norm": 1.151770858930067,
      "learning_rate": 9.989413197626343e-06,
      "loss": 1.0048,
      "step": 3460
    },
    {
      "epoch": 0.2097056868314498,
      "grad_norm": 1.191047547930404,
      "learning_rate": 9.98919544637722e-06,
      "loss": 1.0279,
      "step": 3470
    },
    {
      "epoch": 0.21031002598658366,
      "grad_norm": 1.118603549050447,
      "learning_rate": 9.988975480914618e-06,
      "loss": 1.0249,
      "step": 3480
    },
    {
      "epoch": 0.21091436514171752,
      "grad_norm": 1.2045164357250302,
      "learning_rate": 9.988753301336158e-06,
      "loss": 1.011,
      "step": 3490
    },
    {
      "epoch": 0.2115187042968514,
      "grad_norm": 1.2727860426424333,
      "learning_rate": 9.98852890774044e-06,
      "loss": 1.0109,
      "step": 3500
    },
    {
      "epoch": 0.21212304345198527,
      "grad_norm": 1.2232131903821466,
      "learning_rate": 9.988302300227055e-06,
      "loss": 1.0293,
      "step": 3510
    },
    {
      "epoch": 0.21272738260711913,
      "grad_norm": 1.1581226645248222,
      "learning_rate": 9.98807347889657e-06,
      "loss": 1.0228,
      "step": 3520
    },
    {
      "epoch": 0.21333172176225298,
      "grad_norm": 1.2449768024609256,
      "learning_rate": 9.987842443850536e-06,
      "loss": 1.0234,
      "step": 3530
    },
    {
      "epoch": 0.21393606091738684,
      "grad_norm": 1.1975927583401016,
      "learning_rate": 9.987609195191486e-06,
      "loss": 1.0134,
      "step": 3540
    },
    {
      "epoch": 0.2145404000725207,
      "grad_norm": 1.1875663787990882,
      "learning_rate": 9.987373733022937e-06,
      "loss": 1.0101,
      "step": 3550
    },
    {
      "epoch": 0.21514473922765456,
      "grad_norm": 1.1566686006538525,
      "learning_rate": 9.98713605744939e-06,
      "loss": 1.0206,
      "step": 3560
    },
    {
      "epoch": 0.21574907838278842,
      "grad_norm": 1.13108038867575,
      "learning_rate": 9.98689616857632e-06,
      "loss": 1.0198,
      "step": 3570
    },
    {
      "epoch": 0.21635341753792228,
      "grad_norm": 1.0467233101117432,
      "learning_rate": 9.986654066510196e-06,
      "loss": 1.0267,
      "step": 3580
    },
    {
      "epoch": 0.21695775669305614,
      "grad_norm": 1.0691146534191223,
      "learning_rate": 9.98640975135846e-06,
      "loss": 1.0397,
      "step": 3590
    },
    {
      "epoch": 0.21756209584819,
      "grad_norm": 1.271232751720568,
      "learning_rate": 9.98616322322954e-06,
      "loss": 1.0059,
      "step": 3600
    },
    {
      "epoch": 0.21816643500332386,
      "grad_norm": 1.0941437783791206,
      "learning_rate": 9.985914482232844e-06,
      "loss": 1.0254,
      "step": 3610
    },
    {
      "epoch": 0.21877077415845772,
      "grad_norm": 1.1539838269433296,
      "learning_rate": 9.985663528478769e-06,
      "loss": 1.0156,
      "step": 3620
    },
    {
      "epoch": 0.21937511331359158,
      "grad_norm": 1.1295717857607617,
      "learning_rate": 9.985410362078682e-06,
      "loss": 1.0279,
      "step": 3630
    },
    {
      "epoch": 0.21997945246872544,
      "grad_norm": 1.1362703389129263,
      "learning_rate": 9.985154983144945e-06,
      "loss": 1.0326,
      "step": 3640
    },
    {
      "epoch": 0.2205837916238593,
      "grad_norm": 1.2233926396929016,
      "learning_rate": 9.98489739179089e-06,
      "loss": 1.0251,
      "step": 3650
    },
    {
      "epoch": 0.22118813077899316,
      "grad_norm": 1.1422444246951031,
      "learning_rate": 9.98463758813084e-06,
      "loss": 1.0403,
      "step": 3660
    },
    {
      "epoch": 0.22179246993412705,
      "grad_norm": 1.180709702656107,
      "learning_rate": 9.984375572280096e-06,
      "loss": 1.024,
      "step": 3670
    },
    {
      "epoch": 0.2223968090892609,
      "grad_norm": 1.0663308403722467,
      "learning_rate": 9.98411134435494e-06,
      "loss": 1.02,
      "step": 3680
    },
    {
      "epoch": 0.22300114824439476,
      "grad_norm": 1.0303893693811321,
      "learning_rate": 9.983844904472638e-06,
      "loss": 1.0201,
      "step": 3690
    },
    {
      "epoch": 0.22360548739952862,
      "grad_norm": 1.4702708759476768,
      "learning_rate": 9.983576252751435e-06,
      "loss": 1.0119,
      "step": 3700
    },
    {
      "epoch": 0.22420982655466248,
      "grad_norm": 1.2812013733174148,
      "learning_rate": 9.983305389310558e-06,
      "loss": 1.0253,
      "step": 3710
    },
    {
      "epoch": 0.22481416570979634,
      "grad_norm": 1.2443679370614207,
      "learning_rate": 9.98303231427022e-06,
      "loss": 1.0095,
      "step": 3720
    },
    {
      "epoch": 0.2254185048649302,
      "grad_norm": 1.2591025755974963,
      "learning_rate": 9.982757027751611e-06,
      "loss": 1.0251,
      "step": 3730
    },
    {
      "epoch": 0.22602284402006406,
      "grad_norm": 1.2774968652029854,
      "learning_rate": 9.982479529876904e-06,
      "loss": 1.0263,
      "step": 3740
    },
    {
      "epoch": 0.22662718317519792,
      "grad_norm": 1.172663835151976,
      "learning_rate": 9.982199820769252e-06,
      "loss": 1.0097,
      "step": 3750
    },
    {
      "epoch": 0.22723152233033178,
      "grad_norm": 1.2073777258143432,
      "learning_rate": 9.981917900552788e-06,
      "loss": 1.0081,
      "step": 3760
    },
    {
      "epoch": 0.22783586148546564,
      "grad_norm": 1.1984285625601867,
      "learning_rate": 9.981633769352634e-06,
      "loss": 1.0356,
      "step": 3770
    },
    {
      "epoch": 0.2284402006405995,
      "grad_norm": 1.1323228359391508,
      "learning_rate": 9.981347427294885e-06,
      "loss": 1.0246,
      "step": 3780
    },
    {
      "epoch": 0.22904453979573336,
      "grad_norm": 1.2275135705760998,
      "learning_rate": 9.98105887450662e-06,
      "loss": 1.0136,
      "step": 3790
    },
    {
      "epoch": 0.22964887895086722,
      "grad_norm": 1.3589673185061513,
      "learning_rate": 9.9807681111159e-06,
      "loss": 1.0179,
      "step": 3800
    },
    {
      "epoch": 0.23025321810600108,
      "grad_norm": 1.3504035926727547,
      "learning_rate": 9.980475137251766e-06,
      "loss": 1.0197,
      "step": 3810
    },
    {
      "epoch": 0.23085755726113494,
      "grad_norm": 1.2839034309726234,
      "learning_rate": 9.980179953044239e-06,
      "loss": 1.0381,
      "step": 3820
    },
    {
      "epoch": 0.2314618964162688,
      "grad_norm": 1.285197256321224,
      "learning_rate": 9.979882558624324e-06,
      "loss": 1.013,
      "step": 3830
    },
    {
      "epoch": 0.23206623557140268,
      "grad_norm": 1.3469674248431074,
      "learning_rate": 9.979582954124005e-06,
      "loss": 1.0412,
      "step": 3840
    },
    {
      "epoch": 0.23267057472653654,
      "grad_norm": 1.5282597752166729,
      "learning_rate": 9.979281139676245e-06,
      "loss": 1.0061,
      "step": 3850
    },
    {
      "epoch": 0.2332749138816704,
      "grad_norm": 1.5147290838846537,
      "learning_rate": 9.978977115414993e-06,
      "loss": 1.0016,
      "step": 3860
    },
    {
      "epoch": 0.23387925303680426,
      "grad_norm": 1.5772951028976443,
      "learning_rate": 9.978670881475173e-06,
      "loss": 1.0347,
      "step": 3870
    },
    {
      "epoch": 0.23448359219193812,
      "grad_norm": 1.5110522107773365,
      "learning_rate": 9.978362437992694e-06,
      "loss": 1.0028,
      "step": 3880
    },
    {
      "epoch": 0.23508793134707198,
      "grad_norm": 1.4899685269225527,
      "learning_rate": 9.978051785104442e-06,
      "loss": 1.0226,
      "step": 3890
    },
    {
      "epoch": 0.23569227050220584,
      "grad_norm": 0.9005713878545523,
      "learning_rate": 9.977738922948285e-06,
      "loss": 1.0231,
      "step": 3900
    },
    {
      "epoch": 0.2362966096573397,
      "grad_norm": 0.9030844095327106,
      "learning_rate": 9.977423851663075e-06,
      "loss": 1.0343,
      "step": 3910
    },
    {
      "epoch": 0.23690094881247356,
      "grad_norm": 0.9815562803619603,
      "learning_rate": 9.977106571388637e-06,
      "loss": 1.0279,
      "step": 3920
    },
    {
      "epoch": 0.23750528796760742,
      "grad_norm": 0.9302840544241502,
      "learning_rate": 9.976787082265785e-06,
      "loss": 1.0502,
      "step": 3930
    },
    {
      "epoch": 0.23810962712274128,
      "grad_norm": 0.9790158151791362,
      "learning_rate": 9.976465384436302e-06,
      "loss": 1.0381,
      "step": 3940
    },
    {
      "epoch": 0.23871396627787514,
      "grad_norm": 0.8196333591376035,
      "learning_rate": 9.976141478042967e-06,
      "loss": 1.0325,
      "step": 3950
    },
    {
      "epoch": 0.239318305433009,
      "grad_norm": 0.8263757916793677,
      "learning_rate": 9.975815363229522e-06,
      "loss": 1.0205,
      "step": 3960
    },
    {
      "epoch": 0.23992264458814286,
      "grad_norm": 0.8824787795572878,
      "learning_rate": 9.975487040140702e-06,
      "loss": 1.0135,
      "step": 3970
    },
    {
      "epoch": 0.24052698374327672,
      "grad_norm": 0.793756744958486,
      "learning_rate": 9.975156508922216e-06,
      "loss": 1.0198,
      "step": 3980
    },
    {
      "epoch": 0.24113132289841058,
      "grad_norm": 0.8072068220285223,
      "learning_rate": 9.974823769720756e-06,
      "loss": 0.9994,
      "step": 3990
    },
    {
      "epoch": 0.24173566205354444,
      "grad_norm": 0.857445588862923,
      "learning_rate": 9.97448882268399e-06,
      "loss": 1.0383,
      "step": 4000
    },
    {
      "epoch": 0.24234000120867832,
      "grad_norm": 0.9131111310812462,
      "learning_rate": 9.974151667960568e-06,
      "loss": 1.0185,
      "step": 4010
    },
    {
      "epoch": 0.24294434036381218,
      "grad_norm": 0.8862050022736238,
      "learning_rate": 9.973812305700122e-06,
      "loss": 1.0319,
      "step": 4020
    },
    {
      "epoch": 0.24354867951894604,
      "grad_norm": 0.8330811647742326,
      "learning_rate": 9.973470736053259e-06,
      "loss": 1.0207,
      "step": 4030
    },
    {
      "epoch": 0.2441530186740799,
      "grad_norm": 0.8629570022678702,
      "learning_rate": 9.97312695917157e-06,
      "loss": 1.0304,
      "step": 4040
    },
    {
      "epoch": 0.24475735782921376,
      "grad_norm": 0.6508968864252784,
      "learning_rate": 9.972780975207624e-06,
      "loss": 1.0428,
      "step": 4050
    },
    {
      "epoch": 0.24536169698434762,
      "grad_norm": 0.5707817451030397,
      "learning_rate": 9.972432784314966e-06,
      "loss": 1.0409,
      "step": 4060
    },
    {
      "epoch": 0.24596603613948148,
      "grad_norm": 0.572935983022132,
      "learning_rate": 9.972082386648128e-06,
      "loss": 0.9872,
      "step": 4070
    },
    {
      "epoch": 0.24657037529461534,
      "grad_norm": 0.5663971776134257,
      "learning_rate": 9.971729782362614e-06,
      "loss": 0.9942,
      "step": 4080
    },
    {
      "epoch": 0.2471747144497492,
      "grad_norm": 0.5651070696370811,
      "learning_rate": 9.971374971614914e-06,
      "loss": 1.0043,
      "step": 4090
    },
    {
      "epoch": 0.24777905360488306,
      "grad_norm": 0.636604234486725,
      "learning_rate": 9.97101795456249e-06,
      "loss": 1.0289,
      "step": 4100
    },
    {
      "epoch": 0.24838339276001692,
      "grad_norm": 0.5719926954401137,
      "learning_rate": 9.970658731363788e-06,
      "loss": 1.0167,
      "step": 4110
    },
    {
      "epoch": 0.24898773191515078,
      "grad_norm": 0.6195712000662134,
      "learning_rate": 9.970297302178232e-06,
      "loss": 0.9972,
      "step": 4120
    },
    {
      "epoch": 0.24959207107028464,
      "grad_norm": 0.6320467725781482,
      "learning_rate": 9.969933667166222e-06,
      "loss": 0.9918,
      "step": 4130
    },
    {
      "epoch": 0.2501964102254185,
      "grad_norm": 0.6169397715301146,
      "learning_rate": 9.969567826489146e-06,
      "loss": 0.9968,
      "step": 4140
    },
    {
      "epoch": 0.25080074938055236,
      "grad_norm": 0.884429425115163,
      "learning_rate": 9.969199780309362e-06,
      "loss": 0.9993,
      "step": 4150
    },
    {
      "epoch": 0.2514050885356862,
      "grad_norm": 0.8758021286667406,
      "learning_rate": 9.968829528790207e-06,
      "loss": 1.0023,
      "step": 4160
    },
    {
      "epoch": 0.2520094276908201,
      "grad_norm": 0.9614454188025886,
      "learning_rate": 9.968457072096e-06,
      "loss": 1.02,
      "step": 4170
    },
    {
      "epoch": 0.25261376684595394,
      "grad_norm": 0.8516183537717978,
      "learning_rate": 9.968082410392041e-06,
      "loss": 0.9993,
      "step": 4180
    },
    {
      "epoch": 0.2532181060010878,
      "grad_norm": 0.9340873207567952,
      "learning_rate": 9.967705543844603e-06,
      "loss": 1.0084,
      "step": 4190
    },
    {
      "epoch": 0.25382244515622165,
      "grad_norm": 0.5969120148329723,
      "learning_rate": 9.967326472620942e-06,
      "loss": 1.0358,
      "step": 4200
    },
    {
      "epoch": 0.2544267843113555,
      "grad_norm": 0.5532832780491006,
      "learning_rate": 9.966945196889287e-06,
      "loss": 1.0309,
      "step": 4210
    },
    {
      "epoch": 0.2550311234664894,
      "grad_norm": 0.5279430922926298,
      "learning_rate": 9.966561716818852e-06,
      "loss": 1.0077,
      "step": 4220
    },
    {
      "epoch": 0.25563546262162323,
      "grad_norm": 0.5284510467388175,
      "learning_rate": 9.966176032579824e-06,
      "loss": 1.0122,
      "step": 4230
    },
    {
      "epoch": 0.2562398017767571,
      "grad_norm": 0.5256113886656796,
      "learning_rate": 9.965788144343372e-06,
      "loss": 1.0103,
      "step": 4240
    },
    {
      "epoch": 0.25684414093189095,
      "grad_norm": 0.4921708474117513,
      "learning_rate": 9.96539805228164e-06,
      "loss": 1.0124,
      "step": 4250
    },
    {
      "epoch": 0.2574484800870248,
      "grad_norm": 0.5330450591954504,
      "learning_rate": 9.965005756567754e-06,
      "loss": 1.0309,
      "step": 4260
    },
    {
      "epoch": 0.2580528192421587,
      "grad_norm": 0.5434351708906063,
      "learning_rate": 9.964611257375814e-06,
      "loss": 0.991,
      "step": 4270
    },
    {
      "epoch": 0.2586571583972926,
      "grad_norm": 0.5063139026797504,
      "learning_rate": 9.964214554880898e-06,
      "loss": 0.9929,
      "step": 4280
    },
    {
      "epoch": 0.25926149755242645,
      "grad_norm": 0.5173441985223839,
      "learning_rate": 9.963815649259064e-06,
      "loss": 1.0108,
      "step": 4290
    },
    {
      "epoch": 0.2598658367075603,
      "grad_norm": 0.5302461932782329,
      "learning_rate": 9.963414540687348e-06,
      "loss": 0.9879,
      "step": 4300
    },
    {
      "epoch": 0.26047017586269416,
      "grad_norm": 0.6343881137108661,
      "learning_rate": 9.963011229343762e-06,
      "loss": 1.0073,
      "step": 4310
    },
    {
      "epoch": 0.261074515017828,
      "grad_norm": 0.5665421291453269,
      "learning_rate": 9.962605715407296e-06,
      "loss": 0.9907,
      "step": 4320
    },
    {
      "epoch": 0.2616788541729619,
      "grad_norm": 0.526193747043052,
      "learning_rate": 9.962197999057919e-06,
      "loss": 1.0163,
      "step": 4330
    },
    {
      "epoch": 0.26228319332809574,
      "grad_norm": 0.5122318015872429,
      "learning_rate": 9.961788080476575e-06,
      "loss": 0.9979,
      "step": 4340
    },
    {
      "epoch": 0.2628875324832296,
      "grad_norm": 0.5175499896023672,
      "learning_rate": 9.961375959845187e-06,
      "loss": 1.0058,
      "step": 4350
    },
    {
      "epoch": 0.26349187163836346,
      "grad_norm": 0.5270647330903278,
      "learning_rate": 9.960961637346655e-06,
      "loss": 0.9961,
      "step": 4360
    },
    {
      "epoch": 0.2640962107934973,
      "grad_norm": 0.5003990334724352,
      "learning_rate": 9.960545113164854e-06,
      "loss": 1.0146,
      "step": 4370
    },
    {
      "epoch": 0.2647005499486312,
      "grad_norm": 0.51902917883346,
      "learning_rate": 9.960126387484643e-06,
      "loss": 1.0379,
      "step": 4380
    },
    {
      "epoch": 0.26530488910376504,
      "grad_norm": 0.5345674078416652,
      "learning_rate": 9.959705460491848e-06,
      "loss": 0.9949,
      "step": 4390
    },
    {
      "epoch": 0.2659092282588989,
      "grad_norm": 0.5310656521158249,
      "learning_rate": 9.959282332373281e-06,
      "loss": 0.9944,
      "step": 4400
    },
    {
      "epoch": 0.26651356741403276,
      "grad_norm": 0.513252930661846,
      "learning_rate": 9.958857003316725e-06,
      "loss": 1.0199,
      "step": 4410
    },
    {
      "epoch": 0.2671179065691666,
      "grad_norm": 0.5496207671544353,
      "learning_rate": 9.95842947351094e-06,
      "loss": 0.998,
      "step": 4420
    },
    {
      "epoch": 0.2677222457243005,
      "grad_norm": 0.5423231200415501,
      "learning_rate": 9.957999743145668e-06,
      "loss": 1.0302,
      "step": 4430
    },
    {
      "epoch": 0.26832658487943434,
      "grad_norm": 0.5350558941911503,
      "learning_rate": 9.957567812411621e-06,
      "loss": 1.034,
      "step": 4440
    },
    {
      "epoch": 0.2689309240345682,
      "grad_norm": 0.5517118363987205,
      "learning_rate": 9.957133681500495e-06,
      "loss": 0.9973,
      "step": 4450
    },
    {
      "epoch": 0.26953526318970206,
      "grad_norm": 0.5540472383363395,
      "learning_rate": 9.956697350604954e-06,
      "loss": 0.9975,
      "step": 4460
    },
    {
      "epoch": 0.2701396023448359,
      "grad_norm": 0.5297168446048597,
      "learning_rate": 9.956258819918643e-06,
      "loss": 0.9899,
      "step": 4470
    },
    {
      "epoch": 0.2707439414999698,
      "grad_norm": 0.5390981994209962,
      "learning_rate": 9.955818089636184e-06,
      "loss": 1.0248,
      "step": 4480
    },
    {
      "epoch": 0.27134828065510364,
      "grad_norm": 0.5247451214636177,
      "learning_rate": 9.955375159953171e-06,
      "loss": 1.0333,
      "step": 4490
    },
    {
      "epoch": 0.2719526198102375,
      "grad_norm": 0.5966037843707463,
      "learning_rate": 9.95493003106618e-06,
      "loss": 1.004,
      "step": 4500
    },
    {
      "epoch": 0.27255695896537135,
      "grad_norm": 0.6209620801829305,
      "learning_rate": 9.954482703172761e-06,
      "loss": 1.0113,
      "step": 4510
    },
    {
      "epoch": 0.2731612981205052,
      "grad_norm": 0.6082941355882645,
      "learning_rate": 9.954033176471436e-06,
      "loss": 0.9989,
      "step": 4520
    },
    {
      "epoch": 0.2737656372756391,
      "grad_norm": 0.6614217834772699,
      "learning_rate": 9.953581451161707e-06,
      "loss": 0.9866,
      "step": 4530
    },
    {
      "epoch": 0.27436997643077293,
      "grad_norm": 0.6350953149853057,
      "learning_rate": 9.953127527444049e-06,
      "loss": 1.0094,
      "step": 4540
    },
    {
      "epoch": 0.2749743155859068,
      "grad_norm": 0.7132217019139976,
      "learning_rate": 9.952671405519916e-06,
      "loss": 1.0053,
      "step": 4550
    },
    {
      "epoch": 0.27557865474104065,
      "grad_norm": 0.757638856045083,
      "learning_rate": 9.952213085591734e-06,
      "loss": 1.0183,
      "step": 4560
    },
    {
      "epoch": 0.2761829938961745,
      "grad_norm": 0.7079795324613742,
      "learning_rate": 9.951752567862906e-06,
      "loss": 1.0161,
      "step": 4570
    },
    {
      "epoch": 0.27678733305130837,
      "grad_norm": 0.6734659548315652,
      "learning_rate": 9.951289852537815e-06,
      "loss": 1.0004,
      "step": 4580
    },
    {
      "epoch": 0.27739167220644223,
      "grad_norm": 0.6941761188175806,
      "learning_rate": 9.950824939821809e-06,
      "loss": 1.0075,
      "step": 4590
    },
    {
      "epoch": 0.2779960113615761,
      "grad_norm": 0.6171635347372765,
      "learning_rate": 9.95035782992122e-06,
      "loss": 1.0191,
      "step": 4600
    },
    {
      "epoch": 0.27860035051671,
      "grad_norm": 0.6581497918130428,
      "learning_rate": 9.949888523043351e-06,
      "loss": 1.0392,
      "step": 4610
    },
    {
      "epoch": 0.27920468967184386,
      "grad_norm": 0.6181277058300175,
      "learning_rate": 9.949417019396483e-06,
      "loss": 1.0019,
      "step": 4620
    },
    {
      "epoch": 0.2798090288269777,
      "grad_norm": 0.6137306419982177,
      "learning_rate": 9.94894331918987e-06,
      "loss": 1.009,
      "step": 4630
    },
    {
      "epoch": 0.2804133679821116,
      "grad_norm": 0.6076993344746096,
      "learning_rate": 9.94846742263374e-06,
      "loss": 1.0072,
      "step": 4640
    },
    {
      "epoch": 0.28101770713724544,
      "grad_norm": 0.7929459926101928,
      "learning_rate": 9.947989329939294e-06,
      "loss": 1.0214,
      "step": 4650
    },
    {
      "epoch": 0.2816220462923793,
      "grad_norm": 0.7684915999055225,
      "learning_rate": 9.947509041318715e-06,
      "loss": 1.0013,
      "step": 4660
    },
    {
      "epoch": 0.28222638544751316,
      "grad_norm": 0.8202686356804075,
      "learning_rate": 9.947026556985154e-06,
      "loss": 1.0062,
      "step": 4670
    },
    {
      "epoch": 0.282830724602647,
      "grad_norm": 0.8226214402110111,
      "learning_rate": 9.946541877152738e-06,
      "loss": 1.0075,
      "step": 4680
    },
    {
      "epoch": 0.2834350637577809,
      "grad_norm": 0.7856983742487061,
      "learning_rate": 9.946055002036569e-06,
      "loss": 1.0028,
      "step": 4690
    },
    {
      "epoch": 0.28403940291291474,
      "grad_norm": 0.8505254408387751,
      "learning_rate": 9.945565931852724e-06,
      "loss": 1.0311,
      "step": 4700
    },
    {
      "epoch": 0.2846437420680486,
      "grad_norm": 0.7675633909806334,
      "learning_rate": 9.945074666818251e-06,
      "loss": 0.9977,
      "step": 4710
    },
    {
      "epoch": 0.28524808122318246,
      "grad_norm": 0.7997794300046326,
      "learning_rate": 9.944581207151176e-06,
      "loss": 0.9962,
      "step": 4720
    },
    {
      "epoch": 0.2858524203783163,
      "grad_norm": 0.7720538634296082,
      "learning_rate": 9.944085553070498e-06,
      "loss": 1.0163,
      "step": 4730
    },
    {
      "epoch": 0.2864567595334502,
      "grad_norm": 0.8225570149364019,
      "learning_rate": 9.943587704796185e-06,
      "loss": 1.0324,
      "step": 4740
    },
    {
      "epoch": 0.28706109868858404,
      "grad_norm": 1.1807604509175111,
      "learning_rate": 9.943087662549187e-06,
      "loss": 1.0037,
      "step": 4750
    },
    {
      "epoch": 0.2876654378437179,
      "grad_norm": 1.3134790369548701,
      "learning_rate": 9.942585426551422e-06,
      "loss": 1.0346,
      "step": 4760
    },
    {
      "epoch": 0.28826977699885176,
      "grad_norm": 1.2530231629475332,
      "learning_rate": 9.942080997025787e-06,
      "loss": 0.9849,
      "step": 4770
    },
    {
      "epoch": 0.2888741161539856,
      "grad_norm": 1.3582193716833795,
      "learning_rate": 9.941574374196143e-06,
      "loss": 1.0231,
      "step": 4780
    },
    {
      "epoch": 0.2894784553091195,
      "grad_norm": 1.3835079302012114,
      "learning_rate": 9.941065558287332e-06,
      "loss": 1.016,
      "step": 4790
    },
    {
      "epoch": 0.29008279446425334,
      "grad_norm": 0.967025199831158,
      "learning_rate": 9.940554549525168e-06,
      "loss": 1.032,
      "step": 4800
    },
    {
      "epoch": 0.2906871336193872,
      "grad_norm": 0.9112362947015128,
      "learning_rate": 9.940041348136439e-06,
      "loss": 1.0212,
      "step": 4810
    },
    {
      "epoch": 0.29129147277452105,
      "grad_norm": 1.1436856824772292,
      "learning_rate": 9.939525954348902e-06,
      "loss": 1.0162,
      "step": 4820
    },
    {
      "epoch": 0.2918958119296549,
      "grad_norm": 1.0067462433369134,
      "learning_rate": 9.93900836839129e-06,
      "loss": 1.0107,
      "step": 4830
    },
    {
      "epoch": 0.2925001510847888,
      "grad_norm": 1.1024454895275722,
      "learning_rate": 9.938488590493307e-06,
      "loss": 1.0226,
      "step": 4840
    },
    {
      "epoch": 0.29310449023992263,
      "grad_norm": 2.656804042917146,
      "learning_rate": 9.937966620885637e-06,
      "loss": 1.0367,
      "step": 4850
    },
    {
      "epoch": 0.2937088293950565,
      "grad_norm": 2.370309401156759,
      "learning_rate": 9.937442459799927e-06,
      "loss": 1.0053,
      "step": 4860
    },
    {
      "epoch": 0.29431316855019035,
      "grad_norm": 2.6888181390566563,
      "learning_rate": 9.9369161074688e-06,
      "loss": 1.0289,
      "step": 4870
    },
    {
      "epoch": 0.2949175077053242,
      "grad_norm": 2.5265543603310747,
      "learning_rate": 9.936387564125853e-06,
      "loss": 0.9984,
      "step": 4880
    },
    {
      "epoch": 0.29552184686045807,
      "grad_norm": 2.6560780585820334,
      "learning_rate": 9.935856830005654e-06,
      "loss": 1.0069,
      "step": 4890
    },
    {
      "epoch": 0.29612618601559193,
      "grad_norm": 1.0950460917732527,
      "learning_rate": 9.935323905343746e-06,
      "loss": 0.9931,
      "step": 4900
    },
    {
      "epoch": 0.2967305251707258,
      "grad_norm": 1.3820175656653786,
      "learning_rate": 9.93478879037664e-06,
      "loss": 1.0001,
      "step": 4910
    },
    {
      "epoch": 0.29733486432585965,
      "grad_norm": 1.207982565293866,
      "learning_rate": 9.934251485341822e-06,
      "loss": 1.0235,
      "step": 4920
    },
    {
      "epoch": 0.2979392034809935,
      "grad_norm": 1.0379262494990424,
      "learning_rate": 9.933711990477746e-06,
      "loss": 0.9995,
      "step": 4930
    },
    {
      "epoch": 0.29854354263612737,
      "grad_norm": 1.0061289961358295,
      "learning_rate": 9.933170306023844e-06,
      "loss": 0.9947,
      "step": 4940
    },
    {
      "epoch": 0.2991478817912613,
      "grad_norm": 1.1629824628189593,
      "learning_rate": 9.932626432220515e-06,
      "loss": 1.0218,
      "step": 4950
    },
    {
      "epoch": 0.29975222094639514,
      "grad_norm": 1.0745583116485489,
      "learning_rate": 9.932080369309131e-06,
      "loss": 1.0466,
      "step": 4960
    },
    {
      "epoch": 0.300356560101529,
      "grad_norm": 0.9771012827705086,
      "learning_rate": 9.931532117532036e-06,
      "loss": 0.9814,
      "step": 4970
    },
    {
      "epoch": 0.30096089925666286,
      "grad_norm": 1.049345463556189,
      "learning_rate": 9.930981677132544e-06,
      "loss": 1.004,
      "step": 4980
    },
    {
      "epoch": 0.3015652384117967,
      "grad_norm": 1.0279316367099933,
      "learning_rate": 9.930429048354944e-06,
      "loss": 0.9926,
      "step": 4990
    },
    {
      "epoch": 0.3021695775669306,
      "grad_norm": 1.0842718547937733,
      "learning_rate": 9.929874231444492e-06,
      "loss": 0.9865,
      "step": 5000
    },
    {
      "epoch": 0.30277391672206444,
      "grad_norm": 1.128936697247095,
      "learning_rate": 9.929317226647417e-06,
      "loss": 1.0075,
      "step": 5010
    },
    {
      "epoch": 0.3033782558771983,
      "grad_norm": 1.0606267020366618,
      "learning_rate": 9.92875803421092e-06,
      "loss": 0.9727,
      "step": 5020
    },
    {
      "epoch": 0.30398259503233216,
      "grad_norm": 1.1121105791161146,
      "learning_rate": 9.928196654383167e-06,
      "loss": 1.0158,
      "step": 5030
    },
    {
      "epoch": 0.304586934187466,
      "grad_norm": 1.209476126183107,
      "learning_rate": 9.927633087413304e-06,
      "loss": 1.0174,
      "step": 5040
    },
    {
      "epoch": 0.3051912733425999,
      "grad_norm": 1.1204506000640373,
      "learning_rate": 9.92706733355144e-06,
      "loss": 0.992,
      "step": 5050
    },
    {
      "epoch": 0.30579561249773374,
      "grad_norm": 1.1103095771812197,
      "learning_rate": 9.92649939304866e-06,
      "loss": 0.9754,
      "step": 5060
    },
    {
      "epoch": 0.3063999516528676,
      "grad_norm": 1.0298429237223206,
      "learning_rate": 9.925929266157016e-06,
      "loss": 1.0075,
      "step": 5070
    },
    {
      "epoch": 0.30700429080800146,
      "grad_norm": 1.1301911418300359,
      "learning_rate": 9.92535695312953e-06,
      "loss": 1.0109,
      "step": 5080
    },
    {
      "epoch": 0.3076086299631353,
      "grad_norm": 1.12017833927496,
      "learning_rate": 9.9247824542202e-06,
      "loss": 0.9939,
      "step": 5090
    },
    {
      "epoch": 0.3082129691182692,
      "grad_norm": 1.2825944867534116,
      "learning_rate": 9.924205769683981e-06,
      "loss": 1.0008,
      "step": 5100
    },
    {
      "epoch": 0.30881730827340304,
      "grad_norm": 1.1819104499761663,
      "learning_rate": 9.923626899776815e-06,
      "loss": 1.0218,
      "step": 5110
    },
    {
      "epoch": 0.3094216474285369,
      "grad_norm": 1.2165172294213984,
      "learning_rate": 9.923045844755602e-06,
      "loss": 0.9936,
      "step": 5120
    },
    {
      "epoch": 0.31002598658367075,
      "grad_norm": 1.214728876533914,
      "learning_rate": 9.922462604878216e-06,
      "loss": 1.0375,
      "step": 5130
    },
    {
      "epoch": 0.3106303257388046,
      "grad_norm": 1.1463835104012132,
      "learning_rate": 9.921877180403499e-06,
      "loss": 0.9946,
      "step": 5140
    },
    {
      "epoch": 0.3112346648939385,
      "grad_norm": 0.9960379761045224,
      "learning_rate": 9.921289571591263e-06,
      "loss": 0.9939,
      "step": 5150
    },
    {
      "epoch": 0.31183900404907233,
      "grad_norm": 1.0260036330591795,
      "learning_rate": 9.92069977870229e-06,
      "loss": 0.9972,
      "step": 5160
    },
    {
      "epoch": 0.3124433432042062,
      "grad_norm": 1.1191230761992401,
      "learning_rate": 9.920107801998332e-06,
      "loss": 1.0171,
      "step": 5170
    },
    {
      "epoch": 0.31304768235934005,
      "grad_norm": 0.9602369585823124,
      "learning_rate": 9.91951364174211e-06,
      "loss": 1.0161,
      "step": 5180
    },
    {
      "epoch": 0.3136520215144739,
      "grad_norm": 1.0766618253928928,
      "learning_rate": 9.918917298197313e-06,
      "loss": 0.997,
      "step": 5190
    },
    {
      "epoch": 0.31425636066960777,
      "grad_norm": 1.1470098083236124,
      "learning_rate": 9.918318771628596e-06,
      "loss": 1.0065,
      "step": 5200
    },
    {
      "epoch": 0.31486069982474163,
      "grad_norm": 1.0899646982853766,
      "learning_rate": 9.91771806230159e-06,
      "loss": 1.0135,
      "step": 5210
    },
    {
      "epoch": 0.3154650389798755,
      "grad_norm": 1.1483469336894205,
      "learning_rate": 9.91711517048289e-06,
      "loss": 0.9953,
      "step": 5220
    },
    {
      "epoch": 0.31606937813500935,
      "grad_norm": 1.151446109822897,
      "learning_rate": 9.916510096440059e-06,
      "loss": 1.0166,
      "step": 5230
    },
    {
      "epoch": 0.3166737172901432,
      "grad_norm": 1.170085922207658,
      "learning_rate": 9.91590284044163e-06,
      "loss": 1.0046,
      "step": 5240
    },
    {
      "epoch": 0.31727805644527707,
      "grad_norm": 1.108059940177387,
      "learning_rate": 9.915293402757106e-06,
      "loss": 0.9921,
      "step": 5250
    },
    {
      "epoch": 0.3178823956004109,
      "grad_norm": 1.050844949783572,
      "learning_rate": 9.914681783656956e-06,
      "loss": 0.9938,
      "step": 5260
    },
    {
      "epoch": 0.3184867347555448,
      "grad_norm": 1.0946051348771029,
      "learning_rate": 9.914067983412616e-06,
      "loss": 1.022,
      "step": 5270
    },
    {
      "epoch": 0.31909107391067865,
      "grad_norm": 1.1085941043245624,
      "learning_rate": 9.913452002296492e-06,
      "loss": 1.0047,
      "step": 5280
    },
    {
      "epoch": 0.31969541306581256,
      "grad_norm": 1.0868859335173506,
      "learning_rate": 9.912833840581961e-06,
      "loss": 1.001,
      "step": 5290
    },
    {
      "epoch": 0.3202997522209464,
      "grad_norm": 1.2713784186567478,
      "learning_rate": 9.912213498543359e-06,
      "loss": 0.9886,
      "step": 5300
    },
    {
      "epoch": 0.3209040913760803,
      "grad_norm": 1.3650321918354733,
      "learning_rate": 9.911590976456e-06,
      "loss": 1.0351,
      "step": 5310
    },
    {
      "epoch": 0.32150843053121414,
      "grad_norm": 1.254607633304583,
      "learning_rate": 9.910966274596154e-06,
      "loss": 0.9978,
      "step": 5320
    },
    {
      "epoch": 0.322112769686348,
      "grad_norm": 1.211135276855096,
      "learning_rate": 9.91033939324107e-06,
      "loss": 1.0121,
      "step": 5330
    },
    {
      "epoch": 0.32271710884148186,
      "grad_norm": 1.1815895855599206,
      "learning_rate": 9.90971033266896e-06,
      "loss": 1.0034,
      "step": 5340
    },
    {
      "epoch": 0.3233214479966157,
      "grad_norm": 1.2369326168307309,
      "learning_rate": 9.909079093158997e-06,
      "loss": 1.041,
      "step": 5350
    },
    {
      "epoch": 0.3239257871517496,
      "grad_norm": 1.1079950575888624,
      "learning_rate": 9.908445674991328e-06,
      "loss": 1.0083,
      "step": 5360
    },
    {
      "epoch": 0.32453012630688344,
      "grad_norm": 1.1972584951271126,
      "learning_rate": 9.907810078447066e-06,
      "loss": 0.9926,
      "step": 5370
    },
    {
      "epoch": 0.3251344654620173,
      "grad_norm": 1.1002416731465097,
      "learning_rate": 9.907172303808293e-06,
      "loss": 1.0124,
      "step": 5380
    },
    {
      "epoch": 0.32573880461715116,
      "grad_norm": 1.08929349089889,
      "learning_rate": 9.906532351358047e-06,
      "loss": 1.0079,
      "step": 5390
    },
    {
      "epoch": 0.326343143772285,
      "grad_norm": 1.2489750798564807,
      "learning_rate": 9.905890221380347e-06,
      "loss": 1.0056,
      "step": 5400
    },
    {
      "epoch": 0.3269474829274189,
      "grad_norm": 1.232673278335761,
      "learning_rate": 9.905245914160167e-06,
      "loss": 1.0167,
      "step": 5410
    },
    {
      "epoch": 0.32755182208255273,
      "grad_norm": 1.3690124382242743,
      "learning_rate": 9.904599429983452e-06,
      "loss": 1.037,
      "step": 5420
    },
    {
      "epoch": 0.3281561612376866,
      "grad_norm": 1.2611167113637718,
      "learning_rate": 9.903950769137116e-06,
      "loss": 1.0327,
      "step": 5430
    },
    {
      "epoch": 0.32876050039282045,
      "grad_norm": 1.2642450239980394,
      "learning_rate": 9.903299931909034e-06,
      "loss": 1.0029,
      "step": 5440
    },
    {
      "epoch": 0.3293648395479543,
      "grad_norm": 1.4616179750325198,
      "learning_rate": 9.902646918588048e-06,
      "loss": 0.9931,
      "step": 5450
    },
    {
      "epoch": 0.3299691787030882,
      "grad_norm": 1.4897412660016676,
      "learning_rate": 9.901991729463966e-06,
      "loss": 1.0161,
      "step": 5460
    },
    {
      "epoch": 0.33057351785822203,
      "grad_norm": 1.414515926955237,
      "learning_rate": 9.901334364827564e-06,
      "loss": 0.9849,
      "step": 5470
    },
    {
      "epoch": 0.3311778570133559,
      "grad_norm": 1.3243483131524123,
      "learning_rate": 9.900674824970579e-06,
      "loss": 1.0187,
      "step": 5480
    },
    {
      "epoch": 0.33178219616848975,
      "grad_norm": 1.4291180712523233,
      "learning_rate": 9.900013110185717e-06,
      "loss": 1.0251,
      "step": 5490
    },
    {
      "epoch": 0.3323865353236236,
      "grad_norm": 0.8870838433599929,
      "learning_rate": 9.89934922076665e-06,
      "loss": 0.9663,
      "step": 5500
    },
    {
      "epoch": 0.33299087447875747,
      "grad_norm": 0.8599778729162925,
      "learning_rate": 9.898683157008012e-06,
      "loss": 1.0182,
      "step": 5510
    },
    {
      "epoch": 0.33359521363389133,
      "grad_norm": 0.8685420749064647,
      "learning_rate": 9.898014919205403e-06,
      "loss": 1.0133,
      "step": 5520
    },
    {
      "epoch": 0.3341995527890252,
      "grad_norm": 0.9190668518359403,
      "learning_rate": 9.897344507655387e-06,
      "loss": 1.0025,
      "step": 5530
    },
    {
      "epoch": 0.33480389194415905,
      "grad_norm": 0.8452117875235303,
      "learning_rate": 9.896671922655497e-06,
      "loss": 1.0346,
      "step": 5540
    },
    {
      "epoch": 0.3354082310992929,
      "grad_norm": 0.7608069571147485,
      "learning_rate": 9.895997164504224e-06,
      "loss": 1.0058,
      "step": 5550
    },
    {
      "epoch": 0.33601257025442677,
      "grad_norm": 0.7685634905463093,
      "learning_rate": 9.895320233501027e-06,
      "loss": 0.9829,
      "step": 5560
    },
    {
      "epoch": 0.3366169094095606,
      "grad_norm": 0.7932078532361053,
      "learning_rate": 9.89464112994633e-06,
      "loss": 1.0134,
      "step": 5570
    },
    {
      "epoch": 0.3372212485646945,
      "grad_norm": 0.7729376796762994,
      "learning_rate": 9.893959854141522e-06,
      "loss": 0.9869,
      "step": 5580
    },
    {
      "epoch": 0.33782558771982835,
      "grad_norm": 0.8219890383262062,
      "learning_rate": 9.893276406388952e-06,
      "loss": 1.0172,
      "step": 5590
    },
    {
      "epoch": 0.3384299268749622,
      "grad_norm": 0.8616634454094614,
      "learning_rate": 9.892590786991938e-06,
      "loss": 1.041,
      "step": 5600
    },
    {
      "epoch": 0.33903426603009607,
      "grad_norm": 0.8034396383510313,
      "learning_rate": 9.891902996254755e-06,
      "loss": 0.9573,
      "step": 5610
    },
    {
      "epoch": 0.3396386051852299,
      "grad_norm": 0.939431525813389,
      "learning_rate": 9.89121303448265e-06,
      "loss": 0.9926,
      "step": 5620
    },
    {
      "epoch": 0.34024294434036384,
      "grad_norm": 0.8071969049179112,
      "learning_rate": 9.890520901981825e-06,
      "loss": 1.014,
      "step": 5630
    },
    {
      "epoch": 0.3408472834954977,
      "grad_norm": 0.754025179432674,
      "learning_rate": 9.889826599059453e-06,
      "loss": 1.0067,
      "step": 5640
    },
    {
      "epoch": 0.34145162265063156,
      "grad_norm": 0.5665333715179984,
      "learning_rate": 9.889130126023665e-06,
      "loss": 1.0163,
      "step": 5650
    },
    {
      "epoch": 0.3420559618057654,
      "grad_norm": 0.5668675755503378,
      "learning_rate": 9.888431483183558e-06,
      "loss": 1.0242,
      "step": 5660
    },
    {
      "epoch": 0.3426603009608993,
      "grad_norm": 0.5811077849837687,
      "learning_rate": 9.887730670849192e-06,
      "loss": 0.9957,
      "step": 5670
    },
    {
      "epoch": 0.34326464011603314,
      "grad_norm": 0.54016526581125,
      "learning_rate": 9.887027689331585e-06,
      "loss": 0.9898,
      "step": 5680
    },
    {
      "epoch": 0.343868979271167,
      "grad_norm": 0.5703311662481036,
      "learning_rate": 9.886322538942723e-06,
      "loss": 1.019,
      "step": 5690
    },
    {
      "epoch": 0.34447331842630086,
      "grad_norm": 0.598548325840395,
      "learning_rate": 9.885615219995554e-06,
      "loss": 1.0091,
      "step": 5700
    },
    {
      "epoch": 0.3450776575814347,
      "grad_norm": 0.5577122384284335,
      "learning_rate": 9.88490573280399e-06,
      "loss": 0.9998,
      "step": 5710
    },
    {
      "epoch": 0.3456819967365686,
      "grad_norm": 0.6177372564163169,
      "learning_rate": 9.884194077682895e-06,
      "loss": 1.0114,
      "step": 5720
    },
    {
      "epoch": 0.34628633589170243,
      "grad_norm": 0.5973741439521935,
      "learning_rate": 9.88348025494811e-06,
      "loss": 1.0064,
      "step": 5730
    },
    {
      "epoch": 0.3468906750468363,
      "grad_norm": 0.5968116464639346,
      "learning_rate": 9.882764264916427e-06,
      "loss": 1.0043,
      "step": 5740
    },
    {
      "epoch": 0.34749501420197015,
      "grad_norm": 0.9528692893877637,
      "learning_rate": 9.882046107905604e-06,
      "loss": 1.0059,
      "step": 5750
    },
    {
      "epoch": 0.348099353357104,
      "grad_norm": 0.8641691818073385,
      "learning_rate": 9.881325784234361e-06,
      "loss": 1.0101,
      "step": 5760
    },
    {
      "epoch": 0.3487036925122379,
      "grad_norm": 0.852262168058538,
      "learning_rate": 9.88060329422238e-06,
      "loss": 0.9808,
      "step": 5770
    },
    {
      "epoch": 0.34930803166737173,
      "grad_norm": 0.8819272312678405,
      "learning_rate": 9.879878638190301e-06,
      "loss": 0.9889,
      "step": 5780
    },
    {
      "epoch": 0.3499123708225056,
      "grad_norm": 0.8146998671695789,
      "learning_rate": 9.87915181645973e-06,
      "loss": 1.0029,
      "step": 5790
    },
    {
      "epoch": 0.35051670997763945,
      "grad_norm": 0.5157107777855294,
      "learning_rate": 9.878422829353229e-06,
      "loss": 0.9918,
      "step": 5800
    },
    {
      "epoch": 0.3511210491327733,
      "grad_norm": 0.5327743713540637,
      "learning_rate": 9.877691677194326e-06,
      "loss": 0.978,
      "step": 5810
    },
    {
      "epoch": 0.35172538828790717,
      "grad_norm": 0.5483135621755293,
      "learning_rate": 9.876958360307506e-06,
      "loss": 0.967,
      "step": 5820
    },
    {
      "epoch": 0.35232972744304103,
      "grad_norm": 0.5200417729753583,
      "learning_rate": 9.876222879018218e-06,
      "loss": 0.9996,
      "step": 5830
    },
    {
      "epoch": 0.3529340665981749,
      "grad_norm": 0.5710913007032501,
      "learning_rate": 9.875485233652867e-06,
      "loss": 0.9979,
      "step": 5840
    },
    {
      "epoch": 0.35353840575330875,
      "grad_norm": 0.5065416779787284,
      "learning_rate": 9.874745424538826e-06,
      "loss": 0.9951,
      "step": 5850
    },
    {
      "epoch": 0.3541427449084426,
      "grad_norm": 0.5519496302011243,
      "learning_rate": 9.874003452004416e-06,
      "loss": 1.0332,
      "step": 5860
    },
    {
      "epoch": 0.35474708406357647,
      "grad_norm": 0.5065670748673957,
      "learning_rate": 9.873259316378934e-06,
      "loss": 1.0262,
      "step": 5870
    },
    {
      "epoch": 0.3553514232187103,
      "grad_norm": 0.5191163057311311,
      "learning_rate": 9.872513017992624e-06,
      "loss": 1.0082,
      "step": 5880
    },
    {
      "epoch": 0.3559557623738442,
      "grad_norm": 0.49451164618513554,
      "learning_rate": 9.871764557176698e-06,
      "loss": 0.9948,
      "step": 5890
    },
    {
      "epoch": 0.35656010152897805,
      "grad_norm": 0.49279905161619625,
      "learning_rate": 9.871013934263319e-06,
      "loss": 1.0038,
      "step": 5900
    },
    {
      "epoch": 0.3571644406841119,
      "grad_norm": 0.5371868002364787,
      "learning_rate": 9.870261149585616e-06,
      "loss": 1.0145,
      "step": 5910
    },
    {
      "epoch": 0.35776877983924577,
      "grad_norm": 0.5282710851908254,
      "learning_rate": 9.86950620347768e-06,
      "loss": 1.0055,
      "step": 5920
    },
    {
      "epoch": 0.3583731189943796,
      "grad_norm": 0.5379944232999399,
      "learning_rate": 9.868749096274554e-06,
      "loss": 0.9982,
      "step": 5930
    },
    {
      "epoch": 0.3589774581495135,
      "grad_norm": 0.49212860953679277,
      "learning_rate": 9.867989828312246e-06,
      "loss": 0.9876,
      "step": 5940
    },
    {
      "epoch": 0.35958179730464734,
      "grad_norm": 0.5082399618458712,
      "learning_rate": 9.867228399927717e-06,
      "loss": 1.009,
      "step": 5950
    },
    {
      "epoch": 0.3601861364597812,
      "grad_norm": 0.5328510728537502,
      "learning_rate": 9.866464811458893e-06,
      "loss": 0.9996,
      "step": 5960
    },
    {
      "epoch": 0.36079047561491506,
      "grad_norm": 0.5256073837698226,
      "learning_rate": 9.865699063244655e-06,
      "loss": 1.0132,
      "step": 5970
    },
    {
      "epoch": 0.361394814770049,
      "grad_norm": 0.5242644353528377,
      "learning_rate": 9.864931155624842e-06,
      "loss": 0.9831,
      "step": 5980
    },
    {
      "epoch": 0.36199915392518284,
      "grad_norm": 0.516099706093187,
      "learning_rate": 9.864161088940256e-06,
      "loss": 1.0086,
      "step": 5990
    },
    {
      "epoch": 0.3626034930803167,
      "grad_norm": 0.5776429540423382,
      "learning_rate": 9.863388863532651e-06,
      "loss": 1.0152,
      "step": 6000
    },
    {
      "epoch": 0.36320783223545056,
      "grad_norm": 0.5574160162130866,
      "learning_rate": 9.862614479744743e-06,
      "loss": 1.033,
      "step": 6010
    },
    {
      "epoch": 0.3638121713905844,
      "grad_norm": 0.5841822336746999,
      "learning_rate": 9.861837937920203e-06,
      "loss": 1.0281,
      "step": 6020
    },
    {
      "epoch": 0.3644165105457183,
      "grad_norm": 0.5448786121532411,
      "learning_rate": 9.861059238403665e-06,
      "loss": 1.0115,
      "step": 6030
    },
    {
      "epoch": 0.36502084970085213,
      "grad_norm": 0.5415840371777243,
      "learning_rate": 9.860278381540715e-06,
      "loss": 1.0039,
      "step": 6040
    },
    {
      "epoch": 0.365625188855986,
      "grad_norm": 0.563140715788272,
      "learning_rate": 9.8594953676779e-06,
      "loss": 0.9923,
      "step": 6050
    },
    {
      "epoch": 0.36622952801111985,
      "grad_norm": 0.5569321653314635,
      "learning_rate": 9.858710197162722e-06,
      "loss": 0.9894,
      "step": 6060
    },
    {
      "epoch": 0.3668338671662537,
      "grad_norm": 0.5277713151700876,
      "learning_rate": 9.85792287034364e-06,
      "loss": 1.0008,
      "step": 6070
    },
    {
      "epoch": 0.3674382063213876,
      "grad_norm": 0.48969583395453264,
      "learning_rate": 9.857133387570072e-06,
      "loss": 0.9929,
      "step": 6080
    },
    {
      "epoch": 0.36804254547652143,
      "grad_norm": 0.5288958874123402,
      "learning_rate": 9.856341749192393e-06,
      "loss": 1.0121,
      "step": 6090
    },
    {
      "epoch": 0.3686468846316553,
      "grad_norm": 0.5885370907535019,
      "learning_rate": 9.855547955561931e-06,
      "loss": 1.0031,
      "step": 6100
    },
    {
      "epoch": 0.36925122378678915,
      "grad_norm": 0.6130785511916662,
      "learning_rate": 9.854752007030976e-06,
      "loss": 1.0067,
      "step": 6110
    },
    {
      "epoch": 0.369855562941923,
      "grad_norm": 0.6239670734268047,
      "learning_rate": 9.853953903952767e-06,
      "loss": 1.0097,
      "step": 6120
    },
    {
      "epoch": 0.37045990209705687,
      "grad_norm": 0.6446600588460474,
      "learning_rate": 9.853153646681509e-06,
      "loss": 1.0049,
      "step": 6130
    },
    {
      "epoch": 0.37106424125219073,
      "grad_norm": 0.6324095438719612,
      "learning_rate": 9.852351235572353e-06,
      "loss": 0.9902,
      "step": 6140
    },
    {
      "epoch": 0.3716685804073246,
      "grad_norm": 0.786949200074966,
      "learning_rate": 9.851546670981411e-06,
      "loss": 0.9938,
      "step": 6150
    },
    {
      "epoch": 0.37227291956245845,
      "grad_norm": 0.7316542189995123,
      "learning_rate": 9.850739953265753e-06,
      "loss": 0.9957,
      "step": 6160
    },
    {
      "epoch": 0.3728772587175923,
      "grad_norm": 0.6602545798559167,
      "learning_rate": 9.8499310827834e-06,
      "loss": 0.9915,
      "step": 6170
    },
    {
      "epoch": 0.37348159787272617,
      "grad_norm": 0.6531405610720743,
      "learning_rate": 9.849120059893329e-06,
      "loss": 1.0072,
      "step": 6180
    },
    {
      "epoch": 0.37408593702786,
      "grad_norm": 0.6947368116689114,
      "learning_rate": 9.848306884955475e-06,
      "loss": 1.0362,
      "step": 6190
    },
    {
      "epoch": 0.3746902761829939,
      "grad_norm": 0.6383771212914652,
      "learning_rate": 9.847491558330726e-06,
      "loss": 0.9873,
      "step": 6200
    },
    {
      "epoch": 0.37529461533812775,
      "grad_norm": 0.6405579644703344,
      "learning_rate": 9.846674080380925e-06,
      "loss": 1.0091,
      "step": 6210
    },
    {
      "epoch": 0.3758989544932616,
      "grad_norm": 0.6318280139648661,
      "learning_rate": 9.845854451468868e-06,
      "loss": 0.985,
      "step": 6220
    },
    {
      "epoch": 0.37650329364839547,
      "grad_norm": 0.6339638225372017,
      "learning_rate": 9.845032671958312e-06,
      "loss": 1.0154,
      "step": 6230
    },
    {
      "epoch": 0.3771076328035293,
      "grad_norm": 0.6619136842919839,
      "learning_rate": 9.844208742213963e-06,
      "loss": 1.0029,
      "step": 6240
    },
    {
      "epoch": 0.3777119719586632,
      "grad_norm": 0.7928113525284128,
      "learning_rate": 9.843382662601481e-06,
      "loss": 1.011,
      "step": 6250
    },
    {
      "epoch": 0.37831631111379704,
      "grad_norm": 0.834095380925119,
      "learning_rate": 9.842554433487482e-06,
      "loss": 1.0143,
      "step": 6260
    },
    {
      "epoch": 0.3789206502689309,
      "grad_norm": 0.757658715574136,
      "learning_rate": 9.841724055239535e-06,
      "loss": 1.0268,
      "step": 6270
    },
    {
      "epoch": 0.37952498942406476,
      "grad_norm": 0.7718469183204548,
      "learning_rate": 9.840891528226164e-06,
      "loss": 1.0201,
      "step": 6280
    },
    {
      "epoch": 0.3801293285791986,
      "grad_norm": 0.7627786860683481,
      "learning_rate": 9.840056852816846e-06,
      "loss": 1.034,
      "step": 6290
    },
    {
      "epoch": 0.3807336677343325,
      "grad_norm": 0.7943004246989257,
      "learning_rate": 9.839220029382011e-06,
      "loss": 0.988,
      "step": 6300
    },
    {
      "epoch": 0.38133800688946634,
      "grad_norm": 0.7752178630352604,
      "learning_rate": 9.838381058293043e-06,
      "loss": 0.9986,
      "step": 6310
    },
    {
      "epoch": 0.38194234604460026,
      "grad_norm": 0.7853777183156614,
      "learning_rate": 9.837539939922279e-06,
      "loss": 1.0195,
      "step": 6320
    },
    {
      "epoch": 0.3825466851997341,
      "grad_norm": 0.7621120294237848,
      "learning_rate": 9.836696674643007e-06,
      "loss": 1.006,
      "step": 6330
    },
    {
      "epoch": 0.383151024354868,
      "grad_norm": 0.7774745458608108,
      "learning_rate": 9.835851262829473e-06,
      "loss": 1.0281,
      "step": 6340
    },
    {
      "epoch": 0.38375536351000183,
      "grad_norm": 1.215581564301588,
      "learning_rate": 9.83500370485687e-06,
      "loss": 1.0152,
      "step": 6350
    },
    {
      "epoch": 0.3843597026651357,
      "grad_norm": 1.1912153483343944,
      "learning_rate": 9.834154001101343e-06,
      "loss": 1.0194,
      "step": 6360
    },
    {
      "epoch": 0.38496404182026955,
      "grad_norm": 1.1849026775130922,
      "learning_rate": 9.83330215194e-06,
      "loss": 0.9818,
      "step": 6370
    },
    {
      "epoch": 0.3855683809754034,
      "grad_norm": 1.1765707938172545,
      "learning_rate": 9.832448157750883e-06,
      "loss": 1.0108,
      "step": 6380
    },
    {
      "epoch": 0.3861727201305373,
      "grad_norm": 1.206340887436501,
      "learning_rate": 9.831592018913003e-06,
      "loss": 0.9717,
      "step": 6390
    },
    {
      "epoch": 0.38677705928567113,
      "grad_norm": 0.8389696997546268,
      "learning_rate": 9.830733735806314e-06,
      "loss": 1.0048,
      "step": 6400
    },
    {
      "epoch": 0.387381398440805,
      "grad_norm": 1.081752292506,
      "learning_rate": 9.829873308811724e-06,
      "loss": 1.0191,
      "step": 6410
    },
    {
      "epoch": 0.38798573759593885,
      "grad_norm": 1.2848922573959891,
      "learning_rate": 9.82901073831109e-06,
      "loss": 1.0089,
      "step": 6420
    },
    {
      "epoch": 0.3885900767510727,
      "grad_norm": 1.0491364864125905,
      "learning_rate": 9.828146024687227e-06,
      "loss": 1.0092,
      "step": 6430
    },
    {
      "epoch": 0.38919441590620657,
      "grad_norm": 1.1203027685668425,
      "learning_rate": 9.82727916832389e-06,
      "loss": 1.0081,
      "step": 6440
    },
    {
      "epoch": 0.38979875506134043,
      "grad_norm": 2.949380975637301,
      "learning_rate": 9.826410169605795e-06,
      "loss": 0.982,
      "step": 6450
    },
    {
      "epoch": 0.3904030942164743,
      "grad_norm": 2.5655070580122725,
      "learning_rate": 9.825539028918605e-06,
      "loss": 0.9933,
      "step": 6460
    },
    {
      "epoch": 0.39100743337160815,
      "grad_norm": 2.58481575091607,
      "learning_rate": 9.824665746648933e-06,
      "loss": 1.0009,
      "step": 6470
    },
    {
      "epoch": 0.391611772526742,
      "grad_norm": 2.6736885574250215,
      "learning_rate": 9.823790323184344e-06,
      "loss": 1.0031,
      "step": 6480
    },
    {
      "epoch": 0.39221611168187587,
      "grad_norm": 2.5851065212745823,
      "learning_rate": 9.822912758913352e-06,
      "loss": 1.005,
      "step": 6490
    },
    {
      "epoch": 0.3928204508370097,
      "grad_norm": 1.0358027271238648,
      "learning_rate": 9.82203305422542e-06,
      "loss": 1.0108,
      "step": 6500
    },
    {
      "epoch": 0.3934247899921436,
      "grad_norm": 1.0415077326769058,
      "learning_rate": 9.821151209510965e-06,
      "loss": 1.0075,
      "step": 6510
    },
    {
      "epoch": 0.39402912914727745,
      "grad_norm": 0.8984057919416041,
      "learning_rate": 9.82026722516135e-06,
      "loss": 1.0109,
      "step": 6520
    },
    {
      "epoch": 0.3946334683024113,
      "grad_norm": 1.014800362641625,
      "learning_rate": 9.819381101568887e-06,
      "loss": 0.9858,
      "step": 6530
    },
    {
      "epoch": 0.39523780745754516,
      "grad_norm": 0.9751634770274491,
      "learning_rate": 9.818492839126844e-06,
      "loss": 1.0172,
      "step": 6540
    },
    {
      "epoch": 0.395842146612679,
      "grad_norm": 1.0515289372041692,
      "learning_rate": 9.817602438229429e-06,
      "loss": 1.0039,
      "step": 6550
    },
    {
      "epoch": 0.3964464857678129,
      "grad_norm": 1.167983434940786,
      "learning_rate": 9.816709899271805e-06,
      "loss": 1.0189,
      "step": 6560
    },
    {
      "epoch": 0.39705082492294674,
      "grad_norm": 1.1893825943310294,
      "learning_rate": 9.815815222650082e-06,
      "loss": 1.014,
      "step": 6570
    },
    {
      "epoch": 0.3976551640780806,
      "grad_norm": 1.0242832548952334,
      "learning_rate": 9.814918408761318e-06,
      "loss": 0.9807,
      "step": 6580
    },
    {
      "epoch": 0.39825950323321446,
      "grad_norm": 1.0479578604354343,
      "learning_rate": 9.814019458003523e-06,
      "loss": 0.9694,
      "step": 6590
    },
    {
      "epoch": 0.3988638423883483,
      "grad_norm": 1.2749111022574557,
      "learning_rate": 9.813118370775652e-06,
      "loss": 1.0019,
      "step": 6600
    },
    {
      "epoch": 0.3994681815434822,
      "grad_norm": 1.073361321222981,
      "learning_rate": 9.812215147477608e-06,
      "loss": 0.9982,
      "step": 6610
    },
    {
      "epoch": 0.40007252069861604,
      "grad_norm": 1.047818464589798,
      "learning_rate": 9.811309788510243e-06,
      "loss": 0.9972,
      "step": 6620
    },
    {
      "epoch": 0.4006768598537499,
      "grad_norm": 0.9747534725894509,
      "learning_rate": 9.810402294275359e-06,
      "loss": 1.0049,
      "step": 6630
    },
    {
      "epoch": 0.40128119900888376,
      "grad_norm": 1.050499526975608,
      "learning_rate": 9.8094926651757e-06,
      "loss": 0.9962,
      "step": 6640
    },
    {
      "epoch": 0.4018855381640176,
      "grad_norm": 1.321414260491159,
      "learning_rate": 9.808580901614964e-06,
      "loss": 1.0375,
      "step": 6650
    },
    {
      "epoch": 0.40248987731915153,
      "grad_norm": 1.2177986013454383,
      "learning_rate": 9.807667003997791e-06,
      "loss": 0.9848,
      "step": 6660
    },
    {
      "epoch": 0.4030942164742854,
      "grad_norm": 1.035795192043516,
      "learning_rate": 9.806750972729774e-06,
      "loss": 1.0064,
      "step": 6670
    },
    {
      "epoch": 0.40369855562941925,
      "grad_norm": 1.2967787417060446,
      "learning_rate": 9.805832808217447e-06,
      "loss": 1.0159,
      "step": 6680
    },
    {
      "epoch": 0.4043028947845531,
      "grad_norm": 1.0389867219124678,
      "learning_rate": 9.804912510868289e-06,
      "loss": 0.9841,
      "step": 6690
    },
    {
      "epoch": 0.404907233939687,
      "grad_norm": 1.3409165580157771,
      "learning_rate": 9.803990081090737e-06,
      "loss": 1.0186,
      "step": 6700
    },
    {
      "epoch": 0.40551157309482083,
      "grad_norm": 1.3105392118251176,
      "learning_rate": 9.803065519294162e-06,
      "loss": 1.0251,
      "step": 6710
    },
    {
      "epoch": 0.4061159122499547,
      "grad_norm": 1.3179278079501042,
      "learning_rate": 9.802138825888887e-06,
      "loss": 1.0091,
      "step": 6720
    },
    {
      "epoch": 0.40672025140508855,
      "grad_norm": 1.1782595293863323,
      "learning_rate": 9.801210001286182e-06,
      "loss": 0.9931,
      "step": 6730
    },
    {
      "epoch": 0.4073245905602224,
      "grad_norm": 1.155908582055039,
      "learning_rate": 9.800279045898255e-06,
      "loss": 1.0306,
      "step": 6740
    },
    {
      "epoch": 0.40792892971535627,
      "grad_norm": 1.0862095414943798,
      "learning_rate": 9.799345960138274e-06,
      "loss": 0.9881,
      "step": 6750
    },
    {
      "epoch": 0.40853326887049013,
      "grad_norm": 1.040057157128269,
      "learning_rate": 9.798410744420338e-06,
      "loss": 1.014,
      "step": 6760
    },
    {
      "epoch": 0.409137608025624,
      "grad_norm": 1.0373449963795083,
      "learning_rate": 9.797473399159501e-06,
      "loss": 0.9889,
      "step": 6770
    },
    {
      "epoch": 0.40974194718075785,
      "grad_norm": 0.9810127621452572,
      "learning_rate": 9.796533924771758e-06,
      "loss": 1.0151,
      "step": 6780
    },
    {
      "epoch": 0.4103462863358917,
      "grad_norm": 0.9999966957880234,
      "learning_rate": 9.795592321674046e-06,
      "loss": 0.9847,
      "step": 6790
    },
    {
      "epoch": 0.41095062549102557,
      "grad_norm": 1.21250793554175,
      "learning_rate": 9.794648590284253e-06,
      "loss": 1.0017,
      "step": 6800
    },
    {
      "epoch": 0.4115549646461594,
      "grad_norm": 1.2003842035689833,
      "learning_rate": 9.793702731021207e-06,
      "loss": 1.0143,
      "step": 6810
    },
    {
      "epoch": 0.4121593038012933,
      "grad_norm": 1.018284273328576,
      "learning_rate": 9.792754744304683e-06,
      "loss": 1.004,
      "step": 6820
    },
    {
      "epoch": 0.41276364295642715,
      "grad_norm": 1.0073312646222339,
      "learning_rate": 9.7918046305554e-06,
      "loss": 1.0106,
      "step": 6830
    },
    {
      "epoch": 0.413367982111561,
      "grad_norm": 1.0923757142684147,
      "learning_rate": 9.790852390195017e-06,
      "loss": 1.022,
      "step": 6840
    },
    {
      "epoch": 0.41397232126669486,
      "grad_norm": 1.0304250789790763,
      "learning_rate": 9.789898023646143e-06,
      "loss": 0.9912,
      "step": 6850
    },
    {
      "epoch": 0.4145766604218287,
      "grad_norm": 1.0985006592797248,
      "learning_rate": 9.788941531332327e-06,
      "loss": 0.9899,
      "step": 6860
    },
    {
      "epoch": 0.4151809995769626,
      "grad_norm": 1.0629701582316857,
      "learning_rate": 9.78798291367806e-06,
      "loss": 0.9907,
      "step": 6870
    },
    {
      "epoch": 0.41578533873209644,
      "grad_norm": 1.013124840100367,
      "learning_rate": 9.787022171108782e-06,
      "loss": 1.0111,
      "step": 6880
    },
    {
      "epoch": 0.4163896778872303,
      "grad_norm": 1.1715755837533566,
      "learning_rate": 9.786059304050866e-06,
      "loss": 0.9857,
      "step": 6890
    },
    {
      "epoch": 0.41699401704236416,
      "grad_norm": 1.2565863109710398,
      "learning_rate": 9.78509431293164e-06,
      "loss": 0.9911,
      "step": 6900
    },
    {
      "epoch": 0.417598356197498,
      "grad_norm": 1.3113844912921317,
      "learning_rate": 9.784127198179368e-06,
      "loss": 0.99,
      "step": 6910
    },
    {
      "epoch": 0.4182026953526319,
      "grad_norm": 1.1705994053090263,
      "learning_rate": 9.783157960223253e-06,
      "loss": 1.0015,
      "step": 6920
    },
    {
      "epoch": 0.41880703450776574,
      "grad_norm": 1.165752825855671,
      "learning_rate": 9.782186599493448e-06,
      "loss": 0.9877,
      "step": 6930
    },
    {
      "epoch": 0.4194113736628996,
      "grad_norm": 1.2102328573192687,
      "learning_rate": 9.781213116421044e-06,
      "loss": 0.9932,
      "step": 6940
    },
    {
      "epoch": 0.42001571281803346,
      "grad_norm": 1.080627758934474,
      "learning_rate": 9.780237511438073e-06,
      "loss": 0.9994,
      "step": 6950
    },
    {
      "epoch": 0.4206200519731673,
      "grad_norm": 1.1176859646274222,
      "learning_rate": 9.779259784977514e-06,
      "loss": 1.0068,
      "step": 6960
    },
    {
      "epoch": 0.4212243911283012,
      "grad_norm": 1.1908060720706377,
      "learning_rate": 9.77827993747328e-06,
      "loss": 1.0054,
      "step": 6970
    },
    {
      "epoch": 0.42182873028343504,
      "grad_norm": 1.1166902998641413,
      "learning_rate": 9.777297969360226e-06,
      "loss": 0.9921,
      "step": 6980
    },
    {
      "epoch": 0.4224330694385689,
      "grad_norm": 1.1042008072219143,
      "learning_rate": 9.77631388107416e-06,
      "loss": 1.0087,
      "step": 6990
    },
    {
      "epoch": 0.4230374085937028,
      "grad_norm": 1.1703704357999294,
      "learning_rate": 9.775327673051814e-06,
      "loss": 0.9934,
      "step": 7000
    },
    {
      "epoch": 0.4236417477488367,
      "grad_norm": 1.3204898338176354,
      "learning_rate": 9.774339345730874e-06,
      "loss": 1.0288,
      "step": 7010
    },
    {
      "epoch": 0.42424608690397053,
      "grad_norm": 1.2478257239089718,
      "learning_rate": 9.773348899549959e-06,
      "loss": 1.0127,
      "step": 7020
    },
    {
      "epoch": 0.4248504260591044,
      "grad_norm": 1.1995988774845179,
      "learning_rate": 9.77235633494863e-06,
      "loss": 1.0161,
      "step": 7030
    },
    {
      "epoch": 0.42545476521423825,
      "grad_norm": 1.190152001486567,
      "learning_rate": 9.77136165236739e-06,
      "loss": 0.9929,
      "step": 7040
    },
    {
      "epoch": 0.4260591043693721,
      "grad_norm": 1.468029887672351,
      "learning_rate": 9.770364852247682e-06,
      "loss": 0.9863,
      "step": 7050
    },
    {
      "epoch": 0.42666344352450597,
      "grad_norm": 1.348062017936987,
      "learning_rate": 9.769365935031885e-06,
      "loss": 1.0051,
      "step": 7060
    },
    {
      "epoch": 0.42726778267963983,
      "grad_norm": 1.2737446120875788,
      "learning_rate": 9.768364901163322e-06,
      "loss": 1.0074,
      "step": 7070
    },
    {
      "epoch": 0.4278721218347737,
      "grad_norm": 1.396269211915545,
      "learning_rate": 9.767361751086255e-06,
      "loss": 1.0179,
      "step": 7080
    },
    {
      "epoch": 0.42847646098990755,
      "grad_norm": 1.4199247246451963,
      "learning_rate": 9.76635648524588e-06,
      "loss": 1.0162,
      "step": 7090
    },
    {
      "epoch": 0.4290808001450414,
      "grad_norm": 0.8975674762644424,
      "learning_rate": 9.765349104088337e-06,
      "loss": 1.0163,
      "step": 7100
    },
    {
      "epoch": 0.42968513930017527,
      "grad_norm": 0.821557882217157,
      "learning_rate": 9.764339608060705e-06,
      "loss": 1.012,
      "step": 7110
    },
    {
      "epoch": 0.4302894784553091,
      "grad_norm": 0.8999199060882129,
      "learning_rate": 9.763327997610998e-06,
      "loss": 0.9897,
      "step": 7120
    },
    {
      "epoch": 0.430893817610443,
      "grad_norm": 0.8938660044016248,
      "learning_rate": 9.762314273188171e-06,
      "loss": 1.0005,
      "step": 7130
    },
    {
      "epoch": 0.43149815676557685,
      "grad_norm": 0.8682660718178058,
      "learning_rate": 9.761298435242119e-06,
      "loss": 1.0099,
      "step": 7140
    },
    {
      "epoch": 0.4321024959207107,
      "grad_norm": 0.7362474025654122,
      "learning_rate": 9.76028048422367e-06,
      "loss": 0.9971,
      "step": 7150
    },
    {
      "epoch": 0.43270683507584456,
      "grad_norm": 0.7691450101351018,
      "learning_rate": 9.759260420584592e-06,
      "loss": 1.0095,
      "step": 7160
    },
    {
      "epoch": 0.4333111742309784,
      "grad_norm": 0.73552539237862,
      "learning_rate": 9.758238244777592e-06,
      "loss": 1.0283,
      "step": 7170
    },
    {
      "epoch": 0.4339155133861123,
      "grad_norm": 0.7946866692838345,
      "learning_rate": 9.757213957256316e-06,
      "loss": 1.0123,
      "step": 7180
    },
    {
      "epoch": 0.43451985254124614,
      "grad_norm": 0.8231548083372715,
      "learning_rate": 9.75618755847534e-06,
      "loss": 0.9983,
      "step": 7190
    },
    {
      "epoch": 0.43512419169638,
      "grad_norm": 0.818935982427908,
      "learning_rate": 9.755159048890183e-06,
      "loss": 1.0097,
      "step": 7200
    },
    {
      "epoch": 0.43572853085151386,
      "grad_norm": 0.8766921812361363,
      "learning_rate": 9.754128428957302e-06,
      "loss": 0.9929,
      "step": 7210
    },
    {
      "epoch": 0.4363328700066477,
      "grad_norm": 0.8276226362836122,
      "learning_rate": 9.753095699134084e-06,
      "loss": 0.9913,
      "step": 7220
    },
    {
      "epoch": 0.4369372091617816,
      "grad_norm": 0.840050349505506,
      "learning_rate": 9.752060859878859e-06,
      "loss": 0.9804,
      "step": 7230
    },
    {
      "epoch": 0.43754154831691544,
      "grad_norm": 0.8361762627878246,
      "learning_rate": 9.75102391165089e-06,
      "loss": 0.9882,
      "step": 7240
    },
    {
      "epoch": 0.4381458874720493,
      "grad_norm": 0.5391825704457952,
      "learning_rate": 9.749984854910375e-06,
      "loss": 0.988,
      "step": 7250
    },
    {
      "epoch": 0.43875022662718316,
      "grad_norm": 0.5663380880574846,
      "learning_rate": 9.74894369011845e-06,
      "loss": 1.0155,
      "step": 7260
    },
    {
      "epoch": 0.439354565782317,
      "grad_norm": 0.5136098987146434,
      "learning_rate": 9.747900417737187e-06,
      "loss": 0.9892,
      "step": 7270
    },
    {
      "epoch": 0.4399589049374509,
      "grad_norm": 0.5745091597084321,
      "learning_rate": 9.746855038229589e-06,
      "loss": 1.0059,
      "step": 7280
    },
    {
      "epoch": 0.44056324409258474,
      "grad_norm": 0.543223375127291,
      "learning_rate": 9.7458075520596e-06,
      "loss": 0.9785,
      "step": 7290
    },
    {
      "epoch": 0.4411675832477186,
      "grad_norm": 0.6081263345427493,
      "learning_rate": 9.744757959692094e-06,
      "loss": 1.0006,
      "step": 7300
    },
    {
      "epoch": 0.44177192240285246,
      "grad_norm": 0.6475485365675823,
      "learning_rate": 9.743706261592884e-06,
      "loss": 1.0014,
      "step": 7310
    },
    {
      "epoch": 0.4423762615579863,
      "grad_norm": 0.574667862452884,
      "learning_rate": 9.742652458228715e-06,
      "loss": 1.0145,
      "step": 7320
    },
    {
      "epoch": 0.4429806007131202,
      "grad_norm": 0.5817633067184044,
      "learning_rate": 9.741596550067268e-06,
      "loss": 1.0019,
      "step": 7330
    },
    {
      "epoch": 0.4435849398682541,
      "grad_norm": 0.5737610251519444,
      "learning_rate": 9.740538537577151e-06,
      "loss": 1.017,
      "step": 7340
    },
    {
      "epoch": 0.44418927902338795,
      "grad_norm": 0.8379013520479386,
      "learning_rate": 9.739478421227918e-06,
      "loss": 1.0089,
      "step": 7350
    },
    {
      "epoch": 0.4447936181785218,
      "grad_norm": 0.8653984258209246,
      "learning_rate": 9.738416201490048e-06,
      "loss": 1.0115,
      "step": 7360
    },
    {
      "epoch": 0.44539795733365567,
      "grad_norm": 0.8585524627727098,
      "learning_rate": 9.737351878834957e-06,
      "loss": 1.0033,
      "step": 7370
    },
    {
      "epoch": 0.44600229648878953,
      "grad_norm": 0.8635926564121462,
      "learning_rate": 9.736285453734992e-06,
      "loss": 1.0095,
      "step": 7380
    },
    {
      "epoch": 0.4466066356439234,
      "grad_norm": 0.9364062955477616,
      "learning_rate": 9.735216926663433e-06,
      "loss": 1.0285,
      "step": 7390
    },
    {
      "epoch": 0.44721097479905725,
      "grad_norm": 0.5132235945423931,
      "learning_rate": 9.734146298094497e-06,
      "loss": 0.9964,
      "step": 7400
    },
    {
      "epoch": 0.4478153139541911,
      "grad_norm": 0.5411800684996538,
      "learning_rate": 9.733073568503331e-06,
      "loss": 1.0235,
      "step": 7410
    },
    {
      "epoch": 0.44841965310932497,
      "grad_norm": 0.49242600048418367,
      "learning_rate": 9.731998738366014e-06,
      "loss": 1.0147,
      "step": 7420
    },
    {
      "epoch": 0.4490239922644588,
      "grad_norm": 0.5368617448294334,
      "learning_rate": 9.730921808159552e-06,
      "loss": 0.9992,
      "step": 7430
    },
    {
      "epoch": 0.4496283314195927,
      "grad_norm": 0.540233755894723,
      "learning_rate": 9.729842778361895e-06,
      "loss": 1.0016,
      "step": 7440
    },
    {
      "epoch": 0.45023267057472655,
      "grad_norm": 0.4973184814181504,
      "learning_rate": 9.728761649451917e-06,
      "loss": 0.9999,
      "step": 7450
    },
    {
      "epoch": 0.4508370097298604,
      "grad_norm": 0.5040057736208563,
      "learning_rate": 9.727678421909423e-06,
      "loss": 1.004,
      "step": 7460
    },
    {
      "epoch": 0.45144134888499426,
      "grad_norm": 0.5010346109754319,
      "learning_rate": 9.726593096215154e-06,
      "loss": 1.017,
      "step": 7470
    },
    {
      "epoch": 0.4520456880401281,
      "grad_norm": 0.5159877010609631,
      "learning_rate": 9.725505672850775e-06,
      "loss": 0.9955,
      "step": 7480
    },
    {
      "epoch": 0.452650027195262,
      "grad_norm": 0.5359361237044953,
      "learning_rate": 9.72441615229889e-06,
      "loss": 0.9808,
      "step": 7490
    },
    {
      "epoch": 0.45325436635039584,
      "grad_norm": 0.5420479823701472,
      "learning_rate": 9.723324535043028e-06,
      "loss": 0.9925,
      "step": 7500
    },
    {
      "epoch": 0.4538587055055297,
      "grad_norm": 0.5264705408238455,
      "learning_rate": 9.722230821567652e-06,
      "loss": 0.9956,
      "step": 7510
    },
    {
      "epoch": 0.45446304466066356,
      "grad_norm": 0.5461301540893525,
      "learning_rate": 9.721135012358156e-06,
      "loss": 1.0064,
      "step": 7520
    },
    {
      "epoch": 0.4550673838157974,
      "grad_norm": 0.5015682263575848,
      "learning_rate": 9.720037107900857e-06,
      "loss": 1.0002,
      "step": 7530
    },
    {
      "epoch": 0.4556717229709313,
      "grad_norm": 0.5463703322516429,
      "learning_rate": 9.718937108683012e-06,
      "loss": 1.0059,
      "step": 7540
    },
    {
      "epoch": 0.45627606212606514,
      "grad_norm": 0.5346647708927578,
      "learning_rate": 9.717835015192797e-06,
      "loss": 1.0136,
      "step": 7550
    },
    {
      "epoch": 0.456880401281199,
      "grad_norm": 0.5244071387215217,
      "learning_rate": 9.716730827919327e-06,
      "loss": 0.9906,
      "step": 7560
    },
    {
      "epoch": 0.45748474043633286,
      "grad_norm": 0.5090978820687052,
      "learning_rate": 9.715624547352641e-06,
      "loss": 1.0018,
      "step": 7570
    },
    {
      "epoch": 0.4580890795914667,
      "grad_norm": 0.5368095176163875,
      "learning_rate": 9.714516173983709e-06,
      "loss": 0.9922,
      "step": 7580
    },
    {
      "epoch": 0.4586934187466006,
      "grad_norm": 0.4860640203798647,
      "learning_rate": 9.713405708304427e-06,
      "loss": 0.9966,
      "step": 7590
    },
    {
      "epoch": 0.45929775790173444,
      "grad_norm": 0.45062056663754396,
      "learning_rate": 9.712293150807624e-06,
      "loss": 1.0077,
      "step": 7600
    },
    {
      "epoch": 0.4599020970568683,
      "grad_norm": 0.5264066780993698,
      "learning_rate": 9.711178501987054e-06,
      "loss": 0.9944,
      "step": 7610
    },
    {
      "epoch": 0.46050643621200216,
      "grad_norm": 0.49210314433869495,
      "learning_rate": 9.7100617623374e-06,
      "loss": 1.0085,
      "step": 7620
    },
    {
      "epoch": 0.461110775367136,
      "grad_norm": 0.5360137625880347,
      "learning_rate": 9.708942932354272e-06,
      "loss": 0.9829,
      "step": 7630
    },
    {
      "epoch": 0.4617151145222699,
      "grad_norm": 0.5252314800384593,
      "learning_rate": 9.70782201253421e-06,
      "loss": 0.9812,
      "step": 7640
    },
    {
      "epoch": 0.46231945367740374,
      "grad_norm": 0.5540169774022148,
      "learning_rate": 9.70669900337468e-06,
      "loss": 0.9827,
      "step": 7650
    },
    {
      "epoch": 0.4629237928325376,
      "grad_norm": 0.5071988278808582,
      "learning_rate": 9.705573905374075e-06,
      "loss": 0.9996,
      "step": 7660
    },
    {
      "epoch": 0.46352813198767145,
      "grad_norm": 0.5451423592038422,
      "learning_rate": 9.704446719031714e-06,
      "loss": 1.0267,
      "step": 7670
    },
    {
      "epoch": 0.46413247114280537,
      "grad_norm": 0.5267472331193773,
      "learning_rate": 9.703317444847849e-06,
      "loss": 0.9933,
      "step": 7680
    },
    {
      "epoch": 0.46473681029793923,
      "grad_norm": 0.5561525850421628,
      "learning_rate": 9.702186083323648e-06,
      "loss": 0.9939,
      "step": 7690
    },
    {
      "epoch": 0.4653411494530731,
      "grad_norm": 0.6529114062972717,
      "learning_rate": 9.701052634961213e-06,
      "loss": 0.9883,
      "step": 7700
    },
    {
      "epoch": 0.46594548860820695,
      "grad_norm": 0.5941168324045979,
      "learning_rate": 9.699917100263571e-06,
      "loss": 1.0023,
      "step": 7710
    },
    {
      "epoch": 0.4665498277633408,
      "grad_norm": 0.6355639771440227,
      "learning_rate": 9.698779479734677e-06,
      "loss": 1.0085,
      "step": 7720
    },
    {
      "epoch": 0.46715416691847467,
      "grad_norm": 0.6322387058727607,
      "learning_rate": 9.697639773879404e-06,
      "loss": 0.9798,
      "step": 7730
    },
    {
      "epoch": 0.4677585060736085,
      "grad_norm": 0.6335348266753075,
      "learning_rate": 9.69649798320356e-06,
      "loss": 0.998,
      "step": 7740
    },
    {
      "epoch": 0.4683628452287424,
      "grad_norm": 0.690941642264451,
      "learning_rate": 9.695354108213868e-06,
      "loss": 0.9968,
      "step": 7750
    },
    {
      "epoch": 0.46896718438387625,
      "grad_norm": 0.7029981868793153,
      "learning_rate": 9.694208149417985e-06,
      "loss": 0.9806,
      "step": 7760
    },
    {
      "epoch": 0.4695715235390101,
      "grad_norm": 0.6981601027311941,
      "learning_rate": 9.693060107324493e-06,
      "loss": 0.9959,
      "step": 7770
    },
    {
      "epoch": 0.47017586269414396,
      "grad_norm": 0.7274625431921845,
      "learning_rate": 9.69190998244289e-06,
      "loss": 1.0054,
      "step": 7780
    },
    {
      "epoch": 0.4707802018492778,
      "grad_norm": 0.6867733702462608,
      "learning_rate": 9.690757775283603e-06,
      "loss": 1.0368,
      "step": 7790
    },
    {
      "epoch": 0.4713845410044117,
      "grad_norm": 0.6043675336725368,
      "learning_rate": 9.689603486357986e-06,
      "loss": 1.0021,
      "step": 7800
    },
    {
      "epoch": 0.47198888015954554,
      "grad_norm": 0.6157464021134955,
      "learning_rate": 9.688447116178315e-06,
      "loss": 1.0249,
      "step": 7810
    },
    {
      "epoch": 0.4725932193146794,
      "grad_norm": 0.6427327251905133,
      "learning_rate": 9.687288665257786e-06,
      "loss": 0.9808,
      "step": 7820
    },
    {
      "epoch": 0.47319755846981326,
      "grad_norm": 0.6155938108333662,
      "learning_rate": 9.686128134110525e-06,
      "loss": 1.0035,
      "step": 7830
    },
    {
      "epoch": 0.4738018976249471,
      "grad_norm": 0.6131794954190143,
      "learning_rate": 9.684965523251575e-06,
      "loss": 1.0213,
      "step": 7840
    },
    {
      "epoch": 0.474406236780081,
      "grad_norm": 0.7747597542732607,
      "learning_rate": 9.683800833196907e-06,
      "loss": 0.9688,
      "step": 7850
    },
    {
      "epoch": 0.47501057593521484,
      "grad_norm": 0.7555906781074263,
      "learning_rate": 9.68263406446341e-06,
      "loss": 0.9866,
      "step": 7860
    },
    {
      "epoch": 0.4756149150903487,
      "grad_norm": 0.7527735572152627,
      "learning_rate": 9.6814652175689e-06,
      "loss": 0.9971,
      "step": 7870
    },
    {
      "epoch": 0.47621925424548256,
      "grad_norm": 0.7743973471003973,
      "learning_rate": 9.680294293032112e-06,
      "loss": 1.0072,
      "step": 7880
    },
    {
      "epoch": 0.4768235934006164,
      "grad_norm": 0.8108529448189443,
      "learning_rate": 9.679121291372704e-06,
      "loss": 0.9885,
      "step": 7890
    },
    {
      "epoch": 0.4774279325557503,
      "grad_norm": 0.7460995116908645,
      "learning_rate": 9.677946213111259e-06,
      "loss": 0.9697,
      "step": 7900
    },
    {
      "epoch": 0.47803227171088414,
      "grad_norm": 0.80551425380806,
      "learning_rate": 9.676769058769274e-06,
      "loss": 0.9863,
      "step": 7910
    },
    {
      "epoch": 0.478636610866018,
      "grad_norm": 0.7782410932889495,
      "learning_rate": 9.675589828869177e-06,
      "loss": 0.9947,
      "step": 7920
    },
    {
      "epoch": 0.47924095002115186,
      "grad_norm": 0.8106885579056082,
      "learning_rate": 9.674408523934308e-06,
      "loss": 1.014,
      "step": 7930
    },
    {
      "epoch": 0.4798452891762857,
      "grad_norm": 0.7429403181875742,
      "learning_rate": 9.673225144488934e-06,
      "loss": 0.9875,
      "step": 7940
    },
    {
      "epoch": 0.4804496283314196,
      "grad_norm": 1.2268192425228464,
      "learning_rate": 9.672039691058242e-06,
      "loss": 0.9937,
      "step": 7950
    },
    {
      "epoch": 0.48105396748655344,
      "grad_norm": 1.2759892859249429,
      "learning_rate": 9.670852164168339e-06,
      "loss": 1.0152,
      "step": 7960
    },
    {
      "epoch": 0.4816583066416873,
      "grad_norm": 1.1939470704091346,
      "learning_rate": 9.669662564346246e-06,
      "loss": 0.9663,
      "step": 7970
    },
    {
      "epoch": 0.48226264579682115,
      "grad_norm": 1.1525754224949845,
      "learning_rate": 9.668470892119915e-06,
      "loss": 1.0216,
      "step": 7980
    },
    {
      "epoch": 0.482866984951955,
      "grad_norm": 1.1764028603949268,
      "learning_rate": 9.66727714801821e-06,
      "loss": 1.0079,
      "step": 7990
    },
    {
      "epoch": 0.4834713241070889,
      "grad_norm": 1.1125911325303737,
      "learning_rate": 9.666081332570917e-06,
      "loss": 1.0183,
      "step": 8000
    },
    {
      "epoch": 0.48407566326222273,
      "grad_norm": 0.9207023652756564,
      "learning_rate": 9.66488344630874e-06,
      "loss": 1.0017,
      "step": 8010
    },
    {
      "epoch": 0.48468000241735665,
      "grad_norm": 0.9846740362494372,
      "learning_rate": 9.663683489763305e-06,
      "loss": 1.0082,
      "step": 8020
    },
    {
      "epoch": 0.4852843415724905,
      "grad_norm": 1.2877008149909468,
      "learning_rate": 9.662481463467154e-06,
      "loss": 0.9866,
      "step": 8030
    },
    {
      "epoch": 0.48588868072762437,
      "grad_norm": 0.958446530431782,
      "learning_rate": 9.661277367953747e-06,
      "loss": 0.9959,
      "step": 8040
    },
    {
      "epoch": 0.4864930198827582,
      "grad_norm": 2.4676621726281054,
      "learning_rate": 9.660071203757465e-06,
      "loss": 0.991,
      "step": 8050
    },
    {
      "epoch": 0.4870973590378921,
      "grad_norm": 2.7028175885296237,
      "learning_rate": 9.658862971413606e-06,
      "loss": 0.9903,
      "step": 8060
    },
    {
      "epoch": 0.48770169819302595,
      "grad_norm": 2.836053625288518,
      "learning_rate": 9.657652671458384e-06,
      "loss": 0.9829,
      "step": 8070
    },
    {
      "epoch": 0.4883060373481598,
      "grad_norm": 2.5175583173096814,
      "learning_rate": 9.656440304428934e-06,
      "loss": 0.9914,
      "step": 8080
    },
    {
      "epoch": 0.48891037650329366,
      "grad_norm": 2.4148200191591402,
      "learning_rate": 9.655225870863305e-06,
      "loss": 0.9925,
      "step": 8090
    },
    {
      "epoch": 0.4895147156584275,
      "grad_norm": 0.9786589854518857,
      "learning_rate": 9.654009371300464e-06,
      "loss": 1.0089,
      "step": 8100
    },
    {
      "epoch": 0.4901190548135614,
      "grad_norm": 0.9582223184546904,
      "learning_rate": 9.6527908062803e-06,
      "loss": 0.9749,
      "step": 8110
    },
    {
      "epoch": 0.49072339396869524,
      "grad_norm": 0.9828598576137438,
      "learning_rate": 9.651570176343607e-06,
      "loss": 0.989,
      "step": 8120
    },
    {
      "epoch": 0.4913277331238291,
      "grad_norm": 1.0086996495693867,
      "learning_rate": 9.65034748203211e-06,
      "loss": 1.0111,
      "step": 8130
    },
    {
      "epoch": 0.49193207227896296,
      "grad_norm": 0.9089158258226949,
      "learning_rate": 9.64912272388844e-06,
      "loss": 0.9944,
      "step": 8140
    },
    {
      "epoch": 0.4925364114340968,
      "grad_norm": 0.9903678427113559,
      "learning_rate": 9.647895902456145e-06,
      "loss": 0.9949,
      "step": 8150
    },
    {
      "epoch": 0.4931407505892307,
      "grad_norm": 0.9630889224980455,
      "learning_rate": 9.646667018279693e-06,
      "loss": 0.9862,
      "step": 8160
    },
    {
      "epoch": 0.49374508974436454,
      "grad_norm": 0.9336717507085102,
      "learning_rate": 9.645436071904464e-06,
      "loss": 1.0281,
      "step": 8170
    },
    {
      "epoch": 0.4943494288994984,
      "grad_norm": 1.0469956723088596,
      "learning_rate": 9.644203063876753e-06,
      "loss": 0.9809,
      "step": 8180
    },
    {
      "epoch": 0.49495376805463226,
      "grad_norm": 0.9570505244158888,
      "learning_rate": 9.64296799474377e-06,
      "loss": 1.0014,
      "step": 8190
    },
    {
      "epoch": 0.4955581072097661,
      "grad_norm": 1.0258599596488671,
      "learning_rate": 9.641730865053646e-06,
      "loss": 1.0164,
      "step": 8200
    },
    {
      "epoch": 0.4961624463649,
      "grad_norm": 1.035835625069818,
      "learning_rate": 9.640491675355417e-06,
      "loss": 1.0077,
      "step": 8210
    },
    {
      "epoch": 0.49676678552003384,
      "grad_norm": 1.0276053532929534,
      "learning_rate": 9.63925042619904e-06,
      "loss": 1.0317,
      "step": 8220
    },
    {
      "epoch": 0.4973711246751677,
      "grad_norm": 1.1452176363827613,
      "learning_rate": 9.63800711813538e-06,
      "loss": 0.9864,
      "step": 8230
    },
    {
      "epoch": 0.49797546383030156,
      "grad_norm": 1.0470289586674222,
      "learning_rate": 9.636761751716222e-06,
      "loss": 0.9873,
      "step": 8240
    },
    {
      "epoch": 0.4985798029854354,
      "grad_norm": 1.0735003796994644,
      "learning_rate": 9.63551432749426e-06,
      "loss": 0.9952,
      "step": 8250
    },
    {
      "epoch": 0.4991841421405693,
      "grad_norm": 1.1427966104197986,
      "learning_rate": 9.634264846023106e-06,
      "loss": 1.003,
      "step": 8260
    },
    {
      "epoch": 0.49978848129570314,
      "grad_norm": 1.0781624221613808,
      "learning_rate": 9.63301330785728e-06,
      "loss": 1.0147,
      "step": 8270
    },
    {
      "epoch": 0.500392820450837,
      "grad_norm": 1.148437759316674,
      "learning_rate": 9.631759713552216e-06,
      "loss": 0.9879,
      "step": 8280
    },
    {
      "epoch": 0.5009971596059709,
      "grad_norm": 1.1120292408713583,
      "learning_rate": 9.630504063664261e-06,
      "loss": 0.9853,
      "step": 8290
    },
    {
      "epoch": 0.5016014987611047,
      "grad_norm": 1.1402986570324811,
      "learning_rate": 9.629246358750676e-06,
      "loss": 0.9987,
      "step": 8300
    },
    {
      "epoch": 0.5022058379162386,
      "grad_norm": 1.069535183110544,
      "learning_rate": 9.627986599369634e-06,
      "loss": 0.982,
      "step": 8310
    },
    {
      "epoch": 0.5028101770713724,
      "grad_norm": 1.1654086241178052,
      "learning_rate": 9.626724786080217e-06,
      "loss": 0.9809,
      "step": 8320
    },
    {
      "epoch": 0.5034145162265063,
      "grad_norm": 1.176509443510629,
      "learning_rate": 9.625460919442418e-06,
      "loss": 0.9989,
      "step": 8330
    },
    {
      "epoch": 0.5040188553816402,
      "grad_norm": 1.1800644742301534,
      "learning_rate": 9.624195000017145e-06,
      "loss": 0.9934,
      "step": 8340
    },
    {
      "epoch": 0.5046231945367741,
      "grad_norm": 0.9953993274160409,
      "learning_rate": 9.622927028366216e-06,
      "loss": 0.9909,
      "step": 8350
    },
    {
      "epoch": 0.5052275336919079,
      "grad_norm": 1.0370138623117278,
      "learning_rate": 9.621657005052355e-06,
      "loss": 1.0092,
      "step": 8360
    },
    {
      "epoch": 0.5058318728470418,
      "grad_norm": 0.9922144147056596,
      "learning_rate": 9.620384930639209e-06,
      "loss": 1.002,
      "step": 8370
    },
    {
      "epoch": 0.5064362120021756,
      "grad_norm": 0.9612586208716696,
      "learning_rate": 9.619110805691317e-06,
      "loss": 0.9982,
      "step": 8380
    },
    {
      "epoch": 0.5070405511573095,
      "grad_norm": 1.0259940559255982,
      "learning_rate": 9.617834630774142e-06,
      "loss": 1.0022,
      "step": 8390
    },
    {
      "epoch": 0.5076448903124433,
      "grad_norm": 1.0920260843015757,
      "learning_rate": 9.616556406454054e-06,
      "loss": 0.9877,
      "step": 8400
    },
    {
      "epoch": 0.5082492294675772,
      "grad_norm": 1.1198165725467457,
      "learning_rate": 9.61527613329833e-06,
      "loss": 0.9941,
      "step": 8410
    },
    {
      "epoch": 0.508853568622711,
      "grad_norm": 1.025248313794968,
      "learning_rate": 9.613993811875158e-06,
      "loss": 0.9794,
      "step": 8420
    },
    {
      "epoch": 0.5094579077778449,
      "grad_norm": 1.0482919155273442,
      "learning_rate": 9.612709442753632e-06,
      "loss": 0.9843,
      "step": 8430
    },
    {
      "epoch": 0.5100622469329787,
      "grad_norm": 1.1347725822118602,
      "learning_rate": 9.61142302650376e-06,
      "loss": 1.0234,
      "step": 8440
    },
    {
      "epoch": 0.5106665860881127,
      "grad_norm": 1.0473865614970992,
      "learning_rate": 9.610134563696455e-06,
      "loss": 0.9939,
      "step": 8450
    },
    {
      "epoch": 0.5112709252432465,
      "grad_norm": 0.9657584550175458,
      "learning_rate": 9.608844054903538e-06,
      "loss": 1.0148,
      "step": 8460
    },
    {
      "epoch": 0.5118752643983804,
      "grad_norm": 0.9738617538281833,
      "learning_rate": 9.60755150069774e-06,
      "loss": 0.9968,
      "step": 8470
    },
    {
      "epoch": 0.5124796035535142,
      "grad_norm": 1.0308464966985456,
      "learning_rate": 9.606256901652697e-06,
      "loss": 1.0207,
      "step": 8480
    },
    {
      "epoch": 0.5130839427086481,
      "grad_norm": 1.0433317731026512,
      "learning_rate": 9.604960258342958e-06,
      "loss": 1.0295,
      "step": 8490
    },
    {
      "epoch": 0.5136882818637819,
      "grad_norm": 1.1801333006107229,
      "learning_rate": 9.603661571343971e-06,
      "loss": 0.9851,
      "step": 8500
    },
    {
      "epoch": 0.5142926210189158,
      "grad_norm": 1.2498868118643573,
      "learning_rate": 9.602360841232102e-06,
      "loss": 1.0056,
      "step": 8510
    },
    {
      "epoch": 0.5148969601740496,
      "grad_norm": 1.2145390084331917,
      "learning_rate": 9.601058068584609e-06,
      "loss": 0.9993,
      "step": 8520
    },
    {
      "epoch": 0.5155012993291835,
      "grad_norm": 1.2001517566812112,
      "learning_rate": 9.599753253979669e-06,
      "loss": 1.0024,
      "step": 8530
    },
    {
      "epoch": 0.5161056384843175,
      "grad_norm": 1.2086016201074825,
      "learning_rate": 9.598446397996362e-06,
      "loss": 0.9751,
      "step": 8540
    },
    {
      "epoch": 0.5167099776394513,
      "grad_norm": 1.1466449966340053,
      "learning_rate": 9.597137501214672e-06,
      "loss": 1.0079,
      "step": 8550
    },
    {
      "epoch": 0.5173143167945852,
      "grad_norm": 1.2406782631151254,
      "learning_rate": 9.595826564215488e-06,
      "loss": 0.9984,
      "step": 8560
    },
    {
      "epoch": 0.517918655949719,
      "grad_norm": 1.091992293326711,
      "learning_rate": 9.594513587580608e-06,
      "loss": 0.9997,
      "step": 8570
    },
    {
      "epoch": 0.5185229951048529,
      "grad_norm": 1.0945234033362277,
      "learning_rate": 9.593198571892732e-06,
      "loss": 0.9817,
      "step": 8580
    },
    {
      "epoch": 0.5191273342599867,
      "grad_norm": 1.1029070917753434,
      "learning_rate": 9.591881517735467e-06,
      "loss": 0.9758,
      "step": 8590
    },
    {
      "epoch": 0.5197316734151206,
      "grad_norm": 1.2907186213259112,
      "learning_rate": 9.590562425693325e-06,
      "loss": 0.9859,
      "step": 8600
    },
    {
      "epoch": 0.5203360125702544,
      "grad_norm": 1.2193408801578518,
      "learning_rate": 9.589241296351719e-06,
      "loss": 1.0061,
      "step": 8610
    },
    {
      "epoch": 0.5209403517253883,
      "grad_norm": 1.2730445282162828,
      "learning_rate": 9.587918130296969e-06,
      "loss": 1.0179,
      "step": 8620
    },
    {
      "epoch": 0.5215446908805221,
      "grad_norm": 1.2036104458854633,
      "learning_rate": 9.586592928116301e-06,
      "loss": 1.0322,
      "step": 8630
    },
    {
      "epoch": 0.522149030035656,
      "grad_norm": 1.1623555826965213,
      "learning_rate": 9.585265690397836e-06,
      "loss": 0.9926,
      "step": 8640
    },
    {
      "epoch": 0.5227533691907899,
      "grad_norm": 1.55081919981763,
      "learning_rate": 9.583936417730612e-06,
      "loss": 1.028,
      "step": 8650
    },
    {
      "epoch": 0.5233577083459238,
      "grad_norm": 1.4564423182409123,
      "learning_rate": 9.582605110704557e-06,
      "loss": 0.9915,
      "step": 8660
    },
    {
      "epoch": 0.5239620475010576,
      "grad_norm": 1.3833231922488896,
      "learning_rate": 9.58127176991051e-06,
      "loss": 0.9999,
      "step": 8670
    },
    {
      "epoch": 0.5245663866561915,
      "grad_norm": 1.3857524203326659,
      "learning_rate": 9.579936395940208e-06,
      "loss": 1.0047,
      "step": 8680
    },
    {
      "epoch": 0.5251707258113253,
      "grad_norm": 1.3576348780566196,
      "learning_rate": 9.578598989386293e-06,
      "loss": 1.008,
      "step": 8690
    },
    {
      "epoch": 0.5257750649664592,
      "grad_norm": 0.9212364600268664,
      "learning_rate": 9.577259550842309e-06,
      "loss": 1.0002,
      "step": 8700
    },
    {
      "epoch": 0.526379404121593,
      "grad_norm": 0.9009360363816509,
      "learning_rate": 9.5759180809027e-06,
      "loss": 1.003,
      "step": 8710
    },
    {
      "epoch": 0.5269837432767269,
      "grad_norm": 0.8544273543085027,
      "learning_rate": 9.574574580162815e-06,
      "loss": 0.9805,
      "step": 8720
    },
    {
      "epoch": 0.5275880824318607,
      "grad_norm": 0.8639960734005143,
      "learning_rate": 9.5732290492189e-06,
      "loss": 0.9892,
      "step": 8730
    },
    {
      "epoch": 0.5281924215869946,
      "grad_norm": 0.8431306173063707,
      "learning_rate": 9.571881488668102e-06,
      "loss": 0.977,
      "step": 8740
    },
    {
      "epoch": 0.5287967607421284,
      "grad_norm": 0.7528023093340848,
      "learning_rate": 9.570531899108474e-06,
      "loss": 0.9754,
      "step": 8750
    },
    {
      "epoch": 0.5294010998972624,
      "grad_norm": 0.7696537145424289,
      "learning_rate": 9.569180281138966e-06,
      "loss": 1.0071,
      "step": 8760
    },
    {
      "epoch": 0.5300054390523962,
      "grad_norm": 0.7786988594872661,
      "learning_rate": 9.567826635359427e-06,
      "loss": 1.0069,
      "step": 8770
    },
    {
      "epoch": 0.5306097782075301,
      "grad_norm": 0.753188512965207,
      "learning_rate": 9.566470962370608e-06,
      "loss": 1.0119,
      "step": 8780
    },
    {
      "epoch": 0.5312141173626639,
      "grad_norm": 0.7639611278654884,
      "learning_rate": 9.565113262774159e-06,
      "loss": 1.0059,
      "step": 8790
    },
    {
      "epoch": 0.5318184565177978,
      "grad_norm": 0.8278814556047578,
      "learning_rate": 9.563753537172628e-06,
      "loss": 0.9716,
      "step": 8800
    },
    {
      "epoch": 0.5324227956729316,
      "grad_norm": 0.8358293744224784,
      "learning_rate": 9.562391786169465e-06,
      "loss": 0.9886,
      "step": 8810
    },
    {
      "epoch": 0.5330271348280655,
      "grad_norm": 0.9560295768806172,
      "learning_rate": 9.561028010369017e-06,
      "loss": 0.9874,
      "step": 8820
    },
    {
      "epoch": 0.5336314739831993,
      "grad_norm": 0.8336233591725293,
      "learning_rate": 9.55966221037653e-06,
      "loss": 0.9948,
      "step": 8830
    },
    {
      "epoch": 0.5342358131383332,
      "grad_norm": 0.8508461702610761,
      "learning_rate": 9.558294386798147e-06,
      "loss": 0.9879,
      "step": 8840
    },
    {
      "epoch": 0.534840152293467,
      "grad_norm": 0.49984266226145047,
      "learning_rate": 9.556924540240914e-06,
      "loss": 0.9858,
      "step": 8850
    },
    {
      "epoch": 0.535444491448601,
      "grad_norm": 0.5838740900986712,
      "learning_rate": 9.555552671312772e-06,
      "loss": 1.004,
      "step": 8860
    },
    {
      "epoch": 0.5360488306037348,
      "grad_norm": 0.5152019090813889,
      "learning_rate": 9.554178780622551e-06,
      "loss": 0.9748,
      "step": 8870
    },
    {
      "epoch": 0.5366531697588687,
      "grad_norm": 0.5613467896896484,
      "learning_rate": 9.552802868779993e-06,
      "loss": 0.9837,
      "step": 8880
    },
    {
      "epoch": 0.5372575089140026,
      "grad_norm": 0.5311502872971174,
      "learning_rate": 9.551424936395728e-06,
      "loss": 0.9977,
      "step": 8890
    },
    {
      "epoch": 0.5378618480691364,
      "grad_norm": 0.5815211429254281,
      "learning_rate": 9.550044984081284e-06,
      "loss": 0.9845,
      "step": 8900
    },
    {
      "epoch": 0.5384661872242703,
      "grad_norm": 0.6404477430619733,
      "learning_rate": 9.548663012449088e-06,
      "loss": 0.9824,
      "step": 8910
    },
    {
      "epoch": 0.5390705263794041,
      "grad_norm": 0.6049707245085783,
      "learning_rate": 9.54727902211246e-06,
      "loss": 0.9771,
      "step": 8920
    },
    {
      "epoch": 0.539674865534538,
      "grad_norm": 0.5791098194179001,
      "learning_rate": 9.545893013685618e-06,
      "loss": 0.9835,
      "step": 8930
    },
    {
      "epoch": 0.5402792046896718,
      "grad_norm": 0.6043644346467444,
      "learning_rate": 9.544504987783674e-06,
      "loss": 0.9808,
      "step": 8940
    },
    {
      "epoch": 0.5408835438448057,
      "grad_norm": 0.8286167478370752,
      "learning_rate": 9.543114945022635e-06,
      "loss": 0.9771,
      "step": 8950
    },
    {
      "epoch": 0.5414878829999396,
      "grad_norm": 0.8136288452613204,
      "learning_rate": 9.54172288601941e-06,
      "loss": 1.0124,
      "step": 8960
    },
    {
      "epoch": 0.5420922221550735,
      "grad_norm": 0.7478057049812394,
      "learning_rate": 9.540328811391792e-06,
      "loss": 0.9943,
      "step": 8970
    },
    {
      "epoch": 0.5426965613102073,
      "grad_norm": 0.9397982937807772,
      "learning_rate": 9.538932721758474e-06,
      "loss": 0.9726,
      "step": 8980
    },
    {
      "epoch": 0.5433009004653412,
      "grad_norm": 0.821835902086452,
      "learning_rate": 9.537534617739044e-06,
      "loss": 0.9777,
      "step": 8990
    },
    {
      "epoch": 0.543905239620475,
      "grad_norm": 0.5292199167638089,
      "learning_rate": 9.536134499953984e-06,
      "loss": 1.0081,
      "step": 9000
    },
    {
      "epoch": 0.5445095787756089,
      "grad_norm": 0.5368485696924421,
      "learning_rate": 9.534732369024667e-06,
      "loss": 0.9989,
      "step": 9010
    },
    {
      "epoch": 0.5451139179307427,
      "grad_norm": 0.49688490628385246,
      "learning_rate": 9.533328225573362e-06,
      "loss": 0.9971,
      "step": 9020
    },
    {
      "epoch": 0.5457182570858766,
      "grad_norm": 0.49182087177664563,
      "learning_rate": 9.53192207022323e-06,
      "loss": 0.9835,
      "step": 9030
    },
    {
      "epoch": 0.5463225962410104,
      "grad_norm": 0.5086812831154581,
      "learning_rate": 9.530513903598326e-06,
      "loss": 0.9941,
      "step": 9040
    },
    {
      "epoch": 0.5469269353961443,
      "grad_norm": 0.5462714933115483,
      "learning_rate": 9.529103726323595e-06,
      "loss": 0.9876,
      "step": 9050
    },
    {
      "epoch": 0.5475312745512781,
      "grad_norm": 0.5002948002719604,
      "learning_rate": 9.527691539024877e-06,
      "loss": 0.9989,
      "step": 9060
    },
    {
      "epoch": 0.5481356137064121,
      "grad_norm": 0.4942418792498279,
      "learning_rate": 9.526277342328904e-06,
      "loss": 0.988,
      "step": 9070
    },
    {
      "epoch": 0.5487399528615459,
      "grad_norm": 0.5016699025806897,
      "learning_rate": 9.524861136863297e-06,
      "loss": 0.9862,
      "step": 9080
    },
    {
      "epoch": 0.5493442920166798,
      "grad_norm": 0.5215413382791309,
      "learning_rate": 9.523442923256572e-06,
      "loss": 0.9786,
      "step": 9090
    },
    {
      "epoch": 0.5499486311718136,
      "grad_norm": 0.545135195619704,
      "learning_rate": 9.522022702138134e-06,
      "loss": 0.9931,
      "step": 9100
    },
    {
      "epoch": 0.5505529703269475,
      "grad_norm": 0.519092415968085,
      "learning_rate": 9.52060047413828e-06,
      "loss": 0.9827,
      "step": 9110
    },
    {
      "epoch": 0.5511573094820813,
      "grad_norm": 0.53439021590655,
      "learning_rate": 9.519176239888196e-06,
      "loss": 0.9972,
      "step": 9120
    },
    {
      "epoch": 0.5517616486372152,
      "grad_norm": 0.5422170163674099,
      "learning_rate": 9.51775000001996e-06,
      "loss": 0.98,
      "step": 9130
    },
    {
      "epoch": 0.552365987792349,
      "grad_norm": 0.476965125946667,
      "learning_rate": 9.516321755166542e-06,
      "loss": 0.9971,
      "step": 9140
    },
    {
      "epoch": 0.5529703269474829,
      "grad_norm": 0.5081693326491318,
      "learning_rate": 9.514891505961798e-06,
      "loss": 0.9891,
      "step": 9150
    },
    {
      "epoch": 0.5535746661026167,
      "grad_norm": 0.5383871326489696,
      "learning_rate": 9.513459253040474e-06,
      "loss": 0.9966,
      "step": 9160
    },
    {
      "epoch": 0.5541790052577507,
      "grad_norm": 0.5241087983700397,
      "learning_rate": 9.512024997038207e-06,
      "loss": 1.0265,
      "step": 9170
    },
    {
      "epoch": 0.5547833444128845,
      "grad_norm": 0.5360607049669087,
      "learning_rate": 9.510588738591523e-06,
      "loss": 0.9853,
      "step": 9180
    },
    {
      "epoch": 0.5553876835680184,
      "grad_norm": 0.4941857075784585,
      "learning_rate": 9.509150478337834e-06,
      "loss": 0.9896,
      "step": 9190
    },
    {
      "epoch": 0.5559920227231522,
      "grad_norm": 0.5183668577725419,
      "learning_rate": 9.507710216915444e-06,
      "loss": 0.9833,
      "step": 9200
    },
    {
      "epoch": 0.5565963618782861,
      "grad_norm": 0.5465537464318688,
      "learning_rate": 9.506267954963543e-06,
      "loss": 0.9846,
      "step": 9210
    },
    {
      "epoch": 0.55720070103342,
      "grad_norm": 0.5055592133325096,
      "learning_rate": 9.50482369312221e-06,
      "loss": 0.9812,
      "step": 9220
    },
    {
      "epoch": 0.5578050401885538,
      "grad_norm": 0.5109013938650148,
      "learning_rate": 9.50337743203241e-06,
      "loss": 1.0031,
      "step": 9230
    },
    {
      "epoch": 0.5584093793436877,
      "grad_norm": 0.4979015076590743,
      "learning_rate": 9.501929172335996e-06,
      "loss": 0.9884,
      "step": 9240
    },
    {
      "epoch": 0.5590137184988215,
      "grad_norm": 0.5262878650739787,
      "learning_rate": 9.500478914675709e-06,
      "loss": 1.0129,
      "step": 9250
    },
    {
      "epoch": 0.5596180576539554,
      "grad_norm": 0.5202377604385394,
      "learning_rate": 9.499026659695176e-06,
      "loss": 0.971,
      "step": 9260
    },
    {
      "epoch": 0.5602223968090893,
      "grad_norm": 0.5268814147981004,
      "learning_rate": 9.497572408038909e-06,
      "loss": 0.9883,
      "step": 9270
    },
    {
      "epoch": 0.5608267359642232,
      "grad_norm": 0.5213577286801888,
      "learning_rate": 9.496116160352308e-06,
      "loss": 0.9834,
      "step": 9280
    },
    {
      "epoch": 0.561431075119357,
      "grad_norm": 0.5342746155854908,
      "learning_rate": 9.494657917281658e-06,
      "loss": 0.9699,
      "step": 9290
    },
    {
      "epoch": 0.5620354142744909,
      "grad_norm": 0.6176039596169591,
      "learning_rate": 9.49319767947413e-06,
      "loss": 0.9913,
      "step": 9300
    },
    {
      "epoch": 0.5626397534296247,
      "grad_norm": 0.6478503366597994,
      "learning_rate": 9.491735447577781e-06,
      "loss": 0.9913,
      "step": 9310
    },
    {
      "epoch": 0.5632440925847586,
      "grad_norm": 0.5966945330201456,
      "learning_rate": 9.490271222241552e-06,
      "loss": 0.9715,
      "step": 9320
    },
    {
      "epoch": 0.5638484317398924,
      "grad_norm": 0.5978389545925465,
      "learning_rate": 9.488805004115267e-06,
      "loss": 1.0289,
      "step": 9330
    },
    {
      "epoch": 0.5644527708950263,
      "grad_norm": 0.6286153369876969,
      "learning_rate": 9.487336793849636e-06,
      "loss": 0.9878,
      "step": 9340
    },
    {
      "epoch": 0.5650571100501601,
      "grad_norm": 0.6447773201389241,
      "learning_rate": 9.485866592096254e-06,
      "loss": 0.9879,
      "step": 9350
    },
    {
      "epoch": 0.565661449205294,
      "grad_norm": 0.6832036844551603,
      "learning_rate": 9.4843943995076e-06,
      "loss": 0.9977,
      "step": 9360
    },
    {
      "epoch": 0.5662657883604278,
      "grad_norm": 0.6478441787196813,
      "learning_rate": 9.482920216737035e-06,
      "loss": 0.984,
      "step": 9370
    },
    {
      "epoch": 0.5668701275155618,
      "grad_norm": 0.661678029346523,
      "learning_rate": 9.481444044438803e-06,
      "loss": 1.0017,
      "step": 9380
    },
    {
      "epoch": 0.5674744666706956,
      "grad_norm": 0.6314072896741001,
      "learning_rate": 9.479965883268034e-06,
      "loss": 0.9679,
      "step": 9390
    },
    {
      "epoch": 0.5680788058258295,
      "grad_norm": 0.625023868061403,
      "learning_rate": 9.478485733880736e-06,
      "loss": 0.9934,
      "step": 9400
    },
    {
      "epoch": 0.5686831449809633,
      "grad_norm": 0.6783188143267269,
      "learning_rate": 9.477003596933802e-06,
      "loss": 0.9854,
      "step": 9410
    },
    {
      "epoch": 0.5692874841360972,
      "grad_norm": 0.6059790817111019,
      "learning_rate": 9.475519473085007e-06,
      "loss": 1.0014,
      "step": 9420
    },
    {
      "epoch": 0.569891823291231,
      "grad_norm": 0.6150126760865275,
      "learning_rate": 9.47403336299301e-06,
      "loss": 0.9844,
      "step": 9430
    },
    {
      "epoch": 0.5704961624463649,
      "grad_norm": 0.5849261192824202,
      "learning_rate": 9.472545267317348e-06,
      "loss": 0.9821,
      "step": 9440
    },
    {
      "epoch": 0.5711005016014987,
      "grad_norm": 0.7942566989378923,
      "learning_rate": 9.471055186718439e-06,
      "loss": 0.9714,
      "step": 9450
    },
    {
      "epoch": 0.5717048407566326,
      "grad_norm": 0.7354328013281911,
      "learning_rate": 9.469563121857584e-06,
      "loss": 1.0034,
      "step": 9460
    },
    {
      "epoch": 0.5723091799117664,
      "grad_norm": 0.8356869141980198,
      "learning_rate": 9.468069073396964e-06,
      "loss": 0.9974,
      "step": 9470
    },
    {
      "epoch": 0.5729135190669004,
      "grad_norm": 0.7779732505425232,
      "learning_rate": 9.46657304199964e-06,
      "loss": 0.9797,
      "step": 9480
    },
    {
      "epoch": 0.5735178582220342,
      "grad_norm": 0.7618109936784662,
      "learning_rate": 9.465075028329556e-06,
      "loss": 0.9986,
      "step": 9490
    },
    {
      "epoch": 0.5741221973771681,
      "grad_norm": 0.8120984017758941,
      "learning_rate": 9.463575033051529e-06,
      "loss": 0.9714,
      "step": 9500
    },
    {
      "epoch": 0.5747265365323019,
      "grad_norm": 0.747960001563099,
      "learning_rate": 9.462073056831262e-06,
      "loss": 0.9705,
      "step": 9510
    },
    {
      "epoch": 0.5753308756874358,
      "grad_norm": 0.7586706353112267,
      "learning_rate": 9.460569100335333e-06,
      "loss": 1.005,
      "step": 9520
    },
    {
      "epoch": 0.5759352148425696,
      "grad_norm": 0.7893540322654525,
      "learning_rate": 9.4590631642312e-06,
      "loss": 0.9882,
      "step": 9530
    },
    {
      "epoch": 0.5765395539977035,
      "grad_norm": 0.7753500223338584,
      "learning_rate": 9.457555249187204e-06,
      "loss": 0.9868,
      "step": 9540
    },
    {
      "epoch": 0.5771438931528373,
      "grad_norm": 1.3579732998460725,
      "learning_rate": 9.456045355872558e-06,
      "loss": 1.0081,
      "step": 9550
    },
    {
      "epoch": 0.5777482323079712,
      "grad_norm": 1.3028543465646156,
      "learning_rate": 9.454533484957355e-06,
      "loss": 0.986,
      "step": 9560
    },
    {
      "epoch": 0.5783525714631051,
      "grad_norm": 1.2204451936756202,
      "learning_rate": 9.453019637112563e-06,
      "loss": 0.9885,
      "step": 9570
    },
    {
      "epoch": 0.578956910618239,
      "grad_norm": 1.2688451520885813,
      "learning_rate": 9.451503813010037e-06,
      "loss": 0.9742,
      "step": 9580
    },
    {
      "epoch": 0.5795612497733729,
      "grad_norm": 1.2332251712601991,
      "learning_rate": 9.449986013322494e-06,
      "loss": 0.9972,
      "step": 9590
    },
    {
      "epoch": 0.5801655889285067,
      "grad_norm": 0.9185071923501728,
      "learning_rate": 9.448466238723544e-06,
      "loss": 0.9997,
      "step": 9600
    },
    {
      "epoch": 0.5807699280836406,
      "grad_norm": 1.603028861506127,
      "learning_rate": 9.446944489887661e-06,
      "loss": 0.9892,
      "step": 9610
    },
    {
      "epoch": 0.5813742672387744,
      "grad_norm": 1.0395166278031611,
      "learning_rate": 9.445420767490202e-06,
      "loss": 0.9728,
      "step": 9620
    },
    {
      "epoch": 0.5819786063939083,
      "grad_norm": 0.9347629638545152,
      "learning_rate": 9.443895072207395e-06,
      "loss": 0.9502,
      "step": 9630
    },
    {
      "epoch": 0.5825829455490421,
      "grad_norm": 0.9582723129093502,
      "learning_rate": 9.442367404716346e-06,
      "loss": 1.0183,
      "step": 9640
    },
    {
      "epoch": 0.583187284704176,
      "grad_norm": 2.713079880615707,
      "learning_rate": 9.440837765695039e-06,
      "loss": 0.9794,
      "step": 9650
    },
    {
      "epoch": 0.5837916238593098,
      "grad_norm": 2.4885669667910792,
      "learning_rate": 9.43930615582233e-06,
      "loss": 0.9903,
      "step": 9660
    },
    {
      "epoch": 0.5843959630144437,
      "grad_norm": 2.551078314702416,
      "learning_rate": 9.43777257577795e-06,
      "loss": 1.0104,
      "step": 9670
    },
    {
      "epoch": 0.5850003021695775,
      "grad_norm": 2.4697224232030295,
      "learning_rate": 9.436237026242504e-06,
      "loss": 0.9931,
      "step": 9680
    },
    {
      "epoch": 0.5856046413247115,
      "grad_norm": 3.0647047995108356,
      "learning_rate": 9.434699507897471e-06,
      "loss": 1.0124,
      "step": 9690
    },
    {
      "epoch": 0.5862089804798453,
      "grad_norm": 1.0043624899734567,
      "learning_rate": 9.433160021425206e-06,
      "loss": 0.9837,
      "step": 9700
    },
    {
      "epoch": 0.5868133196349792,
      "grad_norm": 1.0118191531128673,
      "learning_rate": 9.431618567508933e-06,
      "loss": 0.9691,
      "step": 9710
    },
    {
      "epoch": 0.587417658790113,
      "grad_norm": 0.9728501550311862,
      "learning_rate": 9.430075146832753e-06,
      "loss": 0.9932,
      "step": 9720
    },
    {
      "epoch": 0.5880219979452469,
      "grad_norm": 0.9908305095858703,
      "learning_rate": 9.42852976008164e-06,
      "loss": 0.9774,
      "step": 9730
    },
    {
      "epoch": 0.5886263371003807,
      "grad_norm": 0.9506254841230665,
      "learning_rate": 9.426982407941439e-06,
      "loss": 0.9982,
      "step": 9740
    },
    {
      "epoch": 0.5892306762555146,
      "grad_norm": 1.0754394055990935,
      "learning_rate": 9.425433091098866e-06,
      "loss": 0.963,
      "step": 9750
    },
    {
      "epoch": 0.5898350154106484,
      "grad_norm": 1.0204283704917645,
      "learning_rate": 9.423881810241512e-06,
      "loss": 1.0157,
      "step": 9760
    },
    {
      "epoch": 0.5904393545657823,
      "grad_norm": 0.9552442835757272,
      "learning_rate": 9.422328566057837e-06,
      "loss": 1.0116,
      "step": 9770
    },
    {
      "epoch": 0.5910436937209161,
      "grad_norm": 1.0085077374117348,
      "learning_rate": 9.420773359237174e-06,
      "loss": 1.0117,
      "step": 9780
    },
    {
      "epoch": 0.5916480328760501,
      "grad_norm": 1.0391860248819054,
      "learning_rate": 9.419216190469727e-06,
      "loss": 0.9777,
      "step": 9790
    },
    {
      "epoch": 0.5922523720311839,
      "grad_norm": 1.0109958355782462,
      "learning_rate": 9.417657060446569e-06,
      "loss": 0.9784,
      "step": 9800
    },
    {
      "epoch": 0.5928567111863178,
      "grad_norm": 1.1342523088500804,
      "learning_rate": 9.416095969859644e-06,
      "loss": 1.0002,
      "step": 9810
    },
    {
      "epoch": 0.5934610503414516,
      "grad_norm": 1.2380678049192986,
      "learning_rate": 9.41453291940177e-06,
      "loss": 0.9912,
      "step": 9820
    },
    {
      "epoch": 0.5940653894965855,
      "grad_norm": 1.0149853034020109,
      "learning_rate": 9.412967909766629e-06,
      "loss": 0.9745,
      "step": 9830
    },
    {
      "epoch": 0.5946697286517193,
      "grad_norm": 1.049401113645641,
      "learning_rate": 9.411400941648773e-06,
      "loss": 1.0254,
      "step": 9840
    },
    {
      "epoch": 0.5952740678068532,
      "grad_norm": 1.1294305783283562,
      "learning_rate": 9.40983201574363e-06,
      "loss": 0.9863,
      "step": 9850
    },
    {
      "epoch": 0.595878406961987,
      "grad_norm": 1.1648909831717336,
      "learning_rate": 9.408261132747487e-06,
      "loss": 0.9868,
      "step": 9860
    },
    {
      "epoch": 0.5964827461171209,
      "grad_norm": 1.0159094764294985,
      "learning_rate": 9.40668829335751e-06,
      "loss": 0.9692,
      "step": 9870
    },
    {
      "epoch": 0.5970870852722547,
      "grad_norm": 1.066718729195064,
      "learning_rate": 9.405113498271722e-06,
      "loss": 1.006,
      "step": 9880
    },
    {
      "epoch": 0.5976914244273887,
      "grad_norm": 1.2297810928232251,
      "learning_rate": 9.403536748189024e-06,
      "loss": 1.0368,
      "step": 9890
    },
    {
      "epoch": 0.5982957635825226,
      "grad_norm": 1.0228162504015372,
      "learning_rate": 9.401958043809177e-06,
      "loss": 0.9941,
      "step": 9900
    },
    {
      "epoch": 0.5989001027376564,
      "grad_norm": 1.098669404992271,
      "learning_rate": 9.400377385832816e-06,
      "loss": 0.9963,
      "step": 9910
    },
    {
      "epoch": 0.5995044418927903,
      "grad_norm": 1.1280125593355481,
      "learning_rate": 9.398794774961439e-06,
      "loss": 1.0084,
      "step": 9920
    },
    {
      "epoch": 0.6001087810479241,
      "grad_norm": 1.1399829912066743,
      "learning_rate": 9.397210211897409e-06,
      "loss": 0.9923,
      "step": 9930
    },
    {
      "epoch": 0.600713120203058,
      "grad_norm": 1.1029918629585482,
      "learning_rate": 9.39562369734396e-06,
      "loss": 1.0308,
      "step": 9940
    },
    {
      "epoch": 0.6013174593581918,
      "grad_norm": 1.0633303796745397,
      "learning_rate": 9.39403523200519e-06,
      "loss": 0.9832,
      "step": 9950
    },
    {
      "epoch": 0.6019217985133257,
      "grad_norm": 1.0852106183377646,
      "learning_rate": 9.392444816586062e-06,
      "loss": 0.9924,
      "step": 9960
    },
    {
      "epoch": 0.6025261376684595,
      "grad_norm": 1.0236436268379463,
      "learning_rate": 9.390852451792404e-06,
      "loss": 1.0122,
      "step": 9970
    },
    {
      "epoch": 0.6031304768235934,
      "grad_norm": 0.9879030204382597,
      "learning_rate": 9.389258138330914e-06,
      "loss": 1.0021,
      "step": 9980
    },
    {
      "epoch": 0.6037348159787272,
      "grad_norm": 1.0131420022520858,
      "learning_rate": 9.387661876909146e-06,
      "loss": 0.9809,
      "step": 9990
    },
    {
      "epoch": 0.6043391551338612,
      "grad_norm": 1.101723066530562,
      "learning_rate": 9.386063668235527e-06,
      "loss": 0.9892,
      "step": 10000
    },
    {
      "epoch": 0.604943494288995,
      "grad_norm": 1.0696984489293015,
      "learning_rate": 9.384463513019343e-06,
      "loss": 0.9983,
      "step": 10010
    },
    {
      "epoch": 0.6055478334441289,
      "grad_norm": 1.1130438157603528,
      "learning_rate": 9.382861411970747e-06,
      "loss": 0.9795,
      "step": 10020
    },
    {
      "epoch": 0.6061521725992627,
      "grad_norm": 1.2325790126889524,
      "learning_rate": 9.381257365800752e-06,
      "loss": 1.0223,
      "step": 10030
    },
    {
      "epoch": 0.6067565117543966,
      "grad_norm": 1.0193174695653036,
      "learning_rate": 9.379651375221239e-06,
      "loss": 0.9832,
      "step": 10040
    },
    {
      "epoch": 0.6073608509095304,
      "grad_norm": 1.015611141303665,
      "learning_rate": 9.378043440944949e-06,
      "loss": 1.0039,
      "step": 10050
    },
    {
      "epoch": 0.6079651900646643,
      "grad_norm": 1.0920268320292426,
      "learning_rate": 9.376433563685484e-06,
      "loss": 0.9977,
      "step": 10060
    },
    {
      "epoch": 0.6085695292197981,
      "grad_norm": 1.0467741656123792,
      "learning_rate": 9.374821744157312e-06,
      "loss": 0.99,
      "step": 10070
    },
    {
      "epoch": 0.609173868374932,
      "grad_norm": 0.9334623498717842,
      "learning_rate": 9.37320798307576e-06,
      "loss": 1.0022,
      "step": 10080
    },
    {
      "epoch": 0.6097782075300658,
      "grad_norm": 0.9579103125574115,
      "learning_rate": 9.37159228115702e-06,
      "loss": 0.9957,
      "step": 10090
    },
    {
      "epoch": 0.6103825466851998,
      "grad_norm": 1.145755009703452,
      "learning_rate": 9.369974639118142e-06,
      "loss": 1.0031,
      "step": 10100
    },
    {
      "epoch": 0.6109868858403336,
      "grad_norm": 1.2177834882240255,
      "learning_rate": 9.368355057677038e-06,
      "loss": 0.9924,
      "step": 10110
    },
    {
      "epoch": 0.6115912249954675,
      "grad_norm": 1.1834276275165139,
      "learning_rate": 9.366733537552482e-06,
      "loss": 0.9833,
      "step": 10120
    },
    {
      "epoch": 0.6121955641506013,
      "grad_norm": 1.160520941982619,
      "learning_rate": 9.365110079464106e-06,
      "loss": 0.9738,
      "step": 10130
    },
    {
      "epoch": 0.6127999033057352,
      "grad_norm": 1.1972571768669398,
      "learning_rate": 9.363484684132405e-06,
      "loss": 1.0049,
      "step": 10140
    },
    {
      "epoch": 0.613404242460869,
      "grad_norm": 1.1528255829411618,
      "learning_rate": 9.36185735227873e-06,
      "loss": 0.982,
      "step": 10150
    },
    {
      "epoch": 0.6140085816160029,
      "grad_norm": 1.148665291820779,
      "learning_rate": 9.360228084625295e-06,
      "loss": 0.9841,
      "step": 10160
    },
    {
      "epoch": 0.6146129207711367,
      "grad_norm": 1.0587130470191386,
      "learning_rate": 9.358596881895175e-06,
      "loss": 0.9881,
      "step": 10170
    },
    {
      "epoch": 0.6152172599262706,
      "grad_norm": 1.0792518174482242,
      "learning_rate": 9.356963744812294e-06,
      "loss": 0.9851,
      "step": 10180
    },
    {
      "epoch": 0.6158215990814044,
      "grad_norm": 1.1221028189216802,
      "learning_rate": 9.355328674101445e-06,
      "loss": 1.0114,
      "step": 10190
    },
    {
      "epoch": 0.6164259382365384,
      "grad_norm": 1.2120226652942028,
      "learning_rate": 9.353691670488276e-06,
      "loss": 0.998,
      "step": 10200
    },
    {
      "epoch": 0.6170302773916722,
      "grad_norm": 1.209938198372769,
      "learning_rate": 9.35205273469929e-06,
      "loss": 1.0204,
      "step": 10210
    },
    {
      "epoch": 0.6176346165468061,
      "grad_norm": 1.2520259190155245,
      "learning_rate": 9.35041186746185e-06,
      "loss": 0.9981,
      "step": 10220
    },
    {
      "epoch": 0.6182389557019399,
      "grad_norm": 1.2066662035104347,
      "learning_rate": 9.348769069504177e-06,
      "loss": 0.9867,
      "step": 10230
    },
    {
      "epoch": 0.6188432948570738,
      "grad_norm": 1.2675288487407668,
      "learning_rate": 9.347124341555346e-06,
      "loss": 1.0054,
      "step": 10240
    },
    {
      "epoch": 0.6194476340122077,
      "grad_norm": 1.474204975125771,
      "learning_rate": 9.34547768434529e-06,
      "loss": 0.9995,
      "step": 10250
    },
    {
      "epoch": 0.6200519731673415,
      "grad_norm": 1.4503827511507366,
      "learning_rate": 9.343829098604799e-06,
      "loss": 1.01,
      "step": 10260
    },
    {
      "epoch": 0.6206563123224754,
      "grad_norm": 1.4858425013892371,
      "learning_rate": 9.342178585065518e-06,
      "loss": 0.9885,
      "step": 10270
    },
    {
      "epoch": 0.6212606514776092,
      "grad_norm": 1.3819249753680594,
      "learning_rate": 9.340526144459949e-06,
      "loss": 0.9985,
      "step": 10280
    },
    {
      "epoch": 0.6218649906327431,
      "grad_norm": 1.368574954106076,
      "learning_rate": 9.338871777521445e-06,
      "loss": 0.9801,
      "step": 10290
    },
    {
      "epoch": 0.622469329787877,
      "grad_norm": 0.8926510256484639,
      "learning_rate": 9.337215484984218e-06,
      "loss": 0.9876,
      "step": 10300
    },
    {
      "epoch": 0.6230736689430109,
      "grad_norm": 0.8474308596635595,
      "learning_rate": 9.335557267583336e-06,
      "loss": 1.0041,
      "step": 10310
    },
    {
      "epoch": 0.6236780080981447,
      "grad_norm": 0.8574273052321105,
      "learning_rate": 9.333897126054715e-06,
      "loss": 0.9984,
      "step": 10320
    },
    {
      "epoch": 0.6242823472532786,
      "grad_norm": 0.8053143689453196,
      "learning_rate": 9.33223506113513e-06,
      "loss": 0.9763,
      "step": 10330
    },
    {
      "epoch": 0.6248866864084124,
      "grad_norm": 0.8382296713322909,
      "learning_rate": 9.330571073562212e-06,
      "loss": 0.9894,
      "step": 10340
    },
    {
      "epoch": 0.6254910255635463,
      "grad_norm": 0.754298344905235,
      "learning_rate": 9.328905164074436e-06,
      "loss": 0.9663,
      "step": 10350
    },
    {
      "epoch": 0.6260953647186801,
      "grad_norm": 0.7560931894107045,
      "learning_rate": 9.327237333411137e-06,
      "loss": 0.9704,
      "step": 10360
    },
    {
      "epoch": 0.626699703873814,
      "grad_norm": 0.7163717942948837,
      "learning_rate": 9.325567582312502e-06,
      "loss": 0.9649,
      "step": 10370
    },
    {
      "epoch": 0.6273040430289478,
      "grad_norm": 0.7397543008565332,
      "learning_rate": 9.32389591151957e-06,
      "loss": 1.0193,
      "step": 10380
    },
    {
      "epoch": 0.6279083821840817,
      "grad_norm": 0.6857218389504587,
      "learning_rate": 9.32222232177423e-06,
      "loss": 0.9848,
      "step": 10390
    },
    {
      "epoch": 0.6285127213392155,
      "grad_norm": 0.8279028689989585,
      "learning_rate": 9.320546813819224e-06,
      "loss": 0.9858,
      "step": 10400
    },
    {
      "epoch": 0.6291170604943495,
      "grad_norm": 0.811835169761878,
      "learning_rate": 9.318869388398145e-06,
      "loss": 0.9726,
      "step": 10410
    },
    {
      "epoch": 0.6297213996494833,
      "grad_norm": 0.7476098064839439,
      "learning_rate": 9.317190046255441e-06,
      "loss": 0.9839,
      "step": 10420
    },
    {
      "epoch": 0.6303257388046172,
      "grad_norm": 0.8211948668569802,
      "learning_rate": 9.315508788136403e-06,
      "loss": 1.0099,
      "step": 10430
    },
    {
      "epoch": 0.630930077959751,
      "grad_norm": 0.8450834200931795,
      "learning_rate": 9.313825614787178e-06,
      "loss": 1.0062,
      "step": 10440
    },
    {
      "epoch": 0.6315344171148849,
      "grad_norm": 0.5310246813456097,
      "learning_rate": 9.312140526954758e-06,
      "loss": 0.9987,
      "step": 10450
    },
    {
      "epoch": 0.6321387562700187,
      "grad_norm": 0.5425339212418681,
      "learning_rate": 9.310453525386991e-06,
      "loss": 1.0092,
      "step": 10460
    },
    {
      "epoch": 0.6327430954251526,
      "grad_norm": 0.5747138692839948,
      "learning_rate": 9.308764610832573e-06,
      "loss": 1.0095,
      "step": 10470
    },
    {
      "epoch": 0.6333474345802864,
      "grad_norm": 0.5901590026224683,
      "learning_rate": 9.307073784041042e-06,
      "loss": 0.9802,
      "step": 10480
    },
    {
      "epoch": 0.6339517737354203,
      "grad_norm": 0.5281694225065243,
      "learning_rate": 9.305381045762793e-06,
      "loss": 0.9591,
      "step": 10490
    },
    {
      "epoch": 0.6345561128905541,
      "grad_norm": 0.5989764056690364,
      "learning_rate": 9.303686396749066e-06,
      "loss": 0.9845,
      "step": 10500
    },
    {
      "epoch": 0.635160452045688,
      "grad_norm": 0.6003949813112556,
      "learning_rate": 9.301989837751948e-06,
      "loss": 1.0148,
      "step": 10510
    },
    {
      "epoch": 0.6357647912008219,
      "grad_norm": 0.5855380672852845,
      "learning_rate": 9.300291369524376e-06,
      "loss": 0.9782,
      "step": 10520
    },
    {
      "epoch": 0.6363691303559558,
      "grad_norm": 0.573675503149501,
      "learning_rate": 9.298590992820133e-06,
      "loss": 0.9998,
      "step": 10530
    },
    {
      "epoch": 0.6369734695110896,
      "grad_norm": 0.5563569082183251,
      "learning_rate": 9.29688870839385e-06,
      "loss": 0.9856,
      "step": 10540
    },
    {
      "epoch": 0.6375778086662235,
      "grad_norm": 0.8318240048317055,
      "learning_rate": 9.295184517001001e-06,
      "loss": 0.9911,
      "step": 10550
    },
    {
      "epoch": 0.6381821478213573,
      "grad_norm": 0.8333033166195671,
      "learning_rate": 9.29347841939791e-06,
      "loss": 0.9498,
      "step": 10560
    },
    {
      "epoch": 0.6387864869764912,
      "grad_norm": 0.8459279462427219,
      "learning_rate": 9.291770416341748e-06,
      "loss": 0.9955,
      "step": 10570
    },
    {
      "epoch": 0.6393908261316251,
      "grad_norm": 0.924215356492686,
      "learning_rate": 9.290060508590526e-06,
      "loss": 1.0037,
      "step": 10580
    },
    {
      "epoch": 0.6399951652867589,
      "grad_norm": 0.748350916299008,
      "learning_rate": 9.288348696903108e-06,
      "loss": 0.9825,
      "step": 10590
    },
    {
      "epoch": 0.6405995044418928,
      "grad_norm": 0.4819836884833534,
      "learning_rate": 9.286634982039198e-06,
      "loss": 0.9847,
      "step": 10600
    },
    {
      "epoch": 0.6412038435970266,
      "grad_norm": 0.5208292991015511,
      "learning_rate": 9.284919364759343e-06,
      "loss": 0.9964,
      "step": 10610
    },
    {
      "epoch": 0.6418081827521606,
      "grad_norm": 0.532887856663687,
      "learning_rate": 9.28320184582494e-06,
      "loss": 0.9974,
      "step": 10620
    },
    {
      "epoch": 0.6424125219072944,
      "grad_norm": 0.5140749365844139,
      "learning_rate": 9.281482425998228e-06,
      "loss": 0.9812,
      "step": 10630
    },
    {
      "epoch": 0.6430168610624283,
      "grad_norm": 0.49709262973391977,
      "learning_rate": 9.279761106042283e-06,
      "loss": 0.9825,
      "step": 10640
    },
    {
      "epoch": 0.6436212002175621,
      "grad_norm": 0.5071574991468321,
      "learning_rate": 9.278037886721031e-06,
      "loss": 0.9666,
      "step": 10650
    },
    {
      "epoch": 0.644225539372696,
      "grad_norm": 0.4620004023568384,
      "learning_rate": 9.276312768799244e-06,
      "loss": 1.004,
      "step": 10660
    },
    {
      "epoch": 0.6448298785278298,
      "grad_norm": 0.5331264112902342,
      "learning_rate": 9.274585753042529e-06,
      "loss": 0.9988,
      "step": 10670
    },
    {
      "epoch": 0.6454342176829637,
      "grad_norm": 0.5074318556524561,
      "learning_rate": 9.27285684021734e-06,
      "loss": 0.9958,
      "step": 10680
    },
    {
      "epoch": 0.6460385568380975,
      "grad_norm": 0.479338544974117,
      "learning_rate": 9.271126031090969e-06,
      "loss": 0.9722,
      "step": 10690
    },
    {
      "epoch": 0.6466428959932314,
      "grad_norm": 0.4770875357859588,
      "learning_rate": 9.269393326431555e-06,
      "loss": 0.976,
      "step": 10700
    },
    {
      "epoch": 0.6472472351483652,
      "grad_norm": 0.5542886093780625,
      "learning_rate": 9.267658727008075e-06,
      "loss": 0.99,
      "step": 10710
    },
    {
      "epoch": 0.6478515743034992,
      "grad_norm": 0.47188132928993437,
      "learning_rate": 9.265922233590346e-06,
      "loss": 0.9772,
      "step": 10720
    },
    {
      "epoch": 0.648455913458633,
      "grad_norm": 0.505004625991562,
      "learning_rate": 9.264183846949025e-06,
      "loss": 1.0104,
      "step": 10730
    },
    {
      "epoch": 0.6490602526137669,
      "grad_norm": 0.5144095318288245,
      "learning_rate": 9.262443567855615e-06,
      "loss": 0.9922,
      "step": 10740
    },
    {
      "epoch": 0.6496645917689007,
      "grad_norm": 0.5693522566561156,
      "learning_rate": 9.260701397082453e-06,
      "loss": 1.0062,
      "step": 10750
    },
    {
      "epoch": 0.6502689309240346,
      "grad_norm": 0.4663028022486189,
      "learning_rate": 9.258957335402716e-06,
      "loss": 0.9705,
      "step": 10760
    },
    {
      "epoch": 0.6508732700791684,
      "grad_norm": 0.475090514579552,
      "learning_rate": 9.257211383590422e-06,
      "loss": 0.9809,
      "step": 10770
    },
    {
      "epoch": 0.6514776092343023,
      "grad_norm": 0.5183302331900691,
      "learning_rate": 9.255463542420428e-06,
      "loss": 0.9714,
      "step": 10780
    },
    {
      "epoch": 0.6520819483894361,
      "grad_norm": 0.4981893480673195,
      "learning_rate": 9.253713812668432e-06,
      "loss": 0.978,
      "step": 10790
    },
    {
      "epoch": 0.65268628754457,
      "grad_norm": 0.5229191206206769,
      "learning_rate": 9.251962195110961e-06,
      "loss": 0.9806,
      "step": 10800
    },
    {
      "epoch": 0.6532906266997038,
      "grad_norm": 0.4999239946554239,
      "learning_rate": 9.250208690525392e-06,
      "loss": 1.0027,
      "step": 10810
    },
    {
      "epoch": 0.6538949658548378,
      "grad_norm": 0.4848409840368509,
      "learning_rate": 9.24845329968993e-06,
      "loss": 0.9724,
      "step": 10820
    },
    {
      "epoch": 0.6544993050099716,
      "grad_norm": 0.5023465648967292,
      "learning_rate": 9.246696023383618e-06,
      "loss": 1.0,
      "step": 10830
    },
    {
      "epoch": 0.6551036441651055,
      "grad_norm": 0.5589989100230456,
      "learning_rate": 9.244936862386343e-06,
      "loss": 0.9863,
      "step": 10840
    },
    {
      "epoch": 0.6557079833202393,
      "grad_norm": 0.5136965662227517,
      "learning_rate": 9.243175817478822e-06,
      "loss": 0.9911,
      "step": 10850
    },
    {
      "epoch": 0.6563123224753732,
      "grad_norm": 0.5160872271921505,
      "learning_rate": 9.241412889442608e-06,
      "loss": 0.9873,
      "step": 10860
    },
    {
      "epoch": 0.656916661630507,
      "grad_norm": 0.5141559394274802,
      "learning_rate": 9.239648079060095e-06,
      "loss": 0.9845,
      "step": 10870
    },
    {
      "epoch": 0.6575210007856409,
      "grad_norm": 0.4933269971227344,
      "learning_rate": 9.237881387114506e-06,
      "loss": 0.9721,
      "step": 10880
    },
    {
      "epoch": 0.6581253399407747,
      "grad_norm": 0.5753448886446458,
      "learning_rate": 9.2361128143899e-06,
      "loss": 0.9709,
      "step": 10890
    },
    {
      "epoch": 0.6587296790959086,
      "grad_norm": 0.5705198321046384,
      "learning_rate": 9.23434236167118e-06,
      "loss": 0.9922,
      "step": 10900
    },
    {
      "epoch": 0.6593340182510424,
      "grad_norm": 0.595507704789605,
      "learning_rate": 9.232570029744068e-06,
      "loss": 0.9783,
      "step": 10910
    },
    {
      "epoch": 0.6599383574061763,
      "grad_norm": 0.5686412147669858,
      "learning_rate": 9.230795819395132e-06,
      "loss": 1.0247,
      "step": 10920
    },
    {
      "epoch": 0.6605426965613103,
      "grad_norm": 0.6141000987688027,
      "learning_rate": 9.229019731411769e-06,
      "loss": 1.0003,
      "step": 10930
    },
    {
      "epoch": 0.6611470357164441,
      "grad_norm": 0.5601687919154574,
      "learning_rate": 9.22724176658221e-06,
      "loss": 1.0045,
      "step": 10940
    },
    {
      "epoch": 0.661751374871578,
      "grad_norm": 0.6536199635611087,
      "learning_rate": 9.225461925695518e-06,
      "loss": 0.987,
      "step": 10950
    },
    {
      "epoch": 0.6623557140267118,
      "grad_norm": 0.7237551481360515,
      "learning_rate": 9.22368020954159e-06,
      "loss": 1.0059,
      "step": 10960
    },
    {
      "epoch": 0.6629600531818457,
      "grad_norm": 0.6426615334396304,
      "learning_rate": 9.221896618911155e-06,
      "loss": 0.9852,
      "step": 10970
    },
    {
      "epoch": 0.6635643923369795,
      "grad_norm": 0.6895427524504979,
      "learning_rate": 9.220111154595773e-06,
      "loss": 0.9708,
      "step": 10980
    },
    {
      "epoch": 0.6641687314921134,
      "grad_norm": 0.6603713954626483,
      "learning_rate": 9.218323817387835e-06,
      "loss": 0.9681,
      "step": 10990
    },
    {
      "epoch": 0.6647730706472472,
      "grad_norm": 0.6131720081140489,
      "learning_rate": 9.216534608080567e-06,
      "loss": 0.9863,
      "step": 11000
    },
    {
      "epoch": 0.6653774098023811,
      "grad_norm": 0.6334715648239038,
      "learning_rate": 9.214743527468022e-06,
      "loss": 0.9938,
      "step": 11010
    },
    {
      "epoch": 0.6659817489575149,
      "grad_norm": 0.6093963682364176,
      "learning_rate": 9.212950576345083e-06,
      "loss": 0.9778,
      "step": 11020
    },
    {
      "epoch": 0.6665860881126489,
      "grad_norm": 0.6036885481806127,
      "learning_rate": 9.211155755507469e-06,
      "loss": 1.001,
      "step": 11030
    },
    {
      "epoch": 0.6671904272677827,
      "grad_norm": 0.6072954521914475,
      "learning_rate": 9.20935906575172e-06,
      "loss": 1.0085,
      "step": 11040
    },
    {
      "epoch": 0.6677947664229166,
      "grad_norm": 0.7483391104250456,
      "learning_rate": 9.207560507875211e-06,
      "loss": 0.9834,
      "step": 11050
    },
    {
      "epoch": 0.6683991055780504,
      "grad_norm": 0.8143711848042002,
      "learning_rate": 9.205760082676146e-06,
      "loss": 0.9707,
      "step": 11060
    },
    {
      "epoch": 0.6690034447331843,
      "grad_norm": 0.8024408970263641,
      "learning_rate": 9.203957790953557e-06,
      "loss": 0.9966,
      "step": 11070
    },
    {
      "epoch": 0.6696077838883181,
      "grad_norm": 0.7489943244195018,
      "learning_rate": 9.202153633507305e-06,
      "loss": 0.9829,
      "step": 11080
    },
    {
      "epoch": 0.670212123043452,
      "grad_norm": 0.7732069777318934,
      "learning_rate": 9.200347611138076e-06,
      "loss": 0.9986,
      "step": 11090
    },
    {
      "epoch": 0.6708164621985858,
      "grad_norm": 0.7656310561319766,
      "learning_rate": 9.198539724647386e-06,
      "loss": 0.9949,
      "step": 11100
    },
    {
      "epoch": 0.6714208013537197,
      "grad_norm": 0.7255754082020898,
      "learning_rate": 9.19672997483758e-06,
      "loss": 1.008,
      "step": 11110
    },
    {
      "epoch": 0.6720251405088535,
      "grad_norm": 0.7745252880711284,
      "learning_rate": 9.194918362511824e-06,
      "loss": 0.9626,
      "step": 11120
    },
    {
      "epoch": 0.6726294796639875,
      "grad_norm": 0.7779136958327447,
      "learning_rate": 9.19310488847412e-06,
      "loss": 0.9878,
      "step": 11130
    },
    {
      "epoch": 0.6732338188191213,
      "grad_norm": 0.7194593861264345,
      "learning_rate": 9.19128955352929e-06,
      "loss": 1.0045,
      "step": 11140
    },
    {
      "epoch": 0.6738381579742552,
      "grad_norm": 1.1425815135973652,
      "learning_rate": 9.189472358482979e-06,
      "loss": 0.9893,
      "step": 11150
    },
    {
      "epoch": 0.674442497129389,
      "grad_norm": 1.2757276541801958,
      "learning_rate": 9.187653304141664e-06,
      "loss": 0.9717,
      "step": 11160
    },
    {
      "epoch": 0.6750468362845229,
      "grad_norm": 1.2396726750203204,
      "learning_rate": 9.185832391312644e-06,
      "loss": 0.9782,
      "step": 11170
    },
    {
      "epoch": 0.6756511754396567,
      "grad_norm": 1.2196346393467958,
      "learning_rate": 9.184009620804042e-06,
      "loss": 0.9921,
      "step": 11180
    },
    {
      "epoch": 0.6762555145947906,
      "grad_norm": 1.2102711188485422,
      "learning_rate": 9.182184993424809e-06,
      "loss": 1.0139,
      "step": 11190
    },
    {
      "epoch": 0.6768598537499244,
      "grad_norm": 1.0518544448686369,
      "learning_rate": 9.180358509984717e-06,
      "loss": 0.9913,
      "step": 11200
    },
    {
      "epoch": 0.6774641929050583,
      "grad_norm": 1.0281459667255863,
      "learning_rate": 9.178530171294362e-06,
      "loss": 0.9845,
      "step": 11210
    },
    {
      "epoch": 0.6780685320601921,
      "grad_norm": 0.9494328978376024,
      "learning_rate": 9.176699978165162e-06,
      "loss": 0.962,
      "step": 11220
    },
    {
      "epoch": 0.678672871215326,
      "grad_norm": 1.2616756497665285,
      "learning_rate": 9.174867931409364e-06,
      "loss": 0.9726,
      "step": 11230
    },
    {
      "epoch": 0.6792772103704598,
      "grad_norm": 0.9602385959102203,
      "learning_rate": 9.173034031840031e-06,
      "loss": 0.9968,
      "step": 11240
    },
    {
      "epoch": 0.6798815495255938,
      "grad_norm": 2.360136304782415,
      "learning_rate": 9.17119828027105e-06,
      "loss": 1.0026,
      "step": 11250
    },
    {
      "epoch": 0.6804858886807277,
      "grad_norm": 2.5682312858457377,
      "learning_rate": 9.169360677517132e-06,
      "loss": 1.004,
      "step": 11260
    },
    {
      "epoch": 0.6810902278358615,
      "grad_norm": 2.496816033745603,
      "learning_rate": 9.167521224393807e-06,
      "loss": 0.9988,
      "step": 11270
    },
    {
      "epoch": 0.6816945669909954,
      "grad_norm": 2.4382175754864672,
      "learning_rate": 9.16567992171743e-06,
      "loss": 0.9995,
      "step": 11280
    },
    {
      "epoch": 0.6822989061461292,
      "grad_norm": 2.59311472916446,
      "learning_rate": 9.163836770305173e-06,
      "loss": 1.0325,
      "step": 11290
    },
    {
      "epoch": 0.6829032453012631,
      "grad_norm": 0.9643666589556045,
      "learning_rate": 9.161991770975027e-06,
      "loss": 0.9827,
      "step": 11300
    },
    {
      "epoch": 0.6835075844563969,
      "grad_norm": 0.8990940350464417,
      "learning_rate": 9.160144924545809e-06,
      "loss": 0.9751,
      "step": 11310
    },
    {
      "epoch": 0.6841119236115308,
      "grad_norm": 0.99468382355489,
      "learning_rate": 9.158296231837152e-06,
      "loss": 0.9865,
      "step": 11320
    },
    {
      "epoch": 0.6847162627666646,
      "grad_norm": 0.8907197485462354,
      "learning_rate": 9.15644569366951e-06,
      "loss": 1.0031,
      "step": 11330
    },
    {
      "epoch": 0.6853206019217986,
      "grad_norm": 0.9086466011719967,
      "learning_rate": 9.154593310864152e-06,
      "loss": 0.9745,
      "step": 11340
    },
    {
      "epoch": 0.6859249410769324,
      "grad_norm": 0.9451981754943589,
      "learning_rate": 9.152739084243173e-06,
      "loss": 0.9804,
      "step": 11350
    },
    {
      "epoch": 0.6865292802320663,
      "grad_norm": 1.0670467949148794,
      "learning_rate": 9.150883014629478e-06,
      "loss": 1.0091,
      "step": 11360
    },
    {
      "epoch": 0.6871336193872001,
      "grad_norm": 0.8625833050272264,
      "learning_rate": 9.149025102846796e-06,
      "loss": 0.9576,
      "step": 11370
    },
    {
      "epoch": 0.687737958542334,
      "grad_norm": 0.9589680159000987,
      "learning_rate": 9.14716534971967e-06,
      "loss": 0.9996,
      "step": 11380
    },
    {
      "epoch": 0.6883422976974678,
      "grad_norm": 1.1221394032909637,
      "learning_rate": 9.145303756073462e-06,
      "loss": 1.0121,
      "step": 11390
    },
    {
      "epoch": 0.6889466368526017,
      "grad_norm": 1.0797298739274597,
      "learning_rate": 9.143440322734351e-06,
      "loss": 0.986,
      "step": 11400
    },
    {
      "epoch": 0.6895509760077355,
      "grad_norm": 1.0855279220793714,
      "learning_rate": 9.141575050529333e-06,
      "loss": 0.9977,
      "step": 11410
    },
    {
      "epoch": 0.6901553151628694,
      "grad_norm": 1.080811344042126,
      "learning_rate": 9.139707940286217e-06,
      "loss": 0.9786,
      "step": 11420
    },
    {
      "epoch": 0.6907596543180032,
      "grad_norm": 1.1494757087179435,
      "learning_rate": 9.13783899283363e-06,
      "loss": 0.9741,
      "step": 11430
    },
    {
      "epoch": 0.6913639934731372,
      "grad_norm": 0.9931475693939104,
      "learning_rate": 9.135968209001015e-06,
      "loss": 0.979,
      "step": 11440
    },
    {
      "epoch": 0.691968332628271,
      "grad_norm": 1.0369874046079264,
      "learning_rate": 9.13409558961863e-06,
      "loss": 1.0099,
      "step": 11450
    },
    {
      "epoch": 0.6925726717834049,
      "grad_norm": 0.9394206413278563,
      "learning_rate": 9.132221135517543e-06,
      "loss": 1.0058,
      "step": 11460
    },
    {
      "epoch": 0.6931770109385387,
      "grad_norm": 1.1204964671511828,
      "learning_rate": 9.130344847529642e-06,
      "loss": 0.9835,
      "step": 11470
    },
    {
      "epoch": 0.6937813500936726,
      "grad_norm": 1.044209823656874,
      "learning_rate": 9.128466726487629e-06,
      "loss": 0.9953,
      "step": 11480
    },
    {
      "epoch": 0.6943856892488064,
      "grad_norm": 1.019261995820744,
      "learning_rate": 9.126586773225015e-06,
      "loss": 1.0044,
      "step": 11490
    },
    {
      "epoch": 0.6949900284039403,
      "grad_norm": 1.2126115483222022,
      "learning_rate": 9.124704988576126e-06,
      "loss": 0.9862,
      "step": 11500
    },
    {
      "epoch": 0.6955943675590741,
      "grad_norm": 1.3053405603580308,
      "learning_rate": 9.122821373376103e-06,
      "loss": 0.995,
      "step": 11510
    },
    {
      "epoch": 0.696198706714208,
      "grad_norm": 1.0808497663578513,
      "learning_rate": 9.120935928460896e-06,
      "loss": 0.9759,
      "step": 11520
    },
    {
      "epoch": 0.6968030458693418,
      "grad_norm": 1.156907606411265,
      "learning_rate": 9.11904865466727e-06,
      "loss": 1.0087,
      "step": 11530
    },
    {
      "epoch": 0.6974073850244757,
      "grad_norm": 1.1856215445428815,
      "learning_rate": 9.1171595528328e-06,
      "loss": 1.0009,
      "step": 11540
    },
    {
      "epoch": 0.6980117241796095,
      "grad_norm": 0.9628398251299797,
      "learning_rate": 9.115268623795872e-06,
      "loss": 1.0248,
      "step": 11550
    },
    {
      "epoch": 0.6986160633347435,
      "grad_norm": 0.9392412711871685,
      "learning_rate": 9.113375868395684e-06,
      "loss": 1.0008,
      "step": 11560
    },
    {
      "epoch": 0.6992204024898773,
      "grad_norm": 0.936700435997837,
      "learning_rate": 9.111481287472244e-06,
      "loss": 0.9865,
      "step": 11570
    },
    {
      "epoch": 0.6998247416450112,
      "grad_norm": 0.9798770844112202,
      "learning_rate": 9.10958488186637e-06,
      "loss": 0.9682,
      "step": 11580
    },
    {
      "epoch": 0.700429080800145,
      "grad_norm": 1.0060994875143237,
      "learning_rate": 9.10768665241969e-06,
      "loss": 0.9872,
      "step": 11590
    },
    {
      "epoch": 0.7010334199552789,
      "grad_norm": 1.096508664211486,
      "learning_rate": 9.105786599974643e-06,
      "loss": 0.9898,
      "step": 11600
    },
    {
      "epoch": 0.7016377591104128,
      "grad_norm": 1.1152411037856387,
      "learning_rate": 9.103884725374475e-06,
      "loss": 0.9824,
      "step": 11610
    },
    {
      "epoch": 0.7022420982655466,
      "grad_norm": 1.0871318979322648,
      "learning_rate": 9.101981029463238e-06,
      "loss": 1.0027,
      "step": 11620
    },
    {
      "epoch": 0.7028464374206805,
      "grad_norm": 0.9910815261221368,
      "learning_rate": 9.1000755130858e-06,
      "loss": 0.9738,
      "step": 11630
    },
    {
      "epoch": 0.7034507765758143,
      "grad_norm": 0.9908666882350229,
      "learning_rate": 9.09816817708783e-06,
      "loss": 0.981,
      "step": 11640
    },
    {
      "epoch": 0.7040551157309483,
      "grad_norm": 0.959226320905126,
      "learning_rate": 9.096259022315807e-06,
      "loss": 1.0114,
      "step": 11650
    },
    {
      "epoch": 0.7046594548860821,
      "grad_norm": 1.021741168779716,
      "learning_rate": 9.094348049617018e-06,
      "loss": 0.9905,
      "step": 11660
    },
    {
      "epoch": 0.705263794041216,
      "grad_norm": 0.9611665847497872,
      "learning_rate": 9.092435259839556e-06,
      "loss": 0.9831,
      "step": 11670
    },
    {
      "epoch": 0.7058681331963498,
      "grad_norm": 0.9744434453061777,
      "learning_rate": 9.09052065383232e-06,
      "loss": 0.9955,
      "step": 11680
    },
    {
      "epoch": 0.7064724723514837,
      "grad_norm": 0.9913316169570978,
      "learning_rate": 9.088604232445015e-06,
      "loss": 1.0069,
      "step": 11690
    },
    {
      "epoch": 0.7070768115066175,
      "grad_norm": 1.1143855494751294,
      "learning_rate": 9.086685996528153e-06,
      "loss": 0.9878,
      "step": 11700
    },
    {
      "epoch": 0.7076811506617514,
      "grad_norm": 1.1206678887411492,
      "learning_rate": 9.084765946933049e-06,
      "loss": 0.9763,
      "step": 11710
    },
    {
      "epoch": 0.7082854898168852,
      "grad_norm": 1.147611908208286,
      "learning_rate": 9.082844084511826e-06,
      "loss": 0.993,
      "step": 11720
    },
    {
      "epoch": 0.7088898289720191,
      "grad_norm": 1.0563099789647354,
      "learning_rate": 9.080920410117408e-06,
      "loss": 0.9674,
      "step": 11730
    },
    {
      "epoch": 0.7094941681271529,
      "grad_norm": 1.1238285670454469,
      "learning_rate": 9.078994924603528e-06,
      "loss": 0.9699,
      "step": 11740
    },
    {
      "epoch": 0.7100985072822869,
      "grad_norm": 1.0348347262416522,
      "learning_rate": 9.077067628824717e-06,
      "loss": 0.9928,
      "step": 11750
    },
    {
      "epoch": 0.7107028464374207,
      "grad_norm": 1.0877791267266723,
      "learning_rate": 9.075138523636311e-06,
      "loss": 0.9852,
      "step": 11760
    },
    {
      "epoch": 0.7113071855925546,
      "grad_norm": 1.0290753796759835,
      "learning_rate": 9.073207609894453e-06,
      "loss": 0.9978,
      "step": 11770
    },
    {
      "epoch": 0.7119115247476884,
      "grad_norm": 1.0119374235792082,
      "learning_rate": 9.071274888456085e-06,
      "loss": 0.968,
      "step": 11780
    },
    {
      "epoch": 0.7125158639028223,
      "grad_norm": 1.084554789136185,
      "learning_rate": 9.069340360178954e-06,
      "loss": 0.98,
      "step": 11790
    },
    {
      "epoch": 0.7131202030579561,
      "grad_norm": 1.292945247968578,
      "learning_rate": 9.067404025921602e-06,
      "loss": 0.989,
      "step": 11800
    },
    {
      "epoch": 0.71372454221309,
      "grad_norm": 1.2325045084164217,
      "learning_rate": 9.065465886543383e-06,
      "loss": 0.9903,
      "step": 11810
    },
    {
      "epoch": 0.7143288813682238,
      "grad_norm": 1.1954982793614004,
      "learning_rate": 9.063525942904441e-06,
      "loss": 1.0155,
      "step": 11820
    },
    {
      "epoch": 0.7149332205233577,
      "grad_norm": 1.207349366987723,
      "learning_rate": 9.061584195865731e-06,
      "loss": 1.0011,
      "step": 11830
    },
    {
      "epoch": 0.7155375596784915,
      "grad_norm": 1.1640751486901595,
      "learning_rate": 9.059640646289003e-06,
      "loss": 0.9834,
      "step": 11840
    },
    {
      "epoch": 0.7161418988336254,
      "grad_norm": 1.383110403803805,
      "learning_rate": 9.057695295036806e-06,
      "loss": 0.9718,
      "step": 11850
    },
    {
      "epoch": 0.7167462379887592,
      "grad_norm": 1.40955164725621,
      "learning_rate": 9.055748142972492e-06,
      "loss": 0.9896,
      "step": 11860
    },
    {
      "epoch": 0.7173505771438932,
      "grad_norm": 1.3416954089177544,
      "learning_rate": 9.053799190960207e-06,
      "loss": 1.0037,
      "step": 11870
    },
    {
      "epoch": 0.717954916299027,
      "grad_norm": 1.326429827227262,
      "learning_rate": 9.0518484398649e-06,
      "loss": 0.9944,
      "step": 11880
    },
    {
      "epoch": 0.7185592554541609,
      "grad_norm": 1.3703176412803695,
      "learning_rate": 9.049895890552322e-06,
      "loss": 0.9749,
      "step": 11890
    },
    {
      "epoch": 0.7191635946092947,
      "grad_norm": 0.8697629241180893,
      "learning_rate": 9.047941543889014e-06,
      "loss": 1.0026,
      "step": 11900
    },
    {
      "epoch": 0.7197679337644286,
      "grad_norm": 0.86933569591174,
      "learning_rate": 9.045985400742321e-06,
      "loss": 0.9954,
      "step": 11910
    },
    {
      "epoch": 0.7203722729195624,
      "grad_norm": 0.8375890804454419,
      "learning_rate": 9.044027461980381e-06,
      "loss": 0.9912,
      "step": 11920
    },
    {
      "epoch": 0.7209766120746963,
      "grad_norm": 0.8707385495020277,
      "learning_rate": 9.042067728472129e-06,
      "loss": 1.0009,
      "step": 11930
    },
    {
      "epoch": 0.7215809512298301,
      "grad_norm": 0.8135350789072329,
      "learning_rate": 9.040106201087297e-06,
      "loss": 0.9781,
      "step": 11940
    },
    {
      "epoch": 0.722185290384964,
      "grad_norm": 0.7407151800562409,
      "learning_rate": 9.03814288069642e-06,
      "loss": 0.9838,
      "step": 11950
    },
    {
      "epoch": 0.722789629540098,
      "grad_norm": 0.8452069511713939,
      "learning_rate": 9.03617776817082e-06,
      "loss": 0.9963,
      "step": 11960
    },
    {
      "epoch": 0.7233939686952318,
      "grad_norm": 0.8312204615221693,
      "learning_rate": 9.034210864382616e-06,
      "loss": 1.0057,
      "step": 11970
    },
    {
      "epoch": 0.7239983078503657,
      "grad_norm": 0.7671606856646014,
      "learning_rate": 9.032242170204724e-06,
      "loss": 0.988,
      "step": 11980
    },
    {
      "epoch": 0.7246026470054995,
      "grad_norm": 0.7870259571936258,
      "learning_rate": 9.030271686510857e-06,
      "loss": 0.9838,
      "step": 11990
    },
    {
      "epoch": 0.7252069861606334,
      "grad_norm": 0.7722349099886145,
      "learning_rate": 9.028299414175513e-06,
      "loss": 0.9821,
      "step": 12000
    },
    {
      "epoch": 0.7258113253157672,
      "grad_norm": 0.7507826755962207,
      "learning_rate": 9.026325354073994e-06,
      "loss": 0.9839,
      "step": 12010
    },
    {
      "epoch": 0.7264156644709011,
      "grad_norm": 0.7804021600145328,
      "learning_rate": 9.024349507082391e-06,
      "loss": 0.9735,
      "step": 12020
    },
    {
      "epoch": 0.7270200036260349,
      "grad_norm": 0.7444413921959204,
      "learning_rate": 9.022371874077587e-06,
      "loss": 0.9643,
      "step": 12030
    },
    {
      "epoch": 0.7276243427811688,
      "grad_norm": 0.8052815652514865,
      "learning_rate": 9.020392455937258e-06,
      "loss": 0.969,
      "step": 12040
    },
    {
      "epoch": 0.7282286819363026,
      "grad_norm": 0.570848511247743,
      "learning_rate": 9.018411253539878e-06,
      "loss": 1.0035,
      "step": 12050
    },
    {
      "epoch": 0.7288330210914366,
      "grad_norm": 0.547092389803415,
      "learning_rate": 9.016428267764703e-06,
      "loss": 0.992,
      "step": 12060
    },
    {
      "epoch": 0.7294373602465704,
      "grad_norm": 0.5479318492657143,
      "learning_rate": 9.014443499491787e-06,
      "loss": 1.0049,
      "step": 12070
    },
    {
      "epoch": 0.7300416994017043,
      "grad_norm": 0.5392362781407463,
      "learning_rate": 9.012456949601973e-06,
      "loss": 0.9923,
      "step": 12080
    },
    {
      "epoch": 0.7306460385568381,
      "grad_norm": 0.5184908872445889,
      "learning_rate": 9.010468618976897e-06,
      "loss": 0.9784,
      "step": 12090
    },
    {
      "epoch": 0.731250377711972,
      "grad_norm": 0.5660185365497624,
      "learning_rate": 9.008478508498983e-06,
      "loss": 0.9948,
      "step": 12100
    },
    {
      "epoch": 0.7318547168671058,
      "grad_norm": 0.5976364002032862,
      "learning_rate": 9.006486619051444e-06,
      "loss": 0.9854,
      "step": 12110
    },
    {
      "epoch": 0.7324590560222397,
      "grad_norm": 0.6173132792024844,
      "learning_rate": 9.004492951518286e-06,
      "loss": 0.9768,
      "step": 12120
    },
    {
      "epoch": 0.7330633951773735,
      "grad_norm": 0.5808668053590242,
      "learning_rate": 9.0024975067843e-06,
      "loss": 0.9951,
      "step": 12130
    },
    {
      "epoch": 0.7336677343325074,
      "grad_norm": 0.5879149739609679,
      "learning_rate": 9.000500285735068e-06,
      "loss": 0.9587,
      "step": 12140
    },
    {
      "epoch": 0.7342720734876412,
      "grad_norm": 0.7645631638809431,
      "learning_rate": 8.998501289256962e-06,
      "loss": 1.0147,
      "step": 12150
    },
    {
      "epoch": 0.7348764126427751,
      "grad_norm": 0.8201230793337498,
      "learning_rate": 8.996500518237138e-06,
      "loss": 0.9862,
      "step": 12160
    },
    {
      "epoch": 0.735480751797909,
      "grad_norm": 0.8259271342722222,
      "learning_rate": 8.994497973563546e-06,
      "loss": 0.9916,
      "step": 12170
    },
    {
      "epoch": 0.7360850909530429,
      "grad_norm": 0.8422042361479396,
      "learning_rate": 8.99249365612491e-06,
      "loss": 1.003,
      "step": 12180
    },
    {
      "epoch": 0.7366894301081767,
      "grad_norm": 0.8094208736844749,
      "learning_rate": 8.990487566810758e-06,
      "loss": 1.0045,
      "step": 12190
    },
    {
      "epoch": 0.7372937692633106,
      "grad_norm": 0.495061085816348,
      "learning_rate": 8.988479706511392e-06,
      "loss": 0.9684,
      "step": 12200
    },
    {
      "epoch": 0.7378981084184444,
      "grad_norm": 0.46688358966768023,
      "learning_rate": 8.986470076117903e-06,
      "loss": 0.981,
      "step": 12210
    },
    {
      "epoch": 0.7385024475735783,
      "grad_norm": 0.5491548419505625,
      "learning_rate": 8.984458676522173e-06,
      "loss": 1.0009,
      "step": 12220
    },
    {
      "epoch": 0.7391067867287121,
      "grad_norm": 0.5004912743622855,
      "learning_rate": 8.98244550861686e-06,
      "loss": 0.9775,
      "step": 12230
    },
    {
      "epoch": 0.739711125883846,
      "grad_norm": 0.5037586346099409,
      "learning_rate": 8.980430573295412e-06,
      "loss": 0.9937,
      "step": 12240
    },
    {
      "epoch": 0.7403154650389798,
      "grad_norm": 0.50782572699603,
      "learning_rate": 8.978413871452063e-06,
      "loss": 0.9743,
      "step": 12250
    },
    {
      "epoch": 0.7409198041941137,
      "grad_norm": 0.48623540179183794,
      "learning_rate": 8.976395403981826e-06,
      "loss": 0.963,
      "step": 12260
    },
    {
      "epoch": 0.7415241433492475,
      "grad_norm": 0.525237134634794,
      "learning_rate": 8.974375171780503e-06,
      "loss": 0.9786,
      "step": 12270
    },
    {
      "epoch": 0.7421284825043815,
      "grad_norm": 0.4968514246767906,
      "learning_rate": 8.972353175744673e-06,
      "loss": 0.9776,
      "step": 12280
    },
    {
      "epoch": 0.7427328216595154,
      "grad_norm": 0.4954694245732145,
      "learning_rate": 8.970329416771707e-06,
      "loss": 0.9487,
      "step": 12290
    },
    {
      "epoch": 0.7433371608146492,
      "grad_norm": 0.486922501516735,
      "learning_rate": 8.968303895759746e-06,
      "loss": 0.9761,
      "step": 12300
    },
    {
      "epoch": 0.7439414999697831,
      "grad_norm": 0.4824293904099712,
      "learning_rate": 8.966276613607723e-06,
      "loss": 0.9815,
      "step": 12310
    },
    {
      "epoch": 0.7445458391249169,
      "grad_norm": 0.47489455049298224,
      "learning_rate": 8.96424757121535e-06,
      "loss": 0.9642,
      "step": 12320
    },
    {
      "epoch": 0.7451501782800508,
      "grad_norm": 0.5166013727865318,
      "learning_rate": 8.96221676948312e-06,
      "loss": 0.9966,
      "step": 12330
    },
    {
      "epoch": 0.7457545174351846,
      "grad_norm": 0.5397581526337417,
      "learning_rate": 8.960184209312303e-06,
      "loss": 0.972,
      "step": 12340
    },
    {
      "epoch": 0.7463588565903185,
      "grad_norm": 0.519150812071339,
      "learning_rate": 8.958149891604955e-06,
      "loss": 0.9657,
      "step": 12350
    },
    {
      "epoch": 0.7469631957454523,
      "grad_norm": 0.48926311701962183,
      "learning_rate": 8.95611381726391e-06,
      "loss": 0.9734,
      "step": 12360
    },
    {
      "epoch": 0.7475675349005863,
      "grad_norm": 0.4748196966353509,
      "learning_rate": 8.95407598719278e-06,
      "loss": 0.9832,
      "step": 12370
    },
    {
      "epoch": 0.74817187405572,
      "grad_norm": 0.49403967934107046,
      "learning_rate": 8.952036402295957e-06,
      "loss": 0.9835,
      "step": 12380
    },
    {
      "epoch": 0.748776213210854,
      "grad_norm": 0.47845085355691896,
      "learning_rate": 8.949995063478613e-06,
      "loss": 0.9782,
      "step": 12390
    },
    {
      "epoch": 0.7493805523659878,
      "grad_norm": 0.5183590080599418,
      "learning_rate": 8.9479519716467e-06,
      "loss": 0.9688,
      "step": 12400
    },
    {
      "epoch": 0.7499848915211217,
      "grad_norm": 0.5083758736917221,
      "learning_rate": 8.945907127706943e-06,
      "loss": 0.9809,
      "step": 12410
    },
    {
      "epoch": 0.7505892306762555,
      "grad_norm": 0.5217360128388254,
      "learning_rate": 8.943860532566848e-06,
      "loss": 0.9687,
      "step": 12420
    },
    {
      "epoch": 0.7511935698313894,
      "grad_norm": 0.4792187304605603,
      "learning_rate": 8.941812187134697e-06,
      "loss": 0.9694,
      "step": 12430
    },
    {
      "epoch": 0.7517979089865232,
      "grad_norm": 0.4938774550681597,
      "learning_rate": 8.93976209231955e-06,
      "loss": 0.9824,
      "step": 12440
    },
    {
      "epoch": 0.7524022481416571,
      "grad_norm": 0.5427203765930836,
      "learning_rate": 8.937710249031242e-06,
      "loss": 1.008,
      "step": 12450
    },
    {
      "epoch": 0.7530065872967909,
      "grad_norm": 0.5023301042273396,
      "learning_rate": 8.935656658180383e-06,
      "loss": 0.9724,
      "step": 12460
    },
    {
      "epoch": 0.7536109264519248,
      "grad_norm": 0.5194539153980096,
      "learning_rate": 8.933601320678364e-06,
      "loss": 0.9874,
      "step": 12470
    },
    {
      "epoch": 0.7542152656070586,
      "grad_norm": 0.49529764598306425,
      "learning_rate": 8.931544237437345e-06,
      "loss": 0.9902,
      "step": 12480
    },
    {
      "epoch": 0.7548196047621926,
      "grad_norm": 0.5494985257211211,
      "learning_rate": 8.929485409370262e-06,
      "loss": 0.9817,
      "step": 12490
    },
    {
      "epoch": 0.7554239439173264,
      "grad_norm": 0.6035382428142145,
      "learning_rate": 8.927424837390827e-06,
      "loss": 1.0071,
      "step": 12500
    },
    {
      "epoch": 0.7560282830724603,
      "grad_norm": 0.594329703039584,
      "learning_rate": 8.925362522413525e-06,
      "loss": 0.9952,
      "step": 12510
    },
    {
      "epoch": 0.7566326222275941,
      "grad_norm": 0.6182196854564175,
      "learning_rate": 8.923298465353619e-06,
      "loss": 0.972,
      "step": 12520
    },
    {
      "epoch": 0.757236961382728,
      "grad_norm": 0.5803582135895718,
      "learning_rate": 8.921232667127135e-06,
      "loss": 0.9784,
      "step": 12530
    },
    {
      "epoch": 0.7578413005378618,
      "grad_norm": 0.6042346067545898,
      "learning_rate": 8.919165128650883e-06,
      "loss": 0.9726,
      "step": 12540
    },
    {
      "epoch": 0.7584456396929957,
      "grad_norm": 0.6487320621608026,
      "learning_rate": 8.917095850842437e-06,
      "loss": 0.9997,
      "step": 12550
    },
    {
      "epoch": 0.7590499788481295,
      "grad_norm": 0.6322104977852444,
      "learning_rate": 8.915024834620148e-06,
      "loss": 0.9972,
      "step": 12560
    },
    {
      "epoch": 0.7596543180032634,
      "grad_norm": 0.6265604748891366,
      "learning_rate": 8.912952080903133e-06,
      "loss": 0.9646,
      "step": 12570
    },
    {
      "epoch": 0.7602586571583972,
      "grad_norm": 0.6315980662337992,
      "learning_rate": 8.910877590611287e-06,
      "loss": 0.974,
      "step": 12580
    },
    {
      "epoch": 0.7608629963135312,
      "grad_norm": 0.6525734752535817,
      "learning_rate": 8.908801364665271e-06,
      "loss": 0.9962,
      "step": 12590
    },
    {
      "epoch": 0.761467335468665,
      "grad_norm": 0.5984498114217938,
      "learning_rate": 8.90672340398652e-06,
      "loss": 0.992,
      "step": 12600
    },
    {
      "epoch": 0.7620716746237989,
      "grad_norm": 0.5847134619059274,
      "learning_rate": 8.904643709497232e-06,
      "loss": 0.988,
      "step": 12610
    },
    {
      "epoch": 0.7626760137789327,
      "grad_norm": 0.6243311340661921,
      "learning_rate": 8.902562282120384e-06,
      "loss": 0.9838,
      "step": 12620
    },
    {
      "epoch": 0.7632803529340666,
      "grad_norm": 0.6136992415016654,
      "learning_rate": 8.900479122779712e-06,
      "loss": 0.9649,
      "step": 12630
    },
    {
      "epoch": 0.7638846920892005,
      "grad_norm": 0.6181032997009042,
      "learning_rate": 8.89839423239973e-06,
      "loss": 0.9873,
      "step": 12640
    },
    {
      "epoch": 0.7644890312443343,
      "grad_norm": 0.7571176720444768,
      "learning_rate": 8.896307611905712e-06,
      "loss": 0.9689,
      "step": 12650
    },
    {
      "epoch": 0.7650933703994682,
      "grad_norm": 0.8257227649852571,
      "learning_rate": 8.894219262223708e-06,
      "loss": 0.9649,
      "step": 12660
    },
    {
      "epoch": 0.765697709554602,
      "grad_norm": 0.7573265597093679,
      "learning_rate": 8.892129184280529e-06,
      "loss": 0.9717,
      "step": 12670
    },
    {
      "epoch": 0.766302048709736,
      "grad_norm": 0.7765239035915268,
      "learning_rate": 8.890037379003753e-06,
      "loss": 0.9811,
      "step": 12680
    },
    {
      "epoch": 0.7669063878648698,
      "grad_norm": 0.772057018563983,
      "learning_rate": 8.887943847321732e-06,
      "loss": 0.9859,
      "step": 12690
    },
    {
      "epoch": 0.7675107270200037,
      "grad_norm": 0.7485074173903123,
      "learning_rate": 8.885848590163576e-06,
      "loss": 0.9792,
      "step": 12700
    },
    {
      "epoch": 0.7681150661751375,
      "grad_norm": 0.7391935357756217,
      "learning_rate": 8.88375160845916e-06,
      "loss": 0.9741,
      "step": 12710
    },
    {
      "epoch": 0.7687194053302714,
      "grad_norm": 0.7627188576521519,
      "learning_rate": 8.881652903139133e-06,
      "loss": 0.9726,
      "step": 12720
    },
    {
      "epoch": 0.7693237444854052,
      "grad_norm": 0.7719007616210114,
      "learning_rate": 8.879552475134905e-06,
      "loss": 0.9749,
      "step": 12730
    },
    {
      "epoch": 0.7699280836405391,
      "grad_norm": 0.7557094347350467,
      "learning_rate": 8.877450325378644e-06,
      "loss": 0.9872,
      "step": 12740
    },
    {
      "epoch": 0.7705324227956729,
      "grad_norm": 1.1654136615150479,
      "learning_rate": 8.875346454803293e-06,
      "loss": 0.9761,
      "step": 12750
    },
    {
      "epoch": 0.7711367619508068,
      "grad_norm": 1.1643518181661552,
      "learning_rate": 8.873240864342551e-06,
      "loss": 0.9834,
      "step": 12760
    },
    {
      "epoch": 0.7717411011059406,
      "grad_norm": 1.152879144672955,
      "learning_rate": 8.871133554930883e-06,
      "loss": 0.9773,
      "step": 12770
    },
    {
      "epoch": 0.7723454402610745,
      "grad_norm": 1.2105008183838115,
      "learning_rate": 8.869024527503515e-06,
      "loss": 0.9967,
      "step": 12780
    },
    {
      "epoch": 0.7729497794162083,
      "grad_norm": 1.156277855829216,
      "learning_rate": 8.86691378299644e-06,
      "loss": 0.9796,
      "step": 12790
    },
    {
      "epoch": 0.7735541185713423,
      "grad_norm": 0.871129940192965,
      "learning_rate": 8.864801322346409e-06,
      "loss": 0.9751,
      "step": 12800
    },
    {
      "epoch": 0.7741584577264761,
      "grad_norm": 1.0971713398537382,
      "learning_rate": 8.862687146490933e-06,
      "loss": 0.96,
      "step": 12810
    },
    {
      "epoch": 0.77476279688161,
      "grad_norm": 0.9603285046436115,
      "learning_rate": 8.860571256368291e-06,
      "loss": 0.9719,
      "step": 12820
    },
    {
      "epoch": 0.7753671360367438,
      "grad_norm": 1.0284086486563349,
      "learning_rate": 8.858453652917515e-06,
      "loss": 0.9822,
      "step": 12830
    },
    {
      "epoch": 0.7759714751918777,
      "grad_norm": 1.108532051114074,
      "learning_rate": 8.856334337078405e-06,
      "loss": 0.989,
      "step": 12840
    },
    {
      "epoch": 0.7765758143470115,
      "grad_norm": 2.288443083966407,
      "learning_rate": 8.854213309791512e-06,
      "loss": 0.9914,
      "step": 12850
    },
    {
      "epoch": 0.7771801535021454,
      "grad_norm": 2.2618259532924387,
      "learning_rate": 8.852090571998156e-06,
      "loss": 0.9672,
      "step": 12860
    },
    {
      "epoch": 0.7777844926572792,
      "grad_norm": 3.2938088308449385,
      "learning_rate": 8.84996612464041e-06,
      "loss": 0.9994,
      "step": 12870
    },
    {
      "epoch": 0.7783888318124131,
      "grad_norm": 2.4549008753153765,
      "learning_rate": 8.847839968661107e-06,
      "loss": 1.0199,
      "step": 12880
    },
    {
      "epoch": 0.7789931709675469,
      "grad_norm": 2.2624847474969516,
      "learning_rate": 8.845712105003839e-06,
      "loss": 0.9753,
      "step": 12890
    },
    {
      "epoch": 0.7795975101226809,
      "grad_norm": 0.9079545883891622,
      "learning_rate": 8.843582534612953e-06,
      "loss": 0.9892,
      "step": 12900
    },
    {
      "epoch": 0.7802018492778147,
      "grad_norm": 0.8898374237683913,
      "learning_rate": 8.841451258433561e-06,
      "loss": 0.9685,
      "step": 12910
    },
    {
      "epoch": 0.7808061884329486,
      "grad_norm": 1.022380287712484,
      "learning_rate": 8.839318277411525e-06,
      "loss": 0.9908,
      "step": 12920
    },
    {
      "epoch": 0.7814105275880824,
      "grad_norm": 0.8965094544144594,
      "learning_rate": 8.837183592493464e-06,
      "loss": 1.0301,
      "step": 12930
    },
    {
      "epoch": 0.7820148667432163,
      "grad_norm": 1.051180753939061,
      "learning_rate": 8.835047204626754e-06,
      "loss": 0.9689,
      "step": 12940
    },
    {
      "epoch": 0.7826192058983501,
      "grad_norm": 0.9220824188974558,
      "learning_rate": 8.832909114759531e-06,
      "loss": 0.985,
      "step": 12950
    },
    {
      "epoch": 0.783223545053484,
      "grad_norm": 0.9956000678829586,
      "learning_rate": 8.830769323840683e-06,
      "loss": 0.9718,
      "step": 12960
    },
    {
      "epoch": 0.7838278842086179,
      "grad_norm": 0.9865907366189631,
      "learning_rate": 8.828627832819849e-06,
      "loss": 0.9713,
      "step": 12970
    },
    {
      "epoch": 0.7844322233637517,
      "grad_norm": 0.9885251681361801,
      "learning_rate": 8.82648464264743e-06,
      "loss": 1.0016,
      "step": 12980
    },
    {
      "epoch": 0.7850365625188856,
      "grad_norm": 0.8647638258869463,
      "learning_rate": 8.824339754274575e-06,
      "loss": 0.9711,
      "step": 12990
    },
    {
      "epoch": 0.7856409016740195,
      "grad_norm": 0.9766631680171941,
      "learning_rate": 8.822193168653192e-06,
      "loss": 0.9913,
      "step": 13000
    },
    {
      "epoch": 0.7862452408291534,
      "grad_norm": 1.1066695703897402,
      "learning_rate": 8.820044886735937e-06,
      "loss": 0.9758,
      "step": 13010
    },
    {
      "epoch": 0.7868495799842872,
      "grad_norm": 1.0346901679097578,
      "learning_rate": 8.81789490947622e-06,
      "loss": 0.9784,
      "step": 13020
    },
    {
      "epoch": 0.7874539191394211,
      "grad_norm": 1.0138395969210046,
      "learning_rate": 8.815743237828209e-06,
      "loss": 0.9931,
      "step": 13030
    },
    {
      "epoch": 0.7880582582945549,
      "grad_norm": 0.9486189044453003,
      "learning_rate": 8.813589872746817e-06,
      "loss": 0.9708,
      "step": 13040
    },
    {
      "epoch": 0.7886625974496888,
      "grad_norm": 0.9412574776706261,
      "learning_rate": 8.81143481518771e-06,
      "loss": 0.974,
      "step": 13050
    },
    {
      "epoch": 0.7892669366048226,
      "grad_norm": 1.0842521628414044,
      "learning_rate": 8.809278066107307e-06,
      "loss": 0.9794,
      "step": 13060
    },
    {
      "epoch": 0.7898712757599565,
      "grad_norm": 1.0057938488081792,
      "learning_rate": 8.807119626462779e-06,
      "loss": 0.9756,
      "step": 13070
    },
    {
      "epoch": 0.7904756149150903,
      "grad_norm": 0.9875812606361327,
      "learning_rate": 8.804959497212042e-06,
      "loss": 0.9762,
      "step": 13080
    },
    {
      "epoch": 0.7910799540702242,
      "grad_norm": 1.0159773205094567,
      "learning_rate": 8.802797679313766e-06,
      "loss": 0.9839,
      "step": 13090
    },
    {
      "epoch": 0.791684293225358,
      "grad_norm": 1.1402570452683067,
      "learning_rate": 8.800634173727372e-06,
      "loss": 1.0008,
      "step": 13100
    },
    {
      "epoch": 0.792288632380492,
      "grad_norm": 1.2044459552875266,
      "learning_rate": 8.798468981413025e-06,
      "loss": 0.9879,
      "step": 13110
    },
    {
      "epoch": 0.7928929715356258,
      "grad_norm": 1.0625292941885056,
      "learning_rate": 8.79630210333164e-06,
      "loss": 0.9781,
      "step": 13120
    },
    {
      "epoch": 0.7934973106907597,
      "grad_norm": 1.137201790184712,
      "learning_rate": 8.794133540444884e-06,
      "loss": 0.9886,
      "step": 13130
    },
    {
      "epoch": 0.7941016498458935,
      "grad_norm": 1.0583191666183143,
      "learning_rate": 8.791963293715166e-06,
      "loss": 0.9732,
      "step": 13140
    },
    {
      "epoch": 0.7947059890010274,
      "grad_norm": 1.0450695896110342,
      "learning_rate": 8.78979136410565e-06,
      "loss": 0.9813,
      "step": 13150
    },
    {
      "epoch": 0.7953103281561612,
      "grad_norm": 1.0682241880229884,
      "learning_rate": 8.787617752580237e-06,
      "loss": 0.9665,
      "step": 13160
    },
    {
      "epoch": 0.7959146673112951,
      "grad_norm": 1.0131171225012323,
      "learning_rate": 8.785442460103581e-06,
      "loss": 0.9942,
      "step": 13170
    },
    {
      "epoch": 0.7965190064664289,
      "grad_norm": 0.9723856845729653,
      "learning_rate": 8.783265487641083e-06,
      "loss": 0.9935,
      "step": 13180
    },
    {
      "epoch": 0.7971233456215628,
      "grad_norm": 0.9588773108284964,
      "learning_rate": 8.781086836158884e-06,
      "loss": 0.9899,
      "step": 13190
    },
    {
      "epoch": 0.7977276847766966,
      "grad_norm": 1.067355261100459,
      "learning_rate": 8.778906506623875e-06,
      "loss": 0.9924,
      "step": 13200
    },
    {
      "epoch": 0.7983320239318306,
      "grad_norm": 0.9887635902127057,
      "learning_rate": 8.776724500003689e-06,
      "loss": 1.0129,
      "step": 13210
    },
    {
      "epoch": 0.7989363630869644,
      "grad_norm": 1.0662466768371086,
      "learning_rate": 8.774540817266703e-06,
      "loss": 1.001,
      "step": 13220
    },
    {
      "epoch": 0.7995407022420983,
      "grad_norm": 1.01603496144635,
      "learning_rate": 8.772355459382042e-06,
      "loss": 0.98,
      "step": 13230
    },
    {
      "epoch": 0.8001450413972321,
      "grad_norm": 1.0423178226386356,
      "learning_rate": 8.77016842731957e-06,
      "loss": 0.9536,
      "step": 13240
    },
    {
      "epoch": 0.800749380552366,
      "grad_norm": 0.9937477886555593,
      "learning_rate": 8.767979722049898e-06,
      "loss": 0.9841,
      "step": 13250
    },
    {
      "epoch": 0.8013537197074998,
      "grad_norm": 0.9948481782744802,
      "learning_rate": 8.765789344544373e-06,
      "loss": 0.97,
      "step": 13260
    },
    {
      "epoch": 0.8019580588626337,
      "grad_norm": 0.9202628224318222,
      "learning_rate": 8.76359729577509e-06,
      "loss": 0.9814,
      "step": 13270
    },
    {
      "epoch": 0.8025623980177675,
      "grad_norm": 0.9496722799701562,
      "learning_rate": 8.761403576714886e-06,
      "loss": 0.9918,
      "step": 13280
    },
    {
      "epoch": 0.8031667371729014,
      "grad_norm": 0.9319272570842513,
      "learning_rate": 8.759208188337335e-06,
      "loss": 0.9761,
      "step": 13290
    },
    {
      "epoch": 0.8037710763280352,
      "grad_norm": 1.2242296288204617,
      "learning_rate": 8.757011131616754e-06,
      "loss": 0.9912,
      "step": 13300
    },
    {
      "epoch": 0.8043754154831692,
      "grad_norm": 1.1882086142734938,
      "learning_rate": 8.7548124075282e-06,
      "loss": 0.9878,
      "step": 13310
    },
    {
      "epoch": 0.8049797546383031,
      "grad_norm": 1.1022888980543828,
      "learning_rate": 8.752612017047474e-06,
      "loss": 0.9833,
      "step": 13320
    },
    {
      "epoch": 0.8055840937934369,
      "grad_norm": 1.1801299846477953,
      "learning_rate": 8.750409961151113e-06,
      "loss": 0.9836,
      "step": 13330
    },
    {
      "epoch": 0.8061884329485708,
      "grad_norm": 1.146882476748507,
      "learning_rate": 8.748206240816388e-06,
      "loss": 0.9575,
      "step": 13340
    },
    {
      "epoch": 0.8067927721037046,
      "grad_norm": 1.087458660597489,
      "learning_rate": 8.746000857021318e-06,
      "loss": 0.9692,
      "step": 13350
    },
    {
      "epoch": 0.8073971112588385,
      "grad_norm": 1.049939773835628,
      "learning_rate": 8.743793810744655e-06,
      "loss": 1.0044,
      "step": 13360
    },
    {
      "epoch": 0.8080014504139723,
      "grad_norm": 1.075727090349215,
      "learning_rate": 8.741585102965888e-06,
      "loss": 0.9848,
      "step": 13370
    },
    {
      "epoch": 0.8086057895691062,
      "grad_norm": 1.0630622608807763,
      "learning_rate": 8.739374734665248e-06,
      "loss": 0.9773,
      "step": 13380
    },
    {
      "epoch": 0.80921012872424,
      "grad_norm": 1.0728569187017845,
      "learning_rate": 8.737162706823701e-06,
      "loss": 0.995,
      "step": 13390
    },
    {
      "epoch": 0.809814467879374,
      "grad_norm": 1.152574026721421,
      "learning_rate": 8.734949020422946e-06,
      "loss": 0.9752,
      "step": 13400
    },
    {
      "epoch": 0.8104188070345077,
      "grad_norm": 1.1812830246249495,
      "learning_rate": 8.732733676445422e-06,
      "loss": 0.9756,
      "step": 13410
    },
    {
      "epoch": 0.8110231461896417,
      "grad_norm": 1.1278256143246983,
      "learning_rate": 8.730516675874303e-06,
      "loss": 0.9949,
      "step": 13420
    },
    {
      "epoch": 0.8116274853447755,
      "grad_norm": 1.132258464057653,
      "learning_rate": 8.728298019693496e-06,
      "loss": 0.9939,
      "step": 13430
    },
    {
      "epoch": 0.8122318244999094,
      "grad_norm": 1.1585054158947528,
      "learning_rate": 8.726077708887645e-06,
      "loss": 0.9805,
      "step": 13440
    },
    {
      "epoch": 0.8128361636550432,
      "grad_norm": 1.3183845519856896,
      "learning_rate": 8.723855744442131e-06,
      "loss": 0.9891,
      "step": 13450
    },
    {
      "epoch": 0.8134405028101771,
      "grad_norm": 1.294442025174831,
      "learning_rate": 8.721632127343061e-06,
      "loss": 0.9827,
      "step": 13460
    },
    {
      "epoch": 0.8140448419653109,
      "grad_norm": 1.3180376381226397,
      "learning_rate": 8.719406858577283e-06,
      "loss": 0.9779,
      "step": 13470
    },
    {
      "epoch": 0.8146491811204448,
      "grad_norm": 1.3836647676847191,
      "learning_rate": 8.717179939132374e-06,
      "loss": 0.9808,
      "step": 13480
    },
    {
      "epoch": 0.8152535202755786,
      "grad_norm": 1.2321249849783131,
      "learning_rate": 8.714951369996642e-06,
      "loss": 0.9633,
      "step": 13490
    },
    {
      "epoch": 0.8158578594307125,
      "grad_norm": 0.8340901038421106,
      "learning_rate": 8.712721152159136e-06,
      "loss": 0.9618,
      "step": 13500
    },
    {
      "epoch": 0.8164621985858463,
      "grad_norm": 0.8177150348613327,
      "learning_rate": 8.710489286609624e-06,
      "loss": 0.9847,
      "step": 13510
    },
    {
      "epoch": 0.8170665377409803,
      "grad_norm": 0.788853435450462,
      "learning_rate": 8.708255774338616e-06,
      "loss": 0.9649,
      "step": 13520
    },
    {
      "epoch": 0.8176708768961141,
      "grad_norm": 0.8625179587825503,
      "learning_rate": 8.706020616337348e-06,
      "loss": 0.9696,
      "step": 13530
    },
    {
      "epoch": 0.818275216051248,
      "grad_norm": 0.8239071561559175,
      "learning_rate": 8.703783813597785e-06,
      "loss": 0.9589,
      "step": 13540
    },
    {
      "epoch": 0.8188795552063818,
      "grad_norm": 0.7234791914525768,
      "learning_rate": 8.701545367112627e-06,
      "loss": 0.9867,
      "step": 13550
    },
    {
      "epoch": 0.8194838943615157,
      "grad_norm": 0.7324782438751588,
      "learning_rate": 8.699305277875298e-06,
      "loss": 0.9687,
      "step": 13560
    },
    {
      "epoch": 0.8200882335166495,
      "grad_norm": 0.7579427313245379,
      "learning_rate": 8.697063546879952e-06,
      "loss": 0.9833,
      "step": 13570
    },
    {
      "epoch": 0.8206925726717834,
      "grad_norm": 0.7226863976371414,
      "learning_rate": 8.694820175121477e-06,
      "loss": 0.9894,
      "step": 13580
    },
    {
      "epoch": 0.8212969118269172,
      "grad_norm": 0.7178822526135513,
      "learning_rate": 8.692575163595483e-06,
      "loss": 0.9868,
      "step": 13590
    },
    {
      "epoch": 0.8219012509820511,
      "grad_norm": 0.8309495385533435,
      "learning_rate": 8.690328513298309e-06,
      "loss": 0.9874,
      "step": 13600
    },
    {
      "epoch": 0.8225055901371849,
      "grad_norm": 0.7964068295482079,
      "learning_rate": 8.688080225227024e-06,
      "loss": 0.9927,
      "step": 13610
    },
    {
      "epoch": 0.8231099292923189,
      "grad_norm": 0.7134736196963872,
      "learning_rate": 8.685830300379423e-06,
      "loss": 0.9846,
      "step": 13620
    },
    {
      "epoch": 0.8237142684474527,
      "grad_norm": 0.8016370223478746,
      "learning_rate": 8.683578739754023e-06,
      "loss": 0.972,
      "step": 13630
    },
    {
      "epoch": 0.8243186076025866,
      "grad_norm": 0.7828803915266184,
      "learning_rate": 8.681325544350073e-06,
      "loss": 0.9748,
      "step": 13640
    },
    {
      "epoch": 0.8249229467577205,
      "grad_norm": 0.525963738795146,
      "learning_rate": 8.679070715167545e-06,
      "loss": 0.9802,
      "step": 13650
    },
    {
      "epoch": 0.8255272859128543,
      "grad_norm": 0.5119496778046251,
      "learning_rate": 8.676814253207135e-06,
      "loss": 0.9954,
      "step": 13660
    },
    {
      "epoch": 0.8261316250679882,
      "grad_norm": 0.5320327454561167,
      "learning_rate": 8.674556159470266e-06,
      "loss": 0.9531,
      "step": 13670
    },
    {
      "epoch": 0.826735964223122,
      "grad_norm": 0.5298844005491382,
      "learning_rate": 8.672296434959081e-06,
      "loss": 0.9894,
      "step": 13680
    },
    {
      "epoch": 0.8273403033782559,
      "grad_norm": 0.568192299757382,
      "learning_rate": 8.670035080676455e-06,
      "loss": 1.0102,
      "step": 13690
    },
    {
      "epoch": 0.8279446425333897,
      "grad_norm": 0.5795311032175636,
      "learning_rate": 8.667772097625975e-06,
      "loss": 0.9979,
      "step": 13700
    },
    {
      "epoch": 0.8285489816885236,
      "grad_norm": 0.5482009557717502,
      "learning_rate": 8.665507486811962e-06,
      "loss": 0.9798,
      "step": 13710
    },
    {
      "epoch": 0.8291533208436574,
      "grad_norm": 0.5803001889632351,
      "learning_rate": 8.663241249239449e-06,
      "loss": 0.9676,
      "step": 13720
    },
    {
      "epoch": 0.8297576599987914,
      "grad_norm": 0.5585460629729855,
      "learning_rate": 8.6609733859142e-06,
      "loss": 0.9665,
      "step": 13730
    },
    {
      "epoch": 0.8303619991539252,
      "grad_norm": 0.6044475964382725,
      "learning_rate": 8.658703897842693e-06,
      "loss": 0.9917,
      "step": 13740
    },
    {
      "epoch": 0.8309663383090591,
      "grad_norm": 0.8075921621568113,
      "learning_rate": 8.656432786032135e-06,
      "loss": 0.9585,
      "step": 13750
    },
    {
      "epoch": 0.8315706774641929,
      "grad_norm": 0.7213302008928867,
      "learning_rate": 8.654160051490447e-06,
      "loss": 0.9847,
      "step": 13760
    },
    {
      "epoch": 0.8321750166193268,
      "grad_norm": 0.8258436481699448,
      "learning_rate": 8.651885695226273e-06,
      "loss": 0.9876,
      "step": 13770
    },
    {
      "epoch": 0.8327793557744606,
      "grad_norm": 0.7296502453047593,
      "learning_rate": 8.649609718248974e-06,
      "loss": 0.9693,
      "step": 13780
    },
    {
      "epoch": 0.8333836949295945,
      "grad_norm": 0.7565117159939226,
      "learning_rate": 8.647332121568635e-06,
      "loss": 0.9653,
      "step": 13790
    },
    {
      "epoch": 0.8339880340847283,
      "grad_norm": 0.4668739376583184,
      "learning_rate": 8.645052906196059e-06,
      "loss": 0.9756,
      "step": 13800
    },
    {
      "epoch": 0.8345923732398622,
      "grad_norm": 0.512767536719212,
      "learning_rate": 8.64277207314276e-06,
      "loss": 0.9805,
      "step": 13810
    },
    {
      "epoch": 0.835196712394996,
      "grad_norm": 0.513216989615094,
      "learning_rate": 8.640489623420982e-06,
      "loss": 0.9763,
      "step": 13820
    },
    {
      "epoch": 0.83580105155013,
      "grad_norm": 0.5159392361381759,
      "learning_rate": 8.638205558043674e-06,
      "loss": 0.9885,
      "step": 13830
    },
    {
      "epoch": 0.8364053907052638,
      "grad_norm": 0.507925826130658,
      "learning_rate": 8.635919878024514e-06,
      "loss": 0.9812,
      "step": 13840
    },
    {
      "epoch": 0.8370097298603977,
      "grad_norm": 0.5105816874123786,
      "learning_rate": 8.633632584377887e-06,
      "loss": 0.9351,
      "step": 13850
    },
    {
      "epoch": 0.8376140690155315,
      "grad_norm": 0.4843146889888788,
      "learning_rate": 8.6313436781189e-06,
      "loss": 0.9785,
      "step": 13860
    },
    {
      "epoch": 0.8382184081706654,
      "grad_norm": 0.4997169719585392,
      "learning_rate": 8.62905316026337e-06,
      "loss": 0.9948,
      "step": 13870
    },
    {
      "epoch": 0.8388227473257992,
      "grad_norm": 0.5167673178229548,
      "learning_rate": 8.626761031827836e-06,
      "loss": 0.9614,
      "step": 13880
    },
    {
      "epoch": 0.8394270864809331,
      "grad_norm": 0.4983261577437121,
      "learning_rate": 8.624467293829548e-06,
      "loss": 0.9869,
      "step": 13890
    },
    {
      "epoch": 0.8400314256360669,
      "grad_norm": 0.517457209055249,
      "learning_rate": 8.622171947286468e-06,
      "loss": 0.9734,
      "step": 13900
    },
    {
      "epoch": 0.8406357647912008,
      "grad_norm": 0.4624168463259461,
      "learning_rate": 8.61987499321728e-06,
      "loss": 1.0018,
      "step": 13910
    },
    {
      "epoch": 0.8412401039463346,
      "grad_norm": 0.548196103565086,
      "learning_rate": 8.617576432641373e-06,
      "loss": 1.0105,
      "step": 13920
    },
    {
      "epoch": 0.8418444431014686,
      "grad_norm": 0.5151892472790248,
      "learning_rate": 8.615276266578851e-06,
      "loss": 0.989,
      "step": 13930
    },
    {
      "epoch": 0.8424487822566024,
      "grad_norm": 0.5135347974928427,
      "learning_rate": 8.612974496050533e-06,
      "loss": 0.9914,
      "step": 13940
    },
    {
      "epoch": 0.8430531214117363,
      "grad_norm": 0.45531070788787553,
      "learning_rate": 8.610671122077949e-06,
      "loss": 0.9602,
      "step": 13950
    },
    {
      "epoch": 0.8436574605668701,
      "grad_norm": 0.5286391714421553,
      "learning_rate": 8.60836614568334e-06,
      "loss": 0.9808,
      "step": 13960
    },
    {
      "epoch": 0.844261799722004,
      "grad_norm": 0.5216868629946407,
      "learning_rate": 8.60605956788966e-06,
      "loss": 0.9918,
      "step": 13970
    },
    {
      "epoch": 0.8448661388771378,
      "grad_norm": 0.529974168619616,
      "learning_rate": 8.603751389720569e-06,
      "loss": 0.9892,
      "step": 13980
    },
    {
      "epoch": 0.8454704780322717,
      "grad_norm": 0.4881199578200851,
      "learning_rate": 8.601441612200443e-06,
      "loss": 0.983,
      "step": 13990
    },
    {
      "epoch": 0.8460748171874056,
      "grad_norm": 0.5157680614443185,
      "learning_rate": 8.599130236354365e-06,
      "loss": 1.006,
      "step": 14000
    },
    {
      "epoch": 0.8466791563425394,
      "grad_norm": 0.4870445244532973,
      "learning_rate": 8.596817263208125e-06,
      "loss": 0.9823,
      "step": 14010
    },
    {
      "epoch": 0.8472834954976733,
      "grad_norm": 0.5251658980807535,
      "learning_rate": 8.594502693788227e-06,
      "loss": 1.0009,
      "step": 14020
    },
    {
      "epoch": 0.8478878346528071,
      "grad_norm": 0.4754799389422651,
      "learning_rate": 8.592186529121879e-06,
      "loss": 0.9973,
      "step": 14030
    },
    {
      "epoch": 0.8484921738079411,
      "grad_norm": 0.540879750874813,
      "learning_rate": 8.589868770237001e-06,
      "loss": 1.0054,
      "step": 14040
    },
    {
      "epoch": 0.8490965129630749,
      "grad_norm": 0.5270743610518237,
      "learning_rate": 8.58754941816222e-06,
      "loss": 0.9632,
      "step": 14050
    },
    {
      "epoch": 0.8497008521182088,
      "grad_norm": 0.529225460967496,
      "learning_rate": 8.585228473926864e-06,
      "loss": 0.9779,
      "step": 14060
    },
    {
      "epoch": 0.8503051912733426,
      "grad_norm": 0.4733574050188894,
      "learning_rate": 8.582905938560972e-06,
      "loss": 0.9821,
      "step": 14070
    },
    {
      "epoch": 0.8509095304284765,
      "grad_norm": 0.506945004798917,
      "learning_rate": 8.580581813095292e-06,
      "loss": 0.9825,
      "step": 14080
    },
    {
      "epoch": 0.8515138695836103,
      "grad_norm": 0.5504746195997688,
      "learning_rate": 8.578256098561276e-06,
      "loss": 0.9795,
      "step": 14090
    },
    {
      "epoch": 0.8521182087387442,
      "grad_norm": 0.5597103599368471,
      "learning_rate": 8.575928795991076e-06,
      "loss": 0.9658,
      "step": 14100
    },
    {
      "epoch": 0.852722547893878,
      "grad_norm": 0.5759872420755888,
      "learning_rate": 8.573599906417554e-06,
      "loss": 0.9894,
      "step": 14110
    },
    {
      "epoch": 0.8533268870490119,
      "grad_norm": 0.5963806452644624,
      "learning_rate": 8.571269430874277e-06,
      "loss": 0.9641,
      "step": 14120
    },
    {
      "epoch": 0.8539312262041457,
      "grad_norm": 0.6096464374638144,
      "learning_rate": 8.568937370395514e-06,
      "loss": 0.9873,
      "step": 14130
    },
    {
      "epoch": 0.8545355653592797,
      "grad_norm": 0.5738626573172579,
      "learning_rate": 8.566603726016236e-06,
      "loss": 0.9717,
      "step": 14140
    },
    {
      "epoch": 0.8551399045144135,
      "grad_norm": 0.635145077266177,
      "learning_rate": 8.564268498772116e-06,
      "loss": 0.9915,
      "step": 14150
    },
    {
      "epoch": 0.8557442436695474,
      "grad_norm": 0.669204900845668,
      "learning_rate": 8.561931689699538e-06,
      "loss": 1.0097,
      "step": 14160
    },
    {
      "epoch": 0.8563485828246812,
      "grad_norm": 0.6218427952078261,
      "learning_rate": 8.559593299835578e-06,
      "loss": 0.961,
      "step": 14170
    },
    {
      "epoch": 0.8569529219798151,
      "grad_norm": 0.6461805763290213,
      "learning_rate": 8.557253330218019e-06,
      "loss": 0.9888,
      "step": 14180
    },
    {
      "epoch": 0.8575572611349489,
      "grad_norm": 0.701577584718705,
      "learning_rate": 8.55491178188534e-06,
      "loss": 0.9865,
      "step": 14190
    },
    {
      "epoch": 0.8581616002900828,
      "grad_norm": 0.6138595871903099,
      "learning_rate": 8.552568655876728e-06,
      "loss": 0.9571,
      "step": 14200
    },
    {
      "epoch": 0.8587659394452166,
      "grad_norm": 0.5891773854121215,
      "learning_rate": 8.550223953232064e-06,
      "loss": 0.9946,
      "step": 14210
    },
    {
      "epoch": 0.8593702786003505,
      "grad_norm": 0.6039962466404325,
      "learning_rate": 8.547877674991932e-06,
      "loss": 0.9823,
      "step": 14220
    },
    {
      "epoch": 0.8599746177554843,
      "grad_norm": 0.5587377829841708,
      "learning_rate": 8.545529822197615e-06,
      "loss": 0.9743,
      "step": 14230
    },
    {
      "epoch": 0.8605789569106183,
      "grad_norm": 0.6052969075414736,
      "learning_rate": 8.543180395891094e-06,
      "loss": 0.9777,
      "step": 14240
    },
    {
      "epoch": 0.8611832960657521,
      "grad_norm": 0.7745570749066291,
      "learning_rate": 8.540829397115047e-06,
      "loss": 0.9684,
      "step": 14250
    },
    {
      "epoch": 0.861787635220886,
      "grad_norm": 0.7856283138123235,
      "learning_rate": 8.538476826912854e-06,
      "loss": 0.9652,
      "step": 14260
    },
    {
      "epoch": 0.8623919743760198,
      "grad_norm": 0.7686319996115566,
      "learning_rate": 8.536122686328586e-06,
      "loss": 0.9611,
      "step": 14270
    },
    {
      "epoch": 0.8629963135311537,
      "grad_norm": 0.7320404575102034,
      "learning_rate": 8.533766976407016e-06,
      "loss": 0.9808,
      "step": 14280
    },
    {
      "epoch": 0.8636006526862875,
      "grad_norm": 0.7499438577831408,
      "learning_rate": 8.531409698193613e-06,
      "loss": 0.9588,
      "step": 14290
    },
    {
      "epoch": 0.8642049918414214,
      "grad_norm": 0.760692798217358,
      "learning_rate": 8.529050852734542e-06,
      "loss": 0.9815,
      "step": 14300
    },
    {
      "epoch": 0.8648093309965552,
      "grad_norm": 0.7612773578772055,
      "learning_rate": 8.52669044107666e-06,
      "loss": 0.9952,
      "step": 14310
    },
    {
      "epoch": 0.8654136701516891,
      "grad_norm": 0.7304535865932663,
      "learning_rate": 8.524328464267522e-06,
      "loss": 0.9758,
      "step": 14320
    },
    {
      "epoch": 0.866018009306823,
      "grad_norm": 0.7710648651713178,
      "learning_rate": 8.521964923355382e-06,
      "loss": 0.9613,
      "step": 14330
    },
    {
      "epoch": 0.8666223484619568,
      "grad_norm": 0.7772496666920414,
      "learning_rate": 8.519599819389174e-06,
      "loss": 1.012,
      "step": 14340
    },
    {
      "epoch": 0.8672266876170908,
      "grad_norm": 1.2236421890411022,
      "learning_rate": 8.517233153418547e-06,
      "loss": 0.9844,
      "step": 14350
    },
    {
      "epoch": 0.8678310267722246,
      "grad_norm": 1.1564478451423705,
      "learning_rate": 8.514864926493822e-06,
      "loss": 0.9683,
      "step": 14360
    },
    {
      "epoch": 0.8684353659273585,
      "grad_norm": 1.1754332130103715,
      "learning_rate": 8.512495139666026e-06,
      "loss": 0.9908,
      "step": 14370
    },
    {
      "epoch": 0.8690397050824923,
      "grad_norm": 1.1573009779585544,
      "learning_rate": 8.510123793986875e-06,
      "loss": 0.9863,
      "step": 14380
    },
    {
      "epoch": 0.8696440442376262,
      "grad_norm": 1.187821551420355,
      "learning_rate": 8.507750890508777e-06,
      "loss": 0.9468,
      "step": 14390
    },
    {
      "epoch": 0.87024838339276,
      "grad_norm": 0.7814586127959184,
      "learning_rate": 8.505376430284827e-06,
      "loss": 0.9609,
      "step": 14400
    },
    {
      "epoch": 0.8708527225478939,
      "grad_norm": 0.9420540458090267,
      "learning_rate": 8.503000414368816e-06,
      "loss": 0.9523,
      "step": 14410
    },
    {
      "epoch": 0.8714570617030277,
      "grad_norm": 1.0162098669137272,
      "learning_rate": 8.500622843815225e-06,
      "loss": 0.9735,
      "step": 14420
    },
    {
      "epoch": 0.8720614008581616,
      "grad_norm": 1.2102364811172532,
      "learning_rate": 8.498243719679221e-06,
      "loss": 0.9566,
      "step": 14430
    },
    {
      "epoch": 0.8726657400132954,
      "grad_norm": 0.7628880249012625,
      "learning_rate": 8.49586304301667e-06,
      "loss": 0.9707,
      "step": 14440
    },
    {
      "epoch": 0.8732700791684294,
      "grad_norm": 2.473223653843743,
      "learning_rate": 8.493480814884113e-06,
      "loss": 0.9935,
      "step": 14450
    },
    {
      "epoch": 0.8738744183235632,
      "grad_norm": 2.4960605710619737,
      "learning_rate": 8.491097036338792e-06,
      "loss": 0.9972,
      "step": 14460
    },
    {
      "epoch": 0.8744787574786971,
      "grad_norm": 2.5440214608363627,
      "learning_rate": 8.488711708438629e-06,
      "loss": 0.975,
      "step": 14470
    },
    {
      "epoch": 0.8750830966338309,
      "grad_norm": 2.427756414539397,
      "learning_rate": 8.486324832242237e-06,
      "loss": 0.9755,
      "step": 14480
    },
    {
      "epoch": 0.8756874357889648,
      "grad_norm": 2.270768583605908,
      "learning_rate": 8.483936408808917e-06,
      "loss": 0.9847,
      "step": 14490
    },
    {
      "epoch": 0.8762917749440986,
      "grad_norm": 0.9276414380370431,
      "learning_rate": 8.481546439198654e-06,
      "loss": 1.0007,
      "step": 14500
    },
    {
      "epoch": 0.8768961140992325,
      "grad_norm": 0.9322671026768691,
      "learning_rate": 8.479154924472122e-06,
      "loss": 0.9933,
      "step": 14510
    },
    {
      "epoch": 0.8775004532543663,
      "grad_norm": 0.9077576688579291,
      "learning_rate": 8.476761865690679e-06,
      "loss": 0.9487,
      "step": 14520
    },
    {
      "epoch": 0.8781047924095002,
      "grad_norm": 1.0740130002574293,
      "learning_rate": 8.474367263916367e-06,
      "loss": 0.9921,
      "step": 14530
    },
    {
      "epoch": 0.878709131564634,
      "grad_norm": 0.8862733981331156,
      "learning_rate": 8.471971120211917e-06,
      "loss": 1.0023,
      "step": 14540
    },
    {
      "epoch": 0.879313470719768,
      "grad_norm": 1.0373397333430026,
      "learning_rate": 8.46957343564074e-06,
      "loss": 0.9765,
      "step": 14550
    },
    {
      "epoch": 0.8799178098749018,
      "grad_norm": 1.037263547393421,
      "learning_rate": 8.467174211266934e-06,
      "loss": 1.0,
      "step": 14560
    },
    {
      "epoch": 0.8805221490300357,
      "grad_norm": 1.00567796094567,
      "learning_rate": 8.464773448155275e-06,
      "loss": 0.9714,
      "step": 14570
    },
    {
      "epoch": 0.8811264881851695,
      "grad_norm": 0.9418816848410064,
      "learning_rate": 8.46237114737123e-06,
      "loss": 0.9515,
      "step": 14580
    },
    {
      "epoch": 0.8817308273403034,
      "grad_norm": 0.9409106173195751,
      "learning_rate": 8.459967309980943e-06,
      "loss": 0.9732,
      "step": 14590
    },
    {
      "epoch": 0.8823351664954372,
      "grad_norm": 1.019162987597295,
      "learning_rate": 8.45756193705124e-06,
      "loss": 0.9657,
      "step": 14600
    },
    {
      "epoch": 0.8829395056505711,
      "grad_norm": 1.0700939690954803,
      "learning_rate": 8.45515502964963e-06,
      "loss": 0.9953,
      "step": 14610
    },
    {
      "epoch": 0.8835438448057049,
      "grad_norm": 1.0264694737703495,
      "learning_rate": 8.452746588844303e-06,
      "loss": 0.9909,
      "step": 14620
    },
    {
      "epoch": 0.8841481839608388,
      "grad_norm": 1.084242423671368,
      "learning_rate": 8.45033661570413e-06,
      "loss": 0.989,
      "step": 14630
    },
    {
      "epoch": 0.8847525231159726,
      "grad_norm": 1.0984857314914027,
      "learning_rate": 8.447925111298661e-06,
      "loss": 0.9851,
      "step": 14640
    },
    {
      "epoch": 0.8853568622711065,
      "grad_norm": 1.0742926234423478,
      "learning_rate": 8.445512076698125e-06,
      "loss": 0.96,
      "step": 14650
    },
    {
      "epoch": 0.8859612014262404,
      "grad_norm": 1.002027953341183,
      "learning_rate": 8.44309751297343e-06,
      "loss": 0.9951,
      "step": 14660
    },
    {
      "epoch": 0.8865655405813743,
      "grad_norm": 0.9725366946088984,
      "learning_rate": 8.440681421196166e-06,
      "loss": 0.9684,
      "step": 14670
    },
    {
      "epoch": 0.8871698797365082,
      "grad_norm": 1.0216916213642229,
      "learning_rate": 8.438263802438596e-06,
      "loss": 0.9767,
      "step": 14680
    },
    {
      "epoch": 0.887774218891642,
      "grad_norm": 0.998327880324408,
      "learning_rate": 8.435844657773666e-06,
      "loss": 0.9507,
      "step": 14690
    },
    {
      "epoch": 0.8883785580467759,
      "grad_norm": 1.0952169416767867,
      "learning_rate": 8.433423988274994e-06,
      "loss": 0.9838,
      "step": 14700
    },
    {
      "epoch": 0.8889828972019097,
      "grad_norm": 1.2251788461943667,
      "learning_rate": 8.431001795016878e-06,
      "loss": 0.9977,
      "step": 14710
    },
    {
      "epoch": 0.8895872363570436,
      "grad_norm": 1.181693250057835,
      "learning_rate": 8.428578079074294e-06,
      "loss": 0.9537,
      "step": 14720
    },
    {
      "epoch": 0.8901915755121774,
      "grad_norm": 1.0959573473565298,
      "learning_rate": 8.426152841522887e-06,
      "loss": 0.9519,
      "step": 14730
    },
    {
      "epoch": 0.8907959146673113,
      "grad_norm": 1.001507992246485,
      "learning_rate": 8.423726083438986e-06,
      "loss": 0.9588,
      "step": 14740
    },
    {
      "epoch": 0.8914002538224451,
      "grad_norm": 0.9517339318631465,
      "learning_rate": 8.421297805899586e-06,
      "loss": 0.9931,
      "step": 14750
    },
    {
      "epoch": 0.8920045929775791,
      "grad_norm": 0.9001871773562664,
      "learning_rate": 8.418868009982364e-06,
      "loss": 0.988,
      "step": 14760
    },
    {
      "epoch": 0.8926089321327129,
      "grad_norm": 0.9356960550403014,
      "learning_rate": 8.416436696765667e-06,
      "loss": 0.9927,
      "step": 14770
    },
    {
      "epoch": 0.8932132712878468,
      "grad_norm": 0.9811393595806533,
      "learning_rate": 8.414003867328515e-06,
      "loss": 1.0087,
      "step": 14780
    },
    {
      "epoch": 0.8938176104429806,
      "grad_norm": 0.9649555352069363,
      "learning_rate": 8.411569522750603e-06,
      "loss": 0.9988,
      "step": 14790
    },
    {
      "epoch": 0.8944219495981145,
      "grad_norm": 0.9646987723282942,
      "learning_rate": 8.409133664112295e-06,
      "loss": 0.9584,
      "step": 14800
    },
    {
      "epoch": 0.8950262887532483,
      "grad_norm": 0.9961419680305934,
      "learning_rate": 8.406696292494632e-06,
      "loss": 0.9834,
      "step": 14810
    },
    {
      "epoch": 0.8956306279083822,
      "grad_norm": 1.0542166242440507,
      "learning_rate": 8.404257408979322e-06,
      "loss": 0.977,
      "step": 14820
    },
    {
      "epoch": 0.896234967063516,
      "grad_norm": 1.0533188303783554,
      "learning_rate": 8.401817014648748e-06,
      "loss": 0.9689,
      "step": 14830
    },
    {
      "epoch": 0.8968393062186499,
      "grad_norm": 1.0131002601556658,
      "learning_rate": 8.399375110585956e-06,
      "loss": 0.9674,
      "step": 14840
    },
    {
      "epoch": 0.8974436453737837,
      "grad_norm": 0.9709924962791653,
      "learning_rate": 8.396931697874673e-06,
      "loss": 0.9633,
      "step": 14850
    },
    {
      "epoch": 0.8980479845289177,
      "grad_norm": 0.936565669426908,
      "learning_rate": 8.394486777599287e-06,
      "loss": 0.9866,
      "step": 14860
    },
    {
      "epoch": 0.8986523236840515,
      "grad_norm": 0.962284143408487,
      "learning_rate": 8.392040350844858e-06,
      "loss": 0.9735,
      "step": 14870
    },
    {
      "epoch": 0.8992566628391854,
      "grad_norm": 0.9946695017782787,
      "learning_rate": 8.389592418697117e-06,
      "loss": 0.9786,
      "step": 14880
    },
    {
      "epoch": 0.8998610019943192,
      "grad_norm": 0.9691296166065627,
      "learning_rate": 8.387142982242455e-06,
      "loss": 0.9734,
      "step": 14890
    },
    {
      "epoch": 0.9004653411494531,
      "grad_norm": 1.1259013990647193,
      "learning_rate": 8.384692042567942e-06,
      "loss": 0.9981,
      "step": 14900
    },
    {
      "epoch": 0.9010696803045869,
      "grad_norm": 1.1465770908465234,
      "learning_rate": 8.382239600761302e-06,
      "loss": 0.9617,
      "step": 14910
    },
    {
      "epoch": 0.9016740194597208,
      "grad_norm": 1.1654058563708993,
      "learning_rate": 8.379785657910939e-06,
      "loss": 0.9988,
      "step": 14920
    },
    {
      "epoch": 0.9022783586148546,
      "grad_norm": 1.0990433646220246,
      "learning_rate": 8.377330215105917e-06,
      "loss": 0.9676,
      "step": 14930
    },
    {
      "epoch": 0.9028826977699885,
      "grad_norm": 1.0878428754781606,
      "learning_rate": 8.374873273435964e-06,
      "loss": 0.9787,
      "step": 14940
    },
    {
      "epoch": 0.9034870369251223,
      "grad_norm": 1.0830366501084334,
      "learning_rate": 8.372414833991472e-06,
      "loss": 0.9835,
      "step": 14950
    },
    {
      "epoch": 0.9040913760802562,
      "grad_norm": 1.0228113040934184,
      "learning_rate": 8.369954897863507e-06,
      "loss": 0.999,
      "step": 14960
    },
    {
      "epoch": 0.90469571523539,
      "grad_norm": 0.9791965032152091,
      "learning_rate": 8.367493466143786e-06,
      "loss": 0.9921,
      "step": 14970
    },
    {
      "epoch": 0.905300054390524,
      "grad_norm": 1.083387497785294,
      "learning_rate": 8.365030539924704e-06,
      "loss": 0.994,
      "step": 14980
    },
    {
      "epoch": 0.9059043935456578,
      "grad_norm": 0.9877531161131822,
      "learning_rate": 8.362566120299304e-06,
      "loss": 0.9917,
      "step": 14990
    },
    {
      "epoch": 0.9065087327007917,
      "grad_norm": 1.2101812681075603,
      "learning_rate": 8.360100208361306e-06,
      "loss": 0.994,
      "step": 15000
    },
    {
      "epoch": 0.9071130718559255,
      "grad_norm": 1.1916898119621775,
      "learning_rate": 8.357632805205083e-06,
      "loss": 0.9745,
      "step": 15010
    },
    {
      "epoch": 0.9077174110110594,
      "grad_norm": 1.1528099146691606,
      "learning_rate": 8.355163911925672e-06,
      "loss": 0.9811,
      "step": 15020
    },
    {
      "epoch": 0.9083217501661933,
      "grad_norm": 1.1868930183977866,
      "learning_rate": 8.352693529618773e-06,
      "loss": 1.0003,
      "step": 15030
    },
    {
      "epoch": 0.9089260893213271,
      "grad_norm": 1.1441811008731064,
      "learning_rate": 8.350221659380749e-06,
      "loss": 0.9765,
      "step": 15040
    },
    {
      "epoch": 0.909530428476461,
      "grad_norm": 1.2854872430282722,
      "learning_rate": 8.347748302308616e-06,
      "loss": 0.9679,
      "step": 15050
    },
    {
      "epoch": 0.9101347676315948,
      "grad_norm": 1.305410285187446,
      "learning_rate": 8.345273459500055e-06,
      "loss": 0.9696,
      "step": 15060
    },
    {
      "epoch": 0.9107391067867288,
      "grad_norm": 1.4112752533252417,
      "learning_rate": 8.342797132053406e-06,
      "loss": 0.981,
      "step": 15070
    },
    {
      "epoch": 0.9113434459418626,
      "grad_norm": 1.3399073634525698,
      "learning_rate": 8.340319321067668e-06,
      "loss": 0.9652,
      "step": 15080
    },
    {
      "epoch": 0.9119477850969965,
      "grad_norm": 1.3063670610413907,
      "learning_rate": 8.337840027642496e-06,
      "loss": 0.9695,
      "step": 15090
    },
    {
      "epoch": 0.9125521242521303,
      "grad_norm": 0.8581451886576797,
      "learning_rate": 8.335359252878208e-06,
      "loss": 0.9829,
      "step": 15100
    },
    {
      "epoch": 0.9131564634072642,
      "grad_norm": 0.7855715403061322,
      "learning_rate": 8.332876997875772e-06,
      "loss": 0.9806,
      "step": 15110
    },
    {
      "epoch": 0.913760802562398,
      "grad_norm": 0.7467948355223356,
      "learning_rate": 8.33039326373682e-06,
      "loss": 0.965,
      "step": 15120
    },
    {
      "epoch": 0.9143651417175319,
      "grad_norm": 0.7382353306044588,
      "learning_rate": 8.327908051563636e-06,
      "loss": 0.9748,
      "step": 15130
    },
    {
      "epoch": 0.9149694808726657,
      "grad_norm": 0.8474447213633335,
      "learning_rate": 8.325421362459161e-06,
      "loss": 0.9932,
      "step": 15140
    },
    {
      "epoch": 0.9155738200277996,
      "grad_norm": 0.7317917032082506,
      "learning_rate": 8.322933197526994e-06,
      "loss": 0.9835,
      "step": 15150
    },
    {
      "epoch": 0.9161781591829334,
      "grad_norm": 0.7530942217530358,
      "learning_rate": 8.320443557871385e-06,
      "loss": 0.9829,
      "step": 15160
    },
    {
      "epoch": 0.9167824983380674,
      "grad_norm": 0.7154828727948112,
      "learning_rate": 8.31795244459724e-06,
      "loss": 0.9773,
      "step": 15170
    },
    {
      "epoch": 0.9173868374932012,
      "grad_norm": 0.7637348413098434,
      "learning_rate": 8.31545985881012e-06,
      "loss": 0.9862,
      "step": 15180
    },
    {
      "epoch": 0.9179911766483351,
      "grad_norm": 0.7479647119333702,
      "learning_rate": 8.312965801616239e-06,
      "loss": 0.9525,
      "step": 15190
    },
    {
      "epoch": 0.9185955158034689,
      "grad_norm": 0.7942328833650443,
      "learning_rate": 8.310470274122466e-06,
      "loss": 0.9609,
      "step": 15200
    },
    {
      "epoch": 0.9191998549586028,
      "grad_norm": 0.805313807375061,
      "learning_rate": 8.307973277436316e-06,
      "loss": 0.9721,
      "step": 15210
    },
    {
      "epoch": 0.9198041941137366,
      "grad_norm": 0.8151444437111519,
      "learning_rate": 8.305474812665964e-06,
      "loss": 0.9808,
      "step": 15220
    },
    {
      "epoch": 0.9204085332688705,
      "grad_norm": 0.7739572996198226,
      "learning_rate": 8.302974880920232e-06,
      "loss": 0.9535,
      "step": 15230
    },
    {
      "epoch": 0.9210128724240043,
      "grad_norm": 0.7971473423220925,
      "learning_rate": 8.300473483308593e-06,
      "loss": 0.9848,
      "step": 15240
    },
    {
      "epoch": 0.9216172115791382,
      "grad_norm": 0.5593937608199296,
      "learning_rate": 8.297970620941172e-06,
      "loss": 0.9633,
      "step": 15250
    },
    {
      "epoch": 0.922221550734272,
      "grad_norm": 0.5615530705752211,
      "learning_rate": 8.295466294928742e-06,
      "loss": 0.9906,
      "step": 15260
    },
    {
      "epoch": 0.922825889889406,
      "grad_norm": 0.5590789840513106,
      "learning_rate": 8.292960506382732e-06,
      "loss": 0.9785,
      "step": 15270
    },
    {
      "epoch": 0.9234302290445398,
      "grad_norm": 0.5291157134336621,
      "learning_rate": 8.290453256415209e-06,
      "loss": 0.9562,
      "step": 15280
    },
    {
      "epoch": 0.9240345681996737,
      "grad_norm": 0.5409264386781765,
      "learning_rate": 8.287944546138902e-06,
      "loss": 0.9809,
      "step": 15290
    },
    {
      "epoch": 0.9246389073548075,
      "grad_norm": 0.5968634586693193,
      "learning_rate": 8.285434376667173e-06,
      "loss": 0.9751,
      "step": 15300
    },
    {
      "epoch": 0.9252432465099414,
      "grad_norm": 0.6052672579343464,
      "learning_rate": 8.282922749114045e-06,
      "loss": 0.9704,
      "step": 15310
    },
    {
      "epoch": 0.9258475856650752,
      "grad_norm": 0.5767452870876945,
      "learning_rate": 8.280409664594182e-06,
      "loss": 0.9871,
      "step": 15320
    },
    {
      "epoch": 0.9264519248202091,
      "grad_norm": 0.5943979835389582,
      "learning_rate": 8.277895124222893e-06,
      "loss": 0.9507,
      "step": 15330
    },
    {
      "epoch": 0.9270562639753429,
      "grad_norm": 0.5826660040013454,
      "learning_rate": 8.275379129116137e-06,
      "loss": 0.9594,
      "step": 15340
    },
    {
      "epoch": 0.9276606031304768,
      "grad_norm": 0.9398837479641462,
      "learning_rate": 8.272861680390516e-06,
      "loss": 0.9773,
      "step": 15350
    },
    {
      "epoch": 0.9282649422856107,
      "grad_norm": 0.8369343658449666,
      "learning_rate": 8.270342779163278e-06,
      "loss": 0.9626,
      "step": 15360
    },
    {
      "epoch": 0.9288692814407445,
      "grad_norm": 0.7929105923486776,
      "learning_rate": 8.267822426552319e-06,
      "loss": 0.951,
      "step": 15370
    },
    {
      "epoch": 0.9294736205958785,
      "grad_norm": 0.7720117715997297,
      "learning_rate": 8.265300623676168e-06,
      "loss": 0.9613,
      "step": 15380
    },
    {
      "epoch": 0.9300779597510123,
      "grad_norm": 0.737724355421029,
      "learning_rate": 8.262777371654015e-06,
      "loss": 0.9825,
      "step": 15390
    },
    {
      "epoch": 0.9306822989061462,
      "grad_norm": 0.4792498970466182,
      "learning_rate": 8.260252671605678e-06,
      "loss": 0.9938,
      "step": 15400
    },
    {
      "epoch": 0.93128663806128,
      "grad_norm": 0.49639666083524897,
      "learning_rate": 8.257726524651623e-06,
      "loss": 0.989,
      "step": 15410
    },
    {
      "epoch": 0.9318909772164139,
      "grad_norm": 0.4879334070101319,
      "learning_rate": 8.25519893191296e-06,
      "loss": 0.9862,
      "step": 15420
    },
    {
      "epoch": 0.9324953163715477,
      "grad_norm": 0.5236490326331864,
      "learning_rate": 8.25266989451144e-06,
      "loss": 0.9712,
      "step": 15430
    },
    {
      "epoch": 0.9330996555266816,
      "grad_norm": 0.49138545746913104,
      "learning_rate": 8.250139413569452e-06,
      "loss": 0.9815,
      "step": 15440
    },
    {
      "epoch": 0.9337039946818154,
      "grad_norm": 0.4874412389398828,
      "learning_rate": 8.247607490210026e-06,
      "loss": 0.9818,
      "step": 15450
    },
    {
      "epoch": 0.9343083338369493,
      "grad_norm": 0.5483962683415574,
      "learning_rate": 8.24507412555684e-06,
      "loss": 0.9542,
      "step": 15460
    },
    {
      "epoch": 0.9349126729920831,
      "grad_norm": 0.45861596009021244,
      "learning_rate": 8.2425393207342e-06,
      "loss": 0.9741,
      "step": 15470
    },
    {
      "epoch": 0.935517012147217,
      "grad_norm": 0.48667606147713494,
      "learning_rate": 8.240003076867058e-06,
      "loss": 0.9665,
      "step": 15480
    },
    {
      "epoch": 0.9361213513023509,
      "grad_norm": 0.4862374408145362,
      "learning_rate": 8.237465395081006e-06,
      "loss": 0.9853,
      "step": 15490
    },
    {
      "epoch": 0.9367256904574848,
      "grad_norm": 0.49319242139667385,
      "learning_rate": 8.234926276502267e-06,
      "loss": 1.0055,
      "step": 15500
    },
    {
      "epoch": 0.9373300296126186,
      "grad_norm": 0.48575364669378046,
      "learning_rate": 8.232385722257711e-06,
      "loss": 0.9786,
      "step": 15510
    },
    {
      "epoch": 0.9379343687677525,
      "grad_norm": 0.5064431072585477,
      "learning_rate": 8.229843733474838e-06,
      "loss": 0.9661,
      "step": 15520
    },
    {
      "epoch": 0.9385387079228863,
      "grad_norm": 0.5210995718594443,
      "learning_rate": 8.227300311281786e-06,
      "loss": 0.9867,
      "step": 15530
    },
    {
      "epoch": 0.9391430470780202,
      "grad_norm": 0.520543236604751,
      "learning_rate": 8.224755456807331e-06,
      "loss": 0.9645,
      "step": 15540
    },
    {
      "epoch": 0.939747386233154,
      "grad_norm": 0.510032001989098,
      "learning_rate": 8.222209171180883e-06,
      "loss": 0.9837,
      "step": 15550
    },
    {
      "epoch": 0.9403517253882879,
      "grad_norm": 0.5066937166384391,
      "learning_rate": 8.21966145553249e-06,
      "loss": 0.9769,
      "step": 15560
    },
    {
      "epoch": 0.9409560645434217,
      "grad_norm": 0.47529590926820436,
      "learning_rate": 8.217112310992831e-06,
      "loss": 0.964,
      "step": 15570
    },
    {
      "epoch": 0.9415604036985556,
      "grad_norm": 0.5110087145423996,
      "learning_rate": 8.214561738693223e-06,
      "loss": 1.0113,
      "step": 15580
    },
    {
      "epoch": 0.9421647428536895,
      "grad_norm": 0.48570477342435137,
      "learning_rate": 8.212009739765611e-06,
      "loss": 0.9636,
      "step": 15590
    },
    {
      "epoch": 0.9427690820088234,
      "grad_norm": 0.49986980981019213,
      "learning_rate": 8.209456315342578e-06,
      "loss": 0.9752,
      "step": 15600
    },
    {
      "epoch": 0.9433734211639572,
      "grad_norm": 0.4982489644343123,
      "learning_rate": 8.206901466557339e-06,
      "loss": 0.9833,
      "step": 15610
    },
    {
      "epoch": 0.9439777603190911,
      "grad_norm": 0.4925601414432979,
      "learning_rate": 8.204345194543737e-06,
      "loss": 0.9526,
      "step": 15620
    },
    {
      "epoch": 0.9445820994742249,
      "grad_norm": 0.502427741366534,
      "learning_rate": 8.201787500436255e-06,
      "loss": 0.9791,
      "step": 15630
    },
    {
      "epoch": 0.9451864386293588,
      "grad_norm": 0.4837334195618305,
      "learning_rate": 8.199228385369997e-06,
      "loss": 0.9782,
      "step": 15640
    },
    {
      "epoch": 0.9457907777844926,
      "grad_norm": 0.5118359169968741,
      "learning_rate": 8.196667850480705e-06,
      "loss": 0.9898,
      "step": 15650
    },
    {
      "epoch": 0.9463951169396265,
      "grad_norm": 0.5063133883805043,
      "learning_rate": 8.194105896904752e-06,
      "loss": 0.9777,
      "step": 15660
    },
    {
      "epoch": 0.9469994560947603,
      "grad_norm": 0.4863301039086336,
      "learning_rate": 8.19154252577913e-06,
      "loss": 0.9626,
      "step": 15670
    },
    {
      "epoch": 0.9476037952498942,
      "grad_norm": 0.6063423950239919,
      "learning_rate": 8.188977738241475e-06,
      "loss": 0.9454,
      "step": 15680
    },
    {
      "epoch": 0.948208134405028,
      "grad_norm": 0.5343329334189985,
      "learning_rate": 8.18641153543004e-06,
      "loss": 0.967,
      "step": 15690
    },
    {
      "epoch": 0.948812473560162,
      "grad_norm": 0.555430421579258,
      "learning_rate": 8.183843918483707e-06,
      "loss": 1.0006,
      "step": 15700
    },
    {
      "epoch": 0.9494168127152959,
      "grad_norm": 0.6011406882443133,
      "learning_rate": 8.181274888541996e-06,
      "loss": 0.9739,
      "step": 15710
    },
    {
      "epoch": 0.9500211518704297,
      "grad_norm": 0.6132581360553538,
      "learning_rate": 8.178704446745042e-06,
      "loss": 1.0113,
      "step": 15720
    },
    {
      "epoch": 0.9506254910255636,
      "grad_norm": 0.5644882398869236,
      "learning_rate": 8.176132594233615e-06,
      "loss": 0.9852,
      "step": 15730
    },
    {
      "epoch": 0.9512298301806974,
      "grad_norm": 0.5631525429681971,
      "learning_rate": 8.173559332149105e-06,
      "loss": 0.9667,
      "step": 15740
    },
    {
      "epoch": 0.9518341693358313,
      "grad_norm": 0.7062211408913227,
      "learning_rate": 8.17098466163353e-06,
      "loss": 0.9962,
      "step": 15750
    },
    {
      "epoch": 0.9524385084909651,
      "grad_norm": 0.6601969274334052,
      "learning_rate": 8.168408583829534e-06,
      "loss": 0.9723,
      "step": 15760
    },
    {
      "epoch": 0.953042847646099,
      "grad_norm": 0.6670435824440197,
      "learning_rate": 8.165831099880384e-06,
      "loss": 0.9648,
      "step": 15770
    },
    {
      "epoch": 0.9536471868012328,
      "grad_norm": 0.6619443844641391,
      "learning_rate": 8.163252210929973e-06,
      "loss": 0.9806,
      "step": 15780
    },
    {
      "epoch": 0.9542515259563668,
      "grad_norm": 0.6414183110946925,
      "learning_rate": 8.160671918122815e-06,
      "loss": 0.973,
      "step": 15790
    },
    {
      "epoch": 0.9548558651115006,
      "grad_norm": 0.5766717035562999,
      "learning_rate": 8.158090222604052e-06,
      "loss": 0.9957,
      "step": 15800
    },
    {
      "epoch": 0.9554602042666345,
      "grad_norm": 0.5956217786900823,
      "learning_rate": 8.155507125519438e-06,
      "loss": 0.9618,
      "step": 15810
    },
    {
      "epoch": 0.9560645434217683,
      "grad_norm": 0.5895176817887027,
      "learning_rate": 8.152922628015363e-06,
      "loss": 0.964,
      "step": 15820
    },
    {
      "epoch": 0.9566688825769022,
      "grad_norm": 0.6001932027160904,
      "learning_rate": 8.150336731238826e-06,
      "loss": 0.9883,
      "step": 15830
    },
    {
      "epoch": 0.957273221732036,
      "grad_norm": 0.5776169600082943,
      "learning_rate": 8.147749436337455e-06,
      "loss": 0.9806,
      "step": 15840
    },
    {
      "epoch": 0.9578775608871699,
      "grad_norm": 0.7670392590266749,
      "learning_rate": 8.145160744459497e-06,
      "loss": 0.9665,
      "step": 15850
    },
    {
      "epoch": 0.9584819000423037,
      "grad_norm": 0.7478739127084156,
      "learning_rate": 8.142570656753816e-06,
      "loss": 0.9838,
      "step": 15860
    },
    {
      "epoch": 0.9590862391974376,
      "grad_norm": 0.7515754514708176,
      "learning_rate": 8.139979174369898e-06,
      "loss": 0.9662,
      "step": 15870
    },
    {
      "epoch": 0.9596905783525714,
      "grad_norm": 0.791726205386205,
      "learning_rate": 8.137386298457847e-06,
      "loss": 0.9794,
      "step": 15880
    },
    {
      "epoch": 0.9602949175077053,
      "grad_norm": 0.7376190131207087,
      "learning_rate": 8.134792030168386e-06,
      "loss": 0.9614,
      "step": 15890
    },
    {
      "epoch": 0.9608992566628392,
      "grad_norm": 0.7986479819326139,
      "learning_rate": 8.132196370652855e-06,
      "loss": 1.0005,
      "step": 15900
    },
    {
      "epoch": 0.9615035958179731,
      "grad_norm": 0.7598894944051914,
      "learning_rate": 8.129599321063214e-06,
      "loss": 0.9723,
      "step": 15910
    },
    {
      "epoch": 0.9621079349731069,
      "grad_norm": 0.7661687428972335,
      "learning_rate": 8.127000882552036e-06,
      "loss": 0.971,
      "step": 15920
    },
    {
      "epoch": 0.9627122741282408,
      "grad_norm": 0.7380062715591414,
      "learning_rate": 8.124401056272513e-06,
      "loss": 0.9664,
      "step": 15930
    },
    {
      "epoch": 0.9633166132833746,
      "grad_norm": 0.7148929039572601,
      "learning_rate": 8.121799843378452e-06,
      "loss": 0.9599,
      "step": 15940
    },
    {
      "epoch": 0.9639209524385085,
      "grad_norm": 1.1548965140777823,
      "learning_rate": 8.119197245024279e-06,
      "loss": 0.9704,
      "step": 15950
    },
    {
      "epoch": 0.9645252915936423,
      "grad_norm": 1.1355189202619795,
      "learning_rate": 8.116593262365027e-06,
      "loss": 0.9514,
      "step": 15960
    },
    {
      "epoch": 0.9651296307487762,
      "grad_norm": 1.1744731134148332,
      "learning_rate": 8.11398789655635e-06,
      "loss": 0.961,
      "step": 15970
    },
    {
      "epoch": 0.96573396990391,
      "grad_norm": 1.2228545136055289,
      "learning_rate": 8.111381148754515e-06,
      "loss": 0.9806,
      "step": 15980
    },
    {
      "epoch": 0.9663383090590439,
      "grad_norm": 1.1663052920384902,
      "learning_rate": 8.108773020116398e-06,
      "loss": 0.9583,
      "step": 15990
    },
    {
      "epoch": 0.9669426482141777,
      "grad_norm": 0.9580314185195974,
      "learning_rate": 8.106163511799495e-06,
      "loss": 0.9837,
      "step": 16000
    },
    {
      "epoch": 0.9675469873693117,
      "grad_norm": 0.9008400535684915,
      "learning_rate": 8.103552624961907e-06,
      "loss": 0.9615,
      "step": 16010
    },
    {
      "epoch": 0.9681513265244455,
      "grad_norm": 1.240721520907658,
      "learning_rate": 8.100940360762352e-06,
      "loss": 0.9666,
      "step": 16020
    },
    {
      "epoch": 0.9687556656795794,
      "grad_norm": 0.6993395308165153,
      "learning_rate": 8.098326720360155e-06,
      "loss": 0.9863,
      "step": 16030
    },
    {
      "epoch": 0.9693600048347133,
      "grad_norm": 1.1569674427667267,
      "learning_rate": 8.095711704915257e-06,
      "loss": 0.9613,
      "step": 16040
    },
    {
      "epoch": 0.9699643439898471,
      "grad_norm": 2.5312024947088116,
      "learning_rate": 8.093095315588205e-06,
      "loss": 0.9975,
      "step": 16050
    },
    {
      "epoch": 0.970568683144981,
      "grad_norm": 2.6365482277837304,
      "learning_rate": 8.090477553540155e-06,
      "loss": 0.9748,
      "step": 16060
    },
    {
      "epoch": 0.9711730223001148,
      "grad_norm": 2.306362014830411,
      "learning_rate": 8.087858419932876e-06,
      "loss": 1.0037,
      "step": 16070
    },
    {
      "epoch": 0.9717773614552487,
      "grad_norm": 2.633887786204463,
      "learning_rate": 8.085237915928747e-06,
      "loss": 0.9898,
      "step": 16080
    },
    {
      "epoch": 0.9723817006103825,
      "grad_norm": 2.2361695271247406,
      "learning_rate": 8.08261604269075e-06,
      "loss": 0.9712,
      "step": 16090
    },
    {
      "epoch": 0.9729860397655165,
      "grad_norm": 0.9027673205533038,
      "learning_rate": 8.079992801382473e-06,
      "loss": 0.9864,
      "step": 16100
    },
    {
      "epoch": 0.9735903789206503,
      "grad_norm": 0.9485300360821619,
      "learning_rate": 8.07736819316812e-06,
      "loss": 0.9863,
      "step": 16110
    },
    {
      "epoch": 0.9741947180757842,
      "grad_norm": 0.8996714752716645,
      "learning_rate": 8.074742219212495e-06,
      "loss": 0.9629,
      "step": 16120
    },
    {
      "epoch": 0.974799057230918,
      "grad_norm": 1.0196912605373736,
      "learning_rate": 8.072114880681008e-06,
      "loss": 0.9647,
      "step": 16130
    },
    {
      "epoch": 0.9754033963860519,
      "grad_norm": 0.8731143251895254,
      "learning_rate": 8.069486178739677e-06,
      "loss": 0.9737,
      "step": 16140
    },
    {
      "epoch": 0.9760077355411857,
      "grad_norm": 0.9399427677200887,
      "learning_rate": 8.066856114555128e-06,
      "loss": 0.9727,
      "step": 16150
    },
    {
      "epoch": 0.9766120746963196,
      "grad_norm": 0.9481263278394423,
      "learning_rate": 8.064224689294584e-06,
      "loss": 1.0035,
      "step": 16160
    },
    {
      "epoch": 0.9772164138514534,
      "grad_norm": 0.9791490851971432,
      "learning_rate": 8.061591904125878e-06,
      "loss": 0.9685,
      "step": 16170
    },
    {
      "epoch": 0.9778207530065873,
      "grad_norm": 0.9356403527625788,
      "learning_rate": 8.058957760217445e-06,
      "loss": 0.9725,
      "step": 16180
    },
    {
      "epoch": 0.9784250921617211,
      "grad_norm": 0.9882829100865731,
      "learning_rate": 8.05632225873832e-06,
      "loss": 0.9703,
      "step": 16190
    },
    {
      "epoch": 0.979029431316855,
      "grad_norm": 1.1818364373883143,
      "learning_rate": 8.053685400858145e-06,
      "loss": 0.9428,
      "step": 16200
    },
    {
      "epoch": 0.9796337704719889,
      "grad_norm": 1.0711103626098628,
      "learning_rate": 8.05104718774716e-06,
      "loss": 0.9872,
      "step": 16210
    },
    {
      "epoch": 0.9802381096271228,
      "grad_norm": 1.0681541962507033,
      "learning_rate": 8.048407620576214e-06,
      "loss": 0.965,
      "step": 16220
    },
    {
      "epoch": 0.9808424487822566,
      "grad_norm": 1.0982852366265872,
      "learning_rate": 8.045766700516744e-06,
      "loss": 0.9889,
      "step": 16230
    },
    {
      "epoch": 0.9814467879373905,
      "grad_norm": 1.1195015634134953,
      "learning_rate": 8.043124428740798e-06,
      "loss": 0.9883,
      "step": 16240
    },
    {
      "epoch": 0.9820511270925243,
      "grad_norm": 1.0497449594996697,
      "learning_rate": 8.040480806421023e-06,
      "loss": 1.0138,
      "step": 16250
    },
    {
      "epoch": 0.9826554662476582,
      "grad_norm": 1.0369912334742362,
      "learning_rate": 8.037835834730661e-06,
      "loss": 0.9835,
      "step": 16260
    },
    {
      "epoch": 0.983259805402792,
      "grad_norm": 0.9974874217628756,
      "learning_rate": 8.035189514843553e-06,
      "loss": 0.9568,
      "step": 16270
    },
    {
      "epoch": 0.9838641445579259,
      "grad_norm": 1.071505687372954,
      "learning_rate": 8.032541847934145e-06,
      "loss": 0.963,
      "step": 16280
    },
    {
      "epoch": 0.9844684837130597,
      "grad_norm": 1.006831916981404,
      "learning_rate": 8.029892835177473e-06,
      "loss": 0.9906,
      "step": 16290
    },
    {
      "epoch": 0.9850728228681936,
      "grad_norm": 1.0728103298911376,
      "learning_rate": 8.027242477749172e-06,
      "loss": 0.9765,
      "step": 16300
    },
    {
      "epoch": 0.9856771620233274,
      "grad_norm": 1.1279953282506783,
      "learning_rate": 8.02459077682548e-06,
      "loss": 1.0026,
      "step": 16310
    },
    {
      "epoch": 0.9862815011784614,
      "grad_norm": 1.1377071729682078,
      "learning_rate": 8.021937733583223e-06,
      "loss": 0.9762,
      "step": 16320
    },
    {
      "epoch": 0.9868858403335952,
      "grad_norm": 1.1023832208894342,
      "learning_rate": 8.019283349199827e-06,
      "loss": 0.9777,
      "step": 16330
    },
    {
      "epoch": 0.9874901794887291,
      "grad_norm": 1.1830416503673535,
      "learning_rate": 8.016627624853311e-06,
      "loss": 0.9582,
      "step": 16340
    },
    {
      "epoch": 0.9880945186438629,
      "grad_norm": 0.9270252595130956,
      "learning_rate": 8.01397056172229e-06,
      "loss": 0.9931,
      "step": 16350
    },
    {
      "epoch": 0.9886988577989968,
      "grad_norm": 0.917664965218076,
      "learning_rate": 8.011312160985976e-06,
      "loss": 0.9845,
      "step": 16360
    },
    {
      "epoch": 0.9893031969541306,
      "grad_norm": 1.0182963867610455,
      "learning_rate": 8.00865242382417e-06,
      "loss": 0.9831,
      "step": 16370
    },
    {
      "epoch": 0.9899075361092645,
      "grad_norm": 0.9677862270377444,
      "learning_rate": 8.005991351417266e-06,
      "loss": 0.9783,
      "step": 16380
    },
    {
      "epoch": 0.9905118752643984,
      "grad_norm": 0.9756658582285264,
      "learning_rate": 8.003328944946256e-06,
      "loss": 0.98,
      "step": 16390
    },
    {
      "epoch": 0.9911162144195322,
      "grad_norm": 0.9898636351662876,
      "learning_rate": 8.00066520559272e-06,
      "loss": 0.9975,
      "step": 16400
    },
    {
      "epoch": 0.9917205535746662,
      "grad_norm": 1.1825132286439701,
      "learning_rate": 7.998000134538826e-06,
      "loss": 0.9757,
      "step": 16410
    },
    {
      "epoch": 0.9923248927298,
      "grad_norm": 1.0438803315668894,
      "learning_rate": 7.995333732967342e-06,
      "loss": 0.995,
      "step": 16420
    },
    {
      "epoch": 0.9929292318849339,
      "grad_norm": 1.0611379895197603,
      "learning_rate": 7.99266600206162e-06,
      "loss": 0.9482,
      "step": 16430
    },
    {
      "epoch": 0.9935335710400677,
      "grad_norm": 0.968777919224612,
      "learning_rate": 7.989996943005602e-06,
      "loss": 0.9752,
      "step": 16440
    },
    {
      "epoch": 0.9941379101952016,
      "grad_norm": 1.0257813939329161,
      "learning_rate": 7.987326556983823e-06,
      "loss": 0.9553,
      "step": 16450
    },
    {
      "epoch": 0.9947422493503354,
      "grad_norm": 1.0253543776364944,
      "learning_rate": 7.984654845181404e-06,
      "loss": 0.9808,
      "step": 16460
    },
    {
      "epoch": 0.9953465885054693,
      "grad_norm": 0.9311508087482454,
      "learning_rate": 7.981981808784054e-06,
      "loss": 1.0002,
      "step": 16470
    },
    {
      "epoch": 0.9959509276606031,
      "grad_norm": 0.8827879448771927,
      "learning_rate": 7.979307448978075e-06,
      "loss": 0.9839,
      "step": 16480
    },
    {
      "epoch": 0.996555266815737,
      "grad_norm": 0.8781862377620872,
      "learning_rate": 7.976631766950348e-06,
      "loss": 0.9873,
      "step": 16490
    },
    {
      "epoch": 0.9971596059708708,
      "grad_norm": 1.117524151394854,
      "learning_rate": 7.973954763888347e-06,
      "loss": 0.9604,
      "step": 16500
    },
    {
      "epoch": 0.9977639451260047,
      "grad_norm": 1.1513608596965998,
      "learning_rate": 7.971276440980133e-06,
      "loss": 0.9668,
      "step": 16510
    },
    {
      "epoch": 0.9983682842811386,
      "grad_norm": 1.1082309067546794,
      "learning_rate": 7.968596799414347e-06,
      "loss": 0.9759,
      "step": 16520
    },
    {
      "epoch": 0.9989726234362725,
      "grad_norm": 1.116730405972857,
      "learning_rate": 7.965915840380219e-06,
      "loss": 1.0019,
      "step": 16530
    },
    {
      "epoch": 0.9995769625914063,
      "grad_norm": 1.097591218971702,
      "learning_rate": 7.963233565067565e-06,
      "loss": 0.9832,
      "step": 16540
    },
    {
      "epoch": 1.0001813017465402,
      "grad_norm": 1.0222076559815056,
      "learning_rate": 7.96054997466678e-06,
      "loss": 0.9656,
      "step": 16550
    },
    {
      "epoch": 1.000785640901674,
      "grad_norm": 1.1324279892713336,
      "learning_rate": 7.95786507036885e-06,
      "loss": 0.9392,
      "step": 16560
    },
    {
      "epoch": 1.0013899800568078,
      "grad_norm": 1.1583391945287789,
      "learning_rate": 7.955178853365338e-06,
      "loss": 0.9357,
      "step": 16570
    },
    {
      "epoch": 1.0019943192119418,
      "grad_norm": 1.0553472594753273,
      "learning_rate": 7.952491324848392e-06,
      "loss": 0.9361,
      "step": 16580
    },
    {
      "epoch": 1.0025986583670756,
      "grad_norm": 0.9992459487305569,
      "learning_rate": 7.94980248601074e-06,
      "loss": 0.9129,
      "step": 16590
    },
    {
      "epoch": 1.0032029975222094,
      "grad_norm": 1.1827744975887033,
      "learning_rate": 7.947112338045698e-06,
      "loss": 0.9187,
      "step": 16600
    },
    {
      "epoch": 1.0038073366773432,
      "grad_norm": 1.106611197017621,
      "learning_rate": 7.944420882147153e-06,
      "loss": 0.9344,
      "step": 16610
    },
    {
      "epoch": 1.0044116758324773,
      "grad_norm": 1.215207228383985,
      "learning_rate": 7.941728119509582e-06,
      "loss": 0.9387,
      "step": 16620
    },
    {
      "epoch": 1.005016014987611,
      "grad_norm": 1.1347656254059526,
      "learning_rate": 7.939034051328034e-06,
      "loss": 0.9402,
      "step": 16630
    },
    {
      "epoch": 1.0056203541427449,
      "grad_norm": 1.2085534661289723,
      "learning_rate": 7.936338678798143e-06,
      "loss": 0.9462,
      "step": 16640
    },
    {
      "epoch": 1.0062246932978787,
      "grad_norm": 1.4027098713260302,
      "learning_rate": 7.93364200311612e-06,
      "loss": 0.9377,
      "step": 16650
    },
    {
      "epoch": 1.0068290324530127,
      "grad_norm": 1.384610909662134,
      "learning_rate": 7.930944025478755e-06,
      "loss": 0.9426,
      "step": 16660
    },
    {
      "epoch": 1.0074333716081465,
      "grad_norm": 1.4810983091198728,
      "learning_rate": 7.928244747083414e-06,
      "loss": 0.924,
      "step": 16670
    },
    {
      "epoch": 1.0080377107632803,
      "grad_norm": 1.3867969630647643,
      "learning_rate": 7.92554416912804e-06,
      "loss": 0.9265,
      "step": 16680
    },
    {
      "epoch": 1.008642049918414,
      "grad_norm": 1.4354481910321326,
      "learning_rate": 7.922842292811155e-06,
      "loss": 0.9474,
      "step": 16690
    },
    {
      "epoch": 1.0092463890735481,
      "grad_norm": 0.8792760693545016,
      "learning_rate": 7.920139119331859e-06,
      "loss": 0.9709,
      "step": 16700
    },
    {
      "epoch": 1.009850728228682,
      "grad_norm": 0.877800870610991,
      "learning_rate": 7.917434649889823e-06,
      "loss": 0.9437,
      "step": 16710
    },
    {
      "epoch": 1.0104550673838157,
      "grad_norm": 0.830735823659445,
      "learning_rate": 7.914728885685294e-06,
      "loss": 0.9241,
      "step": 16720
    },
    {
      "epoch": 1.0110594065389498,
      "grad_norm": 0.8530426531503883,
      "learning_rate": 7.912021827919097e-06,
      "loss": 0.9646,
      "step": 16730
    },
    {
      "epoch": 1.0116637456940836,
      "grad_norm": 0.88664475936253,
      "learning_rate": 7.909313477792627e-06,
      "loss": 0.9387,
      "step": 16740
    },
    {
      "epoch": 1.0122680848492174,
      "grad_norm": 0.7367816222313015,
      "learning_rate": 7.906603836507853e-06,
      "loss": 0.9344,
      "step": 16750
    },
    {
      "epoch": 1.0128724240043512,
      "grad_norm": 0.7859491510043656,
      "learning_rate": 7.903892905267323e-06,
      "loss": 0.9416,
      "step": 16760
    },
    {
      "epoch": 1.0134767631594852,
      "grad_norm": 0.7543168368386722,
      "learning_rate": 7.90118068527415e-06,
      "loss": 0.9353,
      "step": 16770
    },
    {
      "epoch": 1.014081102314619,
      "grad_norm": 0.7930047723096081,
      "learning_rate": 7.898467177732022e-06,
      "loss": 0.9544,
      "step": 16780
    },
    {
      "epoch": 1.0146854414697528,
      "grad_norm": 0.793617635036497,
      "learning_rate": 7.895752383845201e-06,
      "loss": 0.9378,
      "step": 16790
    },
    {
      "epoch": 1.0152897806248866,
      "grad_norm": 0.8769535221494227,
      "learning_rate": 7.893036304818513e-06,
      "loss": 0.9196,
      "step": 16800
    },
    {
      "epoch": 1.0158941197800206,
      "grad_norm": 0.8206136583910835,
      "learning_rate": 7.890318941857361e-06,
      "loss": 0.925,
      "step": 16810
    },
    {
      "epoch": 1.0164984589351544,
      "grad_norm": 0.7938365201088614,
      "learning_rate": 7.887600296167715e-06,
      "loss": 0.9397,
      "step": 16820
    },
    {
      "epoch": 1.0171027980902883,
      "grad_norm": 0.7925782897904219,
      "learning_rate": 7.884880368956113e-06,
      "loss": 0.9332,
      "step": 16830
    },
    {
      "epoch": 1.017707137245422,
      "grad_norm": 0.7908502049314249,
      "learning_rate": 7.882159161429667e-06,
      "loss": 0.9308,
      "step": 16840
    },
    {
      "epoch": 1.018311476400556,
      "grad_norm": 0.5810920604705304,
      "learning_rate": 7.879436674796051e-06,
      "loss": 0.9241,
      "step": 16850
    },
    {
      "epoch": 1.0189158155556899,
      "grad_norm": 0.5660654629161752,
      "learning_rate": 7.876712910263506e-06,
      "loss": 0.9467,
      "step": 16860
    },
    {
      "epoch": 1.0195201547108237,
      "grad_norm": 0.6036814742101025,
      "learning_rate": 7.873987869040851e-06,
      "loss": 0.9597,
      "step": 16870
    },
    {
      "epoch": 1.0201244938659575,
      "grad_norm": 0.5723685025171404,
      "learning_rate": 7.871261552337458e-06,
      "loss": 0.9177,
      "step": 16880
    },
    {
      "epoch": 1.0207288330210915,
      "grad_norm": 0.6122230853641085,
      "learning_rate": 7.868533961363272e-06,
      "loss": 0.933,
      "step": 16890
    },
    {
      "epoch": 1.0213331721762253,
      "grad_norm": 0.6092007285623798,
      "learning_rate": 7.865805097328804e-06,
      "loss": 0.9155,
      "step": 16900
    },
    {
      "epoch": 1.0219375113313591,
      "grad_norm": 0.6326732563963001,
      "learning_rate": 7.86307496144513e-06,
      "loss": 0.9256,
      "step": 16910
    },
    {
      "epoch": 1.022541850486493,
      "grad_norm": 0.6236000190709885,
      "learning_rate": 7.860343554923884e-06,
      "loss": 0.9498,
      "step": 16920
    },
    {
      "epoch": 1.023146189641627,
      "grad_norm": 0.6126817686032936,
      "learning_rate": 7.857610878977273e-06,
      "loss": 0.95,
      "step": 16930
    },
    {
      "epoch": 1.0237505287967608,
      "grad_norm": 0.6285983823142869,
      "learning_rate": 7.854876934818064e-06,
      "loss": 0.9264,
      "step": 16940
    },
    {
      "epoch": 1.0243548679518946,
      "grad_norm": 0.8143835637091942,
      "learning_rate": 7.85214172365958e-06,
      "loss": 0.9494,
      "step": 16950
    },
    {
      "epoch": 1.0249592071070284,
      "grad_norm": 0.8357300452676808,
      "learning_rate": 7.84940524671572e-06,
      "loss": 0.9428,
      "step": 16960
    },
    {
      "epoch": 1.0255635462621624,
      "grad_norm": 0.8789353994651676,
      "learning_rate": 7.846667505200931e-06,
      "loss": 0.9587,
      "step": 16970
    },
    {
      "epoch": 1.0261678854172962,
      "grad_norm": 0.8293078770478578,
      "learning_rate": 7.843928500330234e-06,
      "loss": 0.9203,
      "step": 16980
    },
    {
      "epoch": 1.02677222457243,
      "grad_norm": 0.8425941491537405,
      "learning_rate": 7.841188233319197e-06,
      "loss": 0.942,
      "step": 16990
    },
    {
      "epoch": 1.0273765637275638,
      "grad_norm": 0.5866636262704408,
      "learning_rate": 7.838446705383958e-06,
      "loss": 0.9254,
      "step": 17000
    },
    {
      "epoch": 1.0279809028826978,
      "grad_norm": 0.5659161865147383,
      "learning_rate": 7.835703917741213e-06,
      "loss": 0.9424,
      "step": 17010
    },
    {
      "epoch": 1.0285852420378316,
      "grad_norm": 0.5326252176534534,
      "learning_rate": 7.832959871608216e-06,
      "loss": 0.9425,
      "step": 17020
    },
    {
      "epoch": 1.0291895811929654,
      "grad_norm": 0.5291294681712005,
      "learning_rate": 7.830214568202777e-06,
      "loss": 0.948,
      "step": 17030
    },
    {
      "epoch": 1.0297939203480992,
      "grad_norm": 0.5418078807948797,
      "learning_rate": 7.827468008743269e-06,
      "loss": 0.9635,
      "step": 17040
    },
    {
      "epoch": 1.0303982595032333,
      "grad_norm": 0.5305533046485595,
      "learning_rate": 7.824720194448617e-06,
      "loss": 0.9186,
      "step": 17050
    },
    {
      "epoch": 1.031002598658367,
      "grad_norm": 0.5329434727869929,
      "learning_rate": 7.821971126538308e-06,
      "loss": 0.9312,
      "step": 17060
    },
    {
      "epoch": 1.0316069378135009,
      "grad_norm": 0.5502276447201018,
      "learning_rate": 7.819220806232381e-06,
      "loss": 0.9439,
      "step": 17070
    },
    {
      "epoch": 1.032211276968635,
      "grad_norm": 0.5500431855661826,
      "learning_rate": 7.816469234751436e-06,
      "loss": 0.9628,
      "step": 17080
    },
    {
      "epoch": 1.0328156161237687,
      "grad_norm": 0.5290719256049004,
      "learning_rate": 7.813716413316621e-06,
      "loss": 0.9596,
      "step": 17090
    },
    {
      "epoch": 1.0334199552789025,
      "grad_norm": 0.5387557996000338,
      "learning_rate": 7.810962343149647e-06,
      "loss": 0.9512,
      "step": 17100
    },
    {
      "epoch": 1.0340242944340363,
      "grad_norm": 0.5536744671060361,
      "learning_rate": 7.808207025472773e-06,
      "loss": 0.9343,
      "step": 17110
    },
    {
      "epoch": 1.0346286335891703,
      "grad_norm": 0.5307958362077472,
      "learning_rate": 7.805450461508812e-06,
      "loss": 0.9232,
      "step": 17120
    },
    {
      "epoch": 1.0352329727443041,
      "grad_norm": 0.5141975412793283,
      "learning_rate": 7.802692652481135e-06,
      "loss": 0.937,
      "step": 17130
    },
    {
      "epoch": 1.035837311899438,
      "grad_norm": 0.5673089058993276,
      "learning_rate": 7.799933599613661e-06,
      "loss": 0.9383,
      "step": 17140
    },
    {
      "epoch": 1.0364416510545718,
      "grad_norm": 0.4963958181023315,
      "learning_rate": 7.797173304130861e-06,
      "loss": 0.9198,
      "step": 17150
    },
    {
      "epoch": 1.0370459902097058,
      "grad_norm": 0.5188908573139808,
      "learning_rate": 7.794411767257763e-06,
      "loss": 0.904,
      "step": 17160
    },
    {
      "epoch": 1.0376503293648396,
      "grad_norm": 0.5139287859896396,
      "learning_rate": 7.791648990219936e-06,
      "loss": 0.9144,
      "step": 17170
    },
    {
      "epoch": 1.0382546685199734,
      "grad_norm": 0.5580327050618912,
      "learning_rate": 7.78888497424351e-06,
      "loss": 0.9203,
      "step": 17180
    },
    {
      "epoch": 1.0388590076751072,
      "grad_norm": 0.5278198038472391,
      "learning_rate": 7.786119720555158e-06,
      "loss": 0.9252,
      "step": 17190
    },
    {
      "epoch": 1.0394633468302412,
      "grad_norm": 0.5591315816805235,
      "learning_rate": 7.783353230382105e-06,
      "loss": 0.9067,
      "step": 17200
    },
    {
      "epoch": 1.040067685985375,
      "grad_norm": 0.5357427652993897,
      "learning_rate": 7.780585504952122e-06,
      "loss": 0.9395,
      "step": 17210
    },
    {
      "epoch": 1.0406720251405088,
      "grad_norm": 0.5134222025354997,
      "learning_rate": 7.777816545493534e-06,
      "loss": 0.9435,
      "step": 17220
    },
    {
      "epoch": 1.0412763642956426,
      "grad_norm": 0.5493819771337071,
      "learning_rate": 7.775046353235206e-06,
      "loss": 0.919,
      "step": 17230
    },
    {
      "epoch": 1.0418807034507767,
      "grad_norm": 0.534374848403984,
      "learning_rate": 7.772274929406556e-06,
      "loss": 0.945,
      "step": 17240
    },
    {
      "epoch": 1.0424850426059105,
      "grad_norm": 0.5265338350839955,
      "learning_rate": 7.769502275237548e-06,
      "loss": 0.9466,
      "step": 17250
    },
    {
      "epoch": 1.0430893817610443,
      "grad_norm": 0.5657623021518893,
      "learning_rate": 7.766728391958689e-06,
      "loss": 0.9243,
      "step": 17260
    },
    {
      "epoch": 1.043693720916178,
      "grad_norm": 0.5416678425204373,
      "learning_rate": 7.763953280801033e-06,
      "loss": 0.9394,
      "step": 17270
    },
    {
      "epoch": 1.044298060071312,
      "grad_norm": 0.5540016282585845,
      "learning_rate": 7.761176942996178e-06,
      "loss": 0.9398,
      "step": 17280
    },
    {
      "epoch": 1.044902399226446,
      "grad_norm": 0.5735650702516203,
      "learning_rate": 7.75839937977627e-06,
      "loss": 0.9326,
      "step": 17290
    },
    {
      "epoch": 1.0455067383815797,
      "grad_norm": 0.6312286886898297,
      "learning_rate": 7.755620592373996e-06,
      "loss": 0.9282,
      "step": 17300
    },
    {
      "epoch": 1.0461110775367135,
      "grad_norm": 0.6334737381977422,
      "learning_rate": 7.752840582022585e-06,
      "loss": 0.9255,
      "step": 17310
    },
    {
      "epoch": 1.0467154166918475,
      "grad_norm": 0.654917849959172,
      "learning_rate": 7.750059349955813e-06,
      "loss": 0.9452,
      "step": 17320
    },
    {
      "epoch": 1.0473197558469813,
      "grad_norm": 0.6271134174635833,
      "learning_rate": 7.747276897407991e-06,
      "loss": 0.931,
      "step": 17330
    },
    {
      "epoch": 1.0479240950021151,
      "grad_norm": 0.6213932038944121,
      "learning_rate": 7.74449322561398e-06,
      "loss": 0.9367,
      "step": 17340
    },
    {
      "epoch": 1.048528434157249,
      "grad_norm": 0.6842261683524189,
      "learning_rate": 7.741708335809178e-06,
      "loss": 0.9407,
      "step": 17350
    },
    {
      "epoch": 1.049132773312383,
      "grad_norm": 0.7512335733068148,
      "learning_rate": 7.738922229229523e-06,
      "loss": 0.9677,
      "step": 17360
    },
    {
      "epoch": 1.0497371124675168,
      "grad_norm": 0.6636142611915227,
      "learning_rate": 7.736134907111492e-06,
      "loss": 0.9028,
      "step": 17370
    },
    {
      "epoch": 1.0503414516226506,
      "grad_norm": 0.719226792283849,
      "learning_rate": 7.733346370692108e-06,
      "loss": 0.9278,
      "step": 17380
    },
    {
      "epoch": 1.0509457907777846,
      "grad_norm": 0.652187295856004,
      "learning_rate": 7.730556621208925e-06,
      "loss": 0.9351,
      "step": 17390
    },
    {
      "epoch": 1.0515501299329184,
      "grad_norm": 0.6589839148089985,
      "learning_rate": 7.727765659900037e-06,
      "loss": 0.9654,
      "step": 17400
    },
    {
      "epoch": 1.0521544690880522,
      "grad_norm": 0.6159802991459832,
      "learning_rate": 7.724973488004081e-06,
      "loss": 0.9108,
      "step": 17410
    },
    {
      "epoch": 1.052758808243186,
      "grad_norm": 0.6811609900169269,
      "learning_rate": 7.722180106760228e-06,
      "loss": 0.9287,
      "step": 17420
    },
    {
      "epoch": 1.05336314739832,
      "grad_norm": 0.6310949608454279,
      "learning_rate": 7.719385517408182e-06,
      "loss": 0.9068,
      "step": 17430
    },
    {
      "epoch": 1.0539674865534538,
      "grad_norm": 0.6288201714726321,
      "learning_rate": 7.716589721188188e-06,
      "loss": 0.945,
      "step": 17440
    },
    {
      "epoch": 1.0545718257085877,
      "grad_norm": 0.7816625286245904,
      "learning_rate": 7.713792719341025e-06,
      "loss": 0.9545,
      "step": 17450
    },
    {
      "epoch": 1.0551761648637215,
      "grad_norm": 0.8997991728124461,
      "learning_rate": 7.710994513108007e-06,
      "loss": 0.919,
      "step": 17460
    },
    {
      "epoch": 1.0557805040188555,
      "grad_norm": 0.8067305936420595,
      "learning_rate": 7.708195103730985e-06,
      "loss": 0.9348,
      "step": 17470
    },
    {
      "epoch": 1.0563848431739893,
      "grad_norm": 0.7829921953321952,
      "learning_rate": 7.705394492452337e-06,
      "loss": 0.9414,
      "step": 17480
    },
    {
      "epoch": 1.056989182329123,
      "grad_norm": 0.7771452511030948,
      "learning_rate": 7.702592680514987e-06,
      "loss": 0.9425,
      "step": 17490
    },
    {
      "epoch": 1.057593521484257,
      "grad_norm": 0.8287772552764033,
      "learning_rate": 7.699789669162375e-06,
      "loss": 0.9423,
      "step": 17500
    },
    {
      "epoch": 1.058197860639391,
      "grad_norm": 0.7865863061813924,
      "learning_rate": 7.696985459638487e-06,
      "loss": 0.9389,
      "step": 17510
    },
    {
      "epoch": 1.0588021997945247,
      "grad_norm": 0.7929486218402757,
      "learning_rate": 7.694180053187835e-06,
      "loss": 0.9106,
      "step": 17520
    },
    {
      "epoch": 1.0594065389496585,
      "grad_norm": 0.7467555816310237,
      "learning_rate": 7.691373451055465e-06,
      "loss": 0.9133,
      "step": 17530
    },
    {
      "epoch": 1.0600108781047923,
      "grad_norm": 0.768235156442969,
      "learning_rate": 7.688565654486946e-06,
      "loss": 0.9181,
      "step": 17540
    },
    {
      "epoch": 1.0606152172599264,
      "grad_norm": 1.2191750462306667,
      "learning_rate": 7.68575666472839e-06,
      "loss": 0.9342,
      "step": 17550
    },
    {
      "epoch": 1.0612195564150602,
      "grad_norm": 1.2345557906807694,
      "learning_rate": 7.68294648302643e-06,
      "loss": 0.9288,
      "step": 17560
    },
    {
      "epoch": 1.061823895570194,
      "grad_norm": 1.2710411362888787,
      "learning_rate": 7.680135110628226e-06,
      "loss": 0.9343,
      "step": 17570
    },
    {
      "epoch": 1.0624282347253278,
      "grad_norm": 1.2138994266559717,
      "learning_rate": 7.677322548781471e-06,
      "loss": 0.9271,
      "step": 17580
    },
    {
      "epoch": 1.0630325738804618,
      "grad_norm": 1.2243402093882982,
      "learning_rate": 7.674508798734388e-06,
      "loss": 0.9165,
      "step": 17590
    },
    {
      "epoch": 1.0636369130355956,
      "grad_norm": 0.9175987045396671,
      "learning_rate": 7.67169386173572e-06,
      "loss": 0.9404,
      "step": 17600
    },
    {
      "epoch": 1.0642412521907294,
      "grad_norm": 0.9958627314941471,
      "learning_rate": 7.668877739034743e-06,
      "loss": 0.9311,
      "step": 17610
    },
    {
      "epoch": 1.0648455913458632,
      "grad_norm": 0.9076448485095544,
      "learning_rate": 7.666060431881258e-06,
      "loss": 0.9257,
      "step": 17620
    },
    {
      "epoch": 1.0654499305009972,
      "grad_norm": 0.9631456639674933,
      "learning_rate": 7.663241941525587e-06,
      "loss": 0.9285,
      "step": 17630
    },
    {
      "epoch": 1.066054269656131,
      "grad_norm": 0.8543313921634847,
      "learning_rate": 7.660422269218584e-06,
      "loss": 0.9267,
      "step": 17640
    },
    {
      "epoch": 1.0666586088112648,
      "grad_norm": 2.401943331264358,
      "learning_rate": 7.657601416211625e-06,
      "loss": 0.9548,
      "step": 17650
    },
    {
      "epoch": 1.0672629479663986,
      "grad_norm": 2.323754105260101,
      "learning_rate": 7.654779383756606e-06,
      "loss": 0.9367,
      "step": 17660
    },
    {
      "epoch": 1.0678672871215327,
      "grad_norm": 2.532305996465686,
      "learning_rate": 7.651956173105952e-06,
      "loss": 0.9371,
      "step": 17670
    },
    {
      "epoch": 1.0684716262766665,
      "grad_norm": 2.6212061626863994,
      "learning_rate": 7.64913178551261e-06,
      "loss": 0.9183,
      "step": 17680
    },
    {
      "epoch": 1.0690759654318003,
      "grad_norm": 2.19291191968829,
      "learning_rate": 7.646306222230046e-06,
      "loss": 0.9642,
      "step": 17690
    },
    {
      "epoch": 1.0696803045869343,
      "grad_norm": 0.9355035176067491,
      "learning_rate": 7.643479484512251e-06,
      "loss": 0.9443,
      "step": 17700
    },
    {
      "epoch": 1.070284643742068,
      "grad_norm": 0.9186575857568213,
      "learning_rate": 7.640651573613735e-06,
      "loss": 0.9281,
      "step": 17710
    },
    {
      "epoch": 1.070888982897202,
      "grad_norm": 0.944859914491086,
      "learning_rate": 7.63782249078953e-06,
      "loss": 0.9608,
      "step": 17720
    },
    {
      "epoch": 1.0714933220523357,
      "grad_norm": 0.9321307889049993,
      "learning_rate": 7.634992237295188e-06,
      "loss": 0.9243,
      "step": 17730
    },
    {
      "epoch": 1.0720976612074695,
      "grad_norm": 0.8925562864372932,
      "learning_rate": 7.63216081438678e-06,
      "loss": 0.9356,
      "step": 17740
    },
    {
      "epoch": 1.0727020003626035,
      "grad_norm": 0.9319896823185124,
      "learning_rate": 7.629328223320898e-06,
      "loss": 0.9441,
      "step": 17750
    },
    {
      "epoch": 1.0733063395177374,
      "grad_norm": 0.9632602171112727,
      "learning_rate": 7.626494465354648e-06,
      "loss": 0.9413,
      "step": 17760
    },
    {
      "epoch": 1.0739106786728712,
      "grad_norm": 0.9679103853513814,
      "learning_rate": 7.623659541745658e-06,
      "loss": 0.9113,
      "step": 17770
    },
    {
      "epoch": 1.0745150178280052,
      "grad_norm": 1.0326943645942648,
      "learning_rate": 7.620823453752073e-06,
      "loss": 0.9324,
      "step": 17780
    },
    {
      "epoch": 1.075119356983139,
      "grad_norm": 1.0656998837801366,
      "learning_rate": 7.617986202632553e-06,
      "loss": 0.9476,
      "step": 17790
    },
    {
      "epoch": 1.0757236961382728,
      "grad_norm": 1.1655298605395468,
      "learning_rate": 7.615147789646272e-06,
      "loss": 0.9116,
      "step": 17800
    },
    {
      "epoch": 1.0763280352934066,
      "grad_norm": 1.0036326709028562,
      "learning_rate": 7.612308216052929e-06,
      "loss": 0.9515,
      "step": 17810
    },
    {
      "epoch": 1.0769323744485406,
      "grad_norm": 1.0340336615446928,
      "learning_rate": 7.6094674831127245e-06,
      "loss": 0.9511,
      "step": 17820
    },
    {
      "epoch": 1.0775367136036744,
      "grad_norm": 0.98383497117508,
      "learning_rate": 7.606625592086385e-06,
      "loss": 0.9323,
      "step": 17830
    },
    {
      "epoch": 1.0781410527588082,
      "grad_norm": 1.0825964160397739,
      "learning_rate": 7.603782544235146e-06,
      "loss": 0.9203,
      "step": 17840
    },
    {
      "epoch": 1.078745391913942,
      "grad_norm": 1.0986722008342131,
      "learning_rate": 7.600938340820756e-06,
      "loss": 0.9309,
      "step": 17850
    },
    {
      "epoch": 1.079349731069076,
      "grad_norm": 1.2457489182430421,
      "learning_rate": 7.598092983105478e-06,
      "loss": 0.9285,
      "step": 17860
    },
    {
      "epoch": 1.0799540702242099,
      "grad_norm": 1.0850639495062309,
      "learning_rate": 7.595246472352088e-06,
      "loss": 0.9296,
      "step": 17870
    },
    {
      "epoch": 1.0805584093793437,
      "grad_norm": 1.0203780363698607,
      "learning_rate": 7.592398809823869e-06,
      "loss": 0.9422,
      "step": 17880
    },
    {
      "epoch": 1.0811627485344775,
      "grad_norm": 1.0045630153118468,
      "learning_rate": 7.589549996784622e-06,
      "loss": 0.9469,
      "step": 17890
    },
    {
      "epoch": 1.0817670876896115,
      "grad_norm": 1.2239430174571362,
      "learning_rate": 7.586700034498652e-06,
      "loss": 0.9596,
      "step": 17900
    },
    {
      "epoch": 1.0823714268447453,
      "grad_norm": 1.1464164762917124,
      "learning_rate": 7.5838489242307785e-06,
      "loss": 0.9251,
      "step": 17910
    },
    {
      "epoch": 1.082975765999879,
      "grad_norm": 1.163584774202159,
      "learning_rate": 7.58099666724633e-06,
      "loss": 0.9421,
      "step": 17920
    },
    {
      "epoch": 1.083580105155013,
      "grad_norm": 1.127372288276386,
      "learning_rate": 7.578143264811142e-06,
      "loss": 0.9379,
      "step": 17930
    },
    {
      "epoch": 1.084184444310147,
      "grad_norm": 1.1477872466713324,
      "learning_rate": 7.575288718191559e-06,
      "loss": 0.9132,
      "step": 17940
    },
    {
      "epoch": 1.0847887834652807,
      "grad_norm": 1.0134197870887063,
      "learning_rate": 7.572433028654436e-06,
      "loss": 0.9392,
      "step": 17950
    },
    {
      "epoch": 1.0853931226204145,
      "grad_norm": 1.0832553901018576,
      "learning_rate": 7.56957619746713e-06,
      "loss": 0.9257,
      "step": 17960
    },
    {
      "epoch": 1.0859974617755483,
      "grad_norm": 1.017673550134316,
      "learning_rate": 7.566718225897509e-06,
      "loss": 0.9117,
      "step": 17970
    },
    {
      "epoch": 1.0866018009306824,
      "grad_norm": 0.9358024022533614,
      "learning_rate": 7.563859115213945e-06,
      "loss": 0.9194,
      "step": 17980
    },
    {
      "epoch": 1.0872061400858162,
      "grad_norm": 0.9594514118907036,
      "learning_rate": 7.560998866685317e-06,
      "loss": 0.9089,
      "step": 17990
    },
    {
      "epoch": 1.08781047924095,
      "grad_norm": 0.9888000686801851,
      "learning_rate": 7.558137481581008e-06,
      "loss": 0.9493,
      "step": 18000
    },
    {
      "epoch": 1.0884148183960838,
      "grad_norm": 1.026695318025682,
      "learning_rate": 7.555274961170906e-06,
      "loss": 0.9409,
      "step": 18010
    },
    {
      "epoch": 1.0890191575512178,
      "grad_norm": 0.9677848897743511,
      "learning_rate": 7.5524113067254e-06,
      "loss": 0.9216,
      "step": 18020
    },
    {
      "epoch": 1.0896234967063516,
      "grad_norm": 1.0562920205313786,
      "learning_rate": 7.549546519515389e-06,
      "loss": 0.9432,
      "step": 18030
    },
    {
      "epoch": 1.0902278358614854,
      "grad_norm": 0.9954802932504392,
      "learning_rate": 7.546680600812267e-06,
      "loss": 0.9435,
      "step": 18040
    },
    {
      "epoch": 1.0908321750166192,
      "grad_norm": 1.0126452259594012,
      "learning_rate": 7.5438135518879355e-06,
      "loss": 0.9378,
      "step": 18050
    },
    {
      "epoch": 1.0914365141717532,
      "grad_norm": 1.013427097497143,
      "learning_rate": 7.540945374014794e-06,
      "loss": 0.9463,
      "step": 18060
    },
    {
      "epoch": 1.092040853326887,
      "grad_norm": 0.9956295192601601,
      "learning_rate": 7.538076068465746e-06,
      "loss": 0.9291,
      "step": 18070
    },
    {
      "epoch": 1.0926451924820209,
      "grad_norm": 0.9452206477910722,
      "learning_rate": 7.535205636514194e-06,
      "loss": 0.9426,
      "step": 18080
    },
    {
      "epoch": 1.0932495316371549,
      "grad_norm": 1.006683965899687,
      "learning_rate": 7.53233407943404e-06,
      "loss": 0.9511,
      "step": 18090
    },
    {
      "epoch": 1.0938538707922887,
      "grad_norm": 1.1452663795381461,
      "learning_rate": 7.529461398499689e-06,
      "loss": 0.9258,
      "step": 18100
    },
    {
      "epoch": 1.0944582099474225,
      "grad_norm": 1.2181680291939803,
      "learning_rate": 7.526587594986037e-06,
      "loss": 0.9289,
      "step": 18110
    },
    {
      "epoch": 1.0950625491025563,
      "grad_norm": 1.0759594597451736,
      "learning_rate": 7.523712670168486e-06,
      "loss": 0.9191,
      "step": 18120
    },
    {
      "epoch": 1.0956668882576903,
      "grad_norm": 1.2041323183873132,
      "learning_rate": 7.5208366253229316e-06,
      "loss": 0.9258,
      "step": 18130
    },
    {
      "epoch": 1.0962712274128241,
      "grad_norm": 1.1827484285363192,
      "learning_rate": 7.517959461725767e-06,
      "loss": 0.908,
      "step": 18140
    },
    {
      "epoch": 1.096875566567958,
      "grad_norm": 1.157332960067366,
      "learning_rate": 7.5150811806538825e-06,
      "loss": 0.9362,
      "step": 18150
    },
    {
      "epoch": 1.0974799057230917,
      "grad_norm": 1.169992858850769,
      "learning_rate": 7.512201783384665e-06,
      "loss": 0.9313,
      "step": 18160
    },
    {
      "epoch": 1.0980842448782258,
      "grad_norm": 1.1282561411447811,
      "learning_rate": 7.509321271195996e-06,
      "loss": 0.9665,
      "step": 18170
    },
    {
      "epoch": 1.0986885840333596,
      "grad_norm": 1.140677578190521,
      "learning_rate": 7.5064396453662515e-06,
      "loss": 0.9547,
      "step": 18180
    },
    {
      "epoch": 1.0992929231884934,
      "grad_norm": 1.0836950332806154,
      "learning_rate": 7.503556907174301e-06,
      "loss": 0.9438,
      "step": 18190
    },
    {
      "epoch": 1.0998972623436272,
      "grad_norm": 1.2107318221539178,
      "learning_rate": 7.500673057899509e-06,
      "loss": 0.9382,
      "step": 18200
    },
    {
      "epoch": 1.1005016014987612,
      "grad_norm": 1.1518057842264933,
      "learning_rate": 7.497788098821734e-06,
      "loss": 0.9082,
      "step": 18210
    },
    {
      "epoch": 1.101105940653895,
      "grad_norm": 1.2321108621606867,
      "learning_rate": 7.494902031221325e-06,
      "loss": 0.9689,
      "step": 18220
    },
    {
      "epoch": 1.1017102798090288,
      "grad_norm": 1.2688642849578091,
      "learning_rate": 7.492014856379123e-06,
      "loss": 0.9489,
      "step": 18230
    },
    {
      "epoch": 1.1023146189641626,
      "grad_norm": 1.1660540709104281,
      "learning_rate": 7.489126575576462e-06,
      "loss": 0.9349,
      "step": 18240
    },
    {
      "epoch": 1.1029189581192966,
      "grad_norm": 1.4189099891583747,
      "learning_rate": 7.486237190095165e-06,
      "loss": 0.9549,
      "step": 18250
    },
    {
      "epoch": 1.1035232972744304,
      "grad_norm": 1.4092109380096982,
      "learning_rate": 7.483346701217548e-06,
      "loss": 0.9426,
      "step": 18260
    },
    {
      "epoch": 1.1041276364295642,
      "grad_norm": 1.388194491260783,
      "learning_rate": 7.4804551102264125e-06,
      "loss": 0.9275,
      "step": 18270
    },
    {
      "epoch": 1.104731975584698,
      "grad_norm": 1.3794214625191168,
      "learning_rate": 7.477562418405053e-06,
      "loss": 0.9363,
      "step": 18280
    },
    {
      "epoch": 1.105336314739832,
      "grad_norm": 1.339498916009595,
      "learning_rate": 7.474668627037251e-06,
      "loss": 0.9384,
      "step": 18290
    },
    {
      "epoch": 1.1059406538949659,
      "grad_norm": 0.8755874501739986,
      "learning_rate": 7.471773737407277e-06,
      "loss": 0.925,
      "step": 18300
    },
    {
      "epoch": 1.1065449930500997,
      "grad_norm": 0.9110506355136735,
      "learning_rate": 7.468877750799887e-06,
      "loss": 0.9342,
      "step": 18310
    },
    {
      "epoch": 1.1071493322052335,
      "grad_norm": 0.8660264542621957,
      "learning_rate": 7.4659806685003245e-06,
      "loss": 0.9439,
      "step": 18320
    },
    {
      "epoch": 1.1077536713603675,
      "grad_norm": 0.8759157981394622,
      "learning_rate": 7.4630824917943186e-06,
      "loss": 0.9426,
      "step": 18330
    },
    {
      "epoch": 1.1083580105155013,
      "grad_norm": 0.8532046255984694,
      "learning_rate": 7.460183221968088e-06,
      "loss": 0.938,
      "step": 18340
    },
    {
      "epoch": 1.1089623496706351,
      "grad_norm": 0.7509445769856646,
      "learning_rate": 7.4572828603083325e-06,
      "loss": 0.9475,
      "step": 18350
    },
    {
      "epoch": 1.109566688825769,
      "grad_norm": 0.8094425513662415,
      "learning_rate": 7.454381408102236e-06,
      "loss": 0.9433,
      "step": 18360
    },
    {
      "epoch": 1.110171027980903,
      "grad_norm": 0.7666016460687293,
      "learning_rate": 7.451478866637469e-06,
      "loss": 0.9206,
      "step": 18370
    },
    {
      "epoch": 1.1107753671360368,
      "grad_norm": 0.7527321293860785,
      "learning_rate": 7.448575237202185e-06,
      "loss": 0.9427,
      "step": 18380
    },
    {
      "epoch": 1.1113797062911706,
      "grad_norm": 0.816718070340287,
      "learning_rate": 7.445670521085017e-06,
      "loss": 0.933,
      "step": 18390
    },
    {
      "epoch": 1.1119840454463046,
      "grad_norm": 0.7946087949364989,
      "learning_rate": 7.442764719575088e-06,
      "loss": 0.9562,
      "step": 18400
    },
    {
      "epoch": 1.1125883846014384,
      "grad_norm": 0.8368392538723526,
      "learning_rate": 7.4398578339619935e-06,
      "loss": 0.9347,
      "step": 18410
    },
    {
      "epoch": 1.1131927237565722,
      "grad_norm": 0.8676632114239737,
      "learning_rate": 7.436949865535814e-06,
      "loss": 0.9412,
      "step": 18420
    },
    {
      "epoch": 1.113797062911706,
      "grad_norm": 0.7942275826430807,
      "learning_rate": 7.434040815587114e-06,
      "loss": 0.9443,
      "step": 18430
    },
    {
      "epoch": 1.1144014020668398,
      "grad_norm": 0.8746305292978679,
      "learning_rate": 7.431130685406933e-06,
      "loss": 0.9267,
      "step": 18440
    },
    {
      "epoch": 1.1150057412219738,
      "grad_norm": 0.5921675189226722,
      "learning_rate": 7.428219476286792e-06,
      "loss": 0.9375,
      "step": 18450
    },
    {
      "epoch": 1.1156100803771076,
      "grad_norm": 0.6047915413956512,
      "learning_rate": 7.425307189518691e-06,
      "loss": 0.9683,
      "step": 18460
    },
    {
      "epoch": 1.1162144195322414,
      "grad_norm": 0.5641124287732586,
      "learning_rate": 7.4223938263951075e-06,
      "loss": 0.9376,
      "step": 18470
    },
    {
      "epoch": 1.1168187586873755,
      "grad_norm": 0.6032880788133669,
      "learning_rate": 7.419479388208997e-06,
      "loss": 0.9294,
      "step": 18480
    },
    {
      "epoch": 1.1174230978425093,
      "grad_norm": 0.5768160398383214,
      "learning_rate": 7.416563876253793e-06,
      "loss": 0.9356,
      "step": 18490
    },
    {
      "epoch": 1.118027436997643,
      "grad_norm": 0.6074259013678874,
      "learning_rate": 7.413647291823402e-06,
      "loss": 0.9392,
      "step": 18500
    },
    {
      "epoch": 1.1186317761527769,
      "grad_norm": 0.6019726796676416,
      "learning_rate": 7.410729636212212e-06,
      "loss": 0.9203,
      "step": 18510
    },
    {
      "epoch": 1.119236115307911,
      "grad_norm": 0.6425817645653109,
      "learning_rate": 7.4078109107150834e-06,
      "loss": 0.918,
      "step": 18520
    },
    {
      "epoch": 1.1198404544630447,
      "grad_norm": 0.6292809675648877,
      "learning_rate": 7.404891116627349e-06,
      "loss": 0.9391,
      "step": 18530
    },
    {
      "epoch": 1.1204447936181785,
      "grad_norm": 0.5967513731120074,
      "learning_rate": 7.401970255244821e-06,
      "loss": 0.9438,
      "step": 18540
    },
    {
      "epoch": 1.1210491327733123,
      "grad_norm": 0.8391683127683519,
      "learning_rate": 7.3990483278637825e-06,
      "loss": 0.9276,
      "step": 18550
    },
    {
      "epoch": 1.1216534719284463,
      "grad_norm": 0.8187967151357829,
      "learning_rate": 7.396125335780988e-06,
      "loss": 0.9311,
      "step": 18560
    },
    {
      "epoch": 1.1222578110835801,
      "grad_norm": 0.8336153283271945,
      "learning_rate": 7.393201280293668e-06,
      "loss": 0.9278,
      "step": 18570
    },
    {
      "epoch": 1.122862150238714,
      "grad_norm": 0.8640297047521126,
      "learning_rate": 7.390276162699524e-06,
      "loss": 0.9316,
      "step": 18580
    },
    {
      "epoch": 1.1234664893938477,
      "grad_norm": 0.8728493256719332,
      "learning_rate": 7.387349984296725e-06,
      "loss": 0.9429,
      "step": 18590
    },
    {
      "epoch": 1.1240708285489818,
      "grad_norm": 0.5105072076303535,
      "learning_rate": 7.384422746383918e-06,
      "loss": 0.9342,
      "step": 18600
    },
    {
      "epoch": 1.1246751677041156,
      "grad_norm": 0.5284821926637777,
      "learning_rate": 7.381494450260214e-06,
      "loss": 0.9371,
      "step": 18610
    },
    {
      "epoch": 1.1252795068592494,
      "grad_norm": 0.5383736889487204,
      "learning_rate": 7.378565097225196e-06,
      "loss": 0.9442,
      "step": 18620
    },
    {
      "epoch": 1.1258838460143832,
      "grad_norm": 0.5487929806086651,
      "learning_rate": 7.375634688578916e-06,
      "loss": 0.9305,
      "step": 18630
    },
    {
      "epoch": 1.1264881851695172,
      "grad_norm": 0.5174109381174217,
      "learning_rate": 7.372703225621896e-06,
      "loss": 0.9543,
      "step": 18640
    },
    {
      "epoch": 1.127092524324651,
      "grad_norm": 0.5238752933855901,
      "learning_rate": 7.36977070965512e-06,
      "loss": 0.9169,
      "step": 18650
    },
    {
      "epoch": 1.1276968634797848,
      "grad_norm": 0.5611088769581871,
      "learning_rate": 7.366837141980048e-06,
      "loss": 0.9332,
      "step": 18660
    },
    {
      "epoch": 1.1283012026349186,
      "grad_norm": 0.5157733749348441,
      "learning_rate": 7.363902523898601e-06,
      "loss": 0.9377,
      "step": 18670
    },
    {
      "epoch": 1.1289055417900526,
      "grad_norm": 0.541693977630682,
      "learning_rate": 7.360966856713166e-06,
      "loss": 0.9345,
      "step": 18680
    },
    {
      "epoch": 1.1295098809451865,
      "grad_norm": 0.5772872167113711,
      "learning_rate": 7.358030141726599e-06,
      "loss": 0.9131,
      "step": 18690
    },
    {
      "epoch": 1.1301142201003203,
      "grad_norm": 0.593293592996897,
      "learning_rate": 7.355092380242217e-06,
      "loss": 0.9362,
      "step": 18700
    },
    {
      "epoch": 1.1307185592554543,
      "grad_norm": 0.4977876548116746,
      "learning_rate": 7.352153573563807e-06,
      "loss": 0.9268,
      "step": 18710
    },
    {
      "epoch": 1.131322898410588,
      "grad_norm": 0.5187828309416557,
      "learning_rate": 7.349213722995614e-06,
      "loss": 0.9382,
      "step": 18720
    },
    {
      "epoch": 1.1319272375657219,
      "grad_norm": 0.6057896594417571,
      "learning_rate": 7.346272829842348e-06,
      "loss": 0.9652,
      "step": 18730
    },
    {
      "epoch": 1.1325315767208557,
      "grad_norm": 0.5627037988929109,
      "learning_rate": 7.343330895409185e-06,
      "loss": 0.9322,
      "step": 18740
    },
    {
      "epoch": 1.1331359158759895,
      "grad_norm": 0.5938214083120487,
      "learning_rate": 7.340387921001759e-06,
      "loss": 0.9507,
      "step": 18750
    },
    {
      "epoch": 1.1337402550311235,
      "grad_norm": 0.497879296733788,
      "learning_rate": 7.337443907926167e-06,
      "loss": 0.9451,
      "step": 18760
    },
    {
      "epoch": 1.1343445941862573,
      "grad_norm": 0.5291685520192279,
      "learning_rate": 7.334498857488968e-06,
      "loss": 0.9368,
      "step": 18770
    },
    {
      "epoch": 1.1349489333413911,
      "grad_norm": 0.5169483284744473,
      "learning_rate": 7.331552770997179e-06,
      "loss": 0.9176,
      "step": 18780
    },
    {
      "epoch": 1.1355532724965252,
      "grad_norm": 0.5121865324391289,
      "learning_rate": 7.328605649758278e-06,
      "loss": 0.9386,
      "step": 18790
    },
    {
      "epoch": 1.136157611651659,
      "grad_norm": 0.5030996319434998,
      "learning_rate": 7.325657495080205e-06,
      "loss": 0.9391,
      "step": 18800
    },
    {
      "epoch": 1.1367619508067928,
      "grad_norm": 0.5134502893425638,
      "learning_rate": 7.322708308271353e-06,
      "loss": 0.9092,
      "step": 18810
    },
    {
      "epoch": 1.1373662899619266,
      "grad_norm": 0.5378254096527265,
      "learning_rate": 7.319758090640579e-06,
      "loss": 0.9194,
      "step": 18820
    },
    {
      "epoch": 1.1379706291170604,
      "grad_norm": 0.5646526258128092,
      "learning_rate": 7.316806843497192e-06,
      "loss": 0.9556,
      "step": 18830
    },
    {
      "epoch": 1.1385749682721944,
      "grad_norm": 0.5201525360244731,
      "learning_rate": 7.313854568150962e-06,
      "loss": 0.9135,
      "step": 18840
    },
    {
      "epoch": 1.1391793074273282,
      "grad_norm": 0.571178709083454,
      "learning_rate": 7.3109012659121134e-06,
      "loss": 0.9372,
      "step": 18850
    },
    {
      "epoch": 1.139783646582462,
      "grad_norm": 0.5924905430176384,
      "learning_rate": 7.3079469380913285e-06,
      "loss": 0.9122,
      "step": 18860
    },
    {
      "epoch": 1.140387985737596,
      "grad_norm": 0.5478826754841538,
      "learning_rate": 7.304991585999741e-06,
      "loss": 0.9494,
      "step": 18870
    },
    {
      "epoch": 1.1409923248927298,
      "grad_norm": 0.5760825458033325,
      "learning_rate": 7.302035210948941e-06,
      "loss": 0.9458,
      "step": 18880
    },
    {
      "epoch": 1.1415966640478636,
      "grad_norm": 0.5914022187486925,
      "learning_rate": 7.2990778142509745e-06,
      "loss": 0.965,
      "step": 18890
    },
    {
      "epoch": 1.1422010032029974,
      "grad_norm": 0.6763849306396178,
      "learning_rate": 7.2961193972183374e-06,
      "loss": 0.9597,
      "step": 18900
    },
    {
      "epoch": 1.1428053423581315,
      "grad_norm": 0.6048262417231475,
      "learning_rate": 7.293159961163983e-06,
      "loss": 0.9086,
      "step": 18910
    },
    {
      "epoch": 1.1434096815132653,
      "grad_norm": 0.6222698001208778,
      "learning_rate": 7.29019950740131e-06,
      "loss": 0.93,
      "step": 18920
    },
    {
      "epoch": 1.144014020668399,
      "grad_norm": 0.6685430569916258,
      "learning_rate": 7.287238037244175e-06,
      "loss": 0.9285,
      "step": 18930
    },
    {
      "epoch": 1.1446183598235329,
      "grad_norm": 0.6072658212484155,
      "learning_rate": 7.284275552006886e-06,
      "loss": 0.9363,
      "step": 18940
    },
    {
      "epoch": 1.145222698978667,
      "grad_norm": 0.6732557639309836,
      "learning_rate": 7.281312053004195e-06,
      "loss": 0.925,
      "step": 18950
    },
    {
      "epoch": 1.1458270381338007,
      "grad_norm": 0.6931346464609389,
      "learning_rate": 7.278347541551309e-06,
      "loss": 0.9415,
      "step": 18960
    },
    {
      "epoch": 1.1464313772889345,
      "grad_norm": 0.6928384248316889,
      "learning_rate": 7.275382018963885e-06,
      "loss": 0.9474,
      "step": 18970
    },
    {
      "epoch": 1.1470357164440683,
      "grad_norm": 0.6798218765972116,
      "learning_rate": 7.272415486558026e-06,
      "loss": 0.952,
      "step": 18980
    },
    {
      "epoch": 1.1476400555992023,
      "grad_norm": 0.6960719947984609,
      "learning_rate": 7.269447945650282e-06,
      "loss": 0.9613,
      "step": 18990
    },
    {
      "epoch": 1.1482443947543362,
      "grad_norm": 0.6430908521862926,
      "learning_rate": 7.266479397557656e-06,
      "loss": 0.9289,
      "step": 19000
    },
    {
      "epoch": 1.14884873390947,
      "grad_norm": 0.6543530075263003,
      "learning_rate": 7.263509843597595e-06,
      "loss": 0.9268,
      "step": 19010
    },
    {
      "epoch": 1.1494530730646038,
      "grad_norm": 0.6470296770223022,
      "learning_rate": 7.260539285087988e-06,
      "loss": 0.9456,
      "step": 19020
    },
    {
      "epoch": 1.1500574122197378,
      "grad_norm": 0.635482263462426,
      "learning_rate": 7.257567723347179e-06,
      "loss": 0.9155,
      "step": 19030
    },
    {
      "epoch": 1.1506617513748716,
      "grad_norm": 0.6649600152854649,
      "learning_rate": 7.254595159693948e-06,
      "loss": 0.9409,
      "step": 19040
    },
    {
      "epoch": 1.1512660905300054,
      "grad_norm": 0.831945972839819,
      "learning_rate": 7.251621595447526e-06,
      "loss": 0.9188,
      "step": 19050
    },
    {
      "epoch": 1.1518704296851392,
      "grad_norm": 0.7504975278972764,
      "learning_rate": 7.248647031927586e-06,
      "loss": 0.9537,
      "step": 19060
    },
    {
      "epoch": 1.1524747688402732,
      "grad_norm": 0.7936176246908174,
      "learning_rate": 7.2456714704542434e-06,
      "loss": 0.9351,
      "step": 19070
    },
    {
      "epoch": 1.153079107995407,
      "grad_norm": 0.8218128454207213,
      "learning_rate": 7.242694912348057e-06,
      "loss": 0.9312,
      "step": 19080
    },
    {
      "epoch": 1.1536834471505408,
      "grad_norm": 0.8156632103920143,
      "learning_rate": 7.23971735893003e-06,
      "loss": 0.9326,
      "step": 19090
    },
    {
      "epoch": 1.1542877863056749,
      "grad_norm": 0.772009988766304,
      "learning_rate": 7.236738811521605e-06,
      "loss": 0.9334,
      "step": 19100
    },
    {
      "epoch": 1.1548921254608087,
      "grad_norm": 0.8043428640497754,
      "learning_rate": 7.233759271444667e-06,
      "loss": 0.9492,
      "step": 19110
    },
    {
      "epoch": 1.1554964646159425,
      "grad_norm": 0.7813595861102836,
      "learning_rate": 7.23077874002154e-06,
      "loss": 0.929,
      "step": 19120
    },
    {
      "epoch": 1.1561008037710763,
      "grad_norm": 0.7419627370864481,
      "learning_rate": 7.22779721857499e-06,
      "loss": 0.93,
      "step": 19130
    },
    {
      "epoch": 1.15670514292621,
      "grad_norm": 0.7824465114149218,
      "learning_rate": 7.22481470842822e-06,
      "loss": 0.9304,
      "step": 19140
    },
    {
      "epoch": 1.157309482081344,
      "grad_norm": 1.1738793303501924,
      "learning_rate": 7.221831210904874e-06,
      "loss": 0.9341,
      "step": 19150
    },
    {
      "epoch": 1.157913821236478,
      "grad_norm": 1.2367200919460224,
      "learning_rate": 7.218846727329033e-06,
      "loss": 0.9379,
      "step": 19160
    },
    {
      "epoch": 1.1585181603916117,
      "grad_norm": 1.2387832930701335,
      "learning_rate": 7.215861259025216e-06,
      "loss": 0.9347,
      "step": 19170
    },
    {
      "epoch": 1.1591224995467457,
      "grad_norm": 1.3907061165091044,
      "learning_rate": 7.212874807318378e-06,
      "loss": 0.9331,
      "step": 19180
    },
    {
      "epoch": 1.1597268387018795,
      "grad_norm": 1.1598510527775059,
      "learning_rate": 7.2098873735339145e-06,
      "loss": 0.9452,
      "step": 19190
    },
    {
      "epoch": 1.1603311778570133,
      "grad_norm": 0.927557777311021,
      "learning_rate": 7.20689895899765e-06,
      "loss": 0.9247,
      "step": 19200
    },
    {
      "epoch": 1.1609355170121471,
      "grad_norm": 0.8370659921602016,
      "learning_rate": 7.2039095650358494e-06,
      "loss": 0.9399,
      "step": 19210
    },
    {
      "epoch": 1.1615398561672812,
      "grad_norm": 0.9688812200462812,
      "learning_rate": 7.200919192975212e-06,
      "loss": 0.928,
      "step": 19220
    },
    {
      "epoch": 1.162144195322415,
      "grad_norm": 1.264778437936213,
      "learning_rate": 7.197927844142869e-06,
      "loss": 0.9392,
      "step": 19230
    },
    {
      "epoch": 1.1627485344775488,
      "grad_norm": 0.9055547884474286,
      "learning_rate": 7.194935519866386e-06,
      "loss": 0.922,
      "step": 19240
    },
    {
      "epoch": 1.1633528736326826,
      "grad_norm": 2.4215463145885345,
      "learning_rate": 7.191942221473762e-06,
      "loss": 0.9328,
      "step": 19250
    },
    {
      "epoch": 1.1639572127878166,
      "grad_norm": 2.504318524119521,
      "learning_rate": 7.188947950293428e-06,
      "loss": 0.9594,
      "step": 19260
    },
    {
      "epoch": 1.1645615519429504,
      "grad_norm": 2.630624707801629,
      "learning_rate": 7.1859527076542465e-06,
      "loss": 0.9327,
      "step": 19270
    },
    {
      "epoch": 1.1651658910980842,
      "grad_norm": 2.548969903935177,
      "learning_rate": 7.1829564948855116e-06,
      "loss": 0.9448,
      "step": 19280
    },
    {
      "epoch": 1.165770230253218,
      "grad_norm": 2.772859776142628,
      "learning_rate": 7.179959313316946e-06,
      "loss": 0.939,
      "step": 19290
    },
    {
      "epoch": 1.166374569408352,
      "grad_norm": 0.9708992928620127,
      "learning_rate": 7.176961164278705e-06,
      "loss": 0.9366,
      "step": 19300
    },
    {
      "epoch": 1.1669789085634859,
      "grad_norm": 0.9322807228718224,
      "learning_rate": 7.173962049101374e-06,
      "loss": 0.9242,
      "step": 19310
    },
    {
      "epoch": 1.1675832477186197,
      "grad_norm": 0.9071571047705957,
      "learning_rate": 7.170961969115964e-06,
      "loss": 0.9272,
      "step": 19320
    },
    {
      "epoch": 1.1681875868737535,
      "grad_norm": 0.9064026522166889,
      "learning_rate": 7.167960925653915e-06,
      "loss": 0.9143,
      "step": 19330
    },
    {
      "epoch": 1.1687919260288875,
      "grad_norm": 0.8856958909723631,
      "learning_rate": 7.164958920047096e-06,
      "loss": 0.9505,
      "step": 19340
    },
    {
      "epoch": 1.1693962651840213,
      "grad_norm": 0.9827473727432444,
      "learning_rate": 7.161955953627801e-06,
      "loss": 0.9409,
      "step": 19350
    },
    {
      "epoch": 1.170000604339155,
      "grad_norm": 1.1319777866603369,
      "learning_rate": 7.158952027728756e-06,
      "loss": 0.9585,
      "step": 19360
    },
    {
      "epoch": 1.170604943494289,
      "grad_norm": 0.9307210123248505,
      "learning_rate": 7.155947143683104e-06,
      "loss": 0.944,
      "step": 19370
    },
    {
      "epoch": 1.171209282649423,
      "grad_norm": 1.0866587614294696,
      "learning_rate": 7.152941302824419e-06,
      "loss": 0.9298,
      "step": 19380
    },
    {
      "epoch": 1.1718136218045567,
      "grad_norm": 0.9725624097208035,
      "learning_rate": 7.149934506486698e-06,
      "loss": 0.9095,
      "step": 19390
    },
    {
      "epoch": 1.1724179609596905,
      "grad_norm": 0.9764327494403443,
      "learning_rate": 7.146926756004362e-06,
      "loss": 0.9074,
      "step": 19400
    },
    {
      "epoch": 1.1730223001148246,
      "grad_norm": 1.0752513846267016,
      "learning_rate": 7.143918052712256e-06,
      "loss": 0.9389,
      "step": 19410
    },
    {
      "epoch": 1.1736266392699584,
      "grad_norm": 0.993877767810036,
      "learning_rate": 7.140908397945649e-06,
      "loss": 0.9595,
      "step": 19420
    },
    {
      "epoch": 1.1742309784250922,
      "grad_norm": 1.0741905300725103,
      "learning_rate": 7.13789779304023e-06,
      "loss": 0.951,
      "step": 19430
    },
    {
      "epoch": 1.174835317580226,
      "grad_norm": 1.0680461873299278,
      "learning_rate": 7.13488623933211e-06,
      "loss": 0.9613,
      "step": 19440
    },
    {
      "epoch": 1.1754396567353598,
      "grad_norm": 1.057180064547026,
      "learning_rate": 7.131873738157823e-06,
      "loss": 0.9306,
      "step": 19450
    },
    {
      "epoch": 1.1760439958904938,
      "grad_norm": 1.0007222482890263,
      "learning_rate": 7.128860290854321e-06,
      "loss": 0.9476,
      "step": 19460
    },
    {
      "epoch": 1.1766483350456276,
      "grad_norm": 1.0306466903209341,
      "learning_rate": 7.125845898758979e-06,
      "loss": 0.9239,
      "step": 19470
    },
    {
      "epoch": 1.1772526742007614,
      "grad_norm": 1.030204917596983,
      "learning_rate": 7.122830563209586e-06,
      "loss": 0.9203,
      "step": 19480
    },
    {
      "epoch": 1.1778570133558954,
      "grad_norm": 1.0916464812747928,
      "learning_rate": 7.119814285544355e-06,
      "loss": 0.9428,
      "step": 19490
    },
    {
      "epoch": 1.1784613525110292,
      "grad_norm": 1.113306921597439,
      "learning_rate": 7.116797067101917e-06,
      "loss": 0.9497,
      "step": 19500
    },
    {
      "epoch": 1.179065691666163,
      "grad_norm": 1.1805570984543383,
      "learning_rate": 7.113778909221316e-06,
      "loss": 0.9494,
      "step": 19510
    },
    {
      "epoch": 1.1796700308212968,
      "grad_norm": 1.1938755354106976,
      "learning_rate": 7.110759813242017e-06,
      "loss": 0.9199,
      "step": 19520
    },
    {
      "epoch": 1.1802743699764306,
      "grad_norm": 1.1266511284695175,
      "learning_rate": 7.1077397805038995e-06,
      "loss": 0.9359,
      "step": 19530
    },
    {
      "epoch": 1.1808787091315647,
      "grad_norm": 1.1392131018580411,
      "learning_rate": 7.10471881234726e-06,
      "loss": 0.9194,
      "step": 19540
    },
    {
      "epoch": 1.1814830482866985,
      "grad_norm": 0.9726978558822639,
      "learning_rate": 7.101696910112806e-06,
      "loss": 0.9195,
      "step": 19550
    },
    {
      "epoch": 1.1820873874418323,
      "grad_norm": 0.9318360286868769,
      "learning_rate": 7.0986740751416675e-06,
      "loss": 0.925,
      "step": 19560
    },
    {
      "epoch": 1.1826917265969663,
      "grad_norm": 0.9999454546800447,
      "learning_rate": 7.095650308775381e-06,
      "loss": 0.9658,
      "step": 19570
    },
    {
      "epoch": 1.1832960657521001,
      "grad_norm": 0.9926969226682162,
      "learning_rate": 7.092625612355901e-06,
      "loss": 0.944,
      "step": 19580
    },
    {
      "epoch": 1.183900404907234,
      "grad_norm": 0.9413123629939818,
      "learning_rate": 7.08959998722559e-06,
      "loss": 0.9432,
      "step": 19590
    },
    {
      "epoch": 1.1845047440623677,
      "grad_norm": 1.0186516597912494,
      "learning_rate": 7.086573434727231e-06,
      "loss": 0.9319,
      "step": 19600
    },
    {
      "epoch": 1.1851090832175017,
      "grad_norm": 1.0415123348345905,
      "learning_rate": 7.083545956204006e-06,
      "loss": 0.9109,
      "step": 19610
    },
    {
      "epoch": 1.1857134223726356,
      "grad_norm": 1.0024155603339588,
      "learning_rate": 7.08051755299952e-06,
      "loss": 0.9388,
      "step": 19620
    },
    {
      "epoch": 1.1863177615277694,
      "grad_norm": 1.1673005148467968,
      "learning_rate": 7.077488226457784e-06,
      "loss": 0.909,
      "step": 19630
    },
    {
      "epoch": 1.1869221006829032,
      "grad_norm": 1.0260071759869203,
      "learning_rate": 7.074457977923214e-06,
      "loss": 0.9143,
      "step": 19640
    },
    {
      "epoch": 1.1875264398380372,
      "grad_norm": 1.0489411635023616,
      "learning_rate": 7.071426808740644e-06,
      "loss": 0.9245,
      "step": 19650
    },
    {
      "epoch": 1.188130778993171,
      "grad_norm": 1.0212859157356728,
      "learning_rate": 7.068394720255311e-06,
      "loss": 0.919,
      "step": 19660
    },
    {
      "epoch": 1.1887351181483048,
      "grad_norm": 1.0004858463039392,
      "learning_rate": 7.065361713812859e-06,
      "loss": 0.9558,
      "step": 19670
    },
    {
      "epoch": 1.1893394573034386,
      "grad_norm": 1.0055819914161925,
      "learning_rate": 7.062327790759344e-06,
      "loss": 0.9453,
      "step": 19680
    },
    {
      "epoch": 1.1899437964585726,
      "grad_norm": 0.9899774815858461,
      "learning_rate": 7.059292952441224e-06,
      "loss": 0.9371,
      "step": 19690
    },
    {
      "epoch": 1.1905481356137064,
      "grad_norm": 1.2108514969546031,
      "learning_rate": 7.056257200205367e-06,
      "loss": 0.9425,
      "step": 19700
    },
    {
      "epoch": 1.1911524747688402,
      "grad_norm": 1.1980472870625456,
      "learning_rate": 7.053220535399043e-06,
      "loss": 0.9336,
      "step": 19710
    },
    {
      "epoch": 1.1917568139239743,
      "grad_norm": 1.1027450959423797,
      "learning_rate": 7.050182959369931e-06,
      "loss": 0.9447,
      "step": 19720
    },
    {
      "epoch": 1.192361153079108,
      "grad_norm": 1.1906947204228908,
      "learning_rate": 7.047144473466114e-06,
      "loss": 0.9286,
      "step": 19730
    },
    {
      "epoch": 1.1929654922342419,
      "grad_norm": 1.1083372382582157,
      "learning_rate": 7.044105079036075e-06,
      "loss": 0.9604,
      "step": 19740
    },
    {
      "epoch": 1.1935698313893757,
      "grad_norm": 1.0228367065038593,
      "learning_rate": 7.0410647774287e-06,
      "loss": 0.9332,
      "step": 19750
    },
    {
      "epoch": 1.1941741705445095,
      "grad_norm": 1.0723768135489609,
      "learning_rate": 7.038023569993284e-06,
      "loss": 0.9349,
      "step": 19760
    },
    {
      "epoch": 1.1947785096996435,
      "grad_norm": 1.0704030557256032,
      "learning_rate": 7.034981458079519e-06,
      "loss": 0.9431,
      "step": 19770
    },
    {
      "epoch": 1.1953828488547773,
      "grad_norm": 1.0754003471430496,
      "learning_rate": 7.0319384430375e-06,
      "loss": 0.9096,
      "step": 19780
    },
    {
      "epoch": 1.195987188009911,
      "grad_norm": 1.0611102234186296,
      "learning_rate": 7.028894526217722e-06,
      "loss": 0.9428,
      "step": 19790
    },
    {
      "epoch": 1.1965915271650451,
      "grad_norm": 1.2388424023247573,
      "learning_rate": 7.02584970897108e-06,
      "loss": 0.9018,
      "step": 19800
    },
    {
      "epoch": 1.197195866320179,
      "grad_norm": 1.243105411071229,
      "learning_rate": 7.022803992648867e-06,
      "loss": 0.922,
      "step": 19810
    },
    {
      "epoch": 1.1978002054753127,
      "grad_norm": 1.1943670106298478,
      "learning_rate": 7.01975737860278e-06,
      "loss": 0.9322,
      "step": 19820
    },
    {
      "epoch": 1.1984045446304465,
      "grad_norm": 1.2435586994894567,
      "learning_rate": 7.016709868184911e-06,
      "loss": 0.9354,
      "step": 19830
    },
    {
      "epoch": 1.1990088837855803,
      "grad_norm": 1.2061295025068992,
      "learning_rate": 7.013661462747751e-06,
      "loss": 0.9454,
      "step": 19840
    },
    {
      "epoch": 1.1996132229407144,
      "grad_norm": 1.3949446935351246,
      "learning_rate": 7.010612163644186e-06,
      "loss": 0.9441,
      "step": 19850
    },
    {
      "epoch": 1.2002175620958482,
      "grad_norm": 1.4460511324635015,
      "learning_rate": 7.007561972227501e-06,
      "loss": 0.9383,
      "step": 19860
    },
    {
      "epoch": 1.200821901250982,
      "grad_norm": 1.4539877051434926,
      "learning_rate": 7.004510889851376e-06,
      "loss": 0.9406,
      "step": 19870
    },
    {
      "epoch": 1.201426240406116,
      "grad_norm": 1.4008324687161817,
      "learning_rate": 7.001458917869887e-06,
      "loss": 0.9286,
      "step": 19880
    },
    {
      "epoch": 1.2020305795612498,
      "grad_norm": 1.4221315242319108,
      "learning_rate": 6.998406057637503e-06,
      "loss": 0.9516,
      "step": 19890
    },
    {
      "epoch": 1.2026349187163836,
      "grad_norm": 0.8890058067800306,
      "learning_rate": 6.995352310509091e-06,
      "loss": 0.9411,
      "step": 19900
    },
    {
      "epoch": 1.2032392578715174,
      "grad_norm": 0.8392562994216797,
      "learning_rate": 6.992297677839908e-06,
      "loss": 0.928,
      "step": 19910
    },
    {
      "epoch": 1.2038435970266514,
      "grad_norm": 0.8909889995816207,
      "learning_rate": 6.989242160985603e-06,
      "loss": 0.9388,
      "step": 19920
    },
    {
      "epoch": 1.2044479361817853,
      "grad_norm": 0.8642233967849721,
      "learning_rate": 6.986185761302224e-06,
      "loss": 0.9408,
      "step": 19930
    },
    {
      "epoch": 1.205052275336919,
      "grad_norm": 0.9460195355213684,
      "learning_rate": 6.983128480146203e-06,
      "loss": 0.9324,
      "step": 19940
    },
    {
      "epoch": 1.2056566144920529,
      "grad_norm": 0.8029051507326859,
      "learning_rate": 6.980070318874367e-06,
      "loss": 0.9353,
      "step": 19950
    },
    {
      "epoch": 1.2062609536471869,
      "grad_norm": 0.7858799944406785,
      "learning_rate": 6.977011278843934e-06,
      "loss": 0.9214,
      "step": 19960
    },
    {
      "epoch": 1.2068652928023207,
      "grad_norm": 0.7510011450562873,
      "learning_rate": 6.973951361412512e-06,
      "loss": 0.9139,
      "step": 19970
    },
    {
      "epoch": 1.2074696319574545,
      "grad_norm": 0.718384309839229,
      "learning_rate": 6.9708905679380956e-06,
      "loss": 0.9307,
      "step": 19980
    },
    {
      "epoch": 1.2080739711125883,
      "grad_norm": 0.7283282319790186,
      "learning_rate": 6.9678288997790725e-06,
      "loss": 0.9201,
      "step": 19990
    },
    {
      "epoch": 1.2086783102677223,
      "grad_norm": 0.8121757501780279,
      "learning_rate": 6.964766358294213e-06,
      "loss": 0.9533,
      "step": 20000
    },
    {
      "epoch": 1.2092826494228561,
      "grad_norm": 0.8782745966915562,
      "learning_rate": 6.961702944842682e-06,
      "loss": 0.9491,
      "step": 20010
    },
    {
      "epoch": 1.20988698857799,
      "grad_norm": 0.8059374276263742,
      "learning_rate": 6.958638660784026e-06,
      "loss": 0.9114,
      "step": 20020
    },
    {
      "epoch": 1.2104913277331237,
      "grad_norm": 0.7598256207588705,
      "learning_rate": 6.955573507478179e-06,
      "loss": 0.9135,
      "step": 20030
    },
    {
      "epoch": 1.2110956668882578,
      "grad_norm": 0.891771194767523,
      "learning_rate": 6.952507486285462e-06,
      "loss": 0.9427,
      "step": 20040
    },
    {
      "epoch": 1.2117000060433916,
      "grad_norm": 0.5764111661821675,
      "learning_rate": 6.94944059856658e-06,
      "loss": 0.9314,
      "step": 20050
    },
    {
      "epoch": 1.2123043451985254,
      "grad_norm": 0.5604924088026544,
      "learning_rate": 6.946372845682622e-06,
      "loss": 0.9319,
      "step": 20060
    },
    {
      "epoch": 1.2129086843536592,
      "grad_norm": 0.5784348320334823,
      "learning_rate": 6.943304228995064e-06,
      "loss": 0.934,
      "step": 20070
    },
    {
      "epoch": 1.2135130235087932,
      "grad_norm": 0.5956243090438809,
      "learning_rate": 6.940234749865763e-06,
      "loss": 0.9385,
      "step": 20080
    },
    {
      "epoch": 1.214117362663927,
      "grad_norm": 0.5837537568852901,
      "learning_rate": 6.937164409656958e-06,
      "loss": 0.9306,
      "step": 20090
    },
    {
      "epoch": 1.2147217018190608,
      "grad_norm": 0.6100500608247255,
      "learning_rate": 6.934093209731272e-06,
      "loss": 0.9442,
      "step": 20100
    },
    {
      "epoch": 1.2153260409741948,
      "grad_norm": 0.6824101665164553,
      "learning_rate": 6.931021151451709e-06,
      "loss": 0.9435,
      "step": 20110
    },
    {
      "epoch": 1.2159303801293286,
      "grad_norm": 0.6058840737726875,
      "learning_rate": 6.9279482361816516e-06,
      "loss": 0.9555,
      "step": 20120
    },
    {
      "epoch": 1.2165347192844624,
      "grad_norm": 0.620996171237021,
      "learning_rate": 6.924874465284869e-06,
      "loss": 0.9101,
      "step": 20130
    },
    {
      "epoch": 1.2171390584395962,
      "grad_norm": 0.6179480078942203,
      "learning_rate": 6.921799840125503e-06,
      "loss": 0.9458,
      "step": 20140
    },
    {
      "epoch": 1.21774339759473,
      "grad_norm": 0.8318185655300202,
      "learning_rate": 6.9187243620680765e-06,
      "loss": 0.9352,
      "step": 20150
    },
    {
      "epoch": 1.218347736749864,
      "grad_norm": 0.8301274637834942,
      "learning_rate": 6.9156480324774946e-06,
      "loss": 0.9199,
      "step": 20160
    },
    {
      "epoch": 1.2189520759049979,
      "grad_norm": 0.7714690574444714,
      "learning_rate": 6.912570852719036e-06,
      "loss": 0.9414,
      "step": 20170
    },
    {
      "epoch": 1.2195564150601317,
      "grad_norm": 0.8167329708260073,
      "learning_rate": 6.90949282415836e-06,
      "loss": 0.9314,
      "step": 20180
    },
    {
      "epoch": 1.2201607542152657,
      "grad_norm": 0.7844816059194251,
      "learning_rate": 6.9064139481615e-06,
      "loss": 0.9349,
      "step": 20190
    },
    {
      "epoch": 1.2207650933703995,
      "grad_norm": 0.5359931103425383,
      "learning_rate": 6.903334226094864e-06,
      "loss": 0.9443,
      "step": 20200
    },
    {
      "epoch": 1.2213694325255333,
      "grad_norm": 0.5407871108862375,
      "learning_rate": 6.900253659325242e-06,
      "loss": 0.9313,
      "step": 20210
    },
    {
      "epoch": 1.2219737716806671,
      "grad_norm": 0.5836312461149332,
      "learning_rate": 6.8971722492197934e-06,
      "loss": 0.9441,
      "step": 20220
    },
    {
      "epoch": 1.2225781108358011,
      "grad_norm": 0.5717076627819502,
      "learning_rate": 6.894089997146052e-06,
      "loss": 0.9215,
      "step": 20230
    },
    {
      "epoch": 1.223182449990935,
      "grad_norm": 0.5421335823155865,
      "learning_rate": 6.891006904471927e-06,
      "loss": 0.9224,
      "step": 20240
    },
    {
      "epoch": 1.2237867891460688,
      "grad_norm": 0.5283379070988189,
      "learning_rate": 6.887922972565702e-06,
      "loss": 0.9173,
      "step": 20250
    },
    {
      "epoch": 1.2243911283012026,
      "grad_norm": 0.515825534070461,
      "learning_rate": 6.884838202796028e-06,
      "loss": 0.9275,
      "step": 20260
    },
    {
      "epoch": 1.2249954674563366,
      "grad_norm": 0.5889256599898429,
      "learning_rate": 6.8817525965319345e-06,
      "loss": 0.9368,
      "step": 20270
    },
    {
      "epoch": 1.2255998066114704,
      "grad_norm": 0.5372474037594597,
      "learning_rate": 6.878666155142818e-06,
      "loss": 0.9296,
      "step": 20280
    },
    {
      "epoch": 1.2262041457666042,
      "grad_norm": 0.5077302684028948,
      "learning_rate": 6.8755788799984435e-06,
      "loss": 0.9367,
      "step": 20290
    },
    {
      "epoch": 1.226808484921738,
      "grad_norm": 0.5538548884538033,
      "learning_rate": 6.872490772468954e-06,
      "loss": 0.9261,
      "step": 20300
    },
    {
      "epoch": 1.227412824076872,
      "grad_norm": 0.5595078322602588,
      "learning_rate": 6.869401833924853e-06,
      "loss": 0.9317,
      "step": 20310
    },
    {
      "epoch": 1.2280171632320058,
      "grad_norm": 0.5356233206774714,
      "learning_rate": 6.866312065737018e-06,
      "loss": 0.9188,
      "step": 20320
    },
    {
      "epoch": 1.2286215023871396,
      "grad_norm": 0.5282016800917761,
      "learning_rate": 6.863221469276693e-06,
      "loss": 0.9466,
      "step": 20330
    },
    {
      "epoch": 1.2292258415422734,
      "grad_norm": 0.5504446506700361,
      "learning_rate": 6.8601300459154895e-06,
      "loss": 0.9336,
      "step": 20340
    },
    {
      "epoch": 1.2298301806974075,
      "grad_norm": 0.5503710980562528,
      "learning_rate": 6.8570377970253885e-06,
      "loss": 0.9008,
      "step": 20350
    },
    {
      "epoch": 1.2304345198525413,
      "grad_norm": 0.5133951305911965,
      "learning_rate": 6.853944723978735e-06,
      "loss": 0.9251,
      "step": 20360
    },
    {
      "epoch": 1.231038859007675,
      "grad_norm": 0.6032558781030611,
      "learning_rate": 6.8508508281482365e-06,
      "loss": 0.9208,
      "step": 20370
    },
    {
      "epoch": 1.2316431981628089,
      "grad_norm": 0.5318308450247039,
      "learning_rate": 6.847756110906974e-06,
      "loss": 0.9241,
      "step": 20380
    },
    {
      "epoch": 1.232247537317943,
      "grad_norm": 0.5349276043564706,
      "learning_rate": 6.844660573628387e-06,
      "loss": 0.925,
      "step": 20390
    },
    {
      "epoch": 1.2328518764730767,
      "grad_norm": 0.5965937715615065,
      "learning_rate": 6.841564217686279e-06,
      "loss": 0.9321,
      "step": 20400
    },
    {
      "epoch": 1.2334562156282105,
      "grad_norm": 0.5105433259348409,
      "learning_rate": 6.838467044454819e-06,
      "loss": 0.9255,
      "step": 20410
    },
    {
      "epoch": 1.2340605547833445,
      "grad_norm": 0.5401507283057976,
      "learning_rate": 6.835369055308536e-06,
      "loss": 0.9587,
      "step": 20420
    },
    {
      "epoch": 1.2346648939384783,
      "grad_norm": 0.529052759553093,
      "learning_rate": 6.832270251622326e-06,
      "loss": 0.9293,
      "step": 20430
    },
    {
      "epoch": 1.2352692330936121,
      "grad_norm": 0.6063008576277052,
      "learning_rate": 6.829170634771442e-06,
      "loss": 0.9491,
      "step": 20440
    },
    {
      "epoch": 1.235873572248746,
      "grad_norm": 0.5676452156183475,
      "learning_rate": 6.826070206131498e-06,
      "loss": 0.947,
      "step": 20450
    },
    {
      "epoch": 1.2364779114038797,
      "grad_norm": 0.5662977986267751,
      "learning_rate": 6.8229689670784695e-06,
      "loss": 0.9373,
      "step": 20460
    },
    {
      "epoch": 1.2370822505590138,
      "grad_norm": 0.531159891950673,
      "learning_rate": 6.8198669189886935e-06,
      "loss": 0.9183,
      "step": 20470
    },
    {
      "epoch": 1.2376865897141476,
      "grad_norm": 0.5932563623654429,
      "learning_rate": 6.8167640632388644e-06,
      "loss": 0.911,
      "step": 20480
    },
    {
      "epoch": 1.2382909288692814,
      "grad_norm": 0.5198335034293413,
      "learning_rate": 6.813660401206033e-06,
      "loss": 0.9282,
      "step": 20490
    },
    {
      "epoch": 1.2388952680244154,
      "grad_norm": 0.5938794346616102,
      "learning_rate": 6.810555934267611e-06,
      "loss": 0.9418,
      "step": 20500
    },
    {
      "epoch": 1.2394996071795492,
      "grad_norm": 0.6935352548280033,
      "learning_rate": 6.807450663801365e-06,
      "loss": 0.9421,
      "step": 20510
    },
    {
      "epoch": 1.240103946334683,
      "grad_norm": 0.6198638091006703,
      "learning_rate": 6.804344591185422e-06,
      "loss": 0.9375,
      "step": 20520
    },
    {
      "epoch": 1.2407082854898168,
      "grad_norm": 0.6465662753867024,
      "learning_rate": 6.8012377177982605e-06,
      "loss": 0.942,
      "step": 20530
    },
    {
      "epoch": 1.2413126246449506,
      "grad_norm": 0.622290026542129,
      "learning_rate": 6.798130045018715e-06,
      "loss": 0.9313,
      "step": 20540
    },
    {
      "epoch": 1.2419169638000847,
      "grad_norm": 0.7142818767669845,
      "learning_rate": 6.795021574225978e-06,
      "loss": 0.9381,
      "step": 20550
    },
    {
      "epoch": 1.2425213029552185,
      "grad_norm": 0.6939265203905661,
      "learning_rate": 6.791912306799594e-06,
      "loss": 0.944,
      "step": 20560
    },
    {
      "epoch": 1.2431256421103523,
      "grad_norm": 0.7295199942711916,
      "learning_rate": 6.7888022441194615e-06,
      "loss": 0.9421,
      "step": 20570
    },
    {
      "epoch": 1.2437299812654863,
      "grad_norm": 0.6712086757588075,
      "learning_rate": 6.785691387565829e-06,
      "loss": 0.9615,
      "step": 20580
    },
    {
      "epoch": 1.24433432042062,
      "grad_norm": 0.7734944438436397,
      "learning_rate": 6.782579738519304e-06,
      "loss": 0.9538,
      "step": 20590
    },
    {
      "epoch": 1.244938659575754,
      "grad_norm": 0.6285076237184368,
      "learning_rate": 6.779467298360837e-06,
      "loss": 0.9103,
      "step": 20600
    },
    {
      "epoch": 1.2455429987308877,
      "grad_norm": 0.6226915890652817,
      "learning_rate": 6.776354068471738e-06,
      "loss": 0.9355,
      "step": 20610
    },
    {
      "epoch": 1.2461473378860217,
      "grad_norm": 0.6456314600247918,
      "learning_rate": 6.7732400502336625e-06,
      "loss": 0.9304,
      "step": 20620
    },
    {
      "epoch": 1.2467516770411555,
      "grad_norm": 0.6399642965634079,
      "learning_rate": 6.770125245028617e-06,
      "loss": 0.928,
      "step": 20630
    },
    {
      "epoch": 1.2473560161962893,
      "grad_norm": 0.6213078373894296,
      "learning_rate": 6.7670096542389576e-06,
      "loss": 0.9167,
      "step": 20640
    },
    {
      "epoch": 1.2479603553514231,
      "grad_norm": 0.7801327277490291,
      "learning_rate": 6.763893279247389e-06,
      "loss": 0.9331,
      "step": 20650
    },
    {
      "epoch": 1.2485646945065572,
      "grad_norm": 0.7996747131481201,
      "learning_rate": 6.760776121436963e-06,
      "loss": 0.9478,
      "step": 20660
    },
    {
      "epoch": 1.249169033661691,
      "grad_norm": 0.8265767279808282,
      "learning_rate": 6.75765818219108e-06,
      "loss": 0.9329,
      "step": 20670
    },
    {
      "epoch": 1.2497733728168248,
      "grad_norm": 0.75948255351935,
      "learning_rate": 6.754539462893488e-06,
      "loss": 0.9371,
      "step": 20680
    },
    {
      "epoch": 1.2503777119719586,
      "grad_norm": 0.8198662867232713,
      "learning_rate": 6.751419964928279e-06,
      "loss": 0.9294,
      "step": 20690
    },
    {
      "epoch": 1.2509820511270926,
      "grad_norm": 0.7870382637790324,
      "learning_rate": 6.74829968967989e-06,
      "loss": 0.9589,
      "step": 20700
    },
    {
      "epoch": 1.2515863902822264,
      "grad_norm": 0.7670199166420851,
      "learning_rate": 6.745178638533106e-06,
      "loss": 0.9447,
      "step": 20710
    },
    {
      "epoch": 1.2521907294373602,
      "grad_norm": 0.7902133057307972,
      "learning_rate": 6.7420568128730545e-06,
      "loss": 0.9126,
      "step": 20720
    },
    {
      "epoch": 1.2527950685924942,
      "grad_norm": 0.7974377089002933,
      "learning_rate": 6.738934214085209e-06,
      "loss": 0.9413,
      "step": 20730
    },
    {
      "epoch": 1.253399407747628,
      "grad_norm": 0.7715117234540104,
      "learning_rate": 6.7358108435553804e-06,
      "loss": 0.9368,
      "step": 20740
    },
    {
      "epoch": 1.2540037469027618,
      "grad_norm": 1.20003798642118,
      "learning_rate": 6.732686702669729e-06,
      "loss": 0.9557,
      "step": 20750
    },
    {
      "epoch": 1.2546080860578956,
      "grad_norm": 1.2226903386953512,
      "learning_rate": 6.729561792814753e-06,
      "loss": 0.9278,
      "step": 20760
    },
    {
      "epoch": 1.2552124252130294,
      "grad_norm": 1.2217402732192495,
      "learning_rate": 6.726436115377291e-06,
      "loss": 0.9397,
      "step": 20770
    },
    {
      "epoch": 1.2558167643681635,
      "grad_norm": 1.2137601271828522,
      "learning_rate": 6.723309671744526e-06,
      "loss": 0.9468,
      "step": 20780
    },
    {
      "epoch": 1.2564211035232973,
      "grad_norm": 1.2663172756265997,
      "learning_rate": 6.7201824633039816e-06,
      "loss": 0.9395,
      "step": 20790
    },
    {
      "epoch": 1.257025442678431,
      "grad_norm": 0.9863430977961953,
      "learning_rate": 6.717054491443511e-06,
      "loss": 0.9478,
      "step": 20800
    },
    {
      "epoch": 1.257629781833565,
      "grad_norm": 1.061233152512319,
      "learning_rate": 6.7139257575513206e-06,
      "loss": 0.9326,
      "step": 20810
    },
    {
      "epoch": 1.258234120988699,
      "grad_norm": 1.0805984098847896,
      "learning_rate": 6.710796263015944e-06,
      "loss": 0.9586,
      "step": 20820
    },
    {
      "epoch": 1.2588384601438327,
      "grad_norm": 1.1130806362391232,
      "learning_rate": 6.707666009226258e-06,
      "loss": 0.9422,
      "step": 20830
    },
    {
      "epoch": 1.2594427992989665,
      "grad_norm": 0.9591251379760726,
      "learning_rate": 6.704534997571473e-06,
      "loss": 0.9274,
      "step": 20840
    },
    {
      "epoch": 1.2600471384541003,
      "grad_norm": 2.5530808287871634,
      "learning_rate": 6.701403229441138e-06,
      "loss": 0.9289,
      "step": 20850
    },
    {
      "epoch": 1.2606514776092343,
      "grad_norm": 2.4412256026625485,
      "learning_rate": 6.698270706225137e-06,
      "loss": 0.9335,
      "step": 20860
    },
    {
      "epoch": 1.2612558167643682,
      "grad_norm": 2.5725695170307636,
      "learning_rate": 6.695137429313691e-06,
      "loss": 0.8941,
      "step": 20870
    },
    {
      "epoch": 1.261860155919502,
      "grad_norm": 2.3312643495843393,
      "learning_rate": 6.69200340009735e-06,
      "loss": 0.927,
      "step": 20880
    },
    {
      "epoch": 1.262464495074636,
      "grad_norm": 3.1210480975513413,
      "learning_rate": 6.6888686199670035e-06,
      "loss": 0.9336,
      "step": 20890
    },
    {
      "epoch": 1.2630688342297698,
      "grad_norm": 0.9626136990919115,
      "learning_rate": 6.6857330903138754e-06,
      "loss": 0.9393,
      "step": 20900
    },
    {
      "epoch": 1.2636731733849036,
      "grad_norm": 1.1870849395147782,
      "learning_rate": 6.6825968125295125e-06,
      "loss": 0.936,
      "step": 20910
    },
    {
      "epoch": 1.2642775125400374,
      "grad_norm": 0.8252889775966359,
      "learning_rate": 6.679459788005805e-06,
      "loss": 0.9573,
      "step": 20920
    },
    {
      "epoch": 1.2648818516951712,
      "grad_norm": 0.9435531385065613,
      "learning_rate": 6.67632201813497e-06,
      "loss": 0.9324,
      "step": 20930
    },
    {
      "epoch": 1.2654861908503052,
      "grad_norm": 1.0158470995515982,
      "learning_rate": 6.673183504309553e-06,
      "loss": 0.9399,
      "step": 20940
    },
    {
      "epoch": 1.266090530005439,
      "grad_norm": 0.9879970036591537,
      "learning_rate": 6.670044247922431e-06,
      "loss": 0.9178,
      "step": 20950
    },
    {
      "epoch": 1.2666948691605728,
      "grad_norm": 1.0109463775278411,
      "learning_rate": 6.666904250366815e-06,
      "loss": 0.9419,
      "step": 20960
    },
    {
      "epoch": 1.2672992083157069,
      "grad_norm": 1.1555863088766083,
      "learning_rate": 6.663763513036237e-06,
      "loss": 0.9269,
      "step": 20970
    },
    {
      "epoch": 1.2679035474708407,
      "grad_norm": 0.9613787263799334,
      "learning_rate": 6.660622037324566e-06,
      "loss": 0.912,
      "step": 20980
    },
    {
      "epoch": 1.2685078866259745,
      "grad_norm": 1.0560038276585342,
      "learning_rate": 6.657479824625989e-06,
      "loss": 0.9286,
      "step": 20990
    },
    {
      "epoch": 1.2691122257811083,
      "grad_norm": 1.074153064961318,
      "learning_rate": 6.6543368763350315e-06,
      "loss": 0.9275,
      "step": 21000
    },
    {
      "epoch": 1.2697165649362423,
      "grad_norm": 0.9760807118304226,
      "learning_rate": 6.651193193846535e-06,
      "loss": 0.9269,
      "step": 21010
    },
    {
      "epoch": 1.270320904091376,
      "grad_norm": 1.058068280322029,
      "learning_rate": 6.648048778555672e-06,
      "loss": 0.9358,
      "step": 21020
    },
    {
      "epoch": 1.27092524324651,
      "grad_norm": 1.0926792005542167,
      "learning_rate": 6.644903631857941e-06,
      "loss": 0.9439,
      "step": 21030
    },
    {
      "epoch": 1.271529582401644,
      "grad_norm": 1.071420110327903,
      "learning_rate": 6.641757755149161e-06,
      "loss": 0.9275,
      "step": 21040
    },
    {
      "epoch": 1.2721339215567777,
      "grad_norm": 1.0070963072937444,
      "learning_rate": 6.638611149825479e-06,
      "loss": 0.9159,
      "step": 21050
    },
    {
      "epoch": 1.2727382607119115,
      "grad_norm": 1.040406979699207,
      "learning_rate": 6.635463817283365e-06,
      "loss": 0.9322,
      "step": 21060
    },
    {
      "epoch": 1.2733425998670453,
      "grad_norm": 1.0978359394302144,
      "learning_rate": 6.632315758919608e-06,
      "loss": 0.9459,
      "step": 21070
    },
    {
      "epoch": 1.2739469390221791,
      "grad_norm": 1.058434784024176,
      "learning_rate": 6.629166976131324e-06,
      "loss": 0.9314,
      "step": 21080
    },
    {
      "epoch": 1.2745512781773132,
      "grad_norm": 1.036345841163535,
      "learning_rate": 6.6260174703159465e-06,
      "loss": 0.924,
      "step": 21090
    },
    {
      "epoch": 1.275155617332447,
      "grad_norm": 1.154005199065569,
      "learning_rate": 6.622867242871231e-06,
      "loss": 0.9297,
      "step": 21100
    },
    {
      "epoch": 1.2757599564875808,
      "grad_norm": 1.1524729230836825,
      "learning_rate": 6.619716295195256e-06,
      "loss": 0.9265,
      "step": 21110
    },
    {
      "epoch": 1.2763642956427148,
      "grad_norm": 1.1713282910504288,
      "learning_rate": 6.616564628686417e-06,
      "loss": 0.911,
      "step": 21120
    },
    {
      "epoch": 1.2769686347978486,
      "grad_norm": 1.1139782251033674,
      "learning_rate": 6.613412244743428e-06,
      "loss": 0.9454,
      "step": 21130
    },
    {
      "epoch": 1.2775729739529824,
      "grad_norm": 1.1869427759294304,
      "learning_rate": 6.610259144765322e-06,
      "loss": 0.9345,
      "step": 21140
    },
    {
      "epoch": 1.2781773131081162,
      "grad_norm": 0.978351285050308,
      "learning_rate": 6.607105330151452e-06,
      "loss": 0.9123,
      "step": 21150
    },
    {
      "epoch": 1.27878165226325,
      "grad_norm": 1.0021079532976485,
      "learning_rate": 6.603950802301485e-06,
      "loss": 0.9241,
      "step": 21160
    },
    {
      "epoch": 1.279385991418384,
      "grad_norm": 1.0861909583578744,
      "learning_rate": 6.600795562615405e-06,
      "loss": 0.9244,
      "step": 21170
    },
    {
      "epoch": 1.2799903305735179,
      "grad_norm": 1.0018027000194207,
      "learning_rate": 6.597639612493516e-06,
      "loss": 0.923,
      "step": 21180
    },
    {
      "epoch": 1.2805946697286517,
      "grad_norm": 1.0369608770113576,
      "learning_rate": 6.594482953336429e-06,
      "loss": 0.9205,
      "step": 21190
    },
    {
      "epoch": 1.2811990088837857,
      "grad_norm": 1.0018049905608608,
      "learning_rate": 6.591325586545081e-06,
      "loss": 0.9666,
      "step": 21200
    },
    {
      "epoch": 1.2818033480389195,
      "grad_norm": 1.0542119094067566,
      "learning_rate": 6.588167513520714e-06,
      "loss": 0.936,
      "step": 21210
    },
    {
      "epoch": 1.2824076871940533,
      "grad_norm": 1.0257850303118394,
      "learning_rate": 6.585008735664884e-06,
      "loss": 0.9171,
      "step": 21220
    },
    {
      "epoch": 1.283012026349187,
      "grad_norm": 1.0629264313566607,
      "learning_rate": 6.581849254379464e-06,
      "loss": 0.9407,
      "step": 21230
    },
    {
      "epoch": 1.283616365504321,
      "grad_norm": 1.0645272350209467,
      "learning_rate": 6.578689071066637e-06,
      "loss": 0.9174,
      "step": 21240
    },
    {
      "epoch": 1.284220704659455,
      "grad_norm": 0.9942172899664239,
      "learning_rate": 6.575528187128897e-06,
      "loss": 0.9428,
      "step": 21250
    },
    {
      "epoch": 1.2848250438145887,
      "grad_norm": 0.9958710522566719,
      "learning_rate": 6.5723666039690515e-06,
      "loss": 0.9182,
      "step": 21260
    },
    {
      "epoch": 1.2854293829697225,
      "grad_norm": 0.9543256694020993,
      "learning_rate": 6.569204322990216e-06,
      "loss": 0.943,
      "step": 21270
    },
    {
      "epoch": 1.2860337221248566,
      "grad_norm": 0.9939418030273146,
      "learning_rate": 6.566041345595814e-06,
      "loss": 0.9319,
      "step": 21280
    },
    {
      "epoch": 1.2866380612799904,
      "grad_norm": 1.0547380875161039,
      "learning_rate": 6.562877673189583e-06,
      "loss": 0.9376,
      "step": 21290
    },
    {
      "epoch": 1.2872424004351242,
      "grad_norm": 1.171081254921601,
      "learning_rate": 6.5597133071755655e-06,
      "loss": 0.9326,
      "step": 21300
    },
    {
      "epoch": 1.287846739590258,
      "grad_norm": 1.1725934465440044,
      "learning_rate": 6.5565482489581115e-06,
      "loss": 0.9466,
      "step": 21310
    },
    {
      "epoch": 1.2884510787453918,
      "grad_norm": 1.2326777667202249,
      "learning_rate": 6.55338249994188e-06,
      "loss": 0.9367,
      "step": 21320
    },
    {
      "epoch": 1.2890554179005258,
      "grad_norm": 1.1597071943780024,
      "learning_rate": 6.550216061531837e-06,
      "loss": 0.9361,
      "step": 21330
    },
    {
      "epoch": 1.2896597570556596,
      "grad_norm": 1.095543226881079,
      "learning_rate": 6.547048935133251e-06,
      "loss": 0.9294,
      "step": 21340
    },
    {
      "epoch": 1.2902640962107936,
      "grad_norm": 1.089831708969902,
      "learning_rate": 6.5438811221516994e-06,
      "loss": 0.9526,
      "step": 21350
    },
    {
      "epoch": 1.2908684353659274,
      "grad_norm": 1.0514813037236386,
      "learning_rate": 6.540712623993063e-06,
      "loss": 0.9274,
      "step": 21360
    },
    {
      "epoch": 1.2914727745210612,
      "grad_norm": 0.995499283392343,
      "learning_rate": 6.537543442063526e-06,
      "loss": 0.9406,
      "step": 21370
    },
    {
      "epoch": 1.292077113676195,
      "grad_norm": 1.1740467112075939,
      "learning_rate": 6.534373577769578e-06,
      "loss": 0.9463,
      "step": 21380
    },
    {
      "epoch": 1.2926814528313288,
      "grad_norm": 1.107057930823461,
      "learning_rate": 6.53120303251801e-06,
      "loss": 0.9476,
      "step": 21390
    },
    {
      "epoch": 1.2932857919864629,
      "grad_norm": 1.1227755034068285,
      "learning_rate": 6.528031807715913e-06,
      "loss": 0.9267,
      "step": 21400
    },
    {
      "epoch": 1.2938901311415967,
      "grad_norm": 1.1532011204220003,
      "learning_rate": 6.524859904770686e-06,
      "loss": 0.9186,
      "step": 21410
    },
    {
      "epoch": 1.2944944702967305,
      "grad_norm": 1.2236358389287254,
      "learning_rate": 6.521687325090023e-06,
      "loss": 0.9327,
      "step": 21420
    },
    {
      "epoch": 1.2950988094518645,
      "grad_norm": 1.2654715751089771,
      "learning_rate": 6.518514070081918e-06,
      "loss": 0.9343,
      "step": 21430
    },
    {
      "epoch": 1.2957031486069983,
      "grad_norm": 1.1260807209923318,
      "learning_rate": 6.5153401411546715e-06,
      "loss": 0.9198,
      "step": 21440
    },
    {
      "epoch": 1.2963074877621321,
      "grad_norm": 1.3782201507384069,
      "learning_rate": 6.512165539716875e-06,
      "loss": 0.9419,
      "step": 21450
    },
    {
      "epoch": 1.296911826917266,
      "grad_norm": 1.4566152824347334,
      "learning_rate": 6.508990267177424e-06,
      "loss": 0.9142,
      "step": 21460
    },
    {
      "epoch": 1.2975161660723997,
      "grad_norm": 1.3611489497109344,
      "learning_rate": 6.505814324945509e-06,
      "loss": 0.9649,
      "step": 21470
    },
    {
      "epoch": 1.2981205052275337,
      "grad_norm": 1.3857152810604922,
      "learning_rate": 6.502637714430619e-06,
      "loss": 0.9406,
      "step": 21480
    },
    {
      "epoch": 1.2987248443826676,
      "grad_norm": 1.4554878354894845,
      "learning_rate": 6.499460437042539e-06,
      "loss": 0.9316,
      "step": 21490
    },
    {
      "epoch": 1.2993291835378014,
      "grad_norm": 0.9060526311398772,
      "learning_rate": 6.496282494191351e-06,
      "loss": 0.9375,
      "step": 21500
    },
    {
      "epoch": 1.2999335226929354,
      "grad_norm": 0.8434628343707185,
      "learning_rate": 6.493103887287429e-06,
      "loss": 0.9107,
      "step": 21510
    },
    {
      "epoch": 1.3005378618480692,
      "grad_norm": 0.9087772227016638,
      "learning_rate": 6.489924617741446e-06,
      "loss": 0.931,
      "step": 21520
    },
    {
      "epoch": 1.301142201003203,
      "grad_norm": 0.8517540136578798,
      "learning_rate": 6.486744686964367e-06,
      "loss": 0.9192,
      "step": 21530
    },
    {
      "epoch": 1.3017465401583368,
      "grad_norm": 0.8659201869544333,
      "learning_rate": 6.483564096367452e-06,
      "loss": 0.9543,
      "step": 21540
    },
    {
      "epoch": 1.3023508793134706,
      "grad_norm": 0.7780229044615488,
      "learning_rate": 6.480382847362251e-06,
      "loss": 0.9592,
      "step": 21550
    },
    {
      "epoch": 1.3029552184686046,
      "grad_norm": 0.7487894351169517,
      "learning_rate": 6.477200941360606e-06,
      "loss": 0.9359,
      "step": 21560
    },
    {
      "epoch": 1.3035595576237384,
      "grad_norm": 0.7718657395863984,
      "learning_rate": 6.474018379774658e-06,
      "loss": 0.954,
      "step": 21570
    },
    {
      "epoch": 1.3041638967788722,
      "grad_norm": 0.7804785430559377,
      "learning_rate": 6.470835164016828e-06,
      "loss": 0.912,
      "step": 21580
    },
    {
      "epoch": 1.3047682359340063,
      "grad_norm": 0.7710334144819115,
      "learning_rate": 6.467651295499835e-06,
      "loss": 0.9562,
      "step": 21590
    },
    {
      "epoch": 1.30537257508914,
      "grad_norm": 0.812678079427488,
      "learning_rate": 6.464466775636684e-06,
      "loss": 0.9337,
      "step": 21600
    },
    {
      "epoch": 1.3059769142442739,
      "grad_norm": 0.8467816023389082,
      "learning_rate": 6.461281605840671e-06,
      "loss": 0.9598,
      "step": 21610
    },
    {
      "epoch": 1.3065812533994077,
      "grad_norm": 0.934919839022193,
      "learning_rate": 6.458095787525379e-06,
      "loss": 0.9299,
      "step": 21620
    },
    {
      "epoch": 1.3071855925545415,
      "grad_norm": 0.8199809488776156,
      "learning_rate": 6.454909322104682e-06,
      "loss": 0.9308,
      "step": 21630
    },
    {
      "epoch": 1.3077899317096755,
      "grad_norm": 0.8362409965440692,
      "learning_rate": 6.4517222109927365e-06,
      "loss": 0.9401,
      "step": 21640
    },
    {
      "epoch": 1.3083942708648093,
      "grad_norm": 0.5492779152089806,
      "learning_rate": 6.448534455603987e-06,
      "loss": 0.9186,
      "step": 21650
    },
    {
      "epoch": 1.308998610019943,
      "grad_norm": 0.5831708502837989,
      "learning_rate": 6.445346057353169e-06,
      "loss": 0.9238,
      "step": 21660
    },
    {
      "epoch": 1.3096029491750771,
      "grad_norm": 0.569977602112014,
      "learning_rate": 6.442157017655296e-06,
      "loss": 0.9309,
      "step": 21670
    },
    {
      "epoch": 1.310207288330211,
      "grad_norm": 0.5492876954598281,
      "learning_rate": 6.438967337925668e-06,
      "loss": 0.94,
      "step": 21680
    },
    {
      "epoch": 1.3108116274853447,
      "grad_norm": 0.560845223605572,
      "learning_rate": 6.435777019579874e-06,
      "loss": 0.94,
      "step": 21690
    },
    {
      "epoch": 1.3114159666404785,
      "grad_norm": 0.6371161033969285,
      "learning_rate": 6.432586064033779e-06,
      "loss": 0.904,
      "step": 21700
    },
    {
      "epoch": 1.3120203057956126,
      "grad_norm": 0.5786682172760027,
      "learning_rate": 6.429394472703539e-06,
      "loss": 0.9374,
      "step": 21710
    },
    {
      "epoch": 1.3126246449507464,
      "grad_norm": 0.6442147002487556,
      "learning_rate": 6.426202247005584e-06,
      "loss": 0.9345,
      "step": 21720
    },
    {
      "epoch": 1.3132289841058802,
      "grad_norm": 0.621374919533826,
      "learning_rate": 6.423009388356629e-06,
      "loss": 0.9392,
      "step": 21730
    },
    {
      "epoch": 1.3138333232610142,
      "grad_norm": 0.6395976854719905,
      "learning_rate": 6.419815898173674e-06,
      "loss": 0.9349,
      "step": 21740
    },
    {
      "epoch": 1.314437662416148,
      "grad_norm": 0.7928798710105224,
      "learning_rate": 6.416621777873992e-06,
      "loss": 0.9164,
      "step": 21750
    },
    {
      "epoch": 1.3150420015712818,
      "grad_norm": 0.8258525432311563,
      "learning_rate": 6.413427028875138e-06,
      "loss": 0.8963,
      "step": 21760
    },
    {
      "epoch": 1.3156463407264156,
      "grad_norm": 0.8690828653097746,
      "learning_rate": 6.41023165259495e-06,
      "loss": 0.9446,
      "step": 21770
    },
    {
      "epoch": 1.3162506798815494,
      "grad_norm": 0.9729642447657476,
      "learning_rate": 6.407035650451541e-06,
      "loss": 0.9444,
      "step": 21780
    },
    {
      "epoch": 1.3168550190366834,
      "grad_norm": 0.7846911672514052,
      "learning_rate": 6.4038390238633e-06,
      "loss": 0.9351,
      "step": 21790
    },
    {
      "epoch": 1.3174593581918173,
      "grad_norm": 0.5747159695914157,
      "learning_rate": 6.400641774248899e-06,
      "loss": 0.9379,
      "step": 21800
    },
    {
      "epoch": 1.318063697346951,
      "grad_norm": 0.5362650986243326,
      "learning_rate": 6.397443903027281e-06,
      "loss": 0.96,
      "step": 21810
    },
    {
      "epoch": 1.318668036502085,
      "grad_norm": 0.5475275595662574,
      "learning_rate": 6.394245411617664e-06,
      "loss": 0.9388,
      "step": 21820
    },
    {
      "epoch": 1.3192723756572189,
      "grad_norm": 0.5570424810631336,
      "learning_rate": 6.3910463014395496e-06,
      "loss": 0.9409,
      "step": 21830
    },
    {
      "epoch": 1.3198767148123527,
      "grad_norm": 0.5283875446225601,
      "learning_rate": 6.387846573912704e-06,
      "loss": 0.9327,
      "step": 21840
    },
    {
      "epoch": 1.3204810539674865,
      "grad_norm": 0.5182538742220805,
      "learning_rate": 6.384646230457173e-06,
      "loss": 0.9101,
      "step": 21850
    },
    {
      "epoch": 1.3210853931226203,
      "grad_norm": 0.5428016082066797,
      "learning_rate": 6.381445272493276e-06,
      "loss": 0.936,
      "step": 21860
    },
    {
      "epoch": 1.3216897322777543,
      "grad_norm": 0.549710069424863,
      "learning_rate": 6.3782437014416e-06,
      "loss": 0.9511,
      "step": 21870
    },
    {
      "epoch": 1.3222940714328881,
      "grad_norm": 0.5480501706869516,
      "learning_rate": 6.37504151872301e-06,
      "loss": 0.9673,
      "step": 21880
    },
    {
      "epoch": 1.322898410588022,
      "grad_norm": 0.5167824004896747,
      "learning_rate": 6.371838725758642e-06,
      "loss": 0.9276,
      "step": 21890
    },
    {
      "epoch": 1.323502749743156,
      "grad_norm": 0.5570338893498351,
      "learning_rate": 6.3686353239698974e-06,
      "loss": 0.9353,
      "step": 21900
    },
    {
      "epoch": 1.3241070888982898,
      "grad_norm": 0.5545426919234545,
      "learning_rate": 6.365431314778454e-06,
      "loss": 0.9216,
      "step": 21910
    },
    {
      "epoch": 1.3247114280534236,
      "grad_norm": 0.5604850610266743,
      "learning_rate": 6.3622266996062564e-06,
      "loss": 0.9615,
      "step": 21920
    },
    {
      "epoch": 1.3253157672085574,
      "grad_norm": 0.5483329374231105,
      "learning_rate": 6.359021479875515e-06,
      "loss": 0.958,
      "step": 21930
    },
    {
      "epoch": 1.3259201063636912,
      "grad_norm": 0.5664060019149888,
      "learning_rate": 6.355815657008717e-06,
      "loss": 0.94,
      "step": 21940
    },
    {
      "epoch": 1.3265244455188252,
      "grad_norm": 0.5447665700072896,
      "learning_rate": 6.352609232428608e-06,
      "loss": 0.9598,
      "step": 21950
    },
    {
      "epoch": 1.327128784673959,
      "grad_norm": 0.5112442863219152,
      "learning_rate": 6.349402207558206e-06,
      "loss": 0.9008,
      "step": 21960
    },
    {
      "epoch": 1.3277331238290928,
      "grad_norm": 0.541148629513141,
      "learning_rate": 6.346194583820795e-06,
      "loss": 0.9335,
      "step": 21970
    },
    {
      "epoch": 1.3283374629842268,
      "grad_norm": 0.5415610638277757,
      "learning_rate": 6.3429863626399245e-06,
      "loss": 0.9308,
      "step": 21980
    },
    {
      "epoch": 1.3289418021393606,
      "grad_norm": 0.5223660122979568,
      "learning_rate": 6.339777545439405e-06,
      "loss": 0.8859,
      "step": 21990
    },
    {
      "epoch": 1.3295461412944944,
      "grad_norm": 0.5690876138385812,
      "learning_rate": 6.33656813364332e-06,
      "loss": 0.94,
      "step": 22000
    },
    {
      "epoch": 1.3301504804496282,
      "grad_norm": 0.5365625441405761,
      "learning_rate": 6.333358128676006e-06,
      "loss": 0.9512,
      "step": 22010
    },
    {
      "epoch": 1.3307548196047623,
      "grad_norm": 0.5493141422193557,
      "learning_rate": 6.330147531962073e-06,
      "loss": 0.9371,
      "step": 22020
    },
    {
      "epoch": 1.331359158759896,
      "grad_norm": 0.5586144619198927,
      "learning_rate": 6.326936344926389e-06,
      "loss": 0.9107,
      "step": 22030
    },
    {
      "epoch": 1.3319634979150299,
      "grad_norm": 0.5249760688552998,
      "learning_rate": 6.323724568994083e-06,
      "loss": 0.9414,
      "step": 22040
    },
    {
      "epoch": 1.332567837070164,
      "grad_norm": 0.5266686959945355,
      "learning_rate": 6.320512205590548e-06,
      "loss": 0.914,
      "step": 22050
    },
    {
      "epoch": 1.3331721762252977,
      "grad_norm": 0.5404602508706192,
      "learning_rate": 6.3172992561414345e-06,
      "loss": 0.9135,
      "step": 22060
    },
    {
      "epoch": 1.3337765153804315,
      "grad_norm": 0.5629736014165325,
      "learning_rate": 6.314085722072656e-06,
      "loss": 0.9095,
      "step": 22070
    },
    {
      "epoch": 1.3343808545355653,
      "grad_norm": 0.5225209520001106,
      "learning_rate": 6.310871604810386e-06,
      "loss": 0.9101,
      "step": 22080
    },
    {
      "epoch": 1.3349851936906991,
      "grad_norm": 0.5800260535790562,
      "learning_rate": 6.307656905781053e-06,
      "loss": 0.9151,
      "step": 22090
    },
    {
      "epoch": 1.3355895328458331,
      "grad_norm": 0.6439688632145437,
      "learning_rate": 6.3044416264113465e-06,
      "loss": 0.9512,
      "step": 22100
    },
    {
      "epoch": 1.336193872000967,
      "grad_norm": 0.5761871565203135,
      "learning_rate": 6.301225768128213e-06,
      "loss": 0.9263,
      "step": 22110
    },
    {
      "epoch": 1.3367982111561008,
      "grad_norm": 0.630189294795375,
      "learning_rate": 6.298009332358857e-06,
      "loss": 0.9544,
      "step": 22120
    },
    {
      "epoch": 1.3374025503112348,
      "grad_norm": 0.5845267401828218,
      "learning_rate": 6.294792320530736e-06,
      "loss": 0.911,
      "step": 22130
    },
    {
      "epoch": 1.3380068894663686,
      "grad_norm": 0.6126744642079395,
      "learning_rate": 6.2915747340715684e-06,
      "loss": 0.9287,
      "step": 22140
    },
    {
      "epoch": 1.3386112286215024,
      "grad_norm": 0.6835343502301523,
      "learning_rate": 6.288356574409321e-06,
      "loss": 0.9261,
      "step": 22150
    },
    {
      "epoch": 1.3392155677766362,
      "grad_norm": 0.7743374738155538,
      "learning_rate": 6.285137842972221e-06,
      "loss": 0.94,
      "step": 22160
    },
    {
      "epoch": 1.33981990693177,
      "grad_norm": 0.6497260732966652,
      "learning_rate": 6.281918541188746e-06,
      "loss": 0.9238,
      "step": 22170
    },
    {
      "epoch": 1.340424246086904,
      "grad_norm": 0.7557369605602523,
      "learning_rate": 6.278698670487628e-06,
      "loss": 0.9287,
      "step": 22180
    },
    {
      "epoch": 1.3410285852420378,
      "grad_norm": 0.6030564395859551,
      "learning_rate": 6.275478232297852e-06,
      "loss": 0.9123,
      "step": 22190
    },
    {
      "epoch": 1.3416329243971716,
      "grad_norm": 0.6327443539400669,
      "learning_rate": 6.272257228048653e-06,
      "loss": 0.9348,
      "step": 22200
    },
    {
      "epoch": 1.3422372635523057,
      "grad_norm": 0.5813519839285867,
      "learning_rate": 6.269035659169515e-06,
      "loss": 0.9157,
      "step": 22210
    },
    {
      "epoch": 1.3428416027074395,
      "grad_norm": 0.6167215685631435,
      "learning_rate": 6.2658135270901834e-06,
      "loss": 0.9327,
      "step": 22220
    },
    {
      "epoch": 1.3434459418625733,
      "grad_norm": 0.6906512066095957,
      "learning_rate": 6.2625908332406404e-06,
      "loss": 0.9225,
      "step": 22230
    },
    {
      "epoch": 1.344050281017707,
      "grad_norm": 0.6067945734676988,
      "learning_rate": 6.2593675790511235e-06,
      "loss": 0.9667,
      "step": 22240
    },
    {
      "epoch": 1.3446546201728409,
      "grad_norm": 0.8204204225475179,
      "learning_rate": 6.25614376595212e-06,
      "loss": 0.9319,
      "step": 22250
    },
    {
      "epoch": 1.345258959327975,
      "grad_norm": 0.7973932270460289,
      "learning_rate": 6.2529193953743615e-06,
      "loss": 0.9263,
      "step": 22260
    },
    {
      "epoch": 1.3458632984831087,
      "grad_norm": 0.8121349243326759,
      "learning_rate": 6.24969446874883e-06,
      "loss": 0.9359,
      "step": 22270
    },
    {
      "epoch": 1.3464676376382425,
      "grad_norm": 0.8015237906389019,
      "learning_rate": 6.246468987506757e-06,
      "loss": 0.9211,
      "step": 22280
    },
    {
      "epoch": 1.3470719767933765,
      "grad_norm": 0.8161251810751748,
      "learning_rate": 6.243242953079611e-06,
      "loss": 0.942,
      "step": 22290
    },
    {
      "epoch": 1.3476763159485103,
      "grad_norm": 0.7685221795581088,
      "learning_rate": 6.240016366899114e-06,
      "loss": 0.9415,
      "step": 22300
    },
    {
      "epoch": 1.3482806551036441,
      "grad_norm": 0.7676335412252995,
      "learning_rate": 6.2367892303972325e-06,
      "loss": 0.9058,
      "step": 22310
    },
    {
      "epoch": 1.348884994258778,
      "grad_norm": 0.8601381268208241,
      "learning_rate": 6.233561545006173e-06,
      "loss": 0.9434,
      "step": 22320
    },
    {
      "epoch": 1.3494893334139118,
      "grad_norm": 0.8024577878501143,
      "learning_rate": 6.230333312158385e-06,
      "loss": 0.9353,
      "step": 22330
    },
    {
      "epoch": 1.3500936725690458,
      "grad_norm": 0.7592737141146152,
      "learning_rate": 6.22710453328657e-06,
      "loss": 0.9227,
      "step": 22340
    },
    {
      "epoch": 1.3506980117241796,
      "grad_norm": 1.185353453725579,
      "learning_rate": 6.223875209823661e-06,
      "loss": 0.9226,
      "step": 22350
    },
    {
      "epoch": 1.3513023508793136,
      "grad_norm": 1.2427121701483452,
      "learning_rate": 6.22064534320284e-06,
      "loss": 0.9243,
      "step": 22360
    },
    {
      "epoch": 1.3519066900344474,
      "grad_norm": 1.2496198440663855,
      "learning_rate": 6.217414934857526e-06,
      "loss": 0.9414,
      "step": 22370
    },
    {
      "epoch": 1.3525110291895812,
      "grad_norm": 1.1937846542128254,
      "learning_rate": 6.21418398622138e-06,
      "loss": 0.9171,
      "step": 22380
    },
    {
      "epoch": 1.353115368344715,
      "grad_norm": 1.1813710547520198,
      "learning_rate": 6.210952498728302e-06,
      "loss": 0.9362,
      "step": 22390
    },
    {
      "epoch": 1.3537197074998488,
      "grad_norm": 1.1441829380985113,
      "learning_rate": 6.2077204738124316e-06,
      "loss": 0.8943,
      "step": 22400
    },
    {
      "epoch": 1.3543240466549828,
      "grad_norm": 0.830010673067638,
      "learning_rate": 6.204487912908148e-06,
      "loss": 0.9165,
      "step": 22410
    },
    {
      "epoch": 1.3549283858101167,
      "grad_norm": 0.8596609071558763,
      "learning_rate": 6.201254817450066e-06,
      "loss": 0.9326,
      "step": 22420
    },
    {
      "epoch": 1.3555327249652505,
      "grad_norm": 0.778697432172059,
      "learning_rate": 6.198021188873041e-06,
      "loss": 0.9492,
      "step": 22430
    },
    {
      "epoch": 1.3561370641203845,
      "grad_norm": 0.8219141931965135,
      "learning_rate": 6.19478702861216e-06,
      "loss": 0.9429,
      "step": 22440
    },
    {
      "epoch": 1.3567414032755183,
      "grad_norm": 2.7015732155787213,
      "learning_rate": 6.191552338102752e-06,
      "loss": 0.9313,
      "step": 22450
    },
    {
      "epoch": 1.357345742430652,
      "grad_norm": 2.247523982015999,
      "learning_rate": 6.188317118780376e-06,
      "loss": 0.9138,
      "step": 22460
    },
    {
      "epoch": 1.357950081585786,
      "grad_norm": 2.5207787575513745,
      "learning_rate": 6.1850813720808265e-06,
      "loss": 0.911,
      "step": 22470
    },
    {
      "epoch": 1.3585544207409197,
      "grad_norm": 2.5163477113093853,
      "learning_rate": 6.181845099440138e-06,
      "loss": 0.9319,
      "step": 22480
    },
    {
      "epoch": 1.3591587598960537,
      "grad_norm": 2.393759477116278,
      "learning_rate": 6.1786083022945695e-06,
      "loss": 0.9434,
      "step": 22490
    },
    {
      "epoch": 1.3597630990511875,
      "grad_norm": 1.0232534242601,
      "learning_rate": 6.175370982080621e-06,
      "loss": 0.9421,
      "step": 22500
    },
    {
      "epoch": 1.3603674382063213,
      "grad_norm": 1.005498263751018,
      "learning_rate": 6.172133140235017e-06,
      "loss": 0.9166,
      "step": 22510
    },
    {
      "epoch": 1.3609717773614554,
      "grad_norm": 0.8964003157565335,
      "learning_rate": 6.168894778194718e-06,
      "loss": 0.9587,
      "step": 22520
    },
    {
      "epoch": 1.3615761165165892,
      "grad_norm": 0.9831363965925539,
      "learning_rate": 6.165655897396916e-06,
      "loss": 0.9371,
      "step": 22530
    },
    {
      "epoch": 1.362180455671723,
      "grad_norm": 1.0242360823936525,
      "learning_rate": 6.1624164992790324e-06,
      "loss": 0.9461,
      "step": 22540
    },
    {
      "epoch": 1.3627847948268568,
      "grad_norm": 0.9251969235022647,
      "learning_rate": 6.159176585278716e-06,
      "loss": 0.9385,
      "step": 22550
    },
    {
      "epoch": 1.3633891339819906,
      "grad_norm": 1.0123875241828533,
      "learning_rate": 6.155936156833846e-06,
      "loss": 0.9301,
      "step": 22560
    },
    {
      "epoch": 1.3639934731371246,
      "grad_norm": 1.0013237217353859,
      "learning_rate": 6.152695215382532e-06,
      "loss": 0.9518,
      "step": 22570
    },
    {
      "epoch": 1.3645978122922584,
      "grad_norm": 1.006583334831408,
      "learning_rate": 6.149453762363106e-06,
      "loss": 0.9242,
      "step": 22580
    },
    {
      "epoch": 1.3652021514473922,
      "grad_norm": 1.0363878873251027,
      "learning_rate": 6.146211799214135e-06,
      "loss": 0.9292,
      "step": 22590
    },
    {
      "epoch": 1.3658064906025262,
      "grad_norm": 1.0217693890776098,
      "learning_rate": 6.142969327374404e-06,
      "loss": 0.9578,
      "step": 22600
    },
    {
      "epoch": 1.36641082975766,
      "grad_norm": 1.052903444545231,
      "learning_rate": 6.139726348282929e-06,
      "loss": 0.929,
      "step": 22610
    },
    {
      "epoch": 1.3670151689127938,
      "grad_norm": 1.0903198118444486,
      "learning_rate": 6.13648286337895e-06,
      "loss": 0.9399,
      "step": 22620
    },
    {
      "epoch": 1.3676195080679276,
      "grad_norm": 1.0931020828190434,
      "learning_rate": 6.13323887410193e-06,
      "loss": 0.9501,
      "step": 22630
    },
    {
      "epoch": 1.3682238472230615,
      "grad_norm": 1.058246207747843,
      "learning_rate": 6.1299943818915585e-06,
      "loss": 0.912,
      "step": 22640
    },
    {
      "epoch": 1.3688281863781955,
      "grad_norm": 0.9972810901398478,
      "learning_rate": 6.126749388187746e-06,
      "loss": 0.9241,
      "step": 22650
    },
    {
      "epoch": 1.3694325255333293,
      "grad_norm": 1.009710890428033,
      "learning_rate": 6.1235038944306255e-06,
      "loss": 0.9346,
      "step": 22660
    },
    {
      "epoch": 1.370036864688463,
      "grad_norm": 1.1395487232627965,
      "learning_rate": 6.120257902060553e-06,
      "loss": 0.9735,
      "step": 22670
    },
    {
      "epoch": 1.3706412038435971,
      "grad_norm": 1.0126362471346355,
      "learning_rate": 6.117011412518107e-06,
      "loss": 0.9277,
      "step": 22680
    },
    {
      "epoch": 1.371245542998731,
      "grad_norm": 1.081490879804867,
      "learning_rate": 6.113764427244082e-06,
      "loss": 0.9024,
      "step": 22690
    },
    {
      "epoch": 1.3718498821538647,
      "grad_norm": 1.1515774599267152,
      "learning_rate": 6.1105169476795e-06,
      "loss": 0.9742,
      "step": 22700
    },
    {
      "epoch": 1.3724542213089985,
      "grad_norm": 1.148094541529228,
      "learning_rate": 6.1072689752655946e-06,
      "loss": 0.9274,
      "step": 22710
    },
    {
      "epoch": 1.3730585604641325,
      "grad_norm": 1.1897230912869639,
      "learning_rate": 6.104020511443822e-06,
      "loss": 0.9272,
      "step": 22720
    },
    {
      "epoch": 1.3736628996192664,
      "grad_norm": 1.1238871800753303,
      "learning_rate": 6.100771557655857e-06,
      "loss": 0.9516,
      "step": 22730
    },
    {
      "epoch": 1.3742672387744002,
      "grad_norm": 1.2501363161240353,
      "learning_rate": 6.097522115343591e-06,
      "loss": 0.9383,
      "step": 22740
    },
    {
      "epoch": 1.3748715779295342,
      "grad_norm": 1.0858864801854375,
      "learning_rate": 6.0942721859491305e-06,
      "loss": 0.9186,
      "step": 22750
    },
    {
      "epoch": 1.375475917084668,
      "grad_norm": 1.0054737063992778,
      "learning_rate": 6.091021770914801e-06,
      "loss": 0.9413,
      "step": 22760
    },
    {
      "epoch": 1.3760802562398018,
      "grad_norm": 1.0130363302378544,
      "learning_rate": 6.087770871683142e-06,
      "loss": 0.9357,
      "step": 22770
    },
    {
      "epoch": 1.3766845953949356,
      "grad_norm": 1.0346428589517516,
      "learning_rate": 6.084519489696907e-06,
      "loss": 0.9363,
      "step": 22780
    },
    {
      "epoch": 1.3772889345500694,
      "grad_norm": 0.9612419586153981,
      "learning_rate": 6.081267626399067e-06,
      "loss": 0.9515,
      "step": 22790
    },
    {
      "epoch": 1.3778932737052034,
      "grad_norm": 1.002256217744848,
      "learning_rate": 6.078015283232803e-06,
      "loss": 0.9199,
      "step": 22800
    },
    {
      "epoch": 1.3784976128603372,
      "grad_norm": 1.038285108587607,
      "learning_rate": 6.0747624616415105e-06,
      "loss": 0.9351,
      "step": 22810
    },
    {
      "epoch": 1.379101952015471,
      "grad_norm": 1.069573709718411,
      "learning_rate": 6.071509163068798e-06,
      "loss": 0.931,
      "step": 22820
    },
    {
      "epoch": 1.379706291170605,
      "grad_norm": 1.1036864995574835,
      "learning_rate": 6.068255388958485e-06,
      "loss": 0.9292,
      "step": 22830
    },
    {
      "epoch": 1.3803106303257389,
      "grad_norm": 1.0905563157319802,
      "learning_rate": 6.0650011407546e-06,
      "loss": 0.944,
      "step": 22840
    },
    {
      "epoch": 1.3809149694808727,
      "grad_norm": 1.043855993639079,
      "learning_rate": 6.061746419901389e-06,
      "loss": 0.9394,
      "step": 22850
    },
    {
      "epoch": 1.3815193086360065,
      "grad_norm": 1.043080650091453,
      "learning_rate": 6.058491227843295e-06,
      "loss": 0.9339,
      "step": 22860
    },
    {
      "epoch": 1.3821236477911403,
      "grad_norm": 0.9964559186294396,
      "learning_rate": 6.0552355660249855e-06,
      "loss": 0.9151,
      "step": 22870
    },
    {
      "epoch": 1.3827279869462743,
      "grad_norm": 1.031683146749357,
      "learning_rate": 6.051979435891324e-06,
      "loss": 0.9368,
      "step": 22880
    },
    {
      "epoch": 1.383332326101408,
      "grad_norm": 0.9972540871676556,
      "learning_rate": 6.048722838887388e-06,
      "loss": 0.9447,
      "step": 22890
    },
    {
      "epoch": 1.383936665256542,
      "grad_norm": 1.1069890579146098,
      "learning_rate": 6.045465776458462e-06,
      "loss": 0.9524,
      "step": 22900
    },
    {
      "epoch": 1.384541004411676,
      "grad_norm": 1.1312219577350453,
      "learning_rate": 6.042208250050035e-06,
      "loss": 0.9265,
      "step": 22910
    },
    {
      "epoch": 1.3851453435668097,
      "grad_norm": 1.106722136380151,
      "learning_rate": 6.0389502611078e-06,
      "loss": 0.9296,
      "step": 22920
    },
    {
      "epoch": 1.3857496827219435,
      "grad_norm": 1.128971079243757,
      "learning_rate": 6.035691811077662e-06,
      "loss": 0.9169,
      "step": 22930
    },
    {
      "epoch": 1.3863540218770773,
      "grad_norm": 1.0760346175620863,
      "learning_rate": 6.032432901405724e-06,
      "loss": 0.9555,
      "step": 22940
    },
    {
      "epoch": 1.3869583610322112,
      "grad_norm": 1.0687632502766016,
      "learning_rate": 6.029173533538298e-06,
      "loss": 0.9219,
      "step": 22950
    },
    {
      "epoch": 1.3875627001873452,
      "grad_norm": 1.0552049540779944,
      "learning_rate": 6.025913708921897e-06,
      "loss": 0.9248,
      "step": 22960
    },
    {
      "epoch": 1.388167039342479,
      "grad_norm": 1.0759021138327665,
      "learning_rate": 6.022653429003234e-06,
      "loss": 0.9546,
      "step": 22970
    },
    {
      "epoch": 1.3887713784976128,
      "grad_norm": 1.1756032399045384,
      "learning_rate": 6.019392695229229e-06,
      "loss": 0.9385,
      "step": 22980
    },
    {
      "epoch": 1.3893757176527468,
      "grad_norm": 1.0496975412144864,
      "learning_rate": 6.016131509047002e-06,
      "loss": 0.9466,
      "step": 22990
    },
    {
      "epoch": 1.3899800568078806,
      "grad_norm": 1.2054130693796088,
      "learning_rate": 6.0128698719038715e-06,
      "loss": 0.9247,
      "step": 23000
    },
    {
      "epoch": 1.3905843959630144,
      "grad_norm": 1.2071874990069786,
      "learning_rate": 6.009607785247356e-06,
      "loss": 0.9208,
      "step": 23010
    },
    {
      "epoch": 1.3911887351181482,
      "grad_norm": 1.1972046327312493,
      "learning_rate": 6.00634525052518e-06,
      "loss": 0.9507,
      "step": 23020
    },
    {
      "epoch": 1.3917930742732822,
      "grad_norm": 1.1407249294411796,
      "learning_rate": 6.003082269185257e-06,
      "loss": 0.9325,
      "step": 23030
    },
    {
      "epoch": 1.392397413428416,
      "grad_norm": 1.22365586421668,
      "learning_rate": 5.999818842675706e-06,
      "loss": 0.9231,
      "step": 23040
    },
    {
      "epoch": 1.3930017525835499,
      "grad_norm": 1.3928813699944156,
      "learning_rate": 5.996554972444842e-06,
      "loss": 0.9358,
      "step": 23050
    },
    {
      "epoch": 1.3936060917386839,
      "grad_norm": 1.3713560730860463,
      "learning_rate": 5.993290659941175e-06,
      "loss": 0.9461,
      "step": 23060
    },
    {
      "epoch": 1.3942104308938177,
      "grad_norm": 1.3616165234332318,
      "learning_rate": 5.990025906613413e-06,
      "loss": 0.9321,
      "step": 23070
    },
    {
      "epoch": 1.3948147700489515,
      "grad_norm": 1.3069118442311392,
      "learning_rate": 5.986760713910458e-06,
      "loss": 0.9523,
      "step": 23080
    },
    {
      "epoch": 1.3954191092040853,
      "grad_norm": 1.4267578130905407,
      "learning_rate": 5.9834950832814085e-06,
      "loss": 0.9243,
      "step": 23090
    },
    {
      "epoch": 1.396023448359219,
      "grad_norm": 0.8370332104615356,
      "learning_rate": 5.980229016175558e-06,
      "loss": 0.9032,
      "step": 23100
    },
    {
      "epoch": 1.3966277875143531,
      "grad_norm": 0.8651910531255127,
      "learning_rate": 5.976962514042392e-06,
      "loss": 0.9363,
      "step": 23110
    },
    {
      "epoch": 1.397232126669487,
      "grad_norm": 0.8778088344901273,
      "learning_rate": 5.9736955783315905e-06,
      "loss": 0.943,
      "step": 23120
    },
    {
      "epoch": 1.3978364658246207,
      "grad_norm": 0.8349847974894837,
      "learning_rate": 5.970428210493024e-06,
      "loss": 0.925,
      "step": 23130
    },
    {
      "epoch": 1.3984408049797548,
      "grad_norm": 0.8478423563184788,
      "learning_rate": 5.967160411976757e-06,
      "loss": 0.9446,
      "step": 23140
    },
    {
      "epoch": 1.3990451441348886,
      "grad_norm": 0.7579262321762854,
      "learning_rate": 5.9638921842330435e-06,
      "loss": 0.9225,
      "step": 23150
    },
    {
      "epoch": 1.3996494832900224,
      "grad_norm": 0.7604510132674285,
      "learning_rate": 5.960623528712328e-06,
      "loss": 0.912,
      "step": 23160
    },
    {
      "epoch": 1.4002538224451562,
      "grad_norm": 0.7626086586725821,
      "learning_rate": 5.957354446865247e-06,
      "loss": 0.937,
      "step": 23170
    },
    {
      "epoch": 1.40085816160029,
      "grad_norm": 0.7236251629362629,
      "learning_rate": 5.954084940142621e-06,
      "loss": 0.9577,
      "step": 23180
    },
    {
      "epoch": 1.401462500755424,
      "grad_norm": 0.832589112317401,
      "learning_rate": 5.950815009995469e-06,
      "loss": 0.9352,
      "step": 23190
    },
    {
      "epoch": 1.4020668399105578,
      "grad_norm": 0.8127616193532974,
      "learning_rate": 5.947544657874984e-06,
      "loss": 0.9426,
      "step": 23200
    },
    {
      "epoch": 1.4026711790656916,
      "grad_norm": 0.8209702452228631,
      "learning_rate": 5.944273885232561e-06,
      "loss": 0.9372,
      "step": 23210
    },
    {
      "epoch": 1.4032755182208256,
      "grad_norm": 0.9392017093993318,
      "learning_rate": 5.94100269351977e-06,
      "loss": 0.9408,
      "step": 23220
    },
    {
      "epoch": 1.4038798573759594,
      "grad_norm": 0.8472141843693116,
      "learning_rate": 5.937731084188372e-06,
      "loss": 0.9187,
      "step": 23230
    },
    {
      "epoch": 1.4044841965310932,
      "grad_norm": 0.7879784182064267,
      "learning_rate": 5.934459058690314e-06,
      "loss": 0.9419,
      "step": 23240
    },
    {
      "epoch": 1.405088535686227,
      "grad_norm": 0.5672758982938461,
      "learning_rate": 5.931186618477727e-06,
      "loss": 0.9197,
      "step": 23250
    },
    {
      "epoch": 1.4056928748413609,
      "grad_norm": 0.5628166222146257,
      "learning_rate": 5.9279137650029226e-06,
      "loss": 0.9241,
      "step": 23260
    },
    {
      "epoch": 1.4062972139964949,
      "grad_norm": 0.5793938583559338,
      "learning_rate": 5.924640499718404e-06,
      "loss": 0.916,
      "step": 23270
    },
    {
      "epoch": 1.4069015531516287,
      "grad_norm": 0.6043414574811208,
      "learning_rate": 5.921366824076846e-06,
      "loss": 0.9331,
      "step": 23280
    },
    {
      "epoch": 1.4075058923067625,
      "grad_norm": 0.5808669327490961,
      "learning_rate": 5.918092739531116e-06,
      "loss": 0.9142,
      "step": 23290
    },
    {
      "epoch": 1.4081102314618965,
      "grad_norm": 0.6602905051195931,
      "learning_rate": 5.914818247534256e-06,
      "loss": 0.9222,
      "step": 23300
    },
    {
      "epoch": 1.4087145706170303,
      "grad_norm": 0.63981220871686,
      "learning_rate": 5.911543349539491e-06,
      "loss": 0.9179,
      "step": 23310
    },
    {
      "epoch": 1.4093189097721641,
      "grad_norm": 0.6635073430666767,
      "learning_rate": 5.9082680470002275e-06,
      "loss": 0.9636,
      "step": 23320
    },
    {
      "epoch": 1.409923248927298,
      "grad_norm": 0.6034233326503531,
      "learning_rate": 5.90499234137005e-06,
      "loss": 0.924,
      "step": 23330
    },
    {
      "epoch": 1.4105275880824317,
      "grad_norm": 0.6030559521565237,
      "learning_rate": 5.901716234102722e-06,
      "loss": 0.9155,
      "step": 23340
    },
    {
      "epoch": 1.4111319272375658,
      "grad_norm": 0.8554880372229265,
      "learning_rate": 5.898439726652185e-06,
      "loss": 0.928,
      "step": 23350
    },
    {
      "epoch": 1.4117362663926996,
      "grad_norm": 0.8497726167709079,
      "learning_rate": 5.895162820472561e-06,
      "loss": 0.9226,
      "step": 23360
    },
    {
      "epoch": 1.4123406055478334,
      "grad_norm": 0.752648043247242,
      "learning_rate": 5.891885517018144e-06,
      "loss": 0.9266,
      "step": 23370
    },
    {
      "epoch": 1.4129449447029674,
      "grad_norm": 0.8587357615882126,
      "learning_rate": 5.888607817743407e-06,
      "loss": 0.9114,
      "step": 23380
    },
    {
      "epoch": 1.4135492838581012,
      "grad_norm": 0.7827360882332269,
      "learning_rate": 5.885329724103e-06,
      "loss": 0.9107,
      "step": 23390
    },
    {
      "epoch": 1.414153623013235,
      "grad_norm": 0.5532713154866552,
      "learning_rate": 5.882051237551747e-06,
      "loss": 0.949,
      "step": 23400
    },
    {
      "epoch": 1.4147579621683688,
      "grad_norm": 0.5370093816865774,
      "learning_rate": 5.878772359544644e-06,
      "loss": 0.9226,
      "step": 23410
    },
    {
      "epoch": 1.4153623013235028,
      "grad_norm": 0.5255402087035534,
      "learning_rate": 5.875493091536863e-06,
      "loss": 0.9219,
      "step": 23420
    },
    {
      "epoch": 1.4159666404786366,
      "grad_norm": 0.5698333088239176,
      "learning_rate": 5.872213434983749e-06,
      "loss": 0.9312,
      "step": 23430
    },
    {
      "epoch": 1.4165709796337704,
      "grad_norm": 0.5639465058228315,
      "learning_rate": 5.86893339134082e-06,
      "loss": 0.9191,
      "step": 23440
    },
    {
      "epoch": 1.4171753187889045,
      "grad_norm": 0.5380120211799528,
      "learning_rate": 5.8656529620637645e-06,
      "loss": 0.9119,
      "step": 23450
    },
    {
      "epoch": 1.4177796579440383,
      "grad_norm": 0.5197435168817296,
      "learning_rate": 5.862372148608442e-06,
      "loss": 0.9255,
      "step": 23460
    },
    {
      "epoch": 1.418383997099172,
      "grad_norm": 0.5276321836999159,
      "learning_rate": 5.859090952430884e-06,
      "loss": 0.9384,
      "step": 23470
    },
    {
      "epoch": 1.4189883362543059,
      "grad_norm": 0.563623102037306,
      "learning_rate": 5.855809374987292e-06,
      "loss": 0.9288,
      "step": 23480
    },
    {
      "epoch": 1.4195926754094397,
      "grad_norm": 0.553209711824374,
      "learning_rate": 5.85252741773403e-06,
      "loss": 0.9575,
      "step": 23490
    },
    {
      "epoch": 1.4201970145645737,
      "grad_norm": 0.5224332909694155,
      "learning_rate": 5.849245082127645e-06,
      "loss": 0.9478,
      "step": 23500
    },
    {
      "epoch": 1.4208013537197075,
      "grad_norm": 0.5567811248645171,
      "learning_rate": 5.845962369624838e-06,
      "loss": 0.9595,
      "step": 23510
    },
    {
      "epoch": 1.4214056928748413,
      "grad_norm": 0.594796326940543,
      "learning_rate": 5.842679281682482e-06,
      "loss": 0.9275,
      "step": 23520
    },
    {
      "epoch": 1.4220100320299753,
      "grad_norm": 0.568043704523285,
      "learning_rate": 5.839395819757618e-06,
      "loss": 0.9243,
      "step": 23530
    },
    {
      "epoch": 1.4226143711851091,
      "grad_norm": 0.536059227424461,
      "learning_rate": 5.836111985307454e-06,
      "loss": 0.9362,
      "step": 23540
    },
    {
      "epoch": 1.423218710340243,
      "grad_norm": 0.519457316071563,
      "learning_rate": 5.832827779789359e-06,
      "loss": 0.9406,
      "step": 23550
    },
    {
      "epoch": 1.4238230494953767,
      "grad_norm": 0.5389675848892215,
      "learning_rate": 5.82954320466087e-06,
      "loss": 0.9209,
      "step": 23560
    },
    {
      "epoch": 1.4244273886505106,
      "grad_norm": 0.5368733627433508,
      "learning_rate": 5.826258261379685e-06,
      "loss": 0.9359,
      "step": 23570
    },
    {
      "epoch": 1.4250317278056446,
      "grad_norm": 0.5247160243873157,
      "learning_rate": 5.82297295140367e-06,
      "loss": 0.9251,
      "step": 23580
    },
    {
      "epoch": 1.4256360669607784,
      "grad_norm": 0.5499268238784991,
      "learning_rate": 5.8196872761908505e-06,
      "loss": 0.9119,
      "step": 23590
    },
    {
      "epoch": 1.4262404061159122,
      "grad_norm": 0.5483747462857096,
      "learning_rate": 5.816401237199415e-06,
      "loss": 0.9131,
      "step": 23600
    },
    {
      "epoch": 1.4268447452710462,
      "grad_norm": 0.553352366434354,
      "learning_rate": 5.813114835887712e-06,
      "loss": 0.9162,
      "step": 23610
    },
    {
      "epoch": 1.42744908442618,
      "grad_norm": 0.5308617286168494,
      "learning_rate": 5.809828073714254e-06,
      "loss": 0.9189,
      "step": 23620
    },
    {
      "epoch": 1.4280534235813138,
      "grad_norm": 0.5472399866509499,
      "learning_rate": 5.8065409521377095e-06,
      "loss": 0.9304,
      "step": 23630
    },
    {
      "epoch": 1.4286577627364476,
      "grad_norm": 0.5343379434090632,
      "learning_rate": 5.80325347261691e-06,
      "loss": 0.9527,
      "step": 23640
    },
    {
      "epoch": 1.4292621018915814,
      "grad_norm": 0.5619202319151952,
      "learning_rate": 5.799965636610844e-06,
      "loss": 0.95,
      "step": 23650
    },
    {
      "epoch": 1.4298664410467155,
      "grad_norm": 0.5710364577741498,
      "learning_rate": 5.796677445578661e-06,
      "loss": 0.9146,
      "step": 23660
    },
    {
      "epoch": 1.4304707802018493,
      "grad_norm": 0.5584617636158474,
      "learning_rate": 5.7933889009796625e-06,
      "loss": 0.927,
      "step": 23670
    },
    {
      "epoch": 1.431075119356983,
      "grad_norm": 0.5435218036747607,
      "learning_rate": 5.7901000042733116e-06,
      "loss": 0.9373,
      "step": 23680
    },
    {
      "epoch": 1.431679458512117,
      "grad_norm": 0.5266136045708651,
      "learning_rate": 5.7868107569192254e-06,
      "loss": 0.9262,
      "step": 23690
    },
    {
      "epoch": 1.432283797667251,
      "grad_norm": 0.6227605025639936,
      "learning_rate": 5.783521160377179e-06,
      "loss": 0.9157,
      "step": 23700
    },
    {
      "epoch": 1.4328881368223847,
      "grad_norm": 0.6550486181648428,
      "learning_rate": 5.7802312161071e-06,
      "loss": 0.9148,
      "step": 23710
    },
    {
      "epoch": 1.4334924759775185,
      "grad_norm": 0.6152904797965045,
      "learning_rate": 5.776940925569072e-06,
      "loss": 0.9301,
      "step": 23720
    },
    {
      "epoch": 1.4340968151326525,
      "grad_norm": 0.6516510877489632,
      "learning_rate": 5.7736502902233314e-06,
      "loss": 0.921,
      "step": 23730
    },
    {
      "epoch": 1.4347011542877863,
      "grad_norm": 0.6418468792766503,
      "learning_rate": 5.770359311530267e-06,
      "loss": 0.9355,
      "step": 23740
    },
    {
      "epoch": 1.4353054934429201,
      "grad_norm": 0.7392416154689126,
      "learning_rate": 5.767067990950422e-06,
      "loss": 0.9358,
      "step": 23750
    },
    {
      "epoch": 1.4359098325980542,
      "grad_norm": 0.6804505711286888,
      "learning_rate": 5.763776329944491e-06,
      "loss": 0.9289,
      "step": 23760
    },
    {
      "epoch": 1.436514171753188,
      "grad_norm": 0.6804828549369927,
      "learning_rate": 5.760484329973314e-06,
      "loss": 0.9448,
      "step": 23770
    },
    {
      "epoch": 1.4371185109083218,
      "grad_norm": 0.7085763300938479,
      "learning_rate": 5.757191992497891e-06,
      "loss": 0.9622,
      "step": 23780
    },
    {
      "epoch": 1.4377228500634556,
      "grad_norm": 0.7162386094733575,
      "learning_rate": 5.753899318979366e-06,
      "loss": 0.9206,
      "step": 23790
    },
    {
      "epoch": 1.4383271892185894,
      "grad_norm": 0.6099898676532602,
      "learning_rate": 5.750606310879034e-06,
      "loss": 0.9333,
      "step": 23800
    },
    {
      "epoch": 1.4389315283737234,
      "grad_norm": 0.6765274733055479,
      "learning_rate": 5.747312969658334e-06,
      "loss": 0.9541,
      "step": 23810
    },
    {
      "epoch": 1.4395358675288572,
      "grad_norm": 0.6447121314069736,
      "learning_rate": 5.74401929677886e-06,
      "loss": 0.9212,
      "step": 23820
    },
    {
      "epoch": 1.440140206683991,
      "grad_norm": 0.6960786656162969,
      "learning_rate": 5.740725293702346e-06,
      "loss": 0.9259,
      "step": 23830
    },
    {
      "epoch": 1.440744545839125,
      "grad_norm": 0.645977388970727,
      "learning_rate": 5.737430961890681e-06,
      "loss": 0.9554,
      "step": 23840
    },
    {
      "epoch": 1.4413488849942588,
      "grad_norm": 0.7459973063815706,
      "learning_rate": 5.734136302805893e-06,
      "loss": 0.9015,
      "step": 23850
    },
    {
      "epoch": 1.4419532241493926,
      "grad_norm": 0.7873226922195327,
      "learning_rate": 5.730841317910155e-06,
      "loss": 0.9378,
      "step": 23860
    },
    {
      "epoch": 1.4425575633045264,
      "grad_norm": 0.7899715681165357,
      "learning_rate": 5.727546008665791e-06,
      "loss": 0.9209,
      "step": 23870
    },
    {
      "epoch": 1.4431619024596603,
      "grad_norm": 0.8172282363303137,
      "learning_rate": 5.72425037653526e-06,
      "loss": 0.9372,
      "step": 23880
    },
    {
      "epoch": 1.4437662416147943,
      "grad_norm": 0.8182090464292833,
      "learning_rate": 5.720954422981175e-06,
      "loss": 0.9328,
      "step": 23890
    },
    {
      "epoch": 1.444370580769928,
      "grad_norm": 0.7834800642413587,
      "learning_rate": 5.717658149466283e-06,
      "loss": 0.9331,
      "step": 23900
    },
    {
      "epoch": 1.4449749199250619,
      "grad_norm": 0.7937840402422004,
      "learning_rate": 5.7143615574534736e-06,
      "loss": 0.9545,
      "step": 23910
    },
    {
      "epoch": 1.445579259080196,
      "grad_norm": 0.8091386230879632,
      "learning_rate": 5.711064648405784e-06,
      "loss": 0.95,
      "step": 23920
    },
    {
      "epoch": 1.4461835982353297,
      "grad_norm": 0.8002209844544873,
      "learning_rate": 5.707767423786385e-06,
      "loss": 0.9071,
      "step": 23930
    },
    {
      "epoch": 1.4467879373904635,
      "grad_norm": 0.7677156548723814,
      "learning_rate": 5.704469885058594e-06,
      "loss": 0.9257,
      "step": 23940
    },
    {
      "epoch": 1.4473922765455973,
      "grad_norm": 1.117799176143842,
      "learning_rate": 5.70117203368586e-06,
      "loss": 0.92,
      "step": 23950
    },
    {
      "epoch": 1.4479966157007311,
      "grad_norm": 1.1564055606972556,
      "learning_rate": 5.697873871131778e-06,
      "loss": 0.9579,
      "step": 23960
    },
    {
      "epoch": 1.4486009548558652,
      "grad_norm": 1.199502041846667,
      "learning_rate": 5.694575398860076e-06,
      "loss": 0.9362,
      "step": 23970
    },
    {
      "epoch": 1.449205294010999,
      "grad_norm": 1.1048179768832782,
      "learning_rate": 5.6912766183346235e-06,
      "loss": 0.9363,
      "step": 23980
    },
    {
      "epoch": 1.4498096331661328,
      "grad_norm": 1.1810427476759027,
      "learning_rate": 5.687977531019425e-06,
      "loss": 0.9381,
      "step": 23990
    },
    {
      "epoch": 1.4504139723212668,
      "grad_norm": 0.9493817329333853,
      "learning_rate": 5.6846781383786196e-06,
      "loss": 0.9377,
      "step": 24000
    },
    {
      "epoch": 1.4510183114764006,
      "grad_norm": 1.0181251025469211,
      "learning_rate": 5.681378441876485e-06,
      "loss": 0.9284,
      "step": 24010
    },
    {
      "epoch": 1.4516226506315344,
      "grad_norm": 0.8571959376934667,
      "learning_rate": 5.678078442977431e-06,
      "loss": 0.9244,
      "step": 24020
    },
    {
      "epoch": 1.4522269897866682,
      "grad_norm": 1.064963072459699,
      "learning_rate": 5.674778143146003e-06,
      "loss": 0.9441,
      "step": 24030
    },
    {
      "epoch": 1.452831328941802,
      "grad_norm": 0.8561229919400791,
      "learning_rate": 5.6714775438468815e-06,
      "loss": 0.9303,
      "step": 24040
    },
    {
      "epoch": 1.453435668096936,
      "grad_norm": 2.4463143228903017,
      "learning_rate": 5.668176646544876e-06,
      "loss": 0.9066,
      "step": 24050
    },
    {
      "epoch": 1.4540400072520698,
      "grad_norm": 2.6358690553350805,
      "learning_rate": 5.66487545270493e-06,
      "loss": 0.9271,
      "step": 24060
    },
    {
      "epoch": 1.4546443464072039,
      "grad_norm": 2.42934031766331,
      "learning_rate": 5.6615739637921205e-06,
      "loss": 0.9306,
      "step": 24070
    },
    {
      "epoch": 1.4552486855623377,
      "grad_norm": 2.7277840927160364,
      "learning_rate": 5.658272181271653e-06,
      "loss": 0.8911,
      "step": 24080
    },
    {
      "epoch": 1.4558530247174715,
      "grad_norm": 2.5010606102575075,
      "learning_rate": 5.654970106608865e-06,
      "loss": 0.9302,
      "step": 24090
    },
    {
      "epoch": 1.4564573638726053,
      "grad_norm": 0.9949991228860524,
      "learning_rate": 5.651667741269222e-06,
      "loss": 0.9133,
      "step": 24100
    },
    {
      "epoch": 1.457061703027739,
      "grad_norm": 1.0082999516017692,
      "learning_rate": 5.64836508671832e-06,
      "loss": 0.9183,
      "step": 24110
    },
    {
      "epoch": 1.457666042182873,
      "grad_norm": 1.0073580807753402,
      "learning_rate": 5.645062144421884e-06,
      "loss": 0.9345,
      "step": 24120
    },
    {
      "epoch": 1.458270381338007,
      "grad_norm": 0.9525189911860953,
      "learning_rate": 5.641758915845762e-06,
      "loss": 0.96,
      "step": 24130
    },
    {
      "epoch": 1.4588747204931407,
      "grad_norm": 1.3000267953353162,
      "learning_rate": 5.638455402455934e-06,
      "loss": 0.9433,
      "step": 24140
    },
    {
      "epoch": 1.4594790596482747,
      "grad_norm": 0.9329578236185453,
      "learning_rate": 5.635151605718507e-06,
      "loss": 0.9184,
      "step": 24150
    },
    {
      "epoch": 1.4600833988034085,
      "grad_norm": 0.9442844028556734,
      "learning_rate": 5.631847527099711e-06,
      "loss": 0.9425,
      "step": 24160
    },
    {
      "epoch": 1.4606877379585423,
      "grad_norm": 0.970976383184132,
      "learning_rate": 5.628543168065899e-06,
      "loss": 0.9327,
      "step": 24170
    },
    {
      "epoch": 1.4612920771136761,
      "grad_norm": 1.0174704510277752,
      "learning_rate": 5.625238530083554e-06,
      "loss": 0.9423,
      "step": 24180
    },
    {
      "epoch": 1.46189641626881,
      "grad_norm": 0.9775932818965495,
      "learning_rate": 5.621933614619279e-06,
      "loss": 0.942,
      "step": 24190
    },
    {
      "epoch": 1.462500755423944,
      "grad_norm": 1.0600828461815932,
      "learning_rate": 5.618628423139801e-06,
      "loss": 0.9406,
      "step": 24200
    },
    {
      "epoch": 1.4631050945790778,
      "grad_norm": 1.0739367834506144,
      "learning_rate": 5.6153229571119706e-06,
      "loss": 0.9423,
      "step": 24210
    },
    {
      "epoch": 1.4637094337342116,
      "grad_norm": 1.0312842310827948,
      "learning_rate": 5.6120172180027565e-06,
      "loss": 0.9104,
      "step": 24220
    },
    {
      "epoch": 1.4643137728893456,
      "grad_norm": 1.0635678647373001,
      "learning_rate": 5.608711207279254e-06,
      "loss": 0.9368,
      "step": 24230
    },
    {
      "epoch": 1.4649181120444794,
      "grad_norm": 1.0344959989680858,
      "learning_rate": 5.6054049264086765e-06,
      "loss": 0.9097,
      "step": 24240
    },
    {
      "epoch": 1.4655224511996132,
      "grad_norm": 1.005835618202722,
      "learning_rate": 5.602098376858356e-06,
      "loss": 0.952,
      "step": 24250
    },
    {
      "epoch": 1.466126790354747,
      "grad_norm": 1.131875223252975,
      "learning_rate": 5.598791560095744e-06,
      "loss": 0.9186,
      "step": 24260
    },
    {
      "epoch": 1.4667311295098808,
      "grad_norm": 1.086194952288562,
      "learning_rate": 5.595484477588415e-06,
      "loss": 0.9403,
      "step": 24270
    },
    {
      "epoch": 1.4673354686650149,
      "grad_norm": 1.0270700446528251,
      "learning_rate": 5.592177130804053e-06,
      "loss": 0.9443,
      "step": 24280
    },
    {
      "epoch": 1.4679398078201487,
      "grad_norm": 1.0910641604620384,
      "learning_rate": 5.588869521210468e-06,
      "loss": 0.947,
      "step": 24290
    },
    {
      "epoch": 1.4685441469752825,
      "grad_norm": 1.0796712728801525,
      "learning_rate": 5.585561650275581e-06,
      "loss": 0.9248,
      "step": 24300
    },
    {
      "epoch": 1.4691484861304165,
      "grad_norm": 1.0995957262770726,
      "learning_rate": 5.582253519467432e-06,
      "loss": 0.8944,
      "step": 24310
    },
    {
      "epoch": 1.4697528252855503,
      "grad_norm": 1.206669149227736,
      "learning_rate": 5.578945130254174e-06,
      "loss": 0.9121,
      "step": 24320
    },
    {
      "epoch": 1.470357164440684,
      "grad_norm": 1.1237455542866643,
      "learning_rate": 5.575636484104075e-06,
      "loss": 0.9272,
      "step": 24330
    },
    {
      "epoch": 1.470961503595818,
      "grad_norm": 1.090345073196265,
      "learning_rate": 5.57232758248552e-06,
      "loss": 0.9206,
      "step": 24340
    },
    {
      "epoch": 1.4715658427509517,
      "grad_norm": 1.0094949587322917,
      "learning_rate": 5.5690184268670035e-06,
      "loss": 0.9283,
      "step": 24350
    },
    {
      "epoch": 1.4721701819060857,
      "grad_norm": 1.0165623788715539,
      "learning_rate": 5.565709018717136e-06,
      "loss": 0.9284,
      "step": 24360
    },
    {
      "epoch": 1.4727745210612195,
      "grad_norm": 1.211764548105409,
      "learning_rate": 5.562399359504637e-06,
      "loss": 0.9333,
      "step": 24370
    },
    {
      "epoch": 1.4733788602163533,
      "grad_norm": 0.998326085486439,
      "learning_rate": 5.559089450698338e-06,
      "loss": 0.9101,
      "step": 24380
    },
    {
      "epoch": 1.4739831993714874,
      "grad_norm": 0.9532644690759089,
      "learning_rate": 5.555779293767185e-06,
      "loss": 0.9267,
      "step": 24390
    },
    {
      "epoch": 1.4745875385266212,
      "grad_norm": 1.1335260433592877,
      "learning_rate": 5.55246889018023e-06,
      "loss": 0.9377,
      "step": 24400
    },
    {
      "epoch": 1.475191877681755,
      "grad_norm": 1.0757158715265807,
      "learning_rate": 5.5491582414066366e-06,
      "loss": 0.9085,
      "step": 24410
    },
    {
      "epoch": 1.4757962168368888,
      "grad_norm": 1.057926732626364,
      "learning_rate": 5.545847348915674e-06,
      "loss": 0.9686,
      "step": 24420
    },
    {
      "epoch": 1.4764005559920228,
      "grad_norm": 1.0654649545148709,
      "learning_rate": 5.542536214176725e-06,
      "loss": 0.9299,
      "step": 24430
    },
    {
      "epoch": 1.4770048951471566,
      "grad_norm": 1.0084903494852366,
      "learning_rate": 5.539224838659275e-06,
      "loss": 0.9437,
      "step": 24440
    },
    {
      "epoch": 1.4776092343022904,
      "grad_norm": 0.9313667488629244,
      "learning_rate": 5.53591322383292e-06,
      "loss": 0.9431,
      "step": 24450
    },
    {
      "epoch": 1.4782135734574244,
      "grad_norm": 1.032862261985383,
      "learning_rate": 5.532601371167356e-06,
      "loss": 0.9411,
      "step": 24460
    },
    {
      "epoch": 1.4788179126125582,
      "grad_norm": 0.9464095432894105,
      "learning_rate": 5.529289282132394e-06,
      "loss": 0.9299,
      "step": 24470
    },
    {
      "epoch": 1.479422251767692,
      "grad_norm": 0.9831548913519023,
      "learning_rate": 5.52597695819794e-06,
      "loss": 0.9284,
      "step": 24480
    },
    {
      "epoch": 1.4800265909228258,
      "grad_norm": 1.0092236694155683,
      "learning_rate": 5.522664400834013e-06,
      "loss": 0.9374,
      "step": 24490
    },
    {
      "epoch": 1.4806309300779597,
      "grad_norm": 1.0630829895520733,
      "learning_rate": 5.519351611510731e-06,
      "loss": 0.9289,
      "step": 24500
    },
    {
      "epoch": 1.4812352692330937,
      "grad_norm": 1.2132130764191726,
      "learning_rate": 5.5160385916983136e-06,
      "loss": 0.9134,
      "step": 24510
    },
    {
      "epoch": 1.4818396083882275,
      "grad_norm": 1.1136866536032546,
      "learning_rate": 5.512725342867087e-06,
      "loss": 0.9187,
      "step": 24520
    },
    {
      "epoch": 1.4824439475433613,
      "grad_norm": 1.1271814258216801,
      "learning_rate": 5.509411866487474e-06,
      "loss": 0.9243,
      "step": 24530
    },
    {
      "epoch": 1.4830482866984953,
      "grad_norm": 1.1097661823471676,
      "learning_rate": 5.506098164030003e-06,
      "loss": 0.9422,
      "step": 24540
    },
    {
      "epoch": 1.4836526258536291,
      "grad_norm": 1.0779943607533187,
      "learning_rate": 5.502784236965301e-06,
      "loss": 0.9351,
      "step": 24550
    },
    {
      "epoch": 1.484256965008763,
      "grad_norm": 1.11532446683327,
      "learning_rate": 5.499470086764094e-06,
      "loss": 0.9371,
      "step": 24560
    },
    {
      "epoch": 1.4848613041638967,
      "grad_norm": 1.114111938319899,
      "learning_rate": 5.496155714897207e-06,
      "loss": 0.9527,
      "step": 24570
    },
    {
      "epoch": 1.4854656433190305,
      "grad_norm": 1.0720829313632556,
      "learning_rate": 5.492841122835565e-06,
      "loss": 0.9186,
      "step": 24580
    },
    {
      "epoch": 1.4860699824741646,
      "grad_norm": 1.1325486265286429,
      "learning_rate": 5.489526312050187e-06,
      "loss": 0.9089,
      "step": 24590
    },
    {
      "epoch": 1.4866743216292984,
      "grad_norm": 1.234902681852651,
      "learning_rate": 5.486211284012196e-06,
      "loss": 0.9357,
      "step": 24600
    },
    {
      "epoch": 1.4872786607844322,
      "grad_norm": 1.153268063464747,
      "learning_rate": 5.482896040192805e-06,
      "loss": 0.884,
      "step": 24610
    },
    {
      "epoch": 1.4878829999395662,
      "grad_norm": 1.1971313011206486,
      "learning_rate": 5.479580582063322e-06,
      "loss": 0.9306,
      "step": 24620
    },
    {
      "epoch": 1.4884873390947,
      "grad_norm": 1.0973762613588873,
      "learning_rate": 5.476264911095158e-06,
      "loss": 0.9316,
      "step": 24630
    },
    {
      "epoch": 1.4890916782498338,
      "grad_norm": 1.212149959098993,
      "learning_rate": 5.472949028759811e-06,
      "loss": 0.9163,
      "step": 24640
    },
    {
      "epoch": 1.4896960174049676,
      "grad_norm": 1.4200996829837007,
      "learning_rate": 5.469632936528875e-06,
      "loss": 0.9177,
      "step": 24650
    },
    {
      "epoch": 1.4903003565601014,
      "grad_norm": 1.4829504571141172,
      "learning_rate": 5.466316635874037e-06,
      "loss": 0.9258,
      "step": 24660
    },
    {
      "epoch": 1.4909046957152354,
      "grad_norm": 1.4280627667598678,
      "learning_rate": 5.463000128267079e-06,
      "loss": 0.9178,
      "step": 24670
    },
    {
      "epoch": 1.4915090348703692,
      "grad_norm": 1.362757596864265,
      "learning_rate": 5.459683415179871e-06,
      "loss": 0.9232,
      "step": 24680
    },
    {
      "epoch": 1.492113374025503,
      "grad_norm": 1.4033937713313966,
      "learning_rate": 5.4563664980843765e-06,
      "loss": 0.9192,
      "step": 24690
    },
    {
      "epoch": 1.492717713180637,
      "grad_norm": 0.7871530104856718,
      "learning_rate": 5.453049378452648e-06,
      "loss": 0.9364,
      "step": 24700
    },
    {
      "epoch": 1.4933220523357709,
      "grad_norm": 0.8376815217719871,
      "learning_rate": 5.449732057756831e-06,
      "loss": 0.9373,
      "step": 24710
    },
    {
      "epoch": 1.4939263914909047,
      "grad_norm": 0.8335830707549651,
      "learning_rate": 5.446414537469158e-06,
      "loss": 0.8991,
      "step": 24720
    },
    {
      "epoch": 1.4945307306460385,
      "grad_norm": 0.8996257796761521,
      "learning_rate": 5.443096819061946e-06,
      "loss": 0.9226,
      "step": 24730
    },
    {
      "epoch": 1.4951350698011725,
      "grad_norm": 0.8344490672476051,
      "learning_rate": 5.43977890400761e-06,
      "loss": 0.9146,
      "step": 24740
    },
    {
      "epoch": 1.4957394089563063,
      "grad_norm": 0.7919301571373561,
      "learning_rate": 5.436460793778642e-06,
      "loss": 0.9407,
      "step": 24750
    },
    {
      "epoch": 1.49634374811144,
      "grad_norm": 0.8424038899184205,
      "learning_rate": 5.433142489847626e-06,
      "loss": 0.9455,
      "step": 24760
    },
    {
      "epoch": 1.4969480872665741,
      "grad_norm": 0.7754519638914394,
      "learning_rate": 5.429823993687234e-06,
      "loss": 0.9509,
      "step": 24770
    },
    {
      "epoch": 1.497552426421708,
      "grad_norm": 0.788339896315405,
      "learning_rate": 5.4265053067702156e-06,
      "loss": 0.9052,
      "step": 24780
    },
    {
      "epoch": 1.4981567655768417,
      "grad_norm": 0.806122611292681,
      "learning_rate": 5.423186430569411e-06,
      "loss": 0.9119,
      "step": 24790
    },
    {
      "epoch": 1.4987611047319755,
      "grad_norm": 0.8061608990913078,
      "learning_rate": 5.419867366557746e-06,
      "loss": 0.942,
      "step": 24800
    },
    {
      "epoch": 1.4993654438871094,
      "grad_norm": 0.8371038554107869,
      "learning_rate": 5.4165481162082255e-06,
      "loss": 0.9446,
      "step": 24810
    },
    {
      "epoch": 1.4999697830422434,
      "grad_norm": 0.8055214948696328,
      "learning_rate": 5.413228680993936e-06,
      "loss": 0.9368,
      "step": 24820
    },
    {
      "epoch": 1.5005741221973772,
      "grad_norm": 0.8338406655848178,
      "learning_rate": 5.409909062388053e-06,
      "loss": 0.9434,
      "step": 24830
    },
    {
      "epoch": 1.501178461352511,
      "grad_norm": 0.8765348187200152,
      "learning_rate": 5.406589261863824e-06,
      "loss": 0.9349,
      "step": 24840
    },
    {
      "epoch": 1.501782800507645,
      "grad_norm": 0.605094709523364,
      "learning_rate": 5.403269280894586e-06,
      "loss": 0.913,
      "step": 24850
    },
    {
      "epoch": 1.5023871396627788,
      "grad_norm": 0.5646927157546151,
      "learning_rate": 5.39994912095375e-06,
      "loss": 0.9299,
      "step": 24860
    },
    {
      "epoch": 1.5029914788179126,
      "grad_norm": 0.5587592563848658,
      "learning_rate": 5.39662878351481e-06,
      "loss": 0.938,
      "step": 24870
    },
    {
      "epoch": 1.5035958179730464,
      "grad_norm": 0.6182669339387755,
      "learning_rate": 5.393308270051336e-06,
      "loss": 0.9581,
      "step": 24880
    },
    {
      "epoch": 1.5042001571281802,
      "grad_norm": 0.5565209161500577,
      "learning_rate": 5.38998758203698e-06,
      "loss": 0.9265,
      "step": 24890
    },
    {
      "epoch": 1.504804496283314,
      "grad_norm": 0.6217507169907869,
      "learning_rate": 5.386666720945466e-06,
      "loss": 0.9345,
      "step": 24900
    },
    {
      "epoch": 1.505408835438448,
      "grad_norm": 0.6406947531859823,
      "learning_rate": 5.383345688250599e-06,
      "loss": 0.9478,
      "step": 24910
    },
    {
      "epoch": 1.506013174593582,
      "grad_norm": 0.6206465542025668,
      "learning_rate": 5.380024485426262e-06,
      "loss": 0.9323,
      "step": 24920
    },
    {
      "epoch": 1.5066175137487159,
      "grad_norm": 0.6410990777965546,
      "learning_rate": 5.376703113946405e-06,
      "loss": 0.924,
      "step": 24930
    },
    {
      "epoch": 1.5072218529038497,
      "grad_norm": 0.6273856086626944,
      "learning_rate": 5.373381575285063e-06,
      "loss": 0.941,
      "step": 24940
    },
    {
      "epoch": 1.5078261920589835,
      "grad_norm": 0.8347017688219457,
      "learning_rate": 5.370059870916338e-06,
      "loss": 0.9294,
      "step": 24950
    },
    {
      "epoch": 1.5084305312141173,
      "grad_norm": 0.8068542653111551,
      "learning_rate": 5.366738002314409e-06,
      "loss": 0.9373,
      "step": 24960
    },
    {
      "epoch": 1.509034870369251,
      "grad_norm": 0.8329047474556546,
      "learning_rate": 5.3634159709535265e-06,
      "loss": 0.939,
      "step": 24970
    },
    {
      "epoch": 1.5096392095243851,
      "grad_norm": 0.8036045907060764,
      "learning_rate": 5.360093778308014e-06,
      "loss": 0.9557,
      "step": 24980
    },
    {
      "epoch": 1.510243548679519,
      "grad_norm": 0.8724936878030212,
      "learning_rate": 5.356771425852265e-06,
      "loss": 0.9445,
      "step": 24990
    },
    {
      "epoch": 1.510847887834653,
      "grad_norm": 0.5718648496158292,
      "learning_rate": 5.353448915060748e-06,
      "loss": 0.932,
      "step": 25000
    },
    {
      "epoch": 1.5114522269897868,
      "grad_norm": 0.5361320371221349,
      "learning_rate": 5.350126247407997e-06,
      "loss": 0.9312,
      "step": 25010
    },
    {
      "epoch": 1.5120565661449206,
      "grad_norm": 0.5433642153068129,
      "learning_rate": 5.3468034243686185e-06,
      "loss": 0.9275,
      "step": 25020
    },
    {
      "epoch": 1.5126609053000544,
      "grad_norm": 0.5327652215574483,
      "learning_rate": 5.343480447417286e-06,
      "loss": 0.9311,
      "step": 25030
    },
    {
      "epoch": 1.5132652444551882,
      "grad_norm": 0.5300971442516206,
      "learning_rate": 5.340157318028743e-06,
      "loss": 0.9274,
      "step": 25040
    },
    {
      "epoch": 1.513869583610322,
      "grad_norm": 0.5568574526881223,
      "learning_rate": 5.336834037677799e-06,
      "loss": 0.9213,
      "step": 25050
    },
    {
      "epoch": 1.514473922765456,
      "grad_norm": 0.5700724867063462,
      "learning_rate": 5.333510607839335e-06,
      "loss": 0.9555,
      "step": 25060
    },
    {
      "epoch": 1.5150782619205898,
      "grad_norm": 0.5264939713962447,
      "learning_rate": 5.33018702998829e-06,
      "loss": 0.9353,
      "step": 25070
    },
    {
      "epoch": 1.5156826010757238,
      "grad_norm": 0.5495008937206775,
      "learning_rate": 5.326863305599676e-06,
      "loss": 0.9292,
      "step": 25080
    },
    {
      "epoch": 1.5162869402308576,
      "grad_norm": 0.511724912429098,
      "learning_rate": 5.323539436148569e-06,
      "loss": 0.9322,
      "step": 25090
    },
    {
      "epoch": 1.5168912793859914,
      "grad_norm": 0.6295904820107148,
      "learning_rate": 5.320215423110106e-06,
      "loss": 0.9234,
      "step": 25100
    },
    {
      "epoch": 1.5174956185411252,
      "grad_norm": 0.5641467392529417,
      "learning_rate": 5.31689126795949e-06,
      "loss": 0.9086,
      "step": 25110
    },
    {
      "epoch": 1.518099957696259,
      "grad_norm": 0.5574782585023906,
      "learning_rate": 5.313566972171986e-06,
      "loss": 0.9093,
      "step": 25120
    },
    {
      "epoch": 1.5187042968513929,
      "grad_norm": 0.5961576601894568,
      "learning_rate": 5.310242537222923e-06,
      "loss": 0.939,
      "step": 25130
    },
    {
      "epoch": 1.5193086360065269,
      "grad_norm": 0.5765480262979534,
      "learning_rate": 5.306917964587692e-06,
      "loss": 0.9469,
      "step": 25140
    },
    {
      "epoch": 1.5199129751616607,
      "grad_norm": 0.5681235217743901,
      "learning_rate": 5.303593255741743e-06,
      "loss": 0.9292,
      "step": 25150
    },
    {
      "epoch": 1.5205173143167947,
      "grad_norm": 0.5480531718419404,
      "learning_rate": 5.300268412160585e-06,
      "loss": 0.92,
      "step": 25160
    },
    {
      "epoch": 1.5211216534719285,
      "grad_norm": 0.5478385749808464,
      "learning_rate": 5.296943435319793e-06,
      "loss": 0.9355,
      "step": 25170
    },
    {
      "epoch": 1.5217259926270623,
      "grad_norm": 0.5405497931577712,
      "learning_rate": 5.293618326694997e-06,
      "loss": 0.9074,
      "step": 25180
    },
    {
      "epoch": 1.5223303317821961,
      "grad_norm": 0.5264046387987237,
      "learning_rate": 5.290293087761884e-06,
      "loss": 0.9423,
      "step": 25190
    },
    {
      "epoch": 1.52293467093733,
      "grad_norm": 0.5529984620973368,
      "learning_rate": 5.286967719996202e-06,
      "loss": 0.9102,
      "step": 25200
    },
    {
      "epoch": 1.5235390100924637,
      "grad_norm": 0.5520220959776307,
      "learning_rate": 5.283642224873755e-06,
      "loss": 0.9308,
      "step": 25210
    },
    {
      "epoch": 1.5241433492475978,
      "grad_norm": 0.5270088411338348,
      "learning_rate": 5.280316603870402e-06,
      "loss": 0.9345,
      "step": 25220
    },
    {
      "epoch": 1.5247476884027316,
      "grad_norm": 0.5604217066139217,
      "learning_rate": 5.2769908584620625e-06,
      "loss": 0.9636,
      "step": 25230
    },
    {
      "epoch": 1.5253520275578656,
      "grad_norm": 0.5531553773663259,
      "learning_rate": 5.273664990124703e-06,
      "loss": 0.9239,
      "step": 25240
    },
    {
      "epoch": 1.5259563667129994,
      "grad_norm": 0.5355421801389098,
      "learning_rate": 5.270339000334354e-06,
      "loss": 0.9179,
      "step": 25250
    },
    {
      "epoch": 1.5265607058681332,
      "grad_norm": 0.615318033585799,
      "learning_rate": 5.267012890567093e-06,
      "loss": 0.9159,
      "step": 25260
    },
    {
      "epoch": 1.527165045023267,
      "grad_norm": 0.5474946878706693,
      "learning_rate": 5.2636866622990545e-06,
      "loss": 0.9374,
      "step": 25270
    },
    {
      "epoch": 1.5277693841784008,
      "grad_norm": 0.53694373818359,
      "learning_rate": 5.260360317006424e-06,
      "loss": 0.9332,
      "step": 25280
    },
    {
      "epoch": 1.5283737233335348,
      "grad_norm": 0.5750388536969242,
      "learning_rate": 5.25703385616544e-06,
      "loss": 0.9409,
      "step": 25290
    },
    {
      "epoch": 1.5289780624886686,
      "grad_norm": 0.6342670899913346,
      "learning_rate": 5.253707281252389e-06,
      "loss": 0.9439,
      "step": 25300
    },
    {
      "epoch": 1.5295824016438027,
      "grad_norm": 0.610913971867008,
      "learning_rate": 5.2503805937436136e-06,
      "loss": 0.9373,
      "step": 25310
    },
    {
      "epoch": 1.5301867407989365,
      "grad_norm": 0.6435650310477017,
      "learning_rate": 5.247053795115502e-06,
      "loss": 0.9065,
      "step": 25320
    },
    {
      "epoch": 1.5307910799540703,
      "grad_norm": 0.6064943663841538,
      "learning_rate": 5.243726886844492e-06,
      "loss": 0.9406,
      "step": 25330
    },
    {
      "epoch": 1.531395419109204,
      "grad_norm": 0.656846394886137,
      "learning_rate": 5.240399870407073e-06,
      "loss": 0.9144,
      "step": 25340
    },
    {
      "epoch": 1.5319997582643379,
      "grad_norm": 0.7094892004234634,
      "learning_rate": 5.23707274727978e-06,
      "loss": 0.9299,
      "step": 25350
    },
    {
      "epoch": 1.5326040974194717,
      "grad_norm": 0.6663549201986683,
      "learning_rate": 5.2337455189391975e-06,
      "loss": 0.9162,
      "step": 25360
    },
    {
      "epoch": 1.5332084365746057,
      "grad_norm": 0.680862343698712,
      "learning_rate": 5.230418186861952e-06,
      "loss": 0.9291,
      "step": 25370
    },
    {
      "epoch": 1.5338127757297395,
      "grad_norm": 0.6830789485686696,
      "learning_rate": 5.22709075252472e-06,
      "loss": 0.9371,
      "step": 25380
    },
    {
      "epoch": 1.5344171148848735,
      "grad_norm": 0.7158287834715885,
      "learning_rate": 5.223763217404224e-06,
      "loss": 0.918,
      "step": 25390
    },
    {
      "epoch": 1.5350214540400073,
      "grad_norm": 0.6771683047073732,
      "learning_rate": 5.220435582977229e-06,
      "loss": 0.9202,
      "step": 25400
    },
    {
      "epoch": 1.5356257931951411,
      "grad_norm": 0.6626398846368972,
      "learning_rate": 5.2171078507205445e-06,
      "loss": 0.9409,
      "step": 25410
    },
    {
      "epoch": 1.536230132350275,
      "grad_norm": 0.6462441093685363,
      "learning_rate": 5.2137800221110245e-06,
      "loss": 0.9356,
      "step": 25420
    },
    {
      "epoch": 1.5368344715054088,
      "grad_norm": 0.6960277359787035,
      "learning_rate": 5.210452098625566e-06,
      "loss": 0.9192,
      "step": 25430
    },
    {
      "epoch": 1.5374388106605426,
      "grad_norm": 0.6286253596716214,
      "learning_rate": 5.207124081741104e-06,
      "loss": 0.9247,
      "step": 25440
    },
    {
      "epoch": 1.5380431498156766,
      "grad_norm": 0.7684302944711767,
      "learning_rate": 5.203795972934621e-06,
      "loss": 0.922,
      "step": 25450
    },
    {
      "epoch": 1.5386474889708104,
      "grad_norm": 0.7947482004036137,
      "learning_rate": 5.200467773683137e-06,
      "loss": 0.94,
      "step": 25460
    },
    {
      "epoch": 1.5392518281259444,
      "grad_norm": 0.838804320386143,
      "learning_rate": 5.197139485463709e-06,
      "loss": 0.9519,
      "step": 25470
    },
    {
      "epoch": 1.5398561672810782,
      "grad_norm": 0.7658220466284251,
      "learning_rate": 5.193811109753443e-06,
      "loss": 0.9228,
      "step": 25480
    },
    {
      "epoch": 1.540460506436212,
      "grad_norm": 0.7647489548855564,
      "learning_rate": 5.190482648029474e-06,
      "loss": 0.9222,
      "step": 25490
    },
    {
      "epoch": 1.5410648455913458,
      "grad_norm": 0.7563827069031019,
      "learning_rate": 5.18715410176898e-06,
      "loss": 0.9153,
      "step": 25500
    },
    {
      "epoch": 1.5416691847464796,
      "grad_norm": 0.8365780095920619,
      "learning_rate": 5.183825472449174e-06,
      "loss": 0.929,
      "step": 25510
    },
    {
      "epoch": 1.5422735239016134,
      "grad_norm": 0.7991861123225084,
      "learning_rate": 5.18049676154731e-06,
      "loss": 0.9076,
      "step": 25520
    },
    {
      "epoch": 1.5428778630567475,
      "grad_norm": 0.8048621616490019,
      "learning_rate": 5.177167970540672e-06,
      "loss": 0.9237,
      "step": 25530
    },
    {
      "epoch": 1.5434822022118813,
      "grad_norm": 0.7734975598678984,
      "learning_rate": 5.173839100906587e-06,
      "loss": 0.9163,
      "step": 25540
    },
    {
      "epoch": 1.5440865413670153,
      "grad_norm": 1.119163306485506,
      "learning_rate": 5.170510154122412e-06,
      "loss": 0.9383,
      "step": 25550
    },
    {
      "epoch": 1.544690880522149,
      "grad_norm": 1.2355977381533663,
      "learning_rate": 5.167181131665537e-06,
      "loss": 0.9101,
      "step": 25560
    },
    {
      "epoch": 1.545295219677283,
      "grad_norm": 1.1578603675013999,
      "learning_rate": 5.163852035013391e-06,
      "loss": 0.9232,
      "step": 25570
    },
    {
      "epoch": 1.5458995588324167,
      "grad_norm": 1.1605758822996983,
      "learning_rate": 5.16052286564343e-06,
      "loss": 0.9378,
      "step": 25580
    },
    {
      "epoch": 1.5465038979875505,
      "grad_norm": 1.1404758708343652,
      "learning_rate": 5.157193625033148e-06,
      "loss": 0.9273,
      "step": 25590
    },
    {
      "epoch": 1.5471082371426845,
      "grad_norm": 0.9145071291879595,
      "learning_rate": 5.1538643146600655e-06,
      "loss": 0.9444,
      "step": 25600
    },
    {
      "epoch": 1.5477125762978183,
      "grad_norm": 1.2991484525941206,
      "learning_rate": 5.150534936001736e-06,
      "loss": 0.9343,
      "step": 25610
    },
    {
      "epoch": 1.5483169154529524,
      "grad_norm": 0.8944752930190453,
      "learning_rate": 5.147205490535746e-06,
      "loss": 0.9198,
      "step": 25620
    },
    {
      "epoch": 1.5489212546080862,
      "grad_norm": 1.3259442828515904,
      "learning_rate": 5.1438759797397065e-06,
      "loss": 0.9613,
      "step": 25630
    },
    {
      "epoch": 1.54952559376322,
      "grad_norm": 1.236584863804695,
      "learning_rate": 5.14054640509126e-06,
      "loss": 0.9494,
      "step": 25640
    },
    {
      "epoch": 1.5501299329183538,
      "grad_norm": 2.1651514961345177,
      "learning_rate": 5.13721676806808e-06,
      "loss": 0.9213,
      "step": 25650
    },
    {
      "epoch": 1.5507342720734876,
      "grad_norm": 2.1903286706820073,
      "learning_rate": 5.1338870701478614e-06,
      "loss": 0.9128,
      "step": 25660
    },
    {
      "epoch": 1.5513386112286214,
      "grad_norm": 2.2900094008757663,
      "learning_rate": 5.130557312808333e-06,
      "loss": 0.9342,
      "step": 25670
    },
    {
      "epoch": 1.5519429503837554,
      "grad_norm": 2.5834108998599925,
      "learning_rate": 5.127227497527245e-06,
      "loss": 0.9273,
      "step": 25680
    },
    {
      "epoch": 1.5525472895388892,
      "grad_norm": 2.376168287666955,
      "learning_rate": 5.123897625782377e-06,
      "loss": 0.9293,
      "step": 25690
    },
    {
      "epoch": 1.5531516286940232,
      "grad_norm": 0.870189873118243,
      "learning_rate": 5.12056769905153e-06,
      "loss": 0.9295,
      "step": 25700
    },
    {
      "epoch": 1.553755967849157,
      "grad_norm": 1.051007151173739,
      "learning_rate": 5.117237718812531e-06,
      "loss": 0.9546,
      "step": 25710
    },
    {
      "epoch": 1.5543603070042908,
      "grad_norm": 0.8648837502793729,
      "learning_rate": 5.1139076865432315e-06,
      "loss": 0.9283,
      "step": 25720
    },
    {
      "epoch": 1.5549646461594246,
      "grad_norm": 0.8946069289994244,
      "learning_rate": 5.110577603721507e-06,
      "loss": 0.9286,
      "step": 25730
    },
    {
      "epoch": 1.5555689853145585,
      "grad_norm": 1.0195353939218657,
      "learning_rate": 5.107247471825253e-06,
      "loss": 0.9302,
      "step": 25740
    },
    {
      "epoch": 1.5561733244696923,
      "grad_norm": 0.9520057049318175,
      "learning_rate": 5.103917292332387e-06,
      "loss": 0.9325,
      "step": 25750
    },
    {
      "epoch": 1.5567776636248263,
      "grad_norm": 0.9999656492662283,
      "learning_rate": 5.100587066720848e-06,
      "loss": 0.9455,
      "step": 25760
    },
    {
      "epoch": 1.55738200277996,
      "grad_norm": 0.9988157968221073,
      "learning_rate": 5.097256796468598e-06,
      "loss": 0.9233,
      "step": 25770
    },
    {
      "epoch": 1.557986341935094,
      "grad_norm": 0.8971390514863067,
      "learning_rate": 5.093926483053613e-06,
      "loss": 0.8944,
      "step": 25780
    },
    {
      "epoch": 1.558590681090228,
      "grad_norm": 0.9408964147464126,
      "learning_rate": 5.0905961279538955e-06,
      "loss": 0.9236,
      "step": 25790
    },
    {
      "epoch": 1.5591950202453617,
      "grad_norm": 1.0587713843456572,
      "learning_rate": 5.087265732647462e-06,
      "loss": 0.9308,
      "step": 25800
    },
    {
      "epoch": 1.5597993594004955,
      "grad_norm": 1.133974684422884,
      "learning_rate": 5.083935298612344e-06,
      "loss": 0.9056,
      "step": 25810
    },
    {
      "epoch": 1.5604036985556293,
      "grad_norm": 1.0537213682437596,
      "learning_rate": 5.0806048273266e-06,
      "loss": 0.9274,
      "step": 25820
    },
    {
      "epoch": 1.5610080377107631,
      "grad_norm": 1.0244925212422469,
      "learning_rate": 5.077274320268295e-06,
      "loss": 0.9218,
      "step": 25830
    },
    {
      "epoch": 1.5616123768658972,
      "grad_norm": 1.0331662872298382,
      "learning_rate": 5.073943778915513e-06,
      "loss": 0.9412,
      "step": 25840
    },
    {
      "epoch": 1.562216716021031,
      "grad_norm": 1.073812897284331,
      "learning_rate": 5.0706132047463566e-06,
      "loss": 0.934,
      "step": 25850
    },
    {
      "epoch": 1.562821055176165,
      "grad_norm": 0.9866836211663912,
      "learning_rate": 5.0672825992389385e-06,
      "loss": 0.9251,
      "step": 25860
    },
    {
      "epoch": 1.5634253943312988,
      "grad_norm": 1.0231239993437653,
      "learning_rate": 5.063951963871388e-06,
      "loss": 0.9305,
      "step": 25870
    },
    {
      "epoch": 1.5640297334864326,
      "grad_norm": 1.0385677366708708,
      "learning_rate": 5.0606213001218474e-06,
      "loss": 0.9206,
      "step": 25880
    },
    {
      "epoch": 1.5646340726415664,
      "grad_norm": 1.036345186275833,
      "learning_rate": 5.0572906094684685e-06,
      "loss": 0.9253,
      "step": 25890
    },
    {
      "epoch": 1.5652384117967002,
      "grad_norm": 1.142363563968631,
      "learning_rate": 5.0539598933894195e-06,
      "loss": 0.918,
      "step": 25900
    },
    {
      "epoch": 1.565842750951834,
      "grad_norm": 1.1499087774908776,
      "learning_rate": 5.050629153362877e-06,
      "loss": 0.9283,
      "step": 25910
    },
    {
      "epoch": 1.566447090106968,
      "grad_norm": 1.2066881531981957,
      "learning_rate": 5.04729839086703e-06,
      "loss": 0.9665,
      "step": 25920
    },
    {
      "epoch": 1.567051429262102,
      "grad_norm": 1.0886914769098222,
      "learning_rate": 5.043967607380076e-06,
      "loss": 0.9144,
      "step": 25930
    },
    {
      "epoch": 1.5676557684172359,
      "grad_norm": 1.085450693017567,
      "learning_rate": 5.0406368043802235e-06,
      "loss": 0.9093,
      "step": 25940
    },
    {
      "epoch": 1.5682601075723697,
      "grad_norm": 0.9891846310233481,
      "learning_rate": 5.037305983345686e-06,
      "loss": 0.9367,
      "step": 25950
    },
    {
      "epoch": 1.5688644467275035,
      "grad_norm": 0.9963790541513797,
      "learning_rate": 5.033975145754691e-06,
      "loss": 0.93,
      "step": 25960
    },
    {
      "epoch": 1.5694687858826373,
      "grad_norm": 0.9865967084762404,
      "learning_rate": 5.030644293085467e-06,
      "loss": 0.9306,
      "step": 25970
    },
    {
      "epoch": 1.570073125037771,
      "grad_norm": 0.9973800819386068,
      "learning_rate": 5.027313426816253e-06,
      "loss": 0.9236,
      "step": 25980
    },
    {
      "epoch": 1.570677464192905,
      "grad_norm": 1.025192235084345,
      "learning_rate": 5.023982548425294e-06,
      "loss": 0.9217,
      "step": 25990
    },
    {
      "epoch": 1.571281803348039,
      "grad_norm": 1.003606142469775,
      "learning_rate": 5.02065165939084e-06,
      "loss": 0.9196,
      "step": 26000
    },
    {
      "epoch": 1.571886142503173,
      "grad_norm": 1.042744971044762,
      "learning_rate": 5.017320761191142e-06,
      "loss": 0.9181,
      "step": 26010
    },
    {
      "epoch": 1.5724904816583067,
      "grad_norm": 1.0393084883227832,
      "learning_rate": 5.013989855304463e-06,
      "loss": 0.9285,
      "step": 26020
    },
    {
      "epoch": 1.5730948208134405,
      "grad_norm": 1.0740141005569745,
      "learning_rate": 5.010658943209061e-06,
      "loss": 0.9157,
      "step": 26030
    },
    {
      "epoch": 1.5736991599685743,
      "grad_norm": 1.0353080282987033,
      "learning_rate": 5.007328026383201e-06,
      "loss": 0.9453,
      "step": 26040
    },
    {
      "epoch": 1.5743034991237081,
      "grad_norm": 0.9779490300492935,
      "learning_rate": 5.0039971063051506e-06,
      "loss": 0.9464,
      "step": 26050
    },
    {
      "epoch": 1.574907838278842,
      "grad_norm": 1.0734494690786855,
      "learning_rate": 5.000666184453178e-06,
      "loss": 0.9097,
      "step": 26060
    },
    {
      "epoch": 1.575512177433976,
      "grad_norm": 0.9897734471954108,
      "learning_rate": 4.997335262305552e-06,
      "loss": 0.9171,
      "step": 26070
    },
    {
      "epoch": 1.5761165165891098,
      "grad_norm": 0.9603379912784669,
      "learning_rate": 4.99400434134054e-06,
      "loss": 0.9248,
      "step": 26080
    },
    {
      "epoch": 1.5767208557442438,
      "grad_norm": 0.9228970233585315,
      "learning_rate": 4.990673423036413e-06,
      "loss": 0.9202,
      "step": 26090
    },
    {
      "epoch": 1.5773251948993776,
      "grad_norm": 1.1794627083838227,
      "learning_rate": 4.987342508871435e-06,
      "loss": 0.916,
      "step": 26100
    },
    {
      "epoch": 1.5779295340545114,
      "grad_norm": 1.1121431300687952,
      "learning_rate": 4.9840116003238755e-06,
      "loss": 0.9596,
      "step": 26110
    },
    {
      "epoch": 1.5785338732096452,
      "grad_norm": 1.1510909770573559,
      "learning_rate": 4.980680698871994e-06,
      "loss": 0.9446,
      "step": 26120
    },
    {
      "epoch": 1.579138212364779,
      "grad_norm": 1.206898411951541,
      "learning_rate": 4.977349805994052e-06,
      "loss": 0.9255,
      "step": 26130
    },
    {
      "epoch": 1.5797425515199128,
      "grad_norm": 1.1342969650408612,
      "learning_rate": 4.974018923168304e-06,
      "loss": 0.9299,
      "step": 26140
    },
    {
      "epoch": 1.5803468906750469,
      "grad_norm": 1.170289375363151,
      "learning_rate": 4.9706880518730035e-06,
      "loss": 0.9391,
      "step": 26150
    },
    {
      "epoch": 1.5809512298301807,
      "grad_norm": 1.0496278904169514,
      "learning_rate": 4.967357193586394e-06,
      "loss": 0.9238,
      "step": 26160
    },
    {
      "epoch": 1.5815555689853147,
      "grad_norm": 1.126234443121289,
      "learning_rate": 4.964026349786718e-06,
      "loss": 0.899,
      "step": 26170
    },
    {
      "epoch": 1.5821599081404485,
      "grad_norm": 1.1347189258346306,
      "learning_rate": 4.960695521952212e-06,
      "loss": 0.9275,
      "step": 26180
    },
    {
      "epoch": 1.5827642472955823,
      "grad_norm": 1.1009282133778555,
      "learning_rate": 4.957364711561098e-06,
      "loss": 0.9168,
      "step": 26190
    },
    {
      "epoch": 1.583368586450716,
      "grad_norm": 1.1225844815673176,
      "learning_rate": 4.954033920091599e-06,
      "loss": 0.939,
      "step": 26200
    },
    {
      "epoch": 1.58397292560585,
      "grad_norm": 1.185183737250558,
      "learning_rate": 4.950703149021926e-06,
      "loss": 0.9269,
      "step": 26210
    },
    {
      "epoch": 1.5845772647609837,
      "grad_norm": 1.170194770014627,
      "learning_rate": 4.947372399830278e-06,
      "loss": 0.9388,
      "step": 26220
    },
    {
      "epoch": 1.5851816039161177,
      "grad_norm": 1.1168819248849862,
      "learning_rate": 4.944041673994849e-06,
      "loss": 0.9574,
      "step": 26230
    },
    {
      "epoch": 1.5857859430712515,
      "grad_norm": 1.1103202715717237,
      "learning_rate": 4.940710972993823e-06,
      "loss": 0.9166,
      "step": 26240
    },
    {
      "epoch": 1.5863902822263856,
      "grad_norm": 1.3970417725624233,
      "learning_rate": 4.937380298305367e-06,
      "loss": 0.9142,
      "step": 26250
    },
    {
      "epoch": 1.5869946213815194,
      "grad_norm": 1.325250298028764,
      "learning_rate": 4.9340496514076426e-06,
      "loss": 0.9168,
      "step": 26260
    },
    {
      "epoch": 1.5875989605366532,
      "grad_norm": 1.4355224563834648,
      "learning_rate": 4.930719033778796e-06,
      "loss": 0.9387,
      "step": 26270
    },
    {
      "epoch": 1.588203299691787,
      "grad_norm": 1.341306523944089,
      "learning_rate": 4.92738844689696e-06,
      "loss": 0.9662,
      "step": 26280
    },
    {
      "epoch": 1.5888076388469208,
      "grad_norm": 1.4293212041784922,
      "learning_rate": 4.924057892240255e-06,
      "loss": 0.8993,
      "step": 26290
    },
    {
      "epoch": 1.5894119780020548,
      "grad_norm": 0.8246833103032736,
      "learning_rate": 4.920727371286787e-06,
      "loss": 0.942,
      "step": 26300
    },
    {
      "epoch": 1.5900163171571886,
      "grad_norm": 0.8227680165610858,
      "learning_rate": 4.917396885514651e-06,
      "loss": 0.9107,
      "step": 26310
    },
    {
      "epoch": 1.5906206563123226,
      "grad_norm": 0.8555478160224953,
      "learning_rate": 4.914066436401915e-06,
      "loss": 0.9282,
      "step": 26320
    },
    {
      "epoch": 1.5912249954674564,
      "grad_norm": 0.8525732833201712,
      "learning_rate": 4.910736025426642e-06,
      "loss": 0.931,
      "step": 26330
    },
    {
      "epoch": 1.5918293346225902,
      "grad_norm": 0.8520978853241207,
      "learning_rate": 4.907405654066876e-06,
      "loss": 0.939,
      "step": 26340
    },
    {
      "epoch": 1.592433673777724,
      "grad_norm": 0.7981743725438954,
      "learning_rate": 4.904075323800637e-06,
      "loss": 0.9308,
      "step": 26350
    },
    {
      "epoch": 1.5930380129328578,
      "grad_norm": 0.7702941241513368,
      "learning_rate": 4.900745036105933e-06,
      "loss": 0.9482,
      "step": 26360
    },
    {
      "epoch": 1.5936423520879917,
      "grad_norm": 0.7776441582442999,
      "learning_rate": 4.897414792460755e-06,
      "loss": 0.9312,
      "step": 26370
    },
    {
      "epoch": 1.5942466912431257,
      "grad_norm": 0.7778515280276301,
      "learning_rate": 4.894084594343065e-06,
      "loss": 0.925,
      "step": 26380
    },
    {
      "epoch": 1.5948510303982595,
      "grad_norm": 0.7723257748622294,
      "learning_rate": 4.890754443230815e-06,
      "loss": 0.9236,
      "step": 26390
    },
    {
      "epoch": 1.5954553695533935,
      "grad_norm": 0.7658686467998368,
      "learning_rate": 4.887424340601929e-06,
      "loss": 0.9251,
      "step": 26400
    },
    {
      "epoch": 1.5960597087085273,
      "grad_norm": 0.7750451266165372,
      "learning_rate": 4.884094287934314e-06,
      "loss": 0.9126,
      "step": 26410
    },
    {
      "epoch": 1.5966640478636611,
      "grad_norm": 0.7497944965314435,
      "learning_rate": 4.880764286705851e-06,
      "loss": 0.9299,
      "step": 26420
    },
    {
      "epoch": 1.597268387018795,
      "grad_norm": 0.8150259219210402,
      "learning_rate": 4.877434338394405e-06,
      "loss": 0.9061,
      "step": 26430
    },
    {
      "epoch": 1.5978727261739287,
      "grad_norm": 0.8855265785179022,
      "learning_rate": 4.874104444477806e-06,
      "loss": 0.9407,
      "step": 26440
    },
    {
      "epoch": 1.5984770653290625,
      "grad_norm": 0.5895912372705706,
      "learning_rate": 4.870774606433871e-06,
      "loss": 0.9172,
      "step": 26450
    },
    {
      "epoch": 1.5990814044841966,
      "grad_norm": 0.5856860585357047,
      "learning_rate": 4.867444825740389e-06,
      "loss": 0.9307,
      "step": 26460
    },
    {
      "epoch": 1.5996857436393304,
      "grad_norm": 0.5948474788306042,
      "learning_rate": 4.864115103875117e-06,
      "loss": 0.9093,
      "step": 26470
    },
    {
      "epoch": 1.6002900827944644,
      "grad_norm": 0.5702327706232243,
      "learning_rate": 4.8607854423157955e-06,
      "loss": 0.9269,
      "step": 26480
    },
    {
      "epoch": 1.6008944219495982,
      "grad_norm": 0.5417826012953707,
      "learning_rate": 4.8574558425401345e-06,
      "loss": 0.9341,
      "step": 26490
    },
    {
      "epoch": 1.601498761104732,
      "grad_norm": 0.646404666609463,
      "learning_rate": 4.854126306025813e-06,
      "loss": 0.8996,
      "step": 26500
    },
    {
      "epoch": 1.6021031002598658,
      "grad_norm": 0.6457680874161628,
      "learning_rate": 4.8507968342504875e-06,
      "loss": 0.9525,
      "step": 26510
    },
    {
      "epoch": 1.6027074394149996,
      "grad_norm": 0.7100155684707448,
      "learning_rate": 4.847467428691783e-06,
      "loss": 0.9508,
      "step": 26520
    },
    {
      "epoch": 1.6033117785701334,
      "grad_norm": 0.6657314539082194,
      "learning_rate": 4.8441380908272935e-06,
      "loss": 0.9223,
      "step": 26530
    },
    {
      "epoch": 1.6039161177252674,
      "grad_norm": 0.6506751308645118,
      "learning_rate": 4.840808822134585e-06,
      "loss": 0.9315,
      "step": 26540
    },
    {
      "epoch": 1.6045204568804012,
      "grad_norm": 0.8521558152037761,
      "learning_rate": 4.837479624091197e-06,
      "loss": 0.9347,
      "step": 26550
    },
    {
      "epoch": 1.6051247960355353,
      "grad_norm": 0.8520698279818828,
      "learning_rate": 4.834150498174628e-06,
      "loss": 0.907,
      "step": 26560
    },
    {
      "epoch": 1.605729135190669,
      "grad_norm": 0.8331079106737275,
      "learning_rate": 4.83082144586235e-06,
      "loss": 0.9355,
      "step": 26570
    },
    {
      "epoch": 1.6063334743458029,
      "grad_norm": 0.8750659000337639,
      "learning_rate": 4.827492468631808e-06,
      "loss": 0.9497,
      "step": 26580
    },
    {
      "epoch": 1.6069378135009367,
      "grad_norm": 0.9443038340742554,
      "learning_rate": 4.824163567960401e-06,
      "loss": 0.9464,
      "step": 26590
    },
    {
      "epoch": 1.6075421526560705,
      "grad_norm": 0.5442946433615261,
      "learning_rate": 4.8208347453255035e-06,
      "loss": 0.8995,
      "step": 26600
    },
    {
      "epoch": 1.6081464918112043,
      "grad_norm": 0.4956088241809183,
      "learning_rate": 4.817506002204455e-06,
      "loss": 0.9353,
      "step": 26610
    },
    {
      "epoch": 1.6087508309663383,
      "grad_norm": 0.529590223032268,
      "learning_rate": 4.814177340074552e-06,
      "loss": 0.9674,
      "step": 26620
    },
    {
      "epoch": 1.6093551701214723,
      "grad_norm": 0.5440362373993227,
      "learning_rate": 4.8108487604130645e-06,
      "loss": 0.9511,
      "step": 26630
    },
    {
      "epoch": 1.6099595092766061,
      "grad_norm": 0.5373674639904178,
      "learning_rate": 4.8075202646972216e-06,
      "loss": 0.908,
      "step": 26640
    },
    {
      "epoch": 1.61056384843174,
      "grad_norm": 0.5233883726919283,
      "learning_rate": 4.804191854404215e-06,
      "loss": 0.9278,
      "step": 26650
    },
    {
      "epoch": 1.6111681875868737,
      "grad_norm": 0.509363665624484,
      "learning_rate": 4.8008635310111995e-06,
      "loss": 0.9426,
      "step": 26660
    },
    {
      "epoch": 1.6117725267420075,
      "grad_norm": 0.5234340054230651,
      "learning_rate": 4.7975352959952875e-06,
      "loss": 0.9186,
      "step": 26670
    },
    {
      "epoch": 1.6123768658971414,
      "grad_norm": 0.5517044750143772,
      "learning_rate": 4.7942071508335605e-06,
      "loss": 0.9571,
      "step": 26680
    },
    {
      "epoch": 1.6129812050522754,
      "grad_norm": 0.5182221386890951,
      "learning_rate": 4.79087909700305e-06,
      "loss": 0.9594,
      "step": 26690
    },
    {
      "epoch": 1.6135855442074092,
      "grad_norm": 0.5921367162961881,
      "learning_rate": 4.787551135980753e-06,
      "loss": 0.9279,
      "step": 26700
    },
    {
      "epoch": 1.6141898833625432,
      "grad_norm": 0.5603120231246512,
      "learning_rate": 4.78422326924363e-06,
      "loss": 0.9334,
      "step": 26710
    },
    {
      "epoch": 1.614794222517677,
      "grad_norm": 0.5437730051893302,
      "learning_rate": 4.7808954982685855e-06,
      "loss": 0.907,
      "step": 26720
    },
    {
      "epoch": 1.6153985616728108,
      "grad_norm": 0.5226481124834631,
      "learning_rate": 4.777567824532495e-06,
      "loss": 0.9209,
      "step": 26730
    },
    {
      "epoch": 1.6160029008279446,
      "grad_norm": 0.5759004935924458,
      "learning_rate": 4.774240249512186e-06,
      "loss": 0.912,
      "step": 26740
    },
    {
      "epoch": 1.6166072399830784,
      "grad_norm": 0.5317313425080706,
      "learning_rate": 4.770912774684439e-06,
      "loss": 0.9251,
      "step": 26750
    },
    {
      "epoch": 1.6172115791382122,
      "grad_norm": 0.5219969850059324,
      "learning_rate": 4.767585401525996e-06,
      "loss": 0.9164,
      "step": 26760
    },
    {
      "epoch": 1.6178159182933463,
      "grad_norm": 0.5398824708514468,
      "learning_rate": 4.76425813151355e-06,
      "loss": 0.9268,
      "step": 26770
    },
    {
      "epoch": 1.61842025744848,
      "grad_norm": 0.5426289328942654,
      "learning_rate": 4.7609309661237485e-06,
      "loss": 0.925,
      "step": 26780
    },
    {
      "epoch": 1.619024596603614,
      "grad_norm": 0.5412258014841264,
      "learning_rate": 4.757603906833192e-06,
      "loss": 0.907,
      "step": 26790
    },
    {
      "epoch": 1.619628935758748,
      "grad_norm": 0.5371693293357467,
      "learning_rate": 4.754276955118439e-06,
      "loss": 0.9224,
      "step": 26800
    },
    {
      "epoch": 1.6202332749138817,
      "grad_norm": 0.5374754509218702,
      "learning_rate": 4.750950112455992e-06,
      "loss": 0.9242,
      "step": 26810
    },
    {
      "epoch": 1.6208376140690155,
      "grad_norm": 0.5361171486338071,
      "learning_rate": 4.747623380322312e-06,
      "loss": 0.9308,
      "step": 26820
    },
    {
      "epoch": 1.6214419532241493,
      "grad_norm": 0.5279326225702382,
      "learning_rate": 4.744296760193809e-06,
      "loss": 0.9221,
      "step": 26830
    },
    {
      "epoch": 1.622046292379283,
      "grad_norm": 0.5294536035839044,
      "learning_rate": 4.740970253546839e-06,
      "loss": 0.9427,
      "step": 26840
    },
    {
      "epoch": 1.6226506315344171,
      "grad_norm": 0.5444931966987818,
      "learning_rate": 4.737643861857715e-06,
      "loss": 0.9213,
      "step": 26850
    },
    {
      "epoch": 1.623254970689551,
      "grad_norm": 0.5492706543591206,
      "learning_rate": 4.734317586602696e-06,
      "loss": 0.9322,
      "step": 26860
    },
    {
      "epoch": 1.623859309844685,
      "grad_norm": 0.5420210930149884,
      "learning_rate": 4.7309914292579845e-06,
      "loss": 0.9302,
      "step": 26870
    },
    {
      "epoch": 1.6244636489998188,
      "grad_norm": 0.5202852539861623,
      "learning_rate": 4.727665391299737e-06,
      "loss": 0.935,
      "step": 26880
    },
    {
      "epoch": 1.6250679881549526,
      "grad_norm": 0.5813698065906621,
      "learning_rate": 4.724339474204058e-06,
      "loss": 0.9117,
      "step": 26890
    },
    {
      "epoch": 1.6256723273100864,
      "grad_norm": 0.6582789140299272,
      "learning_rate": 4.72101367944699e-06,
      "loss": 0.9151,
      "step": 26900
    },
    {
      "epoch": 1.6262766664652202,
      "grad_norm": 0.5730638098774922,
      "learning_rate": 4.717688008504529e-06,
      "loss": 0.9384,
      "step": 26910
    },
    {
      "epoch": 1.626881005620354,
      "grad_norm": 0.7523414520843882,
      "learning_rate": 4.714362462852614e-06,
      "loss": 0.9276,
      "step": 26920
    },
    {
      "epoch": 1.627485344775488,
      "grad_norm": 0.6482797417045263,
      "learning_rate": 4.711037043967126e-06,
      "loss": 0.9324,
      "step": 26930
    },
    {
      "epoch": 1.6280896839306218,
      "grad_norm": 0.6901000510769487,
      "learning_rate": 4.707711753323892e-06,
      "loss": 0.9413,
      "step": 26940
    },
    {
      "epoch": 1.6286940230857558,
      "grad_norm": 0.686288940202985,
      "learning_rate": 4.704386592398684e-06,
      "loss": 0.9447,
      "step": 26950
    },
    {
      "epoch": 1.6292983622408896,
      "grad_norm": 0.7053826906832571,
      "learning_rate": 4.701061562667211e-06,
      "loss": 0.9164,
      "step": 26960
    },
    {
      "epoch": 1.6299027013960234,
      "grad_norm": 0.633892867019406,
      "learning_rate": 4.697736665605127e-06,
      "loss": 0.9141,
      "step": 26970
    },
    {
      "epoch": 1.6305070405511572,
      "grad_norm": 0.7279680293555136,
      "learning_rate": 4.69441190268803e-06,
      "loss": 0.9329,
      "step": 26980
    },
    {
      "epoch": 1.631111379706291,
      "grad_norm": 0.6738772796273035,
      "learning_rate": 4.691087275391454e-06,
      "loss": 0.911,
      "step": 26990
    },
    {
      "epoch": 1.631715718861425,
      "grad_norm": 0.660674340450183,
      "learning_rate": 4.687762785190873e-06,
      "loss": 0.9352,
      "step": 27000
    },
    {
      "epoch": 1.6323200580165589,
      "grad_norm": 0.6186126346725852,
      "learning_rate": 4.6844384335617035e-06,
      "loss": 0.9354,
      "step": 27010
    },
    {
      "epoch": 1.632924397171693,
      "grad_norm": 0.6637131287284315,
      "learning_rate": 4.681114221979297e-06,
      "loss": 0.9383,
      "step": 27020
    },
    {
      "epoch": 1.6335287363268267,
      "grad_norm": 0.6709107255116934,
      "learning_rate": 4.677790151918945e-06,
      "loss": 0.9233,
      "step": 27030
    },
    {
      "epoch": 1.6341330754819605,
      "grad_norm": 0.6410581472298236,
      "learning_rate": 4.674466224855874e-06,
      "loss": 0.9213,
      "step": 27040
    },
    {
      "epoch": 1.6347374146370943,
      "grad_norm": 0.8034656058111185,
      "learning_rate": 4.671142442265252e-06,
      "loss": 0.9119,
      "step": 27050
    },
    {
      "epoch": 1.6353417537922281,
      "grad_norm": 0.7828599156256191,
      "learning_rate": 4.6678188056221744e-06,
      "loss": 0.9412,
      "step": 27060
    },
    {
      "epoch": 1.635946092947362,
      "grad_norm": 0.7989635418415998,
      "learning_rate": 4.6644953164016795e-06,
      "loss": 0.928,
      "step": 27070
    },
    {
      "epoch": 1.636550432102496,
      "grad_norm": 0.7915148899045632,
      "learning_rate": 4.66117197607874e-06,
      "loss": 0.9443,
      "step": 27080
    },
    {
      "epoch": 1.6371547712576298,
      "grad_norm": 0.8030523164757561,
      "learning_rate": 4.6578487861282555e-06,
      "loss": 0.9197,
      "step": 27090
    },
    {
      "epoch": 1.6377591104127638,
      "grad_norm": 0.8126809376864675,
      "learning_rate": 4.654525748025065e-06,
      "loss": 0.932,
      "step": 27100
    },
    {
      "epoch": 1.6383634495678976,
      "grad_norm": 0.7545653873681402,
      "learning_rate": 4.651202863243942e-06,
      "loss": 0.9009,
      "step": 27110
    },
    {
      "epoch": 1.6389677887230314,
      "grad_norm": 0.7669329066945321,
      "learning_rate": 4.6478801332595815e-06,
      "loss": 0.9149,
      "step": 27120
    },
    {
      "epoch": 1.6395721278781652,
      "grad_norm": 0.8014471866527094,
      "learning_rate": 4.644557559546622e-06,
      "loss": 0.9025,
      "step": 27130
    },
    {
      "epoch": 1.640176467033299,
      "grad_norm": 0.7725432760728462,
      "learning_rate": 4.641235143579628e-06,
      "loss": 0.9138,
      "step": 27140
    },
    {
      "epoch": 1.6407808061884328,
      "grad_norm": 1.2191053411337243,
      "learning_rate": 4.6379128868330896e-06,
      "loss": 0.9251,
      "step": 27150
    },
    {
      "epoch": 1.6413851453435668,
      "grad_norm": 1.1848172413315827,
      "learning_rate": 4.634590790781432e-06,
      "loss": 0.9322,
      "step": 27160
    },
    {
      "epoch": 1.6419894844987006,
      "grad_norm": 1.2722634508905812,
      "learning_rate": 4.6312688568990074e-06,
      "loss": 0.9096,
      "step": 27170
    },
    {
      "epoch": 1.6425938236538347,
      "grad_norm": 1.2060298131419236,
      "learning_rate": 4.6279470866600954e-06,
      "loss": 0.928,
      "step": 27180
    },
    {
      "epoch": 1.6431981628089685,
      "grad_norm": 1.2175083913934246,
      "learning_rate": 4.624625481538903e-06,
      "loss": 0.9265,
      "step": 27190
    },
    {
      "epoch": 1.6438025019641023,
      "grad_norm": 0.8684648078225902,
      "learning_rate": 4.621304043009566e-06,
      "loss": 0.929,
      "step": 27200
    },
    {
      "epoch": 1.644406841119236,
      "grad_norm": 0.9919958929756366,
      "learning_rate": 4.61798277254614e-06,
      "loss": 0.9368,
      "step": 27210
    },
    {
      "epoch": 1.6450111802743699,
      "grad_norm": 1.2010800461854234,
      "learning_rate": 4.6146616716226146e-06,
      "loss": 0.926,
      "step": 27220
    },
    {
      "epoch": 1.6456155194295037,
      "grad_norm": 0.7788699100934598,
      "learning_rate": 4.611340741712901e-06,
      "loss": 0.9472,
      "step": 27230
    },
    {
      "epoch": 1.6462198585846377,
      "grad_norm": 1.05596151064457,
      "learning_rate": 4.608019984290828e-06,
      "loss": 0.9349,
      "step": 27240
    },
    {
      "epoch": 1.6468241977397715,
      "grad_norm": 2.497988355890021,
      "learning_rate": 4.604699400830159e-06,
      "loss": 0.9559,
      "step": 27250
    },
    {
      "epoch": 1.6474285368949055,
      "grad_norm": 2.374901210182557,
      "learning_rate": 4.6013789928045734e-06,
      "loss": 0.9379,
      "step": 27260
    },
    {
      "epoch": 1.6480328760500393,
      "grad_norm": 2.3239279043846492,
      "learning_rate": 4.598058761687672e-06,
      "loss": 0.9271,
      "step": 27270
    },
    {
      "epoch": 1.6486372152051731,
      "grad_norm": 2.289244062461535,
      "learning_rate": 4.5947387089529795e-06,
      "loss": 0.9186,
      "step": 27280
    },
    {
      "epoch": 1.649241554360307,
      "grad_norm": 2.227535949967949,
      "learning_rate": 4.591418836073944e-06,
      "loss": 0.9411,
      "step": 27290
    },
    {
      "epoch": 1.6498458935154408,
      "grad_norm": 0.9153778250320579,
      "learning_rate": 4.588099144523927e-06,
      "loss": 0.9096,
      "step": 27300
    },
    {
      "epoch": 1.6504502326705748,
      "grad_norm": 0.9442181791376509,
      "learning_rate": 4.5847796357762155e-06,
      "loss": 0.9383,
      "step": 27310
    },
    {
      "epoch": 1.6510545718257086,
      "grad_norm": 0.8803522298189124,
      "learning_rate": 4.581460311304011e-06,
      "loss": 0.943,
      "step": 27320
    },
    {
      "epoch": 1.6516589109808426,
      "grad_norm": 0.8596887539071623,
      "learning_rate": 4.57814117258044e-06,
      "loss": 0.9484,
      "step": 27330
    },
    {
      "epoch": 1.6522632501359764,
      "grad_norm": 0.8858725049618714,
      "learning_rate": 4.574822221078535e-06,
      "loss": 0.9309,
      "step": 27340
    },
    {
      "epoch": 1.6528675892911102,
      "grad_norm": 1.0154273228509834,
      "learning_rate": 4.571503458271257e-06,
      "loss": 0.9156,
      "step": 27350
    },
    {
      "epoch": 1.653471928446244,
      "grad_norm": 0.9534906211499345,
      "learning_rate": 4.56818488563148e-06,
      "loss": 0.9227,
      "step": 27360
    },
    {
      "epoch": 1.6540762676013778,
      "grad_norm": 0.971604518397412,
      "learning_rate": 4.564866504631987e-06,
      "loss": 0.9372,
      "step": 27370
    },
    {
      "epoch": 1.6546806067565116,
      "grad_norm": 1.010528136988153,
      "learning_rate": 4.561548316745485e-06,
      "loss": 0.9431,
      "step": 27380
    },
    {
      "epoch": 1.6552849459116457,
      "grad_norm": 0.9629646922497487,
      "learning_rate": 4.558230323444592e-06,
      "loss": 0.9488,
      "step": 27390
    },
    {
      "epoch": 1.6558892850667795,
      "grad_norm": 0.984580801392413,
      "learning_rate": 4.554912526201837e-06,
      "loss": 0.9255,
      "step": 27400
    },
    {
      "epoch": 1.6564936242219135,
      "grad_norm": 1.063952951109643,
      "learning_rate": 4.551594926489665e-06,
      "loss": 0.9399,
      "step": 27410
    },
    {
      "epoch": 1.6570979633770473,
      "grad_norm": 1.0083497633968181,
      "learning_rate": 4.548277525780431e-06,
      "loss": 0.9228,
      "step": 27420
    },
    {
      "epoch": 1.657702302532181,
      "grad_norm": 1.0292318641142133,
      "learning_rate": 4.544960325546406e-06,
      "loss": 0.9318,
      "step": 27430
    },
    {
      "epoch": 1.658306641687315,
      "grad_norm": 1.0557518101294667,
      "learning_rate": 4.541643327259766e-06,
      "loss": 0.9208,
      "step": 27440
    },
    {
      "epoch": 1.6589109808424487,
      "grad_norm": 1.1012237967007068,
      "learning_rate": 4.538326532392604e-06,
      "loss": 0.9097,
      "step": 27450
    },
    {
      "epoch": 1.6595153199975825,
      "grad_norm": 1.0371946133413965,
      "learning_rate": 4.535009942416916e-06,
      "loss": 0.9256,
      "step": 27460
    },
    {
      "epoch": 1.6601196591527165,
      "grad_norm": 1.0545601402786666,
      "learning_rate": 4.531693558804611e-06,
      "loss": 0.925,
      "step": 27470
    },
    {
      "epoch": 1.6607239983078503,
      "grad_norm": 1.0306633799473977,
      "learning_rate": 4.528377383027508e-06,
      "loss": 0.9466,
      "step": 27480
    },
    {
      "epoch": 1.6613283374629844,
      "grad_norm": 1.101002036492239,
      "learning_rate": 4.525061416557327e-06,
      "loss": 0.9315,
      "step": 27490
    },
    {
      "epoch": 1.6619326766181182,
      "grad_norm": 1.1638059208900207,
      "learning_rate": 4.521745660865704e-06,
      "loss": 0.9476,
      "step": 27500
    },
    {
      "epoch": 1.662537015773252,
      "grad_norm": 1.1046363665358636,
      "learning_rate": 4.518430117424176e-06,
      "loss": 0.9358,
      "step": 27510
    },
    {
      "epoch": 1.6631413549283858,
      "grad_norm": 1.0586402476588566,
      "learning_rate": 4.515114787704186e-06,
      "loss": 0.9076,
      "step": 27520
    },
    {
      "epoch": 1.6637456940835196,
      "grad_norm": 1.0417259591512342,
      "learning_rate": 4.511799673177081e-06,
      "loss": 0.9257,
      "step": 27530
    },
    {
      "epoch": 1.6643500332386534,
      "grad_norm": 1.0628045004289932,
      "learning_rate": 4.508484775314121e-06,
      "loss": 0.9152,
      "step": 27540
    },
    {
      "epoch": 1.6649543723937874,
      "grad_norm": 0.9754013878320819,
      "learning_rate": 4.505170095586456e-06,
      "loss": 0.9181,
      "step": 27550
    },
    {
      "epoch": 1.6655587115489212,
      "grad_norm": 1.0003173477848608,
      "learning_rate": 4.501855635465151e-06,
      "loss": 0.9003,
      "step": 27560
    },
    {
      "epoch": 1.6661630507040552,
      "grad_norm": 0.9434455139856954,
      "learning_rate": 4.498541396421167e-06,
      "loss": 0.903,
      "step": 27570
    },
    {
      "epoch": 1.666767389859189,
      "grad_norm": 0.9498678319953696,
      "learning_rate": 4.49522737992537e-06,
      "loss": 0.9278,
      "step": 27580
    },
    {
      "epoch": 1.6673717290143228,
      "grad_norm": 0.9903893223510205,
      "learning_rate": 4.4919135874485255e-06,
      "loss": 0.9358,
      "step": 27590
    },
    {
      "epoch": 1.6679760681694566,
      "grad_norm": 1.0547366304881938,
      "learning_rate": 4.488600020461302e-06,
      "loss": 0.9257,
      "step": 27600
    },
    {
      "epoch": 1.6685804073245905,
      "grad_norm": 1.0855267447896846,
      "learning_rate": 4.485286680434263e-06,
      "loss": 0.9188,
      "step": 27610
    },
    {
      "epoch": 1.6691847464797243,
      "grad_norm": 1.14583175579124,
      "learning_rate": 4.481973568837877e-06,
      "loss": 0.9056,
      "step": 27620
    },
    {
      "epoch": 1.6697890856348583,
      "grad_norm": 1.0383279579670204,
      "learning_rate": 4.478660687142509e-06,
      "loss": 0.938,
      "step": 27630
    },
    {
      "epoch": 1.6703934247899923,
      "grad_norm": 0.9973117444619705,
      "learning_rate": 4.47534803681842e-06,
      "loss": 0.9387,
      "step": 27640
    },
    {
      "epoch": 1.6709977639451261,
      "grad_norm": 1.0781597736983577,
      "learning_rate": 4.472035619335768e-06,
      "loss": 0.9365,
      "step": 27650
    },
    {
      "epoch": 1.67160210310026,
      "grad_norm": 1.0086605048833426,
      "learning_rate": 4.468723436164615e-06,
      "loss": 0.9419,
      "step": 27660
    },
    {
      "epoch": 1.6722064422553937,
      "grad_norm": 0.9904528479548492,
      "learning_rate": 4.46541148877491e-06,
      "loss": 0.929,
      "step": 27670
    },
    {
      "epoch": 1.6728107814105275,
      "grad_norm": 1.0233910939852584,
      "learning_rate": 4.462099778636501e-06,
      "loss": 0.9385,
      "step": 27680
    },
    {
      "epoch": 1.6734151205656613,
      "grad_norm": 0.9639341387716622,
      "learning_rate": 4.45878830721913e-06,
      "loss": 0.9355,
      "step": 27690
    },
    {
      "epoch": 1.6740194597207954,
      "grad_norm": 1.1717558786884754,
      "learning_rate": 4.455477075992437e-06,
      "loss": 0.9106,
      "step": 27700
    },
    {
      "epoch": 1.6746237988759292,
      "grad_norm": 1.1384265237157696,
      "learning_rate": 4.452166086425948e-06,
      "loss": 0.9241,
      "step": 27710
    },
    {
      "epoch": 1.6752281380310632,
      "grad_norm": 1.1633279402551957,
      "learning_rate": 4.448855339989088e-06,
      "loss": 0.9319,
      "step": 27720
    },
    {
      "epoch": 1.675832477186197,
      "grad_norm": 1.1612404304794983,
      "learning_rate": 4.445544838151173e-06,
      "loss": 0.8954,
      "step": 27730
    },
    {
      "epoch": 1.6764368163413308,
      "grad_norm": 1.1340392899417444,
      "learning_rate": 4.442234582381405e-06,
      "loss": 0.9116,
      "step": 27740
    },
    {
      "epoch": 1.6770411554964646,
      "grad_norm": 1.0933124543260229,
      "learning_rate": 4.438924574148886e-06,
      "loss": 0.9376,
      "step": 27750
    },
    {
      "epoch": 1.6776454946515984,
      "grad_norm": 1.0506221165887413,
      "learning_rate": 4.435614814922603e-06,
      "loss": 0.9241,
      "step": 27760
    },
    {
      "epoch": 1.6782498338067322,
      "grad_norm": 1.0731449323418225,
      "learning_rate": 4.432305306171429e-06,
      "loss": 0.943,
      "step": 27770
    },
    {
      "epoch": 1.6788541729618662,
      "grad_norm": 1.069398047672288,
      "learning_rate": 4.428996049364132e-06,
      "loss": 0.943,
      "step": 27780
    },
    {
      "epoch": 1.679458512117,
      "grad_norm": 1.1520911903455413,
      "learning_rate": 4.4256870459693675e-06,
      "loss": 0.9369,
      "step": 27790
    },
    {
      "epoch": 1.680062851272134,
      "grad_norm": 1.0763261852922708,
      "learning_rate": 4.422378297455673e-06,
      "loss": 0.8949,
      "step": 27800
    },
    {
      "epoch": 1.6806671904272679,
      "grad_norm": 1.2195842914892054,
      "learning_rate": 4.41906980529148e-06,
      "loss": 0.9251,
      "step": 27810
    },
    {
      "epoch": 1.6812715295824017,
      "grad_norm": 1.1275208778935877,
      "learning_rate": 4.415761570945103e-06,
      "loss": 0.9299,
      "step": 27820
    },
    {
      "epoch": 1.6818758687375355,
      "grad_norm": 1.2393702085132507,
      "learning_rate": 4.4124535958847395e-06,
      "loss": 0.9311,
      "step": 27830
    },
    {
      "epoch": 1.6824802078926693,
      "grad_norm": 1.1479771102291836,
      "learning_rate": 4.409145881578476e-06,
      "loss": 0.9035,
      "step": 27840
    },
    {
      "epoch": 1.683084547047803,
      "grad_norm": 1.375477897581971,
      "learning_rate": 4.405838429494283e-06,
      "loss": 0.9689,
      "step": 27850
    },
    {
      "epoch": 1.683688886202937,
      "grad_norm": 1.2962495106155394,
      "learning_rate": 4.402531241100011e-06,
      "loss": 0.9061,
      "step": 27860
    },
    {
      "epoch": 1.684293225358071,
      "grad_norm": 1.3505056164295135,
      "learning_rate": 4.399224317863396e-06,
      "loss": 0.9073,
      "step": 27870
    },
    {
      "epoch": 1.684897564513205,
      "grad_norm": 1.4685163455523527,
      "learning_rate": 4.39591766125206e-06,
      "loss": 0.951,
      "step": 27880
    },
    {
      "epoch": 1.6855019036683387,
      "grad_norm": 1.3792194479756674,
      "learning_rate": 4.392611272733498e-06,
      "loss": 0.9541,
      "step": 27890
    },
    {
      "epoch": 1.6861062428234725,
      "grad_norm": 0.8647969899655263,
      "learning_rate": 4.389305153775092e-06,
      "loss": 0.9628,
      "step": 27900
    },
    {
      "epoch": 1.6867105819786063,
      "grad_norm": 0.8363252518643338,
      "learning_rate": 4.385999305844106e-06,
      "loss": 0.934,
      "step": 27910
    },
    {
      "epoch": 1.6873149211337402,
      "grad_norm": 0.8748052027046616,
      "learning_rate": 4.382693730407678e-06,
      "loss": 0.9233,
      "step": 27920
    },
    {
      "epoch": 1.687919260288874,
      "grad_norm": 0.8952598102240186,
      "learning_rate": 4.3793884289328305e-06,
      "loss": 0.9224,
      "step": 27930
    },
    {
      "epoch": 1.688523599444008,
      "grad_norm": 0.8234164956760309,
      "learning_rate": 4.37608340288646e-06,
      "loss": 0.9077,
      "step": 27940
    },
    {
      "epoch": 1.6891279385991418,
      "grad_norm": 0.7696182428019646,
      "learning_rate": 4.372778653735344e-06,
      "loss": 0.912,
      "step": 27950
    },
    {
      "epoch": 1.6897322777542758,
      "grad_norm": 0.7701060705873991,
      "learning_rate": 4.369474182946133e-06,
      "loss": 0.9189,
      "step": 27960
    },
    {
      "epoch": 1.6903366169094096,
      "grad_norm": 0.7608107666338965,
      "learning_rate": 4.366169991985363e-06,
      "loss": 0.915,
      "step": 27970
    },
    {
      "epoch": 1.6909409560645434,
      "grad_norm": 0.7742137049554035,
      "learning_rate": 4.362866082319432e-06,
      "loss": 0.9466,
      "step": 27980
    },
    {
      "epoch": 1.6915452952196772,
      "grad_norm": 0.7679991902779509,
      "learning_rate": 4.359562455414624e-06,
      "loss": 0.9095,
      "step": 27990
    },
    {
      "epoch": 1.692149634374811,
      "grad_norm": 0.8544752041076246,
      "learning_rate": 4.356259112737096e-06,
      "loss": 0.9157,
      "step": 28000
    },
    {
      "epoch": 1.692753973529945,
      "grad_norm": 0.7837174651831658,
      "learning_rate": 4.352956055752877e-06,
      "loss": 0.9313,
      "step": 28010
    },
    {
      "epoch": 1.6933583126850789,
      "grad_norm": 0.7913855009297621,
      "learning_rate": 4.349653285927866e-06,
      "loss": 0.9173,
      "step": 28020
    },
    {
      "epoch": 1.6939626518402129,
      "grad_norm": 0.853914467183262,
      "learning_rate": 4.3463508047278404e-06,
      "loss": 0.9082,
      "step": 28030
    },
    {
      "epoch": 1.6945669909953467,
      "grad_norm": 0.8147627434128741,
      "learning_rate": 4.343048613618448e-06,
      "loss": 0.9238,
      "step": 28040
    },
    {
      "epoch": 1.6951713301504805,
      "grad_norm": 0.6046438435047382,
      "learning_rate": 4.3397467140652046e-06,
      "loss": 0.9409,
      "step": 28050
    },
    {
      "epoch": 1.6957756693056143,
      "grad_norm": 0.5683928370681905,
      "learning_rate": 4.3364451075335e-06,
      "loss": 0.9142,
      "step": 28060
    },
    {
      "epoch": 1.696380008460748,
      "grad_norm": 0.548058025415299,
      "learning_rate": 4.333143795488592e-06,
      "loss": 0.9382,
      "step": 28070
    },
    {
      "epoch": 1.696984347615882,
      "grad_norm": 0.5464462211858346,
      "learning_rate": 4.329842779395611e-06,
      "loss": 0.9317,
      "step": 28080
    },
    {
      "epoch": 1.697588686771016,
      "grad_norm": 0.5727019712332703,
      "learning_rate": 4.326542060719549e-06,
      "loss": 0.9278,
      "step": 28090
    },
    {
      "epoch": 1.6981930259261497,
      "grad_norm": 0.6238429816951516,
      "learning_rate": 4.323241640925276e-06,
      "loss": 0.9513,
      "step": 28100
    },
    {
      "epoch": 1.6987973650812838,
      "grad_norm": 0.6336548613948914,
      "learning_rate": 4.319941521477519e-06,
      "loss": 0.9291,
      "step": 28110
    },
    {
      "epoch": 1.6994017042364176,
      "grad_norm": 0.62575323069499,
      "learning_rate": 4.316641703840879e-06,
      "loss": 0.9034,
      "step": 28120
    },
    {
      "epoch": 1.7000060433915514,
      "grad_norm": 0.6296687627610604,
      "learning_rate": 4.313342189479823e-06,
      "loss": 0.9548,
      "step": 28130
    },
    {
      "epoch": 1.7006103825466852,
      "grad_norm": 0.6170402035746204,
      "learning_rate": 4.310042979858677e-06,
      "loss": 0.9655,
      "step": 28140
    },
    {
      "epoch": 1.701214721701819,
      "grad_norm": 0.8211625468584101,
      "learning_rate": 4.306744076441637e-06,
      "loss": 0.9369,
      "step": 28150
    },
    {
      "epoch": 1.7018190608569528,
      "grad_norm": 0.8124031639679851,
      "learning_rate": 4.3034454806927645e-06,
      "loss": 0.9052,
      "step": 28160
    },
    {
      "epoch": 1.7024234000120868,
      "grad_norm": 0.8013036597678181,
      "learning_rate": 4.3001471940759784e-06,
      "loss": 0.9189,
      "step": 28170
    },
    {
      "epoch": 1.7030277391672206,
      "grad_norm": 0.8097437174096903,
      "learning_rate": 4.296849218055066e-06,
      "loss": 0.9561,
      "step": 28180
    },
    {
      "epoch": 1.7036320783223546,
      "grad_norm": 0.847539700590884,
      "learning_rate": 4.293551554093676e-06,
      "loss": 0.9369,
      "step": 28190
    },
    {
      "epoch": 1.7042364174774884,
      "grad_norm": 0.5622933389120811,
      "learning_rate": 4.290254203655313e-06,
      "loss": 0.921,
      "step": 28200
    },
    {
      "epoch": 1.7048407566326222,
      "grad_norm": 0.5273192963739084,
      "learning_rate": 4.286957168203351e-06,
      "loss": 0.9157,
      "step": 28210
    },
    {
      "epoch": 1.705445095787756,
      "grad_norm": 0.519120823987634,
      "learning_rate": 4.283660449201019e-06,
      "loss": 0.9164,
      "step": 28220
    },
    {
      "epoch": 1.7060494349428899,
      "grad_norm": 0.5598572393526107,
      "learning_rate": 4.280364048111405e-06,
      "loss": 0.9161,
      "step": 28230
    },
    {
      "epoch": 1.7066537740980237,
      "grad_norm": 0.5597854324729941,
      "learning_rate": 4.277067966397458e-06,
      "loss": 0.9408,
      "step": 28240
    },
    {
      "epoch": 1.7072581132531577,
      "grad_norm": 0.5216003791321182,
      "learning_rate": 4.273772205521987e-06,
      "loss": 0.9409,
      "step": 28250
    },
    {
      "epoch": 1.7078624524082915,
      "grad_norm": 0.5319526539254827,
      "learning_rate": 4.270476766947652e-06,
      "loss": 0.9162,
      "step": 28260
    },
    {
      "epoch": 1.7084667915634255,
      "grad_norm": 0.5627786996045264,
      "learning_rate": 4.267181652136977e-06,
      "loss": 0.9056,
      "step": 28270
    },
    {
      "epoch": 1.7090711307185593,
      "grad_norm": 0.545375117083162,
      "learning_rate": 4.263886862552341e-06,
      "loss": 0.9209,
      "step": 28280
    },
    {
      "epoch": 1.7096754698736931,
      "grad_norm": 0.5229453450887879,
      "learning_rate": 4.260592399655973e-06,
      "loss": 0.9248,
      "step": 28290
    },
    {
      "epoch": 1.710279809028827,
      "grad_norm": 0.5224402876622195,
      "learning_rate": 4.2572982649099645e-06,
      "loss": 0.9522,
      "step": 28300
    },
    {
      "epoch": 1.7108841481839607,
      "grad_norm": 0.5169319635994611,
      "learning_rate": 4.2540044597762596e-06,
      "loss": 0.9126,
      "step": 28310
    },
    {
      "epoch": 1.7114884873390948,
      "grad_norm": 0.5345270198463166,
      "learning_rate": 4.25071098571665e-06,
      "loss": 0.926,
      "step": 28320
    },
    {
      "epoch": 1.7120928264942286,
      "grad_norm": 0.5314485029825494,
      "learning_rate": 4.247417844192789e-06,
      "loss": 0.9294,
      "step": 28330
    },
    {
      "epoch": 1.7126971656493626,
      "grad_norm": 0.5407995469612804,
      "learning_rate": 4.2441250366661765e-06,
      "loss": 0.9329,
      "step": 28340
    },
    {
      "epoch": 1.7133015048044964,
      "grad_norm": 0.5400833004929693,
      "learning_rate": 4.240832564598168e-06,
      "loss": 0.9247,
      "step": 28350
    },
    {
      "epoch": 1.7139058439596302,
      "grad_norm": 0.5359789955879173,
      "learning_rate": 4.237540429449965e-06,
      "loss": 0.9211,
      "step": 28360
    },
    {
      "epoch": 1.714510183114764,
      "grad_norm": 0.5384733668120537,
      "learning_rate": 4.234248632682625e-06,
      "loss": 0.9368,
      "step": 28370
    },
    {
      "epoch": 1.7151145222698978,
      "grad_norm": 0.5448593145911083,
      "learning_rate": 4.230957175757054e-06,
      "loss": 0.9359,
      "step": 28380
    },
    {
      "epoch": 1.7157188614250316,
      "grad_norm": 0.5494418718750299,
      "learning_rate": 4.2276660601340035e-06,
      "loss": 0.9167,
      "step": 28390
    },
    {
      "epoch": 1.7163232005801656,
      "grad_norm": 0.5134345107264927,
      "learning_rate": 4.224375287274077e-06,
      "loss": 0.9369,
      "step": 28400
    },
    {
      "epoch": 1.7169275397352994,
      "grad_norm": 0.5225699328073128,
      "learning_rate": 4.221084858637729e-06,
      "loss": 0.9326,
      "step": 28410
    },
    {
      "epoch": 1.7175318788904335,
      "grad_norm": 0.5533382379104111,
      "learning_rate": 4.217794775685251e-06,
      "loss": 0.9612,
      "step": 28420
    },
    {
      "epoch": 1.7181362180455673,
      "grad_norm": 0.5230799665949161,
      "learning_rate": 4.21450503987679e-06,
      "loss": 0.9313,
      "step": 28430
    },
    {
      "epoch": 1.718740557200701,
      "grad_norm": 0.522582413903175,
      "learning_rate": 4.211215652672341e-06,
      "loss": 0.9377,
      "step": 28440
    },
    {
      "epoch": 1.7193448963558349,
      "grad_norm": 0.5847133430275941,
      "learning_rate": 4.207926615531732e-06,
      "loss": 0.9233,
      "step": 28450
    },
    {
      "epoch": 1.7199492355109687,
      "grad_norm": 0.5222721781588695,
      "learning_rate": 4.204637929914649e-06,
      "loss": 0.9345,
      "step": 28460
    },
    {
      "epoch": 1.7205535746661025,
      "grad_norm": 0.5471993669412948,
      "learning_rate": 4.201349597280614e-06,
      "loss": 0.9289,
      "step": 28470
    },
    {
      "epoch": 1.7211579138212365,
      "grad_norm": 0.5503192523531073,
      "learning_rate": 4.198061619088995e-06,
      "loss": 0.9509,
      "step": 28480
    },
    {
      "epoch": 1.7217622529763703,
      "grad_norm": 0.5719271620417735,
      "learning_rate": 4.194773996799e-06,
      "loss": 0.9525,
      "step": 28490
    },
    {
      "epoch": 1.7223665921315043,
      "grad_norm": 0.6593825695807976,
      "learning_rate": 4.1914867318696875e-06,
      "loss": 0.9437,
      "step": 28500
    },
    {
      "epoch": 1.7229709312866381,
      "grad_norm": 0.6288954149181362,
      "learning_rate": 4.188199825759944e-06,
      "loss": 0.9183,
      "step": 28510
    },
    {
      "epoch": 1.723575270441772,
      "grad_norm": 0.6646357931438769,
      "learning_rate": 4.184913279928508e-06,
      "loss": 0.9414,
      "step": 28520
    },
    {
      "epoch": 1.7241796095969057,
      "grad_norm": 0.6795611526732371,
      "learning_rate": 4.181627095833955e-06,
      "loss": 0.9253,
      "step": 28530
    },
    {
      "epoch": 1.7247839487520396,
      "grad_norm": 0.6452554054803482,
      "learning_rate": 4.178341274934697e-06,
      "loss": 0.9385,
      "step": 28540
    },
    {
      "epoch": 1.7253882879071734,
      "grad_norm": 0.6954579019471295,
      "learning_rate": 4.175055818688988e-06,
      "loss": 0.9197,
      "step": 28550
    },
    {
      "epoch": 1.7259926270623074,
      "grad_norm": 0.7210926932118932,
      "learning_rate": 4.17177072855492e-06,
      "loss": 0.9176,
      "step": 28560
    },
    {
      "epoch": 1.7265969662174412,
      "grad_norm": 0.7147529530363084,
      "learning_rate": 4.168486005990419e-06,
      "loss": 0.9292,
      "step": 28570
    },
    {
      "epoch": 1.7272013053725752,
      "grad_norm": 0.678852288447761,
      "learning_rate": 4.1652016524532524e-06,
      "loss": 0.9083,
      "step": 28580
    },
    {
      "epoch": 1.727805644527709,
      "grad_norm": 0.6898873988239523,
      "learning_rate": 4.161917669401021e-06,
      "loss": 0.9383,
      "step": 28590
    },
    {
      "epoch": 1.7284099836828428,
      "grad_norm": 0.6146637217519962,
      "learning_rate": 4.158634058291164e-06,
      "loss": 0.8997,
      "step": 28600
    },
    {
      "epoch": 1.7290143228379766,
      "grad_norm": 0.6219568538250213,
      "learning_rate": 4.15535082058095e-06,
      "loss": 0.9241,
      "step": 28610
    },
    {
      "epoch": 1.7296186619931104,
      "grad_norm": 0.6190849352140058,
      "learning_rate": 4.152067957727491e-06,
      "loss": 0.9377,
      "step": 28620
    },
    {
      "epoch": 1.7302230011482442,
      "grad_norm": 0.624568985291949,
      "learning_rate": 4.1487854711877215e-06,
      "loss": 0.9087,
      "step": 28630
    },
    {
      "epoch": 1.7308273403033783,
      "grad_norm": 0.6285974544687487,
      "learning_rate": 4.145503362418416e-06,
      "loss": 0.9186,
      "step": 28640
    },
    {
      "epoch": 1.731431679458512,
      "grad_norm": 0.7786380732845095,
      "learning_rate": 4.142221632876184e-06,
      "loss": 0.9262,
      "step": 28650
    },
    {
      "epoch": 1.732036018613646,
      "grad_norm": 0.7658717757022581,
      "learning_rate": 4.138940284017458e-06,
      "loss": 0.9699,
      "step": 28660
    },
    {
      "epoch": 1.73264035776878,
      "grad_norm": 0.7982729082319859,
      "learning_rate": 4.135659317298508e-06,
      "loss": 0.92,
      "step": 28670
    },
    {
      "epoch": 1.7332446969239137,
      "grad_norm": 0.8501974444355619,
      "learning_rate": 4.132378734175432e-06,
      "loss": 0.9233,
      "step": 28680
    },
    {
      "epoch": 1.7338490360790475,
      "grad_norm": 0.7876602712029557,
      "learning_rate": 4.129098536104161e-06,
      "loss": 0.9178,
      "step": 28690
    },
    {
      "epoch": 1.7344533752341813,
      "grad_norm": 0.753480921958508,
      "learning_rate": 4.1258187245404495e-06,
      "loss": 0.9044,
      "step": 28700
    },
    {
      "epoch": 1.7350577143893153,
      "grad_norm": 0.7808076662796362,
      "learning_rate": 4.1225393009398845e-06,
      "loss": 0.9173,
      "step": 28710
    },
    {
      "epoch": 1.7356620535444491,
      "grad_norm": 0.7909649978679867,
      "learning_rate": 4.11926026675788e-06,
      "loss": 0.9312,
      "step": 28720
    },
    {
      "epoch": 1.7362663926995832,
      "grad_norm": 0.8319812970621725,
      "learning_rate": 4.115981623449677e-06,
      "loss": 0.9449,
      "step": 28730
    },
    {
      "epoch": 1.736870731854717,
      "grad_norm": 0.7961357862062574,
      "learning_rate": 4.112703372470342e-06,
      "loss": 0.9136,
      "step": 28740
    },
    {
      "epoch": 1.7374750710098508,
      "grad_norm": 1.1975285217244283,
      "learning_rate": 4.109425515274771e-06,
      "loss": 0.9304,
      "step": 28750
    },
    {
      "epoch": 1.7380794101649846,
      "grad_norm": 1.1847021062700358,
      "learning_rate": 4.106148053317679e-06,
      "loss": 0.9223,
      "step": 28760
    },
    {
      "epoch": 1.7386837493201184,
      "grad_norm": 1.219984396493548,
      "learning_rate": 4.102870988053611e-06,
      "loss": 0.9238,
      "step": 28770
    },
    {
      "epoch": 1.7392880884752522,
      "grad_norm": 1.20594493272711,
      "learning_rate": 4.099594320936936e-06,
      "loss": 0.9099,
      "step": 28780
    },
    {
      "epoch": 1.7398924276303862,
      "grad_norm": 1.160492299566175,
      "learning_rate": 4.096318053421841e-06,
      "loss": 0.9097,
      "step": 28790
    },
    {
      "epoch": 1.74049676678552,
      "grad_norm": 0.8955515791682743,
      "learning_rate": 4.093042186962341e-06,
      "loss": 0.9198,
      "step": 28800
    },
    {
      "epoch": 1.741101105940654,
      "grad_norm": 0.8412730627715108,
      "learning_rate": 4.0897667230122725e-06,
      "loss": 0.9378,
      "step": 28810
    },
    {
      "epoch": 1.7417054450957878,
      "grad_norm": 1.0874227687016704,
      "learning_rate": 4.0864916630252895e-06,
      "loss": 0.9376,
      "step": 28820
    },
    {
      "epoch": 1.7423097842509216,
      "grad_norm": 0.9910514511056169,
      "learning_rate": 4.08321700845487e-06,
      "loss": 0.9506,
      "step": 28830
    },
    {
      "epoch": 1.7429141234060554,
      "grad_norm": 0.9682200652630557,
      "learning_rate": 4.079942760754313e-06,
      "loss": 0.9301,
      "step": 28840
    },
    {
      "epoch": 1.7435184625611893,
      "grad_norm": 2.209609858110814,
      "learning_rate": 4.076668921376733e-06,
      "loss": 0.936,
      "step": 28850
    },
    {
      "epoch": 1.744122801716323,
      "grad_norm": 2.4471751842831977,
      "learning_rate": 4.073395491775065e-06,
      "loss": 0.9287,
      "step": 28860
    },
    {
      "epoch": 1.744727140871457,
      "grad_norm": 2.2251776326386654,
      "learning_rate": 4.070122473402067e-06,
      "loss": 0.9101,
      "step": 28870
    },
    {
      "epoch": 1.7453314800265909,
      "grad_norm": 2.572562552049361,
      "learning_rate": 4.066849867710304e-06,
      "loss": 0.927,
      "step": 28880
    },
    {
      "epoch": 1.745935819181725,
      "grad_norm": 2.427725408261804,
      "learning_rate": 4.063577676152168e-06,
      "loss": 0.9615,
      "step": 28890
    },
    {
      "epoch": 1.7465401583368587,
      "grad_norm": 0.8817407460860198,
      "learning_rate": 4.060305900179864e-06,
      "loss": 0.9511,
      "step": 28900
    },
    {
      "epoch": 1.7471444974919925,
      "grad_norm": 1.050339568385794,
      "learning_rate": 4.0570345412454084e-06,
      "loss": 0.9193,
      "step": 28910
    },
    {
      "epoch": 1.7477488366471263,
      "grad_norm": 0.9012741801894236,
      "learning_rate": 4.053763600800637e-06,
      "loss": 0.9326,
      "step": 28920
    },
    {
      "epoch": 1.7483531758022601,
      "grad_norm": 0.9800412900520905,
      "learning_rate": 4.050493080297202e-06,
      "loss": 0.898,
      "step": 28930
    },
    {
      "epoch": 1.748957514957394,
      "grad_norm": 0.8971395239471041,
      "learning_rate": 4.04722298118656e-06,
      "loss": 0.9188,
      "step": 28940
    },
    {
      "epoch": 1.749561854112528,
      "grad_norm": 0.973458149617079,
      "learning_rate": 4.043953304919992e-06,
      "loss": 0.9156,
      "step": 28950
    },
    {
      "epoch": 1.7501661932676618,
      "grad_norm": 1.052742223871775,
      "learning_rate": 4.040684052948585e-06,
      "loss": 0.9191,
      "step": 28960
    },
    {
      "epoch": 1.7507705324227958,
      "grad_norm": 0.9611456134382498,
      "learning_rate": 4.037415226723235e-06,
      "loss": 0.9251,
      "step": 28970
    },
    {
      "epoch": 1.7513748715779296,
      "grad_norm": 0.9966884991248691,
      "learning_rate": 4.034146827694657e-06,
      "loss": 0.9452,
      "step": 28980
    },
    {
      "epoch": 1.7519792107330634,
      "grad_norm": 0.8969999396104078,
      "learning_rate": 4.030878857313369e-06,
      "loss": 0.9136,
      "step": 28990
    },
    {
      "epoch": 1.7525835498881972,
      "grad_norm": 1.0350502990855937,
      "learning_rate": 4.027611317029705e-06,
      "loss": 0.9434,
      "step": 29000
    },
    {
      "epoch": 1.753187889043331,
      "grad_norm": 1.1586626710428998,
      "learning_rate": 4.024344208293802e-06,
      "loss": 0.9008,
      "step": 29010
    },
    {
      "epoch": 1.753792228198465,
      "grad_norm": 1.0205193790008382,
      "learning_rate": 4.0210775325556085e-06,
      "loss": 0.9299,
      "step": 29020
    },
    {
      "epoch": 1.7543965673535988,
      "grad_norm": 1.0719006845399093,
      "learning_rate": 4.017811291264884e-06,
      "loss": 0.9464,
      "step": 29030
    },
    {
      "epoch": 1.7550009065087329,
      "grad_norm": 1.0047855323878288,
      "learning_rate": 4.014545485871188e-06,
      "loss": 0.9241,
      "step": 29040
    },
    {
      "epoch": 1.7556052456638667,
      "grad_norm": 0.9992269022985253,
      "learning_rate": 4.011280117823892e-06,
      "loss": 0.9105,
      "step": 29050
    },
    {
      "epoch": 1.7562095848190005,
      "grad_norm": 1.0874729697241927,
      "learning_rate": 4.008015188572175e-06,
      "loss": 0.9412,
      "step": 29060
    },
    {
      "epoch": 1.7568139239741343,
      "grad_norm": 0.9958719412959735,
      "learning_rate": 4.004750699565013e-06,
      "loss": 0.9237,
      "step": 29070
    },
    {
      "epoch": 1.757418263129268,
      "grad_norm": 1.0005849729762109,
      "learning_rate": 4.001486652251194e-06,
      "loss": 0.9353,
      "step": 29080
    },
    {
      "epoch": 1.7580226022844019,
      "grad_norm": 1.061390336705989,
      "learning_rate": 3.99822304807931e-06,
      "loss": 0.9494,
      "step": 29090
    },
    {
      "epoch": 1.758626941439536,
      "grad_norm": 1.1619495194693477,
      "learning_rate": 3.9949598884977505e-06,
      "loss": 0.9428,
      "step": 29100
    },
    {
      "epoch": 1.7592312805946697,
      "grad_norm": 1.187098886512169,
      "learning_rate": 3.991697174954715e-06,
      "loss": 0.9059,
      "step": 29110
    },
    {
      "epoch": 1.7598356197498037,
      "grad_norm": 1.1184433101080973,
      "learning_rate": 3.988434908898198e-06,
      "loss": 0.9486,
      "step": 29120
    },
    {
      "epoch": 1.7604399589049375,
      "grad_norm": 1.078999904905132,
      "learning_rate": 3.985173091776e-06,
      "loss": 0.95,
      "step": 29130
    },
    {
      "epoch": 1.7610442980600713,
      "grad_norm": 1.0633109201823634,
      "learning_rate": 3.981911725035721e-06,
      "loss": 0.912,
      "step": 29140
    },
    {
      "epoch": 1.7616486372152051,
      "grad_norm": 1.04109343203107,
      "learning_rate": 3.978650810124763e-06,
      "loss": 0.9326,
      "step": 29150
    },
    {
      "epoch": 1.762252976370339,
      "grad_norm": 0.9767783925535196,
      "learning_rate": 3.975390348490321e-06,
      "loss": 0.9285,
      "step": 29160
    },
    {
      "epoch": 1.7628573155254728,
      "grad_norm": 0.9541021152711103,
      "learning_rate": 3.972130341579397e-06,
      "loss": 0.9375,
      "step": 29170
    },
    {
      "epoch": 1.7634616546806068,
      "grad_norm": 0.9962720333217153,
      "learning_rate": 3.968870790838789e-06,
      "loss": 0.9357,
      "step": 29180
    },
    {
      "epoch": 1.7640659938357406,
      "grad_norm": 0.9353997194295306,
      "learning_rate": 3.965611697715087e-06,
      "loss": 0.9261,
      "step": 29190
    },
    {
      "epoch": 1.7646703329908746,
      "grad_norm": 0.953982881743324,
      "learning_rate": 3.9623530636546845e-06,
      "loss": 0.9219,
      "step": 29200
    },
    {
      "epoch": 1.7652746721460084,
      "grad_norm": 1.057404587595876,
      "learning_rate": 3.9590948901037715e-06,
      "loss": 0.9209,
      "step": 29210
    },
    {
      "epoch": 1.7658790113011422,
      "grad_norm": 1.0200504242372836,
      "learning_rate": 3.955837178508325e-06,
      "loss": 0.9431,
      "step": 29220
    },
    {
      "epoch": 1.766483350456276,
      "grad_norm": 1.0265144195798135,
      "learning_rate": 3.952579930314128e-06,
      "loss": 0.9393,
      "step": 29230
    },
    {
      "epoch": 1.7670876896114098,
      "grad_norm": 1.0893425750456653,
      "learning_rate": 3.9493231469667495e-06,
      "loss": 0.9288,
      "step": 29240
    },
    {
      "epoch": 1.7676920287665436,
      "grad_norm": 0.9947770454973495,
      "learning_rate": 3.946066829911558e-06,
      "loss": 0.9394,
      "step": 29250
    },
    {
      "epoch": 1.7682963679216777,
      "grad_norm": 1.0256234731425078,
      "learning_rate": 3.942810980593711e-06,
      "loss": 0.9118,
      "step": 29260
    },
    {
      "epoch": 1.7689007070768115,
      "grad_norm": 0.9498250051806073,
      "learning_rate": 3.939555600458162e-06,
      "loss": 0.935,
      "step": 29270
    },
    {
      "epoch": 1.7695050462319455,
      "grad_norm": 0.9382613784732785,
      "learning_rate": 3.936300690949651e-06,
      "loss": 0.9164,
      "step": 29280
    },
    {
      "epoch": 1.7701093853870793,
      "grad_norm": 0.9036835464689572,
      "learning_rate": 3.933046253512714e-06,
      "loss": 0.9168,
      "step": 29290
    },
    {
      "epoch": 1.770713724542213,
      "grad_norm": 1.1260063614778104,
      "learning_rate": 3.929792289591678e-06,
      "loss": 0.9362,
      "step": 29300
    },
    {
      "epoch": 1.771318063697347,
      "grad_norm": 1.112082283564898,
      "learning_rate": 3.926538800630655e-06,
      "loss": 0.9274,
      "step": 29310
    },
    {
      "epoch": 1.7719224028524807,
      "grad_norm": 1.1487598047703247,
      "learning_rate": 3.9232857880735495e-06,
      "loss": 0.9207,
      "step": 29320
    },
    {
      "epoch": 1.7725267420076145,
      "grad_norm": 1.105047230704751,
      "learning_rate": 3.920033253364054e-06,
      "loss": 0.8914,
      "step": 29330
    },
    {
      "epoch": 1.7731310811627485,
      "grad_norm": 1.1355765610830528,
      "learning_rate": 3.916781197945652e-06,
      "loss": 0.9338,
      "step": 29340
    },
    {
      "epoch": 1.7737354203178826,
      "grad_norm": 1.0743764475386823,
      "learning_rate": 3.913529623261606e-06,
      "loss": 0.9306,
      "step": 29350
    },
    {
      "epoch": 1.7743397594730164,
      "grad_norm": 1.0345611859648796,
      "learning_rate": 3.910278530754974e-06,
      "loss": 0.9074,
      "step": 29360
    },
    {
      "epoch": 1.7749440986281502,
      "grad_norm": 1.073161976092642,
      "learning_rate": 3.907027921868596e-06,
      "loss": 0.9304,
      "step": 29370
    },
    {
      "epoch": 1.775548437783284,
      "grad_norm": 1.1161521036350304,
      "learning_rate": 3.903777798045097e-06,
      "loss": 0.9362,
      "step": 29380
    },
    {
      "epoch": 1.7761527769384178,
      "grad_norm": 1.099926306775401,
      "learning_rate": 3.900528160726887e-06,
      "loss": 0.9266,
      "step": 29390
    },
    {
      "epoch": 1.7767571160935516,
      "grad_norm": 1.1106376869829082,
      "learning_rate": 3.897279011356164e-06,
      "loss": 0.9056,
      "step": 29400
    },
    {
      "epoch": 1.7773614552486856,
      "grad_norm": 1.1998338745463926,
      "learning_rate": 3.8940303513749e-06,
      "loss": 0.9048,
      "step": 29410
    },
    {
      "epoch": 1.7779657944038194,
      "grad_norm": 1.1290712572608752,
      "learning_rate": 3.89078218222486e-06,
      "loss": 0.9064,
      "step": 29420
    },
    {
      "epoch": 1.7785701335589534,
      "grad_norm": 1.1499888068921444,
      "learning_rate": 3.887534505347589e-06,
      "loss": 0.8799,
      "step": 29430
    },
    {
      "epoch": 1.7791744727140872,
      "grad_norm": 1.155229460266719,
      "learning_rate": 3.8842873221844065e-06,
      "loss": 0.9328,
      "step": 29440
    },
    {
      "epoch": 1.779778811869221,
      "grad_norm": 1.4312733567142073,
      "learning_rate": 3.8810406341764204e-06,
      "loss": 0.9336,
      "step": 29450
    },
    {
      "epoch": 1.7803831510243548,
      "grad_norm": 1.4028306837952726,
      "learning_rate": 3.877794442764519e-06,
      "loss": 0.9495,
      "step": 29460
    },
    {
      "epoch": 1.7809874901794887,
      "grad_norm": 1.4269343928293714,
      "learning_rate": 3.874548749389362e-06,
      "loss": 0.922,
      "step": 29470
    },
    {
      "epoch": 1.7815918293346225,
      "grad_norm": 1.3734190732144356,
      "learning_rate": 3.871303555491398e-06,
      "loss": 0.9483,
      "step": 29480
    },
    {
      "epoch": 1.7821961684897565,
      "grad_norm": 1.3969604695593751,
      "learning_rate": 3.868058862510848e-06,
      "loss": 0.9114,
      "step": 29490
    },
    {
      "epoch": 1.7828005076448903,
      "grad_norm": 0.8289691556286879,
      "learning_rate": 3.864814671887713e-06,
      "loss": 0.9175,
      "step": 29500
    },
    {
      "epoch": 1.7834048468000243,
      "grad_norm": 0.8170960549761034,
      "learning_rate": 3.861570985061769e-06,
      "loss": 0.907,
      "step": 29510
    },
    {
      "epoch": 1.7840091859551581,
      "grad_norm": 0.8548432678197053,
      "learning_rate": 3.858327803472572e-06,
      "loss": 0.9264,
      "step": 29520
    },
    {
      "epoch": 1.784613525110292,
      "grad_norm": 0.8558348024869225,
      "learning_rate": 3.855085128559448e-06,
      "loss": 0.9356,
      "step": 29530
    },
    {
      "epoch": 1.7852178642654257,
      "grad_norm": 0.8739095987409106,
      "learning_rate": 3.851842961761505e-06,
      "loss": 0.9312,
      "step": 29540
    },
    {
      "epoch": 1.7858222034205595,
      "grad_norm": 0.745830819800408,
      "learning_rate": 3.848601304517622e-06,
      "loss": 0.9196,
      "step": 29550
    },
    {
      "epoch": 1.7864265425756933,
      "grad_norm": 0.7888270545204027,
      "learning_rate": 3.8453601582664505e-06,
      "loss": 0.9096,
      "step": 29560
    },
    {
      "epoch": 1.7870308817308274,
      "grad_norm": 0.8158106354151947,
      "learning_rate": 3.842119524446416e-06,
      "loss": 0.9411,
      "step": 29570
    },
    {
      "epoch": 1.7876352208859612,
      "grad_norm": 0.7702930550041105,
      "learning_rate": 3.838879404495721e-06,
      "loss": 0.9249,
      "step": 29580
    },
    {
      "epoch": 1.7882395600410952,
      "grad_norm": 0.7729868215873599,
      "learning_rate": 3.835639799852332e-06,
      "loss": 0.9203,
      "step": 29590
    },
    {
      "epoch": 1.788843899196229,
      "grad_norm": 0.7555366467832316,
      "learning_rate": 3.832400711953993e-06,
      "loss": 0.9332,
      "step": 29600
    },
    {
      "epoch": 1.7894482383513628,
      "grad_norm": 0.759661534444941,
      "learning_rate": 3.829162142238219e-06,
      "loss": 0.9235,
      "step": 29610
    },
    {
      "epoch": 1.7900525775064966,
      "grad_norm": 0.7744893226266466,
      "learning_rate": 3.825924092142289e-06,
      "loss": 0.9193,
      "step": 29620
    },
    {
      "epoch": 1.7906569166616304,
      "grad_norm": 0.8580460263875727,
      "learning_rate": 3.822686563103257e-06,
      "loss": 0.9258,
      "step": 29630
    },
    {
      "epoch": 1.7912612558167642,
      "grad_norm": 0.8390815248522523,
      "learning_rate": 3.819449556557944e-06,
      "loss": 0.9055,
      "step": 29640
    },
    {
      "epoch": 1.7918655949718982,
      "grad_norm": 0.5784632011880171,
      "learning_rate": 3.816213073942939e-06,
      "loss": 0.9033,
      "step": 29650
    },
    {
      "epoch": 1.792469934127032,
      "grad_norm": 0.5822405172874409,
      "learning_rate": 3.8129771166945973e-06,
      "loss": 0.918,
      "step": 29660
    },
    {
      "epoch": 1.793074273282166,
      "grad_norm": 0.5745599296947681,
      "learning_rate": 3.809741686249043e-06,
      "loss": 0.9184,
      "step": 29670
    },
    {
      "epoch": 1.7936786124372999,
      "grad_norm": 0.5814953579888945,
      "learning_rate": 3.8065067840421678e-06,
      "loss": 0.9241,
      "step": 29680
    },
    {
      "epoch": 1.7942829515924337,
      "grad_norm": 0.6131249058045749,
      "learning_rate": 3.803272411509623e-06,
      "loss": 0.9463,
      "step": 29690
    },
    {
      "epoch": 1.7948872907475675,
      "grad_norm": 0.6544800732532693,
      "learning_rate": 3.80003857008683e-06,
      "loss": 0.9398,
      "step": 29700
    },
    {
      "epoch": 1.7954916299027013,
      "grad_norm": 0.6235371676133413,
      "learning_rate": 3.7968052612089767e-06,
      "loss": 0.9374,
      "step": 29710
    },
    {
      "epoch": 1.7960959690578353,
      "grad_norm": 0.6424185197063359,
      "learning_rate": 3.7935724863110063e-06,
      "loss": 0.9493,
      "step": 29720
    },
    {
      "epoch": 1.796700308212969,
      "grad_norm": 0.6578890377391214,
      "learning_rate": 3.7903402468276313e-06,
      "loss": 0.9155,
      "step": 29730
    },
    {
      "epoch": 1.7973046473681031,
      "grad_norm": 0.636163900304483,
      "learning_rate": 3.7871085441933274e-06,
      "loss": 0.9263,
      "step": 29740
    },
    {
      "epoch": 1.797908986523237,
      "grad_norm": 0.8474354965583315,
      "learning_rate": 3.7838773798423274e-06,
      "loss": 0.9182,
      "step": 29750
    },
    {
      "epoch": 1.7985133256783707,
      "grad_norm": 0.8137597660358254,
      "learning_rate": 3.780646755208627e-06,
      "loss": 0.9044,
      "step": 29760
    },
    {
      "epoch": 1.7991176648335045,
      "grad_norm": 0.8104204356966251,
      "learning_rate": 3.7774166717259852e-06,
      "loss": 0.924,
      "step": 29770
    },
    {
      "epoch": 1.7997220039886384,
      "grad_norm": 0.8995237577266365,
      "learning_rate": 3.774187130827916e-06,
      "loss": 0.9126,
      "step": 29780
    },
    {
      "epoch": 1.8003263431437722,
      "grad_norm": 0.8485054889944411,
      "learning_rate": 3.7709581339476965e-06,
      "loss": 0.9097,
      "step": 29790
    },
    {
      "epoch": 1.8009306822989062,
      "grad_norm": 0.5582675352448696,
      "learning_rate": 3.767729682518363e-06,
      "loss": 0.9325,
      "step": 29800
    },
    {
      "epoch": 1.80153502145404,
      "grad_norm": 0.5299149026002937,
      "learning_rate": 3.7645017779727044e-06,
      "loss": 0.936,
      "step": 29810
    },
    {
      "epoch": 1.802139360609174,
      "grad_norm": 0.549243609096665,
      "learning_rate": 3.761274421743271e-06,
      "loss": 0.9255,
      "step": 29820
    },
    {
      "epoch": 1.8027436997643078,
      "grad_norm": 0.5734958902150582,
      "learning_rate": 3.7580476152623726e-06,
      "loss": 0.9436,
      "step": 29830
    },
    {
      "epoch": 1.8033480389194416,
      "grad_norm": 0.5129167765861892,
      "learning_rate": 3.754821359962066e-06,
      "loss": 0.9043,
      "step": 29840
    },
    {
      "epoch": 1.8039523780745754,
      "grad_norm": 0.5439238123126162,
      "learning_rate": 3.7515956572741715e-06,
      "loss": 0.9355,
      "step": 29850
    },
    {
      "epoch": 1.8045567172297092,
      "grad_norm": 0.5425997075853455,
      "learning_rate": 3.7483705086302634e-06,
      "loss": 0.9224,
      "step": 29860
    },
    {
      "epoch": 1.805161056384843,
      "grad_norm": 0.5666056204932807,
      "learning_rate": 3.7451459154616645e-06,
      "loss": 0.9265,
      "step": 29870
    },
    {
      "epoch": 1.805765395539977,
      "grad_norm": 0.5411669870680204,
      "learning_rate": 3.741921879199457e-06,
      "loss": 0.919,
      "step": 29880
    },
    {
      "epoch": 1.8063697346951109,
      "grad_norm": 0.5466863759604617,
      "learning_rate": 3.7386984012744733e-06,
      "loss": 0.9302,
      "step": 29890
    },
    {
      "epoch": 1.8069740738502449,
      "grad_norm": 0.5334813296440465,
      "learning_rate": 3.7354754831172986e-06,
      "loss": 0.9364,
      "step": 29900
    },
    {
      "epoch": 1.8075784130053787,
      "grad_norm": 0.5420136916352104,
      "learning_rate": 3.732253126158268e-06,
      "loss": 0.9093,
      "step": 29910
    },
    {
      "epoch": 1.8081827521605125,
      "grad_norm": 0.5218457704359801,
      "learning_rate": 3.7290313318274716e-06,
      "loss": 0.9287,
      "step": 29920
    },
    {
      "epoch": 1.8087870913156463,
      "grad_norm": 0.5281767381873376,
      "learning_rate": 3.725810101554745e-06,
      "loss": 0.9239,
      "step": 29930
    },
    {
      "epoch": 1.80939143047078,
      "grad_norm": 0.5066693381495344,
      "learning_rate": 3.722589436769676e-06,
      "loss": 0.9131,
      "step": 29940
    },
    {
      "epoch": 1.809995769625914,
      "grad_norm": 0.5529085145395609,
      "learning_rate": 3.7193693389016038e-06,
      "loss": 0.9193,
      "step": 29950
    },
    {
      "epoch": 1.810600108781048,
      "grad_norm": 0.5456503050449465,
      "learning_rate": 3.716149809379609e-06,
      "loss": 0.9121,
      "step": 29960
    },
    {
      "epoch": 1.8112044479361817,
      "grad_norm": 0.5200752140586525,
      "learning_rate": 3.7129308496325266e-06,
      "loss": 0.921,
      "step": 29970
    },
    {
      "epoch": 1.8118087870913158,
      "grad_norm": 0.5672171447850323,
      "learning_rate": 3.709712461088939e-06,
      "loss": 0.941,
      "step": 29980
    },
    {
      "epoch": 1.8124131262464496,
      "grad_norm": 0.5418840252388737,
      "learning_rate": 3.706494645177168e-06,
      "loss": 0.9063,
      "step": 29990
    },
    {
      "epoch": 1.8130174654015834,
      "grad_norm": 0.5669367221438031,
      "learning_rate": 3.703277403325287e-06,
      "loss": 0.9235,
      "step": 30000
    },
    {
      "epoch": 1.8136218045567172,
      "grad_norm": 0.5397365739890289,
      "learning_rate": 3.7000607369611154e-06,
      "loss": 0.918,
      "step": 30010
    },
    {
      "epoch": 1.814226143711851,
      "grad_norm": 0.5471272823103205,
      "learning_rate": 3.6968446475122146e-06,
      "loss": 0.9394,
      "step": 30020
    },
    {
      "epoch": 1.814830482866985,
      "grad_norm": 0.5022206240626622,
      "learning_rate": 3.6936291364058896e-06,
      "loss": 0.9018,
      "step": 30030
    },
    {
      "epoch": 1.8154348220221188,
      "grad_norm": 0.5689934571594994,
      "learning_rate": 3.6904142050691895e-06,
      "loss": 0.9272,
      "step": 30040
    },
    {
      "epoch": 1.8160391611772528,
      "grad_norm": 0.5836049423164198,
      "learning_rate": 3.687199854928909e-06,
      "loss": 0.9528,
      "step": 30050
    },
    {
      "epoch": 1.8166435003323866,
      "grad_norm": 0.5345911301300967,
      "learning_rate": 3.6839860874115774e-06,
      "loss": 0.9256,
      "step": 30060
    },
    {
      "epoch": 1.8172478394875204,
      "grad_norm": 0.5528500728852526,
      "learning_rate": 3.680772903943474e-06,
      "loss": 0.8837,
      "step": 30070
    },
    {
      "epoch": 1.8178521786426542,
      "grad_norm": 0.5582100586078845,
      "learning_rate": 3.677560305950616e-06,
      "loss": 0.9386,
      "step": 30080
    },
    {
      "epoch": 1.818456517797788,
      "grad_norm": 0.5750340346612108,
      "learning_rate": 3.674348294858756e-06,
      "loss": 0.9268,
      "step": 30090
    },
    {
      "epoch": 1.8190608569529219,
      "grad_norm": 0.5711408078346777,
      "learning_rate": 3.671136872093392e-06,
      "loss": 0.9222,
      "step": 30100
    },
    {
      "epoch": 1.8196651961080559,
      "grad_norm": 0.6112269646455496,
      "learning_rate": 3.667926039079761e-06,
      "loss": 0.9002,
      "step": 30110
    },
    {
      "epoch": 1.8202695352631897,
      "grad_norm": 0.6140953401552849,
      "learning_rate": 3.6647157972428328e-06,
      "loss": 0.9179,
      "step": 30120
    },
    {
      "epoch": 1.8208738744183237,
      "grad_norm": 0.6661035494700247,
      "learning_rate": 3.66150614800732e-06,
      "loss": 0.9118,
      "step": 30130
    },
    {
      "epoch": 1.8214782135734575,
      "grad_norm": 0.637954177846652,
      "learning_rate": 3.6582970927976702e-06,
      "loss": 0.9516,
      "step": 30140
    },
    {
      "epoch": 1.8220825527285913,
      "grad_norm": 0.7360743073942373,
      "learning_rate": 3.655088633038067e-06,
      "loss": 0.9471,
      "step": 30150
    },
    {
      "epoch": 1.8226868918837251,
      "grad_norm": 0.6994191092481861,
      "learning_rate": 3.65188077015243e-06,
      "loss": 0.9444,
      "step": 30160
    },
    {
      "epoch": 1.823291231038859,
      "grad_norm": 0.6573074414989548,
      "learning_rate": 3.6486735055644173e-06,
      "loss": 0.9032,
      "step": 30170
    },
    {
      "epoch": 1.8238955701939927,
      "grad_norm": 0.6775804757434722,
      "learning_rate": 3.6454668406974135e-06,
      "loss": 0.9447,
      "step": 30180
    },
    {
      "epoch": 1.8244999093491268,
      "grad_norm": 0.708305126261626,
      "learning_rate": 3.6422607769745456e-06,
      "loss": 0.9432,
      "step": 30190
    },
    {
      "epoch": 1.8251042485042606,
      "grad_norm": 0.6076070351746073,
      "learning_rate": 3.6390553158186703e-06,
      "loss": 0.9376,
      "step": 30200
    },
    {
      "epoch": 1.8257085876593946,
      "grad_norm": 0.637275892662838,
      "learning_rate": 3.6358504586523736e-06,
      "loss": 0.9233,
      "step": 30210
    },
    {
      "epoch": 1.8263129268145284,
      "grad_norm": 0.6130783908031918,
      "learning_rate": 3.6326462068979797e-06,
      "loss": 0.8998,
      "step": 30220
    },
    {
      "epoch": 1.8269172659696622,
      "grad_norm": 0.665250724142056,
      "learning_rate": 3.629442561977541e-06,
      "loss": 0.9545,
      "step": 30230
    },
    {
      "epoch": 1.827521605124796,
      "grad_norm": 0.654058771807477,
      "learning_rate": 3.626239525312837e-06,
      "loss": 0.9266,
      "step": 30240
    },
    {
      "epoch": 1.8281259442799298,
      "grad_norm": 0.7780445923338133,
      "learning_rate": 3.623037098325384e-06,
      "loss": 0.9113,
      "step": 30250
    },
    {
      "epoch": 1.8287302834350636,
      "grad_norm": 0.8166429109439057,
      "learning_rate": 3.6198352824364256e-06,
      "loss": 0.9344,
      "step": 30260
    },
    {
      "epoch": 1.8293346225901976,
      "grad_norm": 0.7998785126830879,
      "learning_rate": 3.6166340790669303e-06,
      "loss": 0.9297,
      "step": 30270
    },
    {
      "epoch": 1.8299389617453314,
      "grad_norm": 0.7842408660186287,
      "learning_rate": 3.613433489637599e-06,
      "loss": 0.914,
      "step": 30280
    },
    {
      "epoch": 1.8305433009004655,
      "grad_norm": 0.7974565081699884,
      "learning_rate": 3.6102335155688596e-06,
      "loss": 0.9102,
      "step": 30290
    },
    {
      "epoch": 1.8311476400555993,
      "grad_norm": 0.7869548592747461,
      "learning_rate": 3.6070341582808642e-06,
      "loss": 0.9346,
      "step": 30300
    },
    {
      "epoch": 1.831751979210733,
      "grad_norm": 0.8505850628567252,
      "learning_rate": 3.603835419193493e-06,
      "loss": 0.9115,
      "step": 30310
    },
    {
      "epoch": 1.8323563183658669,
      "grad_norm": 0.8235690567516089,
      "learning_rate": 3.600637299726355e-06,
      "loss": 0.947,
      "step": 30320
    },
    {
      "epoch": 1.8329606575210007,
      "grad_norm": 0.835277289975126,
      "learning_rate": 3.5974398012987772e-06,
      "loss": 0.9301,
      "step": 30330
    },
    {
      "epoch": 1.8335649966761345,
      "grad_norm": 0.8111892465325004,
      "learning_rate": 3.594242925329816e-06,
      "loss": 0.9283,
      "step": 30340
    },
    {
      "epoch": 1.8341693358312685,
      "grad_norm": 1.1618986078975626,
      "learning_rate": 3.591046673238251e-06,
      "loss": 0.9071,
      "step": 30350
    },
    {
      "epoch": 1.8347736749864023,
      "grad_norm": 1.1745416774634267,
      "learning_rate": 3.587851046442587e-06,
      "loss": 0.91,
      "step": 30360
    },
    {
      "epoch": 1.8353780141415363,
      "grad_norm": 1.1765824367765059,
      "learning_rate": 3.5846560463610437e-06,
      "loss": 0.9182,
      "step": 30370
    },
    {
      "epoch": 1.8359823532966701,
      "grad_norm": 1.2134141546475001,
      "learning_rate": 3.5814616744115704e-06,
      "loss": 0.9351,
      "step": 30380
    },
    {
      "epoch": 1.836586692451804,
      "grad_norm": 1.2168181895369916,
      "learning_rate": 3.578267932011835e-06,
      "loss": 0.9345,
      "step": 30390
    },
    {
      "epoch": 1.8371910316069378,
      "grad_norm": 0.9185490107846815,
      "learning_rate": 3.5750748205792253e-06,
      "loss": 0.9358,
      "step": 30400
    },
    {
      "epoch": 1.8377953707620716,
      "grad_norm": 1.0675845787646152,
      "learning_rate": 3.571882341530848e-06,
      "loss": 0.9402,
      "step": 30410
    },
    {
      "epoch": 1.8383997099172056,
      "grad_norm": 0.8571488978672929,
      "learning_rate": 3.5686904962835346e-06,
      "loss": 0.9211,
      "step": 30420
    },
    {
      "epoch": 1.8390040490723394,
      "grad_norm": 0.9220315575467384,
      "learning_rate": 3.5654992862538276e-06,
      "loss": 0.9232,
      "step": 30430
    },
    {
      "epoch": 1.8396083882274734,
      "grad_norm": 0.8264268468980654,
      "learning_rate": 3.562308712857992e-06,
      "loss": 0.9162,
      "step": 30440
    },
    {
      "epoch": 1.8402127273826072,
      "grad_norm": 2.2574627287404274,
      "learning_rate": 3.5591187775120134e-06,
      "loss": 0.9166,
      "step": 30450
    },
    {
      "epoch": 1.840817066537741,
      "grad_norm": 2.2135004127350095,
      "learning_rate": 3.555929481631586e-06,
      "loss": 0.9076,
      "step": 30460
    },
    {
      "epoch": 1.8414214056928748,
      "grad_norm": 2.677380574768971,
      "learning_rate": 3.5527408266321262e-06,
      "loss": 0.9314,
      "step": 30470
    },
    {
      "epoch": 1.8420257448480086,
      "grad_norm": 2.447771349329891,
      "learning_rate": 3.549552813928767e-06,
      "loss": 0.9296,
      "step": 30480
    },
    {
      "epoch": 1.8426300840031424,
      "grad_norm": 2.651620784471785,
      "learning_rate": 3.5463654449363505e-06,
      "loss": 0.9351,
      "step": 30490
    },
    {
      "epoch": 1.8432344231582765,
      "grad_norm": 1.0185018940566928,
      "learning_rate": 3.543178721069438e-06,
      "loss": 0.9163,
      "step": 30500
    },
    {
      "epoch": 1.8438387623134103,
      "grad_norm": 0.9436772332216543,
      "learning_rate": 3.539992643742305e-06,
      "loss": 0.9188,
      "step": 30510
    },
    {
      "epoch": 1.8444431014685443,
      "grad_norm": 1.0513732532776594,
      "learning_rate": 3.5368072143689335e-06,
      "loss": 0.9181,
      "step": 30520
    },
    {
      "epoch": 1.845047440623678,
      "grad_norm": 0.9461286171780051,
      "learning_rate": 3.5336224343630267e-06,
      "loss": 0.9295,
      "step": 30530
    },
    {
      "epoch": 1.845651779778812,
      "grad_norm": 0.88231914964534,
      "learning_rate": 3.530438305137993e-06,
      "loss": 0.9117,
      "step": 30540
    },
    {
      "epoch": 1.8462561189339457,
      "grad_norm": 0.9438100821375899,
      "learning_rate": 3.5272548281069563e-06,
      "loss": 0.9268,
      "step": 30550
    },
    {
      "epoch": 1.8468604580890795,
      "grad_norm": 0.9553533991613056,
      "learning_rate": 3.5240720046827458e-06,
      "loss": 0.9232,
      "step": 30560
    },
    {
      "epoch": 1.8474647972442133,
      "grad_norm": 0.9836899162349655,
      "learning_rate": 3.5208898362779086e-06,
      "loss": 0.9366,
      "step": 30570
    },
    {
      "epoch": 1.8480691363993473,
      "grad_norm": 0.9719354566636513,
      "learning_rate": 3.517708324304693e-06,
      "loss": 0.927,
      "step": 30580
    },
    {
      "epoch": 1.8486734755544811,
      "grad_norm": 0.965896212983014,
      "learning_rate": 3.51452747017506e-06,
      "loss": 0.915,
      "step": 30590
    },
    {
      "epoch": 1.8492778147096152,
      "grad_norm": 1.2662065726308087,
      "learning_rate": 3.51134727530068e-06,
      "loss": 0.9268,
      "step": 30600
    },
    {
      "epoch": 1.849882153864749,
      "grad_norm": 1.060891510417651,
      "learning_rate": 3.508167741092926e-06,
      "loss": 0.9491,
      "step": 30610
    },
    {
      "epoch": 1.8504864930198828,
      "grad_norm": 0.9586644529022527,
      "learning_rate": 3.5049888689628827e-06,
      "loss": 0.9221,
      "step": 30620
    },
    {
      "epoch": 1.8510908321750166,
      "grad_norm": 1.0214462184512128,
      "learning_rate": 3.5018106603213404e-06,
      "loss": 0.942,
      "step": 30630
    },
    {
      "epoch": 1.8516951713301504,
      "grad_norm": 1.077764225001201,
      "learning_rate": 3.4986331165787896e-06,
      "loss": 0.906,
      "step": 30640
    },
    {
      "epoch": 1.8522995104852842,
      "grad_norm": 1.0375479696404344,
      "learning_rate": 3.4954562391454317e-06,
      "loss": 0.9185,
      "step": 30650
    },
    {
      "epoch": 1.8529038496404182,
      "grad_norm": 1.0183334429773074,
      "learning_rate": 3.4922800294311727e-06,
      "loss": 0.9199,
      "step": 30660
    },
    {
      "epoch": 1.853508188795552,
      "grad_norm": 1.0247050118179013,
      "learning_rate": 3.4891044888456154e-06,
      "loss": 0.8909,
      "step": 30670
    },
    {
      "epoch": 1.854112527950686,
      "grad_norm": 1.01157509584948,
      "learning_rate": 3.485929618798074e-06,
      "loss": 0.9212,
      "step": 30680
    },
    {
      "epoch": 1.8547168671058198,
      "grad_norm": 1.0678949102994058,
      "learning_rate": 3.482755420697558e-06,
      "loss": 0.8864,
      "step": 30690
    },
    {
      "epoch": 1.8553212062609536,
      "grad_norm": 1.0933914107847948,
      "learning_rate": 3.479581895952786e-06,
      "loss": 0.9284,
      "step": 30700
    },
    {
      "epoch": 1.8559255454160875,
      "grad_norm": 1.113446707454662,
      "learning_rate": 3.476409045972169e-06,
      "loss": 0.9523,
      "step": 30710
    },
    {
      "epoch": 1.8565298845712213,
      "grad_norm": 1.1045607837128297,
      "learning_rate": 3.473236872163826e-06,
      "loss": 0.9193,
      "step": 30720
    },
    {
      "epoch": 1.8571342237263553,
      "grad_norm": 1.527476823151143,
      "learning_rate": 3.4700653759355745e-06,
      "loss": 0.9364,
      "step": 30730
    },
    {
      "epoch": 1.857738562881489,
      "grad_norm": 1.1566134019644148,
      "learning_rate": 3.4668945586949263e-06,
      "loss": 0.93,
      "step": 30740
    },
    {
      "epoch": 1.8583429020366231,
      "grad_norm": 0.9978520510823714,
      "learning_rate": 3.463724421849097e-06,
      "loss": 0.9384,
      "step": 30750
    },
    {
      "epoch": 1.858947241191757,
      "grad_norm": 0.9322939975599035,
      "learning_rate": 3.460554966805002e-06,
      "loss": 0.9166,
      "step": 30760
    },
    {
      "epoch": 1.8595515803468907,
      "grad_norm": 0.9883399437836303,
      "learning_rate": 3.4573861949692456e-06,
      "loss": 0.9175,
      "step": 30770
    },
    {
      "epoch": 1.8601559195020245,
      "grad_norm": 0.9417747363846148,
      "learning_rate": 3.4542181077481372e-06,
      "loss": 0.9653,
      "step": 30780
    },
    {
      "epoch": 1.8607602586571583,
      "grad_norm": 0.9538677075385023,
      "learning_rate": 3.4510507065476796e-06,
      "loss": 0.934,
      "step": 30790
    },
    {
      "epoch": 1.8613645978122921,
      "grad_norm": 1.019560210074955,
      "learning_rate": 3.44788399277357e-06,
      "loss": 0.9129,
      "step": 30800
    },
    {
      "epoch": 1.8619689369674262,
      "grad_norm": 0.9560064580599331,
      "learning_rate": 3.4447179678311997e-06,
      "loss": 0.9131,
      "step": 30810
    },
    {
      "epoch": 1.86257327612256,
      "grad_norm": 1.0449354786666403,
      "learning_rate": 3.4415526331256605e-06,
      "loss": 0.9166,
      "step": 30820
    },
    {
      "epoch": 1.863177615277694,
      "grad_norm": 1.024719714605216,
      "learning_rate": 3.4383879900617278e-06,
      "loss": 0.9068,
      "step": 30830
    },
    {
      "epoch": 1.8637819544328278,
      "grad_norm": 1.0446297480005429,
      "learning_rate": 3.4352240400438785e-06,
      "loss": 0.9253,
      "step": 30840
    },
    {
      "epoch": 1.8643862935879616,
      "grad_norm": 0.9550403219574327,
      "learning_rate": 3.4320607844762817e-06,
      "loss": 0.9093,
      "step": 30850
    },
    {
      "epoch": 1.8649906327430954,
      "grad_norm": 1.00531406267595,
      "learning_rate": 3.42889822476279e-06,
      "loss": 0.9174,
      "step": 30860
    },
    {
      "epoch": 1.8655949718982292,
      "grad_norm": 0.9404242416485341,
      "learning_rate": 3.4257363623069563e-06,
      "loss": 0.956,
      "step": 30870
    },
    {
      "epoch": 1.866199311053363,
      "grad_norm": 0.9447630325252553,
      "learning_rate": 3.4225751985120213e-06,
      "loss": 0.9141,
      "step": 30880
    },
    {
      "epoch": 1.866803650208497,
      "grad_norm": 0.999555036542683,
      "learning_rate": 3.419414734780912e-06,
      "loss": 0.9312,
      "step": 30890
    },
    {
      "epoch": 1.8674079893636308,
      "grad_norm": 1.1077082767806796,
      "learning_rate": 3.416254972516249e-06,
      "loss": 0.9047,
      "step": 30900
    },
    {
      "epoch": 1.8680123285187649,
      "grad_norm": 1.0959927693105036,
      "learning_rate": 3.413095913120342e-06,
      "loss": 0.9432,
      "step": 30910
    },
    {
      "epoch": 1.8686166676738987,
      "grad_norm": 1.1033794414580878,
      "learning_rate": 3.4099375579951844e-06,
      "loss": 0.9132,
      "step": 30920
    },
    {
      "epoch": 1.8692210068290325,
      "grad_norm": 1.1074681876020167,
      "learning_rate": 3.4067799085424608e-06,
      "loss": 0.9266,
      "step": 30930
    },
    {
      "epoch": 1.8698253459841663,
      "grad_norm": 1.1408154445524386,
      "learning_rate": 3.4036229661635413e-06,
      "loss": 0.9297,
      "step": 30940
    },
    {
      "epoch": 1.8704296851393,
      "grad_norm": 1.0881833617425594,
      "learning_rate": 3.400466732259482e-06,
      "loss": 0.8956,
      "step": 30950
    },
    {
      "epoch": 1.8710340242944339,
      "grad_norm": 1.0945169020029208,
      "learning_rate": 3.3973112082310245e-06,
      "loss": 0.9257,
      "step": 30960
    },
    {
      "epoch": 1.871638363449568,
      "grad_norm": 1.0416844339914666,
      "learning_rate": 3.394156395478598e-06,
      "loss": 0.9064,
      "step": 30970
    },
    {
      "epoch": 1.8722427026047017,
      "grad_norm": 1.060926738099387,
      "learning_rate": 3.3910022954023103e-06,
      "loss": 0.9211,
      "step": 30980
    },
    {
      "epoch": 1.8728470417598357,
      "grad_norm": 1.1132565375562784,
      "learning_rate": 3.3878489094019585e-06,
      "loss": 0.9295,
      "step": 30990
    },
    {
      "epoch": 1.8734513809149695,
      "grad_norm": 1.2230042833869643,
      "learning_rate": 3.3846962388770222e-06,
      "loss": 0.912,
      "step": 31000
    },
    {
      "epoch": 1.8740557200701033,
      "grad_norm": 1.1457856142182463,
      "learning_rate": 3.381544285226658e-06,
      "loss": 0.9169,
      "step": 31010
    },
    {
      "epoch": 1.8746600592252372,
      "grad_norm": 1.11155203698507,
      "learning_rate": 3.3783930498497106e-06,
      "loss": 0.9132,
      "step": 31020
    },
    {
      "epoch": 1.875264398380371,
      "grad_norm": 1.1447256528594774,
      "learning_rate": 3.375242534144706e-06,
      "loss": 0.9093,
      "step": 31030
    },
    {
      "epoch": 1.8758687375355048,
      "grad_norm": 1.1454029496747797,
      "learning_rate": 3.3720927395098447e-06,
      "loss": 0.9209,
      "step": 31040
    },
    {
      "epoch": 1.8764730766906388,
      "grad_norm": 1.4096350282334087,
      "learning_rate": 3.3689436673430134e-06,
      "loss": 0.9353,
      "step": 31050
    },
    {
      "epoch": 1.8770774158457728,
      "grad_norm": 1.3979284987475882,
      "learning_rate": 3.3657953190417724e-06,
      "loss": 0.9267,
      "step": 31060
    },
    {
      "epoch": 1.8776817550009066,
      "grad_norm": 1.4316298495958675,
      "learning_rate": 3.3626476960033695e-06,
      "loss": 0.9514,
      "step": 31070
    },
    {
      "epoch": 1.8782860941560404,
      "grad_norm": 1.3647796504721923,
      "learning_rate": 3.3595007996247196e-06,
      "loss": 0.917,
      "step": 31080
    },
    {
      "epoch": 1.8788904333111742,
      "grad_norm": 1.3748002302568774,
      "learning_rate": 3.356354631302423e-06,
      "loss": 0.9336,
      "step": 31090
    },
    {
      "epoch": 1.879494772466308,
      "grad_norm": 0.9362148447253937,
      "learning_rate": 3.353209192432757e-06,
      "loss": 0.9304,
      "step": 31100
    },
    {
      "epoch": 1.8800991116214418,
      "grad_norm": 0.8104977880293032,
      "learning_rate": 3.350064484411668e-06,
      "loss": 0.935,
      "step": 31110
    },
    {
      "epoch": 1.8807034507765759,
      "grad_norm": 0.8236573016929736,
      "learning_rate": 3.3469205086347855e-06,
      "loss": 0.9156,
      "step": 31120
    },
    {
      "epoch": 1.8813077899317097,
      "grad_norm": 0.9094453948721859,
      "learning_rate": 3.3437772664974133e-06,
      "loss": 0.9147,
      "step": 31130
    },
    {
      "epoch": 1.8819121290868437,
      "grad_norm": 0.8413589420763598,
      "learning_rate": 3.3406347593945236e-06,
      "loss": 0.9009,
      "step": 31140
    },
    {
      "epoch": 1.8825164682419775,
      "grad_norm": 0.8132417851350491,
      "learning_rate": 3.3374929887207685e-06,
      "loss": 0.9128,
      "step": 31150
    },
    {
      "epoch": 1.8831208073971113,
      "grad_norm": 0.78740111180985,
      "learning_rate": 3.334351955870472e-06,
      "loss": 0.9158,
      "step": 31160
    },
    {
      "epoch": 1.883725146552245,
      "grad_norm": 0.760767733312395,
      "learning_rate": 3.331211662237628e-06,
      "loss": 0.9234,
      "step": 31170
    },
    {
      "epoch": 1.884329485707379,
      "grad_norm": 0.726505885919191,
      "learning_rate": 3.328072109215905e-06,
      "loss": 0.9045,
      "step": 31180
    },
    {
      "epoch": 1.8849338248625127,
      "grad_norm": 0.780864972848366,
      "learning_rate": 3.3249332981986428e-06,
      "loss": 0.9231,
      "step": 31190
    },
    {
      "epoch": 1.8855381640176467,
      "grad_norm": 0.8307752643958988,
      "learning_rate": 3.3217952305788504e-06,
      "loss": 0.9031,
      "step": 31200
    },
    {
      "epoch": 1.8861425031727805,
      "grad_norm": 0.7988999242594705,
      "learning_rate": 3.3186579077492065e-06,
      "loss": 0.9288,
      "step": 31210
    },
    {
      "epoch": 1.8867468423279146,
      "grad_norm": 0.8253620466099063,
      "learning_rate": 3.315521331102063e-06,
      "loss": 0.9258,
      "step": 31220
    },
    {
      "epoch": 1.8873511814830484,
      "grad_norm": 0.8132882462442603,
      "learning_rate": 3.3123855020294344e-06,
      "loss": 0.9201,
      "step": 31230
    },
    {
      "epoch": 1.8879555206381822,
      "grad_norm": 0.815256218425578,
      "learning_rate": 3.309250421923009e-06,
      "loss": 0.9367,
      "step": 31240
    },
    {
      "epoch": 1.888559859793316,
      "grad_norm": 0.5957479022040869,
      "learning_rate": 3.306116092174143e-06,
      "loss": 0.9337,
      "step": 31250
    },
    {
      "epoch": 1.8891641989484498,
      "grad_norm": 0.5555668046882692,
      "learning_rate": 3.3029825141738525e-06,
      "loss": 0.9293,
      "step": 31260
    },
    {
      "epoch": 1.8897685381035836,
      "grad_norm": 0.6161523499411489,
      "learning_rate": 3.2998496893128274e-06,
      "loss": 0.9179,
      "step": 31270
    },
    {
      "epoch": 1.8903728772587176,
      "grad_norm": 0.551205275812608,
      "learning_rate": 3.296717618981423e-06,
      "loss": 0.9509,
      "step": 31280
    },
    {
      "epoch": 1.8909772164138514,
      "grad_norm": 0.5853362971975156,
      "learning_rate": 3.2935863045696525e-06,
      "loss": 0.931,
      "step": 31290
    },
    {
      "epoch": 1.8915815555689854,
      "grad_norm": 0.6315686524480645,
      "learning_rate": 3.290455747467203e-06,
      "loss": 0.935,
      "step": 31300
    },
    {
      "epoch": 1.8921858947241192,
      "grad_norm": 0.6538485953542557,
      "learning_rate": 3.287325949063419e-06,
      "loss": 0.9343,
      "step": 31310
    },
    {
      "epoch": 1.892790233879253,
      "grad_norm": 0.6710681251782127,
      "learning_rate": 3.2841969107473103e-06,
      "loss": 0.9383,
      "step": 31320
    },
    {
      "epoch": 1.8933945730343869,
      "grad_norm": 0.5931635692804301,
      "learning_rate": 3.281068633907549e-06,
      "loss": 0.9117,
      "step": 31330
    },
    {
      "epoch": 1.8939989121895207,
      "grad_norm": 0.6270552248394079,
      "learning_rate": 3.277941119932473e-06,
      "loss": 0.8916,
      "step": 31340
    },
    {
      "epoch": 1.8946032513446545,
      "grad_norm": 0.7927137373887444,
      "learning_rate": 3.2748143702100733e-06,
      "loss": 0.9119,
      "step": 31350
    },
    {
      "epoch": 1.8952075904997885,
      "grad_norm": 0.8757437613610626,
      "learning_rate": 3.2716883861280095e-06,
      "loss": 0.9516,
      "step": 31360
    },
    {
      "epoch": 1.8958119296549223,
      "grad_norm": 0.8059562565636221,
      "learning_rate": 3.2685631690735987e-06,
      "loss": 0.9228,
      "step": 31370
    },
    {
      "epoch": 1.8964162688100563,
      "grad_norm": 0.876840530022881,
      "learning_rate": 3.2654387204338193e-06,
      "loss": 0.9386,
      "step": 31380
    },
    {
      "epoch": 1.8970206079651901,
      "grad_norm": 0.9382330964159984,
      "learning_rate": 3.262315041595303e-06,
      "loss": 0.92,
      "step": 31390
    },
    {
      "epoch": 1.897624947120324,
      "grad_norm": 0.5300094989498622,
      "learning_rate": 3.2591921339443467e-06,
      "loss": 0.9173,
      "step": 31400
    },
    {
      "epoch": 1.8982292862754577,
      "grad_norm": 0.5590988149399575,
      "learning_rate": 3.256069998866903e-06,
      "loss": 0.9247,
      "step": 31410
    },
    {
      "epoch": 1.8988336254305915,
      "grad_norm": 0.5378023720053102,
      "learning_rate": 3.2529486377485767e-06,
      "loss": 0.9448,
      "step": 31420
    },
    {
      "epoch": 1.8994379645857256,
      "grad_norm": 0.5165235671647236,
      "learning_rate": 3.249828051974637e-06,
      "loss": 0.9086,
      "step": 31430
    },
    {
      "epoch": 1.9000423037408594,
      "grad_norm": 0.5722286288813279,
      "learning_rate": 3.2467082429300047e-06,
      "loss": 0.9351,
      "step": 31440
    },
    {
      "epoch": 1.9006466428959934,
      "grad_norm": 0.5417172030527615,
      "learning_rate": 3.243589211999254e-06,
      "loss": 0.9089,
      "step": 31450
    },
    {
      "epoch": 1.9012509820511272,
      "grad_norm": 0.5927939206647315,
      "learning_rate": 3.240470960566618e-06,
      "loss": 0.9262,
      "step": 31460
    },
    {
      "epoch": 1.901855321206261,
      "grad_norm": 0.5226270742463195,
      "learning_rate": 3.2373534900159833e-06,
      "loss": 0.9084,
      "step": 31470
    },
    {
      "epoch": 1.9024596603613948,
      "grad_norm": 0.567034355947198,
      "learning_rate": 3.2342368017308855e-06,
      "loss": 0.9158,
      "step": 31480
    },
    {
      "epoch": 1.9030639995165286,
      "grad_norm": 0.5339493036133449,
      "learning_rate": 3.231120897094517e-06,
      "loss": 0.9234,
      "step": 31490
    },
    {
      "epoch": 1.9036683386716624,
      "grad_norm": 0.5726067936928292,
      "learning_rate": 3.2280057774897256e-06,
      "loss": 0.9208,
      "step": 31500
    },
    {
      "epoch": 1.9042726778267964,
      "grad_norm": 0.5207599374433414,
      "learning_rate": 3.224891444299001e-06,
      "loss": 0.8899,
      "step": 31510
    },
    {
      "epoch": 1.9048770169819302,
      "grad_norm": 0.5191165272309807,
      "learning_rate": 3.221777898904492e-06,
      "loss": 0.9287,
      "step": 31520
    },
    {
      "epoch": 1.9054813561370643,
      "grad_norm": 0.5486190737368021,
      "learning_rate": 3.2186651426879976e-06,
      "loss": 0.9298,
      "step": 31530
    },
    {
      "epoch": 1.906085695292198,
      "grad_norm": 0.5500060209410718,
      "learning_rate": 3.215553177030961e-06,
      "loss": 0.9397,
      "step": 31540
    },
    {
      "epoch": 1.9066900344473319,
      "grad_norm": 0.4900360459171041,
      "learning_rate": 3.2124420033144783e-06,
      "loss": 0.9346,
      "step": 31550
    },
    {
      "epoch": 1.9072943736024657,
      "grad_norm": 0.5677897663859682,
      "learning_rate": 3.2093316229192968e-06,
      "loss": 0.916,
      "step": 31560
    },
    {
      "epoch": 1.9078987127575995,
      "grad_norm": 0.571046731051788,
      "learning_rate": 3.2062220372258047e-06,
      "loss": 0.9402,
      "step": 31570
    },
    {
      "epoch": 1.9085030519127333,
      "grad_norm": 0.5675328164879303,
      "learning_rate": 3.2031132476140436e-06,
      "loss": 0.9332,
      "step": 31580
    },
    {
      "epoch": 1.9091073910678673,
      "grad_norm": 0.5550489226848586,
      "learning_rate": 3.2000052554636997e-06,
      "loss": 0.9312,
      "step": 31590
    },
    {
      "epoch": 1.9097117302230011,
      "grad_norm": 0.5598651286430216,
      "learning_rate": 3.1968980621541047e-06,
      "loss": 0.9067,
      "step": 31600
    },
    {
      "epoch": 1.9103160693781351,
      "grad_norm": 0.5435394896623816,
      "learning_rate": 3.1937916690642356e-06,
      "loss": 0.9021,
      "step": 31610
    },
    {
      "epoch": 1.910920408533269,
      "grad_norm": 0.5270720182040158,
      "learning_rate": 3.190686077572718e-06,
      "loss": 0.9449,
      "step": 31620
    },
    {
      "epoch": 1.9115247476884027,
      "grad_norm": 0.5531124347734957,
      "learning_rate": 3.1875812890578138e-06,
      "loss": 0.9117,
      "step": 31630
    },
    {
      "epoch": 1.9121290868435366,
      "grad_norm": 0.5626353241899941,
      "learning_rate": 3.1844773048974365e-06,
      "loss": 0.9048,
      "step": 31640
    },
    {
      "epoch": 1.9127334259986704,
      "grad_norm": 0.5357859581854699,
      "learning_rate": 3.181374126469141e-06,
      "loss": 0.9118,
      "step": 31650
    },
    {
      "epoch": 1.9133377651538042,
      "grad_norm": 0.5551032153901071,
      "learning_rate": 3.1782717551501195e-06,
      "loss": 0.9089,
      "step": 31660
    },
    {
      "epoch": 1.9139421043089382,
      "grad_norm": 0.5643872963285096,
      "learning_rate": 3.1751701923172117e-06,
      "loss": 0.9099,
      "step": 31670
    },
    {
      "epoch": 1.914546443464072,
      "grad_norm": 0.5475899852929832,
      "learning_rate": 3.1720694393468976e-06,
      "loss": 0.9397,
      "step": 31680
    },
    {
      "epoch": 1.915150782619206,
      "grad_norm": 0.5910932896915573,
      "learning_rate": 3.168969497615295e-06,
      "loss": 0.9064,
      "step": 31690
    },
    {
      "epoch": 1.9157551217743398,
      "grad_norm": 0.6582655237850477,
      "learning_rate": 3.165870368498164e-06,
      "loss": 0.9493,
      "step": 31700
    },
    {
      "epoch": 1.9163594609294736,
      "grad_norm": 0.5871662553314744,
      "learning_rate": 3.1627720533709028e-06,
      "loss": 0.8935,
      "step": 31710
    },
    {
      "epoch": 1.9169638000846074,
      "grad_norm": 0.6627514787628234,
      "learning_rate": 3.1596745536085503e-06,
      "loss": 0.9068,
      "step": 31720
    },
    {
      "epoch": 1.9175681392397412,
      "grad_norm": 0.6638914032720941,
      "learning_rate": 3.1565778705857804e-06,
      "loss": 0.8906,
      "step": 31730
    },
    {
      "epoch": 1.9181724783948753,
      "grad_norm": 0.6386018044940778,
      "learning_rate": 3.1534820056769066e-06,
      "loss": 0.9355,
      "step": 31740
    },
    {
      "epoch": 1.918776817550009,
      "grad_norm": 0.6921545337885953,
      "learning_rate": 3.1503869602558824e-06,
      "loss": 0.9241,
      "step": 31750
    },
    {
      "epoch": 1.919381156705143,
      "grad_norm": 0.7130650941805561,
      "learning_rate": 3.1472927356962885e-06,
      "loss": 0.9401,
      "step": 31760
    },
    {
      "epoch": 1.919985495860277,
      "grad_norm": 0.7041319097258212,
      "learning_rate": 3.1441993333713506e-06,
      "loss": 0.9269,
      "step": 31770
    },
    {
      "epoch": 1.9205898350154107,
      "grad_norm": 0.6685905670555268,
      "learning_rate": 3.1411067546539277e-06,
      "loss": 0.9088,
      "step": 31780
    },
    {
      "epoch": 1.9211941741705445,
      "grad_norm": 0.6793384253852527,
      "learning_rate": 3.1380150009165066e-06,
      "loss": 0.9367,
      "step": 31790
    },
    {
      "epoch": 1.9217985133256783,
      "grad_norm": 0.6417536972951596,
      "learning_rate": 3.1349240735312157e-06,
      "loss": 0.9253,
      "step": 31800
    },
    {
      "epoch": 1.922402852480812,
      "grad_norm": 0.6182334685408064,
      "learning_rate": 3.1318339738698166e-06,
      "loss": 0.9275,
      "step": 31810
    },
    {
      "epoch": 1.9230071916359461,
      "grad_norm": 0.6451272615634099,
      "learning_rate": 3.1287447033036955e-06,
      "loss": 0.9416,
      "step": 31820
    },
    {
      "epoch": 1.92361153079108,
      "grad_norm": 0.6708339279732021,
      "learning_rate": 3.125656263203879e-06,
      "loss": 0.9096,
      "step": 31830
    },
    {
      "epoch": 1.924215869946214,
      "grad_norm": 0.6310388425930847,
      "learning_rate": 3.122568654941022e-06,
      "loss": 0.9281,
      "step": 31840
    },
    {
      "epoch": 1.9248202091013478,
      "grad_norm": 0.8092513423955408,
      "learning_rate": 3.1194818798854094e-06,
      "loss": 0.9269,
      "step": 31850
    },
    {
      "epoch": 1.9254245482564816,
      "grad_norm": 0.8020465918270393,
      "learning_rate": 3.116395939406957e-06,
      "loss": 0.9286,
      "step": 31860
    },
    {
      "epoch": 1.9260288874116154,
      "grad_norm": 0.7906053954025566,
      "learning_rate": 3.113310834875213e-06,
      "loss": 0.9216,
      "step": 31870
    },
    {
      "epoch": 1.9266332265667492,
      "grad_norm": 0.8188340326221923,
      "learning_rate": 3.1102265676593485e-06,
      "loss": 0.9476,
      "step": 31880
    },
    {
      "epoch": 1.927237565721883,
      "grad_norm": 0.750355734497124,
      "learning_rate": 3.107143139128168e-06,
      "loss": 0.9321,
      "step": 31890
    },
    {
      "epoch": 1.927841904877017,
      "grad_norm": 0.7863238217817754,
      "learning_rate": 3.1040605506501053e-06,
      "loss": 0.9259,
      "step": 31900
    },
    {
      "epoch": 1.9284462440321508,
      "grad_norm": 0.741469798061842,
      "learning_rate": 3.100978803593213e-06,
      "loss": 0.9148,
      "step": 31910
    },
    {
      "epoch": 1.9290505831872848,
      "grad_norm": 0.7765148417313548,
      "learning_rate": 3.097897899325179e-06,
      "loss": 0.9407,
      "step": 31920
    },
    {
      "epoch": 1.9296549223424186,
      "grad_norm": 0.7856535721489776,
      "learning_rate": 3.0948178392133155e-06,
      "loss": 0.9345,
      "step": 31930
    },
    {
      "epoch": 1.9302592614975524,
      "grad_norm": 0.7946613756760031,
      "learning_rate": 3.091738624624554e-06,
      "loss": 0.9231,
      "step": 31940
    },
    {
      "epoch": 1.9308636006526863,
      "grad_norm": 1.128315483264186,
      "learning_rate": 3.088660256925459e-06,
      "loss": 0.9214,
      "step": 31950
    },
    {
      "epoch": 1.93146793980782,
      "grad_norm": 1.1836628003029233,
      "learning_rate": 3.0855827374822134e-06,
      "loss": 0.9404,
      "step": 31960
    },
    {
      "epoch": 1.9320722789629539,
      "grad_norm": 1.1618924226568106,
      "learning_rate": 3.0825060676606266e-06,
      "loss": 0.9083,
      "step": 31970
    },
    {
      "epoch": 1.9326766181180879,
      "grad_norm": 1.199156684298891,
      "learning_rate": 3.0794302488261283e-06,
      "loss": 0.9241,
      "step": 31980
    },
    {
      "epoch": 1.9332809572732217,
      "grad_norm": 1.2470218375580546,
      "learning_rate": 3.076355282343776e-06,
      "loss": 0.9297,
      "step": 31990
    },
    {
      "epoch": 1.9338852964283557,
      "grad_norm": 0.8216792201198375,
      "learning_rate": 3.0732811695782394e-06,
      "loss": 0.9244,
      "step": 32000
    },
    {
      "epoch": 1.9344896355834895,
      "grad_norm": 0.8290002407210196,
      "learning_rate": 3.0702079118938182e-06,
      "loss": 0.9312,
      "step": 32010
    },
    {
      "epoch": 1.9350939747386233,
      "grad_norm": 1.4219101536771561,
      "learning_rate": 3.06713551065443e-06,
      "loss": 0.9356,
      "step": 32020
    },
    {
      "epoch": 1.9356983138937571,
      "grad_norm": 0.8077893072465951,
      "learning_rate": 3.0640639672236123e-06,
      "loss": 0.9127,
      "step": 32030
    },
    {
      "epoch": 1.936302653048891,
      "grad_norm": 0.93829282607337,
      "learning_rate": 3.0609932829645185e-06,
      "loss": 0.9397,
      "step": 32040
    },
    {
      "epoch": 1.9369069922040247,
      "grad_norm": 2.394815372434293,
      "learning_rate": 3.0579234592399243e-06,
      "loss": 0.9344,
      "step": 32050
    },
    {
      "epoch": 1.9375113313591588,
      "grad_norm": 2.4282761480067343,
      "learning_rate": 3.0548544974122257e-06,
      "loss": 0.9401,
      "step": 32060
    },
    {
      "epoch": 1.9381156705142926,
      "grad_norm": 2.25535593102297,
      "learning_rate": 3.0517863988434294e-06,
      "loss": 0.9177,
      "step": 32070
    },
    {
      "epoch": 1.9387200096694266,
      "grad_norm": 2.348962629243723,
      "learning_rate": 3.0487191648951654e-06,
      "loss": 0.9266,
      "step": 32080
    },
    {
      "epoch": 1.9393243488245604,
      "grad_norm": 2.840143765784924,
      "learning_rate": 3.0456527969286757e-06,
      "loss": 0.9464,
      "step": 32090
    },
    {
      "epoch": 1.9399286879796942,
      "grad_norm": 0.875474637443051,
      "learning_rate": 3.042587296304821e-06,
      "loss": 0.9023,
      "step": 32100
    },
    {
      "epoch": 1.940533027134828,
      "grad_norm": 0.9977103761292808,
      "learning_rate": 3.0395226643840735e-06,
      "loss": 0.9262,
      "step": 32110
    },
    {
      "epoch": 1.9411373662899618,
      "grad_norm": 0.9233311417519634,
      "learning_rate": 3.0364589025265257e-06,
      "loss": 0.9267,
      "step": 32120
    },
    {
      "epoch": 1.9417417054450958,
      "grad_norm": 0.8686267646710731,
      "learning_rate": 3.0333960120918772e-06,
      "loss": 0.914,
      "step": 32130
    },
    {
      "epoch": 1.9423460446002296,
      "grad_norm": 0.9021681688215947,
      "learning_rate": 3.030333994439444e-06,
      "loss": 0.938,
      "step": 32140
    },
    {
      "epoch": 1.9429503837553637,
      "grad_norm": 0.897337222074805,
      "learning_rate": 3.0272728509281577e-06,
      "loss": 0.901,
      "step": 32150
    },
    {
      "epoch": 1.9435547229104975,
      "grad_norm": 0.9915272990335809,
      "learning_rate": 3.024212582916556e-06,
      "loss": 0.9353,
      "step": 32160
    },
    {
      "epoch": 1.9441590620656313,
      "grad_norm": 0.9920408114625213,
      "learning_rate": 3.0211531917627922e-06,
      "loss": 0.9181,
      "step": 32170
    },
    {
      "epoch": 1.944763401220765,
      "grad_norm": 0.922240495540258,
      "learning_rate": 3.0180946788246314e-06,
      "loss": 0.907,
      "step": 32180
    },
    {
      "epoch": 1.9453677403758989,
      "grad_norm": 0.8906856500303493,
      "learning_rate": 3.0150370454594425e-06,
      "loss": 0.9126,
      "step": 32190
    },
    {
      "epoch": 1.9459720795310327,
      "grad_norm": 1.0091078276857819,
      "learning_rate": 3.011980293024212e-06,
      "loss": 0.91,
      "step": 32200
    },
    {
      "epoch": 1.9465764186861667,
      "grad_norm": 0.9846864296048661,
      "learning_rate": 3.008924422875531e-06,
      "loss": 0.9033,
      "step": 32210
    },
    {
      "epoch": 1.9471807578413005,
      "grad_norm": 1.1174820841487947,
      "learning_rate": 3.0058694363695994e-06,
      "loss": 0.9246,
      "step": 32220
    },
    {
      "epoch": 1.9477850969964345,
      "grad_norm": 1.055839686448261,
      "learning_rate": 3.0028153348622247e-06,
      "loss": 0.9287,
      "step": 32230
    },
    {
      "epoch": 1.9483894361515683,
      "grad_norm": 0.9914626314915221,
      "learning_rate": 2.999762119708825e-06,
      "loss": 0.9239,
      "step": 32240
    },
    {
      "epoch": 1.9489937753067021,
      "grad_norm": 1.1239900727366239,
      "learning_rate": 2.996709792264419e-06,
      "loss": 0.9302,
      "step": 32250
    },
    {
      "epoch": 1.949598114461836,
      "grad_norm": 0.9709258613994516,
      "learning_rate": 2.993658353883636e-06,
      "loss": 0.9203,
      "step": 32260
    },
    {
      "epoch": 1.9502024536169698,
      "grad_norm": 0.9804946366958124,
      "learning_rate": 2.9906078059207126e-06,
      "loss": 0.9179,
      "step": 32270
    },
    {
      "epoch": 1.9508067927721036,
      "grad_norm": 1.0676886964074659,
      "learning_rate": 2.987558149729483e-06,
      "loss": 0.905,
      "step": 32280
    },
    {
      "epoch": 1.9514111319272376,
      "grad_norm": 0.9875442842327805,
      "learning_rate": 2.984509386663391e-06,
      "loss": 0.914,
      "step": 32290
    },
    {
      "epoch": 1.9520154710823714,
      "grad_norm": 1.059226065297373,
      "learning_rate": 2.9814615180754856e-06,
      "loss": 0.9078,
      "step": 32300
    },
    {
      "epoch": 1.9526198102375054,
      "grad_norm": 1.1558804797938624,
      "learning_rate": 2.9784145453184127e-06,
      "loss": 0.9257,
      "step": 32310
    },
    {
      "epoch": 1.9532241493926392,
      "grad_norm": 1.2244021868212303,
      "learning_rate": 2.9753684697444244e-06,
      "loss": 0.9269,
      "step": 32320
    },
    {
      "epoch": 1.953828488547773,
      "grad_norm": 1.106647973666506,
      "learning_rate": 2.972323292705378e-06,
      "loss": 0.9207,
      "step": 32330
    },
    {
      "epoch": 1.9544328277029068,
      "grad_norm": 1.1065831486996487,
      "learning_rate": 2.9692790155527228e-06,
      "loss": 0.9149,
      "step": 32340
    },
    {
      "epoch": 1.9550371668580406,
      "grad_norm": 0.9404833343662748,
      "learning_rate": 2.966235639637518e-06,
      "loss": 0.9361,
      "step": 32350
    },
    {
      "epoch": 1.9556415060131744,
      "grad_norm": 0.9973618253511899,
      "learning_rate": 2.9631931663104162e-06,
      "loss": 0.9395,
      "step": 32360
    },
    {
      "epoch": 1.9562458451683085,
      "grad_norm": 0.9324200023926333,
      "learning_rate": 2.9601515969216767e-06,
      "loss": 0.9041,
      "step": 32370
    },
    {
      "epoch": 1.9568501843234423,
      "grad_norm": 0.9260919412899182,
      "learning_rate": 2.957110932821148e-06,
      "loss": 0.9484,
      "step": 32380
    },
    {
      "epoch": 1.9574545234785763,
      "grad_norm": 1.0172926808066802,
      "learning_rate": 2.954071175358285e-06,
      "loss": 0.92,
      "step": 32390
    },
    {
      "epoch": 1.95805886263371,
      "grad_norm": 1.0559499410005995,
      "learning_rate": 2.9510323258821383e-06,
      "loss": 0.9218,
      "step": 32400
    },
    {
      "epoch": 1.958663201788844,
      "grad_norm": 1.1322889084296959,
      "learning_rate": 2.9479943857413506e-06,
      "loss": 0.9323,
      "step": 32410
    },
    {
      "epoch": 1.9592675409439777,
      "grad_norm": 1.0526214508943277,
      "learning_rate": 2.944957356284168e-06,
      "loss": 0.921,
      "step": 32420
    },
    {
      "epoch": 1.9598718800991115,
      "grad_norm": 1.0208651283529961,
      "learning_rate": 2.9419212388584305e-06,
      "loss": 0.9079,
      "step": 32430
    },
    {
      "epoch": 1.9604762192542455,
      "grad_norm": 1.0171728187289686,
      "learning_rate": 2.9388860348115684e-06,
      "loss": 0.9198,
      "step": 32440
    },
    {
      "epoch": 1.9610805584093793,
      "grad_norm": 0.99155183818389,
      "learning_rate": 2.935851745490612e-06,
      "loss": 0.9116,
      "step": 32450
    },
    {
      "epoch": 1.9616848975645134,
      "grad_norm": 0.9673745494452112,
      "learning_rate": 2.9328183722421865e-06,
      "loss": 0.9281,
      "step": 32460
    },
    {
      "epoch": 1.9622892367196472,
      "grad_norm": 0.9474513139669102,
      "learning_rate": 2.9297859164125047e-06,
      "loss": 0.9152,
      "step": 32470
    },
    {
      "epoch": 1.962893575874781,
      "grad_norm": 1.0083598815518564,
      "learning_rate": 2.926754379347378e-06,
      "loss": 0.9249,
      "step": 32480
    },
    {
      "epoch": 1.9634979150299148,
      "grad_norm": 1.003871754276492,
      "learning_rate": 2.9237237623922064e-06,
      "loss": 0.9322,
      "step": 32490
    },
    {
      "epoch": 1.9641022541850486,
      "grad_norm": 1.1358619924346611,
      "learning_rate": 2.920694066891984e-06,
      "loss": 0.9181,
      "step": 32500
    },
    {
      "epoch": 1.9647065933401824,
      "grad_norm": 1.1846682303453873,
      "learning_rate": 2.9176652941912924e-06,
      "loss": 0.919,
      "step": 32510
    },
    {
      "epoch": 1.9653109324953164,
      "grad_norm": 1.1408791980114923,
      "learning_rate": 2.9146374456343106e-06,
      "loss": 0.9302,
      "step": 32520
    },
    {
      "epoch": 1.9659152716504502,
      "grad_norm": 1.1349605319128238,
      "learning_rate": 2.9116105225647982e-06,
      "loss": 0.9402,
      "step": 32530
    },
    {
      "epoch": 1.9665196108055842,
      "grad_norm": 1.1148278751903036,
      "learning_rate": 2.90858452632611e-06,
      "loss": 0.9158,
      "step": 32540
    },
    {
      "epoch": 1.967123949960718,
      "grad_norm": 1.1477944399757234,
      "learning_rate": 2.9055594582611922e-06,
      "loss": 0.9219,
      "step": 32550
    },
    {
      "epoch": 1.9677282891158518,
      "grad_norm": 1.0187724455338298,
      "learning_rate": 2.9025353197125703e-06,
      "loss": 0.9258,
      "step": 32560
    },
    {
      "epoch": 1.9683326282709857,
      "grad_norm": 1.0752010407910595,
      "learning_rate": 2.899512112022364e-06,
      "loss": 0.9244,
      "step": 32570
    },
    {
      "epoch": 1.9689369674261195,
      "grad_norm": 1.100695413139785,
      "learning_rate": 2.8964898365322803e-06,
      "loss": 0.9364,
      "step": 32580
    },
    {
      "epoch": 1.9695413065812533,
      "grad_norm": 1.137557064747928,
      "learning_rate": 2.8934684945836055e-06,
      "loss": 0.898,
      "step": 32590
    },
    {
      "epoch": 1.9701456457363873,
      "grad_norm": 1.2278894678868826,
      "learning_rate": 2.890448087517219e-06,
      "loss": 0.931,
      "step": 32600
    },
    {
      "epoch": 1.970749984891521,
      "grad_norm": 1.107111899524758,
      "learning_rate": 2.8874286166735853e-06,
      "loss": 0.9035,
      "step": 32610
    },
    {
      "epoch": 1.9713543240466551,
      "grad_norm": 1.1943072974599098,
      "learning_rate": 2.884410083392745e-06,
      "loss": 0.9118,
      "step": 32620
    },
    {
      "epoch": 1.971958663201789,
      "grad_norm": 1.1005789652895215,
      "learning_rate": 2.881392489014332e-06,
      "loss": 0.9114,
      "step": 32630
    },
    {
      "epoch": 1.9725630023569227,
      "grad_norm": 1.1998377559811042,
      "learning_rate": 2.8783758348775605e-06,
      "loss": 0.8738,
      "step": 32640
    },
    {
      "epoch": 1.9731673415120565,
      "grad_norm": 1.4153646027691513,
      "learning_rate": 2.875360122321224e-06,
      "loss": 0.9223,
      "step": 32650
    },
    {
      "epoch": 1.9737716806671903,
      "grad_norm": 1.3440989087855313,
      "learning_rate": 2.872345352683704e-06,
      "loss": 0.9398,
      "step": 32660
    },
    {
      "epoch": 1.9743760198223241,
      "grad_norm": 1.3037362096791534,
      "learning_rate": 2.869331527302957e-06,
      "loss": 0.9232,
      "step": 32670
    },
    {
      "epoch": 1.9749803589774582,
      "grad_norm": 1.3995213591528144,
      "learning_rate": 2.866318647516527e-06,
      "loss": 0.9195,
      "step": 32680
    },
    {
      "epoch": 1.975584698132592,
      "grad_norm": 1.5130545928590327,
      "learning_rate": 2.8633067146615325e-06,
      "loss": 0.9323,
      "step": 32690
    },
    {
      "epoch": 1.976189037287726,
      "grad_norm": 0.7913265420283543,
      "learning_rate": 2.860295730074676e-06,
      "loss": 0.9215,
      "step": 32700
    },
    {
      "epoch": 1.9767933764428598,
      "grad_norm": 0.8435315743548842,
      "learning_rate": 2.857285695092239e-06,
      "loss": 0.923,
      "step": 32710
    },
    {
      "epoch": 1.9773977155979936,
      "grad_norm": 0.8185880430242493,
      "learning_rate": 2.8542766110500765e-06,
      "loss": 0.9248,
      "step": 32720
    },
    {
      "epoch": 1.9780020547531274,
      "grad_norm": 0.8836776245724742,
      "learning_rate": 2.8512684792836267e-06,
      "loss": 0.915,
      "step": 32730
    },
    {
      "epoch": 1.9786063939082612,
      "grad_norm": 0.8340818409642348,
      "learning_rate": 2.848261301127907e-06,
      "loss": 0.9122,
      "step": 32740
    },
    {
      "epoch": 1.979210733063395,
      "grad_norm": 0.7854443586623282,
      "learning_rate": 2.8452550779175024e-06,
      "loss": 0.9056,
      "step": 32750
    },
    {
      "epoch": 1.979815072218529,
      "grad_norm": 0.7529096818552463,
      "learning_rate": 2.842249810986584e-06,
      "loss": 0.933,
      "step": 32760
    },
    {
      "epoch": 1.980419411373663,
      "grad_norm": 0.7652955543004387,
      "learning_rate": 2.8392455016688948e-06,
      "loss": 0.9283,
      "step": 32770
    },
    {
      "epoch": 1.9810237505287969,
      "grad_norm": 0.7798261430277603,
      "learning_rate": 2.8362421512977483e-06,
      "loss": 0.9265,
      "step": 32780
    },
    {
      "epoch": 1.9816280896839307,
      "grad_norm": 0.7764552270940536,
      "learning_rate": 2.83323976120604e-06,
      "loss": 0.9463,
      "step": 32790
    },
    {
      "epoch": 1.9822324288390645,
      "grad_norm": 0.7727308479756552,
      "learning_rate": 2.830238332726236e-06,
      "loss": 0.9165,
      "step": 32800
    },
    {
      "epoch": 1.9828367679941983,
      "grad_norm": 0.8486144258989697,
      "learning_rate": 2.8272378671903732e-06,
      "loss": 0.9136,
      "step": 32810
    },
    {
      "epoch": 1.983441107149332,
      "grad_norm": 0.7747073893964028,
      "learning_rate": 2.824238365930064e-06,
      "loss": 0.9222,
      "step": 32820
    },
    {
      "epoch": 1.984045446304466,
      "grad_norm": 0.8308668580416605,
      "learning_rate": 2.8212398302764945e-06,
      "loss": 0.9349,
      "step": 32830
    },
    {
      "epoch": 1.9846497854596,
      "grad_norm": 0.9216706197598951,
      "learning_rate": 2.818242261560416e-06,
      "loss": 0.9387,
      "step": 32840
    },
    {
      "epoch": 1.985254124614734,
      "grad_norm": 0.6535766427815558,
      "learning_rate": 2.815245661112157e-06,
      "loss": 0.9105,
      "step": 32850
    },
    {
      "epoch": 1.9858584637698677,
      "grad_norm": 0.5853580374722734,
      "learning_rate": 2.812250030261615e-06,
      "loss": 0.9029,
      "step": 32860
    },
    {
      "epoch": 1.9864628029250015,
      "grad_norm": 0.5444700147799568,
      "learning_rate": 2.809255370338254e-06,
      "loss": 0.9296,
      "step": 32870
    },
    {
      "epoch": 1.9870671420801354,
      "grad_norm": 0.609405547699801,
      "learning_rate": 2.8062616826711087e-06,
      "loss": 0.9414,
      "step": 32880
    },
    {
      "epoch": 1.9876714812352692,
      "grad_norm": 0.5727461239091808,
      "learning_rate": 2.8032689685887866e-06,
      "loss": 0.9269,
      "step": 32890
    },
    {
      "epoch": 1.988275820390403,
      "grad_norm": 0.6674327437341456,
      "learning_rate": 2.8002772294194545e-06,
      "loss": 0.9202,
      "step": 32900
    },
    {
      "epoch": 1.988880159545537,
      "grad_norm": 0.687543998492259,
      "learning_rate": 2.797286466490854e-06,
      "loss": 0.9396,
      "step": 32910
    },
    {
      "epoch": 1.9894844987006708,
      "grad_norm": 0.6250925155005832,
      "learning_rate": 2.7942966811302917e-06,
      "loss": 0.9112,
      "step": 32920
    },
    {
      "epoch": 1.9900888378558048,
      "grad_norm": 0.6006309650330449,
      "learning_rate": 2.7913078746646374e-06,
      "loss": 0.9172,
      "step": 32930
    },
    {
      "epoch": 1.9906931770109386,
      "grad_norm": 0.5888906987637169,
      "learning_rate": 2.78832004842033e-06,
      "loss": 0.9433,
      "step": 32940
    },
    {
      "epoch": 1.9912975161660724,
      "grad_norm": 0.8400342315851975,
      "learning_rate": 2.78533320372337e-06,
      "loss": 0.9066,
      "step": 32950
    },
    {
      "epoch": 1.9919018553212062,
      "grad_norm": 0.8195413255021915,
      "learning_rate": 2.7823473418993276e-06,
      "loss": 0.9111,
      "step": 32960
    },
    {
      "epoch": 1.99250619447634,
      "grad_norm": 0.7897071085945362,
      "learning_rate": 2.7793624642733284e-06,
      "loss": 0.9022,
      "step": 32970
    },
    {
      "epoch": 1.9931105336314738,
      "grad_norm": 0.8736510745012994,
      "learning_rate": 2.776378572170071e-06,
      "loss": 0.9081,
      "step": 32980
    },
    {
      "epoch": 1.9937148727866079,
      "grad_norm": 0.8179960109206366,
      "learning_rate": 2.7733956669138085e-06,
      "loss": 0.9204,
      "step": 32990
    },
    {
      "epoch": 1.9943192119417417,
      "grad_norm": 0.5443723734672266,
      "learning_rate": 2.77041374982836e-06,
      "loss": 0.9338,
      "step": 33000
    },
    {
      "epoch": 1.9949235510968757,
      "grad_norm": 0.5681284356864479,
      "learning_rate": 2.7674328222371085e-06,
      "loss": 0.9135,
      "step": 33010
    },
    {
      "epoch": 1.9955278902520095,
      "grad_norm": 0.5791545706207375,
      "learning_rate": 2.7644528854629904e-06,
      "loss": 0.9029,
      "step": 33020
    },
    {
      "epoch": 1.9961322294071433,
      "grad_norm": 0.5180478658915624,
      "learning_rate": 2.7614739408285084e-06,
      "loss": 0.9349,
      "step": 33030
    },
    {
      "epoch": 1.996736568562277,
      "grad_norm": 0.5630728117922584,
      "learning_rate": 2.758495989655724e-06,
      "loss": 0.9332,
      "step": 33040
    },
    {
      "epoch": 1.997340907717411,
      "grad_norm": 0.5768387302422116,
      "learning_rate": 2.7555190332662585e-06,
      "loss": 0.9387,
      "step": 33050
    },
    {
      "epoch": 1.9979452468725447,
      "grad_norm": 0.5318731060149366,
      "learning_rate": 2.752543072981286e-06,
      "loss": 0.9529,
      "step": 33060
    },
    {
      "epoch": 1.9985495860276787,
      "grad_norm": 0.5904056298317066,
      "learning_rate": 2.749568110121545e-06,
      "loss": 0.9411,
      "step": 33070
    },
    {
      "epoch": 1.9991539251828125,
      "grad_norm": 0.5724855952547209,
      "learning_rate": 2.746594146007331e-06,
      "loss": 0.9129,
      "step": 33080
    },
    {
      "epoch": 1.9997582643379466,
      "grad_norm": 0.5828335575439317,
      "learning_rate": 2.7436211819584914e-06,
      "loss": 0.9302,
      "step": 33090
    },
    {
      "epoch": 2.0003626034930804,
      "grad_norm": 0.5234494648789862,
      "learning_rate": 2.740649219294433e-06,
      "loss": 0.9076,
      "step": 33100
    },
    {
      "epoch": 2.000966942648214,
      "grad_norm": 0.5628551805580915,
      "learning_rate": 2.737678259334121e-06,
      "loss": 0.8893,
      "step": 33110
    },
    {
      "epoch": 2.001571281803348,
      "grad_norm": 0.5467440707166451,
      "learning_rate": 2.7347083033960688e-06,
      "loss": 0.8621,
      "step": 33120
    },
    {
      "epoch": 2.002175620958482,
      "grad_norm": 0.5192012510442349,
      "learning_rate": 2.731739352798348e-06,
      "loss": 0.8876,
      "step": 33130
    },
    {
      "epoch": 2.0027799601136156,
      "grad_norm": 0.5528029113129463,
      "learning_rate": 2.728771408858589e-06,
      "loss": 0.8903,
      "step": 33140
    },
    {
      "epoch": 2.00338429926875,
      "grad_norm": 0.5347676345274723,
      "learning_rate": 2.725804472893964e-06,
      "loss": 0.8721,
      "step": 33150
    },
    {
      "epoch": 2.0039886384238836,
      "grad_norm": 0.5321168639117014,
      "learning_rate": 2.7228385462212076e-06,
      "loss": 0.9036,
      "step": 33160
    },
    {
      "epoch": 2.0045929775790174,
      "grad_norm": 0.5365723743891695,
      "learning_rate": 2.7198736301566044e-06,
      "loss": 0.8944,
      "step": 33170
    },
    {
      "epoch": 2.0051973167341512,
      "grad_norm": 0.5045542509670169,
      "learning_rate": 2.716909726015985e-06,
      "loss": 0.8903,
      "step": 33180
    },
    {
      "epoch": 2.005801655889285,
      "grad_norm": 0.5539428151647481,
      "learning_rate": 2.7139468351147396e-06,
      "loss": 0.8869,
      "step": 33190
    },
    {
      "epoch": 2.006405995044419,
      "grad_norm": 0.5481305980648615,
      "learning_rate": 2.7109849587678005e-06,
      "loss": 0.8765,
      "step": 33200
    },
    {
      "epoch": 2.0070103341995527,
      "grad_norm": 0.5247873972754932,
      "learning_rate": 2.708024098289657e-06,
      "loss": 0.8677,
      "step": 33210
    },
    {
      "epoch": 2.0076146733546865,
      "grad_norm": 0.5199961803702887,
      "learning_rate": 2.7050642549943407e-06,
      "loss": 0.867,
      "step": 33220
    },
    {
      "epoch": 2.0082190125098207,
      "grad_norm": 0.546377177548141,
      "learning_rate": 2.702105430195438e-06,
      "loss": 0.8812,
      "step": 33230
    },
    {
      "epoch": 2.0088233516649545,
      "grad_norm": 0.5367156471007768,
      "learning_rate": 2.6991476252060776e-06,
      "loss": 0.8928,
      "step": 33240
    },
    {
      "epoch": 2.0094276908200883,
      "grad_norm": 0.5662789050973269,
      "learning_rate": 2.6961908413389392e-06,
      "loss": 0.8831,
      "step": 33250
    },
    {
      "epoch": 2.010032029975222,
      "grad_norm": 0.5645438196801373,
      "learning_rate": 2.693235079906252e-06,
      "loss": 0.9111,
      "step": 33260
    },
    {
      "epoch": 2.010636369130356,
      "grad_norm": 0.5568149281941507,
      "learning_rate": 2.690280342219783e-06,
      "loss": 0.8841,
      "step": 33270
    },
    {
      "epoch": 2.0112407082854897,
      "grad_norm": 0.5602450264231862,
      "learning_rate": 2.6873266295908517e-06,
      "loss": 0.8919,
      "step": 33280
    },
    {
      "epoch": 2.0118450474406235,
      "grad_norm": 0.5643730607117805,
      "learning_rate": 2.684373943330324e-06,
      "loss": 0.9105,
      "step": 33290
    },
    {
      "epoch": 2.0124493865957573,
      "grad_norm": 0.6433892215861963,
      "learning_rate": 2.6814222847486016e-06,
      "loss": 0.8791,
      "step": 33300
    },
    {
      "epoch": 2.0130537257508916,
      "grad_norm": 0.621168110091114,
      "learning_rate": 2.6784716551556377e-06,
      "loss": 0.9081,
      "step": 33310
    },
    {
      "epoch": 2.0136580649060254,
      "grad_norm": 0.639073527533395,
      "learning_rate": 2.6755220558609308e-06,
      "loss": 0.8562,
      "step": 33320
    },
    {
      "epoch": 2.014262404061159,
      "grad_norm": 0.6400031150970298,
      "learning_rate": 2.672573488173512e-06,
      "loss": 0.9029,
      "step": 33330
    },
    {
      "epoch": 2.014866743216293,
      "grad_norm": 0.6280523070689569,
      "learning_rate": 2.669625953401963e-06,
      "loss": 0.8873,
      "step": 33340
    },
    {
      "epoch": 2.015471082371427,
      "grad_norm": 0.7477962293475895,
      "learning_rate": 2.6666794528544083e-06,
      "loss": 0.8924,
      "step": 33350
    },
    {
      "epoch": 2.0160754215265606,
      "grad_norm": 0.6903162201229224,
      "learning_rate": 2.663733987838504e-06,
      "loss": 0.8944,
      "step": 33360
    },
    {
      "epoch": 2.0166797606816944,
      "grad_norm": 0.7261448503178681,
      "learning_rate": 2.660789559661455e-06,
      "loss": 0.8977,
      "step": 33370
    },
    {
      "epoch": 2.017284099836828,
      "grad_norm": 0.6786592217617367,
      "learning_rate": 2.6578461696300043e-06,
      "loss": 0.895,
      "step": 33380
    },
    {
      "epoch": 2.0178884389919625,
      "grad_norm": 0.6412472848510454,
      "learning_rate": 2.654903819050435e-06,
      "loss": 0.8645,
      "step": 33390
    },
    {
      "epoch": 2.0184927781470963,
      "grad_norm": 0.6444920737505094,
      "learning_rate": 2.651962509228563e-06,
      "loss": 0.9081,
      "step": 33400
    },
    {
      "epoch": 2.01909711730223,
      "grad_norm": 0.5933777225042867,
      "learning_rate": 2.6490222414697497e-06,
      "loss": 0.8892,
      "step": 33410
    },
    {
      "epoch": 2.019701456457364,
      "grad_norm": 0.6306811462696658,
      "learning_rate": 2.646083017078891e-06,
      "loss": 0.9064,
      "step": 33420
    },
    {
      "epoch": 2.0203057956124977,
      "grad_norm": 0.6293017800076647,
      "learning_rate": 2.643144837360418e-06,
      "loss": 0.8565,
      "step": 33430
    },
    {
      "epoch": 2.0209101347676315,
      "grad_norm": 0.6105091733849285,
      "learning_rate": 2.640207703618302e-06,
      "loss": 0.8871,
      "step": 33440
    },
    {
      "epoch": 2.0215144739227653,
      "grad_norm": 0.8366734512635472,
      "learning_rate": 2.6372716171560455e-06,
      "loss": 0.884,
      "step": 33450
    },
    {
      "epoch": 2.0221188130778995,
      "grad_norm": 0.829892759850703,
      "learning_rate": 2.634336579276692e-06,
      "loss": 0.8935,
      "step": 33460
    },
    {
      "epoch": 2.0227231522330333,
      "grad_norm": 0.7751375567901571,
      "learning_rate": 2.6314025912828113e-06,
      "loss": 0.9244,
      "step": 33470
    },
    {
      "epoch": 2.023327491388167,
      "grad_norm": 0.7849102813434207,
      "learning_rate": 2.6284696544765176e-06,
      "loss": 0.9112,
      "step": 33480
    },
    {
      "epoch": 2.023931830543301,
      "grad_norm": 0.8071163824775153,
      "learning_rate": 2.62553777015945e-06,
      "loss": 0.8801,
      "step": 33490
    },
    {
      "epoch": 2.0245361696984348,
      "grad_norm": 0.8377023058225088,
      "learning_rate": 2.6226069396327836e-06,
      "loss": 0.9004,
      "step": 33500
    },
    {
      "epoch": 2.0251405088535686,
      "grad_norm": 0.7428017653108226,
      "learning_rate": 2.6196771641972298e-06,
      "loss": 0.8914,
      "step": 33510
    },
    {
      "epoch": 2.0257448480087024,
      "grad_norm": 0.8355728375626714,
      "learning_rate": 2.6167484451530234e-06,
      "loss": 0.8964,
      "step": 33520
    },
    {
      "epoch": 2.026349187163836,
      "grad_norm": 0.788687173689808,
      "learning_rate": 2.613820783799936e-06,
      "loss": 0.8756,
      "step": 33530
    },
    {
      "epoch": 2.0269535263189704,
      "grad_norm": 0.7985096227824965,
      "learning_rate": 2.610894181437273e-06,
      "loss": 0.881,
      "step": 33540
    },
    {
      "epoch": 2.027557865474104,
      "grad_norm": 1.2744338711871706,
      "learning_rate": 2.6079686393638592e-06,
      "loss": 0.8874,
      "step": 33550
    },
    {
      "epoch": 2.028162204629238,
      "grad_norm": 1.2668373590520514,
      "learning_rate": 2.6050441588780584e-06,
      "loss": 0.8756,
      "step": 33560
    },
    {
      "epoch": 2.028766543784372,
      "grad_norm": 1.249836076359573,
      "learning_rate": 2.602120741277762e-06,
      "loss": 0.8883,
      "step": 33570
    },
    {
      "epoch": 2.0293708829395056,
      "grad_norm": 1.1500137432809445,
      "learning_rate": 2.599198387860383e-06,
      "loss": 0.8656,
      "step": 33580
    },
    {
      "epoch": 2.0299752220946394,
      "grad_norm": 1.225715597692489,
      "learning_rate": 2.596277099922869e-06,
      "loss": 0.8868,
      "step": 33590
    },
    {
      "epoch": 2.0305795612497732,
      "grad_norm": 0.7786545909646788,
      "learning_rate": 2.5933568787616953e-06,
      "loss": 0.8695,
      "step": 33600
    },
    {
      "epoch": 2.031183900404907,
      "grad_norm": 1.0826214339964257,
      "learning_rate": 2.5904377256728565e-06,
      "loss": 0.897,
      "step": 33610
    },
    {
      "epoch": 2.0317882395600413,
      "grad_norm": 1.0644701152236413,
      "learning_rate": 2.58751964195188e-06,
      "loss": 0.9061,
      "step": 33620
    },
    {
      "epoch": 2.032392578715175,
      "grad_norm": 0.7746719706115537,
      "learning_rate": 2.584602628893819e-06,
      "loss": 0.9032,
      "step": 33630
    },
    {
      "epoch": 2.032996917870309,
      "grad_norm": 0.9177230023310055,
      "learning_rate": 2.5816866877932433e-06,
      "loss": 0.9111,
      "step": 33640
    },
    {
      "epoch": 2.0336012570254427,
      "grad_norm": 2.3606877000249202,
      "learning_rate": 2.5787718199442555e-06,
      "loss": 0.9004,
      "step": 33650
    },
    {
      "epoch": 2.0342055961805765,
      "grad_norm": 2.444030987959464,
      "learning_rate": 2.5758580266404808e-06,
      "loss": 0.894,
      "step": 33660
    },
    {
      "epoch": 2.0348099353357103,
      "grad_norm": 2.2691657418387385,
      "learning_rate": 2.5729453091750617e-06,
      "loss": 0.8995,
      "step": 33670
    },
    {
      "epoch": 2.035414274490844,
      "grad_norm": 2.536120913511076,
      "learning_rate": 2.570033668840669e-06,
      "loss": 0.8879,
      "step": 33680
    },
    {
      "epoch": 2.036018613645978,
      "grad_norm": 2.4728535889157865,
      "learning_rate": 2.567123106929497e-06,
      "loss": 0.8919,
      "step": 33690
    },
    {
      "epoch": 2.036622952801112,
      "grad_norm": 0.9213739755567852,
      "learning_rate": 2.5642136247332517e-06,
      "loss": 0.8649,
      "step": 33700
    },
    {
      "epoch": 2.037227291956246,
      "grad_norm": 0.9047894836272231,
      "learning_rate": 2.5613052235431712e-06,
      "loss": 0.8937,
      "step": 33710
    },
    {
      "epoch": 2.0378316311113798,
      "grad_norm": 0.9934360767452772,
      "learning_rate": 2.5583979046500063e-06,
      "loss": 0.8887,
      "step": 33720
    },
    {
      "epoch": 2.0384359702665136,
      "grad_norm": 0.9132731047293965,
      "learning_rate": 2.5554916693440323e-06,
      "loss": 0.8796,
      "step": 33730
    },
    {
      "epoch": 2.0390403094216474,
      "grad_norm": 0.906634365101205,
      "learning_rate": 2.552586518915038e-06,
      "loss": 0.9015,
      "step": 33740
    },
    {
      "epoch": 2.039644648576781,
      "grad_norm": 1.0039848914728786,
      "learning_rate": 2.549682454652337e-06,
      "loss": 0.8686,
      "step": 33750
    },
    {
      "epoch": 2.040248987731915,
      "grad_norm": 0.9960613291996244,
      "learning_rate": 2.5467794778447586e-06,
      "loss": 0.8727,
      "step": 33760
    },
    {
      "epoch": 2.0408533268870492,
      "grad_norm": 1.0491001038892025,
      "learning_rate": 2.543877589780646e-06,
      "loss": 0.8899,
      "step": 33770
    },
    {
      "epoch": 2.041457666042183,
      "grad_norm": 1.019584010266302,
      "learning_rate": 2.5409767917478633e-06,
      "loss": 0.8816,
      "step": 33780
    },
    {
      "epoch": 2.042062005197317,
      "grad_norm": 0.9578093684340653,
      "learning_rate": 2.538077085033792e-06,
      "loss": 0.8759,
      "step": 33790
    },
    {
      "epoch": 2.0426663443524506,
      "grad_norm": 1.1118673861825716,
      "learning_rate": 2.535178470925323e-06,
      "loss": 0.8812,
      "step": 33800
    },
    {
      "epoch": 2.0432706835075845,
      "grad_norm": 1.0435113535243707,
      "learning_rate": 2.532280950708868e-06,
      "loss": 0.8941,
      "step": 33810
    },
    {
      "epoch": 2.0438750226627183,
      "grad_norm": 1.0694698018499207,
      "learning_rate": 2.5293845256703533e-06,
      "loss": 0.8945,
      "step": 33820
    },
    {
      "epoch": 2.044479361817852,
      "grad_norm": 1.1059337705513197,
      "learning_rate": 2.5264891970952137e-06,
      "loss": 0.8945,
      "step": 33830
    },
    {
      "epoch": 2.045083700972986,
      "grad_norm": 1.0589327270025761,
      "learning_rate": 2.5235949662684027e-06,
      "loss": 0.8863,
      "step": 33840
    },
    {
      "epoch": 2.04568804012812,
      "grad_norm": 1.0755303399925493,
      "learning_rate": 2.520701834474387e-06,
      "loss": 0.8662,
      "step": 33850
    },
    {
      "epoch": 2.046292379283254,
      "grad_norm": 1.1532568856177698,
      "learning_rate": 2.517809802997139e-06,
      "loss": 0.9012,
      "step": 33860
    },
    {
      "epoch": 2.0468967184383877,
      "grad_norm": 1.0807277105029338,
      "learning_rate": 2.51491887312015e-06,
      "loss": 0.8672,
      "step": 33870
    },
    {
      "epoch": 2.0475010575935215,
      "grad_norm": 1.002487627145067,
      "learning_rate": 2.5120290461264213e-06,
      "loss": 0.873,
      "step": 33880
    },
    {
      "epoch": 2.0481053967486553,
      "grad_norm": 1.0167593572073805,
      "learning_rate": 2.509140323298459e-06,
      "loss": 0.8977,
      "step": 33890
    },
    {
      "epoch": 2.048709735903789,
      "grad_norm": 1.1223595145458065,
      "learning_rate": 2.5062527059182853e-06,
      "loss": 0.8774,
      "step": 33900
    },
    {
      "epoch": 2.049314075058923,
      "grad_norm": 1.101351498470469,
      "learning_rate": 2.5033661952674315e-06,
      "loss": 0.8822,
      "step": 33910
    },
    {
      "epoch": 2.0499184142140567,
      "grad_norm": 1.0782198581993534,
      "learning_rate": 2.500480792626932e-06,
      "loss": 0.8993,
      "step": 33920
    },
    {
      "epoch": 2.050522753369191,
      "grad_norm": 1.162541636524804,
      "learning_rate": 2.4975964992773357e-06,
      "loss": 0.887,
      "step": 33930
    },
    {
      "epoch": 2.051127092524325,
      "grad_norm": 1.1544992755469565,
      "learning_rate": 2.4947133164986987e-06,
      "loss": 0.882,
      "step": 33940
    },
    {
      "epoch": 2.0517314316794586,
      "grad_norm": 1.02484742504229,
      "learning_rate": 2.491831245570579e-06,
      "loss": 0.8888,
      "step": 33950
    },
    {
      "epoch": 2.0523357708345924,
      "grad_norm": 1.005488902778665,
      "learning_rate": 2.4889502877720477e-06,
      "loss": 0.8949,
      "step": 33960
    },
    {
      "epoch": 2.052940109989726,
      "grad_norm": 0.9549913290537595,
      "learning_rate": 2.4860704443816753e-06,
      "loss": 0.8779,
      "step": 33970
    },
    {
      "epoch": 2.05354444914486,
      "grad_norm": 1.0144438246472387,
      "learning_rate": 2.4831917166775453e-06,
      "loss": 0.9049,
      "step": 33980
    },
    {
      "epoch": 2.054148788299994,
      "grad_norm": 0.9934721776547466,
      "learning_rate": 2.480314105937238e-06,
      "loss": 0.9205,
      "step": 33990
    },
    {
      "epoch": 2.0547531274551276,
      "grad_norm": 1.1612390010795213,
      "learning_rate": 2.477437613437846e-06,
      "loss": 0.9014,
      "step": 34000
    },
    {
      "epoch": 2.055357466610262,
      "grad_norm": 1.1732583877051292,
      "learning_rate": 2.4745622404559572e-06,
      "loss": 0.9087,
      "step": 34010
    },
    {
      "epoch": 2.0559618057653957,
      "grad_norm": 1.136773696330377,
      "learning_rate": 2.4716879882676692e-06,
      "loss": 0.9088,
      "step": 34020
    },
    {
      "epoch": 2.0565661449205295,
      "grad_norm": 1.0098662837013497,
      "learning_rate": 2.468814858148582e-06,
      "loss": 0.8616,
      "step": 34030
    },
    {
      "epoch": 2.0571704840756633,
      "grad_norm": 1.0702287416437863,
      "learning_rate": 2.465942851373792e-06,
      "loss": 0.8859,
      "step": 34040
    },
    {
      "epoch": 2.057774823230797,
      "grad_norm": 1.1198816440791115,
      "learning_rate": 2.4630719692179027e-06,
      "loss": 0.9049,
      "step": 34050
    },
    {
      "epoch": 2.058379162385931,
      "grad_norm": 1.0496563566526718,
      "learning_rate": 2.4602022129550163e-06,
      "loss": 0.8962,
      "step": 34060
    },
    {
      "epoch": 2.0589835015410647,
      "grad_norm": 1.0561577415803982,
      "learning_rate": 2.4573335838587374e-06,
      "loss": 0.8999,
      "step": 34070
    },
    {
      "epoch": 2.0595878406961985,
      "grad_norm": 1.0572024573012317,
      "learning_rate": 2.454466083202165e-06,
      "loss": 0.8979,
      "step": 34080
    },
    {
      "epoch": 2.0601921798513327,
      "grad_norm": 1.0238078496867966,
      "learning_rate": 2.4515997122579022e-06,
      "loss": 0.8915,
      "step": 34090
    },
    {
      "epoch": 2.0607965190064665,
      "grad_norm": 1.23372882607432,
      "learning_rate": 2.4487344722980523e-06,
      "loss": 0.8691,
      "step": 34100
    },
    {
      "epoch": 2.0614008581616003,
      "grad_norm": 1.1909270512188104,
      "learning_rate": 2.4458703645942082e-06,
      "loss": 0.8984,
      "step": 34110
    },
    {
      "epoch": 2.062005197316734,
      "grad_norm": 1.267788850990393,
      "learning_rate": 2.4430073904174687e-06,
      "loss": 0.9125,
      "step": 34120
    },
    {
      "epoch": 2.062609536471868,
      "grad_norm": 1.296894589043979,
      "learning_rate": 2.4401455510384285e-06,
      "loss": 0.8903,
      "step": 34130
    },
    {
      "epoch": 2.0632138756270018,
      "grad_norm": 1.1630203149076703,
      "learning_rate": 2.4372848477271737e-06,
      "loss": 0.8723,
      "step": 34140
    },
    {
      "epoch": 2.0638182147821356,
      "grad_norm": 1.081595331338194,
      "learning_rate": 2.4344252817532897e-06,
      "loss": 0.8787,
      "step": 34150
    },
    {
      "epoch": 2.06442255393727,
      "grad_norm": 1.0969701820912907,
      "learning_rate": 2.43156685438586e-06,
      "loss": 0.9216,
      "step": 34160
    },
    {
      "epoch": 2.0650268930924036,
      "grad_norm": 1.0998530233601764,
      "learning_rate": 2.4287095668934553e-06,
      "loss": 0.8803,
      "step": 34170
    },
    {
      "epoch": 2.0656312322475374,
      "grad_norm": 1.1125106531216604,
      "learning_rate": 2.4258534205441458e-06,
      "loss": 0.888,
      "step": 34180
    },
    {
      "epoch": 2.0662355714026712,
      "grad_norm": 1.062084138335021,
      "learning_rate": 2.422998416605498e-06,
      "loss": 0.9056,
      "step": 34190
    },
    {
      "epoch": 2.066839910557805,
      "grad_norm": 1.2819693410705182,
      "learning_rate": 2.4201445563445613e-06,
      "loss": 0.8901,
      "step": 34200
    },
    {
      "epoch": 2.067444249712939,
      "grad_norm": 1.287427389826221,
      "learning_rate": 2.417291841027889e-06,
      "loss": 0.9035,
      "step": 34210
    },
    {
      "epoch": 2.0680485888680726,
      "grad_norm": 1.2114528892128853,
      "learning_rate": 2.414440271921517e-06,
      "loss": 0.9046,
      "step": 34220
    },
    {
      "epoch": 2.0686529280232064,
      "grad_norm": 1.224477279906845,
      "learning_rate": 2.4115898502909805e-06,
      "loss": 0.8811,
      "step": 34230
    },
    {
      "epoch": 2.0692572671783407,
      "grad_norm": 1.2633808837607525,
      "learning_rate": 2.408740577401298e-06,
      "loss": 0.9019,
      "step": 34240
    },
    {
      "epoch": 2.0698616063334745,
      "grad_norm": 1.4904601859270488,
      "learning_rate": 2.405892454516984e-06,
      "loss": 0.8796,
      "step": 34250
    },
    {
      "epoch": 2.0704659454886083,
      "grad_norm": 1.4584899747748776,
      "learning_rate": 2.403045482902039e-06,
      "loss": 0.8947,
      "step": 34260
    },
    {
      "epoch": 2.071070284643742,
      "grad_norm": 1.5586887362112432,
      "learning_rate": 2.4001996638199534e-06,
      "loss": 0.8916,
      "step": 34270
    },
    {
      "epoch": 2.071674623798876,
      "grad_norm": 1.5127390082420138,
      "learning_rate": 2.3973549985337092e-06,
      "loss": 0.9092,
      "step": 34280
    },
    {
      "epoch": 2.0722789629540097,
      "grad_norm": 1.4118918722534175,
      "learning_rate": 2.3945114883057707e-06,
      "loss": 0.8828,
      "step": 34290
    },
    {
      "epoch": 2.0728833021091435,
      "grad_norm": 0.942857580073176,
      "learning_rate": 2.3916691343980936e-06,
      "loss": 0.8985,
      "step": 34300
    },
    {
      "epoch": 2.0734876412642773,
      "grad_norm": 0.8738297772014342,
      "learning_rate": 2.3888279380721217e-06,
      "loss": 0.8711,
      "step": 34310
    },
    {
      "epoch": 2.0740919804194116,
      "grad_norm": 0.9276469481384416,
      "learning_rate": 2.3859879005887792e-06,
      "loss": 0.8812,
      "step": 34320
    },
    {
      "epoch": 2.0746963195745454,
      "grad_norm": 0.9057986511621184,
      "learning_rate": 2.3831490232084812e-06,
      "loss": 0.9055,
      "step": 34330
    },
    {
      "epoch": 2.075300658729679,
      "grad_norm": 0.9343820841731831,
      "learning_rate": 2.3803113071911283e-06,
      "loss": 0.9118,
      "step": 34340
    },
    {
      "epoch": 2.075904997884813,
      "grad_norm": 0.8443879267785902,
      "learning_rate": 2.3774747537960995e-06,
      "loss": 0.8846,
      "step": 34350
    },
    {
      "epoch": 2.076509337039947,
      "grad_norm": 0.8045289642936068,
      "learning_rate": 2.374639364282265e-06,
      "loss": 0.8741,
      "step": 34360
    },
    {
      "epoch": 2.0771136761950806,
      "grad_norm": 0.8861799181137933,
      "learning_rate": 2.3718051399079765e-06,
      "loss": 0.9018,
      "step": 34370
    },
    {
      "epoch": 2.0777180153502144,
      "grad_norm": 0.8179872100872368,
      "learning_rate": 2.368972081931064e-06,
      "loss": 0.8882,
      "step": 34380
    },
    {
      "epoch": 2.078322354505348,
      "grad_norm": 0.8141840998347964,
      "learning_rate": 2.3661401916088463e-06,
      "loss": 0.882,
      "step": 34390
    },
    {
      "epoch": 2.0789266936604824,
      "grad_norm": 0.8730579412882683,
      "learning_rate": 2.3633094701981195e-06,
      "loss": 0.8807,
      "step": 34400
    },
    {
      "epoch": 2.0795310328156162,
      "grad_norm": 0.8680190170134608,
      "learning_rate": 2.360479918955166e-06,
      "loss": 0.9077,
      "step": 34410
    },
    {
      "epoch": 2.08013537197075,
      "grad_norm": 0.8948758943263898,
      "learning_rate": 2.3576515391357404e-06,
      "loss": 0.8825,
      "step": 34420
    },
    {
      "epoch": 2.080739711125884,
      "grad_norm": 0.8817060767926409,
      "learning_rate": 2.354824331995085e-06,
      "loss": 0.9152,
      "step": 34430
    },
    {
      "epoch": 2.0813440502810177,
      "grad_norm": 0.8115647503714777,
      "learning_rate": 2.351998298787922e-06,
      "loss": 0.8594,
      "step": 34440
    },
    {
      "epoch": 2.0819483894361515,
      "grad_norm": 0.6287151130532005,
      "learning_rate": 2.3491734407684445e-06,
      "loss": 0.8864,
      "step": 34450
    },
    {
      "epoch": 2.0825527285912853,
      "grad_norm": 0.6523360667762224,
      "learning_rate": 2.346349759190332e-06,
      "loss": 0.8833,
      "step": 34460
    },
    {
      "epoch": 2.083157067746419,
      "grad_norm": 0.6381124356689684,
      "learning_rate": 2.343527255306741e-06,
      "loss": 0.8829,
      "step": 34470
    },
    {
      "epoch": 2.0837614069015533,
      "grad_norm": 0.609362180472817,
      "learning_rate": 2.340705930370302e-06,
      "loss": 0.8946,
      "step": 34480
    },
    {
      "epoch": 2.084365746056687,
      "grad_norm": 0.6220008590249745,
      "learning_rate": 2.3378857856331212e-06,
      "loss": 0.866,
      "step": 34490
    },
    {
      "epoch": 2.084970085211821,
      "grad_norm": 0.6599735606133348,
      "learning_rate": 2.335066822346788e-06,
      "loss": 0.9025,
      "step": 34500
    },
    {
      "epoch": 2.0855744243669547,
      "grad_norm": 0.6549383329723182,
      "learning_rate": 2.332249041762359e-06,
      "loss": 0.88,
      "step": 34510
    },
    {
      "epoch": 2.0861787635220885,
      "grad_norm": 0.6316181413129887,
      "learning_rate": 2.3294324451303723e-06,
      "loss": 0.8542,
      "step": 34520
    },
    {
      "epoch": 2.0867831026772223,
      "grad_norm": 0.7063560196363463,
      "learning_rate": 2.32661703370084e-06,
      "loss": 0.8755,
      "step": 34530
    },
    {
      "epoch": 2.087387441832356,
      "grad_norm": 0.6888164321034947,
      "learning_rate": 2.323802808723242e-06,
      "loss": 0.8887,
      "step": 34540
    },
    {
      "epoch": 2.0879917809874904,
      "grad_norm": 0.8676496105652605,
      "learning_rate": 2.3209897714465394e-06,
      "loss": 0.8867,
      "step": 34550
    },
    {
      "epoch": 2.088596120142624,
      "grad_norm": 0.8836167392541048,
      "learning_rate": 2.318177923119164e-06,
      "loss": 0.872,
      "step": 34560
    },
    {
      "epoch": 2.089200459297758,
      "grad_norm": 0.8751056678380355,
      "learning_rate": 2.3153672649890157e-06,
      "loss": 0.8689,
      "step": 34570
    },
    {
      "epoch": 2.089804798452892,
      "grad_norm": 0.8946280319058736,
      "learning_rate": 2.312557798303471e-06,
      "loss": 0.886,
      "step": 34580
    },
    {
      "epoch": 2.0904091376080256,
      "grad_norm": 0.8520897938908295,
      "learning_rate": 2.3097495243093777e-06,
      "loss": 0.8828,
      "step": 34590
    },
    {
      "epoch": 2.0910134767631594,
      "grad_norm": 0.5982091311112707,
      "learning_rate": 2.3069424442530487e-06,
      "loss": 0.906,
      "step": 34600
    },
    {
      "epoch": 2.091617815918293,
      "grad_norm": 0.584204158757196,
      "learning_rate": 2.304136559380273e-06,
      "loss": 0.8645,
      "step": 34610
    },
    {
      "epoch": 2.092222155073427,
      "grad_norm": 0.6134684130382023,
      "learning_rate": 2.30133187093631e-06,
      "loss": 0.8838,
      "step": 34620
    },
    {
      "epoch": 2.0928264942285613,
      "grad_norm": 0.5829245271987381,
      "learning_rate": 2.2985283801658797e-06,
      "loss": 0.904,
      "step": 34630
    },
    {
      "epoch": 2.093430833383695,
      "grad_norm": 0.6092198710457203,
      "learning_rate": 2.2957260883131795e-06,
      "loss": 0.8722,
      "step": 34640
    },
    {
      "epoch": 2.094035172538829,
      "grad_norm": 0.5747940891469823,
      "learning_rate": 2.292924996621872e-06,
      "loss": 0.8808,
      "step": 34650
    },
    {
      "epoch": 2.0946395116939627,
      "grad_norm": 0.6177267529143505,
      "learning_rate": 2.2901251063350834e-06,
      "loss": 0.8722,
      "step": 34660
    },
    {
      "epoch": 2.0952438508490965,
      "grad_norm": 0.5893416430517816,
      "learning_rate": 2.2873264186954115e-06,
      "loss": 0.887,
      "step": 34670
    },
    {
      "epoch": 2.0958481900042303,
      "grad_norm": 0.6093742520217635,
      "learning_rate": 2.2845289349449208e-06,
      "loss": 0.8737,
      "step": 34680
    },
    {
      "epoch": 2.096452529159364,
      "grad_norm": 0.5943031808818438,
      "learning_rate": 2.2817326563251353e-06,
      "loss": 0.8789,
      "step": 34690
    },
    {
      "epoch": 2.097056868314498,
      "grad_norm": 0.5829180628244706,
      "learning_rate": 2.2789375840770502e-06,
      "loss": 0.887,
      "step": 34700
    },
    {
      "epoch": 2.097661207469632,
      "grad_norm": 0.5635717878567686,
      "learning_rate": 2.2761437194411233e-06,
      "loss": 0.8804,
      "step": 34710
    },
    {
      "epoch": 2.098265546624766,
      "grad_norm": 0.5653942999983343,
      "learning_rate": 2.2733510636572787e-06,
      "loss": 0.8903,
      "step": 34720
    },
    {
      "epoch": 2.0988698857798997,
      "grad_norm": 0.6849553168792376,
      "learning_rate": 2.2705596179648974e-06,
      "loss": 0.8962,
      "step": 34730
    },
    {
      "epoch": 2.0994742249350336,
      "grad_norm": 0.5926090645034051,
      "learning_rate": 2.2677693836028326e-06,
      "loss": 0.8993,
      "step": 34740
    },
    {
      "epoch": 2.1000785640901674,
      "grad_norm": 0.6006080839306511,
      "learning_rate": 2.26498036180939e-06,
      "loss": 0.8497,
      "step": 34750
    },
    {
      "epoch": 2.100682903245301,
      "grad_norm": 0.6345567413536951,
      "learning_rate": 2.262192553822347e-06,
      "loss": 0.8802,
      "step": 34760
    },
    {
      "epoch": 2.101287242400435,
      "grad_norm": 0.5882297662846905,
      "learning_rate": 2.259405960878934e-06,
      "loss": 0.9012,
      "step": 34770
    },
    {
      "epoch": 2.101891581555569,
      "grad_norm": 0.5951227148797412,
      "learning_rate": 2.2566205842158484e-06,
      "loss": 0.8796,
      "step": 34780
    },
    {
      "epoch": 2.102495920710703,
      "grad_norm": 0.5980703035207874,
      "learning_rate": 2.253836425069242e-06,
      "loss": 0.8707,
      "step": 34790
    },
    {
      "epoch": 2.103100259865837,
      "grad_norm": 0.5884591143374369,
      "learning_rate": 2.251053484674732e-06,
      "loss": 0.8852,
      "step": 34800
    },
    {
      "epoch": 2.1037045990209706,
      "grad_norm": 0.600476775170353,
      "learning_rate": 2.2482717642673924e-06,
      "loss": 0.8713,
      "step": 34810
    },
    {
      "epoch": 2.1043089381761044,
      "grad_norm": 0.603379914327248,
      "learning_rate": 2.245491265081752e-06,
      "loss": 0.8797,
      "step": 34820
    },
    {
      "epoch": 2.1049132773312382,
      "grad_norm": 0.5745195459892538,
      "learning_rate": 2.242711988351805e-06,
      "loss": 0.9032,
      "step": 34830
    },
    {
      "epoch": 2.105517616486372,
      "grad_norm": 0.5580419534985216,
      "learning_rate": 2.239933935310998e-06,
      "loss": 0.8726,
      "step": 34840
    },
    {
      "epoch": 2.106121955641506,
      "grad_norm": 0.5902441602622466,
      "learning_rate": 2.237157107192235e-06,
      "loss": 0.8939,
      "step": 34850
    },
    {
      "epoch": 2.10672629479664,
      "grad_norm": 0.6086353032737659,
      "learning_rate": 2.2343815052278767e-06,
      "loss": 0.88,
      "step": 34860
    },
    {
      "epoch": 2.107330633951774,
      "grad_norm": 0.6058424196575252,
      "learning_rate": 2.2316071306497426e-06,
      "loss": 0.88,
      "step": 34870
    },
    {
      "epoch": 2.1079349731069077,
      "grad_norm": 0.6119039244035755,
      "learning_rate": 2.228833984689101e-06,
      "loss": 0.9177,
      "step": 34880
    },
    {
      "epoch": 2.1085393122620415,
      "grad_norm": 0.5851167193550637,
      "learning_rate": 2.2260620685766814e-06,
      "loss": 0.8856,
      "step": 34890
    },
    {
      "epoch": 2.1091436514171753,
      "grad_norm": 0.6488738094814008,
      "learning_rate": 2.2232913835426658e-06,
      "loss": 0.9006,
      "step": 34900
    },
    {
      "epoch": 2.109747990572309,
      "grad_norm": 0.7054329810556527,
      "learning_rate": 2.2205219308166858e-06,
      "loss": 0.8914,
      "step": 34910
    },
    {
      "epoch": 2.110352329727443,
      "grad_norm": 0.6907042944017625,
      "learning_rate": 2.2177537116278306e-06,
      "loss": 0.8717,
      "step": 34920
    },
    {
      "epoch": 2.1109566688825767,
      "grad_norm": 0.6617661198131073,
      "learning_rate": 2.214986727204644e-06,
      "loss": 0.8978,
      "step": 34930
    },
    {
      "epoch": 2.111561008037711,
      "grad_norm": 0.7010738125662763,
      "learning_rate": 2.2122209787751125e-06,
      "loss": 0.8847,
      "step": 34940
    },
    {
      "epoch": 2.1121653471928448,
      "grad_norm": 0.7493688535112966,
      "learning_rate": 2.2094564675666836e-06,
      "loss": 0.8846,
      "step": 34950
    },
    {
      "epoch": 2.1127696863479786,
      "grad_norm": 0.7392628770817403,
      "learning_rate": 2.206693194806253e-06,
      "loss": 0.8885,
      "step": 34960
    },
    {
      "epoch": 2.1133740255031124,
      "grad_norm": 0.7175873337179419,
      "learning_rate": 2.2039311617201627e-06,
      "loss": 0.8825,
      "step": 34970
    },
    {
      "epoch": 2.113978364658246,
      "grad_norm": 0.7198687173415584,
      "learning_rate": 2.201170369534209e-06,
      "loss": 0.8987,
      "step": 34980
    },
    {
      "epoch": 2.11458270381338,
      "grad_norm": 0.7177763240792459,
      "learning_rate": 2.198410819473638e-06,
      "loss": 0.8855,
      "step": 34990
    },
    {
      "epoch": 2.115187042968514,
      "grad_norm": 0.6489087987318607,
      "learning_rate": 2.19565251276314e-06,
      "loss": 0.8662,
      "step": 35000
    },
    {
      "epoch": 2.1157913821236476,
      "grad_norm": 0.6837499991695534,
      "learning_rate": 2.1928954506268595e-06,
      "loss": 0.8978,
      "step": 35010
    },
    {
      "epoch": 2.116395721278782,
      "grad_norm": 0.7071408791622961,
      "learning_rate": 2.1901396342883813e-06,
      "loss": 0.8871,
      "step": 35020
    },
    {
      "epoch": 2.1170000604339156,
      "grad_norm": 0.6910637024332569,
      "learning_rate": 2.1873850649707455e-06,
      "loss": 0.897,
      "step": 35030
    },
    {
      "epoch": 2.1176043995890494,
      "grad_norm": 0.6640508129953245,
      "learning_rate": 2.1846317438964305e-06,
      "loss": 0.9004,
      "step": 35040
    },
    {
      "epoch": 2.1182087387441833,
      "grad_norm": 0.8775869764246423,
      "learning_rate": 2.1818796722873675e-06,
      "loss": 0.8889,
      "step": 35050
    },
    {
      "epoch": 2.118813077899317,
      "grad_norm": 0.8289244115055131,
      "learning_rate": 2.1791288513649316e-06,
      "loss": 0.8961,
      "step": 35060
    },
    {
      "epoch": 2.119417417054451,
      "grad_norm": 0.8565348925654535,
      "learning_rate": 2.176379282349939e-06,
      "loss": 0.9148,
      "step": 35070
    },
    {
      "epoch": 2.1200217562095847,
      "grad_norm": 0.858558783533691,
      "learning_rate": 2.173630966462654e-06,
      "loss": 0.8817,
      "step": 35080
    },
    {
      "epoch": 2.1206260953647185,
      "grad_norm": 0.8202094397338572,
      "learning_rate": 2.170883904922787e-06,
      "loss": 0.8831,
      "step": 35090
    },
    {
      "epoch": 2.1212304345198527,
      "grad_norm": 0.8272451174395579,
      "learning_rate": 2.1681380989494845e-06,
      "loss": 0.8676,
      "step": 35100
    },
    {
      "epoch": 2.1218347736749865,
      "grad_norm": 0.7990505840913317,
      "learning_rate": 2.1653935497613406e-06,
      "loss": 0.8951,
      "step": 35110
    },
    {
      "epoch": 2.1224391128301203,
      "grad_norm": 0.8059993974861749,
      "learning_rate": 2.1626502585763935e-06,
      "loss": 0.8892,
      "step": 35120
    },
    {
      "epoch": 2.123043451985254,
      "grad_norm": 0.8208193456155347,
      "learning_rate": 2.159908226612118e-06,
      "loss": 0.8762,
      "step": 35130
    },
    {
      "epoch": 2.123647791140388,
      "grad_norm": 0.8260571564340111,
      "learning_rate": 2.1571674550854327e-06,
      "loss": 0.8734,
      "step": 35140
    },
    {
      "epoch": 2.1242521302955217,
      "grad_norm": 1.188727260442727,
      "learning_rate": 2.1544279452126994e-06,
      "loss": 0.8925,
      "step": 35150
    },
    {
      "epoch": 2.1248564694506555,
      "grad_norm": 1.2985667446071452,
      "learning_rate": 2.151689698209713e-06,
      "loss": 0.8922,
      "step": 35160
    },
    {
      "epoch": 2.12546080860579,
      "grad_norm": 1.2490945899209687,
      "learning_rate": 2.148952715291715e-06,
      "loss": 0.9058,
      "step": 35170
    },
    {
      "epoch": 2.1260651477609236,
      "grad_norm": 1.252217327619332,
      "learning_rate": 2.1462169976733844e-06,
      "loss": 0.8681,
      "step": 35180
    },
    {
      "epoch": 2.1266694869160574,
      "grad_norm": 1.292078563896392,
      "learning_rate": 2.143482546568833e-06,
      "loss": 0.8904,
      "step": 35190
    },
    {
      "epoch": 2.127273826071191,
      "grad_norm": 1.0783379426951425,
      "learning_rate": 2.140749363191618e-06,
      "loss": 0.9092,
      "step": 35200
    },
    {
      "epoch": 2.127878165226325,
      "grad_norm": 1.1343899003506481,
      "learning_rate": 2.138017448754732e-06,
      "loss": 0.8855,
      "step": 35210
    },
    {
      "epoch": 2.128482504381459,
      "grad_norm": 1.148027935851704,
      "learning_rate": 2.1352868044705998e-06,
      "loss": 0.8948,
      "step": 35220
    },
    {
      "epoch": 2.1290868435365926,
      "grad_norm": 1.037062192508951,
      "learning_rate": 2.1325574315510877e-06,
      "loss": 0.9043,
      "step": 35230
    },
    {
      "epoch": 2.1296911826917264,
      "grad_norm": 1.0306703816982747,
      "learning_rate": 2.129829331207498e-06,
      "loss": 0.8828,
      "step": 35240
    },
    {
      "epoch": 2.1302955218468607,
      "grad_norm": 2.443198515554324,
      "learning_rate": 2.127102504650563e-06,
      "loss": 0.8912,
      "step": 35250
    },
    {
      "epoch": 2.1308998610019945,
      "grad_norm": 2.471623746885376,
      "learning_rate": 2.1243769530904564e-06,
      "loss": 0.9089,
      "step": 35260
    },
    {
      "epoch": 2.1315042001571283,
      "grad_norm": 2.4917903575177496,
      "learning_rate": 2.1216526777367778e-06,
      "loss": 0.8595,
      "step": 35270
    },
    {
      "epoch": 2.132108539312262,
      "grad_norm": 2.352940743333999,
      "learning_rate": 2.1189296797985707e-06,
      "loss": 0.8812,
      "step": 35280
    },
    {
      "epoch": 2.132712878467396,
      "grad_norm": 2.5801568056401427,
      "learning_rate": 2.1162079604843023e-06,
      "loss": 0.8844,
      "step": 35290
    },
    {
      "epoch": 2.1333172176225297,
      "grad_norm": 0.9694831430766778,
      "learning_rate": 2.113487521001879e-06,
      "loss": 0.8973,
      "step": 35300
    },
    {
      "epoch": 2.1339215567776635,
      "grad_norm": 0.9794414191024234,
      "learning_rate": 2.1107683625586335e-06,
      "loss": 0.89,
      "step": 35310
    },
    {
      "epoch": 2.1345258959327973,
      "grad_norm": 1.0859209140806365,
      "learning_rate": 2.1080504863613345e-06,
      "loss": 0.8736,
      "step": 35320
    },
    {
      "epoch": 2.1351302350879315,
      "grad_norm": 0.9369050892777468,
      "learning_rate": 2.105333893616182e-06,
      "loss": 0.8775,
      "step": 35330
    },
    {
      "epoch": 2.1357345742430653,
      "grad_norm": 0.9046471684755009,
      "learning_rate": 2.102618585528802e-06,
      "loss": 0.8821,
      "step": 35340
    },
    {
      "epoch": 2.136338913398199,
      "grad_norm": 0.979791704621407,
      "learning_rate": 2.0999045633042527e-06,
      "loss": 0.8746,
      "step": 35350
    },
    {
      "epoch": 2.136943252553333,
      "grad_norm": 1.0094618054026403,
      "learning_rate": 2.097191828147024e-06,
      "loss": 0.8865,
      "step": 35360
    },
    {
      "epoch": 2.1375475917084668,
      "grad_norm": 1.0224876271721504,
      "learning_rate": 2.094480381261031e-06,
      "loss": 0.8979,
      "step": 35370
    },
    {
      "epoch": 2.1381519308636006,
      "grad_norm": 1.0160810745856237,
      "learning_rate": 2.0917702238496174e-06,
      "loss": 0.8887,
      "step": 35380
    },
    {
      "epoch": 2.1387562700187344,
      "grad_norm": 1.0269846979152855,
      "learning_rate": 2.0890613571155577e-06,
      "loss": 0.892,
      "step": 35390
    },
    {
      "epoch": 2.1393606091738686,
      "grad_norm": 1.0152712709937863,
      "learning_rate": 2.0863537822610525e-06,
      "loss": 0.9,
      "step": 35400
    },
    {
      "epoch": 2.1399649483290024,
      "grad_norm": 1.0511487372225052,
      "learning_rate": 2.0836475004877243e-06,
      "loss": 0.8701,
      "step": 35410
    },
    {
      "epoch": 2.140569287484136,
      "grad_norm": 1.2066776744248506,
      "learning_rate": 2.080942512996628e-06,
      "loss": 0.8761,
      "step": 35420
    },
    {
      "epoch": 2.14117362663927,
      "grad_norm": 1.0498445939232066,
      "learning_rate": 2.0782388209882436e-06,
      "loss": 0.8938,
      "step": 35430
    },
    {
      "epoch": 2.141777965794404,
      "grad_norm": 1.0460228250232422,
      "learning_rate": 2.0755364256624697e-06,
      "loss": 0.9017,
      "step": 35440
    },
    {
      "epoch": 2.1423823049495376,
      "grad_norm": 1.102782905117123,
      "learning_rate": 2.072835328218637e-06,
      "loss": 0.8755,
      "step": 35450
    },
    {
      "epoch": 2.1429866441046714,
      "grad_norm": 1.0861099682857072,
      "learning_rate": 2.070135529855498e-06,
      "loss": 0.8727,
      "step": 35460
    },
    {
      "epoch": 2.1435909832598052,
      "grad_norm": 1.045015745250695,
      "learning_rate": 2.067437031771225e-06,
      "loss": 0.9005,
      "step": 35470
    },
    {
      "epoch": 2.144195322414939,
      "grad_norm": 1.1578892261007294,
      "learning_rate": 2.0647398351634173e-06,
      "loss": 0.8935,
      "step": 35480
    },
    {
      "epoch": 2.1447996615700733,
      "grad_norm": 1.1193433874703707,
      "learning_rate": 2.062043941229097e-06,
      "loss": 0.8957,
      "step": 35490
    },
    {
      "epoch": 2.145404000725207,
      "grad_norm": 1.2570433934283072,
      "learning_rate": 2.0593493511647033e-06,
      "loss": 0.9045,
      "step": 35500
    },
    {
      "epoch": 2.146008339880341,
      "grad_norm": 1.189471684351895,
      "learning_rate": 2.0566560661661034e-06,
      "loss": 0.9022,
      "step": 35510
    },
    {
      "epoch": 2.1466126790354747,
      "grad_norm": 1.1179705995987261,
      "learning_rate": 2.0539640874285783e-06,
      "loss": 0.8733,
      "step": 35520
    },
    {
      "epoch": 2.1472170181906085,
      "grad_norm": 1.2253792804387909,
      "learning_rate": 2.051273416146836e-06,
      "loss": 0.9032,
      "step": 35530
    },
    {
      "epoch": 2.1478213573457423,
      "grad_norm": 1.1463727472970437,
      "learning_rate": 2.048584053514998e-06,
      "loss": 0.8923,
      "step": 35540
    },
    {
      "epoch": 2.148425696500876,
      "grad_norm": 1.011497982764017,
      "learning_rate": 2.045896000726611e-06,
      "loss": 0.8925,
      "step": 35550
    },
    {
      "epoch": 2.1490300356560104,
      "grad_norm": 1.0391118031559496,
      "learning_rate": 2.0432092589746343e-06,
      "loss": 0.8798,
      "step": 35560
    },
    {
      "epoch": 2.149634374811144,
      "grad_norm": 1.0260494795038764,
      "learning_rate": 2.04052382945145e-06,
      "loss": 0.8806,
      "step": 35570
    },
    {
      "epoch": 2.150238713966278,
      "grad_norm": 1.0343086600007603,
      "learning_rate": 2.0378397133488583e-06,
      "loss": 0.861,
      "step": 35580
    },
    {
      "epoch": 2.1508430531214118,
      "grad_norm": 0.9777778462235185,
      "learning_rate": 2.0351569118580707e-06,
      "loss": 0.8888,
      "step": 35590
    },
    {
      "epoch": 2.1514473922765456,
      "grad_norm": 1.0429594046718926,
      "learning_rate": 2.032475426169721e-06,
      "loss": 0.8786,
      "step": 35600
    },
    {
      "epoch": 2.1520517314316794,
      "grad_norm": 1.0806821193356724,
      "learning_rate": 2.0297952574738587e-06,
      "loss": 0.8897,
      "step": 35610
    },
    {
      "epoch": 2.152656070586813,
      "grad_norm": 1.0914073512978002,
      "learning_rate": 2.0271164069599446e-06,
      "loss": 0.8781,
      "step": 35620
    },
    {
      "epoch": 2.153260409741947,
      "grad_norm": 1.1159188438262881,
      "learning_rate": 2.024438875816858e-06,
      "loss": 0.8784,
      "step": 35630
    },
    {
      "epoch": 2.1538647488970812,
      "grad_norm": 1.017125102181146,
      "learning_rate": 2.0217626652328937e-06,
      "loss": 0.8812,
      "step": 35640
    },
    {
      "epoch": 2.154469088052215,
      "grad_norm": 1.0499298551121643,
      "learning_rate": 2.019087776395756e-06,
      "loss": 0.8917,
      "step": 35650
    },
    {
      "epoch": 2.155073427207349,
      "grad_norm": 1.087448243552565,
      "learning_rate": 2.016414210492565e-06,
      "loss": 0.8841,
      "step": 35660
    },
    {
      "epoch": 2.1556777663624827,
      "grad_norm": 1.0237217579134834,
      "learning_rate": 2.013741968709857e-06,
      "loss": 0.9002,
      "step": 35670
    },
    {
      "epoch": 2.1562821055176165,
      "grad_norm": 1.0033457295406665,
      "learning_rate": 2.0110710522335737e-06,
      "loss": 0.8871,
      "step": 35680
    },
    {
      "epoch": 2.1568864446727503,
      "grad_norm": 1.0562248760644997,
      "learning_rate": 2.008401462249074e-06,
      "loss": 0.8781,
      "step": 35690
    },
    {
      "epoch": 2.157490783827884,
      "grad_norm": 1.1886738581592653,
      "learning_rate": 2.0057331999411277e-06,
      "loss": 0.8887,
      "step": 35700
    },
    {
      "epoch": 2.158095122983018,
      "grad_norm": 1.2908916142027507,
      "learning_rate": 2.0030662664939123e-06,
      "loss": 0.8745,
      "step": 35710
    },
    {
      "epoch": 2.158699462138152,
      "grad_norm": 1.236655812966903,
      "learning_rate": 2.000400663091018e-06,
      "loss": 0.8862,
      "step": 35720
    },
    {
      "epoch": 2.159303801293286,
      "grad_norm": 1.2233466413012823,
      "learning_rate": 1.997736390915444e-06,
      "loss": 0.8987,
      "step": 35730
    },
    {
      "epoch": 2.1599081404484197,
      "grad_norm": 1.269474755309957,
      "learning_rate": 1.995073451149601e-06,
      "loss": 0.8662,
      "step": 35740
    },
    {
      "epoch": 2.1605124796035535,
      "grad_norm": 1.1562498483985872,
      "learning_rate": 1.992411844975303e-06,
      "loss": 0.8848,
      "step": 35750
    },
    {
      "epoch": 2.1611168187586873,
      "grad_norm": 1.0306184405875465,
      "learning_rate": 1.9897515735737776e-06,
      "loss": 0.8704,
      "step": 35760
    },
    {
      "epoch": 2.161721157913821,
      "grad_norm": 1.185211482651265,
      "learning_rate": 1.9870926381256563e-06,
      "loss": 0.9017,
      "step": 35770
    },
    {
      "epoch": 2.162325497068955,
      "grad_norm": 1.1722510370499983,
      "learning_rate": 1.9844350398109813e-06,
      "loss": 0.8911,
      "step": 35780
    },
    {
      "epoch": 2.162929836224089,
      "grad_norm": 1.0806332237913934,
      "learning_rate": 1.981778779809196e-06,
      "loss": 0.9087,
      "step": 35790
    },
    {
      "epoch": 2.163534175379223,
      "grad_norm": 1.29751618661275,
      "learning_rate": 1.979123859299157e-06,
      "loss": 0.9006,
      "step": 35800
    },
    {
      "epoch": 2.164138514534357,
      "grad_norm": 1.2403192627717603,
      "learning_rate": 1.976470279459119e-06,
      "loss": 0.9099,
      "step": 35810
    },
    {
      "epoch": 2.1647428536894906,
      "grad_norm": 1.2601712753650698,
      "learning_rate": 1.9738180414667462e-06,
      "loss": 0.8912,
      "step": 35820
    },
    {
      "epoch": 2.1653471928446244,
      "grad_norm": 1.2293630561367561,
      "learning_rate": 1.9711671464991094e-06,
      "loss": 0.8708,
      "step": 35830
    },
    {
      "epoch": 2.165951531999758,
      "grad_norm": 1.3115604141213448,
      "learning_rate": 1.9685175957326757e-06,
      "loss": 0.8968,
      "step": 35840
    },
    {
      "epoch": 2.166555871154892,
      "grad_norm": 1.3863014066553783,
      "learning_rate": 1.965869390343323e-06,
      "loss": 0.8795,
      "step": 35850
    },
    {
      "epoch": 2.167160210310026,
      "grad_norm": 1.5049387392185003,
      "learning_rate": 1.9632225315063302e-06,
      "loss": 0.9095,
      "step": 35860
    },
    {
      "epoch": 2.1677645494651596,
      "grad_norm": 1.5260768768434285,
      "learning_rate": 1.9605770203963754e-06,
      "loss": 0.8952,
      "step": 35870
    },
    {
      "epoch": 2.168368888620294,
      "grad_norm": 1.4635600908834379,
      "learning_rate": 1.957932858187541e-06,
      "loss": 0.8764,
      "step": 35880
    },
    {
      "epoch": 2.1689732277754277,
      "grad_norm": 1.5583928916014775,
      "learning_rate": 1.955290046053314e-06,
      "loss": 0.8696,
      "step": 35890
    },
    {
      "epoch": 2.1695775669305615,
      "grad_norm": 0.9236971476454152,
      "learning_rate": 1.9526485851665755e-06,
      "loss": 0.8734,
      "step": 35900
    },
    {
      "epoch": 2.1701819060856953,
      "grad_norm": 0.9411413085299765,
      "learning_rate": 1.950008476699611e-06,
      "loss": 0.904,
      "step": 35910
    },
    {
      "epoch": 2.170786245240829,
      "grad_norm": 0.896966845381397,
      "learning_rate": 1.947369721824108e-06,
      "loss": 0.8965,
      "step": 35920
    },
    {
      "epoch": 2.171390584395963,
      "grad_norm": 0.9013676365513028,
      "learning_rate": 1.9447323217111456e-06,
      "loss": 0.8783,
      "step": 35930
    },
    {
      "epoch": 2.1719949235510967,
      "grad_norm": 0.9459254683936665,
      "learning_rate": 1.9420962775312095e-06,
      "loss": 0.8958,
      "step": 35940
    },
    {
      "epoch": 2.172599262706231,
      "grad_norm": 0.8271281170881429,
      "learning_rate": 1.939461590454182e-06,
      "loss": 0.8932,
      "step": 35950
    },
    {
      "epoch": 2.1732036018613647,
      "grad_norm": 0.8605198446006623,
      "learning_rate": 1.9368282616493377e-06,
      "loss": 0.8852,
      "step": 35960
    },
    {
      "epoch": 2.1738079410164985,
      "grad_norm": 0.8469156772830047,
      "learning_rate": 1.9341962922853554e-06,
      "loss": 0.8766,
      "step": 35970
    },
    {
      "epoch": 2.1744122801716324,
      "grad_norm": 0.8252595662449176,
      "learning_rate": 1.9315656835303074e-06,
      "loss": 0.8877,
      "step": 35980
    },
    {
      "epoch": 2.175016619326766,
      "grad_norm": 0.8067127772748928,
      "learning_rate": 1.928936436551661e-06,
      "loss": 0.887,
      "step": 35990
    },
    {
      "epoch": 2.1756209584819,
      "grad_norm": 0.9187445385367577,
      "learning_rate": 1.926308552516281e-06,
      "loss": 0.8866,
      "step": 36000
    },
    {
      "epoch": 2.1762252976370338,
      "grad_norm": 0.8328957201464536,
      "learning_rate": 1.9236820325904293e-06,
      "loss": 0.8924,
      "step": 36010
    },
    {
      "epoch": 2.1768296367921676,
      "grad_norm": 0.8257160681123686,
      "learning_rate": 1.9210568779397554e-06,
      "loss": 0.8867,
      "step": 36020
    },
    {
      "epoch": 2.177433975947302,
      "grad_norm": 0.8645447246584572,
      "learning_rate": 1.9184330897293117e-06,
      "loss": 0.871,
      "step": 36030
    },
    {
      "epoch": 2.1780383151024356,
      "grad_norm": 0.8949221317083683,
      "learning_rate": 1.9158106691235368e-06,
      "loss": 0.8901,
      "step": 36040
    },
    {
      "epoch": 2.1786426542575694,
      "grad_norm": 0.6738737702382099,
      "learning_rate": 1.913189617286268e-06,
      "loss": 0.8578,
      "step": 36050
    },
    {
      "epoch": 2.1792469934127032,
      "grad_norm": 0.606883174927583,
      "learning_rate": 1.9105699353807293e-06,
      "loss": 0.8807,
      "step": 36060
    },
    {
      "epoch": 2.179851332567837,
      "grad_norm": 0.6781936866183955,
      "learning_rate": 1.9079516245695425e-06,
      "loss": 0.8804,
      "step": 36070
    },
    {
      "epoch": 2.180455671722971,
      "grad_norm": 0.6492750059631206,
      "learning_rate": 1.9053346860147193e-06,
      "loss": 0.8844,
      "step": 36080
    },
    {
      "epoch": 2.1810600108781046,
      "grad_norm": 0.6279252365467862,
      "learning_rate": 1.9027191208776586e-06,
      "loss": 0.8936,
      "step": 36090
    },
    {
      "epoch": 2.1816643500332384,
      "grad_norm": 0.6969525667350275,
      "learning_rate": 1.9001049303191537e-06,
      "loss": 0.8891,
      "step": 36100
    },
    {
      "epoch": 2.1822686891883727,
      "grad_norm": 0.6642236673433908,
      "learning_rate": 1.8974921154993898e-06,
      "loss": 0.892,
      "step": 36110
    },
    {
      "epoch": 2.1828730283435065,
      "grad_norm": 0.7019656482815692,
      "learning_rate": 1.8948806775779337e-06,
      "loss": 0.8801,
      "step": 36120
    },
    {
      "epoch": 2.1834773674986403,
      "grad_norm": 0.6793730750545127,
      "learning_rate": 1.892270617713749e-06,
      "loss": 0.8739,
      "step": 36130
    },
    {
      "epoch": 2.184081706653774,
      "grad_norm": 0.6654197545780667,
      "learning_rate": 1.889661937065186e-06,
      "loss": 0.8927,
      "step": 36140
    },
    {
      "epoch": 2.184686045808908,
      "grad_norm": 0.8536512125544913,
      "learning_rate": 1.8870546367899773e-06,
      "loss": 0.8843,
      "step": 36150
    },
    {
      "epoch": 2.1852903849640417,
      "grad_norm": 0.9011878101054452,
      "learning_rate": 1.8844487180452502e-06,
      "loss": 0.8835,
      "step": 36160
    },
    {
      "epoch": 2.1858947241191755,
      "grad_norm": 0.9387183601169968,
      "learning_rate": 1.8818441819875172e-06,
      "loss": 0.8852,
      "step": 36170
    },
    {
      "epoch": 2.1864990632743098,
      "grad_norm": 0.9327607919395237,
      "learning_rate": 1.8792410297726726e-06,
      "loss": 0.8864,
      "step": 36180
    },
    {
      "epoch": 2.1871034024294436,
      "grad_norm": 0.8755628055035676,
      "learning_rate": 1.876639262556001e-06,
      "loss": 0.8733,
      "step": 36190
    },
    {
      "epoch": 2.1877077415845774,
      "grad_norm": 0.6338858720646463,
      "learning_rate": 1.8740388814921728e-06,
      "loss": 0.8686,
      "step": 36200
    },
    {
      "epoch": 2.188312080739711,
      "grad_norm": 0.6630273915354651,
      "learning_rate": 1.8714398877352392e-06,
      "loss": 0.8881,
      "step": 36210
    },
    {
      "epoch": 2.188916419894845,
      "grad_norm": 0.6159520225542523,
      "learning_rate": 1.8688422824386388e-06,
      "loss": 0.8839,
      "step": 36220
    },
    {
      "epoch": 2.189520759049979,
      "grad_norm": 0.6385636877440544,
      "learning_rate": 1.8662460667551958e-06,
      "loss": 0.8831,
      "step": 36230
    },
    {
      "epoch": 2.1901250982051126,
      "grad_norm": 0.6467284846128416,
      "learning_rate": 1.8636512418371105e-06,
      "loss": 0.8855,
      "step": 36240
    },
    {
      "epoch": 2.1907294373602464,
      "grad_norm": 0.5977882139510758,
      "learning_rate": 1.8610578088359731e-06,
      "loss": 0.8963,
      "step": 36250
    },
    {
      "epoch": 2.1913337765153806,
      "grad_norm": 0.6083594760850606,
      "learning_rate": 1.8584657689027558e-06,
      "loss": 0.9004,
      "step": 36260
    },
    {
      "epoch": 2.1919381156705144,
      "grad_norm": 0.6018744711703098,
      "learning_rate": 1.8558751231878064e-06,
      "loss": 0.9022,
      "step": 36270
    },
    {
      "epoch": 2.1925424548256482,
      "grad_norm": 0.564524932783717,
      "learning_rate": 1.8532858728408592e-06,
      "loss": 0.8963,
      "step": 36280
    },
    {
      "epoch": 2.193146793980782,
      "grad_norm": 0.5897282314184277,
      "learning_rate": 1.8506980190110307e-06,
      "loss": 0.8824,
      "step": 36290
    },
    {
      "epoch": 2.193751133135916,
      "grad_norm": 0.617904741544674,
      "learning_rate": 1.8481115628468105e-06,
      "loss": 0.8809,
      "step": 36300
    },
    {
      "epoch": 2.1943554722910497,
      "grad_norm": 0.6179938951967063,
      "learning_rate": 1.845526505496076e-06,
      "loss": 0.8952,
      "step": 36310
    },
    {
      "epoch": 2.1949598114461835,
      "grad_norm": 0.6379393710503313,
      "learning_rate": 1.8429428481060769e-06,
      "loss": 0.8773,
      "step": 36320
    },
    {
      "epoch": 2.1955641506013173,
      "grad_norm": 0.5857841852551722,
      "learning_rate": 1.840360591823448e-06,
      "loss": 0.903,
      "step": 36330
    },
    {
      "epoch": 2.1961684897564515,
      "grad_norm": 0.5669331872883006,
      "learning_rate": 1.8377797377941953e-06,
      "loss": 0.8837,
      "step": 36340
    },
    {
      "epoch": 2.1967728289115853,
      "grad_norm": 0.5885944120463181,
      "learning_rate": 1.8352002871637104e-06,
      "loss": 0.8986,
      "step": 36350
    },
    {
      "epoch": 2.197377168066719,
      "grad_norm": 0.5931377916438505,
      "learning_rate": 1.8326222410767536e-06,
      "loss": 0.9044,
      "step": 36360
    },
    {
      "epoch": 2.197981507221853,
      "grad_norm": 0.5748942824991078,
      "learning_rate": 1.8300456006774687e-06,
      "loss": 0.8794,
      "step": 36370
    },
    {
      "epoch": 2.1985858463769867,
      "grad_norm": 0.6105965524242724,
      "learning_rate": 1.827470367109374e-06,
      "loss": 0.9033,
      "step": 36380
    },
    {
      "epoch": 2.1991901855321205,
      "grad_norm": 0.5917392433348321,
      "learning_rate": 1.8248965415153602e-06,
      "loss": 0.8853,
      "step": 36390
    },
    {
      "epoch": 2.1997945246872543,
      "grad_norm": 0.5775980237344719,
      "learning_rate": 1.822324125037696e-06,
      "loss": 0.8947,
      "step": 36400
    },
    {
      "epoch": 2.200398863842388,
      "grad_norm": 0.5873058355473882,
      "learning_rate": 1.8197531188180256e-06,
      "loss": 0.9063,
      "step": 36410
    },
    {
      "epoch": 2.2010032029975224,
      "grad_norm": 0.5701127562582692,
      "learning_rate": 1.8171835239973667e-06,
      "loss": 0.8796,
      "step": 36420
    },
    {
      "epoch": 2.201607542152656,
      "grad_norm": 0.5825694048979243,
      "learning_rate": 1.8146153417161067e-06,
      "loss": 0.8782,
      "step": 36430
    },
    {
      "epoch": 2.20221188130779,
      "grad_norm": 0.5885962454219273,
      "learning_rate": 1.8120485731140107e-06,
      "loss": 0.8796,
      "step": 36440
    },
    {
      "epoch": 2.202816220462924,
      "grad_norm": 0.6351575672730985,
      "learning_rate": 1.8094832193302175e-06,
      "loss": 0.899,
      "step": 36450
    },
    {
      "epoch": 2.2034205596180576,
      "grad_norm": 0.6352902982756858,
      "learning_rate": 1.8069192815032311e-06,
      "loss": 0.8768,
      "step": 36460
    },
    {
      "epoch": 2.2040248987731914,
      "grad_norm": 0.6743008800834629,
      "learning_rate": 1.804356760770934e-06,
      "loss": 0.8969,
      "step": 36470
    },
    {
      "epoch": 2.204629237928325,
      "grad_norm": 0.6510438100399902,
      "learning_rate": 1.8017956582705787e-06,
      "loss": 0.8915,
      "step": 36480
    },
    {
      "epoch": 2.205233577083459,
      "grad_norm": 0.5813652587460086,
      "learning_rate": 1.7992359751387834e-06,
      "loss": 0.8706,
      "step": 36490
    },
    {
      "epoch": 2.2058379162385933,
      "grad_norm": 0.6953466438171458,
      "learning_rate": 1.7966777125115415e-06,
      "loss": 0.8721,
      "step": 36500
    },
    {
      "epoch": 2.206442255393727,
      "grad_norm": 0.6616065017926829,
      "learning_rate": 1.7941208715242165e-06,
      "loss": 0.9004,
      "step": 36510
    },
    {
      "epoch": 2.207046594548861,
      "grad_norm": 0.5938364396116497,
      "learning_rate": 1.7915654533115346e-06,
      "loss": 0.8714,
      "step": 36520
    },
    {
      "epoch": 2.2076509337039947,
      "grad_norm": 0.7179989755939891,
      "learning_rate": 1.7890114590075974e-06,
      "loss": 0.8767,
      "step": 36530
    },
    {
      "epoch": 2.2082552728591285,
      "grad_norm": 0.7358894922676781,
      "learning_rate": 1.7864588897458729e-06,
      "loss": 0.8993,
      "step": 36540
    },
    {
      "epoch": 2.2088596120142623,
      "grad_norm": 0.746367742265286,
      "learning_rate": 1.7839077466591925e-06,
      "loss": 0.8864,
      "step": 36550
    },
    {
      "epoch": 2.209463951169396,
      "grad_norm": 0.7817523231637606,
      "learning_rate": 1.7813580308797613e-06,
      "loss": 0.8861,
      "step": 36560
    },
    {
      "epoch": 2.2100682903245303,
      "grad_norm": 0.729496105777396,
      "learning_rate": 1.7788097435391444e-06,
      "loss": 0.8989,
      "step": 36570
    },
    {
      "epoch": 2.210672629479664,
      "grad_norm": 0.6852444065561719,
      "learning_rate": 1.7762628857682796e-06,
      "loss": 0.884,
      "step": 36580
    },
    {
      "epoch": 2.211276968634798,
      "grad_norm": 0.7175292497845551,
      "learning_rate": 1.7737174586974627e-06,
      "loss": 0.903,
      "step": 36590
    },
    {
      "epoch": 2.2118813077899317,
      "grad_norm": 0.6770318813166235,
      "learning_rate": 1.7711734634563621e-06,
      "loss": 0.9013,
      "step": 36600
    },
    {
      "epoch": 2.2124856469450656,
      "grad_norm": 0.6475624360900526,
      "learning_rate": 1.7686309011740044e-06,
      "loss": 0.8808,
      "step": 36610
    },
    {
      "epoch": 2.2130899861001994,
      "grad_norm": 0.6934941463311876,
      "learning_rate": 1.7660897729787841e-06,
      "loss": 0.8751,
      "step": 36620
    },
    {
      "epoch": 2.213694325255333,
      "grad_norm": 0.6828939367024258,
      "learning_rate": 1.76355007999846e-06,
      "loss": 0.9083,
      "step": 36630
    },
    {
      "epoch": 2.214298664410467,
      "grad_norm": 0.668867271778218,
      "learning_rate": 1.7610118233601497e-06,
      "loss": 0.8602,
      "step": 36640
    },
    {
      "epoch": 2.214903003565601,
      "grad_norm": 0.8615036294875504,
      "learning_rate": 1.758475004190337e-06,
      "loss": 0.8921,
      "step": 36650
    },
    {
      "epoch": 2.215507342720735,
      "grad_norm": 0.848344021406913,
      "learning_rate": 1.7559396236148678e-06,
      "loss": 0.8949,
      "step": 36660
    },
    {
      "epoch": 2.216111681875869,
      "grad_norm": 0.8447896454662474,
      "learning_rate": 1.7534056827589452e-06,
      "loss": 0.9174,
      "step": 36670
    },
    {
      "epoch": 2.2167160210310026,
      "grad_norm": 0.8611124789595431,
      "learning_rate": 1.7508731827471387e-06,
      "loss": 0.8858,
      "step": 36680
    },
    {
      "epoch": 2.2173203601861364,
      "grad_norm": 0.8484346659557226,
      "learning_rate": 1.7483421247033776e-06,
      "loss": 0.9027,
      "step": 36690
    },
    {
      "epoch": 2.2179246993412702,
      "grad_norm": 0.8012577424675125,
      "learning_rate": 1.745812509750946e-06,
      "loss": 0.8949,
      "step": 36700
    },
    {
      "epoch": 2.218529038496404,
      "grad_norm": 0.8111003175741118,
      "learning_rate": 1.7432843390124933e-06,
      "loss": 0.8841,
      "step": 36710
    },
    {
      "epoch": 2.219133377651538,
      "grad_norm": 0.8692906827599644,
      "learning_rate": 1.7407576136100278e-06,
      "loss": 0.9,
      "step": 36720
    },
    {
      "epoch": 2.219737716806672,
      "grad_norm": 0.9031370287133852,
      "learning_rate": 1.738232334664911e-06,
      "loss": 0.8948,
      "step": 36730
    },
    {
      "epoch": 2.220342055961806,
      "grad_norm": 0.8294289203178514,
      "learning_rate": 1.7357085032978676e-06,
      "loss": 0.8735,
      "step": 36740
    },
    {
      "epoch": 2.2209463951169397,
      "grad_norm": 1.232663255690472,
      "learning_rate": 1.733186120628979e-06,
      "loss": 0.8982,
      "step": 36750
    },
    {
      "epoch": 2.2215507342720735,
      "grad_norm": 1.1938677755502394,
      "learning_rate": 1.7306651877776836e-06,
      "loss": 0.8894,
      "step": 36760
    },
    {
      "epoch": 2.2221550734272073,
      "grad_norm": 1.271398043818262,
      "learning_rate": 1.7281457058627733e-06,
      "loss": 0.8972,
      "step": 36770
    },
    {
      "epoch": 2.222759412582341,
      "grad_norm": 1.231630452020186,
      "learning_rate": 1.7256276760023993e-06,
      "loss": 0.89,
      "step": 36780
    },
    {
      "epoch": 2.223363751737475,
      "grad_norm": 1.1888978954029776,
      "learning_rate": 1.7231110993140699e-06,
      "loss": 0.8711,
      "step": 36790
    },
    {
      "epoch": 2.223968090892609,
      "grad_norm": 1.0612803309642933,
      "learning_rate": 1.7205959769146424e-06,
      "loss": 0.9119,
      "step": 36800
    },
    {
      "epoch": 2.224572430047743,
      "grad_norm": 0.8424229073925843,
      "learning_rate": 1.7180823099203365e-06,
      "loss": 0.9146,
      "step": 36810
    },
    {
      "epoch": 2.2251767692028768,
      "grad_norm": 0.8622728513728528,
      "learning_rate": 1.7155700994467173e-06,
      "loss": 0.8964,
      "step": 36820
    },
    {
      "epoch": 2.2257811083580106,
      "grad_norm": 0.9560136887571115,
      "learning_rate": 1.7130593466087125e-06,
      "loss": 0.896,
      "step": 36830
    },
    {
      "epoch": 2.2263854475131444,
      "grad_norm": 0.8185085188081441,
      "learning_rate": 1.7105500525205943e-06,
      "loss": 0.8907,
      "step": 36840
    },
    {
      "epoch": 2.226989786668278,
      "grad_norm": 2.4229031740319957,
      "learning_rate": 1.7080422182959961e-06,
      "loss": 0.8731,
      "step": 36850
    },
    {
      "epoch": 2.227594125823412,
      "grad_norm": 2.7314356927294705,
      "learning_rate": 1.7055358450478948e-06,
      "loss": 0.8778,
      "step": 36860
    },
    {
      "epoch": 2.228198464978546,
      "grad_norm": 2.722034080264612,
      "learning_rate": 1.7030309338886258e-06,
      "loss": 0.8964,
      "step": 36870
    },
    {
      "epoch": 2.2288028041336796,
      "grad_norm": 2.48289751055475,
      "learning_rate": 1.7005274859298731e-06,
      "loss": 0.914,
      "step": 36880
    },
    {
      "epoch": 2.229407143288814,
      "grad_norm": 2.480870019229309,
      "learning_rate": 1.6980255022826691e-06,
      "loss": 0.8956,
      "step": 36890
    },
    {
      "epoch": 2.2300114824439476,
      "grad_norm": 0.9282967713244585,
      "learning_rate": 1.6955249840574001e-06,
      "loss": 0.879,
      "step": 36900
    },
    {
      "epoch": 2.2306158215990814,
      "grad_norm": 0.9122029164231888,
      "learning_rate": 1.6930259323638015e-06,
      "loss": 0.8966,
      "step": 36910
    },
    {
      "epoch": 2.2312201607542153,
      "grad_norm": 0.8933742193952547,
      "learning_rate": 1.6905283483109535e-06,
      "loss": 0.9075,
      "step": 36920
    },
    {
      "epoch": 2.231824499909349,
      "grad_norm": 0.9780188667155987,
      "learning_rate": 1.6880322330072897e-06,
      "loss": 0.8958,
      "step": 36930
    },
    {
      "epoch": 2.232428839064483,
      "grad_norm": 0.8602150819446924,
      "learning_rate": 1.6855375875605923e-06,
      "loss": 0.8611,
      "step": 36940
    },
    {
      "epoch": 2.2330331782196167,
      "grad_norm": 1.0075398662983217,
      "learning_rate": 1.683044413077986e-06,
      "loss": 0.8743,
      "step": 36950
    },
    {
      "epoch": 2.233637517374751,
      "grad_norm": 1.1288389264643472,
      "learning_rate": 1.6805527106659469e-06,
      "loss": 0.8961,
      "step": 36960
    },
    {
      "epoch": 2.2342418565298847,
      "grad_norm": 1.0082437861676605,
      "learning_rate": 1.6780624814302993e-06,
      "loss": 0.9121,
      "step": 36970
    },
    {
      "epoch": 2.2348461956850185,
      "grad_norm": 0.9897001307573534,
      "learning_rate": 1.675573726476208e-06,
      "loss": 0.8894,
      "step": 36980
    },
    {
      "epoch": 2.2354505348401523,
      "grad_norm": 0.9485602597674926,
      "learning_rate": 1.673086446908187e-06,
      "loss": 0.8748,
      "step": 36990
    },
    {
      "epoch": 2.236054873995286,
      "grad_norm": 1.1339738049158197,
      "learning_rate": 1.670600643830098e-06,
      "loss": 0.8893,
      "step": 37000
    },
    {
      "epoch": 2.23665921315042,
      "grad_norm": 1.0643826525869515,
      "learning_rate": 1.668116318345141e-06,
      "loss": 0.8931,
      "step": 37010
    },
    {
      "epoch": 2.2372635523055537,
      "grad_norm": 1.121287043341767,
      "learning_rate": 1.6656334715558658e-06,
      "loss": 0.8763,
      "step": 37020
    },
    {
      "epoch": 2.2378678914606875,
      "grad_norm": 1.0536902328507198,
      "learning_rate": 1.6631521045641658e-06,
      "loss": 0.8792,
      "step": 37030
    },
    {
      "epoch": 2.238472230615822,
      "grad_norm": 0.9963660692357627,
      "learning_rate": 1.660672218471272e-06,
      "loss": 0.8624,
      "step": 37040
    },
    {
      "epoch": 2.2390765697709556,
      "grad_norm": 1.1169132798114147,
      "learning_rate": 1.6581938143777643e-06,
      "loss": 0.8918,
      "step": 37050
    },
    {
      "epoch": 2.2396809089260894,
      "grad_norm": 1.0804784448940277,
      "learning_rate": 1.6557168933835643e-06,
      "loss": 0.8906,
      "step": 37060
    },
    {
      "epoch": 2.240285248081223,
      "grad_norm": 1.0086513349254966,
      "learning_rate": 1.6532414565879302e-06,
      "loss": 0.8807,
      "step": 37070
    },
    {
      "epoch": 2.240889587236357,
      "grad_norm": 1.0226198102864512,
      "learning_rate": 1.650767505089469e-06,
      "loss": 0.8878,
      "step": 37080
    },
    {
      "epoch": 2.241493926391491,
      "grad_norm": 1.0240458853836232,
      "learning_rate": 1.6482950399861203e-06,
      "loss": 0.8862,
      "step": 37090
    },
    {
      "epoch": 2.2420982655466246,
      "grad_norm": 1.1888935201367858,
      "learning_rate": 1.645824062375172e-06,
      "loss": 0.8796,
      "step": 37100
    },
    {
      "epoch": 2.2427026047017584,
      "grad_norm": 1.1268260617394485,
      "learning_rate": 1.6433545733532452e-06,
      "loss": 0.8798,
      "step": 37110
    },
    {
      "epoch": 2.2433069438568927,
      "grad_norm": 1.1905736824302668,
      "learning_rate": 1.6408865740163054e-06,
      "loss": 0.8906,
      "step": 37120
    },
    {
      "epoch": 2.2439112830120265,
      "grad_norm": 1.140063230054329,
      "learning_rate": 1.6384200654596555e-06,
      "loss": 0.8873,
      "step": 37130
    },
    {
      "epoch": 2.2445156221671603,
      "grad_norm": 1.2265198915264868,
      "learning_rate": 1.6359550487779336e-06,
      "loss": 0.8933,
      "step": 37140
    },
    {
      "epoch": 2.245119961322294,
      "grad_norm": 1.0563880127005834,
      "learning_rate": 1.63349152506512e-06,
      "loss": 0.8994,
      "step": 37150
    },
    {
      "epoch": 2.245724300477428,
      "grad_norm": 1.0037048836714033,
      "learning_rate": 1.6310294954145323e-06,
      "loss": 0.8786,
      "step": 37160
    },
    {
      "epoch": 2.2463286396325617,
      "grad_norm": 1.029701021736369,
      "learning_rate": 1.6285689609188199e-06,
      "loss": 0.8863,
      "step": 37170
    },
    {
      "epoch": 2.2469329787876955,
      "grad_norm": 1.042120107209839,
      "learning_rate": 1.6261099226699738e-06,
      "loss": 0.9061,
      "step": 37180
    },
    {
      "epoch": 2.2475373179428297,
      "grad_norm": 1.0140323104236189,
      "learning_rate": 1.623652381759322e-06,
      "loss": 0.915,
      "step": 37190
    },
    {
      "epoch": 2.2481416570979635,
      "grad_norm": 1.1034948303221233,
      "learning_rate": 1.621196339277521e-06,
      "loss": 0.8811,
      "step": 37200
    },
    {
      "epoch": 2.2487459962530973,
      "grad_norm": 1.0019590675621992,
      "learning_rate": 1.618741796314569e-06,
      "loss": 0.8914,
      "step": 37210
    },
    {
      "epoch": 2.249350335408231,
      "grad_norm": 1.1056444831702081,
      "learning_rate": 1.6162887539597977e-06,
      "loss": 0.8927,
      "step": 37220
    },
    {
      "epoch": 2.249954674563365,
      "grad_norm": 1.0650229486425582,
      "learning_rate": 1.6138372133018676e-06,
      "loss": 0.8964,
      "step": 37230
    },
    {
      "epoch": 2.2505590137184988,
      "grad_norm": 1.0637077422666117,
      "learning_rate": 1.6113871754287796e-06,
      "loss": 0.863,
      "step": 37240
    },
    {
      "epoch": 2.2511633528736326,
      "grad_norm": 1.0660076532336324,
      "learning_rate": 1.6089386414278652e-06,
      "loss": 0.8971,
      "step": 37250
    },
    {
      "epoch": 2.2517676920287664,
      "grad_norm": 1.0247836824547185,
      "learning_rate": 1.6064916123857854e-06,
      "loss": 0.8988,
      "step": 37260
    },
    {
      "epoch": 2.2523720311839,
      "grad_norm": 1.0730542327450574,
      "learning_rate": 1.6040460893885374e-06,
      "loss": 0.8575,
      "step": 37270
    },
    {
      "epoch": 2.2529763703390344,
      "grad_norm": 1.1349725238081176,
      "learning_rate": 1.6016020735214494e-06,
      "loss": 0.882,
      "step": 37280
    },
    {
      "epoch": 2.253580709494168,
      "grad_norm": 1.025046365954902,
      "learning_rate": 1.5991595658691778e-06,
      "loss": 0.8891,
      "step": 37290
    },
    {
      "epoch": 2.254185048649302,
      "grad_norm": 1.2105527844032822,
      "learning_rate": 1.596718567515712e-06,
      "loss": 0.8921,
      "step": 37300
    },
    {
      "epoch": 2.254789387804436,
      "grad_norm": 1.3201846423105932,
      "learning_rate": 1.594279079544374e-06,
      "loss": 0.879,
      "step": 37310
    },
    {
      "epoch": 2.2553937269595696,
      "grad_norm": 1.2862273845874301,
      "learning_rate": 1.5918411030378089e-06,
      "loss": 0.8921,
      "step": 37320
    },
    {
      "epoch": 2.2559980661147034,
      "grad_norm": 1.174249035134414,
      "learning_rate": 1.5894046390779982e-06,
      "loss": 0.89,
      "step": 37330
    },
    {
      "epoch": 2.2566024052698372,
      "grad_norm": 1.2148050798508878,
      "learning_rate": 1.5869696887462455e-06,
      "loss": 0.8968,
      "step": 37340
    },
    {
      "epoch": 2.2572067444249715,
      "grad_norm": 1.1568354974623272,
      "learning_rate": 1.5845362531231894e-06,
      "loss": 0.9086,
      "step": 37350
    },
    {
      "epoch": 2.2578110835801053,
      "grad_norm": 1.1810387400072275,
      "learning_rate": 1.5821043332887891e-06,
      "loss": 0.8857,
      "step": 37360
    },
    {
      "epoch": 2.258415422735239,
      "grad_norm": 1.0797959905169021,
      "learning_rate": 1.5796739303223386e-06,
      "loss": 0.8713,
      "step": 37370
    },
    {
      "epoch": 2.259019761890373,
      "grad_norm": 1.0869542518005186,
      "learning_rate": 1.5772450453024513e-06,
      "loss": 0.8874,
      "step": 37380
    },
    {
      "epoch": 2.2596241010455067,
      "grad_norm": 1.1098372976244586,
      "learning_rate": 1.574817679307072e-06,
      "loss": 0.884,
      "step": 37390
    },
    {
      "epoch": 2.2602284402006405,
      "grad_norm": 1.1942924562134536,
      "learning_rate": 1.5723918334134703e-06,
      "loss": 0.8913,
      "step": 37400
    },
    {
      "epoch": 2.2608327793557743,
      "grad_norm": 1.2174895155080612,
      "learning_rate": 1.569967508698242e-06,
      "loss": 0.8973,
      "step": 37410
    },
    {
      "epoch": 2.2614371185109086,
      "grad_norm": 1.2268678157474902,
      "learning_rate": 1.5675447062373035e-06,
      "loss": 0.8915,
      "step": 37420
    },
    {
      "epoch": 2.2620414576660424,
      "grad_norm": 1.21756793256655,
      "learning_rate": 1.5651234271059e-06,
      "loss": 0.8804,
      "step": 37430
    },
    {
      "epoch": 2.262645796821176,
      "grad_norm": 1.2738766299183828,
      "learning_rate": 1.5627036723786022e-06,
      "loss": 0.8803,
      "step": 37440
    },
    {
      "epoch": 2.26325013597631,
      "grad_norm": 1.4434433678876375,
      "learning_rate": 1.5602854431292963e-06,
      "loss": 0.8871,
      "step": 37450
    },
    {
      "epoch": 2.2638544751314438,
      "grad_norm": 1.4766335072600694,
      "learning_rate": 1.5578687404311993e-06,
      "loss": 0.8921,
      "step": 37460
    },
    {
      "epoch": 2.2644588142865776,
      "grad_norm": 1.4760296309205438,
      "learning_rate": 1.5554535653568492e-06,
      "loss": 0.871,
      "step": 37470
    },
    {
      "epoch": 2.2650631534417114,
      "grad_norm": 1.422964959497385,
      "learning_rate": 1.553039918978102e-06,
      "loss": 0.8874,
      "step": 37480
    },
    {
      "epoch": 2.265667492596845,
      "grad_norm": 1.498394552454921,
      "learning_rate": 1.5506278023661398e-06,
      "loss": 0.9058,
      "step": 37490
    },
    {
      "epoch": 2.266271831751979,
      "grad_norm": 0.9678208636011018,
      "learning_rate": 1.548217216591465e-06,
      "loss": 0.8726,
      "step": 37500
    },
    {
      "epoch": 2.2668761709071132,
      "grad_norm": 0.8876549650275329,
      "learning_rate": 1.5458081627238969e-06,
      "loss": 0.8996,
      "step": 37510
    },
    {
      "epoch": 2.267480510062247,
      "grad_norm": 0.9028933332394528,
      "learning_rate": 1.5434006418325792e-06,
      "loss": 0.8744,
      "step": 37520
    },
    {
      "epoch": 2.268084849217381,
      "grad_norm": 0.9120746950777309,
      "learning_rate": 1.5409946549859756e-06,
      "loss": 0.8988,
      "step": 37530
    },
    {
      "epoch": 2.2686891883725147,
      "grad_norm": 0.8974977166314854,
      "learning_rate": 1.538590203251864e-06,
      "loss": 0.8863,
      "step": 37540
    },
    {
      "epoch": 2.2692935275276485,
      "grad_norm": 0.8247723106141008,
      "learning_rate": 1.5361872876973454e-06,
      "loss": 0.887,
      "step": 37550
    },
    {
      "epoch": 2.2698978666827823,
      "grad_norm": 0.8506953062544385,
      "learning_rate": 1.5337859093888403e-06,
      "loss": 0.877,
      "step": 37560
    },
    {
      "epoch": 2.270502205837916,
      "grad_norm": 0.8308649285460079,
      "learning_rate": 1.5313860693920802e-06,
      "loss": 0.8833,
      "step": 37570
    },
    {
      "epoch": 2.2711065449930503,
      "grad_norm": 0.8586671539291736,
      "learning_rate": 1.5289877687721222e-06,
      "loss": 0.8823,
      "step": 37580
    },
    {
      "epoch": 2.271710884148184,
      "grad_norm": 0.8705147847147894,
      "learning_rate": 1.5265910085933323e-06,
      "loss": 0.8728,
      "step": 37590
    },
    {
      "epoch": 2.272315223303318,
      "grad_norm": 0.8386276651409899,
      "learning_rate": 1.524195789919401e-06,
      "loss": 0.8894,
      "step": 37600
    },
    {
      "epoch": 2.2729195624584517,
      "grad_norm": 0.8884548096993902,
      "learning_rate": 1.521802113813326e-06,
      "loss": 0.8633,
      "step": 37610
    },
    {
      "epoch": 2.2735239016135855,
      "grad_norm": 0.9003853829388102,
      "learning_rate": 1.519409981337429e-06,
      "loss": 0.8729,
      "step": 37620
    },
    {
      "epoch": 2.2741282407687193,
      "grad_norm": 0.8834963767649261,
      "learning_rate": 1.5170193935533389e-06,
      "loss": 0.9032,
      "step": 37630
    },
    {
      "epoch": 2.274732579923853,
      "grad_norm": 0.8235260953922231,
      "learning_rate": 1.5146303515220045e-06,
      "loss": 0.8913,
      "step": 37640
    },
    {
      "epoch": 2.275336919078987,
      "grad_norm": 0.6353030770335787,
      "learning_rate": 1.5122428563036883e-06,
      "loss": 0.8934,
      "step": 37650
    },
    {
      "epoch": 2.2759412582341207,
      "grad_norm": 0.6080089849420665,
      "learning_rate": 1.5098569089579611e-06,
      "loss": 0.8735,
      "step": 37660
    },
    {
      "epoch": 2.276545597389255,
      "grad_norm": 0.6717457742950187,
      "learning_rate": 1.5074725105437128e-06,
      "loss": 0.8612,
      "step": 37670
    },
    {
      "epoch": 2.277149936544389,
      "grad_norm": 0.674529580795632,
      "learning_rate": 1.5050896621191447e-06,
      "loss": 0.8731,
      "step": 37680
    },
    {
      "epoch": 2.2777542756995226,
      "grad_norm": 0.6385370616480971,
      "learning_rate": 1.5027083647417657e-06,
      "loss": 0.8854,
      "step": 37690
    },
    {
      "epoch": 2.2783586148546564,
      "grad_norm": 0.6640793926567452,
      "learning_rate": 1.5003286194684008e-06,
      "loss": 0.8795,
      "step": 37700
    },
    {
      "epoch": 2.27896295400979,
      "grad_norm": 0.717967984453388,
      "learning_rate": 1.4979504273551875e-06,
      "loss": 0.8733,
      "step": 37710
    },
    {
      "epoch": 2.279567293164924,
      "grad_norm": 0.6924389178321663,
      "learning_rate": 1.4955737894575683e-06,
      "loss": 0.867,
      "step": 37720
    },
    {
      "epoch": 2.280171632320058,
      "grad_norm": 0.6516681011651431,
      "learning_rate": 1.4931987068303e-06,
      "loss": 0.8627,
      "step": 37730
    },
    {
      "epoch": 2.280775971475192,
      "grad_norm": 0.6436091883452515,
      "learning_rate": 1.4908251805274487e-06,
      "loss": 0.8655,
      "step": 37740
    },
    {
      "epoch": 2.281380310630326,
      "grad_norm": 0.848333541809155,
      "learning_rate": 1.4884532116023914e-06,
      "loss": 0.9009,
      "step": 37750
    },
    {
      "epoch": 2.2819846497854597,
      "grad_norm": 0.8779172399496122,
      "learning_rate": 1.4860828011078087e-06,
      "loss": 0.9046,
      "step": 37760
    },
    {
      "epoch": 2.2825889889405935,
      "grad_norm": 0.8522566568300387,
      "learning_rate": 1.4837139500956938e-06,
      "loss": 0.887,
      "step": 37770
    },
    {
      "epoch": 2.2831933280957273,
      "grad_norm": 0.9434503850020198,
      "learning_rate": 1.4813466596173492e-06,
      "loss": 0.8911,
      "step": 37780
    },
    {
      "epoch": 2.283797667250861,
      "grad_norm": 0.9076135953936881,
      "learning_rate": 1.4789809307233789e-06,
      "loss": 0.8916,
      "step": 37790
    },
    {
      "epoch": 2.284402006405995,
      "grad_norm": 0.5705320045257928,
      "learning_rate": 1.476616764463699e-06,
      "loss": 0.8702,
      "step": 37800
    },
    {
      "epoch": 2.285006345561129,
      "grad_norm": 0.5728535275398304,
      "learning_rate": 1.4742541618875322e-06,
      "loss": 0.8806,
      "step": 37810
    },
    {
      "epoch": 2.285610684716263,
      "grad_norm": 0.5973995018392249,
      "learning_rate": 1.471893124043402e-06,
      "loss": 0.8824,
      "step": 37820
    },
    {
      "epoch": 2.2862150238713967,
      "grad_norm": 0.5630313767600951,
      "learning_rate": 1.4695336519791454e-06,
      "loss": 0.8699,
      "step": 37830
    },
    {
      "epoch": 2.2868193630265305,
      "grad_norm": 0.5959663572778476,
      "learning_rate": 1.4671757467418956e-06,
      "loss": 0.88,
      "step": 37840
    },
    {
      "epoch": 2.2874237021816644,
      "grad_norm": 0.5726324475985635,
      "learning_rate": 1.4648194093780987e-06,
      "loss": 0.8905,
      "step": 37850
    },
    {
      "epoch": 2.288028041336798,
      "grad_norm": 0.5698584527590744,
      "learning_rate": 1.462464640933498e-06,
      "loss": 0.8842,
      "step": 37860
    },
    {
      "epoch": 2.288632380491932,
      "grad_norm": 0.6409374671040744,
      "learning_rate": 1.4601114424531476e-06,
      "loss": 0.8718,
      "step": 37870
    },
    {
      "epoch": 2.2892367196470658,
      "grad_norm": 0.6190529434845924,
      "learning_rate": 1.4577598149813965e-06,
      "loss": 0.8774,
      "step": 37880
    },
    {
      "epoch": 2.2898410588021996,
      "grad_norm": 0.5652780643712649,
      "learning_rate": 1.4554097595619039e-06,
      "loss": 0.8477,
      "step": 37890
    },
    {
      "epoch": 2.290445397957334,
      "grad_norm": 0.6160125339586162,
      "learning_rate": 1.4530612772376295e-06,
      "loss": 0.9001,
      "step": 37900
    },
    {
      "epoch": 2.2910497371124676,
      "grad_norm": 0.6093875474013092,
      "learning_rate": 1.45071436905083e-06,
      "loss": 0.898,
      "step": 37910
    },
    {
      "epoch": 2.2916540762676014,
      "grad_norm": 0.6026820628589228,
      "learning_rate": 1.4483690360430702e-06,
      "loss": 0.8778,
      "step": 37920
    },
    {
      "epoch": 2.2922584154227352,
      "grad_norm": 0.6052810117891112,
      "learning_rate": 1.4460252792552133e-06,
      "loss": 0.882,
      "step": 37930
    },
    {
      "epoch": 2.292862754577869,
      "grad_norm": 0.5933166114948633,
      "learning_rate": 1.4436830997274199e-06,
      "loss": 0.8737,
      "step": 37940
    },
    {
      "epoch": 2.293467093733003,
      "grad_norm": 0.5887759488402405,
      "learning_rate": 1.4413424984991547e-06,
      "loss": 0.8861,
      "step": 37950
    },
    {
      "epoch": 2.2940714328881366,
      "grad_norm": 0.6175401040264422,
      "learning_rate": 1.439003476609182e-06,
      "loss": 0.8962,
      "step": 37960
    },
    {
      "epoch": 2.294675772043271,
      "grad_norm": 0.6058417503897862,
      "learning_rate": 1.4366660350955613e-06,
      "loss": 0.8963,
      "step": 37970
    },
    {
      "epoch": 2.2952801111984047,
      "grad_norm": 0.584681810947484,
      "learning_rate": 1.434330174995654e-06,
      "loss": 0.8829,
      "step": 37980
    },
    {
      "epoch": 2.2958844503535385,
      "grad_norm": 0.6356538065996326,
      "learning_rate": 1.4319958973461202e-06,
      "loss": 0.8875,
      "step": 37990
    },
    {
      "epoch": 2.2964887895086723,
      "grad_norm": 0.6085316972920589,
      "learning_rate": 1.4296632031829139e-06,
      "loss": 0.8757,
      "step": 38000
    },
    {
      "epoch": 2.297093128663806,
      "grad_norm": 0.6136341535863586,
      "learning_rate": 1.42733209354129e-06,
      "loss": 0.8837,
      "step": 38010
    },
    {
      "epoch": 2.29769746781894,
      "grad_norm": 0.6124372399641269,
      "learning_rate": 1.4250025694558e-06,
      "loss": 0.9062,
      "step": 38020
    },
    {
      "epoch": 2.2983018069740737,
      "grad_norm": 0.5718842647616003,
      "learning_rate": 1.4226746319602886e-06,
      "loss": 0.8762,
      "step": 38030
    },
    {
      "epoch": 2.2989061461292075,
      "grad_norm": 0.637893617290052,
      "learning_rate": 1.4203482820878994e-06,
      "loss": 0.8783,
      "step": 38040
    },
    {
      "epoch": 2.2995104852843418,
      "grad_norm": 0.6259416948328,
      "learning_rate": 1.418023520871072e-06,
      "loss": 0.879,
      "step": 38050
    },
    {
      "epoch": 2.3001148244394756,
      "grad_norm": 0.5962914511084001,
      "learning_rate": 1.4157003493415361e-06,
      "loss": 0.885,
      "step": 38060
    },
    {
      "epoch": 2.3007191635946094,
      "grad_norm": 0.6372198512368646,
      "learning_rate": 1.4133787685303213e-06,
      "loss": 0.8727,
      "step": 38070
    },
    {
      "epoch": 2.301323502749743,
      "grad_norm": 0.6117439150049708,
      "learning_rate": 1.4110587794677487e-06,
      "loss": 0.8631,
      "step": 38080
    },
    {
      "epoch": 2.301927841904877,
      "grad_norm": 0.6385353486274702,
      "learning_rate": 1.4087403831834352e-06,
      "loss": 0.8996,
      "step": 38090
    },
    {
      "epoch": 2.302532181060011,
      "grad_norm": 0.6919054891369676,
      "learning_rate": 1.4064235807062853e-06,
      "loss": 0.8966,
      "step": 38100
    },
    {
      "epoch": 2.3031365202151446,
      "grad_norm": 0.6617623573733691,
      "learning_rate": 1.4041083730645043e-06,
      "loss": 0.9033,
      "step": 38110
    },
    {
      "epoch": 2.3037408593702784,
      "grad_norm": 0.6634089563771768,
      "learning_rate": 1.4017947612855814e-06,
      "loss": 0.9012,
      "step": 38120
    },
    {
      "epoch": 2.3043451985254126,
      "grad_norm": 0.6433921208736895,
      "learning_rate": 1.3994827463963045e-06,
      "loss": 0.8721,
      "step": 38130
    },
    {
      "epoch": 2.3049495376805464,
      "grad_norm": 0.6770576897545616,
      "learning_rate": 1.3971723294227468e-06,
      "loss": 0.8775,
      "step": 38140
    },
    {
      "epoch": 2.3055538768356802,
      "grad_norm": 0.7360078605631015,
      "learning_rate": 1.3948635113902781e-06,
      "loss": 0.8926,
      "step": 38150
    },
    {
      "epoch": 2.306158215990814,
      "grad_norm": 0.724226262026073,
      "learning_rate": 1.3925562933235536e-06,
      "loss": 0.8669,
      "step": 38160
    },
    {
      "epoch": 2.306762555145948,
      "grad_norm": 0.7325213655099648,
      "learning_rate": 1.3902506762465212e-06,
      "loss": 0.8991,
      "step": 38170
    },
    {
      "epoch": 2.3073668943010817,
      "grad_norm": 0.7278384354492624,
      "learning_rate": 1.38794666118242e-06,
      "loss": 0.8719,
      "step": 38180
    },
    {
      "epoch": 2.3079712334562155,
      "grad_norm": 0.7897648685006671,
      "learning_rate": 1.3856442491537719e-06,
      "loss": 0.8972,
      "step": 38190
    },
    {
      "epoch": 2.3085755726113497,
      "grad_norm": 0.634141413340625,
      "learning_rate": 1.3833434411823938e-06,
      "loss": 0.8924,
      "step": 38200
    },
    {
      "epoch": 2.3091799117664835,
      "grad_norm": 0.6748259700113949,
      "learning_rate": 1.381044238289389e-06,
      "loss": 0.8769,
      "step": 38210
    },
    {
      "epoch": 2.3097842509216173,
      "grad_norm": 0.6861944331681215,
      "learning_rate": 1.378746641495145e-06,
      "loss": 0.8601,
      "step": 38220
    },
    {
      "epoch": 2.310388590076751,
      "grad_norm": 0.6522450464471466,
      "learning_rate": 1.3764506518193404e-06,
      "loss": 0.9039,
      "step": 38230
    },
    {
      "epoch": 2.310992929231885,
      "grad_norm": 0.6660109867395618,
      "learning_rate": 1.3741562702809414e-06,
      "loss": 0.9032,
      "step": 38240
    },
    {
      "epoch": 2.3115972683870187,
      "grad_norm": 0.8858639158926562,
      "learning_rate": 1.3718634978981948e-06,
      "loss": 0.864,
      "step": 38250
    },
    {
      "epoch": 2.3122016075421525,
      "grad_norm": 0.8688711857797916,
      "learning_rate": 1.3695723356886387e-06,
      "loss": 0.8709,
      "step": 38260
    },
    {
      "epoch": 2.3128059466972863,
      "grad_norm": 0.807033033866591,
      "learning_rate": 1.3672827846690968e-06,
      "loss": 0.8719,
      "step": 38270
    },
    {
      "epoch": 2.31341028585242,
      "grad_norm": 0.8156677950414035,
      "learning_rate": 1.3649948458556717e-06,
      "loss": 0.918,
      "step": 38280
    },
    {
      "epoch": 2.3140146250075544,
      "grad_norm": 0.8481972079943875,
      "learning_rate": 1.3627085202637568e-06,
      "loss": 0.8854,
      "step": 38290
    },
    {
      "epoch": 2.314618964162688,
      "grad_norm": 0.8581066073457636,
      "learning_rate": 1.360423808908029e-06,
      "loss": 0.8695,
      "step": 38300
    },
    {
      "epoch": 2.315223303317822,
      "grad_norm": 0.8723934072113854,
      "learning_rate": 1.3581407128024437e-06,
      "loss": 0.878,
      "step": 38310
    },
    {
      "epoch": 2.315827642472956,
      "grad_norm": 0.854675385101025,
      "learning_rate": 1.3558592329602443e-06,
      "loss": 0.8923,
      "step": 38320
    },
    {
      "epoch": 2.3164319816280896,
      "grad_norm": 0.8577976898983138,
      "learning_rate": 1.3535793703939571e-06,
      "loss": 0.8832,
      "step": 38330
    },
    {
      "epoch": 2.3170363207832234,
      "grad_norm": 0.8515784710989001,
      "learning_rate": 1.3513011261153864e-06,
      "loss": 0.888,
      "step": 38340
    },
    {
      "epoch": 2.317640659938357,
      "grad_norm": 1.1906397458516886,
      "learning_rate": 1.3490245011356212e-06,
      "loss": 0.8631,
      "step": 38350
    },
    {
      "epoch": 2.3182449990934915,
      "grad_norm": 1.3576366931725123,
      "learning_rate": 1.3467494964650346e-06,
      "loss": 0.8744,
      "step": 38360
    },
    {
      "epoch": 2.3188493382486253,
      "grad_norm": 1.2433628693586447,
      "learning_rate": 1.3444761131132738e-06,
      "loss": 0.8631,
      "step": 38370
    },
    {
      "epoch": 2.319453677403759,
      "grad_norm": 1.2087813079594782,
      "learning_rate": 1.342204352089273e-06,
      "loss": 0.8955,
      "step": 38380
    },
    {
      "epoch": 2.320058016558893,
      "grad_norm": 1.2789827972554675,
      "learning_rate": 1.339934214401241e-06,
      "loss": 0.9016,
      "step": 38390
    },
    {
      "epoch": 2.3206623557140267,
      "grad_norm": 1.2571337809254906,
      "learning_rate": 1.337665701056672e-06,
      "loss": 0.8881,
      "step": 38400
    },
    {
      "epoch": 2.3212666948691605,
      "grad_norm": 1.0415830978911638,
      "learning_rate": 1.3353988130623319e-06,
      "loss": 0.9036,
      "step": 38410
    },
    {
      "epoch": 2.3218710340242943,
      "grad_norm": 0.9286746988296491,
      "learning_rate": 1.3331335514242727e-06,
      "loss": 0.9233,
      "step": 38420
    },
    {
      "epoch": 2.3224753731794285,
      "grad_norm": 0.9196636600332607,
      "learning_rate": 1.3308699171478217e-06,
      "loss": 0.8494,
      "step": 38430
    },
    {
      "epoch": 2.3230797123345623,
      "grad_norm": 1.0855610557346511,
      "learning_rate": 1.3286079112375816e-06,
      "loss": 0.9141,
      "step": 38440
    },
    {
      "epoch": 2.323684051489696,
      "grad_norm": 2.3883763207069872,
      "learning_rate": 1.3263475346974346e-06,
      "loss": 0.8796,
      "step": 38450
    },
    {
      "epoch": 2.32428839064483,
      "grad_norm": 2.4485917854873995,
      "learning_rate": 1.3240887885305426e-06,
      "loss": 0.8829,
      "step": 38460
    },
    {
      "epoch": 2.3248927297999638,
      "grad_norm": 2.5911538762668576,
      "learning_rate": 1.3218316737393372e-06,
      "loss": 0.9142,
      "step": 38470
    },
    {
      "epoch": 2.3254970689550976,
      "grad_norm": 2.425973762702792,
      "learning_rate": 1.3195761913255317e-06,
      "loss": 0.8762,
      "step": 38480
    },
    {
      "epoch": 2.3261014081102314,
      "grad_norm": 2.678089605295009,
      "learning_rate": 1.3173223422901143e-06,
      "loss": 0.9103,
      "step": 38490
    },
    {
      "epoch": 2.326705747265365,
      "grad_norm": 0.9620608279141157,
      "learning_rate": 1.3150701276333444e-06,
      "loss": 0.9105,
      "step": 38500
    },
    {
      "epoch": 2.327310086420499,
      "grad_norm": 0.869689154709553,
      "learning_rate": 1.3128195483547596e-06,
      "loss": 0.8866,
      "step": 38510
    },
    {
      "epoch": 2.327914425575633,
      "grad_norm": 0.9514243182061801,
      "learning_rate": 1.3105706054531726e-06,
      "loss": 0.8887,
      "step": 38520
    },
    {
      "epoch": 2.328518764730767,
      "grad_norm": 0.920439849185892,
      "learning_rate": 1.3083232999266654e-06,
      "loss": 0.8732,
      "step": 38530
    },
    {
      "epoch": 2.329123103885901,
      "grad_norm": 0.9496562982987091,
      "learning_rate": 1.3060776327725976e-06,
      "loss": 0.8518,
      "step": 38540
    },
    {
      "epoch": 2.3297274430410346,
      "grad_norm": 1.007130116180434,
      "learning_rate": 1.303833604987601e-06,
      "loss": 0.9023,
      "step": 38550
    },
    {
      "epoch": 2.3303317821961684,
      "grad_norm": 1.0127214602493766,
      "learning_rate": 1.3015912175675766e-06,
      "loss": 0.8947,
      "step": 38560
    },
    {
      "epoch": 2.3309361213513022,
      "grad_norm": 0.9654032967972956,
      "learning_rate": 1.299350471507701e-06,
      "loss": 0.8869,
      "step": 38570
    },
    {
      "epoch": 2.331540460506436,
      "grad_norm": 0.9213049583388885,
      "learning_rate": 1.2971113678024228e-06,
      "loss": 0.8803,
      "step": 38580
    },
    {
      "epoch": 2.3321447996615703,
      "grad_norm": 0.972742456399869,
      "learning_rate": 1.2948739074454563e-06,
      "loss": 0.86,
      "step": 38590
    },
    {
      "epoch": 2.332749138816704,
      "grad_norm": 1.0832606148539412,
      "learning_rate": 1.2926380914297931e-06,
      "loss": 0.8878,
      "step": 38600
    },
    {
      "epoch": 2.333353477971838,
      "grad_norm": 1.0287336097995405,
      "learning_rate": 1.2904039207476927e-06,
      "loss": 0.8918,
      "step": 38610
    },
    {
      "epoch": 2.3339578171269717,
      "grad_norm": 1.1057106361436366,
      "learning_rate": 1.288171396390681e-06,
      "loss": 0.8749,
      "step": 38620
    },
    {
      "epoch": 2.3345621562821055,
      "grad_norm": 1.0174293623061006,
      "learning_rate": 1.2859405193495594e-06,
      "loss": 0.8629,
      "step": 38630
    },
    {
      "epoch": 2.3351664954372393,
      "grad_norm": 1.0853736579345017,
      "learning_rate": 1.2837112906143917e-06,
      "loss": 0.8862,
      "step": 38640
    },
    {
      "epoch": 2.335770834592373,
      "grad_norm": 1.1257346695427284,
      "learning_rate": 1.2814837111745166e-06,
      "loss": 0.8547,
      "step": 38650
    },
    {
      "epoch": 2.336375173747507,
      "grad_norm": 1.0250955148952747,
      "learning_rate": 1.2792577820185343e-06,
      "loss": 0.8891,
      "step": 38660
    },
    {
      "epoch": 2.3369795129026407,
      "grad_norm": 1.0579691840048668,
      "learning_rate": 1.2770335041343195e-06,
      "loss": 0.8891,
      "step": 38670
    },
    {
      "epoch": 2.337583852057775,
      "grad_norm": 1.1262128268216802,
      "learning_rate": 1.2748108785090074e-06,
      "loss": 0.8621,
      "step": 38680
    },
    {
      "epoch": 2.3381881912129088,
      "grad_norm": 1.0215376361463935,
      "learning_rate": 1.2725899061290043e-06,
      "loss": 0.887,
      "step": 38690
    },
    {
      "epoch": 2.3387925303680426,
      "grad_norm": 1.217297751405057,
      "learning_rate": 1.2703705879799833e-06,
      "loss": 0.898,
      "step": 38700
    },
    {
      "epoch": 2.3393968695231764,
      "grad_norm": 1.0979621224373344,
      "learning_rate": 1.2681529250468782e-06,
      "loss": 0.881,
      "step": 38710
    },
    {
      "epoch": 2.34000120867831,
      "grad_norm": 1.2070410107461766,
      "learning_rate": 1.2659369183138936e-06,
      "loss": 0.8649,
      "step": 38720
    },
    {
      "epoch": 2.340605547833444,
      "grad_norm": 1.2016328096140823,
      "learning_rate": 1.2637225687644989e-06,
      "loss": 0.8853,
      "step": 38730
    },
    {
      "epoch": 2.341209886988578,
      "grad_norm": 1.0962450781221647,
      "learning_rate": 1.2615098773814222e-06,
      "loss": 0.8962,
      "step": 38740
    },
    {
      "epoch": 2.341814226143712,
      "grad_norm": 1.0704058699263035,
      "learning_rate": 1.2592988451466615e-06,
      "loss": 0.888,
      "step": 38750
    },
    {
      "epoch": 2.342418565298846,
      "grad_norm": 1.0449913846281933,
      "learning_rate": 1.2570894730414773e-06,
      "loss": 0.8846,
      "step": 38760
    },
    {
      "epoch": 2.3430229044539796,
      "grad_norm": 1.0257455429089837,
      "learning_rate": 1.2548817620463932e-06,
      "loss": 0.8793,
      "step": 38770
    },
    {
      "epoch": 2.3436272436091135,
      "grad_norm": 1.1340190647976194,
      "learning_rate": 1.2526757131411932e-06,
      "loss": 0.8892,
      "step": 38780
    },
    {
      "epoch": 2.3442315827642473,
      "grad_norm": 1.0348871480890511,
      "learning_rate": 1.2504713273049268e-06,
      "loss": 0.8638,
      "step": 38790
    },
    {
      "epoch": 2.344835921919381,
      "grad_norm": 1.1083706930609398,
      "learning_rate": 1.2482686055159054e-06,
      "loss": 0.8747,
      "step": 38800
    },
    {
      "epoch": 2.345440261074515,
      "grad_norm": 1.0484467676271267,
      "learning_rate": 1.2460675487516977e-06,
      "loss": 0.8729,
      "step": 38810
    },
    {
      "epoch": 2.346044600229649,
      "grad_norm": 1.1050274060423304,
      "learning_rate": 1.2438681579891383e-06,
      "loss": 0.8801,
      "step": 38820
    },
    {
      "epoch": 2.346648939384783,
      "grad_norm": 1.167344705455927,
      "learning_rate": 1.2416704342043217e-06,
      "loss": 0.8908,
      "step": 38830
    },
    {
      "epoch": 2.3472532785399167,
      "grad_norm": 1.1648404252980153,
      "learning_rate": 1.2394743783725987e-06,
      "loss": 0.8693,
      "step": 38840
    },
    {
      "epoch": 2.3478576176950505,
      "grad_norm": 1.075533100150982,
      "learning_rate": 1.2372799914685845e-06,
      "loss": 0.89,
      "step": 38850
    },
    {
      "epoch": 2.3484619568501843,
      "grad_norm": 1.0312528503652874,
      "learning_rate": 1.2350872744661523e-06,
      "loss": 0.8822,
      "step": 38860
    },
    {
      "epoch": 2.349066296005318,
      "grad_norm": 0.9737887922759815,
      "learning_rate": 1.2328962283384315e-06,
      "loss": 0.8678,
      "step": 38870
    },
    {
      "epoch": 2.349670635160452,
      "grad_norm": 1.031132776002923,
      "learning_rate": 1.2307068540578149e-06,
      "loss": 0.8839,
      "step": 38880
    },
    {
      "epoch": 2.3502749743155857,
      "grad_norm": 1.0654572364935515,
      "learning_rate": 1.2285191525959478e-06,
      "loss": 0.872,
      "step": 38890
    },
    {
      "epoch": 2.3508793134707195,
      "grad_norm": 1.2851219673422374,
      "learning_rate": 1.226333124923738e-06,
      "loss": 0.8933,
      "step": 38900
    },
    {
      "epoch": 2.351483652625854,
      "grad_norm": 1.309461044217282,
      "learning_rate": 1.224148772011346e-06,
      "loss": 0.9095,
      "step": 38910
    },
    {
      "epoch": 2.3520879917809876,
      "grad_norm": 1.2617117027930045,
      "learning_rate": 1.2219660948281942e-06,
      "loss": 0.9156,
      "step": 38920
    },
    {
      "epoch": 2.3526923309361214,
      "grad_norm": 1.1616053041216488,
      "learning_rate": 1.2197850943429557e-06,
      "loss": 0.9113,
      "step": 38930
    },
    {
      "epoch": 2.353296670091255,
      "grad_norm": 1.256998491359396,
      "learning_rate": 1.2176057715235634e-06,
      "loss": 0.8755,
      "step": 38940
    },
    {
      "epoch": 2.353901009246389,
      "grad_norm": 1.1092277008399234,
      "learning_rate": 1.2154281273372065e-06,
      "loss": 0.8806,
      "step": 38950
    },
    {
      "epoch": 2.354505348401523,
      "grad_norm": 1.2028756460990289,
      "learning_rate": 1.2132521627503236e-06,
      "loss": 0.895,
      "step": 38960
    },
    {
      "epoch": 2.3551096875566566,
      "grad_norm": 1.2170443345590407,
      "learning_rate": 1.2110778787286132e-06,
      "loss": 0.8923,
      "step": 38970
    },
    {
      "epoch": 2.355714026711791,
      "grad_norm": 1.183272829135899,
      "learning_rate": 1.2089052762370278e-06,
      "loss": 0.8597,
      "step": 38980
    },
    {
      "epoch": 2.3563183658669247,
      "grad_norm": 1.1522304044103695,
      "learning_rate": 1.2067343562397693e-06,
      "loss": 0.9039,
      "step": 38990
    },
    {
      "epoch": 2.3569227050220585,
      "grad_norm": 1.1732744962455999,
      "learning_rate": 1.2045651197002973e-06,
      "loss": 0.9066,
      "step": 39000
    },
    {
      "epoch": 2.3575270441771923,
      "grad_norm": 1.2154805582034967,
      "learning_rate": 1.2023975675813237e-06,
      "loss": 0.8804,
      "step": 39010
    },
    {
      "epoch": 2.358131383332326,
      "grad_norm": 1.1846220976986728,
      "learning_rate": 1.2002317008448088e-06,
      "loss": 0.8888,
      "step": 39020
    },
    {
      "epoch": 2.35873572248746,
      "grad_norm": 1.2588579728657623,
      "learning_rate": 1.1980675204519704e-06,
      "loss": 0.8778,
      "step": 39030
    },
    {
      "epoch": 2.3593400616425937,
      "grad_norm": 1.2217056019604282,
      "learning_rate": 1.195905027363276e-06,
      "loss": 0.8897,
      "step": 39040
    },
    {
      "epoch": 2.3599444007977275,
      "grad_norm": 1.3717312497810175,
      "learning_rate": 1.193744222538441e-06,
      "loss": 0.8652,
      "step": 39050
    },
    {
      "epoch": 2.3605487399528613,
      "grad_norm": 1.461396880656895,
      "learning_rate": 1.191585106936436e-06,
      "loss": 0.8894,
      "step": 39060
    },
    {
      "epoch": 2.3611530791079955,
      "grad_norm": 1.4992617779409008,
      "learning_rate": 1.1894276815154815e-06,
      "loss": 0.8884,
      "step": 39070
    },
    {
      "epoch": 2.3617574182631293,
      "grad_norm": 1.415271971506978,
      "learning_rate": 1.187271947233044e-06,
      "loss": 0.8973,
      "step": 39080
    },
    {
      "epoch": 2.362361757418263,
      "grad_norm": 1.4582796354205354,
      "learning_rate": 1.1851179050458428e-06,
      "loss": 0.8746,
      "step": 39090
    },
    {
      "epoch": 2.362966096573397,
      "grad_norm": 0.826188634656245,
      "learning_rate": 1.1829655559098458e-06,
      "loss": 0.8922,
      "step": 39100
    },
    {
      "epoch": 2.3635704357285308,
      "grad_norm": 0.9309698722708558,
      "learning_rate": 1.1808149007802711e-06,
      "loss": 0.8779,
      "step": 39110
    },
    {
      "epoch": 2.3641747748836646,
      "grad_norm": 0.9876997859770517,
      "learning_rate": 1.1786659406115798e-06,
      "loss": 0.8926,
      "step": 39120
    },
    {
      "epoch": 2.3647791140387984,
      "grad_norm": 0.9522632663271489,
      "learning_rate": 1.1765186763574871e-06,
      "loss": 0.8981,
      "step": 39130
    },
    {
      "epoch": 2.3653834531939326,
      "grad_norm": 0.9241515739771305,
      "learning_rate": 1.1743731089709498e-06,
      "loss": 0.896,
      "step": 39140
    },
    {
      "epoch": 2.3659877923490664,
      "grad_norm": 0.8092522033893093,
      "learning_rate": 1.1722292394041769e-06,
      "loss": 0.8717,
      "step": 39150
    },
    {
      "epoch": 2.3665921315042002,
      "grad_norm": 0.8583278389170031,
      "learning_rate": 1.1700870686086191e-06,
      "loss": 0.8783,
      "step": 39160
    },
    {
      "epoch": 2.367196470659334,
      "grad_norm": 0.9271202916869054,
      "learning_rate": 1.167946597534978e-06,
      "loss": 0.8922,
      "step": 39170
    },
    {
      "epoch": 2.367800809814468,
      "grad_norm": 0.8260098433579567,
      "learning_rate": 1.1658078271331951e-06,
      "loss": 0.9027,
      "step": 39180
    },
    {
      "epoch": 2.3684051489696016,
      "grad_norm": 0.8597411978291764,
      "learning_rate": 1.163670758352462e-06,
      "loss": 0.8818,
      "step": 39190
    },
    {
      "epoch": 2.3690094881247354,
      "grad_norm": 0.8610779457857143,
      "learning_rate": 1.161535392141216e-06,
      "loss": 0.8663,
      "step": 39200
    },
    {
      "epoch": 2.3696138272798697,
      "grad_norm": 0.8968116179996009,
      "learning_rate": 1.1594017294471316e-06,
      "loss": 0.8824,
      "step": 39210
    },
    {
      "epoch": 2.3702181664350035,
      "grad_norm": 0.8531697052247004,
      "learning_rate": 1.1572697712171344e-06,
      "loss": 0.8829,
      "step": 39220
    },
    {
      "epoch": 2.3708225055901373,
      "grad_norm": 0.8091335145529137,
      "learning_rate": 1.1551395183973924e-06,
      "loss": 0.8673,
      "step": 39230
    },
    {
      "epoch": 2.371426844745271,
      "grad_norm": 0.8609162752883601,
      "learning_rate": 1.1530109719333127e-06,
      "loss": 0.8782,
      "step": 39240
    },
    {
      "epoch": 2.372031183900405,
      "grad_norm": 0.6486798281733559,
      "learning_rate": 1.1508841327695497e-06,
      "loss": 0.8823,
      "step": 39250
    },
    {
      "epoch": 2.3726355230555387,
      "grad_norm": 0.6535240832818486,
      "learning_rate": 1.1487590018499984e-06,
      "loss": 0.881,
      "step": 39260
    },
    {
      "epoch": 2.3732398622106725,
      "grad_norm": 0.631833198038447,
      "learning_rate": 1.1466355801177942e-06,
      "loss": 0.8714,
      "step": 39270
    },
    {
      "epoch": 2.3738442013658063,
      "grad_norm": 0.6600240251580102,
      "learning_rate": 1.144513868515315e-06,
      "loss": 0.9039,
      "step": 39280
    },
    {
      "epoch": 2.37444854052094,
      "grad_norm": 0.6775966793398034,
      "learning_rate": 1.1423938679841834e-06,
      "loss": 0.8978,
      "step": 39290
    },
    {
      "epoch": 2.3750528796760744,
      "grad_norm": 0.6575718128032229,
      "learning_rate": 1.1402755794652554e-06,
      "loss": 0.886,
      "step": 39300
    },
    {
      "epoch": 2.375657218831208,
      "grad_norm": 0.6841897395708637,
      "learning_rate": 1.1381590038986319e-06,
      "loss": 0.8884,
      "step": 39310
    },
    {
      "epoch": 2.376261557986342,
      "grad_norm": 0.7246653885076295,
      "learning_rate": 1.136044142223655e-06,
      "loss": 0.8934,
      "step": 39320
    },
    {
      "epoch": 2.376865897141476,
      "grad_norm": 0.6588294289115753,
      "learning_rate": 1.133930995378902e-06,
      "loss": 0.9049,
      "step": 39330
    },
    {
      "epoch": 2.3774702362966096,
      "grad_norm": 0.6488178551631584,
      "learning_rate": 1.1318195643021902e-06,
      "loss": 0.8884,
      "step": 39340
    },
    {
      "epoch": 2.3780745754517434,
      "grad_norm": 0.83890880199567,
      "learning_rate": 1.1297098499305798e-06,
      "loss": 0.8731,
      "step": 39350
    },
    {
      "epoch": 2.378678914606877,
      "grad_norm": 0.8835444414153574,
      "learning_rate": 1.1276018532003613e-06,
      "loss": 0.8721,
      "step": 39360
    },
    {
      "epoch": 2.3792832537620114,
      "grad_norm": 0.8018377844467994,
      "learning_rate": 1.1254955750470698e-06,
      "loss": 0.8777,
      "step": 39370
    },
    {
      "epoch": 2.3798875929171452,
      "grad_norm": 1.0536293340339764,
      "learning_rate": 1.1233910164054763e-06,
      "loss": 0.8708,
      "step": 39380
    },
    {
      "epoch": 2.380491932072279,
      "grad_norm": 0.9106179478752364,
      "learning_rate": 1.1212881782095842e-06,
      "loss": 0.8702,
      "step": 39390
    },
    {
      "epoch": 2.381096271227413,
      "grad_norm": 0.6186197212473284,
      "learning_rate": 1.1191870613926398e-06,
      "loss": 0.8834,
      "step": 39400
    },
    {
      "epoch": 2.3817006103825467,
      "grad_norm": 0.5803289380955238,
      "learning_rate": 1.11708766688712e-06,
      "loss": 0.9057,
      "step": 39410
    },
    {
      "epoch": 2.3823049495376805,
      "grad_norm": 0.5947504156256807,
      "learning_rate": 1.1149899956247418e-06,
      "loss": 0.889,
      "step": 39420
    },
    {
      "epoch": 2.3829092886928143,
      "grad_norm": 0.5722482574353475,
      "learning_rate": 1.112894048536453e-06,
      "loss": 0.8937,
      "step": 39430
    },
    {
      "epoch": 2.3835136278479485,
      "grad_norm": 0.6317931416178177,
      "learning_rate": 1.1107998265524394e-06,
      "loss": 0.9303,
      "step": 39440
    },
    {
      "epoch": 2.3841179670030823,
      "grad_norm": 0.6046702724342344,
      "learning_rate": 1.1087073306021222e-06,
      "loss": 0.871,
      "step": 39450
    },
    {
      "epoch": 2.384722306158216,
      "grad_norm": 0.5660446702728901,
      "learning_rate": 1.1066165616141516e-06,
      "loss": 0.8761,
      "step": 39460
    },
    {
      "epoch": 2.38532664531335,
      "grad_norm": 0.5881350501581389,
      "learning_rate": 1.1045275205164157e-06,
      "loss": 0.8777,
      "step": 39470
    },
    {
      "epoch": 2.3859309844684837,
      "grad_norm": 0.6062244346918683,
      "learning_rate": 1.1024402082360363e-06,
      "loss": 0.8708,
      "step": 39480
    },
    {
      "epoch": 2.3865353236236175,
      "grad_norm": 0.6617265414934231,
      "learning_rate": 1.100354625699363e-06,
      "loss": 0.8849,
      "step": 39490
    },
    {
      "epoch": 2.3871396627787513,
      "grad_norm": 0.5730605062061782,
      "learning_rate": 1.0982707738319832e-06,
      "loss": 0.8811,
      "step": 39500
    },
    {
      "epoch": 2.387744001933885,
      "grad_norm": 0.6072893979708712,
      "learning_rate": 1.0961886535587142e-06,
      "loss": 0.888,
      "step": 39510
    },
    {
      "epoch": 2.388348341089019,
      "grad_norm": 0.612568042548487,
      "learning_rate": 1.0941082658036018e-06,
      "loss": 0.8733,
      "step": 39520
    },
    {
      "epoch": 2.388952680244153,
      "grad_norm": 0.6192737226747179,
      "learning_rate": 1.0920296114899275e-06,
      "loss": 0.8851,
      "step": 39530
    },
    {
      "epoch": 2.389557019399287,
      "grad_norm": 0.6149845734577145,
      "learning_rate": 1.089952691540203e-06,
      "loss": 0.8726,
      "step": 39540
    },
    {
      "epoch": 2.390161358554421,
      "grad_norm": 0.6106027207908674,
      "learning_rate": 1.0878775068761666e-06,
      "loss": 0.9031,
      "step": 39550
    },
    {
      "epoch": 2.3907656977095546,
      "grad_norm": 0.5817264493192338,
      "learning_rate": 1.085804058418789e-06,
      "loss": 0.8889,
      "step": 39560
    },
    {
      "epoch": 2.3913700368646884,
      "grad_norm": 0.5719520120056023,
      "learning_rate": 1.0837323470882727e-06,
      "loss": 0.874,
      "step": 39570
    },
    {
      "epoch": 2.391974376019822,
      "grad_norm": 0.60121580275604,
      "learning_rate": 1.0816623738040432e-06,
      "loss": 0.8928,
      "step": 39580
    },
    {
      "epoch": 2.392578715174956,
      "grad_norm": 0.5424523460294342,
      "learning_rate": 1.0795941394847604e-06,
      "loss": 0.887,
      "step": 39590
    },
    {
      "epoch": 2.3931830543300903,
      "grad_norm": 0.5837087164209533,
      "learning_rate": 1.0775276450483108e-06,
      "loss": 0.8938,
      "step": 39600
    },
    {
      "epoch": 2.393787393485224,
      "grad_norm": 0.545301989808708,
      "learning_rate": 1.0754628914118055e-06,
      "loss": 0.8564,
      "step": 39610
    },
    {
      "epoch": 2.394391732640358,
      "grad_norm": 0.5676700215894412,
      "learning_rate": 1.0733998794915873e-06,
      "loss": 0.8823,
      "step": 39620
    },
    {
      "epoch": 2.3949960717954917,
      "grad_norm": 0.5776790473308407,
      "learning_rate": 1.0713386102032258e-06,
      "loss": 0.8815,
      "step": 39630
    },
    {
      "epoch": 2.3956004109506255,
      "grad_norm": 0.5829049078485785,
      "learning_rate": 1.0692790844615131e-06,
      "loss": 0.8918,
      "step": 39640
    },
    {
      "epoch": 2.3962047501057593,
      "grad_norm": 0.6222032243704987,
      "learning_rate": 1.0672213031804718e-06,
      "loss": 0.8856,
      "step": 39650
    },
    {
      "epoch": 2.396809089260893,
      "grad_norm": 0.6482413632166244,
      "learning_rate": 1.0651652672733497e-06,
      "loss": 0.8862,
      "step": 39660
    },
    {
      "epoch": 2.397413428416027,
      "grad_norm": 0.6234854921688038,
      "learning_rate": 1.0631109776526166e-06,
      "loss": 0.8738,
      "step": 39670
    },
    {
      "epoch": 2.3980177675711607,
      "grad_norm": 0.6025737863212187,
      "learning_rate": 1.0610584352299724e-06,
      "loss": 0.9005,
      "step": 39680
    },
    {
      "epoch": 2.398622106726295,
      "grad_norm": 0.5943390356962067,
      "learning_rate": 1.059007640916338e-06,
      "loss": 0.8815,
      "step": 39690
    },
    {
      "epoch": 2.3992264458814287,
      "grad_norm": 0.6665130767977476,
      "learning_rate": 1.056958595621857e-06,
      "loss": 0.8803,
      "step": 39700
    },
    {
      "epoch": 2.3998307850365626,
      "grad_norm": 0.6586745122537215,
      "learning_rate": 1.0549113002559013e-06,
      "loss": 0.9071,
      "step": 39710
    },
    {
      "epoch": 2.4004351241916964,
      "grad_norm": 0.6972941100277277,
      "learning_rate": 1.052865755727066e-06,
      "loss": 0.887,
      "step": 39720
    },
    {
      "epoch": 2.40103946334683,
      "grad_norm": 0.6553379223778856,
      "learning_rate": 1.0508219629431631e-06,
      "loss": 0.9029,
      "step": 39730
    },
    {
      "epoch": 2.401643802501964,
      "grad_norm": 0.7160502767925534,
      "learning_rate": 1.0487799228112344e-06,
      "loss": 0.892,
      "step": 39740
    },
    {
      "epoch": 2.4022481416570978,
      "grad_norm": 0.6923967937788605,
      "learning_rate": 1.046739636237541e-06,
      "loss": 0.8695,
      "step": 39750
    },
    {
      "epoch": 2.402852480812232,
      "grad_norm": 0.7185550380353858,
      "learning_rate": 1.0447011041275629e-06,
      "loss": 0.8833,
      "step": 39760
    },
    {
      "epoch": 2.403456819967366,
      "grad_norm": 0.7149292759973592,
      "learning_rate": 1.0426643273860055e-06,
      "loss": 0.876,
      "step": 39770
    },
    {
      "epoch": 2.4040611591224996,
      "grad_norm": 0.6958120885525948,
      "learning_rate": 1.0406293069167944e-06,
      "loss": 0.8786,
      "step": 39780
    },
    {
      "epoch": 2.4046654982776334,
      "grad_norm": 0.7545752472801516,
      "learning_rate": 1.0385960436230762e-06,
      "loss": 0.8963,
      "step": 39790
    },
    {
      "epoch": 2.4052698374327672,
      "grad_norm": 0.7096128021795056,
      "learning_rate": 1.0365645384072137e-06,
      "loss": 0.8793,
      "step": 39800
    },
    {
      "epoch": 2.405874176587901,
      "grad_norm": 0.6790031370634885,
      "learning_rate": 1.0345347921707942e-06,
      "loss": 0.8669,
      "step": 39810
    },
    {
      "epoch": 2.406478515743035,
      "grad_norm": 0.6656644782473132,
      "learning_rate": 1.0325068058146232e-06,
      "loss": 0.901,
      "step": 39820
    },
    {
      "epoch": 2.407082854898169,
      "grad_norm": 0.6415335294265717,
      "learning_rate": 1.0304805802387225e-06,
      "loss": 0.8818,
      "step": 39830
    },
    {
      "epoch": 2.407687194053303,
      "grad_norm": 0.7207569061893038,
      "learning_rate": 1.0284561163423356e-06,
      "loss": 0.8773,
      "step": 39840
    },
    {
      "epoch": 2.4082915332084367,
      "grad_norm": 0.8244458612890934,
      "learning_rate": 1.0264334150239247e-06,
      "loss": 0.8803,
      "step": 39850
    },
    {
      "epoch": 2.4088958723635705,
      "grad_norm": 0.894079612005884,
      "learning_rate": 1.0244124771811648e-06,
      "loss": 0.9099,
      "step": 39860
    },
    {
      "epoch": 2.4095002115187043,
      "grad_norm": 0.938144724272759,
      "learning_rate": 1.022393303710953e-06,
      "loss": 0.9016,
      "step": 39870
    },
    {
      "epoch": 2.410104550673838,
      "grad_norm": 0.8668587694656729,
      "learning_rate": 1.020375895509404e-06,
      "loss": 0.8839,
      "step": 39880
    },
    {
      "epoch": 2.410708889828972,
      "grad_norm": 0.8593237257260798,
      "learning_rate": 1.0183602534718435e-06,
      "loss": 0.8946,
      "step": 39890
    },
    {
      "epoch": 2.4113132289841057,
      "grad_norm": 0.812151659495348,
      "learning_rate": 1.0163463784928184e-06,
      "loss": 0.8803,
      "step": 39900
    },
    {
      "epoch": 2.4119175681392395,
      "grad_norm": 0.7971137937921556,
      "learning_rate": 1.0143342714660914e-06,
      "loss": 0.8713,
      "step": 39910
    },
    {
      "epoch": 2.4125219072943738,
      "grad_norm": 0.8368856847598072,
      "learning_rate": 1.0123239332846362e-06,
      "loss": 0.8829,
      "step": 39920
    },
    {
      "epoch": 2.4131262464495076,
      "grad_norm": 0.8022886361565047,
      "learning_rate": 1.0103153648406477e-06,
      "loss": 0.9055,
      "step": 39930
    },
    {
      "epoch": 2.4137305856046414,
      "grad_norm": 0.8243992837945359,
      "learning_rate": 1.0083085670255283e-06,
      "loss": 0.8922,
      "step": 39940
    },
    {
      "epoch": 2.414334924759775,
      "grad_norm": 1.2176967055080188,
      "learning_rate": 1.0063035407299016e-06,
      "loss": 0.8728,
      "step": 39950
    },
    {
      "epoch": 2.414939263914909,
      "grad_norm": 1.253705674463632,
      "learning_rate": 1.0043002868435991e-06,
      "loss": 0.8907,
      "step": 39960
    },
    {
      "epoch": 2.415543603070043,
      "grad_norm": 1.2232482221576086,
      "learning_rate": 1.0022988062556704e-06,
      "loss": 0.8921,
      "step": 39970
    },
    {
      "epoch": 2.4161479422251766,
      "grad_norm": 1.2568056364655433,
      "learning_rate": 1.0002990998543737e-06,
      "loss": 0.8961,
      "step": 39980
    },
    {
      "epoch": 2.416752281380311,
      "grad_norm": 1.228900319126794,
      "learning_rate": 9.983011685271832e-07,
      "loss": 0.8949,
      "step": 39990
    },
    {
      "epoch": 2.4173566205354446,
      "grad_norm": 0.7977047349015124,
      "learning_rate": 9.963050131607865e-07,
      "loss": 0.8875,
      "step": 40000
    },
    {
      "epoch": 2.4179609596905784,
      "grad_norm": 0.9469623879299512,
      "learning_rate": 9.943106346410769e-07,
      "loss": 0.8691,
      "step": 40010
    },
    {
      "epoch": 2.4185652988457123,
      "grad_norm": 1.01828106855766,
      "learning_rate": 9.923180338531651e-07,
      "loss": 0.8754,
      "step": 40020
    },
    {
      "epoch": 2.419169638000846,
      "grad_norm": 1.019067869588305,
      "learning_rate": 9.903272116813717e-07,
      "loss": 0.8693,
      "step": 40030
    },
    {
      "epoch": 2.41977397715598,
      "grad_norm": 1.2427818593476803,
      "learning_rate": 9.883381690092248e-07,
      "loss": 0.8637,
      "step": 40040
    },
    {
      "epoch": 2.4203783163111137,
      "grad_norm": 2.51751257552718,
      "learning_rate": 9.863509067194653e-07,
      "loss": 0.8688,
      "step": 40050
    },
    {
      "epoch": 2.4209826554662475,
      "grad_norm": 2.577179746751611,
      "learning_rate": 9.84365425694046e-07,
      "loss": 0.9009,
      "step": 40060
    },
    {
      "epoch": 2.4215869946213813,
      "grad_norm": 2.501405424447074,
      "learning_rate": 9.823817268141233e-07,
      "loss": 0.8959,
      "step": 40070
    },
    {
      "epoch": 2.4221913337765155,
      "grad_norm": 2.88040941430708,
      "learning_rate": 9.803998109600681e-07,
      "loss": 0.8703,
      "step": 40080
    },
    {
      "epoch": 2.4227956729316493,
      "grad_norm": 2.5613001107528564,
      "learning_rate": 9.78419679011458e-07,
      "loss": 0.8895,
      "step": 40090
    },
    {
      "epoch": 2.423400012086783,
      "grad_norm": 0.9349367631368284,
      "learning_rate": 9.7644133184708e-07,
      "loss": 0.8759,
      "step": 40100
    },
    {
      "epoch": 2.424004351241917,
      "grad_norm": 1.009146913955769,
      "learning_rate": 9.744647703449256e-07,
      "loss": 0.9112,
      "step": 40110
    },
    {
      "epoch": 2.4246086903970507,
      "grad_norm": 1.013591341517496,
      "learning_rate": 9.72489995382197e-07,
      "loss": 0.881,
      "step": 40120
    },
    {
      "epoch": 2.4252130295521845,
      "grad_norm": 0.9612964801090413,
      "learning_rate": 9.705170078353043e-07,
      "loss": 0.8847,
      "step": 40130
    },
    {
      "epoch": 2.4258173687073183,
      "grad_norm": 0.9528065448293137,
      "learning_rate": 9.685458085798604e-07,
      "loss": 0.8781,
      "step": 40140
    },
    {
      "epoch": 2.4264217078624526,
      "grad_norm": 1.0473877532677942,
      "learning_rate": 9.665763984906878e-07,
      "loss": 0.88,
      "step": 40150
    },
    {
      "epoch": 2.4270260470175864,
      "grad_norm": 1.0291073107156665,
      "learning_rate": 9.646087784418157e-07,
      "loss": 0.8805,
      "step": 40160
    },
    {
      "epoch": 2.42763038617272,
      "grad_norm": 1.0003495739477302,
      "learning_rate": 9.626429493064743e-07,
      "loss": 0.885,
      "step": 40170
    },
    {
      "epoch": 2.428234725327854,
      "grad_norm": 1.0532050241340538,
      "learning_rate": 9.606789119571047e-07,
      "loss": 0.8901,
      "step": 40180
    },
    {
      "epoch": 2.428839064482988,
      "grad_norm": 0.9979484167797138,
      "learning_rate": 9.587166672653475e-07,
      "loss": 0.888,
      "step": 40190
    },
    {
      "epoch": 2.4294434036381216,
      "grad_norm": 1.0325258340420072,
      "learning_rate": 9.567562161020534e-07,
      "loss": 0.8843,
      "step": 40200
    },
    {
      "epoch": 2.4300477427932554,
      "grad_norm": 1.1206055926553746,
      "learning_rate": 9.5479755933727e-07,
      "loss": 0.9071,
      "step": 40210
    },
    {
      "epoch": 2.4306520819483897,
      "grad_norm": 1.0530992969734247,
      "learning_rate": 9.52840697840256e-07,
      "loss": 0.88,
      "step": 40220
    },
    {
      "epoch": 2.4312564211035235,
      "grad_norm": 0.9979372713695156,
      "learning_rate": 9.508856324794674e-07,
      "loss": 0.8765,
      "step": 40230
    },
    {
      "epoch": 2.4318607602586573,
      "grad_norm": 1.0181891717185223,
      "learning_rate": 9.489323641225667e-07,
      "loss": 0.8849,
      "step": 40240
    },
    {
      "epoch": 2.432465099413791,
      "grad_norm": 1.1239773592157096,
      "learning_rate": 9.469808936364189e-07,
      "loss": 0.8871,
      "step": 40250
    },
    {
      "epoch": 2.433069438568925,
      "grad_norm": 1.0714133073879752,
      "learning_rate": 9.450312218870878e-07,
      "loss": 0.8983,
      "step": 40260
    },
    {
      "epoch": 2.4336737777240587,
      "grad_norm": 1.1042796970049513,
      "learning_rate": 9.430833497398417e-07,
      "loss": 0.8827,
      "step": 40270
    },
    {
      "epoch": 2.4342781168791925,
      "grad_norm": 1.085088893401994,
      "learning_rate": 9.41137278059151e-07,
      "loss": 0.9013,
      "step": 40280
    },
    {
      "epoch": 2.4348824560343263,
      "grad_norm": 1.037640719786448,
      "learning_rate": 9.391930077086837e-07,
      "loss": 0.8811,
      "step": 40290
    },
    {
      "epoch": 2.43548679518946,
      "grad_norm": 1.1339613588055377,
      "learning_rate": 9.372505395513104e-07,
      "loss": 0.8824,
      "step": 40300
    },
    {
      "epoch": 2.4360911343445943,
      "grad_norm": 1.1968594944459072,
      "learning_rate": 9.353098744491046e-07,
      "loss": 0.8694,
      "step": 40310
    },
    {
      "epoch": 2.436695473499728,
      "grad_norm": 1.1535253230895357,
      "learning_rate": 9.333710132633328e-07,
      "loss": 0.8801,
      "step": 40320
    },
    {
      "epoch": 2.437299812654862,
      "grad_norm": 1.1101329881597528,
      "learning_rate": 9.314339568544673e-07,
      "loss": 0.8847,
      "step": 40330
    },
    {
      "epoch": 2.4379041518099958,
      "grad_norm": 1.1202104735969172,
      "learning_rate": 9.294987060821775e-07,
      "loss": 0.8817,
      "step": 40340
    },
    {
      "epoch": 2.4385084909651296,
      "grad_norm": 1.1472877529301357,
      "learning_rate": 9.275652618053294e-07,
      "loss": 0.8971,
      "step": 40350
    },
    {
      "epoch": 2.4391128301202634,
      "grad_norm": 1.14996637819073,
      "learning_rate": 9.256336248819891e-07,
      "loss": 0.8818,
      "step": 40360
    },
    {
      "epoch": 2.439717169275397,
      "grad_norm": 0.9841776466953165,
      "learning_rate": 9.237037961694223e-07,
      "loss": 0.8831,
      "step": 40370
    },
    {
      "epoch": 2.4403215084305314,
      "grad_norm": 1.0384358238508726,
      "learning_rate": 9.217757765240876e-07,
      "loss": 0.8866,
      "step": 40380
    },
    {
      "epoch": 2.440925847585665,
      "grad_norm": 0.9709337804878487,
      "learning_rate": 9.19849566801645e-07,
      "loss": 0.8893,
      "step": 40390
    },
    {
      "epoch": 2.441530186740799,
      "grad_norm": 1.002710911705235,
      "learning_rate": 9.179251678569501e-07,
      "loss": 0.9143,
      "step": 40400
    },
    {
      "epoch": 2.442134525895933,
      "grad_norm": 1.0570992077629024,
      "learning_rate": 9.160025805440526e-07,
      "loss": 0.8816,
      "step": 40410
    },
    {
      "epoch": 2.4427388650510666,
      "grad_norm": 1.0381080922769719,
      "learning_rate": 9.14081805716201e-07,
      "loss": 0.8748,
      "step": 40420
    },
    {
      "epoch": 2.4433432042062004,
      "grad_norm": 1.1464506880121867,
      "learning_rate": 9.121628442258396e-07,
      "loss": 0.8737,
      "step": 40430
    },
    {
      "epoch": 2.4439475433613342,
      "grad_norm": 1.024202977282041,
      "learning_rate": 9.102456969246048e-07,
      "loss": 0.8936,
      "step": 40440
    },
    {
      "epoch": 2.4445518825164685,
      "grad_norm": 1.0392841490146056,
      "learning_rate": 9.083303646633323e-07,
      "loss": 0.8824,
      "step": 40450
    },
    {
      "epoch": 2.4451562216716023,
      "grad_norm": 0.9892722915030926,
      "learning_rate": 9.064168482920472e-07,
      "loss": 0.8893,
      "step": 40460
    },
    {
      "epoch": 2.445760560826736,
      "grad_norm": 1.0548562648819615,
      "learning_rate": 9.045051486599732e-07,
      "loss": 0.881,
      "step": 40470
    },
    {
      "epoch": 2.44636489998187,
      "grad_norm": 1.1086944823651066,
      "learning_rate": 9.025952666155242e-07,
      "loss": 0.8829,
      "step": 40480
    },
    {
      "epoch": 2.4469692391370037,
      "grad_norm": 1.0127080923354776,
      "learning_rate": 9.006872030063101e-07,
      "loss": 0.906,
      "step": 40490
    },
    {
      "epoch": 2.4475735782921375,
      "grad_norm": 1.2698491151972442,
      "learning_rate": 8.987809586791341e-07,
      "loss": 0.8713,
      "step": 40500
    },
    {
      "epoch": 2.4481779174472713,
      "grad_norm": 1.1991252703491568,
      "learning_rate": 8.968765344799879e-07,
      "loss": 0.8815,
      "step": 40510
    },
    {
      "epoch": 2.448782256602405,
      "grad_norm": 1.2431552513571238,
      "learning_rate": 8.949739312540601e-07,
      "loss": 0.8935,
      "step": 40520
    },
    {
      "epoch": 2.449386595757539,
      "grad_norm": 1.1432295897574343,
      "learning_rate": 8.930731498457296e-07,
      "loss": 0.844,
      "step": 40530
    },
    {
      "epoch": 2.449990934912673,
      "grad_norm": 1.2028449447048442,
      "learning_rate": 8.91174191098565e-07,
      "loss": 0.888,
      "step": 40540
    },
    {
      "epoch": 2.450595274067807,
      "grad_norm": 1.1523326084546026,
      "learning_rate": 8.892770558553276e-07,
      "loss": 0.8849,
      "step": 40550
    },
    {
      "epoch": 2.4511996132229408,
      "grad_norm": 1.1790556802267356,
      "learning_rate": 8.873817449579713e-07,
      "loss": 0.8823,
      "step": 40560
    },
    {
      "epoch": 2.4518039523780746,
      "grad_norm": 1.1555191092906105,
      "learning_rate": 8.85488259247636e-07,
      "loss": 0.9068,
      "step": 40570
    },
    {
      "epoch": 2.4524082915332084,
      "grad_norm": 1.082071714979443,
      "learning_rate": 8.835965995646544e-07,
      "loss": 0.8872,
      "step": 40580
    },
    {
      "epoch": 2.453012630688342,
      "grad_norm": 1.1836829788332326,
      "learning_rate": 8.817067667485496e-07,
      "loss": 0.8915,
      "step": 40590
    },
    {
      "epoch": 2.453616969843476,
      "grad_norm": 1.274098974946871,
      "learning_rate": 8.798187616380311e-07,
      "loss": 0.8847,
      "step": 40600
    },
    {
      "epoch": 2.4542213089986102,
      "grad_norm": 1.2647313993593425,
      "learning_rate": 8.779325850709991e-07,
      "loss": 0.8868,
      "step": 40610
    },
    {
      "epoch": 2.454825648153744,
      "grad_norm": 1.2301879884850013,
      "learning_rate": 8.760482378845431e-07,
      "loss": 0.8937,
      "step": 40620
    },
    {
      "epoch": 2.455429987308878,
      "grad_norm": 1.173147950355316,
      "learning_rate": 8.741657209149379e-07,
      "loss": 0.8926,
      "step": 40630
    },
    {
      "epoch": 2.4560343264640117,
      "grad_norm": 1.1804737786496446,
      "learning_rate": 8.722850349976492e-07,
      "loss": 0.8568,
      "step": 40640
    },
    {
      "epoch": 2.4566386656191455,
      "grad_norm": 1.4567761868521123,
      "learning_rate": 8.704061809673286e-07,
      "loss": 0.8648,
      "step": 40650
    },
    {
      "epoch": 2.4572430047742793,
      "grad_norm": 1.532856362103709,
      "learning_rate": 8.685291596578132e-07,
      "loss": 0.917,
      "step": 40660
    },
    {
      "epoch": 2.457847343929413,
      "grad_norm": 1.4067824423288466,
      "learning_rate": 8.666539719021288e-07,
      "loss": 0.8937,
      "step": 40670
    },
    {
      "epoch": 2.458451683084547,
      "grad_norm": 1.4894543584209463,
      "learning_rate": 8.647806185324892e-07,
      "loss": 0.9166,
      "step": 40680
    },
    {
      "epoch": 2.4590560222396807,
      "grad_norm": 1.4080180127720432,
      "learning_rate": 8.629091003802892e-07,
      "loss": 0.8674,
      "step": 40690
    },
    {
      "epoch": 2.459660361394815,
      "grad_norm": 0.9550241482039177,
      "learning_rate": 8.610394182761128e-07,
      "loss": 0.8789,
      "step": 40700
    },
    {
      "epoch": 2.4602647005499487,
      "grad_norm": 0.9170315185809859,
      "learning_rate": 8.591715730497275e-07,
      "loss": 0.8914,
      "step": 40710
    },
    {
      "epoch": 2.4608690397050825,
      "grad_norm": 0.9114295602852198,
      "learning_rate": 8.573055655300877e-07,
      "loss": 0.8684,
      "step": 40720
    },
    {
      "epoch": 2.4614733788602163,
      "grad_norm": 0.905964321453953,
      "learning_rate": 8.554413965453284e-07,
      "loss": 0.8666,
      "step": 40730
    },
    {
      "epoch": 2.46207771801535,
      "grad_norm": 0.9862078466810149,
      "learning_rate": 8.535790669227734e-07,
      "loss": 0.873,
      "step": 40740
    },
    {
      "epoch": 2.462682057170484,
      "grad_norm": 0.8102293660945225,
      "learning_rate": 8.517185774889253e-07,
      "loss": 0.9058,
      "step": 40750
    },
    {
      "epoch": 2.4632863963256177,
      "grad_norm": 0.8671735721374867,
      "learning_rate": 8.498599290694737e-07,
      "loss": 0.8935,
      "step": 40760
    },
    {
      "epoch": 2.463890735480752,
      "grad_norm": 0.9155077692543427,
      "learning_rate": 8.480031224892903e-07,
      "loss": 0.8999,
      "step": 40770
    },
    {
      "epoch": 2.464495074635886,
      "grad_norm": 0.8800269468239498,
      "learning_rate": 8.461481585724302e-07,
      "loss": 0.9024,
      "step": 40780
    },
    {
      "epoch": 2.4650994137910196,
      "grad_norm": 0.7967158437233781,
      "learning_rate": 8.442950381421266e-07,
      "loss": 0.8722,
      "step": 40790
    },
    {
      "epoch": 2.4657037529461534,
      "grad_norm": 0.8383316828539763,
      "learning_rate": 8.424437620207999e-07,
      "loss": 0.869,
      "step": 40800
    },
    {
      "epoch": 2.466308092101287,
      "grad_norm": 0.8556933488620937,
      "learning_rate": 8.40594331030049e-07,
      "loss": 0.8976,
      "step": 40810
    },
    {
      "epoch": 2.466912431256421,
      "grad_norm": 0.8890821834570544,
      "learning_rate": 8.387467459906534e-07,
      "loss": 0.8897,
      "step": 40820
    },
    {
      "epoch": 2.467516770411555,
      "grad_norm": 0.9267038280657559,
      "learning_rate": 8.369010077225759e-07,
      "loss": 0.8711,
      "step": 40830
    },
    {
      "epoch": 2.468121109566689,
      "grad_norm": 0.758990858193993,
      "learning_rate": 8.350571170449584e-07,
      "loss": 0.8776,
      "step": 40840
    },
    {
      "epoch": 2.468725448721823,
      "grad_norm": 0.6752216215519213,
      "learning_rate": 8.332150747761214e-07,
      "loss": 0.9009,
      "step": 40850
    },
    {
      "epoch": 2.4693297878769567,
      "grad_norm": 0.6350222527973223,
      "learning_rate": 8.313748817335671e-07,
      "loss": 0.8785,
      "step": 40860
    },
    {
      "epoch": 2.4699341270320905,
      "grad_norm": 0.6090962901780818,
      "learning_rate": 8.295365387339777e-07,
      "loss": 0.857,
      "step": 40870
    },
    {
      "epoch": 2.4705384661872243,
      "grad_norm": 0.6857908201667366,
      "learning_rate": 8.277000465932106e-07,
      "loss": 0.8858,
      "step": 40880
    },
    {
      "epoch": 2.471142805342358,
      "grad_norm": 0.6642855637090372,
      "learning_rate": 8.25865406126305e-07,
      "loss": 0.8839,
      "step": 40890
    },
    {
      "epoch": 2.471747144497492,
      "grad_norm": 0.7539847947975113,
      "learning_rate": 8.240326181474789e-07,
      "loss": 0.8873,
      "step": 40900
    },
    {
      "epoch": 2.4723514836526257,
      "grad_norm": 0.6629985866564397,
      "learning_rate": 8.222016834701247e-07,
      "loss": 0.8835,
      "step": 40910
    },
    {
      "epoch": 2.4729558228077595,
      "grad_norm": 0.7263953374517576,
      "learning_rate": 8.203726029068149e-07,
      "loss": 0.9003,
      "step": 40920
    },
    {
      "epoch": 2.4735601619628937,
      "grad_norm": 0.6831832263349958,
      "learning_rate": 8.185453772693003e-07,
      "loss": 0.8883,
      "step": 40930
    },
    {
      "epoch": 2.4741645011180275,
      "grad_norm": 0.6567853640438026,
      "learning_rate": 8.167200073685039e-07,
      "loss": 0.8729,
      "step": 40940
    },
    {
      "epoch": 2.4747688402731614,
      "grad_norm": 0.8772422402015878,
      "learning_rate": 8.14896494014531e-07,
      "loss": 0.8885,
      "step": 40950
    },
    {
      "epoch": 2.475373179428295,
      "grad_norm": 0.9163396307998091,
      "learning_rate": 8.13074838016657e-07,
      "loss": 0.9038,
      "step": 40960
    },
    {
      "epoch": 2.475977518583429,
      "grad_norm": 0.9489757718026602,
      "learning_rate": 8.112550401833385e-07,
      "loss": 0.898,
      "step": 40970
    },
    {
      "epoch": 2.4765818577385628,
      "grad_norm": 0.8717296378538169,
      "learning_rate": 8.09437101322203e-07,
      "loss": 0.8783,
      "step": 40980
    },
    {
      "epoch": 2.4771861968936966,
      "grad_norm": 0.8241932637516207,
      "learning_rate": 8.076210222400566e-07,
      "loss": 0.8838,
      "step": 40990
    },
    {
      "epoch": 2.477790536048831,
      "grad_norm": 0.6112766384950763,
      "learning_rate": 8.058068037428757e-07,
      "loss": 0.8835,
      "step": 41000
    },
    {
      "epoch": 2.4783948752039646,
      "grad_norm": 0.5568956321952637,
      "learning_rate": 8.039944466358157e-07,
      "loss": 0.8875,
      "step": 41010
    },
    {
      "epoch": 2.4789992143590984,
      "grad_norm": 0.6046399911718908,
      "learning_rate": 8.021839517232044e-07,
      "loss": 0.8825,
      "step": 41020
    },
    {
      "epoch": 2.4796035535142322,
      "grad_norm": 0.5788564076338315,
      "learning_rate": 8.003753198085401e-07,
      "loss": 0.8876,
      "step": 41030
    },
    {
      "epoch": 2.480207892669366,
      "grad_norm": 0.571656490361716,
      "learning_rate": 7.98568551694498e-07,
      "loss": 0.8819,
      "step": 41040
    },
    {
      "epoch": 2.4808122318245,
      "grad_norm": 0.6007697435233759,
      "learning_rate": 7.967636481829261e-07,
      "loss": 0.9112,
      "step": 41050
    },
    {
      "epoch": 2.4814165709796336,
      "grad_norm": 0.58364374425912,
      "learning_rate": 7.949606100748414e-07,
      "loss": 0.8892,
      "step": 41060
    },
    {
      "epoch": 2.4820209101347674,
      "grad_norm": 0.6208364335185198,
      "learning_rate": 7.931594381704366e-07,
      "loss": 0.9108,
      "step": 41070
    },
    {
      "epoch": 2.4826252492899012,
      "grad_norm": 0.601545399669037,
      "learning_rate": 7.913601332690757e-07,
      "loss": 0.8772,
      "step": 41080
    },
    {
      "epoch": 2.4832295884450355,
      "grad_norm": 0.600802974725024,
      "learning_rate": 7.895626961692915e-07,
      "loss": 0.8986,
      "step": 41090
    },
    {
      "epoch": 2.4838339276001693,
      "grad_norm": 0.607832178324593,
      "learning_rate": 7.877671276687899e-07,
      "loss": 0.8651,
      "step": 41100
    },
    {
      "epoch": 2.484438266755303,
      "grad_norm": 0.6082153709049353,
      "learning_rate": 7.859734285644483e-07,
      "loss": 0.8808,
      "step": 41110
    },
    {
      "epoch": 2.485042605910437,
      "grad_norm": 0.574824514043198,
      "learning_rate": 7.841815996523138e-07,
      "loss": 0.8842,
      "step": 41120
    },
    {
      "epoch": 2.4856469450655707,
      "grad_norm": 0.6439760165299043,
      "learning_rate": 7.823916417276012e-07,
      "loss": 0.8874,
      "step": 41130
    },
    {
      "epoch": 2.4862512842207045,
      "grad_norm": 0.5947019040844654,
      "learning_rate": 7.80603555584698e-07,
      "loss": 0.8586,
      "step": 41140
    },
    {
      "epoch": 2.4868556233758383,
      "grad_norm": 0.6374303381339865,
      "learning_rate": 7.788173420171613e-07,
      "loss": 0.9056,
      "step": 41150
    },
    {
      "epoch": 2.4874599625309726,
      "grad_norm": 0.6085257937966161,
      "learning_rate": 7.77033001817713e-07,
      "loss": 0.9013,
      "step": 41160
    },
    {
      "epoch": 2.4880643016861064,
      "grad_norm": 0.5877577075095965,
      "learning_rate": 7.752505357782475e-07,
      "loss": 0.8769,
      "step": 41170
    },
    {
      "epoch": 2.48866864084124,
      "grad_norm": 0.5851397229057675,
      "learning_rate": 7.734699446898275e-07,
      "loss": 0.8672,
      "step": 41180
    },
    {
      "epoch": 2.489272979996374,
      "grad_norm": 0.6239213774782675,
      "learning_rate": 7.716912293426804e-07,
      "loss": 0.877,
      "step": 41190
    },
    {
      "epoch": 2.489877319151508,
      "grad_norm": 0.5607932588278917,
      "learning_rate": 7.699143905262052e-07,
      "loss": 0.873,
      "step": 41200
    },
    {
      "epoch": 2.4904816583066416,
      "grad_norm": 0.5684729240308406,
      "learning_rate": 7.681394290289634e-07,
      "loss": 0.8824,
      "step": 41210
    },
    {
      "epoch": 2.4910859974617754,
      "grad_norm": 0.5746386240729316,
      "learning_rate": 7.663663456386889e-07,
      "loss": 0.8799,
      "step": 41220
    },
    {
      "epoch": 2.4916903366169096,
      "grad_norm": 0.6256627291758511,
      "learning_rate": 7.645951411422764e-07,
      "loss": 0.8767,
      "step": 41230
    },
    {
      "epoch": 2.4922946757720434,
      "grad_norm": 0.5429073102730873,
      "learning_rate": 7.628258163257918e-07,
      "loss": 0.8826,
      "step": 41240
    },
    {
      "epoch": 2.4928990149271772,
      "grad_norm": 0.6055282159058301,
      "learning_rate": 7.610583719744625e-07,
      "loss": 0.8782,
      "step": 41250
    },
    {
      "epoch": 2.493503354082311,
      "grad_norm": 0.6130839462133468,
      "learning_rate": 7.592928088726837e-07,
      "loss": 0.8743,
      "step": 41260
    },
    {
      "epoch": 2.494107693237445,
      "grad_norm": 0.5888237331924623,
      "learning_rate": 7.57529127804017e-07,
      "loss": 0.8615,
      "step": 41270
    },
    {
      "epoch": 2.4947120323925787,
      "grad_norm": 0.5806026114001572,
      "learning_rate": 7.557673295511853e-07,
      "loss": 0.8764,
      "step": 41280
    },
    {
      "epoch": 2.4953163715477125,
      "grad_norm": 0.5888561537060585,
      "learning_rate": 7.54007414896078e-07,
      "loss": 0.8927,
      "step": 41290
    },
    {
      "epoch": 2.4959207107028463,
      "grad_norm": 0.6635875796227327,
      "learning_rate": 7.522493846197498e-07,
      "loss": 0.8931,
      "step": 41300
    },
    {
      "epoch": 2.49652504985798,
      "grad_norm": 0.7117084599458885,
      "learning_rate": 7.50493239502415e-07,
      "loss": 0.867,
      "step": 41310
    },
    {
      "epoch": 2.4971293890131143,
      "grad_norm": 0.6801507717611001,
      "learning_rate": 7.487389803234552e-07,
      "loss": 0.8589,
      "step": 41320
    },
    {
      "epoch": 2.497733728168248,
      "grad_norm": 0.6690993993859737,
      "learning_rate": 7.469866078614146e-07,
      "loss": 0.8773,
      "step": 41330
    },
    {
      "epoch": 2.498338067323382,
      "grad_norm": 0.6857810396814289,
      "learning_rate": 7.452361228939975e-07,
      "loss": 0.8977,
      "step": 41340
    },
    {
      "epoch": 2.4989424064785157,
      "grad_norm": 0.7193851079727911,
      "learning_rate": 7.434875261980717e-07,
      "loss": 0.893,
      "step": 41350
    },
    {
      "epoch": 2.4995467456336495,
      "grad_norm": 0.7771860953543476,
      "learning_rate": 7.417408185496699e-07,
      "loss": 0.8719,
      "step": 41360
    },
    {
      "epoch": 2.5001510847887833,
      "grad_norm": 0.720979980458029,
      "learning_rate": 7.399960007239815e-07,
      "loss": 0.8971,
      "step": 41370
    },
    {
      "epoch": 2.500755423943917,
      "grad_norm": 0.7508571703625692,
      "learning_rate": 7.382530734953602e-07,
      "loss": 0.8547,
      "step": 41380
    },
    {
      "epoch": 2.5013597630990514,
      "grad_norm": 0.7042127820803035,
      "learning_rate": 7.365120376373219e-07,
      "loss": 0.8799,
      "step": 41390
    },
    {
      "epoch": 2.501964102254185,
      "grad_norm": 0.6328680950770649,
      "learning_rate": 7.347728939225385e-07,
      "loss": 0.8674,
      "step": 41400
    },
    {
      "epoch": 2.502568441409319,
      "grad_norm": 0.6865609669511411,
      "learning_rate": 7.330356431228464e-07,
      "loss": 0.8969,
      "step": 41410
    },
    {
      "epoch": 2.503172780564453,
      "grad_norm": 0.690132915780125,
      "learning_rate": 7.313002860092411e-07,
      "loss": 0.8752,
      "step": 41420
    },
    {
      "epoch": 2.5037771197195866,
      "grad_norm": 0.7021186612842231,
      "learning_rate": 7.295668233518755e-07,
      "loss": 0.894,
      "step": 41430
    },
    {
      "epoch": 2.5043814588747204,
      "grad_norm": 0.6851340951751752,
      "learning_rate": 7.27835255920064e-07,
      "loss": 0.8744,
      "step": 41440
    },
    {
      "epoch": 2.504985798029854,
      "grad_norm": 0.8410684795933202,
      "learning_rate": 7.261055844822795e-07,
      "loss": 0.871,
      "step": 41450
    },
    {
      "epoch": 2.5055901371849885,
      "grad_norm": 0.9147063038305515,
      "learning_rate": 7.243778098061544e-07,
      "loss": 0.8814,
      "step": 41460
    },
    {
      "epoch": 2.506194476340122,
      "grad_norm": 0.8451077764037075,
      "learning_rate": 7.226519326584746e-07,
      "loss": 0.8592,
      "step": 41470
    },
    {
      "epoch": 2.506798815495256,
      "grad_norm": 0.8623229815034531,
      "learning_rate": 7.209279538051911e-07,
      "loss": 0.8701,
      "step": 41480
    },
    {
      "epoch": 2.50740315465039,
      "grad_norm": 0.8513258955684797,
      "learning_rate": 7.192058740114055e-07,
      "loss": 0.8706,
      "step": 41490
    },
    {
      "epoch": 2.5080074938055237,
      "grad_norm": 0.7778377032944958,
      "learning_rate": 7.174856940413826e-07,
      "loss": 0.8934,
      "step": 41500
    },
    {
      "epoch": 2.5086118329606575,
      "grad_norm": 0.8225250114310346,
      "learning_rate": 7.157674146585386e-07,
      "loss": 0.8808,
      "step": 41510
    },
    {
      "epoch": 2.5092161721157913,
      "grad_norm": 0.8286421025559088,
      "learning_rate": 7.14051036625451e-07,
      "loss": 0.8767,
      "step": 41520
    },
    {
      "epoch": 2.509820511270925,
      "grad_norm": 0.8723314979501335,
      "learning_rate": 7.12336560703849e-07,
      "loss": 0.8843,
      "step": 41530
    },
    {
      "epoch": 2.510424850426059,
      "grad_norm": 0.8217767532029071,
      "learning_rate": 7.106239876546217e-07,
      "loss": 0.9186,
      "step": 41540
    },
    {
      "epoch": 2.511029189581193,
      "grad_norm": 1.3176567746767818,
      "learning_rate": 7.089133182378122e-07,
      "loss": 0.8891,
      "step": 41550
    },
    {
      "epoch": 2.511633528736327,
      "grad_norm": 1.2262934350892705,
      "learning_rate": 7.07204553212617e-07,
      "loss": 0.91,
      "step": 41560
    },
    {
      "epoch": 2.5122378678914608,
      "grad_norm": 1.3440158903501784,
      "learning_rate": 7.054976933373897e-07,
      "loss": 0.8893,
      "step": 41570
    },
    {
      "epoch": 2.5128422070465946,
      "grad_norm": 1.233015173445002,
      "learning_rate": 7.037927393696392e-07,
      "loss": 0.8624,
      "step": 41580
    },
    {
      "epoch": 2.5134465462017284,
      "grad_norm": 1.2550374059816882,
      "learning_rate": 7.020896920660242e-07,
      "loss": 0.8986,
      "step": 41590
    },
    {
      "epoch": 2.514050885356862,
      "grad_norm": 0.9814955904656175,
      "learning_rate": 7.003885521823611e-07,
      "loss": 0.8833,
      "step": 41600
    },
    {
      "epoch": 2.514655224511996,
      "grad_norm": 0.905318800146908,
      "learning_rate": 6.9868932047362e-07,
      "loss": 0.8778,
      "step": 41610
    },
    {
      "epoch": 2.51525956366713,
      "grad_norm": 1.1544488775909953,
      "learning_rate": 6.969919976939205e-07,
      "loss": 0.8628,
      "step": 41620
    },
    {
      "epoch": 2.515863902822264,
      "grad_norm": 1.1138147357852284,
      "learning_rate": 6.952965845965393e-07,
      "loss": 0.874,
      "step": 41630
    },
    {
      "epoch": 2.516468241977398,
      "grad_norm": 1.0481084841215758,
      "learning_rate": 6.936030819339035e-07,
      "loss": 0.8833,
      "step": 41640
    },
    {
      "epoch": 2.5170725811325316,
      "grad_norm": 2.4505905400521235,
      "learning_rate": 6.919114904575908e-07,
      "loss": 0.8879,
      "step": 41650
    },
    {
      "epoch": 2.5176769202876654,
      "grad_norm": 2.6286040442490655,
      "learning_rate": 6.902218109183334e-07,
      "loss": 0.8895,
      "step": 41660
    },
    {
      "epoch": 2.5182812594427992,
      "grad_norm": 2.4166159298620835,
      "learning_rate": 6.885340440660154e-07,
      "loss": 0.8793,
      "step": 41670
    },
    {
      "epoch": 2.518885598597933,
      "grad_norm": 2.8147112186088026,
      "learning_rate": 6.868481906496683e-07,
      "loss": 0.8787,
      "step": 41680
    },
    {
      "epoch": 2.5194899377530673,
      "grad_norm": 2.374540844592351,
      "learning_rate": 6.851642514174778e-07,
      "loss": 0.9291,
      "step": 41690
    },
    {
      "epoch": 2.5200942769082006,
      "grad_norm": 0.9637552808959402,
      "learning_rate": 6.834822271167801e-07,
      "loss": 0.8771,
      "step": 41700
    },
    {
      "epoch": 2.520698616063335,
      "grad_norm": 0.9432855559774069,
      "learning_rate": 6.818021184940582e-07,
      "loss": 0.8665,
      "step": 41710
    },
    {
      "epoch": 2.5213029552184687,
      "grad_norm": 0.9156381976104924,
      "learning_rate": 6.801239262949477e-07,
      "loss": 0.8757,
      "step": 41720
    },
    {
      "epoch": 2.5219072943736025,
      "grad_norm": 1.010610201588485,
      "learning_rate": 6.784476512642357e-07,
      "loss": 0.8848,
      "step": 41730
    },
    {
      "epoch": 2.5225116335287363,
      "grad_norm": 0.9177307480678155,
      "learning_rate": 6.767732941458521e-07,
      "loss": 0.8508,
      "step": 41740
    },
    {
      "epoch": 2.52311597268387,
      "grad_norm": 1.0095202052263494,
      "learning_rate": 6.751008556828831e-07,
      "loss": 0.8722,
      "step": 41750
    },
    {
      "epoch": 2.523720311839004,
      "grad_norm": 0.963805904401234,
      "learning_rate": 6.734303366175565e-07,
      "loss": 0.8864,
      "step": 41760
    },
    {
      "epoch": 2.5243246509941377,
      "grad_norm": 1.0159372246572098,
      "learning_rate": 6.717617376912545e-07,
      "loss": 0.8843,
      "step": 41770
    },
    {
      "epoch": 2.524928990149272,
      "grad_norm": 1.0213444690102842,
      "learning_rate": 6.70095059644501e-07,
      "loss": 0.8677,
      "step": 41780
    },
    {
      "epoch": 2.5255333293044058,
      "grad_norm": 1.0811498914152846,
      "learning_rate": 6.684303032169732e-07,
      "loss": 0.891,
      "step": 41790
    },
    {
      "epoch": 2.5261376684595396,
      "grad_norm": 1.059908932999426,
      "learning_rate": 6.667674691474923e-07,
      "loss": 0.8971,
      "step": 41800
    },
    {
      "epoch": 2.5267420076146734,
      "grad_norm": 1.1106257810913194,
      "learning_rate": 6.651065581740257e-07,
      "loss": 0.9171,
      "step": 41810
    },
    {
      "epoch": 2.527346346769807,
      "grad_norm": 1.0327366653796355,
      "learning_rate": 6.634475710336891e-07,
      "loss": 0.881,
      "step": 41820
    },
    {
      "epoch": 2.527950685924941,
      "grad_norm": 0.9996554903594473,
      "learning_rate": 6.617905084627452e-07,
      "loss": 0.8947,
      "step": 41830
    },
    {
      "epoch": 2.528555025080075,
      "grad_norm": 1.0182575251216492,
      "learning_rate": 6.601353711965991e-07,
      "loss": 0.8815,
      "step": 41840
    },
    {
      "epoch": 2.529159364235209,
      "grad_norm": 1.052178439056579,
      "learning_rate": 6.584821599698043e-07,
      "loss": 0.8888,
      "step": 41850
    },
    {
      "epoch": 2.5297637033903424,
      "grad_norm": 1.1635945387098245,
      "learning_rate": 6.568308755160607e-07,
      "loss": 0.8844,
      "step": 41860
    },
    {
      "epoch": 2.5303680425454766,
      "grad_norm": 1.039975162307649,
      "learning_rate": 6.551815185682076e-07,
      "loss": 0.8893,
      "step": 41870
    },
    {
      "epoch": 2.5309723817006105,
      "grad_norm": 1.1949681295089856,
      "learning_rate": 6.535340898582348e-07,
      "loss": 0.9057,
      "step": 41880
    },
    {
      "epoch": 2.5315767208557443,
      "grad_norm": 1.1497988985003074,
      "learning_rate": 6.518885901172744e-07,
      "loss": 0.9093,
      "step": 41890
    },
    {
      "epoch": 2.532181060010878,
      "grad_norm": 1.212665791374069,
      "learning_rate": 6.502450200756005e-07,
      "loss": 0.8834,
      "step": 41900
    },
    {
      "epoch": 2.532785399166012,
      "grad_norm": 1.101735350294155,
      "learning_rate": 6.486033804626324e-07,
      "loss": 0.8831,
      "step": 41910
    },
    {
      "epoch": 2.5333897383211457,
      "grad_norm": 1.098056204967435,
      "learning_rate": 6.469636720069344e-07,
      "loss": 0.8884,
      "step": 41920
    },
    {
      "epoch": 2.5339940774762795,
      "grad_norm": 1.0972572497456845,
      "learning_rate": 6.453258954362091e-07,
      "loss": 0.8803,
      "step": 41930
    },
    {
      "epoch": 2.5345984166314137,
      "grad_norm": 1.1375922649610486,
      "learning_rate": 6.436900514773064e-07,
      "loss": 0.882,
      "step": 41940
    },
    {
      "epoch": 2.5352027557865475,
      "grad_norm": 1.0008639292983053,
      "learning_rate": 6.420561408562176e-07,
      "loss": 0.8748,
      "step": 41950
    },
    {
      "epoch": 2.5358070949416813,
      "grad_norm": 1.0625468113826864,
      "learning_rate": 6.404241642980719e-07,
      "loss": 0.8733,
      "step": 41960
    },
    {
      "epoch": 2.536411434096815,
      "grad_norm": 1.0124885960100483,
      "learning_rate": 6.387941225271454e-07,
      "loss": 0.8765,
      "step": 41970
    },
    {
      "epoch": 2.537015773251949,
      "grad_norm": 1.0346608631156706,
      "learning_rate": 6.371660162668547e-07,
      "loss": 0.9161,
      "step": 41980
    },
    {
      "epoch": 2.5376201124070827,
      "grad_norm": 1.0502572790662368,
      "learning_rate": 6.355398462397533e-07,
      "loss": 0.8675,
      "step": 41990
    },
    {
      "epoch": 2.5382244515622165,
      "grad_norm": 1.1584241430956357,
      "learning_rate": 6.3391561316754e-07,
      "loss": 0.8997,
      "step": 42000
    },
    {
      "epoch": 2.538828790717351,
      "grad_norm": 1.1483278089411093,
      "learning_rate": 6.322933177710511e-07,
      "loss": 0.876,
      "step": 42010
    },
    {
      "epoch": 2.5394331298724846,
      "grad_norm": 1.0485668810213669,
      "learning_rate": 6.30672960770266e-07,
      "loss": 0.881,
      "step": 42020
    },
    {
      "epoch": 2.5400374690276184,
      "grad_norm": 1.1212436329522641,
      "learning_rate": 6.290545428842992e-07,
      "loss": 0.8906,
      "step": 42030
    },
    {
      "epoch": 2.540641808182752,
      "grad_norm": 1.0908228516654648,
      "learning_rate": 6.274380648314099e-07,
      "loss": 0.8777,
      "step": 42040
    },
    {
      "epoch": 2.541246147337886,
      "grad_norm": 1.023040896025016,
      "learning_rate": 6.25823527328992e-07,
      "loss": 0.8981,
      "step": 42050
    },
    {
      "epoch": 2.54185048649302,
      "grad_norm": 1.1038122647020336,
      "learning_rate": 6.242109310935801e-07,
      "loss": 0.9005,
      "step": 42060
    },
    {
      "epoch": 2.5424548256481536,
      "grad_norm": 1.095077367570081,
      "learning_rate": 6.22600276840849e-07,
      "loss": 0.889,
      "step": 42070
    },
    {
      "epoch": 2.543059164803288,
      "grad_norm": 1.0331561983218978,
      "learning_rate": 6.209915652856074e-07,
      "loss": 0.9056,
      "step": 42080
    },
    {
      "epoch": 2.5436635039584212,
      "grad_norm": 1.0257322683385564,
      "learning_rate": 6.19384797141806e-07,
      "loss": 0.8698,
      "step": 42090
    },
    {
      "epoch": 2.5442678431135555,
      "grad_norm": 1.2304851630039246,
      "learning_rate": 6.177799731225315e-07,
      "loss": 0.8865,
      "step": 42100
    },
    {
      "epoch": 2.5448721822686893,
      "grad_norm": 1.1305342969891252,
      "learning_rate": 6.161770939400052e-07,
      "loss": 0.8599,
      "step": 42110
    },
    {
      "epoch": 2.545476521423823,
      "grad_norm": 1.2080891922429504,
      "learning_rate": 6.145761603055894e-07,
      "loss": 0.8815,
      "step": 42120
    },
    {
      "epoch": 2.546080860578957,
      "grad_norm": 1.2639039914736894,
      "learning_rate": 6.129771729297806e-07,
      "loss": 0.8802,
      "step": 42130
    },
    {
      "epoch": 2.5466851997340907,
      "grad_norm": 1.222373367090069,
      "learning_rate": 6.11380132522214e-07,
      "loss": 0.8758,
      "step": 42140
    },
    {
      "epoch": 2.5472895388892245,
      "grad_norm": 1.2147533764602698,
      "learning_rate": 6.097850397916561e-07,
      "loss": 0.8916,
      "step": 42150
    },
    {
      "epoch": 2.5478938780443583,
      "grad_norm": 1.1105430745799367,
      "learning_rate": 6.081918954460131e-07,
      "loss": 0.8801,
      "step": 42160
    },
    {
      "epoch": 2.5484982171994925,
      "grad_norm": 1.1327420871134162,
      "learning_rate": 6.066007001923263e-07,
      "loss": 0.8708,
      "step": 42170
    },
    {
      "epoch": 2.5491025563546263,
      "grad_norm": 1.0907110284284494,
      "learning_rate": 6.050114547367681e-07,
      "loss": 0.8772,
      "step": 42180
    },
    {
      "epoch": 2.54970689550976,
      "grad_norm": 1.1736055008192041,
      "learning_rate": 6.034241597846502e-07,
      "loss": 0.8844,
      "step": 42190
    },
    {
      "epoch": 2.550311234664894,
      "grad_norm": 1.1778309924846977,
      "learning_rate": 6.018388160404176e-07,
      "loss": 0.8752,
      "step": 42200
    },
    {
      "epoch": 2.5509155738200278,
      "grad_norm": 1.2295835842125538,
      "learning_rate": 6.002554242076464e-07,
      "loss": 0.8853,
      "step": 42210
    },
    {
      "epoch": 2.5515199129751616,
      "grad_norm": 1.1878022088962958,
      "learning_rate": 5.986739849890499e-07,
      "loss": 0.903,
      "step": 42220
    },
    {
      "epoch": 2.5521242521302954,
      "grad_norm": 1.272235646274948,
      "learning_rate": 5.970944990864747e-07,
      "loss": 0.8968,
      "step": 42230
    },
    {
      "epoch": 2.5527285912854296,
      "grad_norm": 1.1609381699830803,
      "learning_rate": 5.955169672008965e-07,
      "loss": 0.8673,
      "step": 42240
    },
    {
      "epoch": 2.553332930440563,
      "grad_norm": 1.436489165843789,
      "learning_rate": 5.939413900324292e-07,
      "loss": 0.865,
      "step": 42250
    },
    {
      "epoch": 2.5539372695956972,
      "grad_norm": 1.557865762308893,
      "learning_rate": 5.923677682803148e-07,
      "loss": 0.8939,
      "step": 42260
    },
    {
      "epoch": 2.554541608750831,
      "grad_norm": 1.4514644540897934,
      "learning_rate": 5.907961026429304e-07,
      "loss": 0.8742,
      "step": 42270
    },
    {
      "epoch": 2.555145947905965,
      "grad_norm": 1.4775760792547543,
      "learning_rate": 5.892263938177833e-07,
      "loss": 0.897,
      "step": 42280
    },
    {
      "epoch": 2.5557502870610986,
      "grad_norm": 1.5837154702920653,
      "learning_rate": 5.87658642501514e-07,
      "loss": 0.8827,
      "step": 42290
    },
    {
      "epoch": 2.5563546262162324,
      "grad_norm": 0.975337285406159,
      "learning_rate": 5.860928493898921e-07,
      "loss": 0.8759,
      "step": 42300
    },
    {
      "epoch": 2.5569589653713667,
      "grad_norm": 0.9434887563252474,
      "learning_rate": 5.845290151778188e-07,
      "loss": 0.8999,
      "step": 42310
    },
    {
      "epoch": 2.5575633045265,
      "grad_norm": 0.9366206325953544,
      "learning_rate": 5.82967140559329e-07,
      "loss": 0.8619,
      "step": 42320
    },
    {
      "epoch": 2.5581676436816343,
      "grad_norm": 0.9107071311363936,
      "learning_rate": 5.814072262275822e-07,
      "loss": 0.8843,
      "step": 42330
    },
    {
      "epoch": 2.558771982836768,
      "grad_norm": 0.9325762069290423,
      "learning_rate": 5.798492728748729e-07,
      "loss": 0.8765,
      "step": 42340
    },
    {
      "epoch": 2.559376321991902,
      "grad_norm": 0.9158779710565561,
      "learning_rate": 5.782932811926234e-07,
      "loss": 0.8764,
      "step": 42350
    },
    {
      "epoch": 2.5599806611470357,
      "grad_norm": 0.848763542101446,
      "learning_rate": 5.767392518713844e-07,
      "loss": 0.8982,
      "step": 42360
    },
    {
      "epoch": 2.5605850003021695,
      "grad_norm": 0.8122214444589134,
      "learning_rate": 5.751871856008373e-07,
      "loss": 0.8838,
      "step": 42370
    },
    {
      "epoch": 2.5611893394573033,
      "grad_norm": 0.8983281544339264,
      "learning_rate": 5.736370830697929e-07,
      "loss": 0.894,
      "step": 42380
    },
    {
      "epoch": 2.561793678612437,
      "grad_norm": 0.9133833263149227,
      "learning_rate": 5.720889449661876e-07,
      "loss": 0.8904,
      "step": 42390
    },
    {
      "epoch": 2.5623980177675714,
      "grad_norm": 0.8947611375204686,
      "learning_rate": 5.705427719770884e-07,
      "loss": 0.8994,
      "step": 42400
    },
    {
      "epoch": 2.563002356922705,
      "grad_norm": 0.8521615667422683,
      "learning_rate": 5.689985647886909e-07,
      "loss": 0.8798,
      "step": 42410
    },
    {
      "epoch": 2.563606696077839,
      "grad_norm": 0.9114040080793261,
      "learning_rate": 5.674563240863146e-07,
      "loss": 0.8889,
      "step": 42420
    },
    {
      "epoch": 2.564211035232973,
      "grad_norm": 0.8876995648282615,
      "learning_rate": 5.659160505544093e-07,
      "loss": 0.8976,
      "step": 42430
    },
    {
      "epoch": 2.5648153743881066,
      "grad_norm": 0.9589346831806169,
      "learning_rate": 5.643777448765525e-07,
      "loss": 0.8813,
      "step": 42440
    },
    {
      "epoch": 2.5654197135432404,
      "grad_norm": 0.6397163371256055,
      "learning_rate": 5.628414077354444e-07,
      "loss": 0.8673,
      "step": 42450
    },
    {
      "epoch": 2.566024052698374,
      "grad_norm": 0.6794446596086937,
      "learning_rate": 5.613070398129156e-07,
      "loss": 0.8842,
      "step": 42460
    },
    {
      "epoch": 2.5666283918535084,
      "grad_norm": 0.6133173771038953,
      "learning_rate": 5.597746417899214e-07,
      "loss": 0.8731,
      "step": 42470
    },
    {
      "epoch": 2.567232731008642,
      "grad_norm": 0.7074697632168433,
      "learning_rate": 5.582442143465428e-07,
      "loss": 0.9014,
      "step": 42480
    },
    {
      "epoch": 2.567837070163776,
      "grad_norm": 0.6494350346170464,
      "learning_rate": 5.567157581619848e-07,
      "loss": 0.8825,
      "step": 42490
    },
    {
      "epoch": 2.56844140931891,
      "grad_norm": 0.6523773100339134,
      "learning_rate": 5.551892739145804e-07,
      "loss": 0.8811,
      "step": 42500
    },
    {
      "epoch": 2.5690457484740437,
      "grad_norm": 0.8013430256427662,
      "learning_rate": 5.536647622817842e-07,
      "loss": 0.8806,
      "step": 42510
    },
    {
      "epoch": 2.5696500876291775,
      "grad_norm": 0.6772055551867058,
      "learning_rate": 5.521422239401792e-07,
      "loss": 0.8851,
      "step": 42520
    },
    {
      "epoch": 2.5702544267843113,
      "grad_norm": 0.7228541547978615,
      "learning_rate": 5.506216595654685e-07,
      "loss": 0.8705,
      "step": 42530
    },
    {
      "epoch": 2.570858765939445,
      "grad_norm": 0.7224840143607193,
      "learning_rate": 5.491030698324828e-07,
      "loss": 0.8966,
      "step": 42540
    },
    {
      "epoch": 2.571463105094579,
      "grad_norm": 0.8434937449033829,
      "learning_rate": 5.475864554151727e-07,
      "loss": 0.8889,
      "step": 42550
    },
    {
      "epoch": 2.572067444249713,
      "grad_norm": 0.894371293908137,
      "learning_rate": 5.460718169866163e-07,
      "loss": 0.8678,
      "step": 42560
    },
    {
      "epoch": 2.572671783404847,
      "grad_norm": 0.8435505511396212,
      "learning_rate": 5.445591552190127e-07,
      "loss": 0.8889,
      "step": 42570
    },
    {
      "epoch": 2.5732761225599807,
      "grad_norm": 0.8484808947948894,
      "learning_rate": 5.43048470783682e-07,
      "loss": 0.9071,
      "step": 42580
    },
    {
      "epoch": 2.5738804617151145,
      "grad_norm": 0.8577720676967568,
      "learning_rate": 5.4153976435107e-07,
      "loss": 0.8899,
      "step": 42590
    },
    {
      "epoch": 2.5744848008702483,
      "grad_norm": 0.6387411783552193,
      "learning_rate": 5.400330365907442e-07,
      "loss": 0.9024,
      "step": 42600
    },
    {
      "epoch": 2.575089140025382,
      "grad_norm": 0.6130786577685777,
      "learning_rate": 5.385282881713899e-07,
      "loss": 0.8784,
      "step": 42610
    },
    {
      "epoch": 2.575693479180516,
      "grad_norm": 0.5966086700331396,
      "learning_rate": 5.370255197608193e-07,
      "loss": 0.8861,
      "step": 42620
    },
    {
      "epoch": 2.57629781833565,
      "grad_norm": 0.6386617379497778,
      "learning_rate": 5.355247320259638e-07,
      "loss": 0.8766,
      "step": 42630
    },
    {
      "epoch": 2.5769021574907836,
      "grad_norm": 0.5941256155595729,
      "learning_rate": 5.340259256328734e-07,
      "loss": 0.8766,
      "step": 42640
    },
    {
      "epoch": 2.577506496645918,
      "grad_norm": 0.581963111734866,
      "learning_rate": 5.325291012467232e-07,
      "loss": 0.8734,
      "step": 42650
    },
    {
      "epoch": 2.5781108358010516,
      "grad_norm": 0.6033359411971223,
      "learning_rate": 5.310342595318063e-07,
      "loss": 0.8724,
      "step": 42660
    },
    {
      "epoch": 2.5787151749561854,
      "grad_norm": 0.5910341602160661,
      "learning_rate": 5.295414011515338e-07,
      "loss": 0.8856,
      "step": 42670
    },
    {
      "epoch": 2.579319514111319,
      "grad_norm": 0.5770435225913192,
      "learning_rate": 5.280505267684411e-07,
      "loss": 0.8805,
      "step": 42680
    },
    {
      "epoch": 2.579923853266453,
      "grad_norm": 0.5851012761348724,
      "learning_rate": 5.265616370441806e-07,
      "loss": 0.8776,
      "step": 42690
    },
    {
      "epoch": 2.5805281924215873,
      "grad_norm": 0.6417043600261595,
      "learning_rate": 5.250747326395223e-07,
      "loss": 0.9072,
      "step": 42700
    },
    {
      "epoch": 2.5811325315767206,
      "grad_norm": 0.5720364423893937,
      "learning_rate": 5.235898142143586e-07,
      "loss": 0.8591,
      "step": 42710
    },
    {
      "epoch": 2.581736870731855,
      "grad_norm": 0.589051717202406,
      "learning_rate": 5.221068824276987e-07,
      "loss": 0.899,
      "step": 42720
    },
    {
      "epoch": 2.5823412098869887,
      "grad_norm": 0.6071983701254237,
      "learning_rate": 5.206259379376699e-07,
      "loss": 0.8821,
      "step": 42730
    },
    {
      "epoch": 2.5829455490421225,
      "grad_norm": 0.6067004615332431,
      "learning_rate": 5.191469814015171e-07,
      "loss": 0.8705,
      "step": 42740
    },
    {
      "epoch": 2.5835498881972563,
      "grad_norm": 0.6099898173104981,
      "learning_rate": 5.176700134756057e-07,
      "loss": 0.8987,
      "step": 42750
    },
    {
      "epoch": 2.58415422735239,
      "grad_norm": 0.6123237351748546,
      "learning_rate": 5.161950348154149e-07,
      "loss": 0.8897,
      "step": 42760
    },
    {
      "epoch": 2.584758566507524,
      "grad_norm": 0.5549704567261496,
      "learning_rate": 5.147220460755442e-07,
      "loss": 0.8783,
      "step": 42770
    },
    {
      "epoch": 2.5853629056626577,
      "grad_norm": 0.6176553921963164,
      "learning_rate": 5.132510479097064e-07,
      "loss": 0.8954,
      "step": 42780
    },
    {
      "epoch": 2.585967244817792,
      "grad_norm": 0.6285660020211064,
      "learning_rate": 5.117820409707358e-07,
      "loss": 0.8943,
      "step": 42790
    },
    {
      "epoch": 2.5865715839729257,
      "grad_norm": 0.5749615393245653,
      "learning_rate": 5.103150259105771e-07,
      "loss": 0.8872,
      "step": 42800
    },
    {
      "epoch": 2.5871759231280596,
      "grad_norm": 0.6029564978556262,
      "learning_rate": 5.088500033802957e-07,
      "loss": 0.8717,
      "step": 42810
    },
    {
      "epoch": 2.5877802622831934,
      "grad_norm": 0.606613001232314,
      "learning_rate": 5.073869740300724e-07,
      "loss": 0.8564,
      "step": 42820
    },
    {
      "epoch": 2.588384601438327,
      "grad_norm": 0.6068726818586329,
      "learning_rate": 5.059259385091996e-07,
      "loss": 0.9041,
      "step": 42830
    },
    {
      "epoch": 2.588988940593461,
      "grad_norm": 0.5783701098280523,
      "learning_rate": 5.044668974660888e-07,
      "loss": 0.8963,
      "step": 42840
    },
    {
      "epoch": 2.5895932797485948,
      "grad_norm": 0.5965022551131236,
      "learning_rate": 5.030098515482651e-07,
      "loss": 0.8941,
      "step": 42850
    },
    {
      "epoch": 2.590197618903729,
      "grad_norm": 0.6503457402794336,
      "learning_rate": 5.01554801402367e-07,
      "loss": 0.8986,
      "step": 42860
    },
    {
      "epoch": 2.5908019580588624,
      "grad_norm": 0.6270907151806185,
      "learning_rate": 5.001017476741487e-07,
      "loss": 0.8827,
      "step": 42870
    },
    {
      "epoch": 2.5914062972139966,
      "grad_norm": 0.6263799501859979,
      "learning_rate": 4.986506910084787e-07,
      "loss": 0.8706,
      "step": 42880
    },
    {
      "epoch": 2.5920106363691304,
      "grad_norm": 0.61376824013978,
      "learning_rate": 4.972016320493372e-07,
      "loss": 0.8906,
      "step": 42890
    },
    {
      "epoch": 2.5926149755242642,
      "grad_norm": 0.6611348083256532,
      "learning_rate": 4.957545714398194e-07,
      "loss": 0.897,
      "step": 42900
    },
    {
      "epoch": 2.593219314679398,
      "grad_norm": 0.6661591757414965,
      "learning_rate": 4.943095098221346e-07,
      "loss": 0.8693,
      "step": 42910
    },
    {
      "epoch": 2.593823653834532,
      "grad_norm": 0.6596231070624817,
      "learning_rate": 4.92866447837601e-07,
      "loss": 0.8869,
      "step": 42920
    },
    {
      "epoch": 2.5944279929896656,
      "grad_norm": 0.6718971592695029,
      "learning_rate": 4.914253861266538e-07,
      "loss": 0.8877,
      "step": 42930
    },
    {
      "epoch": 2.5950323321447994,
      "grad_norm": 0.6669725691680815,
      "learning_rate": 4.899863253288395e-07,
      "loss": 0.8919,
      "step": 42940
    },
    {
      "epoch": 2.5956366712999337,
      "grad_norm": 0.7057916200133024,
      "learning_rate": 4.885492660828134e-07,
      "loss": 0.8965,
      "step": 42950
    },
    {
      "epoch": 2.5962410104550675,
      "grad_norm": 0.6905292609717943,
      "learning_rate": 4.871142090263458e-07,
      "loss": 0.872,
      "step": 42960
    },
    {
      "epoch": 2.5968453496102013,
      "grad_norm": 0.7457587929021203,
      "learning_rate": 4.856811547963192e-07,
      "loss": 0.8821,
      "step": 42970
    },
    {
      "epoch": 2.597449688765335,
      "grad_norm": 0.7605233775464636,
      "learning_rate": 4.842501040287229e-07,
      "loss": 0.9034,
      "step": 42980
    },
    {
      "epoch": 2.598054027920469,
      "grad_norm": 0.752038056447917,
      "learning_rate": 4.828210573586605e-07,
      "loss": 0.8651,
      "step": 42990
    },
    {
      "epoch": 2.5986583670756027,
      "grad_norm": 0.6880145727339527,
      "learning_rate": 4.813940154203467e-07,
      "loss": 0.8921,
      "step": 43000
    },
    {
      "epoch": 2.5992627062307365,
      "grad_norm": 0.710689834578039,
      "learning_rate": 4.799689788471034e-07,
      "loss": 0.8807,
      "step": 43010
    },
    {
      "epoch": 2.5998670453858708,
      "grad_norm": 0.6637555466759679,
      "learning_rate": 4.785459482713656e-07,
      "loss": 0.8862,
      "step": 43020
    },
    {
      "epoch": 2.6004713845410046,
      "grad_norm": 0.6753966033832595,
      "learning_rate": 4.771249243246756e-07,
      "loss": 0.869,
      "step": 43030
    },
    {
      "epoch": 2.6010757236961384,
      "grad_norm": 0.6701160009662632,
      "learning_rate": 4.757059076376874e-07,
      "loss": 0.9101,
      "step": 43040
    },
    {
      "epoch": 2.601680062851272,
      "grad_norm": 0.8042100074168831,
      "learning_rate": 4.7428889884016115e-07,
      "loss": 0.8652,
      "step": 43050
    },
    {
      "epoch": 2.602284402006406,
      "grad_norm": 0.8934394930441069,
      "learning_rate": 4.728738985609704e-07,
      "loss": 0.8922,
      "step": 43060
    },
    {
      "epoch": 2.60288874116154,
      "grad_norm": 0.8480391096702197,
      "learning_rate": 4.714609074280924e-07,
      "loss": 0.8703,
      "step": 43070
    },
    {
      "epoch": 2.6034930803166736,
      "grad_norm": 0.8358965834916362,
      "learning_rate": 4.7004992606861586e-07,
      "loss": 0.8897,
      "step": 43080
    },
    {
      "epoch": 2.604097419471808,
      "grad_norm": 0.7900502515232571,
      "learning_rate": 4.686409551087373e-07,
      "loss": 0.8607,
      "step": 43090
    },
    {
      "epoch": 2.604701758626941,
      "grad_norm": 0.7887683312661142,
      "learning_rate": 4.672339951737598e-07,
      "loss": 0.8673,
      "step": 43100
    },
    {
      "epoch": 2.6053060977820754,
      "grad_norm": 0.8198006789645077,
      "learning_rate": 4.6582904688809437e-07,
      "loss": 0.8758,
      "step": 43110
    },
    {
      "epoch": 2.6059104369372093,
      "grad_norm": 0.858222868784393,
      "learning_rate": 4.644261108752596e-07,
      "loss": 0.9041,
      "step": 43120
    },
    {
      "epoch": 2.606514776092343,
      "grad_norm": 0.7841385620649725,
      "learning_rate": 4.630251877578823e-07,
      "loss": 0.8695,
      "step": 43130
    },
    {
      "epoch": 2.607119115247477,
      "grad_norm": 0.815516764291645,
      "learning_rate": 4.616262781576925e-07,
      "loss": 0.9113,
      "step": 43140
    },
    {
      "epoch": 2.6077234544026107,
      "grad_norm": 1.214019958266389,
      "learning_rate": 4.6022938269552953e-07,
      "loss": 0.8726,
      "step": 43150
    },
    {
      "epoch": 2.6083277935577445,
      "grad_norm": 1.2316790980663552,
      "learning_rate": 4.5883450199133873e-07,
      "loss": 0.8578,
      "step": 43160
    },
    {
      "epoch": 2.6089321327128783,
      "grad_norm": 1.187961228354091,
      "learning_rate": 4.5744163666416876e-07,
      "loss": 0.8599,
      "step": 43170
    },
    {
      "epoch": 2.6095364718680125,
      "grad_norm": 1.2259177245333102,
      "learning_rate": 4.560507873321773e-07,
      "loss": 0.9077,
      "step": 43180
    },
    {
      "epoch": 2.6101408110231463,
      "grad_norm": 1.2843016836868726,
      "learning_rate": 4.5466195461262575e-07,
      "loss": 0.8863,
      "step": 43190
    },
    {
      "epoch": 2.61074515017828,
      "grad_norm": 1.1662908017562952,
      "learning_rate": 4.532751391218787e-07,
      "loss": 0.8797,
      "step": 43200
    },
    {
      "epoch": 2.611349489333414,
      "grad_norm": 1.2220459602247087,
      "learning_rate": 4.5189034147540833e-07,
      "loss": 0.8874,
      "step": 43210
    },
    {
      "epoch": 2.6119538284885477,
      "grad_norm": 0.8546693001503075,
      "learning_rate": 4.505075622877908e-07,
      "loss": 0.906,
      "step": 43220
    },
    {
      "epoch": 2.6125581676436815,
      "grad_norm": 0.9619785719140915,
      "learning_rate": 4.4912680217270466e-07,
      "loss": 0.8729,
      "step": 43230
    },
    {
      "epoch": 2.6131625067988153,
      "grad_norm": 0.9340927571487166,
      "learning_rate": 4.4774806174293416e-07,
      "loss": 0.8821,
      "step": 43240
    },
    {
      "epoch": 2.6137668459539496,
      "grad_norm": 2.914872472865227,
      "learning_rate": 4.463713416103682e-07,
      "loss": 0.8898,
      "step": 43250
    },
    {
      "epoch": 2.614371185109083,
      "grad_norm": 2.5256069278283553,
      "learning_rate": 4.4499664238599415e-07,
      "loss": 0.8769,
      "step": 43260
    },
    {
      "epoch": 2.614975524264217,
      "grad_norm": 2.5193237010766163,
      "learning_rate": 4.436239646799084e-07,
      "loss": 0.8784,
      "step": 43270
    },
    {
      "epoch": 2.615579863419351,
      "grad_norm": 2.621712967379699,
      "learning_rate": 4.422533091013076e-07,
      "loss": 0.8881,
      "step": 43280
    },
    {
      "epoch": 2.616184202574485,
      "grad_norm": 2.5873680815964684,
      "learning_rate": 4.408846762584901e-07,
      "loss": 0.8741,
      "step": 43290
    },
    {
      "epoch": 2.6167885417296186,
      "grad_norm": 0.9555779685644418,
      "learning_rate": 4.3951806675885833e-07,
      "loss": 0.8755,
      "step": 43300
    },
    {
      "epoch": 2.6173928808847524,
      "grad_norm": 0.9278433601172189,
      "learning_rate": 4.38153481208915e-07,
      "loss": 0.9185,
      "step": 43310
    },
    {
      "epoch": 2.617997220039886,
      "grad_norm": 0.9825172755445509,
      "learning_rate": 4.367909202142662e-07,
      "loss": 0.8912,
      "step": 43320
    },
    {
      "epoch": 2.61860155919502,
      "grad_norm": 0.9651057214865958,
      "learning_rate": 4.3543038437961814e-07,
      "loss": 0.8777,
      "step": 43330
    },
    {
      "epoch": 2.6192058983501543,
      "grad_norm": 0.9440098579120147,
      "learning_rate": 4.340718743087802e-07,
      "loss": 0.8783,
      "step": 43340
    },
    {
      "epoch": 2.619810237505288,
      "grad_norm": 1.0782410743858655,
      "learning_rate": 4.3271539060465995e-07,
      "loss": 0.8582,
      "step": 43350
    },
    {
      "epoch": 2.620414576660422,
      "grad_norm": 1.0768914130535585,
      "learning_rate": 4.3136093386926816e-07,
      "loss": 0.8689,
      "step": 43360
    },
    {
      "epoch": 2.6210189158155557,
      "grad_norm": 1.0856387250961594,
      "learning_rate": 4.300085047037156e-07,
      "loss": 0.8794,
      "step": 43370
    },
    {
      "epoch": 2.6216232549706895,
      "grad_norm": 1.088311777727129,
      "learning_rate": 4.2865810370821047e-07,
      "loss": 0.8775,
      "step": 43380
    },
    {
      "epoch": 2.6222275941258233,
      "grad_norm": 0.9637473309610657,
      "learning_rate": 4.273097314820651e-07,
      "loss": 0.8921,
      "step": 43390
    },
    {
      "epoch": 2.622831933280957,
      "grad_norm": 1.099713445025246,
      "learning_rate": 4.2596338862368914e-07,
      "loss": 0.8715,
      "step": 43400
    },
    {
      "epoch": 2.6234362724360913,
      "grad_norm": 1.115343801867195,
      "learning_rate": 4.2461907573059067e-07,
      "loss": 0.8768,
      "step": 43410
    },
    {
      "epoch": 2.624040611591225,
      "grad_norm": 1.0546264339069136,
      "learning_rate": 4.2327679339937924e-07,
      "loss": 0.8948,
      "step": 43420
    },
    {
      "epoch": 2.624644950746359,
      "grad_norm": 1.1413067443560065,
      "learning_rate": 4.2193654222576206e-07,
      "loss": 0.8917,
      "step": 43430
    },
    {
      "epoch": 2.6252492899014928,
      "grad_norm": 1.035788299578993,
      "learning_rate": 4.20598322804544e-07,
      "loss": 0.9011,
      "step": 43440
    },
    {
      "epoch": 2.6258536290566266,
      "grad_norm": 1.0783188574807157,
      "learning_rate": 4.192621357296295e-07,
      "loss": 0.8849,
      "step": 43450
    },
    {
      "epoch": 2.6264579682117604,
      "grad_norm": 1.0561851845741754,
      "learning_rate": 4.179279815940207e-07,
      "loss": 0.8742,
      "step": 43460
    },
    {
      "epoch": 2.627062307366894,
      "grad_norm": 1.1756603887056123,
      "learning_rate": 4.1659586098981854e-07,
      "loss": 0.9047,
      "step": 43470
    },
    {
      "epoch": 2.6276666465220284,
      "grad_norm": 1.162210507205473,
      "learning_rate": 4.1526577450821835e-07,
      "loss": 0.8824,
      "step": 43480
    },
    {
      "epoch": 2.6282709856771618,
      "grad_norm": 1.126249856683156,
      "learning_rate": 4.1393772273951583e-07,
      "loss": 0.8689,
      "step": 43490
    },
    {
      "epoch": 2.628875324832296,
      "grad_norm": 1.089311651312746,
      "learning_rate": 4.126117062731039e-07,
      "loss": 0.8807,
      "step": 43500
    },
    {
      "epoch": 2.62947966398743,
      "grad_norm": 1.1112939970484306,
      "learning_rate": 4.1128772569746845e-07,
      "loss": 0.8915,
      "step": 43510
    },
    {
      "epoch": 2.6300840031425636,
      "grad_norm": 1.180490640411348,
      "learning_rate": 4.0996578160019617e-07,
      "loss": 0.8845,
      "step": 43520
    },
    {
      "epoch": 2.6306883422976974,
      "grad_norm": 1.197374232502728,
      "learning_rate": 4.086458745679678e-07,
      "loss": 0.8817,
      "step": 43530
    },
    {
      "epoch": 2.6312926814528312,
      "grad_norm": 1.1542862155577798,
      "learning_rate": 4.073280051865597e-07,
      "loss": 0.8955,
      "step": 43540
    },
    {
      "epoch": 2.631897020607965,
      "grad_norm": 1.0446893088197602,
      "learning_rate": 4.060121740408457e-07,
      "loss": 0.8953,
      "step": 43550
    },
    {
      "epoch": 2.632501359763099,
      "grad_norm": 1.0192611349747334,
      "learning_rate": 4.04698381714792e-07,
      "loss": 0.8775,
      "step": 43560
    },
    {
      "epoch": 2.633105698918233,
      "grad_norm": 1.0228223880126996,
      "learning_rate": 4.0338662879146453e-07,
      "loss": 0.9087,
      "step": 43570
    },
    {
      "epoch": 2.633710038073367,
      "grad_norm": 1.0225605695194147,
      "learning_rate": 4.0207691585301855e-07,
      "loss": 0.8836,
      "step": 43580
    },
    {
      "epoch": 2.6343143772285007,
      "grad_norm": 1.0866276505740302,
      "learning_rate": 4.007692434807098e-07,
      "loss": 0.888,
      "step": 43590
    },
    {
      "epoch": 2.6349187163836345,
      "grad_norm": 0.9970753469976296,
      "learning_rate": 3.9946361225488283e-07,
      "loss": 0.8826,
      "step": 43600
    },
    {
      "epoch": 2.6355230555387683,
      "grad_norm": 1.0409089782300487,
      "learning_rate": 3.9816002275498036e-07,
      "loss": 0.883,
      "step": 43610
    },
    {
      "epoch": 2.636127394693902,
      "grad_norm": 1.0217474566699665,
      "learning_rate": 3.9685847555953784e-07,
      "loss": 0.8677,
      "step": 43620
    },
    {
      "epoch": 2.636731733849036,
      "grad_norm": 1.0013617352147786,
      "learning_rate": 3.9555897124618383e-07,
      "loss": 0.8838,
      "step": 43630
    },
    {
      "epoch": 2.63733607300417,
      "grad_norm": 1.0664941232312735,
      "learning_rate": 3.942615103916403e-07,
      "loss": 0.87,
      "step": 43640
    },
    {
      "epoch": 2.6379404121593035,
      "grad_norm": 1.0355345230974287,
      "learning_rate": 3.9296609357172333e-07,
      "loss": 0.8669,
      "step": 43650
    },
    {
      "epoch": 2.6385447513144378,
      "grad_norm": 1.0596592072884867,
      "learning_rate": 3.9167272136134027e-07,
      "loss": 0.8864,
      "step": 43660
    },
    {
      "epoch": 2.6391490904695716,
      "grad_norm": 1.0823909732558148,
      "learning_rate": 3.9038139433449194e-07,
      "loss": 0.884,
      "step": 43670
    },
    {
      "epoch": 2.6397534296247054,
      "grad_norm": 1.013513754511572,
      "learning_rate": 3.8909211306427274e-07,
      "loss": 0.8765,
      "step": 43680
    },
    {
      "epoch": 2.640357768779839,
      "grad_norm": 1.0390876464100904,
      "learning_rate": 3.878048781228666e-07,
      "loss": 0.9131,
      "step": 43690
    },
    {
      "epoch": 2.640962107934973,
      "grad_norm": 1.144995132013146,
      "learning_rate": 3.86519690081551e-07,
      "loss": 0.8762,
      "step": 43700
    },
    {
      "epoch": 2.6415664470901072,
      "grad_norm": 1.2570537438358822,
      "learning_rate": 3.852365495106952e-07,
      "loss": 0.8672,
      "step": 43710
    },
    {
      "epoch": 2.6421707862452406,
      "grad_norm": 1.322219189103477,
      "learning_rate": 3.8395545697975765e-07,
      "loss": 0.9006,
      "step": 43720
    },
    {
      "epoch": 2.642775125400375,
      "grad_norm": 1.1290146874942655,
      "learning_rate": 3.8267641305729076e-07,
      "loss": 0.8995,
      "step": 43730
    },
    {
      "epoch": 2.6433794645555087,
      "grad_norm": 1.2326782837518608,
      "learning_rate": 3.813994183109365e-07,
      "loss": 0.8499,
      "step": 43740
    },
    {
      "epoch": 2.6439838037106425,
      "grad_norm": 1.1724561433574527,
      "learning_rate": 3.8012447330742607e-07,
      "loss": 0.8974,
      "step": 43750
    },
    {
      "epoch": 2.6445881428657763,
      "grad_norm": 1.0728561939648071,
      "learning_rate": 3.7885157861258295e-07,
      "loss": 0.8862,
      "step": 43760
    },
    {
      "epoch": 2.64519248202091,
      "grad_norm": 1.1908841490343347,
      "learning_rate": 3.775807347913207e-07,
      "loss": 0.8719,
      "step": 43770
    },
    {
      "epoch": 2.645796821176044,
      "grad_norm": 1.184293877908691,
      "learning_rate": 3.7631194240764113e-07,
      "loss": 0.8736,
      "step": 43780
    },
    {
      "epoch": 2.6464011603311777,
      "grad_norm": 1.0766110307497285,
      "learning_rate": 3.750452020246359e-07,
      "loss": 0.9241,
      "step": 43790
    },
    {
      "epoch": 2.647005499486312,
      "grad_norm": 1.2148393304143459,
      "learning_rate": 3.737805142044887e-07,
      "loss": 0.8712,
      "step": 43800
    },
    {
      "epoch": 2.6476098386414457,
      "grad_norm": 1.220223015092784,
      "learning_rate": 3.725178795084677e-07,
      "loss": 0.8798,
      "step": 43810
    },
    {
      "epoch": 2.6482141777965795,
      "grad_norm": 1.2335915789870364,
      "learning_rate": 3.7125729849693427e-07,
      "loss": 0.8833,
      "step": 43820
    },
    {
      "epoch": 2.6488185169517133,
      "grad_norm": 1.168483822697791,
      "learning_rate": 3.6999877172933517e-07,
      "loss": 0.8762,
      "step": 43830
    },
    {
      "epoch": 2.649422856106847,
      "grad_norm": 1.2127118443731855,
      "learning_rate": 3.687422997642071e-07,
      "loss": 0.8826,
      "step": 43840
    },
    {
      "epoch": 2.650027195261981,
      "grad_norm": 1.4665009668811448,
      "learning_rate": 3.674878831591744e-07,
      "loss": 0.8889,
      "step": 43850
    },
    {
      "epoch": 2.6506315344171147,
      "grad_norm": 1.4576280159055806,
      "learning_rate": 3.662355224709491e-07,
      "loss": 0.8869,
      "step": 43860
    },
    {
      "epoch": 2.651235873572249,
      "grad_norm": 1.5118235914781029,
      "learning_rate": 3.6498521825533205e-07,
      "loss": 0.8879,
      "step": 43870
    },
    {
      "epoch": 2.6518402127273824,
      "grad_norm": 1.512515345737943,
      "learning_rate": 3.6373697106720896e-07,
      "loss": 0.8716,
      "step": 43880
    },
    {
      "epoch": 2.6524445518825166,
      "grad_norm": 1.4423984504415237,
      "learning_rate": 3.6249078146055493e-07,
      "loss": 0.9107,
      "step": 43890
    },
    {
      "epoch": 2.6530488910376504,
      "grad_norm": 0.9584133630320645,
      "learning_rate": 3.612466499884315e-07,
      "loss": 0.9106,
      "step": 43900
    },
    {
      "epoch": 2.653653230192784,
      "grad_norm": 0.992578958991291,
      "learning_rate": 3.6000457720298533e-07,
      "loss": 0.8947,
      "step": 43910
    },
    {
      "epoch": 2.654257569347918,
      "grad_norm": 0.9178430826637624,
      "learning_rate": 3.5876456365545045e-07,
      "loss": 0.8865,
      "step": 43920
    },
    {
      "epoch": 2.654861908503052,
      "grad_norm": 0.9150553484729788,
      "learning_rate": 3.575266098961483e-07,
      "loss": 0.864,
      "step": 43930
    },
    {
      "epoch": 2.6554662476581856,
      "grad_norm": 0.9124823216881625,
      "learning_rate": 3.562907164744833e-07,
      "loss": 0.8972,
      "step": 43940
    },
    {
      "epoch": 2.6560705868133194,
      "grad_norm": 0.8724897906392935,
      "learning_rate": 3.550568839389473e-07,
      "loss": 0.8749,
      "step": 43950
    },
    {
      "epoch": 2.6566749259684537,
      "grad_norm": 0.852852251174776,
      "learning_rate": 3.5382511283711895e-07,
      "loss": 0.8836,
      "step": 43960
    },
    {
      "epoch": 2.6572792651235875,
      "grad_norm": 0.7995311589105277,
      "learning_rate": 3.5259540371565816e-07,
      "loss": 0.8756,
      "step": 43970
    },
    {
      "epoch": 2.6578836042787213,
      "grad_norm": 0.8614177345583277,
      "learning_rate": 3.5136775712031337e-07,
      "loss": 0.8895,
      "step": 43980
    },
    {
      "epoch": 2.658487943433855,
      "grad_norm": 0.8870556743029577,
      "learning_rate": 3.50142173595916e-07,
      "loss": 0.8741,
      "step": 43990
    },
    {
      "epoch": 2.659092282588989,
      "grad_norm": 0.8167234999583096,
      "learning_rate": 3.4891865368638146e-07,
      "loss": 0.8791,
      "step": 44000
    },
    {
      "epoch": 2.6596966217441227,
      "grad_norm": 0.9378867466529495,
      "learning_rate": 3.4769719793471046e-07,
      "loss": 0.8743,
      "step": 44010
    },
    {
      "epoch": 2.6603009608992565,
      "grad_norm": 0.9029683678887714,
      "learning_rate": 3.464778068829883e-07,
      "loss": 0.8942,
      "step": 44020
    },
    {
      "epoch": 2.6609053000543907,
      "grad_norm": 0.8960525968167036,
      "learning_rate": 3.452604810723803e-07,
      "loss": 0.8717,
      "step": 44030
    },
    {
      "epoch": 2.6615096392095245,
      "grad_norm": 0.8767426540642903,
      "learning_rate": 3.4404522104313943e-07,
      "loss": 0.9051,
      "step": 44040
    },
    {
      "epoch": 2.6621139783646584,
      "grad_norm": 0.6662176844034844,
      "learning_rate": 3.428320273346003e-07,
      "loss": 0.8832,
      "step": 44050
    },
    {
      "epoch": 2.662718317519792,
      "grad_norm": 0.5742983863112889,
      "learning_rate": 3.4162090048517946e-07,
      "loss": 0.8631,
      "step": 44060
    },
    {
      "epoch": 2.663322656674926,
      "grad_norm": 0.6743823834285331,
      "learning_rate": 3.404118410323781e-07,
      "loss": 0.9191,
      "step": 44070
    },
    {
      "epoch": 2.6639269958300598,
      "grad_norm": 0.6424213514599196,
      "learning_rate": 3.3920484951277753e-07,
      "loss": 0.8545,
      "step": 44080
    },
    {
      "epoch": 2.6645313349851936,
      "grad_norm": 0.6378953679641663,
      "learning_rate": 3.379999264620443e-07,
      "loss": 0.8853,
      "step": 44090
    },
    {
      "epoch": 2.665135674140328,
      "grad_norm": 0.7228214564536835,
      "learning_rate": 3.3679707241492334e-07,
      "loss": 0.8864,
      "step": 44100
    },
    {
      "epoch": 2.665740013295461,
      "grad_norm": 0.6955572224582335,
      "learning_rate": 3.3559628790524544e-07,
      "loss": 0.8942,
      "step": 44110
    },
    {
      "epoch": 2.6663443524505954,
      "grad_norm": 0.7149970854094722,
      "learning_rate": 3.343975734659194e-07,
      "loss": 0.885,
      "step": 44120
    },
    {
      "epoch": 2.6669486916057292,
      "grad_norm": 0.6964161244538115,
      "learning_rate": 3.3320092962893623e-07,
      "loss": 0.9111,
      "step": 44130
    },
    {
      "epoch": 2.667553030760863,
      "grad_norm": 0.66458410337221,
      "learning_rate": 3.320063569253701e-07,
      "loss": 0.8625,
      "step": 44140
    },
    {
      "epoch": 2.668157369915997,
      "grad_norm": 0.9769026065755836,
      "learning_rate": 3.308138558853746e-07,
      "loss": 0.8595,
      "step": 44150
    },
    {
      "epoch": 2.6687617090711306,
      "grad_norm": 0.8453732839139658,
      "learning_rate": 3.296234270381821e-07,
      "loss": 0.8915,
      "step": 44160
    },
    {
      "epoch": 2.6693660482262644,
      "grad_norm": 0.8869206614138567,
      "learning_rate": 3.284350709121076e-07,
      "loss": 0.8803,
      "step": 44170
    },
    {
      "epoch": 2.6699703873813982,
      "grad_norm": 0.8997987974377994,
      "learning_rate": 3.272487880345465e-07,
      "loss": 0.8486,
      "step": 44180
    },
    {
      "epoch": 2.6705747265365325,
      "grad_norm": 0.8727690731104922,
      "learning_rate": 3.2606457893197164e-07,
      "loss": 0.8935,
      "step": 44190
    },
    {
      "epoch": 2.6711790656916663,
      "grad_norm": 0.6311554818261784,
      "learning_rate": 3.248824441299375e-07,
      "loss": 0.8877,
      "step": 44200
    },
    {
      "epoch": 2.6717834048468,
      "grad_norm": 0.6123527286505754,
      "learning_rate": 3.2370238415307896e-07,
      "loss": 0.8723,
      "step": 44210
    },
    {
      "epoch": 2.672387744001934,
      "grad_norm": 0.5859084938790945,
      "learning_rate": 3.2252439952510704e-07,
      "loss": 0.9013,
      "step": 44220
    },
    {
      "epoch": 2.6729920831570677,
      "grad_norm": 0.6217074710737848,
      "learning_rate": 3.2134849076881303e-07,
      "loss": 0.8574,
      "step": 44230
    },
    {
      "epoch": 2.6735964223122015,
      "grad_norm": 0.6026215002550583,
      "learning_rate": 3.2017465840606944e-07,
      "loss": 0.8838,
      "step": 44240
    },
    {
      "epoch": 2.6742007614673353,
      "grad_norm": 0.5772391787459733,
      "learning_rate": 3.190029029578229e-07,
      "loss": 0.8674,
      "step": 44250
    },
    {
      "epoch": 2.6748051006224696,
      "grad_norm": 0.5874932384324928,
      "learning_rate": 3.178332249441013e-07,
      "loss": 0.8982,
      "step": 44260
    },
    {
      "epoch": 2.675409439777603,
      "grad_norm": 0.5791916985148398,
      "learning_rate": 3.1666562488401054e-07,
      "loss": 0.8939,
      "step": 44270
    },
    {
      "epoch": 2.676013778932737,
      "grad_norm": 0.5647132992319281,
      "learning_rate": 3.1550010329573164e-07,
      "loss": 0.8773,
      "step": 44280
    },
    {
      "epoch": 2.676618118087871,
      "grad_norm": 0.6378354821034709,
      "learning_rate": 3.143366606965259e-07,
      "loss": 0.8915,
      "step": 44290
    },
    {
      "epoch": 2.677222457243005,
      "grad_norm": 0.603575427971721,
      "learning_rate": 3.13175297602733e-07,
      "loss": 0.899,
      "step": 44300
    },
    {
      "epoch": 2.6778267963981386,
      "grad_norm": 0.5995792293066782,
      "learning_rate": 3.120160145297646e-07,
      "loss": 0.8828,
      "step": 44310
    },
    {
      "epoch": 2.6784311355532724,
      "grad_norm": 0.5512564719336104,
      "learning_rate": 3.1085881199211576e-07,
      "loss": 0.8638,
      "step": 44320
    },
    {
      "epoch": 2.679035474708406,
      "grad_norm": 0.5958366573850185,
      "learning_rate": 3.097036905033518e-07,
      "loss": 0.853,
      "step": 44330
    },
    {
      "epoch": 2.67963981386354,
      "grad_norm": 0.59003121469836,
      "learning_rate": 3.0855065057612044e-07,
      "loss": 0.9033,
      "step": 44340
    },
    {
      "epoch": 2.6802441530186742,
      "grad_norm": 0.5710681931885819,
      "learning_rate": 3.0739969272214065e-07,
      "loss": 0.8762,
      "step": 44350
    },
    {
      "epoch": 2.680848492173808,
      "grad_norm": 0.6160427067410187,
      "learning_rate": 3.0625081745221107e-07,
      "loss": 0.889,
      "step": 44360
    },
    {
      "epoch": 2.681452831328942,
      "grad_norm": 0.6452060402213153,
      "learning_rate": 3.0510402527620277e-07,
      "loss": 0.9081,
      "step": 44370
    },
    {
      "epoch": 2.6820571704840757,
      "grad_norm": 0.5753217129010045,
      "learning_rate": 3.0395931670306523e-07,
      "loss": 0.8761,
      "step": 44380
    },
    {
      "epoch": 2.6826615096392095,
      "grad_norm": 0.6068957855421832,
      "learning_rate": 3.0281669224082266e-07,
      "loss": 0.8763,
      "step": 44390
    },
    {
      "epoch": 2.6832658487943433,
      "grad_norm": 0.568098102051787,
      "learning_rate": 3.016761523965728e-07,
      "loss": 0.871,
      "step": 44400
    },
    {
      "epoch": 2.683870187949477,
      "grad_norm": 0.5885370471373237,
      "learning_rate": 3.00537697676489e-07,
      "loss": 0.8936,
      "step": 44410
    },
    {
      "epoch": 2.6844745271046113,
      "grad_norm": 0.6300200683297026,
      "learning_rate": 2.994013285858205e-07,
      "loss": 0.9193,
      "step": 44420
    },
    {
      "epoch": 2.685078866259745,
      "grad_norm": 0.6383619882100581,
      "learning_rate": 2.982670456288883e-07,
      "loss": 0.9028,
      "step": 44430
    },
    {
      "epoch": 2.685683205414879,
      "grad_norm": 0.5682472920606244,
      "learning_rate": 2.971348493090903e-07,
      "loss": 0.8699,
      "step": 44440
    },
    {
      "epoch": 2.6862875445700127,
      "grad_norm": 0.5809286243912853,
      "learning_rate": 2.960047401288979e-07,
      "loss": 0.8814,
      "step": 44450
    },
    {
      "epoch": 2.6868918837251465,
      "grad_norm": 0.5964210974375643,
      "learning_rate": 2.948767185898532e-07,
      "loss": 0.9095,
      "step": 44460
    },
    {
      "epoch": 2.6874962228802803,
      "grad_norm": 0.6269819978363419,
      "learning_rate": 2.9375078519257525e-07,
      "loss": 0.8782,
      "step": 44470
    },
    {
      "epoch": 2.688100562035414,
      "grad_norm": 0.6112862451973233,
      "learning_rate": 2.9262694043675545e-07,
      "loss": 0.8683,
      "step": 44480
    },
    {
      "epoch": 2.6887049011905484,
      "grad_norm": 0.6450045891635381,
      "learning_rate": 2.915051848211592e-07,
      "loss": 0.8693,
      "step": 44490
    },
    {
      "epoch": 2.6893092403456818,
      "grad_norm": 0.6429482315688143,
      "learning_rate": 2.9038551884362056e-07,
      "loss": 0.8825,
      "step": 44500
    },
    {
      "epoch": 2.689913579500816,
      "grad_norm": 0.677920814906551,
      "learning_rate": 2.8926794300105143e-07,
      "loss": 0.8721,
      "step": 44510
    },
    {
      "epoch": 2.69051791865595,
      "grad_norm": 0.6701491222126793,
      "learning_rate": 2.8815245778943403e-07,
      "loss": 0.8963,
      "step": 44520
    },
    {
      "epoch": 2.6911222578110836,
      "grad_norm": 0.698093394607616,
      "learning_rate": 2.8703906370382117e-07,
      "loss": 0.8962,
      "step": 44530
    },
    {
      "epoch": 2.6917265969662174,
      "grad_norm": 0.6838647105102715,
      "learning_rate": 2.859277612383399e-07,
      "loss": 0.8791,
      "step": 44540
    },
    {
      "epoch": 2.692330936121351,
      "grad_norm": 0.7423617329646944,
      "learning_rate": 2.848185508861889e-07,
      "loss": 0.8867,
      "step": 44550
    },
    {
      "epoch": 2.692935275276485,
      "grad_norm": 0.7438536977114678,
      "learning_rate": 2.8371143313963566e-07,
      "loss": 0.9049,
      "step": 44560
    },
    {
      "epoch": 2.693539614431619,
      "grad_norm": 0.773209881306369,
      "learning_rate": 2.826064084900232e-07,
      "loss": 0.8889,
      "step": 44570
    },
    {
      "epoch": 2.694143953586753,
      "grad_norm": 0.7129533364584698,
      "learning_rate": 2.815034774277614e-07,
      "loss": 0.8677,
      "step": 44580
    },
    {
      "epoch": 2.694748292741887,
      "grad_norm": 0.7511512987060771,
      "learning_rate": 2.8040264044233436e-07,
      "loss": 0.9059,
      "step": 44590
    },
    {
      "epoch": 2.6953526318970207,
      "grad_norm": 0.6833981161092568,
      "learning_rate": 2.79303898022294e-07,
      "loss": 0.8738,
      "step": 44600
    },
    {
      "epoch": 2.6959569710521545,
      "grad_norm": 0.7292251167777419,
      "learning_rate": 2.782072506552658e-07,
      "loss": 0.8777,
      "step": 44610
    },
    {
      "epoch": 2.6965613102072883,
      "grad_norm": 0.7130669561404509,
      "learning_rate": 2.771126988279421e-07,
      "loss": 0.899,
      "step": 44620
    },
    {
      "epoch": 2.697165649362422,
      "grad_norm": 0.6802975319099562,
      "learning_rate": 2.760202430260872e-07,
      "loss": 0.9004,
      "step": 44630
    },
    {
      "epoch": 2.697769988517556,
      "grad_norm": 0.6302671018604316,
      "learning_rate": 2.7492988373453653e-07,
      "loss": 0.8678,
      "step": 44640
    },
    {
      "epoch": 2.69837432767269,
      "grad_norm": 0.8747536511635973,
      "learning_rate": 2.7384162143719137e-07,
      "loss": 0.8874,
      "step": 44650
    },
    {
      "epoch": 2.6989786668278235,
      "grad_norm": 0.8461739229698654,
      "learning_rate": 2.7275545661702485e-07,
      "loss": 0.88,
      "step": 44660
    },
    {
      "epoch": 2.6995830059829578,
      "grad_norm": 0.8363056138119715,
      "learning_rate": 2.716713897560802e-07,
      "loss": 0.8871,
      "step": 44670
    },
    {
      "epoch": 2.7001873451380916,
      "grad_norm": 0.7926902738712706,
      "learning_rate": 2.705894213354665e-07,
      "loss": 0.8753,
      "step": 44680
    },
    {
      "epoch": 2.7007916842932254,
      "grad_norm": 0.8376912435964293,
      "learning_rate": 2.69509551835363e-07,
      "loss": 0.9015,
      "step": 44690
    },
    {
      "epoch": 2.701396023448359,
      "grad_norm": 0.8350907894895543,
      "learning_rate": 2.684317817350196e-07,
      "loss": 0.896,
      "step": 44700
    },
    {
      "epoch": 2.702000362603493,
      "grad_norm": 0.8464860003006531,
      "learning_rate": 2.6735611151275045e-07,
      "loss": 0.8931,
      "step": 44710
    },
    {
      "epoch": 2.702604701758627,
      "grad_norm": 0.9090875024563327,
      "learning_rate": 2.662825416459408e-07,
      "loss": 0.9092,
      "step": 44720
    },
    {
      "epoch": 2.7032090409137606,
      "grad_norm": 0.8212991164160085,
      "learning_rate": 2.652110726110435e-07,
      "loss": 0.8921,
      "step": 44730
    },
    {
      "epoch": 2.703813380068895,
      "grad_norm": 0.8379777426125715,
      "learning_rate": 2.641417048835765e-07,
      "loss": 0.8857,
      "step": 44740
    },
    {
      "epoch": 2.7044177192240286,
      "grad_norm": 1.148210674383701,
      "learning_rate": 2.6307443893812847e-07,
      "loss": 0.8621,
      "step": 44750
    },
    {
      "epoch": 2.7050220583791624,
      "grad_norm": 1.2261794452709474,
      "learning_rate": 2.620092752483533e-07,
      "loss": 0.8967,
      "step": 44760
    },
    {
      "epoch": 2.7056263975342962,
      "grad_norm": 1.240665228130346,
      "learning_rate": 2.6094621428697285e-07,
      "loss": 0.8668,
      "step": 44770
    },
    {
      "epoch": 2.70623073668943,
      "grad_norm": 1.208723166751661,
      "learning_rate": 2.5988525652577424e-07,
      "loss": 0.882,
      "step": 44780
    },
    {
      "epoch": 2.706835075844564,
      "grad_norm": 1.2517706033107603,
      "learning_rate": 2.588264024356146e-07,
      "loss": 0.8762,
      "step": 44790
    },
    {
      "epoch": 2.7074394149996976,
      "grad_norm": 0.9902902650343767,
      "learning_rate": 2.577696524864132e-07,
      "loss": 0.9,
      "step": 44800
    },
    {
      "epoch": 2.708043754154832,
      "grad_norm": 0.9776933450606815,
      "learning_rate": 2.5671500714715745e-07,
      "loss": 0.8984,
      "step": 44810
    },
    {
      "epoch": 2.7086480933099657,
      "grad_norm": 0.9092755011009269,
      "learning_rate": 2.55662466885902e-07,
      "loss": 0.8988,
      "step": 44820
    },
    {
      "epoch": 2.7092524324650995,
      "grad_norm": 1.0110562276040946,
      "learning_rate": 2.5461203216976593e-07,
      "loss": 0.9084,
      "step": 44830
    },
    {
      "epoch": 2.7098567716202333,
      "grad_norm": 1.0129500837342975,
      "learning_rate": 2.535637034649324e-07,
      "loss": 0.8799,
      "step": 44840
    },
    {
      "epoch": 2.710461110775367,
      "grad_norm": 2.7061545308058244,
      "learning_rate": 2.5251748123665376e-07,
      "loss": 0.8893,
      "step": 44850
    },
    {
      "epoch": 2.711065449930501,
      "grad_norm": 2.143043160935743,
      "learning_rate": 2.51473365949243e-07,
      "loss": 0.8886,
      "step": 44860
    },
    {
      "epoch": 2.7116697890856347,
      "grad_norm": 2.4614840484862843,
      "learning_rate": 2.5043135806608223e-07,
      "loss": 0.8875,
      "step": 44870
    },
    {
      "epoch": 2.712274128240769,
      "grad_norm": 2.6432759170363687,
      "learning_rate": 2.493914580496143e-07,
      "loss": 0.8858,
      "step": 44880
    },
    {
      "epoch": 2.7128784673959023,
      "grad_norm": 2.4554370560241487,
      "learning_rate": 2.483536663613506e-07,
      "loss": 0.8749,
      "step": 44890
    },
    {
      "epoch": 2.7134828065510366,
      "grad_norm": 0.9459670926428778,
      "learning_rate": 2.473179834618639e-07,
      "loss": 0.88,
      "step": 44900
    },
    {
      "epoch": 2.7140871457061704,
      "grad_norm": 0.9580246587333454,
      "learning_rate": 2.462844098107914e-07,
      "loss": 0.8697,
      "step": 44910
    },
    {
      "epoch": 2.714691484861304,
      "grad_norm": 0.9955754639792951,
      "learning_rate": 2.4525294586683625e-07,
      "loss": 0.8816,
      "step": 44920
    },
    {
      "epoch": 2.715295824016438,
      "grad_norm": 0.919740970255983,
      "learning_rate": 2.442235920877628e-07,
      "loss": 0.8999,
      "step": 44930
    },
    {
      "epoch": 2.715900163171572,
      "grad_norm": 1.0107015176502936,
      "learning_rate": 2.431963489304001e-07,
      "loss": 0.8485,
      "step": 44940
    },
    {
      "epoch": 2.7165045023267056,
      "grad_norm": 1.0184455406520803,
      "learning_rate": 2.421712168506418e-07,
      "loss": 0.8958,
      "step": 44950
    },
    {
      "epoch": 2.7171088414818394,
      "grad_norm": 0.9825152806957556,
      "learning_rate": 2.411481963034412e-07,
      "loss": 0.892,
      "step": 44960
    },
    {
      "epoch": 2.7177131806369736,
      "grad_norm": 0.9392867403119376,
      "learning_rate": 2.401272877428179e-07,
      "loss": 0.8554,
      "step": 44970
    },
    {
      "epoch": 2.7183175197921075,
      "grad_norm": 1.088299104437561,
      "learning_rate": 2.391084916218533e-07,
      "loss": 0.8907,
      "step": 44980
    },
    {
      "epoch": 2.7189218589472413,
      "grad_norm": 0.9794911351986175,
      "learning_rate": 2.380918083926892e-07,
      "loss": 0.8677,
      "step": 44990
    },
    {
      "epoch": 2.719526198102375,
      "grad_norm": 1.1727643450334087,
      "learning_rate": 2.370772385065323e-07,
      "loss": 0.8818,
      "step": 45000
    },
    {
      "epoch": 2.720130537257509,
      "grad_norm": 1.0177159573572903,
      "learning_rate": 2.360647824136514e-07,
      "loss": 0.8842,
      "step": 45010
    },
    {
      "epoch": 2.7207348764126427,
      "grad_norm": 0.9805753057868669,
      "learning_rate": 2.3505444056337478e-07,
      "loss": 0.8814,
      "step": 45020
    },
    {
      "epoch": 2.7213392155677765,
      "grad_norm": 1.5267478949796094,
      "learning_rate": 2.3404621340409383e-07,
      "loss": 0.8929,
      "step": 45030
    },
    {
      "epoch": 2.7219435547229107,
      "grad_norm": 1.0990778320044334,
      "learning_rate": 2.3304010138326282e-07,
      "loss": 0.8991,
      "step": 45040
    },
    {
      "epoch": 2.7225478938780445,
      "grad_norm": 1.0014123143834517,
      "learning_rate": 2.3203610494739514e-07,
      "loss": 0.9033,
      "step": 45050
    },
    {
      "epoch": 2.7231522330331783,
      "grad_norm": 1.11621162461148,
      "learning_rate": 2.3103422454206548e-07,
      "loss": 0.8799,
      "step": 45060
    },
    {
      "epoch": 2.723756572188312,
      "grad_norm": 1.149514812100179,
      "learning_rate": 2.3003446061191204e-07,
      "loss": 0.8965,
      "step": 45070
    },
    {
      "epoch": 2.724360911343446,
      "grad_norm": 1.0592214190440168,
      "learning_rate": 2.2903681360062934e-07,
      "loss": 0.8391,
      "step": 45080
    },
    {
      "epoch": 2.7249652504985797,
      "grad_norm": 1.045205437885987,
      "learning_rate": 2.2804128395097659e-07,
      "loss": 0.8756,
      "step": 45090
    },
    {
      "epoch": 2.7255695896537135,
      "grad_norm": 1.1695799392856787,
      "learning_rate": 2.270478721047714e-07,
      "loss": 0.9016,
      "step": 45100
    },
    {
      "epoch": 2.726173928808848,
      "grad_norm": 1.2092449162116197,
      "learning_rate": 2.2605657850289064e-07,
      "loss": 0.9097,
      "step": 45110
    },
    {
      "epoch": 2.726778267963981,
      "grad_norm": 1.1930576511306474,
      "learning_rate": 2.250674035852729e-07,
      "loss": 0.8626,
      "step": 45120
    },
    {
      "epoch": 2.7273826071191154,
      "grad_norm": 1.106632166487915,
      "learning_rate": 2.2408034779091537e-07,
      "loss": 0.9029,
      "step": 45130
    },
    {
      "epoch": 2.727986946274249,
      "grad_norm": 1.1287378717292094,
      "learning_rate": 2.2309541155787595e-07,
      "loss": 0.8802,
      "step": 45140
    },
    {
      "epoch": 2.728591285429383,
      "grad_norm": 1.0761266096183384,
      "learning_rate": 2.2211259532326945e-07,
      "loss": 0.9043,
      "step": 45150
    },
    {
      "epoch": 2.729195624584517,
      "grad_norm": 0.9740322270351403,
      "learning_rate": 2.2113189952327197e-07,
      "loss": 0.884,
      "step": 45160
    },
    {
      "epoch": 2.7297999637396506,
      "grad_norm": 1.114333437812144,
      "learning_rate": 2.2015332459311921e-07,
      "loss": 0.872,
      "step": 45170
    },
    {
      "epoch": 2.7304043028947844,
      "grad_norm": 1.0057318949706762,
      "learning_rate": 2.191768709671027e-07,
      "loss": 0.8771,
      "step": 45180
    },
    {
      "epoch": 2.7310086420499182,
      "grad_norm": 0.9943468202093073,
      "learning_rate": 2.1820253907857526e-07,
      "loss": 0.8777,
      "step": 45190
    },
    {
      "epoch": 2.7316129812050525,
      "grad_norm": 1.037643341148123,
      "learning_rate": 2.172303293599476e-07,
      "loss": 0.8726,
      "step": 45200
    },
    {
      "epoch": 2.7322173203601863,
      "grad_norm": 1.058745903853557,
      "learning_rate": 2.1626024224268628e-07,
      "loss": 0.8731,
      "step": 45210
    },
    {
      "epoch": 2.73282165951532,
      "grad_norm": 1.110355961210582,
      "learning_rate": 2.1529227815731858e-07,
      "loss": 0.8574,
      "step": 45220
    },
    {
      "epoch": 2.733425998670454,
      "grad_norm": 1.0564040481363202,
      "learning_rate": 2.143264375334292e-07,
      "loss": 0.8697,
      "step": 45230
    },
    {
      "epoch": 2.7340303378255877,
      "grad_norm": 1.0220958855142999,
      "learning_rate": 2.1336272079965857e-07,
      "loss": 0.9043,
      "step": 45240
    },
    {
      "epoch": 2.7346346769807215,
      "grad_norm": 1.0016145974739459,
      "learning_rate": 2.1240112838370686e-07,
      "loss": 0.8948,
      "step": 45250
    },
    {
      "epoch": 2.7352390161358553,
      "grad_norm": 1.163346595977779,
      "learning_rate": 2.1144166071233042e-07,
      "loss": 0.8947,
      "step": 45260
    },
    {
      "epoch": 2.7358433552909895,
      "grad_norm": 1.0993512992455559,
      "learning_rate": 2.1048431821134196e-07,
      "loss": 0.8819,
      "step": 45270
    },
    {
      "epoch": 2.736447694446123,
      "grad_norm": 1.005474838410543,
      "learning_rate": 2.0952910130561165e-07,
      "loss": 0.8862,
      "step": 45280
    },
    {
      "epoch": 2.737052033601257,
      "grad_norm": 1.0095370829479087,
      "learning_rate": 2.0857601041906816e-07,
      "loss": 0.9066,
      "step": 45290
    },
    {
      "epoch": 2.737656372756391,
      "grad_norm": 1.139606821521737,
      "learning_rate": 2.0762504597469258e-07,
      "loss": 0.894,
      "step": 45300
    },
    {
      "epoch": 2.7382607119115248,
      "grad_norm": 1.1910161161297665,
      "learning_rate": 2.0667620839452508e-07,
      "loss": 0.8878,
      "step": 45310
    },
    {
      "epoch": 2.7388650510666586,
      "grad_norm": 1.1200032044713917,
      "learning_rate": 2.0572949809966326e-07,
      "loss": 0.9016,
      "step": 45320
    },
    {
      "epoch": 2.7394693902217924,
      "grad_norm": 1.1618684672210273,
      "learning_rate": 2.0478491551025604e-07,
      "loss": 0.903,
      "step": 45330
    },
    {
      "epoch": 2.740073729376926,
      "grad_norm": 1.2146107547250926,
      "learning_rate": 2.0384246104551252e-07,
      "loss": 0.8837,
      "step": 45340
    },
    {
      "epoch": 2.74067806853206,
      "grad_norm": 1.1648456698217036,
      "learning_rate": 2.0290213512369594e-07,
      "loss": 0.8893,
      "step": 45350
    },
    {
      "epoch": 2.7412824076871942,
      "grad_norm": 1.1862111453265956,
      "learning_rate": 2.0196393816212246e-07,
      "loss": 0.8706,
      "step": 45360
    },
    {
      "epoch": 2.741886746842328,
      "grad_norm": 1.1763491035258002,
      "learning_rate": 2.0102787057716788e-07,
      "loss": 0.8933,
      "step": 45370
    },
    {
      "epoch": 2.742491085997462,
      "grad_norm": 1.157534575240225,
      "learning_rate": 2.0009393278425882e-07,
      "loss": 0.8698,
      "step": 45380
    },
    {
      "epoch": 2.7430954251525956,
      "grad_norm": 1.277193657313442,
      "learning_rate": 1.9916212519788037e-07,
      "loss": 0.8913,
      "step": 45390
    },
    {
      "epoch": 2.7436997643077294,
      "grad_norm": 1.155230518209369,
      "learning_rate": 1.9823244823156783e-07,
      "loss": 0.8665,
      "step": 45400
    },
    {
      "epoch": 2.7443041034628632,
      "grad_norm": 1.248104079354863,
      "learning_rate": 1.9730490229791567e-07,
      "loss": 0.8799,
      "step": 45410
    },
    {
      "epoch": 2.744908442617997,
      "grad_norm": 1.141762068994002,
      "learning_rate": 1.9637948780856898e-07,
      "loss": 0.8693,
      "step": 45420
    },
    {
      "epoch": 2.7455127817731313,
      "grad_norm": 1.2753267008002964,
      "learning_rate": 1.954562051742287e-07,
      "loss": 0.8851,
      "step": 45430
    },
    {
      "epoch": 2.746117120928265,
      "grad_norm": 1.2235220363401962,
      "learning_rate": 1.9453505480464985e-07,
      "loss": 0.8827,
      "step": 45440
    },
    {
      "epoch": 2.746721460083399,
      "grad_norm": 1.4858072217758969,
      "learning_rate": 1.936160371086393e-07,
      "loss": 0.8857,
      "step": 45450
    },
    {
      "epoch": 2.7473257992385327,
      "grad_norm": 1.5156490892263077,
      "learning_rate": 1.9269915249405912e-07,
      "loss": 0.9048,
      "step": 45460
    },
    {
      "epoch": 2.7479301383936665,
      "grad_norm": 1.4707327254645535,
      "learning_rate": 1.917844013678255e-07,
      "loss": 0.8756,
      "step": 45470
    },
    {
      "epoch": 2.7485344775488003,
      "grad_norm": 1.4603464920498683,
      "learning_rate": 1.908717841359048e-07,
      "loss": 0.9005,
      "step": 45480
    },
    {
      "epoch": 2.749138816703934,
      "grad_norm": 1.4919514207653686,
      "learning_rate": 1.8996130120331914e-07,
      "loss": 0.9005,
      "step": 45490
    },
    {
      "epoch": 2.7497431558590684,
      "grad_norm": 1.0327461395228872,
      "learning_rate": 1.8905295297414194e-07,
      "loss": 0.8768,
      "step": 45500
    },
    {
      "epoch": 2.7503474950142017,
      "grad_norm": 0.9820293648659041,
      "learning_rate": 1.8814673985150078e-07,
      "loss": 0.8936,
      "step": 45510
    },
    {
      "epoch": 2.750951834169336,
      "grad_norm": 0.9016692349610924,
      "learning_rate": 1.872426622375728e-07,
      "loss": 0.8827,
      "step": 45520
    },
    {
      "epoch": 2.75155617332447,
      "grad_norm": 0.937904234239,
      "learning_rate": 1.8634072053359042e-07,
      "loss": 0.8903,
      "step": 45530
    },
    {
      "epoch": 2.7521605124796036,
      "grad_norm": 0.9393857965950156,
      "learning_rate": 1.854409151398373e-07,
      "loss": 0.8987,
      "step": 45540
    },
    {
      "epoch": 2.7527648516347374,
      "grad_norm": 0.8576501034287979,
      "learning_rate": 1.845432464556468e-07,
      "loss": 0.8759,
      "step": 45550
    },
    {
      "epoch": 2.753369190789871,
      "grad_norm": 0.8383746397929337,
      "learning_rate": 1.8364771487940747e-07,
      "loss": 0.8778,
      "step": 45560
    },
    {
      "epoch": 2.753973529945005,
      "grad_norm": 0.8676877785678067,
      "learning_rate": 1.8275432080855747e-07,
      "loss": 0.8918,
      "step": 45570
    },
    {
      "epoch": 2.754577869100139,
      "grad_norm": 0.8245797010954128,
      "learning_rate": 1.8186306463958636e-07,
      "loss": 0.8806,
      "step": 45580
    },
    {
      "epoch": 2.755182208255273,
      "grad_norm": 0.7980094066911018,
      "learning_rate": 1.8097394676803437e-07,
      "loss": 0.8833,
      "step": 45590
    },
    {
      "epoch": 2.755786547410407,
      "grad_norm": 0.8814633362664858,
      "learning_rate": 1.800869675884953e-07,
      "loss": 0.8948,
      "step": 45600
    },
    {
      "epoch": 2.7563908865655407,
      "grad_norm": 0.8482749726288346,
      "learning_rate": 1.7920212749461097e-07,
      "loss": 0.9,
      "step": 45610
    },
    {
      "epoch": 2.7569952257206745,
      "grad_norm": 0.8838429290291643,
      "learning_rate": 1.78319426879075e-07,
      "loss": 0.8853,
      "step": 45620
    },
    {
      "epoch": 2.7575995648758083,
      "grad_norm": 0.9175655452716482,
      "learning_rate": 1.7743886613363125e-07,
      "loss": 0.8832,
      "step": 45630
    },
    {
      "epoch": 2.758203904030942,
      "grad_norm": 0.8627250630323202,
      "learning_rate": 1.7656044564907437e-07,
      "loss": 0.8966,
      "step": 45640
    },
    {
      "epoch": 2.758808243186076,
      "grad_norm": 0.6406229430466438,
      "learning_rate": 1.756841658152486e-07,
      "loss": 0.8836,
      "step": 45650
    },
    {
      "epoch": 2.75941258234121,
      "grad_norm": 0.6381578530896204,
      "learning_rate": 1.74810027021049e-07,
      "loss": 0.8661,
      "step": 45660
    },
    {
      "epoch": 2.7600169214963435,
      "grad_norm": 0.6559950695303506,
      "learning_rate": 1.739380296544191e-07,
      "loss": 0.8814,
      "step": 45670
    },
    {
      "epoch": 2.7606212606514777,
      "grad_norm": 0.6323016875435511,
      "learning_rate": 1.730681741023532e-07,
      "loss": 0.8697,
      "step": 45680
    },
    {
      "epoch": 2.7612255998066115,
      "grad_norm": 0.6496213978397954,
      "learning_rate": 1.7220046075089536e-07,
      "loss": 0.914,
      "step": 45690
    },
    {
      "epoch": 2.7618299389617453,
      "grad_norm": 0.6226583041147095,
      "learning_rate": 1.7133488998513692e-07,
      "loss": 0.871,
      "step": 45700
    },
    {
      "epoch": 2.762434278116879,
      "grad_norm": 0.708741715191671,
      "learning_rate": 1.7047146218922062e-07,
      "loss": 0.8799,
      "step": 45710
    },
    {
      "epoch": 2.763038617272013,
      "grad_norm": 0.7229626178654985,
      "learning_rate": 1.6961017774633714e-07,
      "loss": 0.893,
      "step": 45720
    },
    {
      "epoch": 2.763642956427147,
      "grad_norm": 0.6892131594267331,
      "learning_rate": 1.6875103703872518e-07,
      "loss": 0.8807,
      "step": 45730
    },
    {
      "epoch": 2.7642472955822806,
      "grad_norm": 0.7166797767440123,
      "learning_rate": 1.6789404044767366e-07,
      "loss": 0.8758,
      "step": 45740
    },
    {
      "epoch": 2.764851634737415,
      "grad_norm": 0.8957928780555156,
      "learning_rate": 1.670391883535194e-07,
      "loss": 0.8889,
      "step": 45750
    },
    {
      "epoch": 2.7654559738925486,
      "grad_norm": 0.9388549397024907,
      "learning_rate": 1.6618648113564616e-07,
      "loss": 0.8551,
      "step": 45760
    },
    {
      "epoch": 2.7660603130476824,
      "grad_norm": 0.8756157168474912,
      "learning_rate": 1.6533591917248737e-07,
      "loss": 0.9027,
      "step": 45770
    },
    {
      "epoch": 2.766664652202816,
      "grad_norm": 0.9666364572877677,
      "learning_rate": 1.6448750284152438e-07,
      "loss": 0.8836,
      "step": 45780
    },
    {
      "epoch": 2.76726899135795,
      "grad_norm": 0.9060640149069816,
      "learning_rate": 1.6364123251928487e-07,
      "loss": 0.8699,
      "step": 45790
    },
    {
      "epoch": 2.767873330513084,
      "grad_norm": 0.5688728041654626,
      "learning_rate": 1.6279710858134623e-07,
      "loss": 0.8861,
      "step": 45800
    },
    {
      "epoch": 2.7684776696682176,
      "grad_norm": 0.5856732913104161,
      "learning_rate": 1.6195513140233098e-07,
      "loss": 0.8692,
      "step": 45810
    },
    {
      "epoch": 2.769082008823352,
      "grad_norm": 0.5667721982668187,
      "learning_rate": 1.611153013559108e-07,
      "loss": 0.8675,
      "step": 45820
    },
    {
      "epoch": 2.7696863479784857,
      "grad_norm": 0.5945509810361672,
      "learning_rate": 1.6027761881480307e-07,
      "loss": 0.8905,
      "step": 45830
    },
    {
      "epoch": 2.7702906871336195,
      "grad_norm": 0.6282107389067669,
      "learning_rate": 1.5944208415077323e-07,
      "loss": 0.8943,
      "step": 45840
    },
    {
      "epoch": 2.7708950262887533,
      "grad_norm": 0.5977481253744285,
      "learning_rate": 1.586086977346335e-07,
      "loss": 0.8843,
      "step": 45850
    },
    {
      "epoch": 2.771499365443887,
      "grad_norm": 0.571685794633865,
      "learning_rate": 1.5777745993624028e-07,
      "loss": 0.8882,
      "step": 45860
    },
    {
      "epoch": 2.772103704599021,
      "grad_norm": 0.6065710704633975,
      "learning_rate": 1.5694837112450067e-07,
      "loss": 0.8631,
      "step": 45870
    },
    {
      "epoch": 2.7727080437541547,
      "grad_norm": 0.5964393739660016,
      "learning_rate": 1.5612143166736426e-07,
      "loss": 0.8975,
      "step": 45880
    },
    {
      "epoch": 2.773312382909289,
      "grad_norm": 0.562478649284115,
      "learning_rate": 1.5529664193182858e-07,
      "loss": 0.8839,
      "step": 45890
    },
    {
      "epoch": 2.7739167220644223,
      "grad_norm": 0.5672029556953706,
      "learning_rate": 1.5447400228393695e-07,
      "loss": 0.872,
      "step": 45900
    },
    {
      "epoch": 2.7745210612195566,
      "grad_norm": 0.5942124996834438,
      "learning_rate": 1.5365351308877785e-07,
      "loss": 0.8649,
      "step": 45910
    },
    {
      "epoch": 2.7751254003746904,
      "grad_norm": 0.6086671159664907,
      "learning_rate": 1.528351747104856e-07,
      "loss": 0.8903,
      "step": 45920
    },
    {
      "epoch": 2.775729739529824,
      "grad_norm": 0.5986512656298831,
      "learning_rate": 1.5201898751224075e-07,
      "loss": 0.896,
      "step": 45930
    },
    {
      "epoch": 2.776334078684958,
      "grad_norm": 0.5987795832523434,
      "learning_rate": 1.512049518562686e-07,
      "loss": 0.8869,
      "step": 45940
    },
    {
      "epoch": 2.7769384178400918,
      "grad_norm": 0.5903942970017256,
      "learning_rate": 1.5039306810383848e-07,
      "loss": 0.8771,
      "step": 45950
    },
    {
      "epoch": 2.7775427569952256,
      "grad_norm": 0.5643675712703505,
      "learning_rate": 1.4958333661526713e-07,
      "loss": 0.8796,
      "step": 45960
    },
    {
      "epoch": 2.7781470961503594,
      "grad_norm": 0.5488452337788334,
      "learning_rate": 1.4877575774991493e-07,
      "loss": 0.8992,
      "step": 45970
    },
    {
      "epoch": 2.7787514353054936,
      "grad_norm": 0.5680377663391452,
      "learning_rate": 1.4797033186618513e-07,
      "loss": 0.8511,
      "step": 45980
    },
    {
      "epoch": 2.7793557744606274,
      "grad_norm": 0.5759190022229727,
      "learning_rate": 1.4716705932152798e-07,
      "loss": 0.8712,
      "step": 45990
    },
    {
      "epoch": 2.7799601136157612,
      "grad_norm": 0.598760258147706,
      "learning_rate": 1.4636594047243823e-07,
      "loss": 0.8885,
      "step": 46000
    },
    {
      "epoch": 2.780564452770895,
      "grad_norm": 0.5826932912802716,
      "learning_rate": 1.455669756744521e-07,
      "loss": 0.888,
      "step": 46010
    },
    {
      "epoch": 2.781168791926029,
      "grad_norm": 0.561753710232754,
      "learning_rate": 1.4477016528215259e-07,
      "loss": 0.88,
      "step": 46020
    },
    {
      "epoch": 2.7817731310811626,
      "grad_norm": 0.5581961395981817,
      "learning_rate": 1.4397550964916574e-07,
      "loss": 0.904,
      "step": 46030
    },
    {
      "epoch": 2.7823774702362964,
      "grad_norm": 0.568751185955918,
      "learning_rate": 1.4318300912815997e-07,
      "loss": 0.8561,
      "step": 46040
    },
    {
      "epoch": 2.7829818093914307,
      "grad_norm": 0.5907567741634793,
      "learning_rate": 1.4239266407084895e-07,
      "loss": 0.8942,
      "step": 46050
    },
    {
      "epoch": 2.7835861485465645,
      "grad_norm": 0.597394406216666,
      "learning_rate": 1.416044748279899e-07,
      "loss": 0.9099,
      "step": 46060
    },
    {
      "epoch": 2.7841904877016983,
      "grad_norm": 0.6043521594637002,
      "learning_rate": 1.4081844174938076e-07,
      "loss": 0.8799,
      "step": 46070
    },
    {
      "epoch": 2.784794826856832,
      "grad_norm": 0.5979420961631614,
      "learning_rate": 1.4003456518386582e-07,
      "loss": 0.8755,
      "step": 46080
    },
    {
      "epoch": 2.785399166011966,
      "grad_norm": 0.6415000168194408,
      "learning_rate": 1.3925284547933126e-07,
      "loss": 0.8785,
      "step": 46090
    },
    {
      "epoch": 2.7860035051670997,
      "grad_norm": 0.7312409200714377,
      "learning_rate": 1.38473282982704e-07,
      "loss": 0.9052,
      "step": 46100
    },
    {
      "epoch": 2.7866078443222335,
      "grad_norm": 0.676311515345982,
      "learning_rate": 1.376958780399562e-07,
      "loss": 0.8847,
      "step": 46110
    },
    {
      "epoch": 2.7872121834773678,
      "grad_norm": 0.7411261653744863,
      "learning_rate": 1.3692063099610187e-07,
      "loss": 0.8878,
      "step": 46120
    },
    {
      "epoch": 2.787816522632501,
      "grad_norm": 0.6782575006435092,
      "learning_rate": 1.3614754219519577e-07,
      "loss": 0.8748,
      "step": 46130
    },
    {
      "epoch": 2.7884208617876354,
      "grad_norm": 0.6904883125058752,
      "learning_rate": 1.3537661198033735e-07,
      "loss": 0.8878,
      "step": 46140
    },
    {
      "epoch": 2.789025200942769,
      "grad_norm": 0.7617510133608474,
      "learning_rate": 1.3460784069366573e-07,
      "loss": 0.8855,
      "step": 46150
    },
    {
      "epoch": 2.789629540097903,
      "grad_norm": 0.7200044659977112,
      "learning_rate": 1.338412286763635e-07,
      "loss": 0.9089,
      "step": 46160
    },
    {
      "epoch": 2.790233879253037,
      "grad_norm": 0.7772490473931158,
      "learning_rate": 1.3307677626865413e-07,
      "loss": 0.8796,
      "step": 46170
    },
    {
      "epoch": 2.7908382184081706,
      "grad_norm": 0.7099030513901076,
      "learning_rate": 1.3231448380980284e-07,
      "loss": 0.8831,
      "step": 46180
    },
    {
      "epoch": 2.7914425575633044,
      "grad_norm": 0.7022387939697349,
      "learning_rate": 1.3155435163811681e-07,
      "loss": 0.8799,
      "step": 46190
    },
    {
      "epoch": 2.792046896718438,
      "grad_norm": 0.7088038703085388,
      "learning_rate": 1.3079638009094286e-07,
      "loss": 0.9003,
      "step": 46200
    },
    {
      "epoch": 2.7926512358735724,
      "grad_norm": 0.6788177111866023,
      "learning_rate": 1.3004056950467135e-07,
      "loss": 0.897,
      "step": 46210
    },
    {
      "epoch": 2.7932555750287063,
      "grad_norm": 0.7283795620806107,
      "learning_rate": 1.2928692021473231e-07,
      "loss": 0.8723,
      "step": 46220
    },
    {
      "epoch": 2.79385991418384,
      "grad_norm": 0.6838402674598276,
      "learning_rate": 1.2853543255559542e-07,
      "loss": 0.8841,
      "step": 46230
    },
    {
      "epoch": 2.794464253338974,
      "grad_norm": 0.7045179880227102,
      "learning_rate": 1.2778610686077274e-07,
      "loss": 0.8916,
      "step": 46240
    },
    {
      "epoch": 2.7950685924941077,
      "grad_norm": 0.7993308797096246,
      "learning_rate": 1.270389434628172e-07,
      "loss": 0.8572,
      "step": 46250
    },
    {
      "epoch": 2.7956729316492415,
      "grad_norm": 0.8571924094544897,
      "learning_rate": 1.262939426933202e-07,
      "loss": 0.8867,
      "step": 46260
    },
    {
      "epoch": 2.7962772708043753,
      "grad_norm": 0.8306968904325757,
      "learning_rate": 1.255511048829139e-07,
      "loss": 0.8717,
      "step": 46270
    },
    {
      "epoch": 2.7968816099595095,
      "grad_norm": 0.8465291540343134,
      "learning_rate": 1.2481043036127238e-07,
      "loss": 0.8838,
      "step": 46280
    },
    {
      "epoch": 2.797485949114643,
      "grad_norm": 0.8117943044821256,
      "learning_rate": 1.2407191945710718e-07,
      "loss": 0.8646,
      "step": 46290
    },
    {
      "epoch": 2.798090288269777,
      "grad_norm": 0.8373645276096434,
      "learning_rate": 1.2333557249817107e-07,
      "loss": 0.8978,
      "step": 46300
    },
    {
      "epoch": 2.798694627424911,
      "grad_norm": 0.8100752631612739,
      "learning_rate": 1.226013898112566e-07,
      "loss": 0.8922,
      "step": 46310
    },
    {
      "epoch": 2.7992989665800447,
      "grad_norm": 0.8183693837632486,
      "learning_rate": 1.218693717221947e-07,
      "loss": 0.9128,
      "step": 46320
    },
    {
      "epoch": 2.7999033057351785,
      "grad_norm": 0.8012229286627883,
      "learning_rate": 1.211395185558556e-07,
      "loss": 0.8729,
      "step": 46330
    },
    {
      "epoch": 2.8005076448903123,
      "grad_norm": 0.8390485626989796,
      "learning_rate": 1.2041183063615126e-07,
      "loss": 0.8774,
      "step": 46340
    },
    {
      "epoch": 2.801111984045446,
      "grad_norm": 1.3003713487067985,
      "learning_rate": 1.196863082860289e-07,
      "loss": 0.8884,
      "step": 46350
    },
    {
      "epoch": 2.80171632320058,
      "grad_norm": 1.1966820725481797,
      "learning_rate": 1.1896295182747709e-07,
      "loss": 0.8729,
      "step": 46360
    },
    {
      "epoch": 2.802320662355714,
      "grad_norm": 1.2283574882036994,
      "learning_rate": 1.1824176158152346e-07,
      "loss": 0.9012,
      "step": 46370
    },
    {
      "epoch": 2.802925001510848,
      "grad_norm": 1.2925235894327627,
      "learning_rate": 1.175227378682331e-07,
      "loss": 0.8767,
      "step": 46380
    },
    {
      "epoch": 2.803529340665982,
      "grad_norm": 1.3404863705472452,
      "learning_rate": 1.1680588100670964e-07,
      "loss": 0.8973,
      "step": 46390
    },
    {
      "epoch": 2.8041336798211156,
      "grad_norm": 0.8819776330937119,
      "learning_rate": 1.1609119131509583e-07,
      "loss": 0.8977,
      "step": 46400
    },
    {
      "epoch": 2.8047380189762494,
      "grad_norm": 1.3152709206599904,
      "learning_rate": 1.1537866911057183e-07,
      "loss": 0.8782,
      "step": 46410
    },
    {
      "epoch": 2.805342358131383,
      "grad_norm": 1.1882071994148973,
      "learning_rate": 1.146683147093558e-07,
      "loss": 0.8647,
      "step": 46420
    },
    {
      "epoch": 2.805946697286517,
      "grad_norm": 0.7900719459903853,
      "learning_rate": 1.1396012842670556e-07,
      "loss": 0.8659,
      "step": 46430
    },
    {
      "epoch": 2.8065510364416513,
      "grad_norm": 1.024499184682017,
      "learning_rate": 1.132541105769136e-07,
      "loss": 0.8795,
      "step": 46440
    },
    {
      "epoch": 2.807155375596785,
      "grad_norm": 2.2376591793600547,
      "learning_rate": 1.1255026147331316e-07,
      "loss": 0.8906,
      "step": 46450
    },
    {
      "epoch": 2.807759714751919,
      "grad_norm": 2.556276121537409,
      "learning_rate": 1.1184858142827382e-07,
      "loss": 0.8678,
      "step": 46460
    },
    {
      "epoch": 2.8083640539070527,
      "grad_norm": 2.3118055345112056,
      "learning_rate": 1.1114907075320147e-07,
      "loss": 0.8959,
      "step": 46470
    },
    {
      "epoch": 2.8089683930621865,
      "grad_norm": 2.5783981980041775,
      "learning_rate": 1.1045172975854057e-07,
      "loss": 0.8735,
      "step": 46480
    },
    {
      "epoch": 2.8095727322173203,
      "grad_norm": 2.346724448745516,
      "learning_rate": 1.0975655875377244e-07,
      "loss": 0.8707,
      "step": 46490
    },
    {
      "epoch": 2.810177071372454,
      "grad_norm": 1.043861123232615,
      "learning_rate": 1.090635580474153e-07,
      "loss": 0.8718,
      "step": 46500
    },
    {
      "epoch": 2.8107814105275883,
      "grad_norm": 0.9994293790810496,
      "learning_rate": 1.0837272794702314e-07,
      "loss": 0.9039,
      "step": 46510
    },
    {
      "epoch": 2.8113857496827217,
      "grad_norm": 0.9405530801990281,
      "learning_rate": 1.0768406875918847e-07,
      "loss": 0.8895,
      "step": 46520
    },
    {
      "epoch": 2.811990088837856,
      "grad_norm": 0.8708076821707352,
      "learning_rate": 1.0699758078953904e-07,
      "loss": 0.8929,
      "step": 46530
    },
    {
      "epoch": 2.8125944279929898,
      "grad_norm": 0.9292307378624338,
      "learning_rate": 1.0631326434273948e-07,
      "loss": 0.873,
      "step": 46540
    },
    {
      "epoch": 2.8131987671481236,
      "grad_norm": 0.9734444244612962,
      "learning_rate": 1.0563111972249018e-07,
      "loss": 0.8856,
      "step": 46550
    },
    {
      "epoch": 2.8138031063032574,
      "grad_norm": 0.9675862271215961,
      "learning_rate": 1.04951147231529e-07,
      "loss": 0.8655,
      "step": 46560
    },
    {
      "epoch": 2.814407445458391,
      "grad_norm": 0.9877806037397211,
      "learning_rate": 1.0427334717162841e-07,
      "loss": 0.8882,
      "step": 46570
    },
    {
      "epoch": 2.815011784613525,
      "grad_norm": 0.9748356763372517,
      "learning_rate": 1.0359771984359668e-07,
      "loss": 0.855,
      "step": 46580
    },
    {
      "epoch": 2.8156161237686588,
      "grad_norm": 1.1603181515046335,
      "learning_rate": 1.0292426554727953e-07,
      "loss": 0.8828,
      "step": 46590
    },
    {
      "epoch": 2.816220462923793,
      "grad_norm": 1.0489701291607032,
      "learning_rate": 1.0225298458155563e-07,
      "loss": 0.8747,
      "step": 46600
    },
    {
      "epoch": 2.816824802078927,
      "grad_norm": 1.1263964630292564,
      "learning_rate": 1.0158387724434171e-07,
      "loss": 0.8943,
      "step": 46610
    },
    {
      "epoch": 2.8174291412340606,
      "grad_norm": 1.0173615825873077,
      "learning_rate": 1.009169438325891e-07,
      "loss": 0.8525,
      "step": 46620
    },
    {
      "epoch": 2.8180334803891944,
      "grad_norm": 1.1670095734263473,
      "learning_rate": 1.002521846422827e-07,
      "loss": 0.8893,
      "step": 46630
    },
    {
      "epoch": 2.8186378195443282,
      "grad_norm": 1.0737130949007663,
      "learning_rate": 9.958959996844486e-08,
      "loss": 0.8706,
      "step": 46640
    },
    {
      "epoch": 2.819242158699462,
      "grad_norm": 1.168097788390359,
      "learning_rate": 9.892919010513147e-08,
      "loss": 0.8948,
      "step": 46650
    },
    {
      "epoch": 2.819846497854596,
      "grad_norm": 1.1803821624601578,
      "learning_rate": 9.827095534543307e-08,
      "loss": 0.9075,
      "step": 46660
    },
    {
      "epoch": 2.82045083700973,
      "grad_norm": 1.0261548376509666,
      "learning_rate": 9.761489598147655e-08,
      "loss": 0.8846,
      "step": 46670
    },
    {
      "epoch": 2.8210551761648635,
      "grad_norm": 1.1563884522704913,
      "learning_rate": 9.696101230442067e-08,
      "loss": 0.8839,
      "step": 46680
    },
    {
      "epoch": 2.8216595153199977,
      "grad_norm": 1.1706999991284068,
      "learning_rate": 9.630930460446163e-08,
      "loss": 0.8911,
      "step": 46690
    },
    {
      "epoch": 2.8222638544751315,
      "grad_norm": 1.311721728642041,
      "learning_rate": 9.565977317082698e-08,
      "loss": 0.8861,
      "step": 46700
    },
    {
      "epoch": 2.8228681936302653,
      "grad_norm": 1.2769329052758165,
      "learning_rate": 9.501241829178109e-08,
      "loss": 0.9037,
      "step": 46710
    },
    {
      "epoch": 2.823472532785399,
      "grad_norm": 1.2206326222224786,
      "learning_rate": 9.43672402546203e-08,
      "loss": 0.9067,
      "step": 46720
    },
    {
      "epoch": 2.824076871940533,
      "grad_norm": 1.181323136426952,
      "learning_rate": 9.37242393456761e-08,
      "loss": 0.8912,
      "step": 46730
    },
    {
      "epoch": 2.8246812110956667,
      "grad_norm": 1.2200265128827024,
      "learning_rate": 9.308341585031356e-08,
      "loss": 0.875,
      "step": 46740
    },
    {
      "epoch": 2.8252855502508005,
      "grad_norm": 1.014588840435437,
      "learning_rate": 9.244477005293129e-08,
      "loss": 0.8915,
      "step": 46750
    },
    {
      "epoch": 2.8258898894059348,
      "grad_norm": 0.9571812270972648,
      "learning_rate": 9.180830223696036e-08,
      "loss": 0.9121,
      "step": 46760
    },
    {
      "epoch": 2.8264942285610686,
      "grad_norm": 1.0493186897081646,
      "learning_rate": 9.117401268486814e-08,
      "loss": 0.8826,
      "step": 46770
    },
    {
      "epoch": 2.8270985677162024,
      "grad_norm": 1.0642303083861215,
      "learning_rate": 9.054190167815114e-08,
      "loss": 0.8859,
      "step": 46780
    },
    {
      "epoch": 2.827702906871336,
      "grad_norm": 1.0278710984688342,
      "learning_rate": 8.991196949734327e-08,
      "loss": 0.8623,
      "step": 46790
    },
    {
      "epoch": 2.82830724602647,
      "grad_norm": 1.0879670282751883,
      "learning_rate": 8.928421642200869e-08,
      "loss": 0.8858,
      "step": 46800
    },
    {
      "epoch": 2.828911585181604,
      "grad_norm": 1.0288407306604375,
      "learning_rate": 8.865864273074565e-08,
      "loss": 0.8781,
      "step": 46810
    },
    {
      "epoch": 2.8295159243367376,
      "grad_norm": 1.0787730109493485,
      "learning_rate": 8.803524870118374e-08,
      "loss": 0.885,
      "step": 46820
    },
    {
      "epoch": 2.830120263491872,
      "grad_norm": 1.077160638388282,
      "learning_rate": 8.741403460998721e-08,
      "loss": 0.8704,
      "step": 46830
    },
    {
      "epoch": 2.8307246026470056,
      "grad_norm": 1.0204152608102433,
      "learning_rate": 8.679500073285274e-08,
      "loss": 0.8982,
      "step": 46840
    },
    {
      "epoch": 2.8313289418021395,
      "grad_norm": 1.021120619901927,
      "learning_rate": 8.617814734450669e-08,
      "loss": 0.895,
      "step": 46850
    },
    {
      "epoch": 2.8319332809572733,
      "grad_norm": 1.0219136918527307,
      "learning_rate": 8.556347471871118e-08,
      "loss": 0.8772,
      "step": 46860
    },
    {
      "epoch": 2.832537620112407,
      "grad_norm": 1.0673649062757689,
      "learning_rate": 8.495098312825856e-08,
      "loss": 0.8896,
      "step": 46870
    },
    {
      "epoch": 2.833141959267541,
      "grad_norm": 1.1195832357987203,
      "learning_rate": 8.434067284497304e-08,
      "loss": 0.9054,
      "step": 46880
    },
    {
      "epoch": 2.8337462984226747,
      "grad_norm": 1.0242138680746011,
      "learning_rate": 8.373254413971243e-08,
      "loss": 0.8871,
      "step": 46890
    },
    {
      "epoch": 2.834350637577809,
      "grad_norm": 1.230775567341354,
      "learning_rate": 8.312659728236471e-08,
      "loss": 0.8855,
      "step": 46900
    },
    {
      "epoch": 2.8349549767329423,
      "grad_norm": 1.2705193051277628,
      "learning_rate": 8.252283254184979e-08,
      "loss": 0.8959,
      "step": 46910
    },
    {
      "epoch": 2.8355593158880765,
      "grad_norm": 1.170057308394656,
      "learning_rate": 8.192125018611996e-08,
      "loss": 0.9018,
      "step": 46920
    },
    {
      "epoch": 2.8361636550432103,
      "grad_norm": 1.2200485884000922,
      "learning_rate": 8.132185048215835e-08,
      "loss": 0.9074,
      "step": 46930
    },
    {
      "epoch": 2.836767994198344,
      "grad_norm": 1.360498759458052,
      "learning_rate": 8.072463369597994e-08,
      "loss": 0.8915,
      "step": 46940
    },
    {
      "epoch": 2.837372333353478,
      "grad_norm": 1.1551758157361498,
      "learning_rate": 8.012960009262993e-08,
      "loss": 0.874,
      "step": 46950
    },
    {
      "epoch": 2.8379766725086117,
      "grad_norm": 1.1353881574746296,
      "learning_rate": 7.953674993618599e-08,
      "loss": 0.8864,
      "step": 46960
    },
    {
      "epoch": 2.8385810116637455,
      "grad_norm": 1.1125432873158816,
      "learning_rate": 7.894608348975541e-08,
      "loss": 0.8791,
      "step": 46970
    },
    {
      "epoch": 2.8391853508188793,
      "grad_norm": 1.2469623921709778,
      "learning_rate": 7.835760101547684e-08,
      "loss": 0.8754,
      "step": 46980
    },
    {
      "epoch": 2.8397896899740136,
      "grad_norm": 1.1181508124100181,
      "learning_rate": 7.777130277452082e-08,
      "loss": 0.868,
      "step": 46990
    },
    {
      "epoch": 2.8403940291291474,
      "grad_norm": 1.2532438025038035,
      "learning_rate": 7.718718902708589e-08,
      "loss": 0.8857,
      "step": 47000
    },
    {
      "epoch": 2.840998368284281,
      "grad_norm": 1.2132650148866018,
      "learning_rate": 7.66052600324041e-08,
      "loss": 0.8832,
      "step": 47010
    },
    {
      "epoch": 2.841602707439415,
      "grad_norm": 1.2552440728721452,
      "learning_rate": 7.602551604873665e-08,
      "loss": 0.8784,
      "step": 47020
    },
    {
      "epoch": 2.842207046594549,
      "grad_norm": 1.249658242160488,
      "learning_rate": 7.544795733337384e-08,
      "loss": 0.9048,
      "step": 47030
    },
    {
      "epoch": 2.8428113857496826,
      "grad_norm": 1.2234935983664768,
      "learning_rate": 7.487258414263732e-08,
      "loss": 0.8924,
      "step": 47040
    },
    {
      "epoch": 2.8434157249048164,
      "grad_norm": 1.4385600249385777,
      "learning_rate": 7.429939673188003e-08,
      "loss": 0.8917,
      "step": 47050
    },
    {
      "epoch": 2.8440200640599507,
      "grad_norm": 1.4487850586067208,
      "learning_rate": 7.372839535548182e-08,
      "loss": 0.8882,
      "step": 47060
    },
    {
      "epoch": 2.844624403215084,
      "grad_norm": 1.4570299716789237,
      "learning_rate": 7.315958026685555e-08,
      "loss": 0.9054,
      "step": 47070
    },
    {
      "epoch": 2.8452287423702183,
      "grad_norm": 1.3553314367497842,
      "learning_rate": 7.25929517184415e-08,
      "loss": 0.8788,
      "step": 47080
    },
    {
      "epoch": 2.845833081525352,
      "grad_norm": 1.506834707658984,
      "learning_rate": 7.202850996171018e-08,
      "loss": 0.8908,
      "step": 47090
    },
    {
      "epoch": 2.846437420680486,
      "grad_norm": 0.907679281978945,
      "learning_rate": 7.146625524716233e-08,
      "loss": 0.8972,
      "step": 47100
    },
    {
      "epoch": 2.8470417598356197,
      "grad_norm": 0.920633696020761,
      "learning_rate": 7.090618782432778e-08,
      "loss": 0.9007,
      "step": 47110
    },
    {
      "epoch": 2.8476460989907535,
      "grad_norm": 0.9474354164229694,
      "learning_rate": 7.034830794176439e-08,
      "loss": 0.8777,
      "step": 47120
    },
    {
      "epoch": 2.8482504381458877,
      "grad_norm": 0.8976442716170314,
      "learning_rate": 6.979261584706132e-08,
      "loss": 0.8723,
      "step": 47130
    },
    {
      "epoch": 2.848854777301021,
      "grad_norm": 0.9841409651818649,
      "learning_rate": 6.923911178683518e-08,
      "loss": 0.8794,
      "step": 47140
    },
    {
      "epoch": 2.8494591164561553,
      "grad_norm": 0.8804013532486229,
      "learning_rate": 6.868779600673226e-08,
      "loss": 0.889,
      "step": 47150
    },
    {
      "epoch": 2.850063455611289,
      "grad_norm": 0.8852889331177264,
      "learning_rate": 6.813866875142683e-08,
      "loss": 0.8951,
      "step": 47160
    },
    {
      "epoch": 2.850667794766423,
      "grad_norm": 0.8098427558182276,
      "learning_rate": 6.759173026462285e-08,
      "loss": 0.8771,
      "step": 47170
    },
    {
      "epoch": 2.8512721339215568,
      "grad_norm": 0.8850582175581926,
      "learning_rate": 6.704698078905281e-08,
      "loss": 0.9061,
      "step": 47180
    },
    {
      "epoch": 2.8518764730766906,
      "grad_norm": 0.8148389397287765,
      "learning_rate": 6.650442056647721e-08,
      "loss": 0.8766,
      "step": 47190
    },
    {
      "epoch": 2.8524808122318244,
      "grad_norm": 0.8667722304988269,
      "learning_rate": 6.596404983768512e-08,
      "loss": 0.859,
      "step": 47200
    },
    {
      "epoch": 2.853085151386958,
      "grad_norm": 0.8414301200296546,
      "learning_rate": 6.54258688424947e-08,
      "loss": 0.8863,
      "step": 47210
    },
    {
      "epoch": 2.8536894905420924,
      "grad_norm": 0.8366656816231236,
      "learning_rate": 6.488987781975042e-08,
      "loss": 0.9028,
      "step": 47220
    },
    {
      "epoch": 2.8542938296972262,
      "grad_norm": 0.7745611905990553,
      "learning_rate": 6.435607700732649e-08,
      "loss": 0.8702,
      "step": 47230
    },
    {
      "epoch": 2.85489816885236,
      "grad_norm": 0.8242775179155521,
      "learning_rate": 6.38244666421256e-08,
      "loss": 0.8738,
      "step": 47240
    },
    {
      "epoch": 2.855502508007494,
      "grad_norm": 0.6496646518401458,
      "learning_rate": 6.329504696007571e-08,
      "loss": 0.8632,
      "step": 47250
    },
    {
      "epoch": 2.8561068471626276,
      "grad_norm": 0.6345989415982591,
      "learning_rate": 6.276781819613498e-08,
      "loss": 0.8774,
      "step": 47260
    },
    {
      "epoch": 2.8567111863177614,
      "grad_norm": 0.6235959709723341,
      "learning_rate": 6.224278058428901e-08,
      "loss": 0.8847,
      "step": 47270
    },
    {
      "epoch": 2.8573155254728952,
      "grad_norm": 0.6753358926569923,
      "learning_rate": 6.171993435754974e-08,
      "loss": 0.9056,
      "step": 47280
    },
    {
      "epoch": 2.8579198646280295,
      "grad_norm": 0.6275729174849449,
      "learning_rate": 6.119927974795659e-08,
      "loss": 0.8847,
      "step": 47290
    },
    {
      "epoch": 2.858524203783163,
      "grad_norm": 0.7072188642614241,
      "learning_rate": 6.068081698657857e-08,
      "loss": 0.8856,
      "step": 47300
    },
    {
      "epoch": 2.859128542938297,
      "grad_norm": 0.6865434161367749,
      "learning_rate": 6.016454630350888e-08,
      "loss": 0.908,
      "step": 47310
    },
    {
      "epoch": 2.859732882093431,
      "grad_norm": 0.6583015188338257,
      "learning_rate": 5.965046792786977e-08,
      "loss": 0.8702,
      "step": 47320
    },
    {
      "epoch": 2.8603372212485647,
      "grad_norm": 0.7072343862163731,
      "learning_rate": 5.913858208781043e-08,
      "loss": 0.8943,
      "step": 47330
    },
    {
      "epoch": 2.8609415604036985,
      "grad_norm": 0.6780368034978116,
      "learning_rate": 5.862888901050634e-08,
      "loss": 0.8667,
      "step": 47340
    },
    {
      "epoch": 2.8615458995588323,
      "grad_norm": 0.8229330433280384,
      "learning_rate": 5.812138892216046e-08,
      "loss": 0.9096,
      "step": 47350
    },
    {
      "epoch": 2.862150238713966,
      "grad_norm": 0.8321494776305733,
      "learning_rate": 5.761608204800206e-08,
      "loss": 0.8775,
      "step": 47360
    },
    {
      "epoch": 2.8627545778691,
      "grad_norm": 0.8588862094950755,
      "learning_rate": 5.7112968612286765e-08,
      "loss": 0.8831,
      "step": 47370
    },
    {
      "epoch": 2.863358917024234,
      "grad_norm": 0.9435169344902588,
      "learning_rate": 5.661204883829763e-08,
      "loss": 0.8716,
      "step": 47380
    },
    {
      "epoch": 2.863963256179368,
      "grad_norm": 0.9734734147603962,
      "learning_rate": 5.61133229483446e-08,
      "loss": 0.8815,
      "step": 47390
    },
    {
      "epoch": 2.864567595334502,
      "grad_norm": 0.573786678514249,
      "learning_rate": 5.561679116376117e-08,
      "loss": 0.8752,
      "step": 47400
    },
    {
      "epoch": 2.8651719344896356,
      "grad_norm": 0.6191535919663005,
      "learning_rate": 5.512245370490998e-08,
      "loss": 0.8826,
      "step": 47410
    },
    {
      "epoch": 2.8657762736447694,
      "grad_norm": 0.6142616597429817,
      "learning_rate": 5.463031079117942e-08,
      "loss": 0.8775,
      "step": 47420
    },
    {
      "epoch": 2.866380612799903,
      "grad_norm": 0.6139070308326352,
      "learning_rate": 5.414036264098199e-08,
      "loss": 0.8661,
      "step": 47430
    },
    {
      "epoch": 2.866984951955037,
      "grad_norm": 0.6339430008827052,
      "learning_rate": 5.365260947175877e-08,
      "loss": 0.86,
      "step": 47440
    },
    {
      "epoch": 2.8675892911101712,
      "grad_norm": 0.6014077080311432,
      "learning_rate": 5.316705149997492e-08,
      "loss": 0.888,
      "step": 47450
    },
    {
      "epoch": 2.868193630265305,
      "grad_norm": 0.5961247251355861,
      "learning_rate": 5.2683688941121966e-08,
      "loss": 0.8806,
      "step": 47460
    },
    {
      "epoch": 2.868797969420439,
      "grad_norm": 0.5974730287761562,
      "learning_rate": 5.2202522009716647e-08,
      "loss": 0.8811,
      "step": 47470
    },
    {
      "epoch": 2.8694023085755727,
      "grad_norm": 0.564147768905625,
      "learning_rate": 5.172355091930259e-08,
      "loss": 0.8977,
      "step": 47480
    },
    {
      "epoch": 2.8700066477307065,
      "grad_norm": 0.590412262492525,
      "learning_rate": 5.124677588244698e-08,
      "loss": 0.8791,
      "step": 47490
    },
    {
      "epoch": 2.8706109868858403,
      "grad_norm": 0.5636910791058279,
      "learning_rate": 5.077219711074388e-08,
      "loss": 0.8761,
      "step": 47500
    },
    {
      "epoch": 2.871215326040974,
      "grad_norm": 0.586483313663193,
      "learning_rate": 5.029981481481261e-08,
      "loss": 0.887,
      "step": 47510
    },
    {
      "epoch": 2.8718196651961083,
      "grad_norm": 0.5932535536691216,
      "learning_rate": 4.982962920429657e-08,
      "loss": 0.9017,
      "step": 47520
    },
    {
      "epoch": 2.8724240043512417,
      "grad_norm": 0.6065631818589902,
      "learning_rate": 4.93616404878644e-08,
      "loss": 0.8784,
      "step": 47530
    },
    {
      "epoch": 2.873028343506376,
      "grad_norm": 0.6011846516671374,
      "learning_rate": 4.889584887321164e-08,
      "loss": 0.8696,
      "step": 47540
    },
    {
      "epoch": 2.8736326826615097,
      "grad_norm": 0.5803400381946026,
      "learning_rate": 4.84322545670568e-08,
      "loss": 0.8775,
      "step": 47550
    },
    {
      "epoch": 2.8742370218166435,
      "grad_norm": 0.5703820696290958,
      "learning_rate": 4.797085777514365e-08,
      "loss": 0.8762,
      "step": 47560
    },
    {
      "epoch": 2.8748413609717773,
      "grad_norm": 0.5768360571919828,
      "learning_rate": 4.751165870224117e-08,
      "loss": 0.8815,
      "step": 47570
    },
    {
      "epoch": 2.875445700126911,
      "grad_norm": 0.5850061461151962,
      "learning_rate": 4.7054657552143e-08,
      "loss": 0.8584,
      "step": 47580
    },
    {
      "epoch": 2.876050039282045,
      "grad_norm": 0.5914027341537174,
      "learning_rate": 4.65998545276658e-08,
      "loss": 0.8738,
      "step": 47590
    },
    {
      "epoch": 2.8766543784371787,
      "grad_norm": 0.6017551131997185,
      "learning_rate": 4.614724983065311e-08,
      "loss": 0.913,
      "step": 47600
    },
    {
      "epoch": 2.877258717592313,
      "grad_norm": 0.5979821184315797,
      "learning_rate": 4.569684366197147e-08,
      "loss": 0.8994,
      "step": 47610
    },
    {
      "epoch": 2.877863056747447,
      "grad_norm": 0.5962390370357233,
      "learning_rate": 4.5248636221511543e-08,
      "loss": 0.8763,
      "step": 47620
    },
    {
      "epoch": 2.8784673959025806,
      "grad_norm": 0.6200529278695698,
      "learning_rate": 4.4802627708188105e-08,
      "loss": 0.8984,
      "step": 47630
    },
    {
      "epoch": 2.8790717350577144,
      "grad_norm": 0.6336552635198061,
      "learning_rate": 4.4358818319942266e-08,
      "loss": 0.8778,
      "step": 47640
    },
    {
      "epoch": 2.879676074212848,
      "grad_norm": 0.6037715073948364,
      "learning_rate": 4.391720825373535e-08,
      "loss": 0.8739,
      "step": 47650
    },
    {
      "epoch": 2.880280413367982,
      "grad_norm": 0.6212368508294195,
      "learning_rate": 4.3477797705556156e-08,
      "loss": 0.8719,
      "step": 47660
    },
    {
      "epoch": 2.880884752523116,
      "grad_norm": 0.6252612210117532,
      "learning_rate": 4.3040586870415346e-08,
      "loss": 0.8989,
      "step": 47670
    },
    {
      "epoch": 2.88148909167825,
      "grad_norm": 0.5922084095551822,
      "learning_rate": 4.2605575942347154e-08,
      "loss": 0.8888,
      "step": 47680
    },
    {
      "epoch": 2.8820934308333834,
      "grad_norm": 0.6152783038963051,
      "learning_rate": 4.217276511441104e-08,
      "loss": 0.8819,
      "step": 47690
    },
    {
      "epoch": 2.8826977699885177,
      "grad_norm": 0.6756744366354618,
      "learning_rate": 4.1742154578688906e-08,
      "loss": 0.8694,
      "step": 47700
    },
    {
      "epoch": 2.8833021091436515,
      "grad_norm": 0.6987691784481876,
      "learning_rate": 4.131374452628678e-08,
      "loss": 0.8725,
      "step": 47710
    },
    {
      "epoch": 2.8839064482987853,
      "grad_norm": 0.6437287514862596,
      "learning_rate": 4.0887535147333125e-08,
      "loss": 0.9037,
      "step": 47720
    },
    {
      "epoch": 2.884510787453919,
      "grad_norm": 0.6385553672862551,
      "learning_rate": 4.046352663098107e-08,
      "loss": 0.8581,
      "step": 47730
    },
    {
      "epoch": 2.885115126609053,
      "grad_norm": 0.691475279436729,
      "learning_rate": 4.004171916540567e-08,
      "loss": 0.8842,
      "step": 47740
    },
    {
      "epoch": 2.8857194657641867,
      "grad_norm": 0.7065817536846365,
      "learning_rate": 3.962211293780549e-08,
      "loss": 0.8901,
      "step": 47750
    },
    {
      "epoch": 2.8863238049193205,
      "grad_norm": 0.7850325936349539,
      "learning_rate": 3.9204708134404355e-08,
      "loss": 0.8884,
      "step": 47760
    },
    {
      "epoch": 2.8869281440744547,
      "grad_norm": 0.7554187281668162,
      "learning_rate": 3.87895049404452e-08,
      "loss": 0.8723,
      "step": 47770
    },
    {
      "epoch": 2.8875324832295886,
      "grad_norm": 0.7844783819668697,
      "learning_rate": 3.837650354019673e-08,
      "loss": 0.8867,
      "step": 47780
    },
    {
      "epoch": 2.8881368223847224,
      "grad_norm": 0.7187229970698894,
      "learning_rate": 3.796570411694955e-08,
      "loss": 0.87,
      "step": 47790
    },
    {
      "epoch": 2.888741161539856,
      "grad_norm": 0.6568169544101209,
      "learning_rate": 3.755710685301728e-08,
      "loss": 0.8736,
      "step": 47800
    },
    {
      "epoch": 2.88934550069499,
      "grad_norm": 0.7062961033925668,
      "learning_rate": 3.7150711929735964e-08,
      "loss": 0.8994,
      "step": 47810
    },
    {
      "epoch": 2.8899498398501238,
      "grad_norm": 0.6620164459446279,
      "learning_rate": 3.674651952746522e-08,
      "loss": 0.8818,
      "step": 47820
    },
    {
      "epoch": 2.8905541790052576,
      "grad_norm": 0.6931280661951346,
      "learning_rate": 3.634452982558489e-08,
      "loss": 0.8825,
      "step": 47830
    },
    {
      "epoch": 2.891158518160392,
      "grad_norm": 0.6840717502640693,
      "learning_rate": 3.594474300249895e-08,
      "loss": 0.8766,
      "step": 47840
    },
    {
      "epoch": 2.8917628573155256,
      "grad_norm": 0.866572671607689,
      "learning_rate": 3.5547159235634344e-08,
      "loss": 0.893,
      "step": 47850
    },
    {
      "epoch": 2.8923671964706594,
      "grad_norm": 0.8739253775433515,
      "learning_rate": 3.515177870143938e-08,
      "loss": 0.9044,
      "step": 47860
    },
    {
      "epoch": 2.8929715356257932,
      "grad_norm": 0.8193532824668441,
      "learning_rate": 3.4758601575383686e-08,
      "loss": 0.8765,
      "step": 47870
    },
    {
      "epoch": 2.893575874780927,
      "grad_norm": 0.8031726221399412,
      "learning_rate": 3.4367628031959924e-08,
      "loss": 0.8693,
      "step": 47880
    },
    {
      "epoch": 2.894180213936061,
      "grad_norm": 0.8798552597475637,
      "learning_rate": 3.397885824468428e-08,
      "loss": 0.8782,
      "step": 47890
    },
    {
      "epoch": 2.8947845530911946,
      "grad_norm": 0.8233590042699582,
      "learning_rate": 3.359229238609207e-08,
      "loss": 0.8886,
      "step": 47900
    },
    {
      "epoch": 2.895388892246329,
      "grad_norm": 0.7913200132565852,
      "learning_rate": 3.320793062774275e-08,
      "loss": 0.882,
      "step": 47910
    },
    {
      "epoch": 2.8959932314014623,
      "grad_norm": 0.8136735262509437,
      "learning_rate": 3.282577314021596e-08,
      "loss": 0.8817,
      "step": 47920
    },
    {
      "epoch": 2.8965975705565965,
      "grad_norm": 0.8197318711129136,
      "learning_rate": 3.2445820093114365e-08,
      "loss": 0.8688,
      "step": 47930
    },
    {
      "epoch": 2.8972019097117303,
      "grad_norm": 0.8022732632023085,
      "learning_rate": 3.2068071655061426e-08,
      "loss": 0.8935,
      "step": 47940
    },
    {
      "epoch": 2.897806248866864,
      "grad_norm": 1.3033310600292838,
      "learning_rate": 3.169252799370304e-08,
      "loss": 0.8767,
      "step": 47950
    },
    {
      "epoch": 2.898410588021998,
      "grad_norm": 1.1757968660352565,
      "learning_rate": 3.131918927570532e-08,
      "loss": 0.8692,
      "step": 47960
    },
    {
      "epoch": 2.8990149271771317,
      "grad_norm": 1.2695810916233092,
      "learning_rate": 3.0948055666757404e-08,
      "loss": 0.8999,
      "step": 47970
    },
    {
      "epoch": 2.8996192663322655,
      "grad_norm": 1.283341008376628,
      "learning_rate": 3.0579127331569204e-08,
      "loss": 0.8636,
      "step": 47980
    },
    {
      "epoch": 2.9002236054873993,
      "grad_norm": 1.1818688417448828,
      "learning_rate": 3.02124044338703e-08,
      "loss": 0.887,
      "step": 47990
    },
    {
      "epoch": 2.9008279446425336,
      "grad_norm": 0.9026133091409173,
      "learning_rate": 2.98478871364144e-08,
      "loss": 0.9065,
      "step": 48000
    },
    {
      "epoch": 2.9014322837976674,
      "grad_norm": 1.047480065952849,
      "learning_rate": 2.948557560097487e-08,
      "loss": 0.8782,
      "step": 48010
    },
    {
      "epoch": 2.902036622952801,
      "grad_norm": 0.9303797384411031,
      "learning_rate": 2.9125469988345312e-08,
      "loss": 0.8851,
      "step": 48020
    },
    {
      "epoch": 2.902640962107935,
      "grad_norm": 0.9802651332582745,
      "learning_rate": 2.8767570458341775e-08,
      "loss": 0.8957,
      "step": 48030
    },
    {
      "epoch": 2.903245301263069,
      "grad_norm": 0.8896807070238727,
      "learning_rate": 2.841187716980054e-08,
      "loss": 0.8687,
      "step": 48040
    },
    {
      "epoch": 2.9038496404182026,
      "grad_norm": 2.500513041071473,
      "learning_rate": 2.8058390280578662e-08,
      "loss": 0.8643,
      "step": 48050
    },
    {
      "epoch": 2.9044539795733364,
      "grad_norm": 2.3990638653765615,
      "learning_rate": 2.7707109947554545e-08,
      "loss": 0.8858,
      "step": 48060
    },
    {
      "epoch": 2.9050583187284706,
      "grad_norm": 2.7711905370017194,
      "learning_rate": 2.735803632662737e-08,
      "loss": 0.8753,
      "step": 48070
    },
    {
      "epoch": 2.905662657883604,
      "grad_norm": 2.7376534823592937,
      "learning_rate": 2.701116957271599e-08,
      "loss": 0.8835,
      "step": 48080
    },
    {
      "epoch": 2.9062669970387383,
      "grad_norm": 2.57357269387293,
      "learning_rate": 2.6666509839760602e-08,
      "loss": 0.9,
      "step": 48090
    },
    {
      "epoch": 2.906871336193872,
      "grad_norm": 1.020523450808735,
      "learning_rate": 2.6324057280721627e-08,
      "loss": 0.869,
      "step": 48100
    },
    {
      "epoch": 2.907475675349006,
      "grad_norm": 0.8953639467395756,
      "learning_rate": 2.5983812047580824e-08,
      "loss": 0.8761,
      "step": 48110
    },
    {
      "epoch": 2.9080800145041397,
      "grad_norm": 0.9106429177907563,
      "learning_rate": 2.5645774291338522e-08,
      "loss": 0.8938,
      "step": 48120
    },
    {
      "epoch": 2.9086843536592735,
      "grad_norm": 0.9257714897382723,
      "learning_rate": 2.5309944162017486e-08,
      "loss": 0.8866,
      "step": 48130
    },
    {
      "epoch": 2.9092886928144077,
      "grad_norm": 0.9245962575317234,
      "learning_rate": 2.4976321808659054e-08,
      "loss": 0.87,
      "step": 48140
    },
    {
      "epoch": 2.909893031969541,
      "grad_norm": 1.0882567199495,
      "learning_rate": 2.4644907379325345e-08,
      "loss": 0.903,
      "step": 48150
    },
    {
      "epoch": 2.9104973711246753,
      "grad_norm": 1.0279581207737698,
      "learning_rate": 2.4315701021099258e-08,
      "loss": 0.8672,
      "step": 48160
    },
    {
      "epoch": 2.911101710279809,
      "grad_norm": 1.0636935510244523,
      "learning_rate": 2.3988702880082815e-08,
      "loss": 0.8889,
      "step": 48170
    },
    {
      "epoch": 2.911706049434943,
      "grad_norm": 0.9586831525230265,
      "learning_rate": 2.3663913101397707e-08,
      "loss": 0.8729,
      "step": 48180
    },
    {
      "epoch": 2.9123103885900767,
      "grad_norm": 1.1843819294379931,
      "learning_rate": 2.3341331829187518e-08,
      "loss": 0.8678,
      "step": 48190
    },
    {
      "epoch": 2.9129147277452105,
      "grad_norm": 0.9836143279433839,
      "learning_rate": 2.302095920661329e-08,
      "loss": 0.8816,
      "step": 48200
    },
    {
      "epoch": 2.9135190669003443,
      "grad_norm": 1.028957684726055,
      "learning_rate": 2.27027953758574e-08,
      "loss": 0.8645,
      "step": 48210
    },
    {
      "epoch": 2.914123406055478,
      "grad_norm": 1.089780173699639,
      "learning_rate": 2.2386840478121342e-08,
      "loss": 0.8551,
      "step": 48220
    },
    {
      "epoch": 2.9147277452106124,
      "grad_norm": 1.0001868740979094,
      "learning_rate": 2.2073094653626836e-08,
      "loss": 0.8663,
      "step": 48230
    },
    {
      "epoch": 2.915332084365746,
      "grad_norm": 1.0603245706445,
      "learning_rate": 2.1761558041614174e-08,
      "loss": 0.8895,
      "step": 48240
    },
    {
      "epoch": 2.91593642352088,
      "grad_norm": 1.1025176359618816,
      "learning_rate": 2.1452230780343863e-08,
      "loss": 0.8776,
      "step": 48250
    },
    {
      "epoch": 2.916540762676014,
      "grad_norm": 1.1304699483778677,
      "learning_rate": 2.1145113007096652e-08,
      "loss": 0.8793,
      "step": 48260
    },
    {
      "epoch": 2.9171451018311476,
      "grad_norm": 1.1529694127797536,
      "learning_rate": 2.0840204858171287e-08,
      "loss": 0.9053,
      "step": 48270
    },
    {
      "epoch": 2.9177494409862814,
      "grad_norm": 1.1131795130265054,
      "learning_rate": 2.0537506468886192e-08,
      "loss": 0.896,
      "step": 48280
    },
    {
      "epoch": 2.918353780141415,
      "grad_norm": 1.1483771805940317,
      "learning_rate": 2.0237017973580576e-08,
      "loss": 0.8878,
      "step": 48290
    },
    {
      "epoch": 2.9189581192965495,
      "grad_norm": 1.2378147073709478,
      "learning_rate": 1.9938739505610537e-08,
      "loss": 0.9153,
      "step": 48300
    },
    {
      "epoch": 2.919562458451683,
      "grad_norm": 1.1276109034490809,
      "learning_rate": 1.9642671197353525e-08,
      "loss": 0.8782,
      "step": 48310
    },
    {
      "epoch": 2.920166797606817,
      "grad_norm": 1.1804055671026048,
      "learning_rate": 1.934881318020443e-08,
      "loss": 0.8926,
      "step": 48320
    },
    {
      "epoch": 2.920771136761951,
      "grad_norm": 1.1453111272567047,
      "learning_rate": 1.9057165584578375e-08,
      "loss": 0.8842,
      "step": 48330
    },
    {
      "epoch": 2.9213754759170847,
      "grad_norm": 1.1895832535428406,
      "learning_rate": 1.8767728539909047e-08,
      "loss": 0.8727,
      "step": 48340
    },
    {
      "epoch": 2.9219798150722185,
      "grad_norm": 1.054963680278825,
      "learning_rate": 1.8480502174649252e-08,
      "loss": 0.8765,
      "step": 48350
    },
    {
      "epoch": 2.9225841542273523,
      "grad_norm": 1.1353672251489595,
      "learning_rate": 1.8195486616269797e-08,
      "loss": 0.881,
      "step": 48360
    },
    {
      "epoch": 2.923188493382486,
      "grad_norm": 0.9942376652737575,
      "learning_rate": 1.7912681991261728e-08,
      "loss": 0.887,
      "step": 48370
    },
    {
      "epoch": 2.92379283253762,
      "grad_norm": 1.041258106423287,
      "learning_rate": 1.763208842513464e-08,
      "loss": 0.8939,
      "step": 48380
    },
    {
      "epoch": 2.924397171692754,
      "grad_norm": 1.0094817412444108,
      "learning_rate": 1.7353706042415042e-08,
      "loss": 0.8919,
      "step": 48390
    },
    {
      "epoch": 2.925001510847888,
      "grad_norm": 1.1270933287404508,
      "learning_rate": 1.7077534966650767e-08,
      "loss": 0.8719,
      "step": 48400
    },
    {
      "epoch": 2.9256058500030218,
      "grad_norm": 1.1054662519323466,
      "learning_rate": 1.680357532040655e-08,
      "loss": 0.866,
      "step": 48410
    },
    {
      "epoch": 2.9262101891581556,
      "grad_norm": 1.043973034769781,
      "learning_rate": 1.653182722526625e-08,
      "loss": 0.8775,
      "step": 48420
    },
    {
      "epoch": 2.9268145283132894,
      "grad_norm": 1.0751898591350448,
      "learning_rate": 1.6262290801831726e-08,
      "loss": 0.8867,
      "step": 48430
    },
    {
      "epoch": 2.927418867468423,
      "grad_norm": 1.0749551779575743,
      "learning_rate": 1.5994966169724512e-08,
      "loss": 0.8628,
      "step": 48440
    },
    {
      "epoch": 2.928023206623557,
      "grad_norm": 1.1076709077630909,
      "learning_rate": 1.5729853447583044e-08,
      "loss": 0.8811,
      "step": 48450
    },
    {
      "epoch": 2.928627545778691,
      "grad_norm": 1.0501163117814052,
      "learning_rate": 1.546695275306487e-08,
      "loss": 0.8985,
      "step": 48460
    },
    {
      "epoch": 2.929231884933825,
      "grad_norm": 1.057995129125902,
      "learning_rate": 1.52062642028461e-08,
      "loss": 0.8875,
      "step": 48470
    },
    {
      "epoch": 2.929836224088959,
      "grad_norm": 1.0435990242800204,
      "learning_rate": 1.4947787912620304e-08,
      "loss": 0.8826,
      "step": 48480
    },
    {
      "epoch": 2.9304405632440926,
      "grad_norm": 1.0624525626936585,
      "learning_rate": 1.4691523997100165e-08,
      "loss": 0.8561,
      "step": 48490
    },
    {
      "epoch": 2.9310449023992264,
      "grad_norm": 1.2022924755368345,
      "learning_rate": 1.4437472570016376e-08,
      "loss": 0.8812,
      "step": 48500
    },
    {
      "epoch": 2.9316492415543602,
      "grad_norm": 1.2850191129122155,
      "learning_rate": 1.4185633744116523e-08,
      "loss": 0.8835,
      "step": 48510
    },
    {
      "epoch": 2.932253580709494,
      "grad_norm": 1.2049387789591448,
      "learning_rate": 1.3936007631167314e-08,
      "loss": 0.8672,
      "step": 48520
    },
    {
      "epoch": 2.9328579198646283,
      "grad_norm": 1.29483095187272,
      "learning_rate": 1.3688594341953465e-08,
      "loss": 0.8955,
      "step": 48530
    },
    {
      "epoch": 2.9334622590197617,
      "grad_norm": 1.195476951143966,
      "learning_rate": 1.3443393986277143e-08,
      "loss": 0.8993,
      "step": 48540
    },
    {
      "epoch": 2.934066598174896,
      "grad_norm": 1.1108318720046964,
      "learning_rate": 1.3200406672959077e-08,
      "loss": 0.8686,
      "step": 48550
    },
    {
      "epoch": 2.9346709373300297,
      "grad_norm": 1.1775040355041206,
      "learning_rate": 1.2959632509837449e-08,
      "loss": 0.9028,
      "step": 48560
    },
    {
      "epoch": 2.9352752764851635,
      "grad_norm": 1.0719683668734474,
      "learning_rate": 1.2721071603767899e-08,
      "loss": 0.878,
      "step": 48570
    },
    {
      "epoch": 2.9358796156402973,
      "grad_norm": 1.1538732366865068,
      "learning_rate": 1.248472406062462e-08,
      "loss": 0.8844,
      "step": 48580
    },
    {
      "epoch": 2.936483954795431,
      "grad_norm": 1.146548691738027,
      "learning_rate": 1.2250589985298155e-08,
      "loss": 0.8938,
      "step": 48590
    },
    {
      "epoch": 2.937088293950565,
      "grad_norm": 1.2013913050531153,
      "learning_rate": 1.2018669481698719e-08,
      "loss": 0.873,
      "step": 48600
    },
    {
      "epoch": 2.9376926331056987,
      "grad_norm": 1.2745246508822037,
      "learning_rate": 1.1788962652752312e-08,
      "loss": 0.8971,
      "step": 48610
    },
    {
      "epoch": 2.938296972260833,
      "grad_norm": 1.3084135195341944,
      "learning_rate": 1.1561469600403496e-08,
      "loss": 0.908,
      "step": 48620
    },
    {
      "epoch": 2.9389013114159668,
      "grad_norm": 1.2582654235480115,
      "learning_rate": 1.133619042561429e-08,
      "loss": 0.8644,
      "step": 48630
    },
    {
      "epoch": 2.9395056505711006,
      "grad_norm": 1.2688377667219746,
      "learning_rate": 1.1113125228363608e-08,
      "loss": 0.8864,
      "step": 48640
    },
    {
      "epoch": 2.9401099897262344,
      "grad_norm": 1.4564923069321634,
      "learning_rate": 1.0892274107647816e-08,
      "loss": 0.8878,
      "step": 48650
    },
    {
      "epoch": 2.940714328881368,
      "grad_norm": 1.4155443907845306,
      "learning_rate": 1.0673637161481842e-08,
      "loss": 0.8757,
      "step": 48660
    },
    {
      "epoch": 2.941318668036502,
      "grad_norm": 1.4997561993402064,
      "learning_rate": 1.045721448689696e-08,
      "loss": 0.9184,
      "step": 48670
    },
    {
      "epoch": 2.941923007191636,
      "grad_norm": 1.4623064230799816,
      "learning_rate": 1.024300617994134e-08,
      "loss": 0.8879,
      "step": 48680
    },
    {
      "epoch": 2.94252734634677,
      "grad_norm": 1.514055213760748,
      "learning_rate": 1.003101233568171e-08,
      "loss": 0.9062,
      "step": 48690
    },
    {
      "epoch": 2.9431316855019034,
      "grad_norm": 0.8963237880393616,
      "learning_rate": 9.821233048200596e-09,
      "loss": 0.8772,
      "step": 48700
    },
    {
      "epoch": 2.9437360246570377,
      "grad_norm": 0.8952764990034612,
      "learning_rate": 9.613668410599075e-09,
      "loss": 0.8718,
      "step": 48710
    },
    {
      "epoch": 2.9443403638121715,
      "grad_norm": 0.8975042641198281,
      "learning_rate": 9.408318514994574e-09,
      "loss": 0.9018,
      "step": 48720
    },
    {
      "epoch": 2.9449447029673053,
      "grad_norm": 0.9170974345942527,
      "learning_rate": 9.205183452520861e-09,
      "loss": 0.8969,
      "step": 48730
    },
    {
      "epoch": 2.945549042122439,
      "grad_norm": 0.9367375294242539,
      "learning_rate": 9.004263313330818e-09,
      "loss": 0.8792,
      "step": 48740
    },
    {
      "epoch": 2.946153381277573,
      "grad_norm": 0.813725206000509,
      "learning_rate": 8.805558186592566e-09,
      "loss": 0.9126,
      "step": 48750
    },
    {
      "epoch": 2.9467577204327067,
      "grad_norm": 0.8199137608373107,
      "learning_rate": 8.609068160492229e-09,
      "loss": 0.8703,
      "step": 48760
    },
    {
      "epoch": 2.9473620595878405,
      "grad_norm": 0.8002124846668378,
      "learning_rate": 8.414793322231718e-09,
      "loss": 0.8982,
      "step": 48770
    },
    {
      "epoch": 2.9479663987429747,
      "grad_norm": 0.8350032478867254,
      "learning_rate": 8.222733758030955e-09,
      "loss": 0.8955,
      "step": 48780
    },
    {
      "epoch": 2.9485707378981085,
      "grad_norm": 0.8259746165134852,
      "learning_rate": 8.032889553125644e-09,
      "loss": 0.8691,
      "step": 48790
    },
    {
      "epoch": 2.9491750770532423,
      "grad_norm": 0.8701461749155597,
      "learning_rate": 7.84526079177006e-09,
      "loss": 0.9011,
      "step": 48800
    },
    {
      "epoch": 2.949779416208376,
      "grad_norm": 0.8032006560678746,
      "learning_rate": 7.659847557233702e-09,
      "loss": 0.8733,
      "step": 48810
    },
    {
      "epoch": 2.95038375536351,
      "grad_norm": 0.8425075793643152,
      "learning_rate": 7.476649931802971e-09,
      "loss": 0.8741,
      "step": 48820
    },
    {
      "epoch": 2.9509880945186437,
      "grad_norm": 0.8846479838314857,
      "learning_rate": 7.295667996781719e-09,
      "loss": 0.8907,
      "step": 48830
    },
    {
      "epoch": 2.9515924336737775,
      "grad_norm": 0.9065296894692451,
      "learning_rate": 7.116901832489587e-09,
      "loss": 0.8846,
      "step": 48840
    },
    {
      "epoch": 2.952196772828912,
      "grad_norm": 0.6548291879523653,
      "learning_rate": 6.9403515182642215e-09,
      "loss": 0.8959,
      "step": 48850
    },
    {
      "epoch": 2.9528011119840456,
      "grad_norm": 0.6751675350049904,
      "learning_rate": 6.766017132457947e-09,
      "loss": 0.8735,
      "step": 48860
    },
    {
      "epoch": 2.9534054511391794,
      "grad_norm": 0.6479372730098051,
      "learning_rate": 6.593898752440542e-09,
      "loss": 0.8913,
      "step": 48870
    },
    {
      "epoch": 2.954009790294313,
      "grad_norm": 0.6775292365042954,
      "learning_rate": 6.42399645459979e-09,
      "loss": 0.8717,
      "step": 48880
    },
    {
      "epoch": 2.954614129449447,
      "grad_norm": 0.5785354213231347,
      "learning_rate": 6.256310314337044e-09,
      "loss": 0.8861,
      "step": 48890
    },
    {
      "epoch": 2.955218468604581,
      "grad_norm": 0.7234655571692659,
      "learning_rate": 6.090840406072773e-09,
      "loss": 0.8894,
      "step": 48900
    },
    {
      "epoch": 2.9558228077597146,
      "grad_norm": 0.6965411068752153,
      "learning_rate": 5.9275868032426795e-09,
      "loss": 0.8905,
      "step": 48910
    },
    {
      "epoch": 2.956427146914849,
      "grad_norm": 0.668334419531779,
      "learning_rate": 5.766549578298252e-09,
      "loss": 0.8803,
      "step": 48920
    },
    {
      "epoch": 2.9570314860699822,
      "grad_norm": 0.6929757132506598,
      "learning_rate": 5.607728802708434e-09,
      "loss": 0.8901,
      "step": 48930
    },
    {
      "epoch": 2.9576358252251165,
      "grad_norm": 0.6791935173477074,
      "learning_rate": 5.4511245469590635e-09,
      "loss": 0.8911,
      "step": 48940
    },
    {
      "epoch": 2.9582401643802503,
      "grad_norm": 0.8245425632724706,
      "learning_rate": 5.296736880550102e-09,
      "loss": 0.8655,
      "step": 48950
    },
    {
      "epoch": 2.958844503535384,
      "grad_norm": 0.8980330785952828,
      "learning_rate": 5.144565871999518e-09,
      "loss": 0.8862,
      "step": 48960
    },
    {
      "epoch": 2.959448842690518,
      "grad_norm": 0.95339056150385,
      "learning_rate": 4.994611588841624e-09,
      "loss": 0.8908,
      "step": 48970
    },
    {
      "epoch": 2.9600531818456517,
      "grad_norm": 0.8792979966516095,
      "learning_rate": 4.846874097625409e-09,
      "loss": 0.8749,
      "step": 48980
    },
    {
      "epoch": 2.9606575210007855,
      "grad_norm": 0.8419969610509209,
      "learning_rate": 4.701353463917868e-09,
      "loss": 0.9028,
      "step": 48990
    },
    {
      "epoch": 2.9612618601559193,
      "grad_norm": 0.5745110143833063,
      "learning_rate": 4.558049752300675e-09,
      "loss": 0.8741,
      "step": 49000
    },
    {
      "epoch": 2.9618661993110535,
      "grad_norm": 0.5823552936408872,
      "learning_rate": 4.416963026372955e-09,
      "loss": 0.8848,
      "step": 49010
    },
    {
      "epoch": 2.9624705384661874,
      "grad_norm": 0.5926183017853066,
      "learning_rate": 4.278093348748513e-09,
      "loss": 0.8727,
      "step": 49020
    },
    {
      "epoch": 2.963074877621321,
      "grad_norm": 0.6096019853074819,
      "learning_rate": 4.141440781058048e-09,
      "loss": 0.8729,
      "step": 49030
    },
    {
      "epoch": 2.963679216776455,
      "grad_norm": 0.5786432726139871,
      "learning_rate": 4.007005383948603e-09,
      "loss": 0.8788,
      "step": 49040
    },
    {
      "epoch": 2.9642835559315888,
      "grad_norm": 0.5666405146170521,
      "learning_rate": 3.874787217083009e-09,
      "loss": 0.9124,
      "step": 49050
    },
    {
      "epoch": 2.9648878950867226,
      "grad_norm": 0.5930436680869728,
      "learning_rate": 3.744786339139328e-09,
      "loss": 0.8833,
      "step": 49060
    },
    {
      "epoch": 2.9654922342418564,
      "grad_norm": 0.6061420937100483,
      "learning_rate": 3.6170028078119647e-09,
      "loss": 0.8826,
      "step": 49070
    },
    {
      "epoch": 2.9660965733969906,
      "grad_norm": 0.6213514978877125,
      "learning_rate": 3.491436679812221e-09,
      "loss": 0.9099,
      "step": 49080
    },
    {
      "epoch": 2.966700912552124,
      "grad_norm": 0.5847448638982462,
      "learning_rate": 3.3680880108666327e-09,
      "loss": 0.8754,
      "step": 49090
    },
    {
      "epoch": 2.9673052517072582,
      "grad_norm": 0.5895139162136194,
      "learning_rate": 3.246956855716965e-09,
      "loss": 0.8768,
      "step": 49100
    },
    {
      "epoch": 2.967909590862392,
      "grad_norm": 0.5746931854544143,
      "learning_rate": 3.128043268121328e-09,
      "loss": 0.8735,
      "step": 49110
    },
    {
      "epoch": 2.968513930017526,
      "grad_norm": 0.5931281529327385,
      "learning_rate": 3.0113473008541727e-09,
      "loss": 0.872,
      "step": 49120
    },
    {
      "epoch": 2.9691182691726596,
      "grad_norm": 0.603838851133666,
      "learning_rate": 2.896869005705183e-09,
      "loss": 0.9015,
      "step": 49130
    },
    {
      "epoch": 2.9697226083277934,
      "grad_norm": 0.5827662201168953,
      "learning_rate": 2.7846084334803846e-09,
      "loss": 0.8803,
      "step": 49140
    },
    {
      "epoch": 2.9703269474829277,
      "grad_norm": 0.5778573144543692,
      "learning_rate": 2.674565634001036e-09,
      "loss": 0.8891,
      "step": 49150
    },
    {
      "epoch": 2.970931286638061,
      "grad_norm": 0.5735814384398921,
      "learning_rate": 2.566740656103628e-09,
      "loss": 0.8775,
      "step": 49160
    },
    {
      "epoch": 2.9715356257931953,
      "grad_norm": 0.6061507113458374,
      "learning_rate": 2.4611335476421027e-09,
      "loss": 0.8833,
      "step": 49170
    },
    {
      "epoch": 2.972139964948329,
      "grad_norm": 0.6176135240296653,
      "learning_rate": 2.357744355484526e-09,
      "loss": 0.8965,
      "step": 49180
    },
    {
      "epoch": 2.972744304103463,
      "grad_norm": 0.5741498086221855,
      "learning_rate": 2.2565731255153045e-09,
      "loss": 0.879,
      "step": 49190
    },
    {
      "epoch": 2.9733486432585967,
      "grad_norm": 0.6250699824204273,
      "learning_rate": 2.1576199026346333e-09,
      "loss": 0.895,
      "step": 49200
    },
    {
      "epoch": 2.9739529824137305,
      "grad_norm": 0.6011562023441367,
      "learning_rate": 2.060884730757384e-09,
      "loss": 0.9049,
      "step": 49210
    },
    {
      "epoch": 2.9745573215688643,
      "grad_norm": 0.6023919580072014,
      "learning_rate": 1.9663676528158814e-09,
      "loss": 0.8621,
      "step": 49220
    },
    {
      "epoch": 2.975161660723998,
      "grad_norm": 0.6182059212995448,
      "learning_rate": 1.8740687107560164e-09,
      "loss": 0.8642,
      "step": 49230
    },
    {
      "epoch": 2.9757659998791324,
      "grad_norm": 0.585521058002708,
      "learning_rate": 1.7839879455411325e-09,
      "loss": 0.8711,
      "step": 49240
    },
    {
      "epoch": 2.976370339034266,
      "grad_norm": 0.6068836425468965,
      "learning_rate": 1.6961253971481406e-09,
      "loss": 0.9029,
      "step": 49250
    },
    {
      "epoch": 2.9769746781894,
      "grad_norm": 0.6135498475674649,
      "learning_rate": 1.610481104571404e-09,
      "loss": 0.8818,
      "step": 49260
    },
    {
      "epoch": 2.977579017344534,
      "grad_norm": 0.6011288764524885,
      "learning_rate": 1.5270551058194083e-09,
      "loss": 0.8619,
      "step": 49270
    },
    {
      "epoch": 2.9781833564996676,
      "grad_norm": 0.6289414057745571,
      "learning_rate": 1.4458474379169806e-09,
      "loss": 0.9104,
      "step": 49280
    },
    {
      "epoch": 2.9787876956548014,
      "grad_norm": 0.6623175758365316,
      "learning_rate": 1.3668581369047362e-09,
      "loss": 0.8659,
      "step": 49290
    },
    {
      "epoch": 2.979392034809935,
      "grad_norm": 0.6602552204338815,
      "learning_rate": 1.290087237837967e-09,
      "loss": 0.8834,
      "step": 49300
    },
    {
      "epoch": 2.9799963739650694,
      "grad_norm": 0.6990209477739195,
      "learning_rate": 1.2155347747871971e-09,
      "loss": 0.893,
      "step": 49310
    },
    {
      "epoch": 2.980600713120203,
      "grad_norm": 0.6904709034302106,
      "learning_rate": 1.1432007808398482e-09,
      "loss": 0.9,
      "step": 49320
    },
    {
      "epoch": 2.981205052275337,
      "grad_norm": 0.7048984983315532,
      "learning_rate": 1.0730852880969088e-09,
      "loss": 0.9005,
      "step": 49330
    },
    {
      "epoch": 2.981809391430471,
      "grad_norm": 0.6966414093410382,
      "learning_rate": 1.0051883276762653e-09,
      "loss": 0.8624,
      "step": 49340
    },
    {
      "epoch": 2.9824137305856047,
      "grad_norm": 0.8403987749537117,
      "learning_rate": 9.395099297110355e-10,
      "loss": 0.8767,
      "step": 49350
    },
    {
      "epoch": 2.9830180697407385,
      "grad_norm": 0.7515781844389828,
      "learning_rate": 8.760501233484597e-10,
      "loss": 0.8738,
      "step": 49360
    },
    {
      "epoch": 2.9836224088958723,
      "grad_norm": 0.7397597873393668,
      "learning_rate": 8.148089367532308e-10,
      "loss": 0.8998,
      "step": 49370
    },
    {
      "epoch": 2.984226748051006,
      "grad_norm": 0.7297769888916055,
      "learning_rate": 7.557863971036084e-10,
      "loss": 0.883,
      "step": 49380
    },
    {
      "epoch": 2.98483108720614,
      "grad_norm": 0.7152450326570774,
      "learning_rate": 6.989825305936393e-10,
      "loss": 0.8995,
      "step": 49390
    },
    {
      "epoch": 2.985435426361274,
      "grad_norm": 0.7379537255250919,
      "learning_rate": 6.44397362433713e-10,
      "loss": 0.8877,
      "step": 49400
    },
    {
      "epoch": 2.986039765516408,
      "grad_norm": 0.8258330177937314,
      "learning_rate": 5.920309168483407e-10,
      "loss": 0.8945,
      "step": 49410
    },
    {
      "epoch": 2.9866441046715417,
      "grad_norm": 0.721822043141609,
      "learning_rate": 5.418832170783761e-10,
      "loss": 0.8922,
      "step": 49420
    },
    {
      "epoch": 2.9872484438266755,
      "grad_norm": 0.6690799268238671,
      "learning_rate": 4.939542853787948e-10,
      "loss": 0.8755,
      "step": 49430
    },
    {
      "epoch": 2.9878527829818093,
      "grad_norm": 0.6239633496165857,
      "learning_rate": 4.482441430203599e-10,
      "loss": 0.8715,
      "step": 49440
    },
    {
      "epoch": 2.988457122136943,
      "grad_norm": 0.8337487714495238,
      "learning_rate": 4.0475281029017654e-10,
      "loss": 0.8863,
      "step": 49450
    },
    {
      "epoch": 2.989061461292077,
      "grad_norm": 0.862269789654611,
      "learning_rate": 3.6348030648947206e-10,
      "loss": 0.8955,
      "step": 49460
    },
    {
      "epoch": 2.989665800447211,
      "grad_norm": 0.8550932641644109,
      "learning_rate": 3.24426649934706e-10,
      "loss": 0.8948,
      "step": 49470
    },
    {
      "epoch": 2.990270139602345,
      "grad_norm": 0.8354210161578469,
      "learning_rate": 2.875918579581249e-10,
      "loss": 0.8778,
      "step": 49480
    },
    {
      "epoch": 2.990874478757479,
      "grad_norm": 0.9036768662399685,
      "learning_rate": 2.5297594690720797e-10,
      "loss": 0.8821,
      "step": 49490
    },
    {
      "epoch": 2.9914788179126126,
      "grad_norm": 0.8855291970675718,
      "learning_rate": 2.205789321446661e-10,
      "loss": 0.8829,
      "step": 49500
    },
    {
      "epoch": 2.9920831570677464,
      "grad_norm": 0.8539480963425093,
      "learning_rate": 1.904008280478875e-10,
      "loss": 0.8799,
      "step": 49510
    },
    {
      "epoch": 2.99268749622288,
      "grad_norm": 0.8335814596473017,
      "learning_rate": 1.624416480106028e-10,
      "loss": 0.8832,
      "step": 49520
    },
    {
      "epoch": 2.993291835378014,
      "grad_norm": 0.8132456530146003,
      "learning_rate": 1.3670140444010937e-10,
      "loss": 0.8832,
      "step": 49530
    },
    {
      "epoch": 2.9938961745331483,
      "grad_norm": 0.8449260546807574,
      "learning_rate": 1.1318010876115725e-10,
      "loss": 0.8905,
      "step": 49540
    },
    {
      "epoch": 2.9945005136882816,
      "grad_norm": 1.2398274573494132,
      "learning_rate": 9.187777141206333e-11,
      "loss": 0.896,
      "step": 49550
    },
    {
      "epoch": 2.995104852843416,
      "grad_norm": 1.0779526193921622,
      "learning_rate": 7.279440184637665e-11,
      "loss": 0.883,
      "step": 49560
    },
    {
      "epoch": 2.9957091919985497,
      "grad_norm": 1.3347277897980097,
      "learning_rate": 5.5930008533988667e-11,
      "loss": 0.9023,
      "step": 49570
    },
    {
      "epoch": 2.9963135311536835,
      "grad_norm": 1.240232862369335,
      "learning_rate": 4.12845989594679e-11,
      "loss": 0.8884,
      "step": 49580
    },
    {
      "epoch": 2.9969178703088173,
      "grad_norm": 1.222159608521956,
      "learning_rate": 2.8858179621504835e-11,
      "loss": 0.8993,
      "step": 49590
    },
    {
      "epoch": 2.997522209463951,
      "grad_norm": 0.8417692759125358,
      "learning_rate": 1.8650756036242555e-11,
      "loss": 0.8806,
      "step": 49600
    },
    {
      "epoch": 2.998126548619085,
      "grad_norm": 0.9887157162262514,
      "learning_rate": 1.0662332732280789e-11,
      "loss": 0.8635,
      "step": 49610
    },
    {
      "epoch": 2.9987308877742187,
      "grad_norm": 0.9703831343529619,
      "learning_rate": 4.892913256226984e-12,
      "loss": 0.8827,
      "step": 49620
    },
    {
      "epoch": 2.999335226929353,
      "grad_norm": 1.0022784710492407,
      "learning_rate": 1.3425001677003224e-12,
      "loss": 0.8881,
      "step": 49630
    },
    {
      "epoch": 2.9999395660844868,
      "grad_norm": 0.951152020027502,
      "learning_rate": 1.1095043217501655e-14,
      "loss": 0.8982,
      "step": 49640
    },
    {
      "epoch": 3.0,
      "step": 49641,
      "total_flos": 1475049923543040.0,
      "train_loss": 0.949762989045323,
      "train_runtime": 45551.2119,
      "train_samples_per_second": 139.491,
      "train_steps_per_second": 1.09
    }
  ],
  "logging_steps": 10,
  "max_steps": 49641,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1475049923543040.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}