{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 25.0,
  "eval_steps": 8310,
  "global_step": 207750,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00012033694344163658,
      "eval_loss": 3.815816879272461,
      "eval_runtime": 112.4917,
      "eval_samples_per_second": 65.667,
      "eval_steps_per_second": 8.214,
      "step": 1
    },
    {
      "epoch": 0.0012033694344163659,
      "grad_norm": 5.570976257324219,
      "learning_rate": 3.8513383400731756e-07,
      "loss": 3.7364,
      "step": 10
    },
    {
      "epoch": 0.0024067388688327317,
      "grad_norm": 4.148311614990234,
      "learning_rate": 7.702676680146351e-07,
      "loss": 3.8316,
      "step": 20
    },
    {
      "epoch": 0.0036101083032490976,
      "grad_norm": 5.7906084060668945,
      "learning_rate": 1.1554015020219527e-06,
      "loss": 3.8671,
      "step": 30
    },
    {
      "epoch": 0.0048134777376654635,
      "grad_norm": 3.5879058837890625,
      "learning_rate": 1.5405353360292703e-06,
      "loss": 3.7271,
      "step": 40
    },
    {
      "epoch": 0.006016847172081829,
      "grad_norm": 8.534287452697754,
      "learning_rate": 1.925669170036588e-06,
      "loss": 3.7304,
      "step": 50
    },
    {
      "epoch": 0.007220216606498195,
      "grad_norm": 4.045993804931641,
      "learning_rate": 2.3108030040439054e-06,
      "loss": 3.6963,
      "step": 60
    },
    {
      "epoch": 0.00842358604091456,
      "grad_norm": 3.4035534858703613,
      "learning_rate": 2.695936838051223e-06,
      "loss": 3.6405,
      "step": 70
    },
    {
      "epoch": 0.009626955475330927,
      "grad_norm": 5.412262439727783,
      "learning_rate": 3.0810706720585405e-06,
      "loss": 3.4811,
      "step": 80
    },
    {
      "epoch": 0.010830324909747292,
      "grad_norm": 3.6570475101470947,
      "learning_rate": 3.466204506065858e-06,
      "loss": 3.4819,
      "step": 90
    },
    {
      "epoch": 0.012033694344163659,
      "grad_norm": 6.75805139541626,
      "learning_rate": 3.851338340073176e-06,
      "loss": 3.5358,
      "step": 100
    },
    {
      "epoch": 0.013237063778580024,
      "grad_norm": 4.606562614440918,
      "learning_rate": 4.236472174080494e-06,
      "loss": 3.3392,
      "step": 110
    },
    {
      "epoch": 0.01444043321299639,
      "grad_norm": 3.2331383228302,
      "learning_rate": 4.621606008087811e-06,
      "loss": 3.4324,
      "step": 120
    },
    {
      "epoch": 0.015643802647412757,
      "grad_norm": 5.177463054656982,
      "learning_rate": 5.006739842095128e-06,
      "loss": 3.2908,
      "step": 130
    },
    {
      "epoch": 0.01684717208182912,
      "grad_norm": 3.7441446781158447,
      "learning_rate": 5.391873676102446e-06,
      "loss": 3.2742,
      "step": 140
    },
    {
      "epoch": 0.018050541516245487,
      "grad_norm": 8.747030258178711,
      "learning_rate": 5.777007510109764e-06,
      "loss": 3.3198,
      "step": 150
    },
    {
      "epoch": 0.019253910950661854,
      "grad_norm": 4.705621242523193,
      "learning_rate": 6.162141344117081e-06,
      "loss": 3.267,
      "step": 160
    },
    {
      "epoch": 0.02045728038507822,
      "grad_norm": 3.8838236331939697,
      "learning_rate": 6.547275178124398e-06,
      "loss": 3.1912,
      "step": 170
    },
    {
      "epoch": 0.021660649819494584,
      "grad_norm": 5.968881607055664,
      "learning_rate": 6.932409012131716e-06,
      "loss": 3.2778,
      "step": 180
    },
    {
      "epoch": 0.02286401925391095,
      "grad_norm": 4.607678413391113,
      "learning_rate": 7.317542846139033e-06,
      "loss": 3.2558,
      "step": 190
    },
    {
      "epoch": 0.024067388688327317,
      "grad_norm": 7.0164594650268555,
      "learning_rate": 7.702676680146351e-06,
      "loss": 3.273,
      "step": 200
    },
    {
      "epoch": 0.02527075812274368,
      "grad_norm": 5.379019737243652,
      "learning_rate": 8.08781051415367e-06,
      "loss": 3.0331,
      "step": 210
    },
    {
      "epoch": 0.026474127557160047,
      "grad_norm": 4.538290023803711,
      "learning_rate": 8.472944348160987e-06,
      "loss": 3.2632,
      "step": 220
    },
    {
      "epoch": 0.027677496991576414,
      "grad_norm": 6.160151481628418,
      "learning_rate": 8.858078182168304e-06,
      "loss": 3.1785,
      "step": 230
    },
    {
      "epoch": 0.02888086642599278,
      "grad_norm": 4.559308052062988,
      "learning_rate": 9.243212016175622e-06,
      "loss": 3.2087,
      "step": 240
    },
    {
      "epoch": 0.030084235860409144,
      "grad_norm": 7.7826104164123535,
      "learning_rate": 9.62834585018294e-06,
      "loss": 3.2275,
      "step": 250
    },
    {
      "epoch": 0.031287605294825514,
      "grad_norm": 4.852511882781982,
      "learning_rate": 1.0013479684190256e-05,
      "loss": 3.0636,
      "step": 260
    },
    {
      "epoch": 0.032490974729241874,
      "grad_norm": 4.055009365081787,
      "learning_rate": 1.0398613518197574e-05,
      "loss": 3.1303,
      "step": 270
    },
    {
      "epoch": 0.03369434416365824,
      "grad_norm": 6.477360725402832,
      "learning_rate": 1.0783747352204892e-05,
      "loss": 3.0855,
      "step": 280
    },
    {
      "epoch": 0.03489771359807461,
      "grad_norm": 4.41857385635376,
      "learning_rate": 1.1168881186212208e-05,
      "loss": 3.1672,
      "step": 290
    },
    {
      "epoch": 0.036101083032490974,
      "grad_norm": 8.869633674621582,
      "learning_rate": 1.1554015020219528e-05,
      "loss": 3.0891,
      "step": 300
    },
    {
      "epoch": 0.03730445246690734,
      "grad_norm": 6.276860237121582,
      "learning_rate": 1.1939148854226844e-05,
      "loss": 3.0637,
      "step": 310
    },
    {
      "epoch": 0.03850782190132371,
      "grad_norm": 3.9365427494049072,
      "learning_rate": 1.2324282688234162e-05,
      "loss": 3.0161,
      "step": 320
    },
    {
      "epoch": 0.039711191335740074,
      "grad_norm": 5.813107490539551,
      "learning_rate": 1.270941652224148e-05,
      "loss": 3.002,
      "step": 330
    },
    {
      "epoch": 0.04091456077015644,
      "grad_norm": 4.736392498016357,
      "learning_rate": 1.3094550356248796e-05,
      "loss": 3.1451,
      "step": 340
    },
    {
      "epoch": 0.0421179302045728,
      "grad_norm": 7.715930938720703,
      "learning_rate": 1.3479684190256114e-05,
      "loss": 3.0218,
      "step": 350
    },
    {
      "epoch": 0.04332129963898917,
      "grad_norm": 5.3318328857421875,
      "learning_rate": 1.3864818024263432e-05,
      "loss": 2.9514,
      "step": 360
    },
    {
      "epoch": 0.044524669073405534,
      "grad_norm": 4.3995466232299805,
      "learning_rate": 1.4249951858270749e-05,
      "loss": 3.0154,
      "step": 370
    },
    {
      "epoch": 0.0457280385078219,
      "grad_norm": 5.849190711975098,
      "learning_rate": 1.4635085692278067e-05,
      "loss": 2.9262,
      "step": 380
    },
    {
      "epoch": 0.04693140794223827,
      "grad_norm": 5.229336261749268,
      "learning_rate": 1.5020219526285385e-05,
      "loss": 2.9599,
      "step": 390
    },
    {
      "epoch": 0.048134777376654635,
      "grad_norm": 7.649044513702393,
      "learning_rate": 1.5405353360292703e-05,
      "loss": 3.0405,
      "step": 400
    },
    {
      "epoch": 0.049338146811071,
      "grad_norm": 6.180991172790527,
      "learning_rate": 1.579048719430002e-05,
      "loss": 3.0349,
      "step": 410
    },
    {
      "epoch": 0.05054151624548736,
      "grad_norm": 3.732105016708374,
      "learning_rate": 1.617562102830734e-05,
      "loss": 3.0039,
      "step": 420
    },
    {
      "epoch": 0.05174488567990373,
      "grad_norm": 6.500801086425781,
      "learning_rate": 1.6560754862314655e-05,
      "loss": 3.1118,
      "step": 430
    },
    {
      "epoch": 0.052948255114320095,
      "grad_norm": 4.652256011962891,
      "learning_rate": 1.6945888696321974e-05,
      "loss": 2.8496,
      "step": 440
    },
    {
      "epoch": 0.05415162454873646,
      "grad_norm": 7.485434055328369,
      "learning_rate": 1.733102253032929e-05,
      "loss": 3.0901,
      "step": 450
    },
    {
      "epoch": 0.05535499398315283,
      "grad_norm": 5.527116298675537,
      "learning_rate": 1.7716156364336607e-05,
      "loss": 3.0363,
      "step": 460
    },
    {
      "epoch": 0.056558363417569195,
      "grad_norm": 3.4333620071411133,
      "learning_rate": 1.8101290198343927e-05,
      "loss": 2.9691,
      "step": 470
    },
    {
      "epoch": 0.05776173285198556,
      "grad_norm": 6.362924575805664,
      "learning_rate": 1.8486424032351243e-05,
      "loss": 2.9124,
      "step": 480
    },
    {
      "epoch": 0.05896510228640193,
      "grad_norm": 4.653800010681152,
      "learning_rate": 1.887155786635856e-05,
      "loss": 2.9238,
      "step": 490
    },
    {
      "epoch": 0.06016847172081829,
      "grad_norm": 7.520432949066162,
      "learning_rate": 1.925669170036588e-05,
      "loss": 3.0473,
      "step": 500
    },
    {
      "epoch": 0.061371841155234655,
      "grad_norm": 5.268134593963623,
      "learning_rate": 1.9641825534373195e-05,
      "loss": 2.9194,
      "step": 510
    },
    {
      "epoch": 0.06257521058965103,
      "grad_norm": 3.7793736457824707,
      "learning_rate": 2.002695936838051e-05,
      "loss": 2.8513,
      "step": 520
    },
    {
      "epoch": 0.06377858002406739,
      "grad_norm": 4.988191604614258,
      "learning_rate": 2.041209320238783e-05,
      "loss": 2.9907,
      "step": 530
    },
    {
      "epoch": 0.06498194945848375,
      "grad_norm": 3.8745498657226562,
      "learning_rate": 2.0797227036395148e-05,
      "loss": 2.9702,
      "step": 540
    },
    {
      "epoch": 0.06618531889290012,
      "grad_norm": 6.353143215179443,
      "learning_rate": 2.1182360870402464e-05,
      "loss": 2.8286,
      "step": 550
    },
    {
      "epoch": 0.06738868832731648,
      "grad_norm": 5.642485618591309,
      "learning_rate": 2.1567494704409784e-05,
      "loss": 2.8682,
      "step": 560
    },
    {
      "epoch": 0.06859205776173286,
      "grad_norm": 3.590162992477417,
      "learning_rate": 2.19526285384171e-05,
      "loss": 2.8957,
      "step": 570
    },
    {
      "epoch": 0.06979542719614922,
      "grad_norm": 5.774114608764648,
      "learning_rate": 2.2337762372424416e-05,
      "loss": 2.8963,
      "step": 580
    },
    {
      "epoch": 0.07099879663056559,
      "grad_norm": 4.017606258392334,
      "learning_rate": 2.2722896206431736e-05,
      "loss": 2.8724,
      "step": 590
    },
    {
      "epoch": 0.07220216606498195,
      "grad_norm": 7.137066841125488,
      "learning_rate": 2.3108030040439056e-05,
      "loss": 2.9016,
      "step": 600
    },
    {
      "epoch": 0.07340553549939831,
      "grad_norm": 4.75874137878418,
      "learning_rate": 2.3493163874446372e-05,
      "loss": 2.8465,
      "step": 610
    },
    {
      "epoch": 0.07460890493381468,
      "grad_norm": 3.671997547149658,
      "learning_rate": 2.3878297708453688e-05,
      "loss": 2.9338,
      "step": 620
    },
    {
      "epoch": 0.07581227436823104,
      "grad_norm": 5.378019332885742,
      "learning_rate": 2.4263431542461008e-05,
      "loss": 2.8506,
      "step": 630
    },
    {
      "epoch": 0.07701564380264742,
      "grad_norm": 4.3311662673950195,
      "learning_rate": 2.4648565376468324e-05,
      "loss": 2.9701,
      "step": 640
    },
    {
      "epoch": 0.07821901323706378,
      "grad_norm": 5.478089809417725,
      "learning_rate": 2.503369921047564e-05,
      "loss": 3.0287,
      "step": 650
    },
    {
      "epoch": 0.07942238267148015,
      "grad_norm": 4.632708549499512,
      "learning_rate": 2.541883304448296e-05,
      "loss": 2.8668,
      "step": 660
    },
    {
      "epoch": 0.08062575210589651,
      "grad_norm": 3.374662399291992,
      "learning_rate": 2.5803966878490276e-05,
      "loss": 2.8472,
      "step": 670
    },
    {
      "epoch": 0.08182912154031288,
      "grad_norm": 5.122900485992432,
      "learning_rate": 2.6189100712497593e-05,
      "loss": 2.8916,
      "step": 680
    },
    {
      "epoch": 0.08303249097472924,
      "grad_norm": 4.146032810211182,
      "learning_rate": 2.6574234546504912e-05,
      "loss": 2.9094,
      "step": 690
    },
    {
      "epoch": 0.0842358604091456,
      "grad_norm": 6.2430925369262695,
      "learning_rate": 2.695936838051223e-05,
      "loss": 2.8602,
      "step": 700
    },
    {
      "epoch": 0.08543922984356198,
      "grad_norm": 5.070657730102539,
      "learning_rate": 2.7344502214519545e-05,
      "loss": 2.8657,
      "step": 710
    },
    {
      "epoch": 0.08664259927797834,
      "grad_norm": 3.6268534660339355,
      "learning_rate": 2.7729636048526865e-05,
      "loss": 2.9479,
      "step": 720
    },
    {
      "epoch": 0.08784596871239471,
      "grad_norm": 5.12242317199707,
      "learning_rate": 2.811476988253418e-05,
      "loss": 2.9425,
      "step": 730
    },
    {
      "epoch": 0.08904933814681107,
      "grad_norm": 3.7903456687927246,
      "learning_rate": 2.8499903716541497e-05,
      "loss": 2.8374,
      "step": 740
    },
    {
      "epoch": 0.09025270758122744,
      "grad_norm": 6.039148330688477,
      "learning_rate": 2.8885037550548817e-05,
      "loss": 2.7863,
      "step": 750
    },
    {
      "epoch": 0.0914560770156438,
      "grad_norm": 4.456302165985107,
      "learning_rate": 2.9270171384556133e-05,
      "loss": 2.6648,
      "step": 760
    },
    {
      "epoch": 0.09265944645006016,
      "grad_norm": 3.65207576751709,
      "learning_rate": 2.965530521856345e-05,
      "loss": 2.8954,
      "step": 770
    },
    {
      "epoch": 0.09386281588447654,
      "grad_norm": 4.908219337463379,
      "learning_rate": 3.004043905257077e-05,
      "loss": 2.8401,
      "step": 780
    },
    {
      "epoch": 0.0950661853188929,
      "grad_norm": 4.265645503997803,
      "learning_rate": 3.0425572886578085e-05,
      "loss": 2.9104,
      "step": 790
    },
    {
      "epoch": 0.09626955475330927,
      "grad_norm": 6.547182559967041,
      "learning_rate": 3.0810706720585405e-05,
      "loss": 2.8935,
      "step": 800
    },
    {
      "epoch": 0.09747292418772563,
      "grad_norm": 4.835933208465576,
      "learning_rate": 3.119584055459272e-05,
      "loss": 2.6432,
      "step": 810
    },
    {
      "epoch": 0.098676293622142,
      "grad_norm": 3.267350435256958,
      "learning_rate": 3.158097438860004e-05,
      "loss": 2.8195,
      "step": 820
    },
    {
      "epoch": 0.09987966305655836,
      "grad_norm": 4.60970401763916,
      "learning_rate": 3.1966108222607354e-05,
      "loss": 2.6222,
      "step": 830
    },
    {
      "epoch": 0.10108303249097472,
      "grad_norm": 4.191972255706787,
      "learning_rate": 3.235124205661468e-05,
      "loss": 2.7868,
      "step": 840
    },
    {
      "epoch": 0.1022864019253911,
      "grad_norm": 6.102560997009277,
      "learning_rate": 3.273637589062199e-05,
      "loss": 2.7755,
      "step": 850
    },
    {
      "epoch": 0.10348977135980746,
      "grad_norm": 4.372929573059082,
      "learning_rate": 3.312150972462931e-05,
      "loss": 2.823,
      "step": 860
    },
    {
      "epoch": 0.10469314079422383,
      "grad_norm": 3.2868309020996094,
      "learning_rate": 3.350664355863663e-05,
      "loss": 2.6892,
      "step": 870
    },
    {
      "epoch": 0.10589651022864019,
      "grad_norm": 4.598580837249756,
      "learning_rate": 3.389177739264395e-05,
      "loss": 2.8837,
      "step": 880
    },
    {
      "epoch": 0.10709987966305656,
      "grad_norm": 3.940325975418091,
      "learning_rate": 3.4276911226651265e-05,
      "loss": 2.8312,
      "step": 890
    },
    {
      "epoch": 0.10830324909747292,
      "grad_norm": 6.136560916900635,
      "learning_rate": 3.466204506065858e-05,
      "loss": 2.747,
      "step": 900
    },
    {
      "epoch": 0.1095066185318893,
      "grad_norm": 4.117788314819336,
      "learning_rate": 3.50471788946659e-05,
      "loss": 2.6159,
      "step": 910
    },
    {
      "epoch": 0.11070998796630566,
      "grad_norm": 4.032705307006836,
      "learning_rate": 3.5432312728673214e-05,
      "loss": 2.7615,
      "step": 920
    },
    {
      "epoch": 0.11191335740072202,
      "grad_norm": 5.094011306762695,
      "learning_rate": 3.581744656268054e-05,
      "loss": 2.7753,
      "step": 930
    },
    {
      "epoch": 0.11311672683513839,
      "grad_norm": 3.8412024974823,
      "learning_rate": 3.6202580396687854e-05,
      "loss": 2.7813,
      "step": 940
    },
    {
      "epoch": 0.11432009626955475,
      "grad_norm": 4.643109321594238,
      "learning_rate": 3.658771423069517e-05,
      "loss": 2.9567,
      "step": 950
    },
    {
      "epoch": 0.11552346570397112,
      "grad_norm": 4.502158164978027,
      "learning_rate": 3.6972848064702486e-05,
      "loss": 2.6247,
      "step": 960
    },
    {
      "epoch": 0.11672683513838748,
      "grad_norm": 2.5785179138183594,
      "learning_rate": 3.73579818987098e-05,
      "loss": 2.7178,
      "step": 970
    },
    {
      "epoch": 0.11793020457280386,
      "grad_norm": 4.689826011657715,
      "learning_rate": 3.774311573271712e-05,
      "loss": 2.8354,
      "step": 980
    },
    {
      "epoch": 0.11913357400722022,
      "grad_norm": 4.020382404327393,
      "learning_rate": 3.812824956672444e-05,
      "loss": 2.6598,
      "step": 990
    },
    {
      "epoch": 0.12033694344163658,
      "grad_norm": 5.820950031280518,
      "learning_rate": 3.851338340073176e-05,
      "loss": 2.8073,
      "step": 1000
    },
    {
      "epoch": 0.12154031287605295,
      "grad_norm": 3.5636305809020996,
      "learning_rate": 3.8898517234739074e-05,
      "loss": 2.5984,
      "step": 1010
    },
    {
      "epoch": 0.12274368231046931,
      "grad_norm": 3.0366146564483643,
      "learning_rate": 3.928365106874639e-05,
      "loss": 2.7126,
      "step": 1020
    },
    {
      "epoch": 0.12394705174488568,
      "grad_norm": 5.1952385902404785,
      "learning_rate": 3.966878490275371e-05,
      "loss": 2.7211,
      "step": 1030
    },
    {
      "epoch": 0.12515042117930206,
      "grad_norm": 4.054571151733398,
      "learning_rate": 4.005391873676102e-05,
      "loss": 2.6443,
      "step": 1040
    },
    {
      "epoch": 0.1263537906137184,
      "grad_norm": 6.402095794677734,
      "learning_rate": 4.0439052570768346e-05,
      "loss": 2.7943,
      "step": 1050
    },
    {
      "epoch": 0.12755716004813478,
      "grad_norm": 4.076451301574707,
      "learning_rate": 4.082418640477566e-05,
      "loss": 2.4575,
      "step": 1060
    },
    {
      "epoch": 0.12876052948255115,
      "grad_norm": 3.4446961879730225,
      "learning_rate": 4.120932023878298e-05,
      "loss": 2.6509,
      "step": 1070
    },
    {
      "epoch": 0.1299638989169675,
      "grad_norm": 5.1258625984191895,
      "learning_rate": 4.1594454072790295e-05,
      "loss": 2.6947,
      "step": 1080
    },
    {
      "epoch": 0.13116726835138387,
      "grad_norm": 4.072508335113525,
      "learning_rate": 4.197958790679761e-05,
      "loss": 2.7828,
      "step": 1090
    },
    {
      "epoch": 0.13237063778580024,
      "grad_norm": 6.007488250732422,
      "learning_rate": 4.236472174080493e-05,
      "loss": 2.8042,
      "step": 1100
    },
    {
      "epoch": 0.13357400722021662,
      "grad_norm": 4.070166110992432,
      "learning_rate": 4.274985557481225e-05,
      "loss": 2.6465,
      "step": 1110
    },
    {
      "epoch": 0.13477737665463296,
      "grad_norm": 3.5824296474456787,
      "learning_rate": 4.313498940881957e-05,
      "loss": 2.6448,
      "step": 1120
    },
    {
      "epoch": 0.13598074608904934,
      "grad_norm": 4.705931186676025,
      "learning_rate": 4.3520123242826883e-05,
      "loss": 2.6401,
      "step": 1130
    },
    {
      "epoch": 0.1371841155234657,
      "grad_norm": 3.6984827518463135,
      "learning_rate": 4.39052570768342e-05,
      "loss": 2.5397,
      "step": 1140
    },
    {
      "epoch": 0.13838748495788206,
      "grad_norm": 5.334888458251953,
      "learning_rate": 4.4290390910841516e-05,
      "loss": 2.5313,
      "step": 1150
    },
    {
      "epoch": 0.13959085439229843,
      "grad_norm": 4.262475490570068,
      "learning_rate": 4.467552474484883e-05,
      "loss": 2.6548,
      "step": 1160
    },
    {
      "epoch": 0.1407942238267148,
      "grad_norm": 3.3582048416137695,
      "learning_rate": 4.506065857885615e-05,
      "loss": 2.5978,
      "step": 1170
    },
    {
      "epoch": 0.14199759326113118,
      "grad_norm": 4.077620983123779,
      "learning_rate": 4.544579241286347e-05,
      "loss": 2.716,
      "step": 1180
    },
    {
      "epoch": 0.14320096269554752,
      "grad_norm": 3.881532669067383,
      "learning_rate": 4.5830926246870795e-05,
      "loss": 2.7427,
      "step": 1190
    },
    {
      "epoch": 0.1444043321299639,
      "grad_norm": 4.926413059234619,
      "learning_rate": 4.621606008087811e-05,
      "loss": 2.5799,
      "step": 1200
    },
    {
      "epoch": 0.14560770156438027,
      "grad_norm": 3.9430980682373047,
      "learning_rate": 4.660119391488543e-05,
      "loss": 2.4586,
      "step": 1210
    },
    {
      "epoch": 0.14681107099879662,
      "grad_norm": 2.823546886444092,
      "learning_rate": 4.6986327748892744e-05,
      "loss": 2.6815,
      "step": 1220
    },
    {
      "epoch": 0.148014440433213,
      "grad_norm": 4.250391483306885,
      "learning_rate": 4.737146158290006e-05,
      "loss": 2.5357,
      "step": 1230
    },
    {
      "epoch": 0.14921780986762936,
      "grad_norm": 5.145439624786377,
      "learning_rate": 4.7756595416907376e-05,
      "loss": 2.4226,
      "step": 1240
    },
    {
      "epoch": 0.15042117930204574,
      "grad_norm": 6.212322235107422,
      "learning_rate": 4.81417292509147e-05,
      "loss": 2.6047,
      "step": 1250
    },
    {
      "epoch": 0.15162454873646208,
      "grad_norm": 4.376655578613281,
      "learning_rate": 4.8526863084922016e-05,
      "loss": 2.3968,
      "step": 1260
    },
    {
      "epoch": 0.15282791817087846,
      "grad_norm": 3.185271978378296,
      "learning_rate": 4.891199691892933e-05,
      "loss": 2.6196,
      "step": 1270
    },
    {
      "epoch": 0.15403128760529483,
      "grad_norm": 4.398382663726807,
      "learning_rate": 4.929713075293665e-05,
      "loss": 2.5637,
      "step": 1280
    },
    {
      "epoch": 0.1552346570397112,
      "grad_norm": 3.7962944507598877,
      "learning_rate": 4.9682264586943964e-05,
      "loss": 2.4871,
      "step": 1290
    },
    {
      "epoch": 0.15643802647412755,
      "grad_norm": 5.663314342498779,
      "learning_rate": 5.006739842095128e-05,
      "loss": 2.6053,
      "step": 1300
    },
    {
      "epoch": 0.15764139590854392,
      "grad_norm": 4.08873987197876,
      "learning_rate": 5.04525322549586e-05,
      "loss": 2.3832,
      "step": 1310
    },
    {
      "epoch": 0.1588447653429603,
      "grad_norm": 3.060068130493164,
      "learning_rate": 5.083766608896592e-05,
      "loss": 2.5909,
      "step": 1320
    },
    {
      "epoch": 0.16004813477737664,
      "grad_norm": 3.5497021675109863,
      "learning_rate": 5.122279992297323e-05,
      "loss": 2.4943,
      "step": 1330
    },
    {
      "epoch": 0.16125150421179302,
      "grad_norm": 3.9688732624053955,
      "learning_rate": 5.160793375698055e-05,
      "loss": 2.4693,
      "step": 1340
    },
    {
      "epoch": 0.1624548736462094,
      "grad_norm": 5.243858337402344,
      "learning_rate": 5.1993067590987876e-05,
      "loss": 2.5234,
      "step": 1350
    },
    {
      "epoch": 0.16365824308062576,
      "grad_norm": 4.390367031097412,
      "learning_rate": 5.2378201424995185e-05,
      "loss": 2.5178,
      "step": 1360
    },
    {
      "epoch": 0.1648616125150421,
      "grad_norm": 2.778979778289795,
      "learning_rate": 5.276333525900251e-05,
      "loss": 2.7806,
      "step": 1370
    },
    {
      "epoch": 0.16606498194945848,
      "grad_norm": 5.424134254455566,
      "learning_rate": 5.3148469093009825e-05,
      "loss": 2.6223,
      "step": 1380
    },
    {
      "epoch": 0.16726835138387486,
      "grad_norm": 3.2314844131469727,
      "learning_rate": 5.353360292701715e-05,
      "loss": 2.4702,
      "step": 1390
    },
    {
      "epoch": 0.1684717208182912,
      "grad_norm": 5.493102550506592,
      "learning_rate": 5.391873676102446e-05,
      "loss": 2.6778,
      "step": 1400
    },
    {
      "epoch": 0.16967509025270758,
      "grad_norm": 4.03582239151001,
      "learning_rate": 5.430387059503178e-05,
      "loss": 2.4206,
      "step": 1410
    },
    {
      "epoch": 0.17087845968712395,
      "grad_norm": 3.435091018676758,
      "learning_rate": 5.468900442903909e-05,
      "loss": 2.7106,
      "step": 1420
    },
    {
      "epoch": 0.17208182912154033,
      "grad_norm": 4.965870380401611,
      "learning_rate": 5.507413826304641e-05,
      "loss": 2.4973,
      "step": 1430
    },
    {
      "epoch": 0.17328519855595667,
      "grad_norm": 3.4605584144592285,
      "learning_rate": 5.545927209705373e-05,
      "loss": 2.5029,
      "step": 1440
    },
    {
      "epoch": 0.17448856799037304,
      "grad_norm": 5.244032382965088,
      "learning_rate": 5.584440593106105e-05,
      "loss": 2.5749,
      "step": 1450
    },
    {
      "epoch": 0.17569193742478942,
      "grad_norm": 4.023014068603516,
      "learning_rate": 5.622953976506836e-05,
      "loss": 2.5202,
      "step": 1460
    },
    {
      "epoch": 0.17689530685920576,
      "grad_norm": 2.943232297897339,
      "learning_rate": 5.6614673599075685e-05,
      "loss": 2.6431,
      "step": 1470
    },
    {
      "epoch": 0.17809867629362214,
      "grad_norm": 5.903708457946777,
      "learning_rate": 5.6999807433082994e-05,
      "loss": 2.5852,
      "step": 1480
    },
    {
      "epoch": 0.1793020457280385,
      "grad_norm": 4.0138444900512695,
      "learning_rate": 5.738494126709032e-05,
      "loss": 2.4998,
      "step": 1490
    },
    {
      "epoch": 0.18050541516245489,
      "grad_norm": 4.759198188781738,
      "learning_rate": 5.7770075101097634e-05,
      "loss": 2.4394,
      "step": 1500
    },
    {
      "epoch": 0.18170878459687123,
      "grad_norm": 4.2235188484191895,
      "learning_rate": 5.815520893510496e-05,
      "loss": 2.6585,
      "step": 1510
    },
    {
      "epoch": 0.1829121540312876,
      "grad_norm": 3.335575580596924,
      "learning_rate": 5.8540342769112266e-05,
      "loss": 2.4832,
      "step": 1520
    },
    {
      "epoch": 0.18411552346570398,
      "grad_norm": 4.41403341293335,
      "learning_rate": 5.892547660311959e-05,
      "loss": 2.3476,
      "step": 1530
    },
    {
      "epoch": 0.18531889290012032,
      "grad_norm": 3.597169876098633,
      "learning_rate": 5.93106104371269e-05,
      "loss": 2.4265,
      "step": 1540
    },
    {
      "epoch": 0.1865222623345367,
      "grad_norm": 5.240777015686035,
      "learning_rate": 5.969574427113422e-05,
      "loss": 2.4832,
      "step": 1550
    },
    {
      "epoch": 0.18772563176895307,
      "grad_norm": 4.511013984680176,
      "learning_rate": 6.008087810514154e-05,
      "loss": 2.456,
      "step": 1560
    },
    {
      "epoch": 0.18892900120336945,
      "grad_norm": 3.002626895904541,
      "learning_rate": 6.046601193914886e-05,
      "loss": 2.3578,
      "step": 1570
    },
    {
      "epoch": 0.1901323706377858,
      "grad_norm": 4.632908821105957,
      "learning_rate": 6.085114577315617e-05,
      "loss": 2.4645,
      "step": 1580
    },
    {
      "epoch": 0.19133574007220217,
      "grad_norm": 3.4771511554718018,
      "learning_rate": 6.12362796071635e-05,
      "loss": 2.4285,
      "step": 1590
    },
    {
      "epoch": 0.19253910950661854,
      "grad_norm": 6.660020351409912,
      "learning_rate": 6.162141344117081e-05,
      "loss": 2.6111,
      "step": 1600
    },
    {
      "epoch": 0.19374247894103488,
      "grad_norm": 3.7338061332702637,
      "learning_rate": 6.200654727517813e-05,
      "loss": 2.4041,
      "step": 1610
    },
    {
      "epoch": 0.19494584837545126,
      "grad_norm": 3.1107709407806396,
      "learning_rate": 6.239168110918544e-05,
      "loss": 2.5913,
      "step": 1620
    },
    {
      "epoch": 0.19614921780986763,
      "grad_norm": 4.126433372497559,
      "learning_rate": 6.277681494319276e-05,
      "loss": 2.5915,
      "step": 1630
    },
    {
      "epoch": 0.197352587244284,
      "grad_norm": 4.266234874725342,
      "learning_rate": 6.316194877720008e-05,
      "loss": 2.3163,
      "step": 1640
    },
    {
      "epoch": 0.19855595667870035,
      "grad_norm": 5.36894416809082,
      "learning_rate": 6.354708261120739e-05,
      "loss": 2.5416,
      "step": 1650
    },
    {
      "epoch": 0.19975932611311673,
      "grad_norm": 3.78735613822937,
      "learning_rate": 6.393221644521471e-05,
      "loss": 2.463,
      "step": 1660
    },
    {
      "epoch": 0.2009626955475331,
      "grad_norm": 3.2801718711853027,
      "learning_rate": 6.431735027922204e-05,
      "loss": 2.5265,
      "step": 1670
    },
    {
      "epoch": 0.20216606498194944,
      "grad_norm": 4.672789096832275,
      "learning_rate": 6.470248411322935e-05,
      "loss": 2.3468,
      "step": 1680
    },
    {
      "epoch": 0.20336943441636582,
      "grad_norm": 4.022416591644287,
      "learning_rate": 6.508761794723667e-05,
      "loss": 2.4621,
      "step": 1690
    },
    {
      "epoch": 0.2045728038507822,
      "grad_norm": 5.079505443572998,
      "learning_rate": 6.547275178124399e-05,
      "loss": 2.6353,
      "step": 1700
    },
    {
      "epoch": 0.20577617328519857,
      "grad_norm": 4.310054779052734,
      "learning_rate": 6.58578856152513e-05,
      "loss": 2.3306,
      "step": 1710
    },
    {
      "epoch": 0.2069795427196149,
      "grad_norm": 2.7068967819213867,
      "learning_rate": 6.624301944925862e-05,
      "loss": 2.5586,
      "step": 1720
    },
    {
      "epoch": 0.20818291215403129,
      "grad_norm": 4.833555221557617,
      "learning_rate": 6.662815328326594e-05,
      "loss": 2.5028,
      "step": 1730
    },
    {
      "epoch": 0.20938628158844766,
      "grad_norm": 3.597747802734375,
      "learning_rate": 6.701328711727327e-05,
      "loss": 2.4376,
      "step": 1740
    },
    {
      "epoch": 0.21058965102286403,
      "grad_norm": 5.5854010581970215,
      "learning_rate": 6.739842095128057e-05,
      "loss": 2.5534,
      "step": 1750
    },
    {
      "epoch": 0.21179302045728038,
      "grad_norm": 3.6864216327667236,
      "learning_rate": 6.77835547852879e-05,
      "loss": 2.4125,
      "step": 1760
    },
    {
      "epoch": 0.21299638989169675,
      "grad_norm": 2.955962657928467,
      "learning_rate": 6.81686886192952e-05,
      "loss": 2.5775,
      "step": 1770
    },
    {
      "epoch": 0.21419975932611313,
      "grad_norm": 5.092808723449707,
      "learning_rate": 6.855382245330253e-05,
      "loss": 2.3907,
      "step": 1780
    },
    {
      "epoch": 0.21540312876052947,
      "grad_norm": 3.32344913482666,
      "learning_rate": 6.893895628730985e-05,
      "loss": 2.5718,
      "step": 1790
    },
    {
      "epoch": 0.21660649819494585,
      "grad_norm": 4.911040782928467,
      "learning_rate": 6.932409012131716e-05,
      "loss": 2.5209,
      "step": 1800
    },
    {
      "epoch": 0.21780986762936222,
      "grad_norm": 3.9270997047424316,
      "learning_rate": 6.970922395532448e-05,
      "loss": 2.4563,
      "step": 1810
    },
    {
      "epoch": 0.2190132370637786,
      "grad_norm": 2.8259575366973877,
      "learning_rate": 7.00943577893318e-05,
      "loss": 2.5424,
      "step": 1820
    },
    {
      "epoch": 0.22021660649819494,
      "grad_norm": 4.02173376083374,
      "learning_rate": 7.047949162333911e-05,
      "loss": 2.4887,
      "step": 1830
    },
    {
      "epoch": 0.2214199759326113,
      "grad_norm": 4.133205890655518,
      "learning_rate": 7.086462545734643e-05,
      "loss": 2.5443,
      "step": 1840
    },
    {
      "epoch": 0.2226233453670277,
      "grad_norm": 5.094078540802002,
      "learning_rate": 7.124975929135374e-05,
      "loss": 2.5202,
      "step": 1850
    },
    {
      "epoch": 0.22382671480144403,
      "grad_norm": 4.140336513519287,
      "learning_rate": 7.163489312536107e-05,
      "loss": 2.4474,
      "step": 1860
    },
    {
      "epoch": 0.2250300842358604,
      "grad_norm": 2.8555362224578857,
      "learning_rate": 7.202002695936838e-05,
      "loss": 2.3452,
      "step": 1870
    },
    {
      "epoch": 0.22623345367027678,
      "grad_norm": 4.063607692718506,
      "learning_rate": 7.240516079337571e-05,
      "loss": 2.3434,
      "step": 1880
    },
    {
      "epoch": 0.22743682310469315,
      "grad_norm": 3.575995683670044,
      "learning_rate": 7.279029462738301e-05,
      "loss": 2.2894,
      "step": 1890
    },
    {
      "epoch": 0.2286401925391095,
      "grad_norm": 4.515583515167236,
      "learning_rate": 7.317542846139034e-05,
      "loss": 2.4199,
      "step": 1900
    },
    {
      "epoch": 0.22984356197352587,
      "grad_norm": 4.865725994110107,
      "learning_rate": 7.356056229539766e-05,
      "loss": 2.2267,
      "step": 1910
    },
    {
      "epoch": 0.23104693140794225,
      "grad_norm": 2.9381701946258545,
      "learning_rate": 7.394569612940497e-05,
      "loss": 2.3357,
      "step": 1920
    },
    {
      "epoch": 0.2322503008423586,
      "grad_norm": 5.303056240081787,
      "learning_rate": 7.433082996341229e-05,
      "loss": 2.4184,
      "step": 1930
    },
    {
      "epoch": 0.23345367027677497,
      "grad_norm": 3.5984139442443848,
      "learning_rate": 7.47159637974196e-05,
      "loss": 2.3659,
      "step": 1940
    },
    {
      "epoch": 0.23465703971119134,
      "grad_norm": 5.596217632293701,
      "learning_rate": 7.510109763142692e-05,
      "loss": 2.4371,
      "step": 1950
    },
    {
      "epoch": 0.2358604091456077,
      "grad_norm": 4.290953159332275,
      "learning_rate": 7.548623146543424e-05,
      "loss": 2.2033,
      "step": 1960
    },
    {
      "epoch": 0.23706377858002406,
      "grad_norm": 3.4179747104644775,
      "learning_rate": 7.587136529944155e-05,
      "loss": 2.385,
      "step": 1970
    },
    {
      "epoch": 0.23826714801444043,
      "grad_norm": 3.991859197616577,
      "learning_rate": 7.625649913344888e-05,
      "loss": 2.3238,
      "step": 1980
    },
    {
      "epoch": 0.2394705174488568,
      "grad_norm": 3.4576947689056396,
      "learning_rate": 7.664163296745619e-05,
      "loss": 2.3791,
      "step": 1990
    },
    {
      "epoch": 0.24067388688327315,
      "grad_norm": 4.9071149826049805,
      "learning_rate": 7.702676680146352e-05,
      "loss": 2.3563,
      "step": 2000
    },
    {
      "epoch": 0.24187725631768953,
      "grad_norm": 3.402372360229492,
      "learning_rate": 7.741190063547082e-05,
      "loss": 2.3858,
      "step": 2010
    },
    {
      "epoch": 0.2430806257521059,
      "grad_norm": 2.9735138416290283,
      "learning_rate": 7.779703446947815e-05,
      "loss": 2.4769,
      "step": 2020
    },
    {
      "epoch": 0.24428399518652227,
      "grad_norm": 4.522787570953369,
      "learning_rate": 7.818216830348546e-05,
      "loss": 2.259,
      "step": 2030
    },
    {
      "epoch": 0.24548736462093862,
      "grad_norm": 3.260927200317383,
      "learning_rate": 7.856730213749278e-05,
      "loss": 2.4279,
      "step": 2040
    },
    {
      "epoch": 0.246690734055355,
      "grad_norm": 4.7217326164245605,
      "learning_rate": 7.895243597150011e-05,
      "loss": 2.3933,
      "step": 2050
    },
    {
      "epoch": 0.24789410348977137,
      "grad_norm": 4.655685901641846,
      "learning_rate": 7.933756980550741e-05,
      "loss": 2.3567,
      "step": 2060
    },
    {
      "epoch": 0.2490974729241877,
      "grad_norm": 3.0860252380371094,
      "learning_rate": 7.972270363951474e-05,
      "loss": 2.488,
      "step": 2070
    },
    {
      "epoch": 0.2503008423586041,
      "grad_norm": 5.234461784362793,
      "learning_rate": 8.010783747352205e-05,
      "loss": 2.2966,
      "step": 2080
    },
    {
      "epoch": 0.25150421179302046,
      "grad_norm": 3.013174057006836,
      "learning_rate": 8.049297130752938e-05,
      "loss": 2.4847,
      "step": 2090
    },
    {
      "epoch": 0.2527075812274368,
      "grad_norm": 5.257478713989258,
      "learning_rate": 8.087810514153669e-05,
      "loss": 2.3881,
      "step": 2100
    },
    {
      "epoch": 0.2539109506618532,
      "grad_norm": 4.6101789474487305,
      "learning_rate": 8.126323897554401e-05,
      "loss": 2.2844,
      "step": 2110
    },
    {
      "epoch": 0.25511432009626955,
      "grad_norm": 3.3464407920837402,
      "learning_rate": 8.164837280955133e-05,
      "loss": 2.4395,
      "step": 2120
    },
    {
      "epoch": 0.2563176895306859,
      "grad_norm": 4.377286434173584,
      "learning_rate": 8.203350664355864e-05,
      "loss": 2.0636,
      "step": 2130
    },
    {
      "epoch": 0.2575210589651023,
      "grad_norm": 3.287522554397583,
      "learning_rate": 8.241864047756596e-05,
      "loss": 2.3959,
      "step": 2140
    },
    {
      "epoch": 0.25872442839951865,
      "grad_norm": 4.91052770614624,
      "learning_rate": 8.280377431157327e-05,
      "loss": 2.5249,
      "step": 2150
    },
    {
      "epoch": 0.259927797833935,
      "grad_norm": 3.9140472412109375,
      "learning_rate": 8.318890814558059e-05,
      "loss": 2.3234,
      "step": 2160
    },
    {
      "epoch": 0.2611311672683514,
      "grad_norm": 2.5841314792633057,
      "learning_rate": 8.357404197958792e-05,
      "loss": 2.3596,
      "step": 2170
    },
    {
      "epoch": 0.26233453670276774,
      "grad_norm": 5.444457530975342,
      "learning_rate": 8.395917581359522e-05,
      "loss": 2.4814,
      "step": 2180
    },
    {
      "epoch": 0.26353790613718414,
      "grad_norm": 3.8762309551239014,
      "learning_rate": 8.434430964760255e-05,
      "loss": 2.2627,
      "step": 2190
    },
    {
      "epoch": 0.2647412755716005,
      "grad_norm": 5.452149868011475,
      "learning_rate": 8.472944348160986e-05,
      "loss": 2.5046,
      "step": 2200
    },
    {
      "epoch": 0.26594464500601683,
      "grad_norm": 5.255434513092041,
      "learning_rate": 8.511457731561719e-05,
      "loss": 2.1281,
      "step": 2210
    },
    {
      "epoch": 0.26714801444043323,
      "grad_norm": 2.345582962036133,
      "learning_rate": 8.54997111496245e-05,
      "loss": 2.3343,
      "step": 2220
    },
    {
      "epoch": 0.2683513838748496,
      "grad_norm": 5.703465938568115,
      "learning_rate": 8.588484498363182e-05,
      "loss": 2.4233,
      "step": 2230
    },
    {
      "epoch": 0.2695547533092659,
      "grad_norm": 3.678798198699951,
      "learning_rate": 8.626997881763913e-05,
      "loss": 2.1464,
      "step": 2240
    },
    {
      "epoch": 0.27075812274368233,
      "grad_norm": 5.3430023193359375,
      "learning_rate": 8.665511265164645e-05,
      "loss": 2.3755,
      "step": 2250
    },
    {
      "epoch": 0.2719614921780987,
      "grad_norm": 4.224210739135742,
      "learning_rate": 8.704024648565377e-05,
      "loss": 2.068,
      "step": 2260
    },
    {
      "epoch": 0.273164861612515,
      "grad_norm": 3.020923137664795,
      "learning_rate": 8.742538031966108e-05,
      "loss": 2.2785,
      "step": 2270
    },
    {
      "epoch": 0.2743682310469314,
      "grad_norm": 5.135770320892334,
      "learning_rate": 8.78105141536684e-05,
      "loss": 2.43,
      "step": 2280
    },
    {
      "epoch": 0.27557160048134777,
      "grad_norm": 3.6533546447753906,
      "learning_rate": 8.819564798767572e-05,
      "loss": 2.1872,
      "step": 2290
    },
    {
      "epoch": 0.2767749699157641,
      "grad_norm": 5.642998218536377,
      "learning_rate": 8.858078182168303e-05,
      "loss": 2.4434,
      "step": 2300
    },
    {
      "epoch": 0.2779783393501805,
      "grad_norm": 3.8212432861328125,
      "learning_rate": 8.896591565569036e-05,
      "loss": 2.3333,
      "step": 2310
    },
    {
      "epoch": 0.27918170878459686,
      "grad_norm": 2.579754114151001,
      "learning_rate": 8.935104948969766e-05,
      "loss": 2.4356,
      "step": 2320
    },
    {
      "epoch": 0.28038507821901326,
      "grad_norm": 4.244583606719971,
      "learning_rate": 8.9736183323705e-05,
      "loss": 2.4217,
      "step": 2330
    },
    {
      "epoch": 0.2815884476534296,
      "grad_norm": 3.4745070934295654,
      "learning_rate": 9.01213171577123e-05,
      "loss": 2.4493,
      "step": 2340
    },
    {
      "epoch": 0.28279181708784595,
      "grad_norm": 5.872068881988525,
      "learning_rate": 9.050645099171963e-05,
      "loss": 2.2935,
      "step": 2350
    },
    {
      "epoch": 0.28399518652226236,
      "grad_norm": 3.9975595474243164,
      "learning_rate": 9.089158482572694e-05,
      "loss": 2.184,
      "step": 2360
    },
    {
      "epoch": 0.2851985559566787,
      "grad_norm": 3.1640396118164062,
      "learning_rate": 9.127671865973426e-05,
      "loss": 2.4179,
      "step": 2370
    },
    {
      "epoch": 0.28640192539109505,
      "grad_norm": 4.480766773223877,
      "learning_rate": 9.166185249374159e-05,
      "loss": 2.0609,
      "step": 2380
    },
    {
      "epoch": 0.28760529482551145,
      "grad_norm": 3.7833733558654785,
      "learning_rate": 9.204698632774889e-05,
      "loss": 2.349,
      "step": 2390
    },
    {
      "epoch": 0.2888086642599278,
      "grad_norm": 6.1306352615356445,
      "learning_rate": 9.243212016175622e-05,
      "loss": 2.373,
      "step": 2400
    },
    {
      "epoch": 0.29001203369434414,
      "grad_norm": 4.305640697479248,
      "learning_rate": 9.281725399576352e-05,
      "loss": 2.2492,
      "step": 2410
    },
    {
      "epoch": 0.29121540312876054,
      "grad_norm": 2.8770713806152344,
      "learning_rate": 9.320238782977085e-05,
      "loss": 2.1706,
      "step": 2420
    },
    {
      "epoch": 0.2924187725631769,
      "grad_norm": 4.647495269775391,
      "learning_rate": 9.358752166377817e-05,
      "loss": 2.1634,
      "step": 2430
    },
    {
      "epoch": 0.29362214199759323,
      "grad_norm": 3.7371299266815186,
      "learning_rate": 9.397265549778549e-05,
      "loss": 2.3072,
      "step": 2440
    },
    {
      "epoch": 0.29482551143200963,
      "grad_norm": 5.267301559448242,
      "learning_rate": 9.43577893317928e-05,
      "loss": 2.514,
      "step": 2450
    },
    {
      "epoch": 0.296028880866426,
      "grad_norm": 4.16841459274292,
      "learning_rate": 9.474292316580012e-05,
      "loss": 2.085,
      "step": 2460
    },
    {
      "epoch": 0.2972322503008424,
      "grad_norm": 3.1590590476989746,
      "learning_rate": 9.512805699980744e-05,
      "loss": 2.3315,
      "step": 2470
    },
    {
      "epoch": 0.29843561973525873,
      "grad_norm": 4.402718544006348,
      "learning_rate": 9.551319083381475e-05,
      "loss": 2.2205,
      "step": 2480
    },
    {
      "epoch": 0.2996389891696751,
      "grad_norm": 3.7688190937042236,
      "learning_rate": 9.589832466782207e-05,
      "loss": 2.3368,
      "step": 2490
    },
    {
      "epoch": 0.3008423586040915,
      "grad_norm": 3.71555757522583,
      "learning_rate": 9.62834585018294e-05,
      "loss": 2.2572,
      "step": 2500
    },
    {
      "epoch": 0.3020457280385078,
      "grad_norm": 4.018877983093262,
      "learning_rate": 9.66685923358367e-05,
      "loss": 2.2546,
      "step": 2510
    },
    {
      "epoch": 0.30324909747292417,
      "grad_norm": 3.1480233669281006,
      "learning_rate": 9.705372616984403e-05,
      "loss": 2.4473,
      "step": 2520
    },
    {
      "epoch": 0.30445246690734057,
      "grad_norm": 4.867123603820801,
      "learning_rate": 9.743886000385133e-05,
      "loss": 2.2999,
      "step": 2530
    },
    {
      "epoch": 0.3056558363417569,
      "grad_norm": 3.555675745010376,
      "learning_rate": 9.782399383785866e-05,
      "loss": 2.3267,
      "step": 2540
    },
    {
      "epoch": 0.30685920577617326,
      "grad_norm": 5.9477362632751465,
      "learning_rate": 9.820912767186598e-05,
      "loss": 2.3903,
      "step": 2550
    },
    {
      "epoch": 0.30806257521058966,
      "grad_norm": 3.783031940460205,
      "learning_rate": 9.85942615058733e-05,
      "loss": 2.3415,
      "step": 2560
    },
    {
      "epoch": 0.309265944645006,
      "grad_norm": 3.087137222290039,
      "learning_rate": 9.897939533988061e-05,
      "loss": 2.4655,
      "step": 2570
    },
    {
      "epoch": 0.3104693140794224,
      "grad_norm": 5.595900058746338,
      "learning_rate": 9.936452917388793e-05,
      "loss": 2.3899,
      "step": 2580
    },
    {
      "epoch": 0.31167268351383876,
      "grad_norm": 3.25927734375,
      "learning_rate": 9.974966300789525e-05,
      "loss": 2.3068,
      "step": 2590
    },
    {
      "epoch": 0.3128760529482551,
      "grad_norm": 5.424354553222656,
      "learning_rate": 0.00010013479684190256,
      "loss": 2.4695,
      "step": 2600
    },
    {
      "epoch": 0.3140794223826715,
      "grad_norm": 3.853463888168335,
      "learning_rate": 0.00010051993067590988,
      "loss": 2.206,
      "step": 2610
    },
    {
      "epoch": 0.31528279181708785,
      "grad_norm": 3.2356932163238525,
      "learning_rate": 0.0001009050645099172,
      "loss": 2.1866,
      "step": 2620
    },
    {
      "epoch": 0.3164861612515042,
      "grad_norm": 4.676336288452148,
      "learning_rate": 0.00010129019834392452,
      "loss": 2.4018,
      "step": 2630
    },
    {
      "epoch": 0.3176895306859206,
      "grad_norm": 4.09110689163208,
      "learning_rate": 0.00010167533217793184,
      "loss": 2.1298,
      "step": 2640
    },
    {
      "epoch": 0.31889290012033694,
      "grad_norm": 4.968358993530273,
      "learning_rate": 0.00010206046601193914,
      "loss": 2.1658,
      "step": 2650
    },
    {
      "epoch": 0.3200962695547533,
      "grad_norm": 3.5447754859924316,
      "learning_rate": 0.00010244559984594646,
      "loss": 2.1519,
      "step": 2660
    },
    {
      "epoch": 0.3212996389891697,
      "grad_norm": 3.910879373550415,
      "learning_rate": 0.00010283073367995379,
      "loss": 2.5106,
      "step": 2670
    },
    {
      "epoch": 0.32250300842358604,
      "grad_norm": 4.198660373687744,
      "learning_rate": 0.0001032158675139611,
      "loss": 2.3276,
      "step": 2680
    },
    {
      "epoch": 0.3237063778580024,
      "grad_norm": 3.775038003921509,
      "learning_rate": 0.00010360100134796842,
      "loss": 2.1432,
      "step": 2690
    },
    {
      "epoch": 0.3249097472924188,
      "grad_norm": 5.7064900398254395,
      "learning_rate": 0.00010398613518197575,
      "loss": 2.3454,
      "step": 2700
    },
    {
      "epoch": 0.32611311672683513,
      "grad_norm": 4.214547157287598,
      "learning_rate": 0.00010437126901598307,
      "loss": 2.0752,
      "step": 2710
    },
    {
      "epoch": 0.32731648616125153,
      "grad_norm": 2.9629077911376953,
      "learning_rate": 0.00010475640284999037,
      "loss": 2.0854,
      "step": 2720
    },
    {
      "epoch": 0.3285198555956679,
      "grad_norm": 5.487660884857178,
      "learning_rate": 0.00010514153668399769,
      "loss": 2.279,
      "step": 2730
    },
    {
      "epoch": 0.3297232250300842,
      "grad_norm": 4.312379360198975,
      "learning_rate": 0.00010552667051800502,
      "loss": 2.289,
      "step": 2740
    },
    {
      "epoch": 0.3309265944645006,
      "grad_norm": 5.722112655639648,
      "learning_rate": 0.00010591180435201233,
      "loss": 2.447,
      "step": 2750
    },
    {
      "epoch": 0.33212996389891697,
      "grad_norm": 3.474245309829712,
      "learning_rate": 0.00010629693818601965,
      "loss": 2.2201,
      "step": 2760
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 2.9945054054260254,
      "learning_rate": 0.00010668207202002695,
      "loss": 2.0959,
      "step": 2770
    },
    {
      "epoch": 0.3345367027677497,
      "grad_norm": 3.9089252948760986,
      "learning_rate": 0.0001070672058540343,
      "loss": 2.1162,
      "step": 2780
    },
    {
      "epoch": 0.33574007220216606,
      "grad_norm": 3.8732986450195312,
      "learning_rate": 0.0001074523396880416,
      "loss": 2.4375,
      "step": 2790
    },
    {
      "epoch": 0.3369434416365824,
      "grad_norm": 4.811378002166748,
      "learning_rate": 0.00010783747352204891,
      "loss": 2.3234,
      "step": 2800
    },
    {
      "epoch": 0.3381468110709988,
      "grad_norm": 4.7229108810424805,
      "learning_rate": 0.00010822260735605623,
      "loss": 1.9816,
      "step": 2810
    },
    {
      "epoch": 0.33935018050541516,
      "grad_norm": 3.5141379833221436,
      "learning_rate": 0.00010860774119006356,
      "loss": 2.3154,
      "step": 2820
    },
    {
      "epoch": 0.3405535499398315,
      "grad_norm": 4.4615888595581055,
      "learning_rate": 0.00010899287502407088,
      "loss": 2.3536,
      "step": 2830
    },
    {
      "epoch": 0.3417569193742479,
      "grad_norm": 4.35577392578125,
      "learning_rate": 0.00010937800885807818,
      "loss": 2.0574,
      "step": 2840
    },
    {
      "epoch": 0.34296028880866425,
      "grad_norm": 5.919861793518066,
      "learning_rate": 0.0001097631426920855,
      "loss": 2.2634,
      "step": 2850
    },
    {
      "epoch": 0.34416365824308065,
      "grad_norm": 4.538342475891113,
      "learning_rate": 0.00011014827652609283,
      "loss": 2.0598,
      "step": 2860
    },
    {
      "epoch": 0.345367027677497,
      "grad_norm": 3.16560435295105,
      "learning_rate": 0.00011053341036010014,
      "loss": 2.2131,
      "step": 2870
    },
    {
      "epoch": 0.34657039711191334,
      "grad_norm": 4.720436096191406,
      "learning_rate": 0.00011091854419410746,
      "loss": 2.2599,
      "step": 2880
    },
    {
      "epoch": 0.34777376654632974,
      "grad_norm": 4.121572971343994,
      "learning_rate": 0.00011130367802811476,
      "loss": 2.0998,
      "step": 2890
    },
    {
      "epoch": 0.3489771359807461,
      "grad_norm": 4.9068732261657715,
      "learning_rate": 0.0001116888118621221,
      "loss": 2.1145,
      "step": 2900
    },
    {
      "epoch": 0.35018050541516244,
      "grad_norm": 4.335965156555176,
      "learning_rate": 0.00011207394569612941,
      "loss": 2.2301,
      "step": 2910
    },
    {
      "epoch": 0.35138387484957884,
      "grad_norm": 3.4327359199523926,
      "learning_rate": 0.00011245907953013672,
      "loss": 2.3188,
      "step": 2920
    },
    {
      "epoch": 0.3525872442839952,
      "grad_norm": 5.5908894538879395,
      "learning_rate": 0.00011284421336414404,
      "loss": 2.1288,
      "step": 2930
    },
    {
      "epoch": 0.35379061371841153,
      "grad_norm": 3.7546348571777344,
      "learning_rate": 0.00011322934719815137,
      "loss": 2.1982,
      "step": 2940
    },
    {
      "epoch": 0.35499398315282793,
      "grad_norm": 5.478588104248047,
      "learning_rate": 0.00011361448103215869,
      "loss": 2.1034,
      "step": 2950
    },
    {
      "epoch": 0.3561973525872443,
      "grad_norm": 4.286993503570557,
      "learning_rate": 0.00011399961486616599,
      "loss": 2.279,
      "step": 2960
    },
    {
      "epoch": 0.3574007220216607,
      "grad_norm": 3.498656749725342,
      "learning_rate": 0.0001143847487001733,
      "loss": 2.1784,
      "step": 2970
    },
    {
      "epoch": 0.358604091456077,
      "grad_norm": 4.852539539337158,
      "learning_rate": 0.00011476988253418063,
      "loss": 2.3248,
      "step": 2980
    },
    {
      "epoch": 0.35980746089049337,
      "grad_norm": 3.67238450050354,
      "learning_rate": 0.00011515501636818795,
      "loss": 2.0548,
      "step": 2990
    },
    {
      "epoch": 0.36101083032490977,
      "grad_norm": 5.410921573638916,
      "learning_rate": 0.00011554015020219527,
      "loss": 2.395,
      "step": 3000
    },
    {
      "epoch": 0.3622141997593261,
      "grad_norm": 4.061553478240967,
      "learning_rate": 0.00011592528403620257,
      "loss": 2.1706,
      "step": 3010
    },
    {
      "epoch": 0.36341756919374246,
      "grad_norm": 2.949221611022949,
      "learning_rate": 0.00011631041787020991,
      "loss": 2.3203,
      "step": 3020
    },
    {
      "epoch": 0.36462093862815886,
      "grad_norm": 4.697376251220703,
      "learning_rate": 0.00011669555170421722,
      "loss": 2.3691,
      "step": 3030
    },
    {
      "epoch": 0.3658243080625752,
      "grad_norm": 3.6829640865325928,
      "learning_rate": 0.00011708068553822453,
      "loss": 2.2573,
      "step": 3040
    },
    {
      "epoch": 0.36702767749699156,
      "grad_norm": 5.870731830596924,
      "learning_rate": 0.00011746581937223186,
      "loss": 2.3537,
      "step": 3050
    },
    {
      "epoch": 0.36823104693140796,
      "grad_norm": 3.713322877883911,
      "learning_rate": 0.00011785095320623918,
      "loss": 2.1367,
      "step": 3060
    },
    {
      "epoch": 0.3694344163658243,
      "grad_norm": 3.511606454849243,
      "learning_rate": 0.0001182360870402465,
      "loss": 2.2126,
      "step": 3070
    },
    {
      "epoch": 0.37063778580024065,
      "grad_norm": 6.14950704574585,
      "learning_rate": 0.0001186212208742538,
      "loss": 2.2284,
      "step": 3080
    },
    {
      "epoch": 0.37184115523465705,
      "grad_norm": 3.878574848175049,
      "learning_rate": 0.00011900635470826114,
      "loss": 2.1863,
      "step": 3090
    },
    {
      "epoch": 0.3730445246690734,
      "grad_norm": 7.3028059005737305,
      "learning_rate": 0.00011939148854226844,
      "loss": 2.2284,
      "step": 3100
    },
    {
      "epoch": 0.3742478941034898,
      "grad_norm": 8.766607284545898,
      "learning_rate": 0.00011977662237627576,
      "loss": 2.2654,
      "step": 3110
    },
    {
      "epoch": 0.37545126353790614,
      "grad_norm": 3.1363110542297363,
      "learning_rate": 0.00012016175621028308,
      "loss": 2.2604,
      "step": 3120
    },
    {
      "epoch": 0.3766546329723225,
      "grad_norm": 5.78159761428833,
      "learning_rate": 0.0001205468900442904,
      "loss": 2.2574,
      "step": 3130
    },
    {
      "epoch": 0.3778580024067389,
      "grad_norm": 3.4998180866241455,
      "learning_rate": 0.00012093202387829772,
      "loss": 2.0765,
      "step": 3140
    },
    {
      "epoch": 0.37906137184115524,
      "grad_norm": 5.023599147796631,
      "learning_rate": 0.00012131715771230503,
      "loss": 2.1099,
      "step": 3150
    },
    {
      "epoch": 0.3802647412755716,
      "grad_norm": 4.230167865753174,
      "learning_rate": 0.00012170229154631234,
      "loss": 2.068,
      "step": 3160
    },
    {
      "epoch": 0.381468110709988,
      "grad_norm": 3.3696231842041016,
      "learning_rate": 0.00012208742538031967,
      "loss": 2.1214,
      "step": 3170
    },
    {
      "epoch": 0.38267148014440433,
      "grad_norm": 4.373711109161377,
      "learning_rate": 0.000122472559214327,
      "loss": 2.0965,
      "step": 3180
    },
    {
      "epoch": 0.3838748495788207,
      "grad_norm": 3.5923421382904053,
      "learning_rate": 0.0001228576930483343,
      "loss": 2.1647,
      "step": 3190
    },
    {
      "epoch": 0.3850782190132371,
      "grad_norm": 5.419064044952393,
      "learning_rate": 0.00012324282688234162,
      "loss": 2.2482,
      "step": 3200
    },
    {
      "epoch": 0.3862815884476534,
      "grad_norm": 4.109050273895264,
      "learning_rate": 0.00012362796071634894,
      "loss": 2.1084,
      "step": 3210
    },
    {
      "epoch": 0.38748495788206977,
      "grad_norm": 3.0200486183166504,
      "learning_rate": 0.00012401309455035625,
      "loss": 2.0833,
      "step": 3220
    },
    {
      "epoch": 0.38868832731648617,
      "grad_norm": 5.205567836761475,
      "learning_rate": 0.00012439822838436357,
      "loss": 2.223,
      "step": 3230
    },
    {
      "epoch": 0.3898916967509025,
      "grad_norm": 3.1712558269500732,
      "learning_rate": 0.00012478336221837089,
      "loss": 1.9621,
      "step": 3240
    },
    {
      "epoch": 0.3910950661853189,
      "grad_norm": 6.77772331237793,
      "learning_rate": 0.0001251684960523782,
      "loss": 2.1003,
      "step": 3250
    },
    {
      "epoch": 0.39229843561973526,
      "grad_norm": 4.719583988189697,
      "learning_rate": 0.00012555362988638552,
      "loss": 2.3107,
      "step": 3260
    },
    {
      "epoch": 0.3935018050541516,
      "grad_norm": 2.557326316833496,
      "learning_rate": 0.00012593876372039283,
      "loss": 2.2969,
      "step": 3270
    },
    {
      "epoch": 0.394705174488568,
      "grad_norm": 3.8585081100463867,
      "learning_rate": 0.00012632389755440015,
      "loss": 2.2111,
      "step": 3280
    },
    {
      "epoch": 0.39590854392298436,
      "grad_norm": 3.757640838623047,
      "learning_rate": 0.0001267090313884075,
      "loss": 2.1898,
      "step": 3290
    },
    {
      "epoch": 0.3971119133574007,
      "grad_norm": 5.546092987060547,
      "learning_rate": 0.00012709416522241478,
      "loss": 2.2393,
      "step": 3300
    },
    {
      "epoch": 0.3983152827918171,
      "grad_norm": 3.8220839500427246,
      "learning_rate": 0.0001274792990564221,
      "loss": 2.1682,
      "step": 3310
    },
    {
      "epoch": 0.39951865222623345,
      "grad_norm": 3.2766096591949463,
      "learning_rate": 0.00012786443289042942,
      "loss": 2.2139,
      "step": 3320
    },
    {
      "epoch": 0.4007220216606498,
      "grad_norm": 5.108669757843018,
      "learning_rate": 0.00012824956672443676,
      "loss": 2.0244,
      "step": 3330
    },
    {
      "epoch": 0.4019253910950662,
      "grad_norm": 3.889655113220215,
      "learning_rate": 0.00012863470055844408,
      "loss": 2.1231,
      "step": 3340
    },
    {
      "epoch": 0.40312876052948254,
      "grad_norm": 5.427848815917969,
      "learning_rate": 0.00012901983439245136,
      "loss": 2.158,
      "step": 3350
    },
    {
      "epoch": 0.4043321299638989,
      "grad_norm": 3.919846773147583,
      "learning_rate": 0.0001294049682264587,
      "loss": 2.1693,
      "step": 3360
    },
    {
      "epoch": 0.4055354993983153,
      "grad_norm": 2.8813560009002686,
      "learning_rate": 0.00012979010206046602,
      "loss": 2.2194,
      "step": 3370
    },
    {
      "epoch": 0.40673886883273164,
      "grad_norm": 5.581686019897461,
      "learning_rate": 0.00013017523589447334,
      "loss": 2.2722,
      "step": 3380
    },
    {
      "epoch": 0.40794223826714804,
      "grad_norm": 3.22434401512146,
      "learning_rate": 0.00013056036972848066,
      "loss": 2.3132,
      "step": 3390
    },
    {
      "epoch": 0.4091456077015644,
      "grad_norm": 5.471015930175781,
      "learning_rate": 0.00013094550356248797,
      "loss": 2.3227,
      "step": 3400
    },
    {
      "epoch": 0.41034897713598073,
      "grad_norm": 4.294218063354492,
      "learning_rate": 0.0001313306373964953,
      "loss": 2.3273,
      "step": 3410
    },
    {
      "epoch": 0.41155234657039713,
      "grad_norm": 4.315289497375488,
      "learning_rate": 0.0001317157712305026,
      "loss": 2.0973,
      "step": 3420
    },
    {
      "epoch": 0.4127557160048135,
      "grad_norm": 4.89292573928833,
      "learning_rate": 0.00013210090506450992,
      "loss": 2.2171,
      "step": 3430
    },
    {
      "epoch": 0.4139590854392298,
      "grad_norm": 3.9335169792175293,
      "learning_rate": 0.00013248603889851724,
      "loss": 2.363,
      "step": 3440
    },
    {
      "epoch": 0.4151624548736462,
      "grad_norm": 5.29353666305542,
      "learning_rate": 0.00013287117273252455,
      "loss": 2.3166,
      "step": 3450
    },
    {
      "epoch": 0.41636582430806257,
      "grad_norm": 4.394747257232666,
      "learning_rate": 0.00013325630656653187,
      "loss": 2.1407,
      "step": 3460
    },
    {
      "epoch": 0.4175691937424789,
      "grad_norm": 2.707184314727783,
      "learning_rate": 0.0001336414404005392,
      "loss": 2.3482,
      "step": 3470
    },
    {
      "epoch": 0.4187725631768953,
      "grad_norm": 5.04945182800293,
      "learning_rate": 0.00013402657423454653,
      "loss": 2.2553,
      "step": 3480
    },
    {
      "epoch": 0.41997593261131166,
      "grad_norm": 3.828322410583496,
      "learning_rate": 0.00013441170806855382,
      "loss": 2.0523,
      "step": 3490
    },
    {
      "epoch": 0.42117930204572807,
      "grad_norm": 6.2465925216674805,
      "learning_rate": 0.00013479684190256114,
      "loss": 2.4015,
      "step": 3500
    },
    {
      "epoch": 0.4223826714801444,
      "grad_norm": 5.26687479019165,
      "learning_rate": 0.00013518197573656845,
      "loss": 2.117,
      "step": 3510
    },
    {
      "epoch": 0.42358604091456076,
      "grad_norm": 4.215747356414795,
      "learning_rate": 0.0001355671095705758,
      "loss": 2.2339,
      "step": 3520
    },
    {
      "epoch": 0.42478941034897716,
      "grad_norm": 4.485904216766357,
      "learning_rate": 0.0001359522434045831,
      "loss": 2.2601,
      "step": 3530
    },
    {
      "epoch": 0.4259927797833935,
      "grad_norm": 4.049141883850098,
      "learning_rate": 0.0001363373772385904,
      "loss": 2.0584,
      "step": 3540
    },
    {
      "epoch": 0.42719614921780985,
      "grad_norm": 5.23921537399292,
      "learning_rate": 0.00013672251107259772,
      "loss": 2.1419,
      "step": 3550
    },
    {
      "epoch": 0.42839951865222625,
      "grad_norm": 5.058278560638428,
      "learning_rate": 0.00013710764490660506,
      "loss": 2.1201,
      "step": 3560
    },
    {
      "epoch": 0.4296028880866426,
      "grad_norm": 5.548385143280029,
      "learning_rate": 0.00013749277874061238,
      "loss": 2.3672,
      "step": 3570
    },
    {
      "epoch": 0.43080625752105894,
      "grad_norm": 4.900745868682861,
      "learning_rate": 0.0001378779125746197,
      "loss": 2.2147,
      "step": 3580
    },
    {
      "epoch": 0.43200962695547535,
      "grad_norm": 3.4081718921661377,
      "learning_rate": 0.00013826304640862698,
      "loss": 2.1727,
      "step": 3590
    },
    {
      "epoch": 0.4332129963898917,
      "grad_norm": 5.376938343048096,
      "learning_rate": 0.00013864818024263433,
      "loss": 2.0224,
      "step": 3600
    },
    {
      "epoch": 0.43441636582430804,
      "grad_norm": 4.5283708572387695,
      "learning_rate": 0.00013903331407664164,
      "loss": 2.2583,
      "step": 3610
    },
    {
      "epoch": 0.43561973525872444,
      "grad_norm": 3.4518930912017822,
      "learning_rate": 0.00013941844791064896,
      "loss": 2.27,
      "step": 3620
    },
    {
      "epoch": 0.4368231046931408,
      "grad_norm": 5.660602569580078,
      "learning_rate": 0.00013980358174465628,
      "loss": 2.0659,
      "step": 3630
    },
    {
      "epoch": 0.4380264741275572,
      "grad_norm": 3.406203269958496,
      "learning_rate": 0.0001401887155786636,
      "loss": 2.2138,
      "step": 3640
    },
    {
      "epoch": 0.43922984356197353,
      "grad_norm": 5.18516731262207,
      "learning_rate": 0.0001405738494126709,
      "loss": 2.3397,
      "step": 3650
    },
    {
      "epoch": 0.4404332129963899,
      "grad_norm": 4.447347164154053,
      "learning_rate": 0.00014095898324667822,
      "loss": 1.9989,
      "step": 3660
    },
    {
      "epoch": 0.4416365824308063,
      "grad_norm": 3.6428494453430176,
      "learning_rate": 0.00014134411708068554,
      "loss": 2.2966,
      "step": 3670
    },
    {
      "epoch": 0.4428399518652226,
      "grad_norm": 5.201837539672852,
      "learning_rate": 0.00014172925091469286,
      "loss": 2.2646,
      "step": 3680
    },
    {
      "epoch": 0.44404332129963897,
      "grad_norm": 4.126829147338867,
      "learning_rate": 0.00014211438474870017,
      "loss": 2.4197,
      "step": 3690
    },
    {
      "epoch": 0.4452466907340554,
      "grad_norm": 6.108100414276123,
      "learning_rate": 0.0001424995185827075,
      "loss": 2.1061,
      "step": 3700
    },
    {
      "epoch": 0.4464500601684717,
      "grad_norm": 4.69503116607666,
      "learning_rate": 0.00014288465241671483,
      "loss": 2.1273,
      "step": 3710
    },
    {
      "epoch": 0.44765342960288806,
      "grad_norm": 2.9712235927581787,
      "learning_rate": 0.00014326978625072215,
      "loss": 2.1659,
      "step": 3720
    },
    {
      "epoch": 0.44885679903730447,
      "grad_norm": 4.880232334136963,
      "learning_rate": 0.00014365492008472944,
      "loss": 2.1072,
      "step": 3730
    },
    {
      "epoch": 0.4500601684717208,
      "grad_norm": 4.117298126220703,
      "learning_rate": 0.00014404005391873675,
      "loss": 2.132,
      "step": 3740
    },
    {
      "epoch": 0.45126353790613716,
      "grad_norm": 6.599010944366455,
      "learning_rate": 0.0001444251877527441,
      "loss": 2.2534,
      "step": 3750
    },
    {
      "epoch": 0.45246690734055356,
      "grad_norm": 4.913687705993652,
      "learning_rate": 0.00014481032158675141,
      "loss": 2.2075,
      "step": 3760
    },
    {
      "epoch": 0.4536702767749699,
      "grad_norm": 3.421839714050293,
      "learning_rate": 0.00014519545542075873,
      "loss": 2.0303,
      "step": 3770
    },
    {
      "epoch": 0.4548736462093863,
      "grad_norm": 4.960007667541504,
      "learning_rate": 0.00014558058925476602,
      "loss": 2.1867,
      "step": 3780
    },
    {
      "epoch": 0.45607701564380265,
      "grad_norm": 4.16343879699707,
      "learning_rate": 0.00014596572308877336,
      "loss": 2.2308,
      "step": 3790
    },
    {
      "epoch": 0.457280385078219,
      "grad_norm": 5.702826023101807,
      "learning_rate": 0.00014635085692278068,
      "loss": 2.183,
      "step": 3800
    },
    {
      "epoch": 0.4584837545126354,
      "grad_norm": 4.250916957855225,
      "learning_rate": 0.000146735990756788,
      "loss": 2.0379,
      "step": 3810
    },
    {
      "epoch": 0.45968712394705175,
      "grad_norm": 3.0171141624450684,
      "learning_rate": 0.0001471211245907953,
      "loss": 2.2479,
      "step": 3820
    },
    {
      "epoch": 0.4608904933814681,
      "grad_norm": 5.237485885620117,
      "learning_rate": 0.00014750625842480263,
      "loss": 2.1692,
      "step": 3830
    },
    {
      "epoch": 0.4620938628158845,
      "grad_norm": 4.266112804412842,
      "learning_rate": 0.00014789139225880994,
      "loss": 2.2255,
      "step": 3840
    },
    {
      "epoch": 0.46329723225030084,
      "grad_norm": 5.257189750671387,
      "learning_rate": 0.00014827652609281726,
      "loss": 2.0497,
      "step": 3850
    },
    {
      "epoch": 0.4645006016847172,
      "grad_norm": 3.7063605785369873,
      "learning_rate": 0.00014866165992682458,
      "loss": 2.1092,
      "step": 3860
    },
    {
      "epoch": 0.4657039711191336,
      "grad_norm": 3.2208852767944336,
      "learning_rate": 0.0001490467937608319,
      "loss": 2.2522,
      "step": 3870
    },
    {
      "epoch": 0.46690734055354993,
      "grad_norm": 5.302590370178223,
      "learning_rate": 0.0001494319275948392,
      "loss": 2.1958,
      "step": 3880
    },
    {
      "epoch": 0.4681107099879663,
      "grad_norm": 3.7522192001342773,
      "learning_rate": 0.00014981706142884653,
      "loss": 2.2607,
      "step": 3890
    },
    {
      "epoch": 0.4693140794223827,
      "grad_norm": 6.742669105529785,
      "learning_rate": 0.00015020219526285384,
      "loss": 2.4315,
      "step": 3900
    },
    {
      "epoch": 0.470517448856799,
      "grad_norm": 3.790123462677002,
      "learning_rate": 0.00015058732909686119,
      "loss": 2.0291,
      "step": 3910
    },
    {
      "epoch": 0.4717208182912154,
      "grad_norm": 3.7009456157684326,
      "learning_rate": 0.00015097246293086847,
      "loss": 2.2023,
      "step": 3920
    },
    {
      "epoch": 0.4729241877256318,
      "grad_norm": 5.0835089683532715,
      "learning_rate": 0.0001513575967648758,
      "loss": 2.1321,
      "step": 3930
    },
    {
      "epoch": 0.4741275571600481,
      "grad_norm": 3.5800023078918457,
      "learning_rate": 0.0001517427305988831,
      "loss": 2.1163,
      "step": 3940
    },
    {
      "epoch": 0.4753309265944645,
      "grad_norm": 5.33997917175293,
      "learning_rate": 0.00015212786443289045,
      "loss": 2.0387,
      "step": 3950
    },
    {
      "epoch": 0.47653429602888087,
      "grad_norm": 4.518845558166504,
      "learning_rate": 0.00015251299826689777,
      "loss": 2.0584,
      "step": 3960
    },
    {
      "epoch": 0.4777376654632972,
      "grad_norm": 3.687009334564209,
      "learning_rate": 0.00015289813210090506,
      "loss": 2.2745,
      "step": 3970
    },
    {
      "epoch": 0.4789410348977136,
      "grad_norm": 4.833267688751221,
      "learning_rate": 0.00015328326593491237,
      "loss": 2.3606,
      "step": 3980
    },
    {
      "epoch": 0.48014440433212996,
      "grad_norm": 3.7437093257904053,
      "learning_rate": 0.00015366839976891972,
      "loss": 2.2791,
      "step": 3990
    },
    {
      "epoch": 0.4813477737665463,
      "grad_norm": 5.243849754333496,
      "learning_rate": 0.00015405353360292703,
      "loss": 2.2659,
      "step": 4000
    },
    {
      "epoch": 0.4825511432009627,
      "grad_norm": 4.449324131011963,
      "learning_rate": 0.00015443866743693435,
      "loss": 2.2892,
      "step": 4010
    },
    {
      "epoch": 0.48375451263537905,
      "grad_norm": 3.0258562564849854,
      "learning_rate": 0.00015482380127094164,
      "loss": 1.9254,
      "step": 4020
    },
    {
      "epoch": 0.48495788206979545,
      "grad_norm": 7.307828426361084,
      "learning_rate": 0.00015520893510494898,
      "loss": 2.2143,
      "step": 4030
    },
    {
      "epoch": 0.4861612515042118,
      "grad_norm": 3.417057991027832,
      "learning_rate": 0.0001555940689389563,
      "loss": 2.1549,
      "step": 4040
    },
    {
      "epoch": 0.48736462093862815,
      "grad_norm": 6.394173622131348,
      "learning_rate": 0.00015597920277296361,
      "loss": 2.403,
      "step": 4050
    },
    {
      "epoch": 0.48856799037304455,
      "grad_norm": 4.979101181030273,
      "learning_rate": 0.00015636433660697093,
      "loss": 2.0301,
      "step": 4060
    },
    {
      "epoch": 0.4897713598074609,
      "grad_norm": 3.3656721115112305,
      "learning_rate": 0.00015674947044097825,
      "loss": 2.2258,
      "step": 4070
    },
    {
      "epoch": 0.49097472924187724,
      "grad_norm": 5.95798397064209,
      "learning_rate": 0.00015713460427498556,
      "loss": 2.4844,
      "step": 4080
    },
    {
      "epoch": 0.49217809867629364,
      "grad_norm": 3.4689602851867676,
      "learning_rate": 0.00015751973810899288,
      "loss": 2.0246,
      "step": 4090
    },
    {
      "epoch": 0.49338146811071,
      "grad_norm": 5.611385345458984,
      "learning_rate": 0.00015790487194300022,
      "loss": 2.1991,
      "step": 4100
    },
    {
      "epoch": 0.49458483754512633,
      "grad_norm": 4.616299152374268,
      "learning_rate": 0.0001582900057770075,
      "loss": 2.0429,
      "step": 4110
    },
    {
      "epoch": 0.49578820697954273,
      "grad_norm": 3.6518077850341797,
      "learning_rate": 0.00015867513961101483,
      "loss": 2.4328,
      "step": 4120
    },
    {
      "epoch": 0.4969915764139591,
      "grad_norm": 5.260908603668213,
      "learning_rate": 0.00015906027344502214,
      "loss": 2.2006,
      "step": 4130
    },
    {
      "epoch": 0.4981949458483754,
      "grad_norm": 3.864971160888672,
      "learning_rate": 0.0001594454072790295,
      "loss": 2.2522,
      "step": 4140
    },
    {
      "epoch": 0.4993983152827918,
      "grad_norm": 7.640333652496338,
      "learning_rate": 0.0001598305411130368,
      "loss": 2.2165,
      "step": 4150
    },
    {
      "epoch": 0.5006016847172082,
      "grad_norm": 4.940099716186523,
      "learning_rate": 0.0001602156749470441,
      "loss": 2.1367,
      "step": 4160
    },
    {
      "epoch": 0.5018050541516246,
      "grad_norm": 2.9353463649749756,
      "learning_rate": 0.0001606008087810514,
      "loss": 2.2242,
      "step": 4170
    },
    {
      "epoch": 0.5030084235860409,
      "grad_norm": 4.441645622253418,
      "learning_rate": 0.00016098594261505875,
      "loss": 2.244,
      "step": 4180
    },
    {
      "epoch": 0.5042117930204573,
      "grad_norm": 3.1046464443206787,
      "learning_rate": 0.00016137107644906607,
      "loss": 2.1284,
      "step": 4190
    },
    {
      "epoch": 0.5054151624548736,
      "grad_norm": 4.9619903564453125,
      "learning_rate": 0.00016175621028307339,
      "loss": 2.3206,
      "step": 4200
    },
    {
      "epoch": 0.50661853188929,
      "grad_norm": 4.747543811798096,
      "learning_rate": 0.00016214134411708067,
      "loss": 2.1086,
      "step": 4210
    },
    {
      "epoch": 0.5078219013237064,
      "grad_norm": 3.993290424346924,
      "learning_rate": 0.00016252647795108802,
      "loss": 2.2709,
      "step": 4220
    },
    {
      "epoch": 0.5090252707581228,
      "grad_norm": 4.79225492477417,
      "learning_rate": 0.00016291161178509533,
      "loss": 2.1154,
      "step": 4230
    },
    {
      "epoch": 0.5102286401925391,
      "grad_norm": 3.7596824169158936,
      "learning_rate": 0.00016329674561910265,
      "loss": 2.2315,
      "step": 4240
    },
    {
      "epoch": 0.5114320096269555,
      "grad_norm": 7.182191848754883,
      "learning_rate": 0.00016368187945310997,
      "loss": 2.2752,
      "step": 4250
    },
    {
      "epoch": 0.5126353790613718,
      "grad_norm": 3.6913094520568848,
      "learning_rate": 0.00016406701328711728,
      "loss": 2.1639,
      "step": 4260
    },
    {
      "epoch": 0.5138387484957883,
      "grad_norm": 3.209007501602173,
      "learning_rate": 0.0001644521471211246,
      "loss": 2.2458,
      "step": 4270
    },
    {
      "epoch": 0.5150421179302046,
      "grad_norm": 4.694979667663574,
      "learning_rate": 0.00016483728095513192,
      "loss": 2.3178,
      "step": 4280
    },
    {
      "epoch": 0.516245487364621,
      "grad_norm": 4.524529457092285,
      "learning_rate": 0.00016522241478913923,
      "loss": 2.0244,
      "step": 4290
    },
    {
      "epoch": 0.5174488567990373,
      "grad_norm": 5.364401340484619,
      "learning_rate": 0.00016560754862314655,
      "loss": 2.3895,
      "step": 4300
    },
    {
      "epoch": 0.5186522262334536,
      "grad_norm": 4.454987049102783,
      "learning_rate": 0.00016599268245715386,
      "loss": 2.1674,
      "step": 4310
    },
    {
      "epoch": 0.51985559566787,
      "grad_norm": 3.4239635467529297,
      "learning_rate": 0.00016637781629116118,
      "loss": 2.2167,
      "step": 4320
    },
    {
      "epoch": 0.5210589651022864,
      "grad_norm": 6.292297840118408,
      "learning_rate": 0.0001667629501251685,
      "loss": 2.3381,
      "step": 4330
    },
    {
      "epoch": 0.5222623345367028,
      "grad_norm": 6.415258407592773,
      "learning_rate": 0.00016714808395917584,
      "loss": 2.422,
      "step": 4340
    },
    {
      "epoch": 0.5234657039711191,
      "grad_norm": 5.473740577697754,
      "learning_rate": 0.00016753321779318313,
      "loss": 2.0074,
      "step": 4350
    },
    {
      "epoch": 0.5246690734055355,
      "grad_norm": 3.851654052734375,
      "learning_rate": 0.00016791835162719045,
      "loss": 2.1131,
      "step": 4360
    },
    {
      "epoch": 0.5258724428399518,
      "grad_norm": 2.981456995010376,
      "learning_rate": 0.0001683034854611978,
      "loss": 2.0493,
      "step": 4370
    },
    {
      "epoch": 0.5270758122743683,
      "grad_norm": 5.323829174041748,
      "learning_rate": 0.0001686886192952051,
      "loss": 2.1481,
      "step": 4380
    },
    {
      "epoch": 0.5282791817087846,
      "grad_norm": 3.30295991897583,
      "learning_rate": 0.00016907375312921242,
      "loss": 2.2369,
      "step": 4390
    },
    {
      "epoch": 0.529482551143201,
      "grad_norm": 5.954172134399414,
      "learning_rate": 0.0001694588869632197,
      "loss": 2.188,
      "step": 4400
    },
    {
      "epoch": 0.5306859205776173,
      "grad_norm": 4.6917595863342285,
      "learning_rate": 0.00016984402079722705,
      "loss": 2.1564,
      "step": 4410
    },
    {
      "epoch": 0.5318892900120337,
      "grad_norm": 4.059054851531982,
      "learning_rate": 0.00017022915463123437,
      "loss": 2.1733,
      "step": 4420
    },
    {
      "epoch": 0.53309265944645,
      "grad_norm": 4.693612575531006,
      "learning_rate": 0.0001706142884652417,
      "loss": 2.049,
      "step": 4430
    },
    {
      "epoch": 0.5342960288808665,
      "grad_norm": 3.6918413639068604,
      "learning_rate": 0.000170999422299249,
      "loss": 2.2558,
      "step": 4440
    },
    {
      "epoch": 0.5354993983152828,
      "grad_norm": 6.020305156707764,
      "learning_rate": 0.00017138455613325632,
      "loss": 2.2933,
      "step": 4450
    },
    {
      "epoch": 0.5367027677496992,
      "grad_norm": 4.025328636169434,
      "learning_rate": 0.00017176968996726364,
      "loss": 2.2216,
      "step": 4460
    },
    {
      "epoch": 0.5379061371841155,
      "grad_norm": 3.8254079818725586,
      "learning_rate": 0.00017215482380127095,
      "loss": 2.3882,
      "step": 4470
    },
    {
      "epoch": 0.5391095066185319,
      "grad_norm": 4.954911231994629,
      "learning_rate": 0.00017253995763527827,
      "loss": 2.3088,
      "step": 4480
    },
    {
      "epoch": 0.5403128760529483,
      "grad_norm": 4.960634708404541,
      "learning_rate": 0.00017292509146928558,
      "loss": 2.2734,
      "step": 4490
    },
    {
      "epoch": 0.5415162454873647,
      "grad_norm": 5.230152130126953,
      "learning_rate": 0.0001733102253032929,
      "loss": 2.1023,
      "step": 4500
    },
    {
      "epoch": 0.542719614921781,
      "grad_norm": 4.745293140411377,
      "learning_rate": 0.00017369535913730022,
      "loss": 2.2787,
      "step": 4510
    },
    {
      "epoch": 0.5439229843561973,
      "grad_norm": 3.2070212364196777,
      "learning_rate": 0.00017408049297130753,
      "loss": 2.4571,
      "step": 4520
    },
    {
      "epoch": 0.5451263537906137,
      "grad_norm": 4.792518138885498,
      "learning_rate": 0.00017446562680531488,
      "loss": 2.3549,
      "step": 4530
    },
    {
      "epoch": 0.54632972322503,
      "grad_norm": 4.161391735076904,
      "learning_rate": 0.00017485076063932217,
      "loss": 2.2024,
      "step": 4540
    },
    {
      "epoch": 0.5475330926594465,
      "grad_norm": 6.9995012283325195,
      "learning_rate": 0.00017523589447332948,
      "loss": 2.3532,
      "step": 4550
    },
    {
      "epoch": 0.5487364620938628,
      "grad_norm": 4.922509670257568,
      "learning_rate": 0.0001756210283073368,
      "loss": 2.0588,
      "step": 4560
    },
    {
      "epoch": 0.5499398315282792,
      "grad_norm": 5.178070068359375,
      "learning_rate": 0.00017600616214134414,
      "loss": 2.3225,
      "step": 4570
    },
    {
      "epoch": 0.5511432009626955,
      "grad_norm": 4.753723621368408,
      "learning_rate": 0.00017639129597535143,
      "loss": 2.1332,
      "step": 4580
    },
    {
      "epoch": 0.5523465703971119,
      "grad_norm": 4.199524402618408,
      "learning_rate": 0.00017677642980935875,
      "loss": 2.1207,
      "step": 4590
    },
    {
      "epoch": 0.5535499398315282,
      "grad_norm": 6.0876007080078125,
      "learning_rate": 0.00017716156364336606,
      "loss": 2.1182,
      "step": 4600
    },
    {
      "epoch": 0.5547533092659447,
      "grad_norm": 6.212006568908691,
      "learning_rate": 0.0001775466974773734,
      "loss": 2.2262,
      "step": 4610
    },
    {
      "epoch": 0.555956678700361,
      "grad_norm": 3.594230890274048,
      "learning_rate": 0.00017793183131138072,
      "loss": 2.2525,
      "step": 4620
    },
    {
      "epoch": 0.5571600481347774,
      "grad_norm": 5.565987586975098,
      "learning_rate": 0.000178316965145388,
      "loss": 2.0538,
      "step": 4630
    },
    {
      "epoch": 0.5583634175691937,
      "grad_norm": 5.315800666809082,
      "learning_rate": 0.00017870209897939533,
      "loss": 1.9768,
      "step": 4640
    },
    {
      "epoch": 0.5595667870036101,
      "grad_norm": 5.953965187072754,
      "learning_rate": 0.00017908723281340267,
      "loss": 2.4237,
      "step": 4650
    },
    {
      "epoch": 0.5607701564380265,
      "grad_norm": 4.337621212005615,
      "learning_rate": 0.00017947236664741,
      "loss": 1.9684,
      "step": 4660
    },
    {
      "epoch": 0.5619735258724429,
      "grad_norm": 3.0972256660461426,
      "learning_rate": 0.0001798575004814173,
      "loss": 2.2303,
      "step": 4670
    },
    {
      "epoch": 0.5631768953068592,
      "grad_norm": 4.8460893630981445,
      "learning_rate": 0.0001802426343154246,
      "loss": 2.3611,
      "step": 4680
    },
    {
      "epoch": 0.5643802647412756,
      "grad_norm": 3.7181291580200195,
      "learning_rate": 0.00018062776814943194,
      "loss": 2.2953,
      "step": 4690
    },
    {
      "epoch": 0.5655836341756919,
      "grad_norm": 6.203132152557373,
      "learning_rate": 0.00018101290198343925,
      "loss": 2.1678,
      "step": 4700
    },
    {
      "epoch": 0.5667870036101083,
      "grad_norm": 4.8794941902160645,
      "learning_rate": 0.00018139803581744657,
      "loss": 2.2847,
      "step": 4710
    },
    {
      "epoch": 0.5679903730445247,
      "grad_norm": 3.7808096408843994,
      "learning_rate": 0.0001817831696514539,
      "loss": 2.2479,
      "step": 4720
    },
    {
      "epoch": 0.5691937424789411,
      "grad_norm": 4.418076038360596,
      "learning_rate": 0.0001821683034854612,
      "loss": 2.2857,
      "step": 4730
    },
    {
      "epoch": 0.5703971119133574,
      "grad_norm": 3.5367400646209717,
      "learning_rate": 0.00018255343731946852,
      "loss": 2.3112,
      "step": 4740
    },
    {
      "epoch": 0.5716004813477737,
      "grad_norm": 6.713764667510986,
      "learning_rate": 0.00018293857115347584,
      "loss": 2.3626,
      "step": 4750
    },
    {
      "epoch": 0.5728038507821901,
      "grad_norm": 4.047953128814697,
      "learning_rate": 0.00018332370498748318,
      "loss": 2.2012,
      "step": 4760
    },
    {
      "epoch": 0.5740072202166066,
      "grad_norm": 3.482131242752075,
      "learning_rate": 0.00018370883882149047,
      "loss": 2.2053,
      "step": 4770
    },
    {
      "epoch": 0.5752105896510229,
      "grad_norm": 4.9270339012146,
      "learning_rate": 0.00018409397265549778,
      "loss": 2.2103,
      "step": 4780
    },
    {
      "epoch": 0.5764139590854392,
      "grad_norm": 4.626224517822266,
      "learning_rate": 0.0001844791064895051,
      "loss": 2.1866,
      "step": 4790
    },
    {
      "epoch": 0.5776173285198556,
      "grad_norm": 5.537825584411621,
      "learning_rate": 0.00018486424032351244,
      "loss": 2.2666,
      "step": 4800
    },
    {
      "epoch": 0.5788206979542719,
      "grad_norm": 4.913242340087891,
      "learning_rate": 0.00018524937415751976,
      "loss": 2.08,
      "step": 4810
    },
    {
      "epoch": 0.5800240673886883,
      "grad_norm": 4.09318733215332,
      "learning_rate": 0.00018563450799152705,
      "loss": 2.2275,
      "step": 4820
    },
    {
      "epoch": 0.5812274368231047,
      "grad_norm": 5.854251861572266,
      "learning_rate": 0.00018601964182553437,
      "loss": 2.2121,
      "step": 4830
    },
    {
      "epoch": 0.5824308062575211,
      "grad_norm": 4.641722679138184,
      "learning_rate": 0.0001864047756595417,
      "loss": 2.3117,
      "step": 4840
    },
    {
      "epoch": 0.5836341756919374,
      "grad_norm": 5.240549564361572,
      "learning_rate": 0.00018678990949354903,
      "loss": 2.3387,
      "step": 4850
    },
    {
      "epoch": 0.5848375451263538,
      "grad_norm": 5.191614627838135,
      "learning_rate": 0.00018717504332755634,
      "loss": 2.2299,
      "step": 4860
    },
    {
      "epoch": 0.5860409145607701,
      "grad_norm": 3.6426289081573486,
      "learning_rate": 0.00018756017716156363,
      "loss": 2.3794,
      "step": 4870
    },
    {
      "epoch": 0.5872442839951865,
      "grad_norm": 5.039050579071045,
      "learning_rate": 0.00018794531099557097,
      "loss": 2.1374,
      "step": 4880
    },
    {
      "epoch": 0.5884476534296029,
      "grad_norm": 5.175183296203613,
      "learning_rate": 0.0001883304448295783,
      "loss": 2.1648,
      "step": 4890
    },
    {
      "epoch": 0.5896510228640193,
      "grad_norm": 6.7789387702941895,
      "learning_rate": 0.0001887155786635856,
      "loss": 2.2178,
      "step": 4900
    },
    {
      "epoch": 0.5908543922984356,
      "grad_norm": 5.478110313415527,
      "learning_rate": 0.00018910071249759292,
      "loss": 2.2999,
      "step": 4910
    },
    {
      "epoch": 0.592057761732852,
      "grad_norm": 3.7008423805236816,
      "learning_rate": 0.00018948584633160024,
      "loss": 2.1704,
      "step": 4920
    },
    {
      "epoch": 0.5932611311672683,
      "grad_norm": 5.629617214202881,
      "learning_rate": 0.00018987098016560756,
      "loss": 2.2908,
      "step": 4930
    },
    {
      "epoch": 0.5944645006016848,
      "grad_norm": 3.8162500858306885,
      "learning_rate": 0.00019025611399961487,
      "loss": 2.3308,
      "step": 4940
    },
    {
      "epoch": 0.5956678700361011,
      "grad_norm": 5.998647689819336,
      "learning_rate": 0.0001906412478336222,
      "loss": 2.2056,
      "step": 4950
    },
    {
      "epoch": 0.5968712394705175,
      "grad_norm": 4.687641620635986,
      "learning_rate": 0.0001910263816676295,
      "loss": 2.11,
      "step": 4960
    },
    {
      "epoch": 0.5980746089049338,
      "grad_norm": 3.5128839015960693,
      "learning_rate": 0.00019141151550163682,
      "loss": 2.22,
      "step": 4970
    },
    {
      "epoch": 0.5992779783393501,
      "grad_norm": 5.303665637969971,
      "learning_rate": 0.00019179664933564414,
      "loss": 2.15,
      "step": 4980
    },
    {
      "epoch": 0.6004813477737665,
      "grad_norm": 3.855205535888672,
      "learning_rate": 0.00019218178316965145,
      "loss": 2.2459,
      "step": 4990
    },
    {
      "epoch": 0.601684717208183,
      "grad_norm": 5.189699649810791,
      "learning_rate": 0.0001925669170036588,
      "loss": 2.3758,
      "step": 5000
    },
    {
      "epoch": 0.6028880866425993,
      "grad_norm": 4.4476823806762695,
      "learning_rate": 0.00019295205083766609,
      "loss": 2.0236,
      "step": 5010
    },
    {
      "epoch": 0.6040914560770156,
      "grad_norm": 4.877256870269775,
      "learning_rate": 0.0001933371846716734,
      "loss": 2.2245,
      "step": 5020
    },
    {
      "epoch": 0.605294825511432,
      "grad_norm": 6.751873016357422,
      "learning_rate": 0.00019372231850568075,
      "loss": 2.3168,
      "step": 5030
    },
    {
      "epoch": 0.6064981949458483,
      "grad_norm": 3.8965351581573486,
      "learning_rate": 0.00019410745233968806,
      "loss": 2.2746,
      "step": 5040
    },
    {
      "epoch": 0.6077015643802648,
      "grad_norm": 5.491021156311035,
      "learning_rate": 0.00019449258617369538,
      "loss": 2.3079,
      "step": 5050
    },
    {
      "epoch": 0.6089049338146811,
      "grad_norm": 3.6740262508392334,
      "learning_rate": 0.00019487772000770267,
      "loss": 2.1978,
      "step": 5060
    },
    {
      "epoch": 0.6101083032490975,
      "grad_norm": 3.504803419113159,
      "learning_rate": 0.00019526285384171,
      "loss": 2.4303,
      "step": 5070
    },
    {
      "epoch": 0.6113116726835138,
      "grad_norm": 4.8340840339660645,
      "learning_rate": 0.00019564798767571733,
      "loss": 2.1761,
      "step": 5080
    },
    {
      "epoch": 0.6125150421179302,
      "grad_norm": 4.319956302642822,
      "learning_rate": 0.00019603312150972464,
      "loss": 2.1507,
      "step": 5090
    },
    {
      "epoch": 0.6137184115523465,
      "grad_norm": 6.505810260772705,
      "learning_rate": 0.00019641825534373196,
      "loss": 2.3415,
      "step": 5100
    },
    {
      "epoch": 0.614921780986763,
      "grad_norm": 5.259067058563232,
      "learning_rate": 0.00019680338917773928,
      "loss": 2.236,
      "step": 5110
    },
    {
      "epoch": 0.6161251504211793,
      "grad_norm": 3.3339712619781494,
      "learning_rate": 0.0001971885230117466,
      "loss": 2.3776,
      "step": 5120
    },
    {
      "epoch": 0.6173285198555957,
      "grad_norm": 5.416616916656494,
      "learning_rate": 0.0001975736568457539,
      "loss": 2.0913,
      "step": 5130
    },
    {
      "epoch": 0.618531889290012,
      "grad_norm": 3.966322183609009,
      "learning_rate": 0.00019795879067976123,
      "loss": 2.1771,
      "step": 5140
    },
    {
      "epoch": 0.6197352587244284,
      "grad_norm": 6.432679176330566,
      "learning_rate": 0.00019834392451376854,
      "loss": 2.3724,
      "step": 5150
    },
    {
      "epoch": 0.6209386281588448,
      "grad_norm": 3.8340489864349365,
      "learning_rate": 0.00019872905834777586,
      "loss": 2.0687,
      "step": 5160
    },
    {
      "epoch": 0.6221419975932612,
      "grad_norm": 3.3140413761138916,
      "learning_rate": 0.00019911419218178317,
      "loss": 2.3506,
      "step": 5170
    },
    {
      "epoch": 0.6233453670276775,
      "grad_norm": 5.370321273803711,
      "learning_rate": 0.0001994993260157905,
      "loss": 2.1462,
      "step": 5180
    },
    {
      "epoch": 0.6245487364620939,
      "grad_norm": 5.8055925369262695,
      "learning_rate": 0.00019988445984979783,
      "loss": 2.4433,
      "step": 5190
    },
    {
      "epoch": 0.6257521058965102,
      "grad_norm": 5.655837059020996,
      "learning_rate": 0.00019999999996454245,
      "loss": 2.3347,
      "step": 5200
    },
    {
      "epoch": 0.6269554753309265,
      "grad_norm": 5.972043514251709,
      "learning_rate": 0.00019999999979087286,
      "loss": 2.0611,
      "step": 5210
    },
    {
      "epoch": 0.628158844765343,
      "grad_norm": 3.918379306793213,
      "learning_rate": 0.00019999999947247852,
      "loss": 2.4486,
      "step": 5220
    },
    {
      "epoch": 0.6293622141997594,
      "grad_norm": 6.398720741271973,
      "learning_rate": 0.0001999999990093595,
      "loss": 2.1424,
      "step": 5230
    },
    {
      "epoch": 0.6305655836341757,
      "grad_norm": 4.77095890045166,
      "learning_rate": 0.00019999999840151583,
      "loss": 2.3111,
      "step": 5240
    },
    {
      "epoch": 0.631768953068592,
      "grad_norm": 6.796426773071289,
      "learning_rate": 0.00019999999764894747,
      "loss": 2.4209,
      "step": 5250
    },
    {
      "epoch": 0.6329723225030084,
      "grad_norm": 7.171083450317383,
      "learning_rate": 0.00019999999675165438,
      "loss": 2.3567,
      "step": 5260
    },
    {
      "epoch": 0.6341756919374247,
      "grad_norm": 3.3638195991516113,
      "learning_rate": 0.00019999999570963664,
      "loss": 2.0822,
      "step": 5270
    },
    {
      "epoch": 0.6353790613718412,
      "grad_norm": 5.260736465454102,
      "learning_rate": 0.0001999999945228942,
      "loss": 2.2258,
      "step": 5280
    },
    {
      "epoch": 0.6365824308062575,
      "grad_norm": 3.7317612171173096,
      "learning_rate": 0.00019999999319142707,
      "loss": 2.2966,
      "step": 5290
    },
    {
      "epoch": 0.6377858002406739,
      "grad_norm": 6.7515645027160645,
      "learning_rate": 0.0001999999917152353,
      "loss": 2.3575,
      "step": 5300
    },
    {
      "epoch": 0.6389891696750902,
      "grad_norm": 6.711028575897217,
      "learning_rate": 0.00019999999009431884,
      "loss": 2.2671,
      "step": 5310
    },
    {
      "epoch": 0.6401925391095066,
      "grad_norm": 3.392085552215576,
      "learning_rate": 0.00019999998832867768,
      "loss": 2.3201,
      "step": 5320
    },
    {
      "epoch": 0.641395908543923,
      "grad_norm": 6.115386009216309,
      "learning_rate": 0.00019999998641831185,
      "loss": 2.422,
      "step": 5330
    },
    {
      "epoch": 0.6425992779783394,
      "grad_norm": 3.648564100265503,
      "learning_rate": 0.00019999998436322138,
      "loss": 2.4774,
      "step": 5340
    },
    {
      "epoch": 0.6438026474127557,
      "grad_norm": 5.814185619354248,
      "learning_rate": 0.0001999999821634062,
      "loss": 2.4827,
      "step": 5350
    },
    {
      "epoch": 0.6450060168471721,
      "grad_norm": 4.432163715362549,
      "learning_rate": 0.0001999999798188664,
      "loss": 2.0722,
      "step": 5360
    },
    {
      "epoch": 0.6462093862815884,
      "grad_norm": 4.281014919281006,
      "learning_rate": 0.00019999997732960193,
      "loss": 2.2363,
      "step": 5370
    },
    {
      "epoch": 0.6474127557160048,
      "grad_norm": 7.3086066246032715,
      "learning_rate": 0.0001999999746956128,
      "loss": 2.1118,
      "step": 5380
    },
    {
      "epoch": 0.6486161251504212,
      "grad_norm": 4.297363758087158,
      "learning_rate": 0.000199999971916899,
      "loss": 2.2896,
      "step": 5390
    },
    {
      "epoch": 0.6498194945848376,
      "grad_norm": 5.7083048820495605,
      "learning_rate": 0.00019999996899346058,
      "loss": 2.3377,
      "step": 5400
    },
    {
      "epoch": 0.6510228640192539,
      "grad_norm": 4.992364883422852,
      "learning_rate": 0.00019999996592529748,
      "loss": 2.321,
      "step": 5410
    },
    {
      "epoch": 0.6522262334536703,
      "grad_norm": 3.141991138458252,
      "learning_rate": 0.00019999996271240978,
      "loss": 2.2665,
      "step": 5420
    },
    {
      "epoch": 0.6534296028880866,
      "grad_norm": 5.258912086486816,
      "learning_rate": 0.00019999995935479742,
      "loss": 2.3438,
      "step": 5430
    },
    {
      "epoch": 0.6546329723225031,
      "grad_norm": 4.900660514831543,
      "learning_rate": 0.00019999995585246044,
      "loss": 2.0806,
      "step": 5440
    },
    {
      "epoch": 0.6558363417569194,
      "grad_norm": 6.445699691772461,
      "learning_rate": 0.00019999995220539883,
      "loss": 2.4233,
      "step": 5450
    },
    {
      "epoch": 0.6570397111913358,
      "grad_norm": 4.976849555969238,
      "learning_rate": 0.00019999994841361259,
      "loss": 2.1497,
      "step": 5460
    },
    {
      "epoch": 0.6582430806257521,
      "grad_norm": 3.247731924057007,
      "learning_rate": 0.00019999994447710175,
      "loss": 2.4414,
      "step": 5470
    },
    {
      "epoch": 0.6594464500601684,
      "grad_norm": 5.598658084869385,
      "learning_rate": 0.0001999999403958663,
      "loss": 2.3372,
      "step": 5480
    },
    {
      "epoch": 0.6606498194945848,
      "grad_norm": 4.76513671875,
      "learning_rate": 0.00019999993616990624,
      "loss": 2.2463,
      "step": 5490
    },
    {
      "epoch": 0.6618531889290012,
      "grad_norm": 6.446969509124756,
      "learning_rate": 0.0001999999317992216,
      "loss": 2.3092,
      "step": 5500
    },
    {
      "epoch": 0.6630565583634176,
      "grad_norm": 4.556867599487305,
      "learning_rate": 0.00019999992728381236,
      "loss": 2.085,
      "step": 5510
    },
    {
      "epoch": 0.6642599277978339,
      "grad_norm": 3.7941901683807373,
      "learning_rate": 0.00019999992262367851,
      "loss": 2.3755,
      "step": 5520
    },
    {
      "epoch": 0.6654632972322503,
      "grad_norm": 4.857911586761475,
      "learning_rate": 0.0001999999178188201,
      "loss": 2.2633,
      "step": 5530
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 3.6080713272094727,
      "learning_rate": 0.00019999991286923715,
      "loss": 2.0844,
      "step": 5540
    },
    {
      "epoch": 0.6678700361010831,
      "grad_norm": 6.234077453613281,
      "learning_rate": 0.0001999999077749296,
      "loss": 2.0721,
      "step": 5550
    },
    {
      "epoch": 0.6690734055354994,
      "grad_norm": 4.698109149932861,
      "learning_rate": 0.00019999990253589752,
      "loss": 2.2047,
      "step": 5560
    },
    {
      "epoch": 0.6702767749699158,
      "grad_norm": 3.984668731689453,
      "learning_rate": 0.00019999989715214087,
      "loss": 2.1264,
      "step": 5570
    },
    {
      "epoch": 0.6714801444043321,
      "grad_norm": 5.2652997970581055,
      "learning_rate": 0.0001999998916236597,
      "loss": 2.2808,
      "step": 5580
    },
    {
      "epoch": 0.6726835138387485,
      "grad_norm": 3.186023712158203,
      "learning_rate": 0.00019999988595045397,
      "loss": 2.305,
      "step": 5590
    },
    {
      "epoch": 0.6738868832731648,
      "grad_norm": 6.204652786254883,
      "learning_rate": 0.00019999988013252374,
      "loss": 2.2583,
      "step": 5600
    },
    {
      "epoch": 0.6750902527075813,
      "grad_norm": 5.521336555480957,
      "learning_rate": 0.00019999987416986897,
      "loss": 2.2934,
      "step": 5610
    },
    {
      "epoch": 0.6762936221419976,
      "grad_norm": 3.410998582839966,
      "learning_rate": 0.00019999986806248972,
      "loss": 2.1635,
      "step": 5620
    },
    {
      "epoch": 0.677496991576414,
      "grad_norm": 5.09378719329834,
      "learning_rate": 0.00019999986181038593,
      "loss": 2.1468,
      "step": 5630
    },
    {
      "epoch": 0.6787003610108303,
      "grad_norm": 3.893676280975342,
      "learning_rate": 0.00019999985541355772,
      "loss": 2.1738,
      "step": 5640
    },
    {
      "epoch": 0.6799037304452467,
      "grad_norm": 5.081244468688965,
      "learning_rate": 0.000199999848872005,
      "loss": 2.1461,
      "step": 5650
    },
    {
      "epoch": 0.681107099879663,
      "grad_norm": 4.685601711273193,
      "learning_rate": 0.00019999984218572778,
      "loss": 2.179,
      "step": 5660
    },
    {
      "epoch": 0.6823104693140795,
      "grad_norm": 2.979675054550171,
      "learning_rate": 0.00019999983535472614,
      "loss": 2.3638,
      "step": 5670
    },
    {
      "epoch": 0.6835138387484958,
      "grad_norm": 7.460003852844238,
      "learning_rate": 0.00019999982837900002,
      "loss": 2.2055,
      "step": 5680
    },
    {
      "epoch": 0.6847172081829122,
      "grad_norm": 4.4045538902282715,
      "learning_rate": 0.00019999982125854947,
      "loss": 2.0693,
      "step": 5690
    },
    {
      "epoch": 0.6859205776173285,
      "grad_norm": 5.677947044372559,
      "learning_rate": 0.0001999998139933745,
      "loss": 2.4137,
      "step": 5700
    },
    {
      "epoch": 0.6871239470517448,
      "grad_norm": 4.1160888671875,
      "learning_rate": 0.00019999980658347508,
      "loss": 2.2255,
      "step": 5710
    },
    {
      "epoch": 0.6883273164861613,
      "grad_norm": 4.297820091247559,
      "learning_rate": 0.00019999979902885127,
      "loss": 2.1269,
      "step": 5720
    },
    {
      "epoch": 0.6895306859205776,
      "grad_norm": 4.658998012542725,
      "learning_rate": 0.00019999979132950306,
      "loss": 2.3559,
      "step": 5730
    },
    {
      "epoch": 0.690734055354994,
      "grad_norm": 3.612006425857544,
      "learning_rate": 0.00019999978348543044,
      "loss": 2.2368,
      "step": 5740
    },
    {
      "epoch": 0.6919374247894103,
      "grad_norm": 5.840548515319824,
      "learning_rate": 0.00019999977549663348,
      "loss": 2.1801,
      "step": 5750
    },
    {
      "epoch": 0.6931407942238267,
      "grad_norm": 4.654530048370361,
      "learning_rate": 0.00019999976736311215,
      "loss": 2.2934,
      "step": 5760
    },
    {
      "epoch": 0.694344163658243,
      "grad_norm": 3.498783826828003,
      "learning_rate": 0.00019999975908486646,
      "loss": 2.0995,
      "step": 5770
    },
    {
      "epoch": 0.6955475330926595,
      "grad_norm": 5.616212368011475,
      "learning_rate": 0.00019999975066189643,
      "loss": 2.1991,
      "step": 5780
    },
    {
      "epoch": 0.6967509025270758,
      "grad_norm": 4.105483531951904,
      "learning_rate": 0.00019999974209420208,
      "loss": 2.3082,
      "step": 5790
    },
    {
      "epoch": 0.6979542719614922,
      "grad_norm": 6.668478012084961,
      "learning_rate": 0.00019999973338178338,
      "loss": 2.3708,
      "step": 5800
    },
    {
      "epoch": 0.6991576413959085,
      "grad_norm": 5.358325481414795,
      "learning_rate": 0.00019999972452464042,
      "loss": 2.2654,
      "step": 5810
    },
    {
      "epoch": 0.7003610108303249,
      "grad_norm": 3.947488784790039,
      "learning_rate": 0.00019999971552277313,
      "loss": 2.3865,
      "step": 5820
    },
    {
      "epoch": 0.7015643802647413,
      "grad_norm": 5.846844673156738,
      "learning_rate": 0.0001999997063761816,
      "loss": 2.2507,
      "step": 5830
    },
    {
      "epoch": 0.7027677496991577,
      "grad_norm": 3.6681337356567383,
      "learning_rate": 0.0001999996970848658,
      "loss": 2.1866,
      "step": 5840
    },
    {
      "epoch": 0.703971119133574,
      "grad_norm": 7.481490135192871,
      "learning_rate": 0.00019999968764882573,
      "loss": 2.29,
      "step": 5850
    },
    {
      "epoch": 0.7051744885679904,
      "grad_norm": 4.638186931610107,
      "learning_rate": 0.00019999967806806144,
      "loss": 2.1477,
      "step": 5860
    },
    {
      "epoch": 0.7063778580024067,
      "grad_norm": 3.310225486755371,
      "learning_rate": 0.0001999996683425729,
      "loss": 2.228,
      "step": 5870
    },
    {
      "epoch": 0.7075812274368231,
      "grad_norm": 5.039708614349365,
      "learning_rate": 0.0001999996584723602,
      "loss": 2.1469,
      "step": 5880
    },
    {
      "epoch": 0.7087845968712395,
      "grad_norm": 3.915811061859131,
      "learning_rate": 0.00019999964845742328,
      "loss": 2.2734,
      "step": 5890
    },
    {
      "epoch": 0.7099879663056559,
      "grad_norm": 6.468316555023193,
      "learning_rate": 0.00019999963829776216,
      "loss": 2.1917,
      "step": 5900
    },
    {
      "epoch": 0.7111913357400722,
      "grad_norm": 4.221285820007324,
      "learning_rate": 0.00019999962799337689,
      "loss": 2.1898,
      "step": 5910
    },
    {
      "epoch": 0.7123947051744886,
      "grad_norm": 3.1073007583618164,
      "learning_rate": 0.00019999961754426745,
      "loss": 2.1003,
      "step": 5920
    },
    {
      "epoch": 0.7135980746089049,
      "grad_norm": 5.455918788909912,
      "learning_rate": 0.00019999960695043392,
      "loss": 2.3484,
      "step": 5930
    },
    {
      "epoch": 0.7148014440433214,
      "grad_norm": 4.238619327545166,
      "learning_rate": 0.00019999959621187622,
      "loss": 2.1655,
      "step": 5940
    },
    {
      "epoch": 0.7160048134777377,
      "grad_norm": 5.345495700836182,
      "learning_rate": 0.00019999958532859445,
      "loss": 2.3268,
      "step": 5950
    },
    {
      "epoch": 0.717208182912154,
      "grad_norm": 4.478791236877441,
      "learning_rate": 0.00019999957430058855,
      "loss": 2.1228,
      "step": 5960
    },
    {
      "epoch": 0.7184115523465704,
      "grad_norm": 5.313588619232178,
      "learning_rate": 0.0001999995631278586,
      "loss": 2.1756,
      "step": 5970
    },
    {
      "epoch": 0.7196149217809867,
      "grad_norm": 5.476967811584473,
      "learning_rate": 0.00019999955181040463,
      "loss": 2.2398,
      "step": 5980
    },
    {
      "epoch": 0.7208182912154031,
      "grad_norm": 4.409827709197998,
      "learning_rate": 0.00019999954034822658,
      "loss": 2.2433,
      "step": 5990
    },
    {
      "epoch": 0.7220216606498195,
      "grad_norm": 6.954128265380859,
      "learning_rate": 0.0001999995287413245,
      "loss": 2.2805,
      "step": 6000
    },
    {
      "epoch": 0.7232250300842359,
      "grad_norm": 4.963774681091309,
      "learning_rate": 0.00019999951698969844,
      "loss": 2.0606,
      "step": 6010
    },
    {
      "epoch": 0.7244283995186522,
      "grad_norm": 3.1691906452178955,
      "learning_rate": 0.00019999950509334838,
      "loss": 2.2873,
      "step": 6020
    },
    {
      "epoch": 0.7256317689530686,
      "grad_norm": 5.281974792480469,
      "learning_rate": 0.00019999949305227435,
      "loss": 2.4136,
      "step": 6030
    },
    {
      "epoch": 0.7268351383874849,
      "grad_norm": 4.72628116607666,
      "learning_rate": 0.00019999948086647632,
      "loss": 2.264,
      "step": 6040
    },
    {
      "epoch": 0.7280385078219013,
      "grad_norm": 5.913517951965332,
      "learning_rate": 0.00019999946853595439,
      "loss": 2.2188,
      "step": 6050
    },
    {
      "epoch": 0.7292418772563177,
      "grad_norm": 4.688936233520508,
      "learning_rate": 0.00019999945606070856,
      "loss": 2.0428,
      "step": 6060
    },
    {
      "epoch": 0.7304452466907341,
      "grad_norm": 3.245565414428711,
      "learning_rate": 0.0001999994434407388,
      "loss": 2.3174,
      "step": 6070
    },
    {
      "epoch": 0.7316486161251504,
      "grad_norm": 4.206037998199463,
      "learning_rate": 0.0001999994306760452,
      "loss": 2.0825,
      "step": 6080
    },
    {
      "epoch": 0.7328519855595668,
      "grad_norm": 3.948371648788452,
      "learning_rate": 0.00019999941776662768,
      "loss": 2.2976,
      "step": 6090
    },
    {
      "epoch": 0.7340553549939831,
      "grad_norm": 6.331472873687744,
      "learning_rate": 0.00019999940471248634,
      "loss": 2.3223,
      "step": 6100
    },
    {
      "epoch": 0.7352587244283996,
      "grad_norm": 4.219931125640869,
      "learning_rate": 0.00019999939151362116,
      "loss": 2.1751,
      "step": 6110
    },
    {
      "epoch": 0.7364620938628159,
      "grad_norm": 3.756378173828125,
      "learning_rate": 0.00019999937817003215,
      "loss": 2.216,
      "step": 6120
    },
    {
      "epoch": 0.7376654632972323,
      "grad_norm": 4.940024375915527,
      "learning_rate": 0.0001999993646817194,
      "loss": 2.3813,
      "step": 6130
    },
    {
      "epoch": 0.7388688327316486,
      "grad_norm": 4.085123062133789,
      "learning_rate": 0.00019999935104868285,
      "loss": 2.3866,
      "step": 6140
    },
    {
      "epoch": 0.740072202166065,
      "grad_norm": 7.11965274810791,
      "learning_rate": 0.00019999933727092256,
      "loss": 2.6845,
      "step": 6150
    },
    {
      "epoch": 0.7412755716004813,
      "grad_norm": 5.067838668823242,
      "learning_rate": 0.0001999993233484385,
      "loss": 2.228,
      "step": 6160
    },
    {
      "epoch": 0.7424789410348978,
      "grad_norm": 4.1947808265686035,
      "learning_rate": 0.0001999993092812308,
      "loss": 2.297,
      "step": 6170
    },
    {
      "epoch": 0.7436823104693141,
      "grad_norm": 5.650806903839111,
      "learning_rate": 0.00019999929506929936,
      "loss": 2.324,
      "step": 6180
    },
    {
      "epoch": 0.7448856799037304,
      "grad_norm": 4.293774127960205,
      "learning_rate": 0.00019999928071264429,
      "loss": 2.1975,
      "step": 6190
    },
    {
      "epoch": 0.7460890493381468,
      "grad_norm": 6.2978315353393555,
      "learning_rate": 0.00019999926621126554,
      "loss": 2.3262,
      "step": 6200
    },
    {
      "epoch": 0.7472924187725631,
      "grad_norm": 5.376358985900879,
      "learning_rate": 0.00019999925156516317,
      "loss": 1.9778,
      "step": 6210
    },
    {
      "epoch": 0.7484957882069796,
      "grad_norm": 3.434368848800659,
      "learning_rate": 0.00019999923677433722,
      "loss": 2.3768,
      "step": 6220
    },
    {
      "epoch": 0.7496991576413959,
      "grad_norm": 4.872026443481445,
      "learning_rate": 0.00019999922183878762,
      "loss": 2.1686,
      "step": 6230
    },
    {
      "epoch": 0.7509025270758123,
      "grad_norm": 4.505314350128174,
      "learning_rate": 0.00019999920675851452,
      "loss": 2.2798,
      "step": 6240
    },
    {
      "epoch": 0.7521058965102286,
      "grad_norm": 7.478349208831787,
      "learning_rate": 0.00019999919153351785,
      "loss": 2.1283,
      "step": 6250
    },
    {
      "epoch": 0.753309265944645,
      "grad_norm": 5.635042667388916,
      "learning_rate": 0.00019999917616379768,
      "loss": 2.1289,
      "step": 6260
    },
    {
      "epoch": 0.7545126353790613,
      "grad_norm": 3.7895894050598145,
      "learning_rate": 0.000199999160649354,
      "loss": 2.206,
      "step": 6270
    },
    {
      "epoch": 0.7557160048134778,
      "grad_norm": 4.9072065353393555,
      "learning_rate": 0.00019999914499018684,
      "loss": 2.5128,
      "step": 6280
    },
    {
      "epoch": 0.7569193742478941,
      "grad_norm": 4.7990899085998535,
      "learning_rate": 0.00019999912918629623,
      "loss": 2.2511,
      "step": 6290
    },
    {
      "epoch": 0.7581227436823105,
      "grad_norm": 6.253719329833984,
      "learning_rate": 0.00019999911323768222,
      "loss": 2.3181,
      "step": 6300
    },
    {
      "epoch": 0.7593261131167268,
      "grad_norm": 4.957713603973389,
      "learning_rate": 0.00019999909714434475,
      "loss": 2.1288,
      "step": 6310
    },
    {
      "epoch": 0.7605294825511432,
      "grad_norm": 4.218201637268066,
      "learning_rate": 0.00019999908090628394,
      "loss": 2.1878,
      "step": 6320
    },
    {
      "epoch": 0.7617328519855595,
      "grad_norm": 5.5473432540893555,
      "learning_rate": 0.00019999906452349974,
      "loss": 2.0831,
      "step": 6330
    },
    {
      "epoch": 0.762936221419976,
      "grad_norm": 4.122081279754639,
      "learning_rate": 0.00019999904799599224,
      "loss": 2.2415,
      "step": 6340
    },
    {
      "epoch": 0.7641395908543923,
      "grad_norm": 5.6238555908203125,
      "learning_rate": 0.00019999903132376142,
      "loss": 2.1524,
      "step": 6350
    },
    {
      "epoch": 0.7653429602888087,
      "grad_norm": 4.484196662902832,
      "learning_rate": 0.00019999901450680732,
      "loss": 1.9236,
      "step": 6360
    },
    {
      "epoch": 0.766546329723225,
      "grad_norm": 3.451127290725708,
      "learning_rate": 0.00019999899754512992,
      "loss": 2.1563,
      "step": 6370
    },
    {
      "epoch": 0.7677496991576414,
      "grad_norm": 4.9223408699035645,
      "learning_rate": 0.0001999989804387293,
      "loss": 2.3576,
      "step": 6380
    },
    {
      "epoch": 0.7689530685920578,
      "grad_norm": 5.084750175476074,
      "learning_rate": 0.0001999989631876055,
      "loss": 2.1415,
      "step": 6390
    },
    {
      "epoch": 0.7701564380264742,
      "grad_norm": 6.764801979064941,
      "learning_rate": 0.00019999894579175847,
      "loss": 2.262,
      "step": 6400
    },
    {
      "epoch": 0.7713598074608905,
      "grad_norm": 5.669933795928955,
      "learning_rate": 0.00019999892825118827,
      "loss": 2.0838,
      "step": 6410
    },
    {
      "epoch": 0.7725631768953068,
      "grad_norm": 3.124565362930298,
      "learning_rate": 0.00019999891056589496,
      "loss": 2.2526,
      "step": 6420
    },
    {
      "epoch": 0.7737665463297232,
      "grad_norm": 5.475543022155762,
      "learning_rate": 0.0001999988927358785,
      "loss": 2.192,
      "step": 6430
    },
    {
      "epoch": 0.7749699157641395,
      "grad_norm": 4.148515701293945,
      "learning_rate": 0.000199998874761139,
      "loss": 2.2276,
      "step": 6440
    },
    {
      "epoch": 0.776173285198556,
      "grad_norm": 5.512551784515381,
      "learning_rate": 0.00019999885664167644,
      "loss": 2.1467,
      "step": 6450
    },
    {
      "epoch": 0.7773766546329723,
      "grad_norm": 6.695659160614014,
      "learning_rate": 0.0001999988383774908,
      "loss": 2.007,
      "step": 6460
    },
    {
      "epoch": 0.7785800240673887,
      "grad_norm": 3.3527722358703613,
      "learning_rate": 0.0001999988199685822,
      "loss": 2.0665,
      "step": 6470
    },
    {
      "epoch": 0.779783393501805,
      "grad_norm": 4.57830286026001,
      "learning_rate": 0.00019999880141495057,
      "loss": 2.1413,
      "step": 6480
    },
    {
      "epoch": 0.7809867629362214,
      "grad_norm": 4.720484733581543,
      "learning_rate": 0.00019999878271659602,
      "loss": 2.275,
      "step": 6490
    },
    {
      "epoch": 0.7821901323706378,
      "grad_norm": 6.585578441619873,
      "learning_rate": 0.00019999876387351853,
      "loss": 2.3317,
      "step": 6500
    },
    {
      "epoch": 0.7833935018050542,
      "grad_norm": 4.008000373840332,
      "learning_rate": 0.00019999874488571813,
      "loss": 2.0169,
      "step": 6510
    },
    {
      "epoch": 0.7845968712394705,
      "grad_norm": 3.417729377746582,
      "learning_rate": 0.0001999987257531949,
      "loss": 2.3291,
      "step": 6520
    },
    {
      "epoch": 0.7858002406738869,
      "grad_norm": 5.9073710441589355,
      "learning_rate": 0.0001999987064759488,
      "loss": 2.368,
      "step": 6530
    },
    {
      "epoch": 0.7870036101083032,
      "grad_norm": 3.870997190475464,
      "learning_rate": 0.00019999868705397988,
      "loss": 1.9916,
      "step": 6540
    },
    {
      "epoch": 0.7882069795427196,
      "grad_norm": 5.63076639175415,
      "learning_rate": 0.00019999866748728816,
      "loss": 2.2776,
      "step": 6550
    },
    {
      "epoch": 0.789410348977136,
      "grad_norm": 4.0659613609313965,
      "learning_rate": 0.0001999986477758737,
      "loss": 2.1265,
      "step": 6560
    },
    {
      "epoch": 0.7906137184115524,
      "grad_norm": 4.5430145263671875,
      "learning_rate": 0.00019999862791973648,
      "loss": 2.353,
      "step": 6570
    },
    {
      "epoch": 0.7918170878459687,
      "grad_norm": 6.095127582550049,
      "learning_rate": 0.00019999860791887657,
      "loss": 2.1899,
      "step": 6580
    },
    {
      "epoch": 0.7930204572803851,
      "grad_norm": 4.152281284332275,
      "learning_rate": 0.000199998587773294,
      "loss": 2.2989,
      "step": 6590
    },
    {
      "epoch": 0.7942238267148014,
      "grad_norm": 5.399205684661865,
      "learning_rate": 0.00019999856748298877,
      "loss": 2.436,
      "step": 6600
    },
    {
      "epoch": 0.7954271961492179,
      "grad_norm": 4.7240095138549805,
      "learning_rate": 0.00019999854704796092,
      "loss": 2.2057,
      "step": 6610
    },
    {
      "epoch": 0.7966305655836342,
      "grad_norm": 3.8349926471710205,
      "learning_rate": 0.0001999985264682105,
      "loss": 2.1351,
      "step": 6620
    },
    {
      "epoch": 0.7978339350180506,
      "grad_norm": 5.826876163482666,
      "learning_rate": 0.0001999985057437375,
      "loss": 2.3605,
      "step": 6630
    },
    {
      "epoch": 0.7990373044524669,
      "grad_norm": 3.7913665771484375,
      "learning_rate": 0.00019999848487454197,
      "loss": 2.1542,
      "step": 6640
    },
    {
      "epoch": 0.8002406738868832,
      "grad_norm": 5.375580787658691,
      "learning_rate": 0.000199998463860624,
      "loss": 2.0194,
      "step": 6650
    },
    {
      "epoch": 0.8014440433212996,
      "grad_norm": 5.2479166984558105,
      "learning_rate": 0.0001999984427019835,
      "loss": 2.1781,
      "step": 6660
    },
    {
      "epoch": 0.802647412755716,
      "grad_norm": 4.459693908691406,
      "learning_rate": 0.00019999842139862062,
      "loss": 2.3453,
      "step": 6670
    },
    {
      "epoch": 0.8038507821901324,
      "grad_norm": 5.684396266937256,
      "learning_rate": 0.0001999983999505353,
      "loss": 2.3262,
      "step": 6680
    },
    {
      "epoch": 0.8050541516245487,
      "grad_norm": 4.626715660095215,
      "learning_rate": 0.00019999837835772759,
      "loss": 2.2126,
      "step": 6690
    },
    {
      "epoch": 0.8062575210589651,
      "grad_norm": 6.133049964904785,
      "learning_rate": 0.00019999835662019757,
      "loss": 2.2424,
      "step": 6700
    },
    {
      "epoch": 0.8074608904933814,
      "grad_norm": 4.258509159088135,
      "learning_rate": 0.0001999983347379452,
      "loss": 1.99,
      "step": 6710
    },
    {
      "epoch": 0.8086642599277978,
      "grad_norm": 3.301567554473877,
      "learning_rate": 0.0001999983127109706,
      "loss": 2.1121,
      "step": 6720
    },
    {
      "epoch": 0.8098676293622142,
      "grad_norm": 5.264902591705322,
      "learning_rate": 0.00019999829053927374,
      "loss": 2.1922,
      "step": 6730
    },
    {
      "epoch": 0.8110709987966306,
      "grad_norm": 3.8825745582580566,
      "learning_rate": 0.00019999826822285466,
      "loss": 2.1548,
      "step": 6740
    },
    {
      "epoch": 0.8122743682310469,
      "grad_norm": 5.56144905090332,
      "learning_rate": 0.00019999824576171337,
      "loss": 2.3099,
      "step": 6750
    },
    {
      "epoch": 0.8134777376654633,
      "grad_norm": 4.648536205291748,
      "learning_rate": 0.00019999822315584998,
      "loss": 2.091,
      "step": 6760
    },
    {
      "epoch": 0.8146811070998796,
      "grad_norm": 3.6888937950134277,
      "learning_rate": 0.00019999820040526444,
      "loss": 2.1071,
      "step": 6770
    },
    {
      "epoch": 0.8158844765342961,
      "grad_norm": 4.9940314292907715,
      "learning_rate": 0.00019999817750995683,
      "loss": 2.1822,
      "step": 6780
    },
    {
      "epoch": 0.8170878459687124,
      "grad_norm": 4.372590065002441,
      "learning_rate": 0.00019999815446992717,
      "loss": 2.2962,
      "step": 6790
    },
    {
      "epoch": 0.8182912154031288,
      "grad_norm": 5.009139060974121,
      "learning_rate": 0.00019999813128517547,
      "loss": 2.3153,
      "step": 6800
    },
    {
      "epoch": 0.8194945848375451,
      "grad_norm": 4.939500331878662,
      "learning_rate": 0.00019999810795570178,
      "loss": 2.2014,
      "step": 6810
    },
    {
      "epoch": 0.8206979542719615,
      "grad_norm": 4.308764457702637,
      "learning_rate": 0.00019999808448150616,
      "loss": 2.1361,
      "step": 6820
    },
    {
      "epoch": 0.8219013237063778,
      "grad_norm": 6.462900161743164,
      "learning_rate": 0.0001999980608625886,
      "loss": 2.1952,
      "step": 6830
    },
    {
      "epoch": 0.8231046931407943,
      "grad_norm": 5.643688201904297,
      "learning_rate": 0.0001999980370989492,
      "loss": 2.0308,
      "step": 6840
    },
    {
      "epoch": 0.8243080625752106,
      "grad_norm": 6.7328104972839355,
      "learning_rate": 0.00019999801319058792,
      "loss": 2.11,
      "step": 6850
    },
    {
      "epoch": 0.825511432009627,
      "grad_norm": 5.818809986114502,
      "learning_rate": 0.00019999798913750483,
      "loss": 1.944,
      "step": 6860
    },
    {
      "epoch": 0.8267148014440433,
      "grad_norm": 3.0272462368011475,
      "learning_rate": 0.00019999796493969996,
      "loss": 2.1284,
      "step": 6870
    },
    {
      "epoch": 0.8279181708784596,
      "grad_norm": 6.05486536026001,
      "learning_rate": 0.00019999794059717335,
      "loss": 2.267,
      "step": 6880
    },
    {
      "epoch": 0.8291215403128761,
      "grad_norm": 3.904543161392212,
      "learning_rate": 0.00019999791610992505,
      "loss": 2.2204,
      "step": 6890
    },
    {
      "epoch": 0.8303249097472925,
      "grad_norm": 5.603869915008545,
      "learning_rate": 0.00019999789147795506,
      "loss": 2.2056,
      "step": 6900
    },
    {
      "epoch": 0.8315282791817088,
      "grad_norm": 5.127676963806152,
      "learning_rate": 0.00019999786670126343,
      "loss": 2.2686,
      "step": 6910
    },
    {
      "epoch": 0.8327316486161251,
      "grad_norm": 3.6282613277435303,
      "learning_rate": 0.0001999978417798502,
      "loss": 2.2868,
      "step": 6920
    },
    {
      "epoch": 0.8339350180505415,
      "grad_norm": 5.270694732666016,
      "learning_rate": 0.0001999978167137154,
      "loss": 2.2679,
      "step": 6930
    },
    {
      "epoch": 0.8351383874849578,
      "grad_norm": 3.836949110031128,
      "learning_rate": 0.00019999779150285907,
      "loss": 2.2726,
      "step": 6940
    },
    {
      "epoch": 0.8363417569193743,
      "grad_norm": 5.457431793212891,
      "learning_rate": 0.00019999776614728124,
      "loss": 2.2211,
      "step": 6950
    },
    {
      "epoch": 0.8375451263537906,
      "grad_norm": 5.116433620452881,
      "learning_rate": 0.00019999774064698197,
      "loss": 2.2302,
      "step": 6960
    },
    {
      "epoch": 0.838748495788207,
      "grad_norm": 3.554002285003662,
      "learning_rate": 0.00019999771500196127,
      "loss": 2.4411,
      "step": 6970
    },
    {
      "epoch": 0.8399518652226233,
      "grad_norm": 6.70152473449707,
      "learning_rate": 0.00019999768921221919,
      "loss": 2.1905,
      "step": 6980
    },
    {
      "epoch": 0.8411552346570397,
      "grad_norm": 4.895505428314209,
      "learning_rate": 0.0001999976632777558,
      "loss": 2.4138,
      "step": 6990
    },
    {
      "epoch": 0.8423586040914561,
      "grad_norm": 6.639925003051758,
      "learning_rate": 0.00019999763719857105,
      "loss": 2.3516,
      "step": 7000
    },
    {
      "epoch": 0.8435619735258725,
      "grad_norm": 4.094420433044434,
      "learning_rate": 0.00019999761097466508,
      "loss": 2.1786,
      "step": 7010
    },
    {
      "epoch": 0.8447653429602888,
      "grad_norm": 4.961806774139404,
      "learning_rate": 0.00019999758460603783,
      "loss": 2.2067,
      "step": 7020
    },
    {
      "epoch": 0.8459687123947052,
      "grad_norm": 4.95901346206665,
      "learning_rate": 0.0001999975580926894,
      "loss": 2.1451,
      "step": 7030
    },
    {
      "epoch": 0.8471720818291215,
      "grad_norm": 4.088491439819336,
      "learning_rate": 0.00019999753143461982,
      "loss": 2.1258,
      "step": 7040
    },
    {
      "epoch": 0.8483754512635379,
      "grad_norm": 5.4671525955200195,
      "learning_rate": 0.00019999750463182912,
      "loss": 2.2137,
      "step": 7050
    },
    {
      "epoch": 0.8495788206979543,
      "grad_norm": 4.36328649520874,
      "learning_rate": 0.00019999747768431736,
      "loss": 2.0348,
      "step": 7060
    },
    {
      "epoch": 0.8507821901323707,
      "grad_norm": 3.508300542831421,
      "learning_rate": 0.00019999745059208455,
      "loss": 2.0429,
      "step": 7070
    },
    {
      "epoch": 0.851985559566787,
      "grad_norm": 5.880530834197998,
      "learning_rate": 0.00019999742335513073,
      "loss": 2.1445,
      "step": 7080
    },
    {
      "epoch": 0.8531889290012034,
      "grad_norm": 4.054240703582764,
      "learning_rate": 0.00019999739597345597,
      "loss": 2.1468,
      "step": 7090
    },
    {
      "epoch": 0.8543922984356197,
      "grad_norm": 7.524439334869385,
      "learning_rate": 0.00019999736844706029,
      "loss": 2.3211,
      "step": 7100
    },
    {
      "epoch": 0.855595667870036,
      "grad_norm": 4.426852703094482,
      "learning_rate": 0.0001999973407759437,
      "loss": 2.0554,
      "step": 7110
    },
    {
      "epoch": 0.8567990373044525,
      "grad_norm": 3.4823224544525146,
      "learning_rate": 0.00019999731296010628,
      "loss": 2.2347,
      "step": 7120
    },
    {
      "epoch": 0.8580024067388689,
      "grad_norm": 5.5711894035339355,
      "learning_rate": 0.0001999972849995481,
      "loss": 2.1123,
      "step": 7130
    },
    {
      "epoch": 0.8592057761732852,
      "grad_norm": 3.9582128524780273,
      "learning_rate": 0.00019999725689426908,
      "loss": 2.0099,
      "step": 7140
    },
    {
      "epoch": 0.8604091456077015,
      "grad_norm": 5.960570335388184,
      "learning_rate": 0.00019999722864426941,
      "loss": 2.2012,
      "step": 7150
    },
    {
      "epoch": 0.8616125150421179,
      "grad_norm": 5.0911126136779785,
      "learning_rate": 0.00019999720024954903,
      "loss": 2.112,
      "step": 7160
    },
    {
      "epoch": 0.8628158844765343,
      "grad_norm": 4.357975482940674,
      "learning_rate": 0.00019999717171010804,
      "loss": 2.2499,
      "step": 7170
    },
    {
      "epoch": 0.8640192539109507,
      "grad_norm": 5.801352500915527,
      "learning_rate": 0.0001999971430259464,
      "loss": 2.2858,
      "step": 7180
    },
    {
      "epoch": 0.865222623345367,
      "grad_norm": 4.905974388122559,
      "learning_rate": 0.00019999711419706426,
      "loss": 2.1957,
      "step": 7190
    },
    {
      "epoch": 0.8664259927797834,
      "grad_norm": 6.450088977813721,
      "learning_rate": 0.00019999708522346158,
      "loss": 2.3496,
      "step": 7200
    },
    {
      "epoch": 0.8676293622141997,
      "grad_norm": 5.748147487640381,
      "learning_rate": 0.00019999705610513843,
      "loss": 2.2512,
      "step": 7210
    },
    {
      "epoch": 0.8688327316486161,
      "grad_norm": 3.711469888687134,
      "learning_rate": 0.00019999702684209486,
      "loss": 2.2549,
      "step": 7220
    },
    {
      "epoch": 0.8700361010830325,
      "grad_norm": 6.755938529968262,
      "learning_rate": 0.0001999969974343309,
      "loss": 2.4081,
      "step": 7230
    },
    {
      "epoch": 0.8712394705174489,
      "grad_norm": 5.398969650268555,
      "learning_rate": 0.0001999969678818466,
      "loss": 2.2644,
      "step": 7240
    },
    {
      "epoch": 0.8724428399518652,
      "grad_norm": 6.106263160705566,
      "learning_rate": 0.000199996938184642,
      "loss": 2.387,
      "step": 7250
    },
    {
      "epoch": 0.8736462093862816,
      "grad_norm": 4.6583709716796875,
      "learning_rate": 0.00019999690834271715,
      "loss": 2.0932,
      "step": 7260
    },
    {
      "epoch": 0.8748495788206979,
      "grad_norm": 2.823810338973999,
      "learning_rate": 0.0001999968783560721,
      "loss": 2.1311,
      "step": 7270
    },
    {
      "epoch": 0.8760529482551144,
      "grad_norm": 5.98362398147583,
      "learning_rate": 0.0001999968482247068,
      "loss": 2.1842,
      "step": 7280
    },
    {
      "epoch": 0.8772563176895307,
      "grad_norm": 3.847651720046997,
      "learning_rate": 0.00019999681794862143,
      "loss": 2.2082,
      "step": 7290
    },
    {
      "epoch": 0.8784596871239471,
      "grad_norm": 8.450338363647461,
      "learning_rate": 0.00019999678752781602,
      "loss": 2.2407,
      "step": 7300
    },
    {
      "epoch": 0.8796630565583634,
      "grad_norm": 6.108689785003662,
      "learning_rate": 0.0001999967569622905,
      "loss": 2.1711,
      "step": 7310
    },
    {
      "epoch": 0.8808664259927798,
      "grad_norm": 3.1826913356781006,
      "learning_rate": 0.000199996726252045,
      "loss": 2.3579,
      "step": 7320
    },
    {
      "epoch": 0.8820697954271961,
      "grad_norm": 4.625455856323242,
      "learning_rate": 0.00019999669539707955,
      "loss": 2.0504,
      "step": 7330
    },
    {
      "epoch": 0.8832731648616126,
      "grad_norm": 5.346113681793213,
      "learning_rate": 0.0001999966643973942,
      "loss": 2.075,
      "step": 7340
    },
    {
      "epoch": 0.8844765342960289,
      "grad_norm": 5.861973285675049,
      "learning_rate": 0.00019999663325298897,
      "loss": 2.0196,
      "step": 7350
    },
    {
      "epoch": 0.8856799037304453,
      "grad_norm": 4.9203972816467285,
      "learning_rate": 0.00019999660196386396,
      "loss": 2.0732,
      "step": 7360
    },
    {
      "epoch": 0.8868832731648616,
      "grad_norm": 3.6702864170074463,
      "learning_rate": 0.00019999657053001917,
      "loss": 2.1506,
      "step": 7370
    },
    {
      "epoch": 0.8880866425992779,
      "grad_norm": 4.6571431159973145,
      "learning_rate": 0.00019999653895145462,
      "loss": 2.2488,
      "step": 7380
    },
    {
      "epoch": 0.8892900120336944,
      "grad_norm": 4.352538585662842,
      "learning_rate": 0.00019999650722817044,
      "loss": 2.3038,
      "step": 7390
    },
    {
      "epoch": 0.8904933814681107,
      "grad_norm": 7.005280494689941,
      "learning_rate": 0.0001999964753601666,
      "loss": 2.4323,
      "step": 7400
    },
    {
      "epoch": 0.8916967509025271,
      "grad_norm": 4.320887088775635,
      "learning_rate": 0.00019999644334744315,
      "loss": 2.2703,
      "step": 7410
    },
    {
      "epoch": 0.8929001203369434,
      "grad_norm": 3.865842580795288,
      "learning_rate": 0.00019999641119000015,
      "loss": 2.3429,
      "step": 7420
    },
    {
      "epoch": 0.8941034897713598,
      "grad_norm": 5.025181293487549,
      "learning_rate": 0.00019999637888783772,
      "loss": 2.0387,
      "step": 7430
    },
    {
      "epoch": 0.8953068592057761,
      "grad_norm": 5.02504301071167,
      "learning_rate": 0.00019999634644095578,
      "loss": 2.357,
      "step": 7440
    },
    {
      "epoch": 0.8965102286401926,
      "grad_norm": 6.646202087402344,
      "learning_rate": 0.0001999963138493545,
      "loss": 2.4796,
      "step": 7450
    },
    {
      "epoch": 0.8977135980746089,
      "grad_norm": 4.324987888336182,
      "learning_rate": 0.0001999962811130338,
      "loss": 2.2562,
      "step": 7460
    },
    {
      "epoch": 0.8989169675090253,
      "grad_norm": 3.8772072792053223,
      "learning_rate": 0.00019999624823199383,
      "loss": 2.0955,
      "step": 7470
    },
    {
      "epoch": 0.9001203369434416,
      "grad_norm": 5.725499153137207,
      "learning_rate": 0.0001999962152062346,
      "loss": 2.0614,
      "step": 7480
    },
    {
      "epoch": 0.901323706377858,
      "grad_norm": 4.010166645050049,
      "learning_rate": 0.00019999618203575615,
      "loss": 2.168,
      "step": 7490
    },
    {
      "epoch": 0.9025270758122743,
      "grad_norm": 6.155289173126221,
      "learning_rate": 0.00019999614872055852,
      "loss": 2.3418,
      "step": 7500
    },
    {
      "epoch": 0.9037304452466908,
      "grad_norm": 4.861774444580078,
      "learning_rate": 0.0001999961152606418,
      "loss": 2.0682,
      "step": 7510
    },
    {
      "epoch": 0.9049338146811071,
      "grad_norm": 4.0041823387146,
      "learning_rate": 0.000199996081656006,
      "loss": 2.0419,
      "step": 7520
    },
    {
      "epoch": 0.9061371841155235,
      "grad_norm": 7.270376682281494,
      "learning_rate": 0.00019999604790665117,
      "loss": 1.976,
      "step": 7530
    },
    {
      "epoch": 0.9073405535499398,
      "grad_norm": 4.057343006134033,
      "learning_rate": 0.0001999960140125774,
      "loss": 2.3294,
      "step": 7540
    },
    {
      "epoch": 0.9085439229843562,
      "grad_norm": 5.586544036865234,
      "learning_rate": 0.00019999597997378467,
      "loss": 2.2769,
      "step": 7550
    },
    {
      "epoch": 0.9097472924187726,
      "grad_norm": 4.569085121154785,
      "learning_rate": 0.0001999959457902731,
      "loss": 2.2375,
      "step": 7560
    },
    {
      "epoch": 0.910950661853189,
      "grad_norm": 3.8890581130981445,
      "learning_rate": 0.0001999959114620427,
      "loss": 2.1112,
      "step": 7570
    },
    {
      "epoch": 0.9121540312876053,
      "grad_norm": 6.173394203186035,
      "learning_rate": 0.00019999587698909355,
      "loss": 2.1902,
      "step": 7580
    },
    {
      "epoch": 0.9133574007220217,
      "grad_norm": 5.014927387237549,
      "learning_rate": 0.00019999584237142566,
      "loss": 2.1683,
      "step": 7590
    },
    {
      "epoch": 0.914560770156438,
      "grad_norm": 5.847899913787842,
      "learning_rate": 0.0001999958076090391,
      "loss": 2.3996,
      "step": 7600
    },
    {
      "epoch": 0.9157641395908543,
      "grad_norm": 5.280930995941162,
      "learning_rate": 0.00019999577270193392,
      "loss": 2.0922,
      "step": 7610
    },
    {
      "epoch": 0.9169675090252708,
      "grad_norm": 3.5625391006469727,
      "learning_rate": 0.00019999573765011011,
      "loss": 2.1937,
      "step": 7620
    },
    {
      "epoch": 0.9181708784596871,
      "grad_norm": 4.805070400238037,
      "learning_rate": 0.00019999570245356787,
      "loss": 2.1327,
      "step": 7630
    },
    {
      "epoch": 0.9193742478941035,
      "grad_norm": 5.546348571777344,
      "learning_rate": 0.0001999956671123071,
      "loss": 2.2091,
      "step": 7640
    },
    {
      "epoch": 0.9205776173285198,
      "grad_norm": 7.768810749053955,
      "learning_rate": 0.00019999563162632794,
      "loss": 2.5238,
      "step": 7650
    },
    {
      "epoch": 0.9217809867629362,
      "grad_norm": 5.4269561767578125,
      "learning_rate": 0.00019999559599563044,
      "loss": 2.0429,
      "step": 7660
    },
    {
      "epoch": 0.9229843561973526,
      "grad_norm": 4.423208713531494,
      "learning_rate": 0.00019999556022021458,
      "loss": 2.3963,
      "step": 7670
    },
    {
      "epoch": 0.924187725631769,
      "grad_norm": 4.935546398162842,
      "learning_rate": 0.00019999552430008044,
      "loss": 2.1661,
      "step": 7680
    },
    {
      "epoch": 0.9253910950661853,
      "grad_norm": 4.095705986022949,
      "learning_rate": 0.00019999548823522814,
      "loss": 1.9773,
      "step": 7690
    },
    {
      "epoch": 0.9265944645006017,
      "grad_norm": 6.3755340576171875,
      "learning_rate": 0.00019999545202565764,
      "loss": 2.0078,
      "step": 7700
    },
    {
      "epoch": 0.927797833935018,
      "grad_norm": 5.321127891540527,
      "learning_rate": 0.00019999541567136905,
      "loss": 2.2288,
      "step": 7710
    },
    {
      "epoch": 0.9290012033694344,
      "grad_norm": 3.6329126358032227,
      "learning_rate": 0.0001999953791723624,
      "loss": 2.1282,
      "step": 7720
    },
    {
      "epoch": 0.9302045728038508,
      "grad_norm": 7.373484134674072,
      "learning_rate": 0.00019999534252863776,
      "loss": 2.215,
      "step": 7730
    },
    {
      "epoch": 0.9314079422382672,
      "grad_norm": 4.53382682800293,
      "learning_rate": 0.0001999953057401952,
      "loss": 2.2439,
      "step": 7740
    },
    {
      "epoch": 0.9326113116726835,
      "grad_norm": 5.549760341644287,
      "learning_rate": 0.00019999526880703467,
      "loss": 2.4746,
      "step": 7750
    },
    {
      "epoch": 0.9338146811070999,
      "grad_norm": 5.585546016693115,
      "learning_rate": 0.00019999523172915633,
      "loss": 2.1636,
      "step": 7760
    },
    {
      "epoch": 0.9350180505415162,
      "grad_norm": 4.102782726287842,
      "learning_rate": 0.0001999951945065602,
      "loss": 2.2475,
      "step": 7770
    },
    {
      "epoch": 0.9362214199759326,
      "grad_norm": 5.998604774475098,
      "learning_rate": 0.00019999515713924637,
      "loss": 2.262,
      "step": 7780
    },
    {
      "epoch": 0.937424789410349,
      "grad_norm": 4.92228364944458,
      "learning_rate": 0.00019999511962721483,
      "loss": 2.1335,
      "step": 7790
    },
    {
      "epoch": 0.9386281588447654,
      "grad_norm": 6.664450168609619,
      "learning_rate": 0.0001999950819704657,
      "loss": 2.1727,
      "step": 7800
    },
    {
      "epoch": 0.9398315282791817,
      "grad_norm": 4.493361949920654,
      "learning_rate": 0.00019999504416899893,
      "loss": 2.0277,
      "step": 7810
    },
    {
      "epoch": 0.941034897713598,
      "grad_norm": 3.9881207942962646,
      "learning_rate": 0.00019999500622281467,
      "loss": 1.8935,
      "step": 7820
    },
    {
      "epoch": 0.9422382671480144,
      "grad_norm": 5.744288921356201,
      "learning_rate": 0.00019999496813191296,
      "loss": 2.204,
      "step": 7830
    },
    {
      "epoch": 0.9434416365824309,
      "grad_norm": 3.685032367706299,
      "learning_rate": 0.00019999492989629386,
      "loss": 2.3238,
      "step": 7840
    },
    {
      "epoch": 0.9446450060168472,
      "grad_norm": 6.588531970977783,
      "learning_rate": 0.00019999489151595738,
      "loss": 2.0914,
      "step": 7850
    },
    {
      "epoch": 0.9458483754512635,
      "grad_norm": 5.450845718383789,
      "learning_rate": 0.0001999948529909036,
      "loss": 1.9164,
      "step": 7860
    },
    {
      "epoch": 0.9470517448856799,
      "grad_norm": 3.3784396648406982,
      "learning_rate": 0.0001999948143211326,
      "loss": 2.0871,
      "step": 7870
    },
    {
      "epoch": 0.9482551143200962,
      "grad_norm": 6.645673751831055,
      "learning_rate": 0.0001999947755066444,
      "loss": 2.1626,
      "step": 7880
    },
    {
      "epoch": 0.9494584837545126,
      "grad_norm": 3.890578508377075,
      "learning_rate": 0.0001999947365474391,
      "loss": 2.1516,
      "step": 7890
    },
    {
      "epoch": 0.950661853188929,
      "grad_norm": 6.042820453643799,
      "learning_rate": 0.0001999946974435167,
      "loss": 2.4586,
      "step": 7900
    },
    {
      "epoch": 0.9518652226233454,
      "grad_norm": 4.843379974365234,
      "learning_rate": 0.0001999946581948773,
      "loss": 2.0618,
      "step": 7910
    },
    {
      "epoch": 0.9530685920577617,
      "grad_norm": 3.511101484298706,
      "learning_rate": 0.00019999461880152093,
      "loss": 2.3543,
      "step": 7920
    },
    {
      "epoch": 0.9542719614921781,
      "grad_norm": 5.819178104400635,
      "learning_rate": 0.00019999457926344767,
      "loss": 2.1943,
      "step": 7930
    },
    {
      "epoch": 0.9554753309265944,
      "grad_norm": 4.102616310119629,
      "learning_rate": 0.00019999453958065755,
      "loss": 2.1075,
      "step": 7940
    },
    {
      "epoch": 0.9566787003610109,
      "grad_norm": 6.817751884460449,
      "learning_rate": 0.00019999449975315064,
      "loss": 2.2178,
      "step": 7950
    },
    {
      "epoch": 0.9578820697954272,
      "grad_norm": 5.780127048492432,
      "learning_rate": 0.00019999445978092703,
      "loss": 1.9242,
      "step": 7960
    },
    {
      "epoch": 0.9590854392298436,
      "grad_norm": 4.017819881439209,
      "learning_rate": 0.00019999441966398674,
      "loss": 2.4252,
      "step": 7970
    },
    {
      "epoch": 0.9602888086642599,
      "grad_norm": 6.4633331298828125,
      "learning_rate": 0.00019999437940232983,
      "loss": 2.0193,
      "step": 7980
    },
    {
      "epoch": 0.9614921780986763,
      "grad_norm": 4.9790496826171875,
      "learning_rate": 0.00019999433899595633,
      "loss": 2.1775,
      "step": 7990
    },
    {
      "epoch": 0.9626955475330926,
      "grad_norm": 6.155992031097412,
      "learning_rate": 0.0001999942984448664,
      "loss": 2.3467,
      "step": 8000
    },
    {
      "epoch": 0.9638989169675091,
      "grad_norm": 4.168816566467285,
      "learning_rate": 0.00019999425774906,
      "loss": 2.1639,
      "step": 8010
    },
    {
      "epoch": 0.9651022864019254,
      "grad_norm": 4.128564357757568,
      "learning_rate": 0.00019999421690853722,
      "loss": 2.0396,
      "step": 8020
    },
    {
      "epoch": 0.9663056558363418,
      "grad_norm": 5.868348598480225,
      "learning_rate": 0.00019999417592329812,
      "loss": 1.9762,
      "step": 8030
    },
    {
      "epoch": 0.9675090252707581,
      "grad_norm": 4.708838939666748,
      "learning_rate": 0.00019999413479334277,
      "loss": 2.1418,
      "step": 8040
    },
    {
      "epoch": 0.9687123947051745,
      "grad_norm": 6.427163600921631,
      "learning_rate": 0.0001999940935186712,
      "loss": 2.26,
      "step": 8050
    },
    {
      "epoch": 0.9699157641395909,
      "grad_norm": 4.871311664581299,
      "learning_rate": 0.00019999405209928352,
      "loss": 2.2123,
      "step": 8060
    },
    {
      "epoch": 0.9711191335740073,
      "grad_norm": 3.6231939792633057,
      "learning_rate": 0.0001999940105351797,
      "loss": 2.2806,
      "step": 8070
    },
    {
      "epoch": 0.9723225030084236,
      "grad_norm": 5.265206813812256,
      "learning_rate": 0.0001999939688263599,
      "loss": 2.1661,
      "step": 8080
    },
    {
      "epoch": 0.97352587244284,
      "grad_norm": 4.007277488708496,
      "learning_rate": 0.00019999392697282416,
      "loss": 2.1418,
      "step": 8090
    },
    {
      "epoch": 0.9747292418772563,
      "grad_norm": 5.873310089111328,
      "learning_rate": 0.0001999938849745725,
      "loss": 2.5009,
      "step": 8100
    },
    {
      "epoch": 0.9759326113116726,
      "grad_norm": 5.166054725646973,
      "learning_rate": 0.000199993842831605,
      "loss": 2.2151,
      "step": 8110
    },
    {
      "epoch": 0.9771359807460891,
      "grad_norm": 4.688544750213623,
      "learning_rate": 0.00019999380054392172,
      "loss": 2.1889,
      "step": 8120
    },
    {
      "epoch": 0.9783393501805054,
      "grad_norm": 5.526015758514404,
      "learning_rate": 0.00019999375811152273,
      "loss": 2.2725,
      "step": 8130
    },
    {
      "epoch": 0.9795427196149218,
      "grad_norm": 4.228044033050537,
      "learning_rate": 0.00019999371553440807,
      "loss": 2.3354,
      "step": 8140
    },
    {
      "epoch": 0.9807460890493381,
      "grad_norm": 5.739175319671631,
      "learning_rate": 0.00019999367281257785,
      "loss": 2.2704,
      "step": 8150
    },
    {
      "epoch": 0.9819494584837545,
      "grad_norm": 4.4686665534973145,
      "learning_rate": 0.00019999362994603206,
      "loss": 2.1563,
      "step": 8160
    },
    {
      "epoch": 0.9831528279181708,
      "grad_norm": 4.358283042907715,
      "learning_rate": 0.00019999358693477079,
      "loss": 2.3027,
      "step": 8170
    },
    {
      "epoch": 0.9843561973525873,
      "grad_norm": 4.596220016479492,
      "learning_rate": 0.00019999354377879414,
      "loss": 2.0057,
      "step": 8180
    },
    {
      "epoch": 0.9855595667870036,
      "grad_norm": 4.477689743041992,
      "learning_rate": 0.0001999935004781021,
      "loss": 2.0681,
      "step": 8190
    },
    {
      "epoch": 0.98676293622142,
      "grad_norm": 6.645779132843018,
      "learning_rate": 0.00019999345703269485,
      "loss": 2.4531,
      "step": 8200
    },
    {
      "epoch": 0.9879663056558363,
      "grad_norm": 4.308712959289551,
      "learning_rate": 0.00019999341344257234,
      "loss": 2.1801,
      "step": 8210
    },
    {
      "epoch": 0.9891696750902527,
      "grad_norm": 3.8493010997772217,
      "learning_rate": 0.0001999933697077347,
      "loss": 2.239,
      "step": 8220
    },
    {
      "epoch": 0.9903730445246691,
      "grad_norm": 5.303299427032471,
      "learning_rate": 0.00019999332582818193,
      "loss": 2.235,
      "step": 8230
    },
    {
      "epoch": 0.9915764139590855,
      "grad_norm": 4.349549770355225,
      "learning_rate": 0.00019999328180391414,
      "loss": 2.1783,
      "step": 8240
    },
    {
      "epoch": 0.9927797833935018,
      "grad_norm": 7.270279884338379,
      "learning_rate": 0.00019999323763493138,
      "loss": 2.3507,
      "step": 8250
    },
    {
      "epoch": 0.9939831528279182,
      "grad_norm": 4.912819862365723,
      "learning_rate": 0.00019999319332123373,
      "loss": 2.1266,
      "step": 8260
    },
    {
      "epoch": 0.9951865222623345,
      "grad_norm": 3.0928843021392822,
      "learning_rate": 0.00019999314886282122,
      "loss": 2.4408,
      "step": 8270
    },
    {
      "epoch": 0.9963898916967509,
      "grad_norm": 7.019705772399902,
      "learning_rate": 0.00019999310425969398,
      "loss": 2.2645,
      "step": 8280
    },
    {
      "epoch": 0.9975932611311673,
      "grad_norm": 4.224442005157471,
      "learning_rate": 0.000199993059511852,
      "loss": 2.2539,
      "step": 8290
    },
    {
      "epoch": 0.9987966305655837,
      "grad_norm": 5.672049045562744,
      "learning_rate": 0.0001999930146192954,
      "loss": 2.1419,
      "step": 8300
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.581302165985107,
      "learning_rate": 0.00019999296958202417,
      "loss": 2.1409,
      "step": 8310
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.080259084701538,
      "eval_runtime": 118.705,
      "eval_samples_per_second": 62.23,
      "eval_steps_per_second": 7.784,
      "step": 8310
    },
    {
      "epoch": 1.0012033694344165,
      "grad_norm": 4.850121974945068,
      "learning_rate": 0.00019999292440003847,
      "loss": 1.776,
      "step": 8320
    },
    {
      "epoch": 1.0024067388688327,
      "grad_norm": 3.2496631145477295,
      "learning_rate": 0.0001999928790733383,
      "loss": 2.1302,
      "step": 8330
    },
    {
      "epoch": 1.0036101083032491,
      "grad_norm": 5.475099086761475,
      "learning_rate": 0.00019999283360192375,
      "loss": 1.7934,
      "step": 8340
    },
    {
      "epoch": 1.0048134777376654,
      "grad_norm": 4.7173380851745605,
      "learning_rate": 0.00019999278798579488,
      "loss": 2.1552,
      "step": 8350
    },
    {
      "epoch": 1.0060168471720818,
      "grad_norm": 6.577690601348877,
      "learning_rate": 0.00019999274222495177,
      "loss": 2.0915,
      "step": 8360
    },
    {
      "epoch": 1.0072202166064983,
      "grad_norm": 4.1316375732421875,
      "learning_rate": 0.00019999269631939447,
      "loss": 1.8381,
      "step": 8370
    },
    {
      "epoch": 1.0084235860409145,
      "grad_norm": 4.386764049530029,
      "learning_rate": 0.00019999265026912307,
      "loss": 2.001,
      "step": 8380
    },
    {
      "epoch": 1.009626955475331,
      "grad_norm": 5.818033695220947,
      "learning_rate": 0.0001999926040741376,
      "loss": 1.9157,
      "step": 8390
    },
    {
      "epoch": 1.0108303249097472,
      "grad_norm": 4.358513832092285,
      "learning_rate": 0.00019999255773443813,
      "loss": 2.1462,
      "step": 8400
    },
    {
      "epoch": 1.0120336943441637,
      "grad_norm": 6.169714450836182,
      "learning_rate": 0.00019999251125002476,
      "loss": 1.9187,
      "step": 8410
    },
    {
      "epoch": 1.01323706377858,
      "grad_norm": 3.796625852584839,
      "learning_rate": 0.00019999246462089755,
      "loss": 1.7694,
      "step": 8420
    },
    {
      "epoch": 1.0144404332129964,
      "grad_norm": 3.1594159603118896,
      "learning_rate": 0.0001999924178470565,
      "loss": 1.9839,
      "step": 8430
    },
    {
      "epoch": 1.0156438026474128,
      "grad_norm": 5.432295322418213,
      "learning_rate": 0.00019999237092850177,
      "loss": 1.8924,
      "step": 8440
    },
    {
      "epoch": 1.016847172081829,
      "grad_norm": 3.7919535636901855,
      "learning_rate": 0.0001999923238652334,
      "loss": 2.0057,
      "step": 8450
    },
    {
      "epoch": 1.0180505415162455,
      "grad_norm": 6.825023651123047,
      "learning_rate": 0.00019999227665725146,
      "loss": 2.2807,
      "step": 8460
    },
    {
      "epoch": 1.0192539109506618,
      "grad_norm": 4.782787322998047,
      "learning_rate": 0.000199992229304556,
      "loss": 1.9329,
      "step": 8470
    },
    {
      "epoch": 1.0204572803850782,
      "grad_norm": 3.1457180976867676,
      "learning_rate": 0.00019999218180714707,
      "loss": 2.1785,
      "step": 8480
    },
    {
      "epoch": 1.0216606498194947,
      "grad_norm": 5.4611945152282715,
      "learning_rate": 0.00019999213416502479,
      "loss": 1.9522,
      "step": 8490
    },
    {
      "epoch": 1.022864019253911,
      "grad_norm": 4.97552490234375,
      "learning_rate": 0.0001999920863781892,
      "loss": 2.0244,
      "step": 8500
    },
    {
      "epoch": 1.0240673886883274,
      "grad_norm": 5.632757663726807,
      "learning_rate": 0.0001999920384466404,
      "loss": 2.0238,
      "step": 8510
    },
    {
      "epoch": 1.0252707581227436,
      "grad_norm": 5.427265167236328,
      "learning_rate": 0.0001999919903703784,
      "loss": 2.0926,
      "step": 8520
    },
    {
      "epoch": 1.02647412755716,
      "grad_norm": 4.070896625518799,
      "learning_rate": 0.00019999194214940328,
      "loss": 2.0128,
      "step": 8530
    },
    {
      "epoch": 1.0276774969915765,
      "grad_norm": 5.544242858886719,
      "learning_rate": 0.00019999189378371517,
      "loss": 2.0243,
      "step": 8540
    },
    {
      "epoch": 1.0288808664259927,
      "grad_norm": 3.901855230331421,
      "learning_rate": 0.0001999918452733141,
      "loss": 1.9422,
      "step": 8550
    },
    {
      "epoch": 1.0300842358604092,
      "grad_norm": 5.984253883361816,
      "learning_rate": 0.00019999179661820015,
      "loss": 1.9022,
      "step": 8560
    },
    {
      "epoch": 1.0312876052948254,
      "grad_norm": 4.536534786224365,
      "learning_rate": 0.00019999174781837336,
      "loss": 1.8929,
      "step": 8570
    },
    {
      "epoch": 1.032490974729242,
      "grad_norm": 3.4476656913757324,
      "learning_rate": 0.00019999169887383383,
      "loss": 1.9753,
      "step": 8580
    },
    {
      "epoch": 1.0336943441636581,
      "grad_norm": 5.469036102294922,
      "learning_rate": 0.00019999164978458164,
      "loss": 1.8637,
      "step": 8590
    },
    {
      "epoch": 1.0348977135980746,
      "grad_norm": 3.7836496829986572,
      "learning_rate": 0.00019999160055061683,
      "loss": 1.9978,
      "step": 8600
    },
    {
      "epoch": 1.036101083032491,
      "grad_norm": 6.112541198730469,
      "learning_rate": 0.0001999915511719395,
      "loss": 2.142,
      "step": 8610
    },
    {
      "epoch": 1.0373044524669073,
      "grad_norm": 5.205537796020508,
      "learning_rate": 0.0001999915016485497,
      "loss": 2.1229,
      "step": 8620
    },
    {
      "epoch": 1.0385078219013237,
      "grad_norm": 4.5776519775390625,
      "learning_rate": 0.0001999914519804475,
      "loss": 2.0338,
      "step": 8630
    },
    {
      "epoch": 1.03971119133574,
      "grad_norm": 7.718768119812012,
      "learning_rate": 0.000199991402167633,
      "loss": 1.9866,
      "step": 8640
    },
    {
      "epoch": 1.0409145607701564,
      "grad_norm": 4.679311275482178,
      "learning_rate": 0.00019999135221010624,
      "loss": 2.1309,
      "step": 8650
    },
    {
      "epoch": 1.0421179302045729,
      "grad_norm": 6.655802249908447,
      "learning_rate": 0.0001999913021078673,
      "loss": 1.8992,
      "step": 8660
    },
    {
      "epoch": 1.0433212996389891,
      "grad_norm": 5.339411735534668,
      "learning_rate": 0.0001999912518609163,
      "loss": 2.2236,
      "step": 8670
    },
    {
      "epoch": 1.0445246690734056,
      "grad_norm": 3.4371402263641357,
      "learning_rate": 0.00019999120146925323,
      "loss": 1.9414,
      "step": 8680
    },
    {
      "epoch": 1.0457280385078218,
      "grad_norm": 5.796459197998047,
      "learning_rate": 0.0001999911509328782,
      "loss": 2.1834,
      "step": 8690
    },
    {
      "epoch": 1.0469314079422383,
      "grad_norm": 4.311466693878174,
      "learning_rate": 0.0001999911002517913,
      "loss": 2.0023,
      "step": 8700
    },
    {
      "epoch": 1.0481347773766547,
      "grad_norm": 6.57638692855835,
      "learning_rate": 0.00019999104942599263,
      "loss": 1.9436,
      "step": 8710
    },
    {
      "epoch": 1.049338146811071,
      "grad_norm": 6.028642654418945,
      "learning_rate": 0.00019999099845548218,
      "loss": 1.7673,
      "step": 8720
    },
    {
      "epoch": 1.0505415162454874,
      "grad_norm": 3.5979578495025635,
      "learning_rate": 0.00019999094734026007,
      "loss": 2.0021,
      "step": 8730
    },
    {
      "epoch": 1.0517448856799037,
      "grad_norm": 6.466935634613037,
      "learning_rate": 0.00019999089608032638,
      "loss": 2.0623,
      "step": 8740
    },
    {
      "epoch": 1.05294825511432,
      "grad_norm": 4.704542636871338,
      "learning_rate": 0.00019999084467568117,
      "loss": 1.9281,
      "step": 8750
    },
    {
      "epoch": 1.0541516245487366,
      "grad_norm": 6.5226311683654785,
      "learning_rate": 0.00019999079312632453,
      "loss": 2.0969,
      "step": 8760
    },
    {
      "epoch": 1.0553549939831528,
      "grad_norm": 5.316224575042725,
      "learning_rate": 0.0001999907414322565,
      "loss": 1.9609,
      "step": 8770
    },
    {
      "epoch": 1.0565583634175693,
      "grad_norm": 3.4058215618133545,
      "learning_rate": 0.00019999068959347725,
      "loss": 2.2148,
      "step": 8780
    },
    {
      "epoch": 1.0577617328519855,
      "grad_norm": 6.79750394821167,
      "learning_rate": 0.00019999063760998671,
      "loss": 1.8923,
      "step": 8790
    },
    {
      "epoch": 1.058965102286402,
      "grad_norm": 4.632766246795654,
      "learning_rate": 0.00019999058548178506,
      "loss": 2.0139,
      "step": 8800
    },
    {
      "epoch": 1.0601684717208182,
      "grad_norm": 5.880581378936768,
      "learning_rate": 0.00019999053320887236,
      "loss": 1.9082,
      "step": 8810
    },
    {
      "epoch": 1.0613718411552346,
      "grad_norm": 4.673683166503906,
      "learning_rate": 0.00019999048079124865,
      "loss": 2.0224,
      "step": 8820
    },
    {
      "epoch": 1.062575210589651,
      "grad_norm": 3.230836868286133,
      "learning_rate": 0.00019999042822891404,
      "loss": 2.106,
      "step": 8830
    },
    {
      "epoch": 1.0637785800240673,
      "grad_norm": 5.293460369110107,
      "learning_rate": 0.00019999037552186858,
      "loss": 2.36,
      "step": 8840
    },
    {
      "epoch": 1.0649819494584838,
      "grad_norm": 4.429105281829834,
      "learning_rate": 0.00019999032267011235,
      "loss": 1.9762,
      "step": 8850
    },
    {
      "epoch": 1.0661853188929,
      "grad_norm": 6.840360641479492,
      "learning_rate": 0.00019999026967364546,
      "loss": 1.9054,
      "step": 8860
    },
    {
      "epoch": 1.0673886883273165,
      "grad_norm": 4.287208557128906,
      "learning_rate": 0.00019999021653246797,
      "loss": 1.9889,
      "step": 8870
    },
    {
      "epoch": 1.068592057761733,
      "grad_norm": 2.9769177436828613,
      "learning_rate": 0.00019999016324657992,
      "loss": 1.941,
      "step": 8880
    },
    {
      "epoch": 1.0697954271961492,
      "grad_norm": 7.548506259918213,
      "learning_rate": 0.00019999010981598146,
      "loss": 2.0448,
      "step": 8890
    },
    {
      "epoch": 1.0709987966305656,
      "grad_norm": 4.892628192901611,
      "learning_rate": 0.0001999900562406726,
      "loss": 2.0519,
      "step": 8900
    },
    {
      "epoch": 1.0722021660649819,
      "grad_norm": 7.174940586090088,
      "learning_rate": 0.0001999900025206534,
      "loss": 2.0359,
      "step": 8910
    },
    {
      "epoch": 1.0734055354993983,
      "grad_norm": 4.764652252197266,
      "learning_rate": 0.00019998994865592405,
      "loss": 1.812,
      "step": 8920
    },
    {
      "epoch": 1.0746089049338148,
      "grad_norm": 3.470702648162842,
      "learning_rate": 0.00019998989464648453,
      "loss": 2.1387,
      "step": 8930
    },
    {
      "epoch": 1.075812274368231,
      "grad_norm": 5.508899688720703,
      "learning_rate": 0.00019998984049233494,
      "loss": 1.8764,
      "step": 8940
    },
    {
      "epoch": 1.0770156438026475,
      "grad_norm": 4.350720405578613,
      "learning_rate": 0.00019998978619347537,
      "loss": 1.9498,
      "step": 8950
    },
    {
      "epoch": 1.0782190132370637,
      "grad_norm": 7.807206630706787,
      "learning_rate": 0.0001999897317499059,
      "loss": 2.1676,
      "step": 8960
    },
    {
      "epoch": 1.0794223826714802,
      "grad_norm": 5.922019958496094,
      "learning_rate": 0.00019998967716162658,
      "loss": 1.8992,
      "step": 8970
    },
    {
      "epoch": 1.0806257521058966,
      "grad_norm": 3.9038050174713135,
      "learning_rate": 0.0001999896224286375,
      "loss": 2.2461,
      "step": 8980
    },
    {
      "epoch": 1.0818291215403129,
      "grad_norm": 4.654329299926758,
      "learning_rate": 0.00019998956755093878,
      "loss": 2.008,
      "step": 8990
    },
    {
      "epoch": 1.0830324909747293,
      "grad_norm": 4.3668036460876465,
      "learning_rate": 0.00019998951252853047,
      "loss": 1.918,
      "step": 9000
    },
    {
      "epoch": 1.0842358604091455,
      "grad_norm": 6.6642842292785645,
      "learning_rate": 0.00019998945736141263,
      "loss": 1.9944,
      "step": 9010
    },
    {
      "epoch": 1.085439229843562,
      "grad_norm": 4.998341083526611,
      "learning_rate": 0.00019998940204958538,
      "loss": 2.0143,
      "step": 9020
    },
    {
      "epoch": 1.0866425992779782,
      "grad_norm": 3.4509310722351074,
      "learning_rate": 0.00019998934659304875,
      "loss": 2.1054,
      "step": 9030
    },
    {
      "epoch": 1.0878459687123947,
      "grad_norm": 4.551707744598389,
      "learning_rate": 0.00019998929099180286,
      "loss": 2.0293,
      "step": 9040
    },
    {
      "epoch": 1.0890493381468112,
      "grad_norm": 3.933088779449463,
      "learning_rate": 0.0001999892352458478,
      "loss": 2.0668,
      "step": 9050
    },
    {
      "epoch": 1.0902527075812274,
      "grad_norm": 6.708771228790283,
      "learning_rate": 0.00019998917935518362,
      "loss": 2.1052,
      "step": 9060
    },
    {
      "epoch": 1.0914560770156438,
      "grad_norm": 5.076292991638184,
      "learning_rate": 0.0001999891233198104,
      "loss": 2.0848,
      "step": 9070
    },
    {
      "epoch": 1.09265944645006,
      "grad_norm": 3.7987160682678223,
      "learning_rate": 0.00019998906713972824,
      "loss": 1.8708,
      "step": 9080
    },
    {
      "epoch": 1.0938628158844765,
      "grad_norm": 5.70545768737793,
      "learning_rate": 0.0001999890108149372,
      "loss": 2.0238,
      "step": 9090
    },
    {
      "epoch": 1.095066185318893,
      "grad_norm": 5.972437858581543,
      "learning_rate": 0.00019998895434543738,
      "loss": 1.8914,
      "step": 9100
    },
    {
      "epoch": 1.0962695547533092,
      "grad_norm": 5.777250289916992,
      "learning_rate": 0.00019998889773122888,
      "loss": 2.1697,
      "step": 9110
    },
    {
      "epoch": 1.0974729241877257,
      "grad_norm": 4.514420032501221,
      "learning_rate": 0.0001999888409723117,
      "loss": 2.0899,
      "step": 9120
    },
    {
      "epoch": 1.098676293622142,
      "grad_norm": 3.5541980266571045,
      "learning_rate": 0.00019998878406868605,
      "loss": 1.9751,
      "step": 9130
    },
    {
      "epoch": 1.0998796630565584,
      "grad_norm": 5.4803266525268555,
      "learning_rate": 0.00019998872702035192,
      "loss": 1.9228,
      "step": 9140
    },
    {
      "epoch": 1.1010830324909748,
      "grad_norm": 4.44549560546875,
      "learning_rate": 0.0001999886698273094,
      "loss": 1.8906,
      "step": 9150
    },
    {
      "epoch": 1.102286401925391,
      "grad_norm": 5.911207675933838,
      "learning_rate": 0.0001999886124895586,
      "loss": 2.2008,
      "step": 9160
    },
    {
      "epoch": 1.1034897713598075,
      "grad_norm": 5.027959823608398,
      "learning_rate": 0.0001999885550070996,
      "loss": 1.9231,
      "step": 9170
    },
    {
      "epoch": 1.1046931407942238,
      "grad_norm": 4.950850009918213,
      "learning_rate": 0.00019998849737993247,
      "loss": 2.0911,
      "step": 9180
    },
    {
      "epoch": 1.1058965102286402,
      "grad_norm": 5.364449977874756,
      "learning_rate": 0.0001999884396080573,
      "loss": 1.9605,
      "step": 9190
    },
    {
      "epoch": 1.1070998796630565,
      "grad_norm": 4.605388641357422,
      "learning_rate": 0.00019998838169147415,
      "loss": 2.1481,
      "step": 9200
    },
    {
      "epoch": 1.108303249097473,
      "grad_norm": 5.884681224822998,
      "learning_rate": 0.00019998832363018313,
      "loss": 1.8628,
      "step": 9210
    },
    {
      "epoch": 1.1095066185318894,
      "grad_norm": 4.382067680358887,
      "learning_rate": 0.00019998826542418435,
      "loss": 1.9347,
      "step": 9220
    },
    {
      "epoch": 1.1107099879663056,
      "grad_norm": 3.439744472503662,
      "learning_rate": 0.00019998820707347784,
      "loss": 2.1023,
      "step": 9230
    },
    {
      "epoch": 1.111913357400722,
      "grad_norm": 6.21846866607666,
      "learning_rate": 0.00019998814857806368,
      "loss": 2.0217,
      "step": 9240
    },
    {
      "epoch": 1.1131167268351383,
      "grad_norm": 4.801575660705566,
      "learning_rate": 0.00019998808993794203,
      "loss": 2.0597,
      "step": 9250
    },
    {
      "epoch": 1.1143200962695547,
      "grad_norm": 6.326198577880859,
      "learning_rate": 0.0001999880311531129,
      "loss": 2.066,
      "step": 9260
    },
    {
      "epoch": 1.1155234657039712,
      "grad_norm": 4.852086544036865,
      "learning_rate": 0.0001999879722235764,
      "loss": 1.8671,
      "step": 9270
    },
    {
      "epoch": 1.1167268351383874,
      "grad_norm": 4.770281791687012,
      "learning_rate": 0.00019998791314933265,
      "loss": 2.1625,
      "step": 9280
    },
    {
      "epoch": 1.117930204572804,
      "grad_norm": 5.579623699188232,
      "learning_rate": 0.0001999878539303817,
      "loss": 1.9572,
      "step": 9290
    },
    {
      "epoch": 1.1191335740072201,
      "grad_norm": 3.959735155105591,
      "learning_rate": 0.00019998779456672364,
      "loss": 2.0381,
      "step": 9300
    },
    {
      "epoch": 1.1203369434416366,
      "grad_norm": 8.383445739746094,
      "learning_rate": 0.0001999877350583585,
      "loss": 2.0783,
      "step": 9310
    },
    {
      "epoch": 1.121540312876053,
      "grad_norm": 4.759198188781738,
      "learning_rate": 0.00019998767540528646,
      "loss": 2.0799,
      "step": 9320
    },
    {
      "epoch": 1.1227436823104693,
      "grad_norm": 3.45686674118042,
      "learning_rate": 0.0001999876156075076,
      "loss": 2.1875,
      "step": 9330
    },
    {
      "epoch": 1.1239470517448857,
      "grad_norm": 5.586175441741943,
      "learning_rate": 0.0001999875556650219,
      "loss": 2.1816,
      "step": 9340
    },
    {
      "epoch": 1.125150421179302,
      "grad_norm": 4.440593242645264,
      "learning_rate": 0.00019998749557782958,
      "loss": 1.8717,
      "step": 9350
    },
    {
      "epoch": 1.1263537906137184,
      "grad_norm": 7.6358489990234375,
      "learning_rate": 0.00019998743534593066,
      "loss": 2.0476,
      "step": 9360
    },
    {
      "epoch": 1.1275571600481347,
      "grad_norm": 4.324075222015381,
      "learning_rate": 0.00019998737496932522,
      "loss": 1.9486,
      "step": 9370
    },
    {
      "epoch": 1.1287605294825511,
      "grad_norm": 4.16616153717041,
      "learning_rate": 0.00019998731444801339,
      "loss": 2.0713,
      "step": 9380
    },
    {
      "epoch": 1.1299638989169676,
      "grad_norm": 5.545494079589844,
      "learning_rate": 0.0001999872537819952,
      "loss": 2.1747,
      "step": 9390
    },
    {
      "epoch": 1.1311672683513838,
      "grad_norm": 4.747119426727295,
      "learning_rate": 0.00019998719297127077,
      "loss": 2.1822,
      "step": 9400
    },
    {
      "epoch": 1.1323706377858003,
      "grad_norm": 6.263423919677734,
      "learning_rate": 0.00019998713201584018,
      "loss": 2.0005,
      "step": 9410
    },
    {
      "epoch": 1.1335740072202167,
      "grad_norm": 5.5049896240234375,
      "learning_rate": 0.00019998707091570353,
      "loss": 1.9868,
      "step": 9420
    },
    {
      "epoch": 1.134777376654633,
      "grad_norm": 2.4110701084136963,
      "learning_rate": 0.00019998700967086092,
      "loss": 1.8796,
      "step": 9430
    },
    {
      "epoch": 1.1359807460890494,
      "grad_norm": 5.396789073944092,
      "learning_rate": 0.00019998694828131242,
      "loss": 2.1325,
      "step": 9440
    },
    {
      "epoch": 1.1371841155234657,
      "grad_norm": 4.019585609436035,
      "learning_rate": 0.0001999868867470581,
      "loss": 2.0537,
      "step": 9450
    },
    {
      "epoch": 1.1383874849578821,
      "grad_norm": 7.141306400299072,
      "learning_rate": 0.00019998682506809808,
      "loss": 1.94,
      "step": 9460
    },
    {
      "epoch": 1.1395908543922983,
      "grad_norm": 5.070889949798584,
      "learning_rate": 0.00019998676324443242,
      "loss": 1.8988,
      "step": 9470
    },
    {
      "epoch": 1.1407942238267148,
      "grad_norm": 3.384098768234253,
      "learning_rate": 0.00019998670127606126,
      "loss": 1.9526,
      "step": 9480
    },
    {
      "epoch": 1.1419975932611313,
      "grad_norm": 6.994525909423828,
      "learning_rate": 0.00019998663916298464,
      "loss": 2.0269,
      "step": 9490
    },
    {
      "epoch": 1.1432009626955475,
      "grad_norm": 3.67915415763855,
      "learning_rate": 0.00019998657690520265,
      "loss": 2.1269,
      "step": 9500
    },
    {
      "epoch": 1.144404332129964,
      "grad_norm": 7.615024089813232,
      "learning_rate": 0.00019998651450271542,
      "loss": 1.9923,
      "step": 9510
    },
    {
      "epoch": 1.1456077015643802,
      "grad_norm": 5.535434246063232,
      "learning_rate": 0.000199986451955523,
      "loss": 1.8611,
      "step": 9520
    },
    {
      "epoch": 1.1468110709987966,
      "grad_norm": 3.3452887535095215,
      "learning_rate": 0.00019998638926362552,
      "loss": 2.0757,
      "step": 9530
    },
    {
      "epoch": 1.1480144404332129,
      "grad_norm": 4.985581398010254,
      "learning_rate": 0.00019998632642702303,
      "loss": 2.0092,
      "step": 9540
    },
    {
      "epoch": 1.1492178098676293,
      "grad_norm": 4.592456817626953,
      "learning_rate": 0.00019998626344571566,
      "loss": 2.1094,
      "step": 9550
    },
    {
      "epoch": 1.1504211793020458,
      "grad_norm": 7.482632637023926,
      "learning_rate": 0.00019998620031970349,
      "loss": 2.0469,
      "step": 9560
    },
    {
      "epoch": 1.151624548736462,
      "grad_norm": 4.598583221435547,
      "learning_rate": 0.00019998613704898656,
      "loss": 2.1008,
      "step": 9570
    },
    {
      "epoch": 1.1528279181708785,
      "grad_norm": 3.9577908515930176,
      "learning_rate": 0.00019998607363356502,
      "loss": 2.0616,
      "step": 9580
    },
    {
      "epoch": 1.154031287605295,
      "grad_norm": 5.127796649932861,
      "learning_rate": 0.00019998601007343897,
      "loss": 2.1729,
      "step": 9590
    },
    {
      "epoch": 1.1552346570397112,
      "grad_norm": 4.676178455352783,
      "learning_rate": 0.00019998594636860847,
      "loss": 2.0056,
      "step": 9600
    },
    {
      "epoch": 1.1564380264741276,
      "grad_norm": 5.499486923217773,
      "learning_rate": 0.0001999858825190736,
      "loss": 2.1094,
      "step": 9610
    },
    {
      "epoch": 1.1576413959085439,
      "grad_norm": 4.813460826873779,
      "learning_rate": 0.0001999858185248345,
      "loss": 1.8542,
      "step": 9620
    },
    {
      "epoch": 1.1588447653429603,
      "grad_norm": 3.5271053314208984,
      "learning_rate": 0.00019998575438589117,
      "loss": 2.1226,
      "step": 9630
    },
    {
      "epoch": 1.1600481347773766,
      "grad_norm": 6.317688941955566,
      "learning_rate": 0.00019998569010224384,
      "loss": 1.9584,
      "step": 9640
    },
    {
      "epoch": 1.161251504211793,
      "grad_norm": 4.959247589111328,
      "learning_rate": 0.00019998562567389251,
      "loss": 1.9842,
      "step": 9650
    },
    {
      "epoch": 1.1624548736462095,
      "grad_norm": 6.804013252258301,
      "learning_rate": 0.00019998556110083728,
      "loss": 2.1276,
      "step": 9660
    },
    {
      "epoch": 1.1636582430806257,
      "grad_norm": 4.854011535644531,
      "learning_rate": 0.00019998549638307827,
      "loss": 1.8973,
      "step": 9670
    },
    {
      "epoch": 1.1648616125150422,
      "grad_norm": 3.1386466026306152,
      "learning_rate": 0.00019998543152061556,
      "loss": 1.9201,
      "step": 9680
    },
    {
      "epoch": 1.1660649819494584,
      "grad_norm": 5.494327545166016,
      "learning_rate": 0.00019998536651344925,
      "loss": 2.0264,
      "step": 9690
    },
    {
      "epoch": 1.1672683513838749,
      "grad_norm": 3.908991575241089,
      "learning_rate": 0.00019998530136157943,
      "loss": 2.0907,
      "step": 9700
    },
    {
      "epoch": 1.168471720818291,
      "grad_norm": 7.346467018127441,
      "learning_rate": 0.00019998523606500616,
      "loss": 2.0548,
      "step": 9710
    },
    {
      "epoch": 1.1696750902527075,
      "grad_norm": 4.82690954208374,
      "learning_rate": 0.0001999851706237296,
      "loss": 1.9129,
      "step": 9720
    },
    {
      "epoch": 1.170878459687124,
      "grad_norm": 3.32192063331604,
      "learning_rate": 0.0001999851050377498,
      "loss": 2.0197,
      "step": 9730
    },
    {
      "epoch": 1.1720818291215402,
      "grad_norm": 4.709057807922363,
      "learning_rate": 0.0001999850393070669,
      "loss": 2.1231,
      "step": 9740
    },
    {
      "epoch": 1.1732851985559567,
      "grad_norm": 4.261242866516113,
      "learning_rate": 0.0001999849734316809,
      "loss": 2.1172,
      "step": 9750
    },
    {
      "epoch": 1.1744885679903732,
      "grad_norm": 6.338689804077148,
      "learning_rate": 0.00019998490741159202,
      "loss": 2.1408,
      "step": 9760
    },
    {
      "epoch": 1.1756919374247894,
      "grad_norm": 5.811868667602539,
      "learning_rate": 0.00019998484124680025,
      "loss": 2.0779,
      "step": 9770
    },
    {
      "epoch": 1.1768953068592058,
      "grad_norm": 3.0623340606689453,
      "learning_rate": 0.00019998477493730573,
      "loss": 2.2344,
      "step": 9780
    },
    {
      "epoch": 1.178098676293622,
      "grad_norm": 5.205389022827148,
      "learning_rate": 0.0001999847084831086,
      "loss": 2.041,
      "step": 9790
    },
    {
      "epoch": 1.1793020457280385,
      "grad_norm": 6.058391094207764,
      "learning_rate": 0.0001999846418842089,
      "loss": 2.0143,
      "step": 9800
    },
    {
      "epoch": 1.1805054151624548,
      "grad_norm": 7.069987773895264,
      "learning_rate": 0.0001999845751406067,
      "loss": 2.1039,
      "step": 9810
    },
    {
      "epoch": 1.1817087845968712,
      "grad_norm": 4.928779125213623,
      "learning_rate": 0.00019998450825230214,
      "loss": 1.8592,
      "step": 9820
    },
    {
      "epoch": 1.1829121540312877,
      "grad_norm": 3.740241765975952,
      "learning_rate": 0.00019998444121929534,
      "loss": 2.1091,
      "step": 9830
    },
    {
      "epoch": 1.184115523465704,
      "grad_norm": 9.451654434204102,
      "learning_rate": 0.00019998437404158636,
      "loss": 2.158,
      "step": 9840
    },
    {
      "epoch": 1.1853188929001204,
      "grad_norm": 4.026190280914307,
      "learning_rate": 0.0001999843067191753,
      "loss": 2.0388,
      "step": 9850
    },
    {
      "epoch": 1.1865222623345366,
      "grad_norm": 6.800570011138916,
      "learning_rate": 0.00019998423925206226,
      "loss": 2.0448,
      "step": 9860
    },
    {
      "epoch": 1.187725631768953,
      "grad_norm": 4.920462131500244,
      "learning_rate": 0.00019998417164024735,
      "loss": 1.8885,
      "step": 9870
    },
    {
      "epoch": 1.1889290012033695,
      "grad_norm": 3.1672539710998535,
      "learning_rate": 0.00019998410388373064,
      "loss": 1.9803,
      "step": 9880
    },
    {
      "epoch": 1.1901323706377858,
      "grad_norm": 7.906315803527832,
      "learning_rate": 0.00019998403598251225,
      "loss": 2.2293,
      "step": 9890
    },
    {
      "epoch": 1.1913357400722022,
      "grad_norm": 5.510423183441162,
      "learning_rate": 0.0001999839679365923,
      "loss": 2.0506,
      "step": 9900
    },
    {
      "epoch": 1.1925391095066185,
      "grad_norm": 5.803864002227783,
      "learning_rate": 0.00019998389974597086,
      "loss": 2.1935,
      "step": 9910
    },
    {
      "epoch": 1.193742478941035,
      "grad_norm": 5.4688310623168945,
      "learning_rate": 0.000199983831410648,
      "loss": 1.8745,
      "step": 9920
    },
    {
      "epoch": 1.1949458483754514,
      "grad_norm": 4.938477993011475,
      "learning_rate": 0.00019998376293062385,
      "loss": 2.1407,
      "step": 9930
    },
    {
      "epoch": 1.1961492178098676,
      "grad_norm": 5.239728927612305,
      "learning_rate": 0.00019998369430589854,
      "loss": 2.0674,
      "step": 9940
    },
    {
      "epoch": 1.197352587244284,
      "grad_norm": 4.851728439331055,
      "learning_rate": 0.00019998362553647213,
      "loss": 1.9441,
      "step": 9950
    },
    {
      "epoch": 1.1985559566787003,
      "grad_norm": 7.1798415184021,
      "learning_rate": 0.00019998355662234475,
      "loss": 1.9134,
      "step": 9960
    },
    {
      "epoch": 1.1997593261131168,
      "grad_norm": 5.0848188400268555,
      "learning_rate": 0.00019998348756351643,
      "loss": 1.9767,
      "step": 9970
    },
    {
      "epoch": 1.200962695547533,
      "grad_norm": 4.1638288497924805,
      "learning_rate": 0.00019998341835998735,
      "loss": 1.8312,
      "step": 9980
    },
    {
      "epoch": 1.2021660649819494,
      "grad_norm": 5.722476959228516,
      "learning_rate": 0.00019998334901175758,
      "loss": 1.9314,
      "step": 9990
    },
    {
      "epoch": 1.203369434416366,
      "grad_norm": 4.66650390625,
      "learning_rate": 0.0001999832795188272,
      "loss": 2.1877,
      "step": 10000
    },
    {
      "epoch": 1.2045728038507821,
      "grad_norm": 8.986128807067871,
      "learning_rate": 0.00019998320988119634,
      "loss": 2.2097,
      "step": 10010
    },
    {
      "epoch": 1.2057761732851986,
      "grad_norm": 5.32340669631958,
      "learning_rate": 0.00019998314009886508,
      "loss": 2.1554,
      "step": 10020
    },
    {
      "epoch": 1.2069795427196148,
      "grad_norm": 3.7524874210357666,
      "learning_rate": 0.00019998307017183355,
      "loss": 1.9542,
      "step": 10030
    },
    {
      "epoch": 1.2081829121540313,
      "grad_norm": 5.027027606964111,
      "learning_rate": 0.00019998300010010182,
      "loss": 2.1156,
      "step": 10040
    },
    {
      "epoch": 1.2093862815884477,
      "grad_norm": 3.9726953506469727,
      "learning_rate": 0.00019998292988367003,
      "loss": 1.9822,
      "step": 10050
    },
    {
      "epoch": 1.210589651022864,
      "grad_norm": 7.025004863739014,
      "learning_rate": 0.00019998285952253824,
      "loss": 2.0503,
      "step": 10060
    },
    {
      "epoch": 1.2117930204572804,
      "grad_norm": 4.493628025054932,
      "learning_rate": 0.00019998278901670654,
      "loss": 2.0269,
      "step": 10070
    },
    {
      "epoch": 1.2129963898916967,
      "grad_norm": 3.804781675338745,
      "learning_rate": 0.00019998271836617508,
      "loss": 2.0031,
      "step": 10080
    },
    {
      "epoch": 1.2141997593261131,
      "grad_norm": 5.55324649810791,
      "learning_rate": 0.00019998264757094397,
      "loss": 1.8584,
      "step": 10090
    },
    {
      "epoch": 1.2154031287605296,
      "grad_norm": 4.546861171722412,
      "learning_rate": 0.00019998257663101323,
      "loss": 1.9519,
      "step": 10100
    },
    {
      "epoch": 1.2166064981949458,
      "grad_norm": 6.647198677062988,
      "learning_rate": 0.00019998250554638305,
      "loss": 1.9068,
      "step": 10110
    },
    {
      "epoch": 1.2178098676293623,
      "grad_norm": 6.195066452026367,
      "learning_rate": 0.0001999824343170535,
      "loss": 1.8015,
      "step": 10120
    },
    {
      "epoch": 1.2190132370637785,
      "grad_norm": 3.3782405853271484,
      "learning_rate": 0.00019998236294302467,
      "loss": 2.2316,
      "step": 10130
    },
    {
      "epoch": 1.220216606498195,
      "grad_norm": 4.880107879638672,
      "learning_rate": 0.00019998229142429667,
      "loss": 1.9407,
      "step": 10140
    },
    {
      "epoch": 1.2214199759326112,
      "grad_norm": 4.86598539352417,
      "learning_rate": 0.00019998221976086963,
      "loss": 2.1939,
      "step": 10150
    },
    {
      "epoch": 1.2226233453670277,
      "grad_norm": 6.2081804275512695,
      "learning_rate": 0.00019998214795274361,
      "loss": 2.1162,
      "step": 10160
    },
    {
      "epoch": 1.2238267148014441,
      "grad_norm": 4.876759052276611,
      "learning_rate": 0.00019998207599991873,
      "loss": 1.8673,
      "step": 10170
    },
    {
      "epoch": 1.2250300842358604,
      "grad_norm": 3.492619752883911,
      "learning_rate": 0.00019998200390239514,
      "loss": 2.2299,
      "step": 10180
    },
    {
      "epoch": 1.2262334536702768,
      "grad_norm": 5.097721576690674,
      "learning_rate": 0.0001999819316601729,
      "loss": 1.9695,
      "step": 10190
    },
    {
      "epoch": 1.2274368231046933,
      "grad_norm": 4.120977401733398,
      "learning_rate": 0.00019998185927325212,
      "loss": 2.0585,
      "step": 10200
    },
    {
      "epoch": 1.2286401925391095,
      "grad_norm": 6.539756774902344,
      "learning_rate": 0.00019998178674163287,
      "loss": 2.2924,
      "step": 10210
    },
    {
      "epoch": 1.229843561973526,
      "grad_norm": 5.448577404022217,
      "learning_rate": 0.00019998171406531532,
      "loss": 1.9213,
      "step": 10220
    },
    {
      "epoch": 1.2310469314079422,
      "grad_norm": 4.067614555358887,
      "learning_rate": 0.00019998164124429956,
      "loss": 1.9187,
      "step": 10230
    },
    {
      "epoch": 1.2322503008423586,
      "grad_norm": 5.904695510864258,
      "learning_rate": 0.00019998156827858563,
      "loss": 2.2251,
      "step": 10240
    },
    {
      "epoch": 1.2334536702767749,
      "grad_norm": 3.8210763931274414,
      "learning_rate": 0.00019998149516817373,
      "loss": 1.8841,
      "step": 10250
    },
    {
      "epoch": 1.2346570397111913,
      "grad_norm": 5.61317777633667,
      "learning_rate": 0.00019998142191306388,
      "loss": 1.9608,
      "step": 10260
    },
    {
      "epoch": 1.2358604091456078,
      "grad_norm": 5.043588161468506,
      "learning_rate": 0.00019998134851325627,
      "loss": 2.0538,
      "step": 10270
    },
    {
      "epoch": 1.237063778580024,
      "grad_norm": 3.8707103729248047,
      "learning_rate": 0.00019998127496875093,
      "loss": 2.2254,
      "step": 10280
    },
    {
      "epoch": 1.2382671480144405,
      "grad_norm": 6.206582546234131,
      "learning_rate": 0.00019998120127954802,
      "loss": 1.947,
      "step": 10290
    },
    {
      "epoch": 1.2394705174488567,
      "grad_norm": 4.542145729064941,
      "learning_rate": 0.00019998112744564763,
      "loss": 2.0519,
      "step": 10300
    },
    {
      "epoch": 1.2406738868832732,
      "grad_norm": 6.438401699066162,
      "learning_rate": 0.00019998105346704986,
      "loss": 2.0104,
      "step": 10310
    },
    {
      "epoch": 1.2418772563176894,
      "grad_norm": 4.163731098175049,
      "learning_rate": 0.00019998097934375482,
      "loss": 2.0364,
      "step": 10320
    },
    {
      "epoch": 1.2430806257521059,
      "grad_norm": 3.883605480194092,
      "learning_rate": 0.00019998090507576262,
      "loss": 2.0702,
      "step": 10330
    },
    {
      "epoch": 1.2442839951865223,
      "grad_norm": 5.311922073364258,
      "learning_rate": 0.00019998083066307337,
      "loss": 2.0248,
      "step": 10340
    },
    {
      "epoch": 1.2454873646209386,
      "grad_norm": 4.283480167388916,
      "learning_rate": 0.0001999807561056872,
      "loss": 2.0493,
      "step": 10350
    },
    {
      "epoch": 1.246690734055355,
      "grad_norm": 6.248193264007568,
      "learning_rate": 0.00019998068140360413,
      "loss": 1.9874,
      "step": 10360
    },
    {
      "epoch": 1.2478941034897715,
      "grad_norm": 5.551257133483887,
      "learning_rate": 0.00019998060655682435,
      "loss": 2.0025,
      "step": 10370
    },
    {
      "epoch": 1.2490974729241877,
      "grad_norm": 2.876491069793701,
      "learning_rate": 0.00019998053156534797,
      "loss": 2.0228,
      "step": 10380
    },
    {
      "epoch": 1.2503008423586042,
      "grad_norm": 4.5422515869140625,
      "learning_rate": 0.00019998045642917505,
      "loss": 1.7866,
      "step": 10390
    },
    {
      "epoch": 1.2515042117930204,
      "grad_norm": 5.026985168457031,
      "learning_rate": 0.00019998038114830575,
      "loss": 1.7722,
      "step": 10400
    },
    {
      "epoch": 1.2527075812274369,
      "grad_norm": 7.3900909423828125,
      "learning_rate": 0.00019998030572274014,
      "loss": 2.1746,
      "step": 10410
    },
    {
      "epoch": 1.253910950661853,
      "grad_norm": 5.162228584289551,
      "learning_rate": 0.00019998023015247834,
      "loss": 1.858,
      "step": 10420
    },
    {
      "epoch": 1.2551143200962696,
      "grad_norm": 2.786094903945923,
      "learning_rate": 0.00019998015443752046,
      "loss": 1.9791,
      "step": 10430
    },
    {
      "epoch": 1.256317689530686,
      "grad_norm": 6.485592842102051,
      "learning_rate": 0.00019998007857786662,
      "loss": 2.0075,
      "step": 10440
    },
    {
      "epoch": 1.2575210589651022,
      "grad_norm": 4.226631164550781,
      "learning_rate": 0.00019998000257351694,
      "loss": 1.9714,
      "step": 10450
    },
    {
      "epoch": 1.2587244283995187,
      "grad_norm": 8.003928184509277,
      "learning_rate": 0.0001999799264244715,
      "loss": 2.0834,
      "step": 10460
    },
    {
      "epoch": 1.259927797833935,
      "grad_norm": 5.195785045623779,
      "learning_rate": 0.0001999798501307304,
      "loss": 1.7946,
      "step": 10470
    },
    {
      "epoch": 1.2611311672683514,
      "grad_norm": 3.2900009155273438,
      "learning_rate": 0.0001999797736922938,
      "loss": 2.2332,
      "step": 10480
    },
    {
      "epoch": 1.2623345367027676,
      "grad_norm": 7.40859317779541,
      "learning_rate": 0.00019997969710916177,
      "loss": 2.1342,
      "step": 10490
    },
    {
      "epoch": 1.263537906137184,
      "grad_norm": 5.1050004959106445,
      "learning_rate": 0.00019997962038133444,
      "loss": 1.8853,
      "step": 10500
    },
    {
      "epoch": 1.2647412755716005,
      "grad_norm": 6.883108139038086,
      "learning_rate": 0.0001999795435088119,
      "loss": 2.1931,
      "step": 10510
    },
    {
      "epoch": 1.2659446450060168,
      "grad_norm": 5.1271281242370605,
      "learning_rate": 0.0001999794664915943,
      "loss": 1.9036,
      "step": 10520
    },
    {
      "epoch": 1.2671480144404332,
      "grad_norm": 3.9865400791168213,
      "learning_rate": 0.00019997938932968172,
      "loss": 2.1866,
      "step": 10530
    },
    {
      "epoch": 1.2683513838748497,
      "grad_norm": 5.753084182739258,
      "learning_rate": 0.00019997931202307427,
      "loss": 2.1757,
      "step": 10540
    },
    {
      "epoch": 1.269554753309266,
      "grad_norm": 4.798121929168701,
      "learning_rate": 0.0001999792345717721,
      "loss": 2.052,
      "step": 10550
    },
    {
      "epoch": 1.2707581227436824,
      "grad_norm": 6.817185401916504,
      "learning_rate": 0.00019997915697577526,
      "loss": 2.0153,
      "step": 10560
    },
    {
      "epoch": 1.2719614921780986,
      "grad_norm": 4.873058795928955,
      "learning_rate": 0.0001999790792350839,
      "loss": 2.3704,
      "step": 10570
    },
    {
      "epoch": 1.273164861612515,
      "grad_norm": 3.428636074066162,
      "learning_rate": 0.00019997900134969814,
      "loss": 2.3233,
      "step": 10580
    },
    {
      "epoch": 1.2743682310469313,
      "grad_norm": 6.904481887817383,
      "learning_rate": 0.00019997892331961807,
      "loss": 1.882,
      "step": 10590
    },
    {
      "epoch": 1.2755716004813478,
      "grad_norm": 4.372970104217529,
      "learning_rate": 0.00019997884514484385,
      "loss": 2.1738,
      "step": 10600
    },
    {
      "epoch": 1.2767749699157642,
      "grad_norm": 6.636587619781494,
      "learning_rate": 0.0001999787668253755,
      "loss": 2.4223,
      "step": 10610
    },
    {
      "epoch": 1.2779783393501805,
      "grad_norm": 5.977717876434326,
      "learning_rate": 0.00019997868836121323,
      "loss": 1.9153,
      "step": 10620
    },
    {
      "epoch": 1.279181708784597,
      "grad_norm": 3.2894933223724365,
      "learning_rate": 0.0001999786097523571,
      "loss": 2.2322,
      "step": 10630
    },
    {
      "epoch": 1.2803850782190134,
      "grad_norm": 6.185766696929932,
      "learning_rate": 0.00019997853099880722,
      "loss": 1.8874,
      "step": 10640
    },
    {
      "epoch": 1.2815884476534296,
      "grad_norm": 4.085166931152344,
      "learning_rate": 0.00019997845210056373,
      "loss": 1.968,
      "step": 10650
    },
    {
      "epoch": 1.2827918170878458,
      "grad_norm": 6.705855369567871,
      "learning_rate": 0.0001999783730576268,
      "loss": 1.9561,
      "step": 10660
    },
    {
      "epoch": 1.2839951865222623,
      "grad_norm": 5.585948467254639,
      "learning_rate": 0.00019997829386999641,
      "loss": 2.0039,
      "step": 10670
    },
    {
      "epoch": 1.2851985559566788,
      "grad_norm": 3.6671464443206787,
      "learning_rate": 0.00019997821453767274,
      "loss": 2.0347,
      "step": 10680
    },
    {
      "epoch": 1.286401925391095,
      "grad_norm": 4.681477069854736,
      "learning_rate": 0.00019997813506065595,
      "loss": 1.9104,
      "step": 10690
    },
    {
      "epoch": 1.2876052948255114,
      "grad_norm": 5.2901387214660645,
      "learning_rate": 0.0001999780554389461,
      "loss": 1.9701,
      "step": 10700
    },
    {
      "epoch": 1.288808664259928,
      "grad_norm": 5.807519435882568,
      "learning_rate": 0.0001999779756725433,
      "loss": 1.8904,
      "step": 10710
    },
    {
      "epoch": 1.2900120336943441,
      "grad_norm": 4.371450424194336,
      "learning_rate": 0.0001999778957614477,
      "loss": 1.6964,
      "step": 10720
    },
    {
      "epoch": 1.2912154031287606,
      "grad_norm": 4.2103590965271,
      "learning_rate": 0.0001999778157056594,
      "loss": 2.2595,
      "step": 10730
    },
    {
      "epoch": 1.2924187725631768,
      "grad_norm": 6.793369293212891,
      "learning_rate": 0.00019997773550517854,
      "loss": 2.0873,
      "step": 10740
    },
    {
      "epoch": 1.2936221419975933,
      "grad_norm": 5.259834289550781,
      "learning_rate": 0.00019997765516000519,
      "loss": 2.1054,
      "step": 10750
    },
    {
      "epoch": 1.2948255114320095,
      "grad_norm": 6.473832130432129,
      "learning_rate": 0.0001999775746701395,
      "loss": 1.9038,
      "step": 10760
    },
    {
      "epoch": 1.296028880866426,
      "grad_norm": 5.644038200378418,
      "learning_rate": 0.00019997749403558154,
      "loss": 1.9461,
      "step": 10770
    },
    {
      "epoch": 1.2972322503008424,
      "grad_norm": 3.3906755447387695,
      "learning_rate": 0.0001999774132563315,
      "loss": 1.9609,
      "step": 10780
    },
    {
      "epoch": 1.2984356197352587,
      "grad_norm": 6.086814880371094,
      "learning_rate": 0.00019997733233238947,
      "loss": 2.0266,
      "step": 10790
    },
    {
      "epoch": 1.2996389891696751,
      "grad_norm": 3.8884189128875732,
      "learning_rate": 0.00019997725126375552,
      "loss": 1.8661,
      "step": 10800
    },
    {
      "epoch": 1.3008423586040916,
      "grad_norm": 5.768379211425781,
      "learning_rate": 0.00019997717005042982,
      "loss": 2.0649,
      "step": 10810
    },
    {
      "epoch": 1.3020457280385078,
      "grad_norm": 4.84769344329834,
      "learning_rate": 0.00019997708869241246,
      "loss": 2.1423,
      "step": 10820
    },
    {
      "epoch": 1.303249097472924,
      "grad_norm": 4.503974437713623,
      "learning_rate": 0.0001999770071897036,
      "loss": 1.9141,
      "step": 10830
    },
    {
      "epoch": 1.3044524669073405,
      "grad_norm": 5.6907267570495605,
      "learning_rate": 0.00019997692554230332,
      "loss": 1.7882,
      "step": 10840
    },
    {
      "epoch": 1.305655836341757,
      "grad_norm": 4.08046293258667,
      "learning_rate": 0.00019997684375021172,
      "loss": 2.0279,
      "step": 10850
    },
    {
      "epoch": 1.3068592057761732,
      "grad_norm": 6.6718430519104,
      "learning_rate": 0.00019997676181342895,
      "loss": 2.0306,
      "step": 10860
    },
    {
      "epoch": 1.3080625752105897,
      "grad_norm": 5.364537715911865,
      "learning_rate": 0.00019997667973195514,
      "loss": 1.7702,
      "step": 10870
    },
    {
      "epoch": 1.3092659446450061,
      "grad_norm": 4.512383460998535,
      "learning_rate": 0.00019997659750579038,
      "loss": 1.9526,
      "step": 10880
    },
    {
      "epoch": 1.3104693140794224,
      "grad_norm": 5.895017147064209,
      "learning_rate": 0.0001999765151349348,
      "loss": 1.8212,
      "step": 10890
    },
    {
      "epoch": 1.3116726835138388,
      "grad_norm": 4.3670125007629395,
      "learning_rate": 0.0001999764326193885,
      "loss": 1.8725,
      "step": 10900
    },
    {
      "epoch": 1.312876052948255,
      "grad_norm": 6.285079002380371,
      "learning_rate": 0.00019997634995915168,
      "loss": 2.2052,
      "step": 10910
    },
    {
      "epoch": 1.3140794223826715,
      "grad_norm": 5.09568977355957,
      "learning_rate": 0.00019997626715422432,
      "loss": 1.7704,
      "step": 10920
    },
    {
      "epoch": 1.3152827918170877,
      "grad_norm": 3.565420389175415,
      "learning_rate": 0.00019997618420460666,
      "loss": 2.2321,
      "step": 10930
    },
    {
      "epoch": 1.3164861612515042,
      "grad_norm": 5.460746765136719,
      "learning_rate": 0.00019997610111029877,
      "loss": 2.1075,
      "step": 10940
    },
    {
      "epoch": 1.3176895306859207,
      "grad_norm": 4.214285373687744,
      "learning_rate": 0.0001999760178713008,
      "loss": 1.9047,
      "step": 10950
    },
    {
      "epoch": 1.3188929001203369,
      "grad_norm": 5.36951208114624,
      "learning_rate": 0.00019997593448761283,
      "loss": 2.1852,
      "step": 10960
    },
    {
      "epoch": 1.3200962695547533,
      "grad_norm": 6.230038166046143,
      "learning_rate": 0.000199975850959235,
      "loss": 2.2681,
      "step": 10970
    },
    {
      "epoch": 1.3212996389891698,
      "grad_norm": 3.9167563915252686,
      "learning_rate": 0.00019997576728616743,
      "loss": 1.9569,
      "step": 10980
    },
    {
      "epoch": 1.322503008423586,
      "grad_norm": 6.026401042938232,
      "learning_rate": 0.00019997568346841025,
      "loss": 1.9088,
      "step": 10990
    },
    {
      "epoch": 1.3237063778580023,
      "grad_norm": 4.840241432189941,
      "learning_rate": 0.00019997559950596356,
      "loss": 2.1188,
      "step": 11000
    },
    {
      "epoch": 1.3249097472924187,
      "grad_norm": 6.2873969078063965,
      "learning_rate": 0.0001999755153988275,
      "loss": 2.1796,
      "step": 11010
    },
    {
      "epoch": 1.3261131167268352,
      "grad_norm": 5.201504707336426,
      "learning_rate": 0.00019997543114700216,
      "loss": 1.8979,
      "step": 11020
    },
    {
      "epoch": 1.3273164861612514,
      "grad_norm": 4.130815505981445,
      "learning_rate": 0.00019997534675048772,
      "loss": 2.1705,
      "step": 11030
    },
    {
      "epoch": 1.3285198555956679,
      "grad_norm": 6.854120254516602,
      "learning_rate": 0.00019997526220928426,
      "loss": 2.1597,
      "step": 11040
    },
    {
      "epoch": 1.3297232250300843,
      "grad_norm": 3.4538803100585938,
      "learning_rate": 0.00019997517752339192,
      "loss": 1.741,
      "step": 11050
    },
    {
      "epoch": 1.3309265944645006,
      "grad_norm": 5.753960609436035,
      "learning_rate": 0.00019997509269281083,
      "loss": 2.0376,
      "step": 11060
    },
    {
      "epoch": 1.332129963898917,
      "grad_norm": 4.628495216369629,
      "learning_rate": 0.00019997500771754107,
      "loss": 1.7842,
      "step": 11070
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 3.723304271697998,
      "learning_rate": 0.00019997492259758278,
      "loss": 2.2054,
      "step": 11080
    },
    {
      "epoch": 1.3345367027677497,
      "grad_norm": 5.684840202331543,
      "learning_rate": 0.00019997483733293612,
      "loss": 1.8107,
      "step": 11090
    },
    {
      "epoch": 1.335740072202166,
      "grad_norm": 4.158824443817139,
      "learning_rate": 0.0001999747519236012,
      "loss": 1.9651,
      "step": 11100
    },
    {
      "epoch": 1.3369434416365824,
      "grad_norm": 6.488555431365967,
      "learning_rate": 0.00019997466636957812,
      "loss": 2.1014,
      "step": 11110
    },
    {
      "epoch": 1.3381468110709989,
      "grad_norm": 4.762753009796143,
      "learning_rate": 0.00019997458067086697,
      "loss": 2.0527,
      "step": 11120
    },
    {
      "epoch": 1.339350180505415,
      "grad_norm": 3.0060322284698486,
      "learning_rate": 0.00019997449482746796,
      "loss": 2.0986,
      "step": 11130
    },
    {
      "epoch": 1.3405535499398316,
      "grad_norm": 5.863572120666504,
      "learning_rate": 0.00019997440883938117,
      "loss": 2.0484,
      "step": 11140
    },
    {
      "epoch": 1.341756919374248,
      "grad_norm": 4.870255947113037,
      "learning_rate": 0.00019997432270660675,
      "loss": 2.0285,
      "step": 11150
    },
    {
      "epoch": 1.3429602888086642,
      "grad_norm": 6.171783924102783,
      "learning_rate": 0.00019997423642914477,
      "loss": 1.9961,
      "step": 11160
    },
    {
      "epoch": 1.3441636582430807,
      "grad_norm": 5.617040157318115,
      "learning_rate": 0.00019997415000699538,
      "loss": 2.1632,
      "step": 11170
    },
    {
      "epoch": 1.345367027677497,
      "grad_norm": 3.242630958557129,
      "learning_rate": 0.00019997406344015873,
      "loss": 1.8618,
      "step": 11180
    },
    {
      "epoch": 1.3465703971119134,
      "grad_norm": 6.868119239807129,
      "learning_rate": 0.00019997397672863492,
      "loss": 1.9079,
      "step": 11190
    },
    {
      "epoch": 1.3477737665463296,
      "grad_norm": 4.686202526092529,
      "learning_rate": 0.0001999738898724241,
      "loss": 2.25,
      "step": 11200
    },
    {
      "epoch": 1.348977135980746,
      "grad_norm": 5.856929779052734,
      "learning_rate": 0.00019997380287152636,
      "loss": 2.1734,
      "step": 11210
    },
    {
      "epoch": 1.3501805054151625,
      "grad_norm": 4.6299872398376465,
      "learning_rate": 0.00019997371572594186,
      "loss": 1.8238,
      "step": 11220
    },
    {
      "epoch": 1.3513838748495788,
      "grad_norm": 3.755962610244751,
      "learning_rate": 0.00019997362843567068,
      "loss": 1.9503,
      "step": 11230
    },
    {
      "epoch": 1.3525872442839952,
      "grad_norm": 5.344730377197266,
      "learning_rate": 0.00019997354100071302,
      "loss": 1.9743,
      "step": 11240
    },
    {
      "epoch": 1.3537906137184115,
      "grad_norm": 4.1081953048706055,
      "learning_rate": 0.0001999734534210689,
      "loss": 1.8849,
      "step": 11250
    },
    {
      "epoch": 1.354993983152828,
      "grad_norm": 7.932389736175537,
      "learning_rate": 0.0001999733656967386,
      "loss": 2.0345,
      "step": 11260
    },
    {
      "epoch": 1.3561973525872442,
      "grad_norm": 6.176091194152832,
      "learning_rate": 0.00019997327782772208,
      "loss": 1.8347,
      "step": 11270
    },
    {
      "epoch": 1.3574007220216606,
      "grad_norm": 4.27577543258667,
      "learning_rate": 0.00019997318981401958,
      "loss": 2.1063,
      "step": 11280
    },
    {
      "epoch": 1.358604091456077,
      "grad_norm": 5.581092357635498,
      "learning_rate": 0.0001999731016556312,
      "loss": 1.9629,
      "step": 11290
    },
    {
      "epoch": 1.3598074608904933,
      "grad_norm": 5.33603048324585,
      "learning_rate": 0.00019997301335255703,
      "loss": 2.0722,
      "step": 11300
    },
    {
      "epoch": 1.3610108303249098,
      "grad_norm": 6.872206687927246,
      "learning_rate": 0.00019997292490479723,
      "loss": 2.0029,
      "step": 11310
    },
    {
      "epoch": 1.3622141997593262,
      "grad_norm": 5.270725250244141,
      "learning_rate": 0.00019997283631235195,
      "loss": 2.0149,
      "step": 11320
    },
    {
      "epoch": 1.3634175691937425,
      "grad_norm": 3.964169502258301,
      "learning_rate": 0.0001999727475752213,
      "loss": 1.9302,
      "step": 11330
    },
    {
      "epoch": 1.364620938628159,
      "grad_norm": 4.73685359954834,
      "learning_rate": 0.00019997265869340538,
      "loss": 2.0308,
      "step": 11340
    },
    {
      "epoch": 1.3658243080625752,
      "grad_norm": 4.951168060302734,
      "learning_rate": 0.0001999725696669043,
      "loss": 1.9565,
      "step": 11350
    },
    {
      "epoch": 1.3670276774969916,
      "grad_norm": 6.389776229858398,
      "learning_rate": 0.0001999724804957183,
      "loss": 2.0621,
      "step": 11360
    },
    {
      "epoch": 1.3682310469314078,
      "grad_norm": 5.1416473388671875,
      "learning_rate": 0.00019997239117984741,
      "loss": 2.0558,
      "step": 11370
    },
    {
      "epoch": 1.3694344163658243,
      "grad_norm": 3.951632499694824,
      "learning_rate": 0.00019997230171929178,
      "loss": 1.9674,
      "step": 11380
    },
    {
      "epoch": 1.3706377858002408,
      "grad_norm": 6.3243184089660645,
      "learning_rate": 0.00019997221211405156,
      "loss": 2.1929,
      "step": 11390
    },
    {
      "epoch": 1.371841155234657,
      "grad_norm": 4.366563320159912,
      "learning_rate": 0.0001999721223641269,
      "loss": 2.033,
      "step": 11400
    },
    {
      "epoch": 1.3730445246690735,
      "grad_norm": 7.604349136352539,
      "learning_rate": 0.00019997203246951783,
      "loss": 2.2521,
      "step": 11410
    },
    {
      "epoch": 1.37424789410349,
      "grad_norm": 5.672321319580078,
      "learning_rate": 0.00019997194243022456,
      "loss": 2.0388,
      "step": 11420
    },
    {
      "epoch": 1.3754512635379061,
      "grad_norm": 3.863300085067749,
      "learning_rate": 0.00019997185224624722,
      "loss": 1.8918,
      "step": 11430
    },
    {
      "epoch": 1.3766546329723224,
      "grad_norm": 5.572450160980225,
      "learning_rate": 0.00019997176191758595,
      "loss": 1.9658,
      "step": 11440
    },
    {
      "epoch": 1.3778580024067388,
      "grad_norm": 4.11118745803833,
      "learning_rate": 0.00019997167144424082,
      "loss": 1.8633,
      "step": 11450
    },
    {
      "epoch": 1.3790613718411553,
      "grad_norm": 5.848665237426758,
      "learning_rate": 0.000199971580826212,
      "loss": 2.0335,
      "step": 11460
    },
    {
      "epoch": 1.3802647412755715,
      "grad_norm": 4.966587066650391,
      "learning_rate": 0.00019997149006349964,
      "loss": 2.0401,
      "step": 11470
    },
    {
      "epoch": 1.381468110709988,
      "grad_norm": 4.319188117980957,
      "learning_rate": 0.00019997139915610385,
      "loss": 2.2575,
      "step": 11480
    },
    {
      "epoch": 1.3826714801444044,
      "grad_norm": 6.347306251525879,
      "learning_rate": 0.00019997130810402478,
      "loss": 2.1662,
      "step": 11490
    },
    {
      "epoch": 1.3838748495788207,
      "grad_norm": 4.402041435241699,
      "learning_rate": 0.00019997121690726254,
      "loss": 2.0651,
      "step": 11500
    },
    {
      "epoch": 1.3850782190132371,
      "grad_norm": 6.501856327056885,
      "learning_rate": 0.00019997112556581725,
      "loss": 2.3038,
      "step": 11510
    },
    {
      "epoch": 1.3862815884476534,
      "grad_norm": 5.262118339538574,
      "learning_rate": 0.00019997103407968906,
      "loss": 2.0157,
      "step": 11520
    },
    {
      "epoch": 1.3874849578820698,
      "grad_norm": 3.0442757606506348,
      "learning_rate": 0.0001999709424488781,
      "loss": 1.7683,
      "step": 11530
    },
    {
      "epoch": 1.388688327316486,
      "grad_norm": 5.854463577270508,
      "learning_rate": 0.00019997085067338451,
      "loss": 1.9958,
      "step": 11540
    },
    {
      "epoch": 1.3898916967509025,
      "grad_norm": 4.829777717590332,
      "learning_rate": 0.00019997075875320843,
      "loss": 2.06,
      "step": 11550
    },
    {
      "epoch": 1.391095066185319,
      "grad_norm": 7.081327438354492,
      "learning_rate": 0.00019997066668834995,
      "loss": 1.9274,
      "step": 11560
    },
    {
      "epoch": 1.3922984356197352,
      "grad_norm": 6.35081672668457,
      "learning_rate": 0.00019997057447880927,
      "loss": 2.1442,
      "step": 11570
    },
    {
      "epoch": 1.3935018050541517,
      "grad_norm": 3.7289793491363525,
      "learning_rate": 0.00019997048212458645,
      "loss": 1.9846,
      "step": 11580
    },
    {
      "epoch": 1.3947051744885681,
      "grad_norm": 6.378157615661621,
      "learning_rate": 0.00019997038962568165,
      "loss": 2.0232,
      "step": 11590
    },
    {
      "epoch": 1.3959085439229844,
      "grad_norm": 4.062281131744385,
      "learning_rate": 0.00019997029698209508,
      "loss": 1.9276,
      "step": 11600
    },
    {
      "epoch": 1.3971119133574006,
      "grad_norm": 7.422173976898193,
      "learning_rate": 0.00019997020419382678,
      "loss": 2.1896,
      "step": 11610
    },
    {
      "epoch": 1.398315282791817,
      "grad_norm": 4.6294355392456055,
      "learning_rate": 0.0001999701112608769,
      "loss": 1.7339,
      "step": 11620
    },
    {
      "epoch": 1.3995186522262335,
      "grad_norm": 3.527700662612915,
      "learning_rate": 0.00019997001818324557,
      "loss": 1.9711,
      "step": 11630
    },
    {
      "epoch": 1.4007220216606497,
      "grad_norm": 5.8211445808410645,
      "learning_rate": 0.00019996992496093297,
      "loss": 2.1136,
      "step": 11640
    },
    {
      "epoch": 1.4019253910950662,
      "grad_norm": 3.9967219829559326,
      "learning_rate": 0.0001999698315939392,
      "loss": 1.9309,
      "step": 11650
    },
    {
      "epoch": 1.4031287605294827,
      "grad_norm": 6.780445575714111,
      "learning_rate": 0.0001999697380822644,
      "loss": 1.9845,
      "step": 11660
    },
    {
      "epoch": 1.404332129963899,
      "grad_norm": 5.2751545906066895,
      "learning_rate": 0.0001999696444259087,
      "loss": 1.8692,
      "step": 11670
    },
    {
      "epoch": 1.4055354993983153,
      "grad_norm": 3.81404709815979,
      "learning_rate": 0.00019996955062487226,
      "loss": 2.137,
      "step": 11680
    },
    {
      "epoch": 1.4067388688327316,
      "grad_norm": 5.983846664428711,
      "learning_rate": 0.0001999694566791552,
      "loss": 2.0787,
      "step": 11690
    },
    {
      "epoch": 1.407942238267148,
      "grad_norm": 4.335493087768555,
      "learning_rate": 0.0001999693625887576,
      "loss": 2.0673,
      "step": 11700
    },
    {
      "epoch": 1.4091456077015643,
      "grad_norm": 7.844127655029297,
      "learning_rate": 0.0001999692683536797,
      "loss": 2.2025,
      "step": 11710
    },
    {
      "epoch": 1.4103489771359807,
      "grad_norm": 4.600785255432129,
      "learning_rate": 0.0001999691739739216,
      "loss": 1.9487,
      "step": 11720
    },
    {
      "epoch": 1.4115523465703972,
      "grad_norm": 4.707550525665283,
      "learning_rate": 0.0001999690794494834,
      "loss": 1.9785,
      "step": 11730
    },
    {
      "epoch": 1.4127557160048134,
      "grad_norm": 6.067191123962402,
      "learning_rate": 0.00019996898478036527,
      "loss": 1.9159,
      "step": 11740
    },
    {
      "epoch": 1.4139590854392299,
      "grad_norm": 3.8539209365844727,
      "learning_rate": 0.00019996888996656734,
      "loss": 2.062,
      "step": 11750
    },
    {
      "epoch": 1.4151624548736463,
      "grad_norm": 6.2311248779296875,
      "learning_rate": 0.00019996879500808974,
      "loss": 2.0352,
      "step": 11760
    },
    {
      "epoch": 1.4163658243080626,
      "grad_norm": 4.776411533355713,
      "learning_rate": 0.00019996869990493262,
      "loss": 1.9179,
      "step": 11770
    },
    {
      "epoch": 1.4175691937424788,
      "grad_norm": 4.044094562530518,
      "learning_rate": 0.00019996860465709606,
      "loss": 2.0612,
      "step": 11780
    },
    {
      "epoch": 1.4187725631768953,
      "grad_norm": 5.775789260864258,
      "learning_rate": 0.00019996850926458032,
      "loss": 1.9522,
      "step": 11790
    },
    {
      "epoch": 1.4199759326113117,
      "grad_norm": 4.458516597747803,
      "learning_rate": 0.0001999684137273854,
      "loss": 2.096,
      "step": 11800
    },
    {
      "epoch": 1.421179302045728,
      "grad_norm": 5.984865665435791,
      "learning_rate": 0.00019996831804551155,
      "loss": 2.133,
      "step": 11810
    },
    {
      "epoch": 1.4223826714801444,
      "grad_norm": 5.28660249710083,
      "learning_rate": 0.00019996822221895886,
      "loss": 2.02,
      "step": 11820
    },
    {
      "epoch": 1.4235860409145609,
      "grad_norm": 3.703310251235962,
      "learning_rate": 0.00019996812624772746,
      "loss": 2.0494,
      "step": 11830
    },
    {
      "epoch": 1.424789410348977,
      "grad_norm": 5.994476795196533,
      "learning_rate": 0.0001999680301318175,
      "loss": 1.7702,
      "step": 11840
    },
    {
      "epoch": 1.4259927797833936,
      "grad_norm": 4.5734405517578125,
      "learning_rate": 0.00019996793387122913,
      "loss": 1.9721,
      "step": 11850
    },
    {
      "epoch": 1.4271961492178098,
      "grad_norm": 6.583423137664795,
      "learning_rate": 0.00019996783746596247,
      "loss": 1.9945,
      "step": 11860
    },
    {
      "epoch": 1.4283995186522263,
      "grad_norm": 4.987894058227539,
      "learning_rate": 0.00019996774091601764,
      "loss": 2.0069,
      "step": 11870
    },
    {
      "epoch": 1.4296028880866425,
      "grad_norm": 3.197599172592163,
      "learning_rate": 0.00019996764422139482,
      "loss": 2.0022,
      "step": 11880
    },
    {
      "epoch": 1.430806257521059,
      "grad_norm": 4.9459991455078125,
      "learning_rate": 0.00019996754738209416,
      "loss": 1.9842,
      "step": 11890
    },
    {
      "epoch": 1.4320096269554754,
      "grad_norm": 4.139046669006348,
      "learning_rate": 0.00019996745039811578,
      "loss": 2.0101,
      "step": 11900
    },
    {
      "epoch": 1.4332129963898916,
      "grad_norm": 5.977879047393799,
      "learning_rate": 0.00019996735326945978,
      "loss": 2.1852,
      "step": 11910
    },
    {
      "epoch": 1.434416365824308,
      "grad_norm": 4.959321022033691,
      "learning_rate": 0.0001999672559961264,
      "loss": 1.9614,
      "step": 11920
    },
    {
      "epoch": 1.4356197352587245,
      "grad_norm": 3.7874855995178223,
      "learning_rate": 0.00019996715857811568,
      "loss": 2.0799,
      "step": 11930
    },
    {
      "epoch": 1.4368231046931408,
      "grad_norm": 6.308355331420898,
      "learning_rate": 0.0001999670610154278,
      "loss": 2.1004,
      "step": 11940
    },
    {
      "epoch": 1.4380264741275572,
      "grad_norm": 4.808499813079834,
      "learning_rate": 0.0001999669633080629,
      "loss": 2.0538,
      "step": 11950
    },
    {
      "epoch": 1.4392298435619735,
      "grad_norm": 6.995471477508545,
      "learning_rate": 0.00019996686545602113,
      "loss": 2.045,
      "step": 11960
    },
    {
      "epoch": 1.44043321299639,
      "grad_norm": 4.54448127746582,
      "learning_rate": 0.00019996676745930264,
      "loss": 1.9182,
      "step": 11970
    },
    {
      "epoch": 1.4416365824308062,
      "grad_norm": 3.7178962230682373,
      "learning_rate": 0.00019996666931790754,
      "loss": 1.9286,
      "step": 11980
    },
    {
      "epoch": 1.4428399518652226,
      "grad_norm": 4.949057102203369,
      "learning_rate": 0.00019996657103183598,
      "loss": 1.9982,
      "step": 11990
    },
    {
      "epoch": 1.444043321299639,
      "grad_norm": 4.729292869567871,
      "learning_rate": 0.0001999664726010881,
      "loss": 2.0252,
      "step": 12000
    },
    {
      "epoch": 1.4452466907340553,
      "grad_norm": 7.523972511291504,
      "learning_rate": 0.00019996637402566408,
      "loss": 1.9543,
      "step": 12010
    },
    {
      "epoch": 1.4464500601684718,
      "grad_norm": 4.342970848083496,
      "learning_rate": 0.00019996627530556403,
      "loss": 1.7837,
      "step": 12020
    },
    {
      "epoch": 1.447653429602888,
      "grad_norm": 3.445206642150879,
      "learning_rate": 0.00019996617644078813,
      "loss": 2.2453,
      "step": 12030
    },
    {
      "epoch": 1.4488567990373045,
      "grad_norm": 6.204970836639404,
      "learning_rate": 0.00019996607743133645,
      "loss": 2.159,
      "step": 12040
    },
    {
      "epoch": 1.4500601684717207,
      "grad_norm": 4.197068214416504,
      "learning_rate": 0.0001999659782772092,
      "loss": 1.8933,
      "step": 12050
    },
    {
      "epoch": 1.4512635379061372,
      "grad_norm": 7.213578224182129,
      "learning_rate": 0.00019996587897840647,
      "loss": 2.0458,
      "step": 12060
    },
    {
      "epoch": 1.4524669073405536,
      "grad_norm": 5.023141384124756,
      "learning_rate": 0.00019996577953492844,
      "loss": 2.1336,
      "step": 12070
    },
    {
      "epoch": 1.4536702767749698,
      "grad_norm": 2.6657612323760986,
      "learning_rate": 0.00019996567994677524,
      "loss": 2.043,
      "step": 12080
    },
    {
      "epoch": 1.4548736462093863,
      "grad_norm": 5.107360363006592,
      "learning_rate": 0.00019996558021394704,
      "loss": 2.0071,
      "step": 12090
    },
    {
      "epoch": 1.4560770156438028,
      "grad_norm": 4.615115165710449,
      "learning_rate": 0.00019996548033644396,
      "loss": 2.0489,
      "step": 12100
    },
    {
      "epoch": 1.457280385078219,
      "grad_norm": 7.913209438323975,
      "learning_rate": 0.00019996538031426614,
      "loss": 2.4675,
      "step": 12110
    },
    {
      "epoch": 1.4584837545126355,
      "grad_norm": 5.84035062789917,
      "learning_rate": 0.00019996528014741376,
      "loss": 2.0735,
      "step": 12120
    },
    {
      "epoch": 1.4596871239470517,
      "grad_norm": 4.441844940185547,
      "learning_rate": 0.0001999651798358869,
      "loss": 2.1426,
      "step": 12130
    },
    {
      "epoch": 1.4608904933814681,
      "grad_norm": 5.56676721572876,
      "learning_rate": 0.00019996507937968577,
      "loss": 2.0858,
      "step": 12140
    },
    {
      "epoch": 1.4620938628158844,
      "grad_norm": 4.233429431915283,
      "learning_rate": 0.00019996497877881048,
      "loss": 2.0057,
      "step": 12150
    },
    {
      "epoch": 1.4632972322503008,
      "grad_norm": 7.260183334350586,
      "learning_rate": 0.0001999648780332612,
      "loss": 2.0318,
      "step": 12160
    },
    {
      "epoch": 1.4645006016847173,
      "grad_norm": 5.844688415527344,
      "learning_rate": 0.00019996477714303806,
      "loss": 1.8661,
      "step": 12170
    },
    {
      "epoch": 1.4657039711191335,
      "grad_norm": 3.717721700668335,
      "learning_rate": 0.0001999646761081412,
      "loss": 2.1405,
      "step": 12180
    },
    {
      "epoch": 1.46690734055355,
      "grad_norm": 6.341575622558594,
      "learning_rate": 0.00019996457492857075,
      "loss": 2.1799,
      "step": 12190
    },
    {
      "epoch": 1.4681107099879662,
      "grad_norm": 3.2736339569091797,
      "learning_rate": 0.0001999644736043269,
      "loss": 1.8986,
      "step": 12200
    },
    {
      "epoch": 1.4693140794223827,
      "grad_norm": 6.800692558288574,
      "learning_rate": 0.00019996437213540977,
      "loss": 1.9837,
      "step": 12210
    },
    {
      "epoch": 1.470517448856799,
      "grad_norm": 5.0401201248168945,
      "learning_rate": 0.00019996427052181953,
      "loss": 1.9493,
      "step": 12220
    },
    {
      "epoch": 1.4717208182912154,
      "grad_norm": 3.1545989513397217,
      "learning_rate": 0.00019996416876355628,
      "loss": 2.1774,
      "step": 12230
    },
    {
      "epoch": 1.4729241877256318,
      "grad_norm": 6.337563991546631,
      "learning_rate": 0.00019996406686062024,
      "loss": 2.2252,
      "step": 12240
    },
    {
      "epoch": 1.474127557160048,
      "grad_norm": 4.547989368438721,
      "learning_rate": 0.00019996396481301146,
      "loss": 2.0871,
      "step": 12250
    },
    {
      "epoch": 1.4753309265944645,
      "grad_norm": 8.071065902709961,
      "learning_rate": 0.0001999638626207302,
      "loss": 1.9863,
      "step": 12260
    },
    {
      "epoch": 1.476534296028881,
      "grad_norm": 5.691819190979004,
      "learning_rate": 0.00019996376028377652,
      "loss": 2.1144,
      "step": 12270
    },
    {
      "epoch": 1.4777376654632972,
      "grad_norm": 4.698114395141602,
      "learning_rate": 0.0001999636578021506,
      "loss": 2.1699,
      "step": 12280
    },
    {
      "epoch": 1.4789410348977137,
      "grad_norm": 6.224732398986816,
      "learning_rate": 0.00019996355517585258,
      "loss": 2.1088,
      "step": 12290
    },
    {
      "epoch": 1.48014440433213,
      "grad_norm": 5.283705711364746,
      "learning_rate": 0.00019996345240488262,
      "loss": 2.0153,
      "step": 12300
    },
    {
      "epoch": 1.4813477737665464,
      "grad_norm": 6.128474712371826,
      "learning_rate": 0.0001999633494892409,
      "loss": 2.1222,
      "step": 12310
    },
    {
      "epoch": 1.4825511432009626,
      "grad_norm": 4.254839897155762,
      "learning_rate": 0.00019996324642892747,
      "loss": 1.9586,
      "step": 12320
    },
    {
      "epoch": 1.483754512635379,
      "grad_norm": 3.5428590774536133,
      "learning_rate": 0.0001999631432239426,
      "loss": 2.1966,
      "step": 12330
    },
    {
      "epoch": 1.4849578820697955,
      "grad_norm": 4.892740249633789,
      "learning_rate": 0.00019996303987428635,
      "loss": 2.0384,
      "step": 12340
    },
    {
      "epoch": 1.4861612515042117,
      "grad_norm": 4.1757659912109375,
      "learning_rate": 0.0001999629363799589,
      "loss": 2.1789,
      "step": 12350
    },
    {
      "epoch": 1.4873646209386282,
      "grad_norm": 6.969892978668213,
      "learning_rate": 0.00019996283274096042,
      "loss": 2.1338,
      "step": 12360
    },
    {
      "epoch": 1.4885679903730447,
      "grad_norm": 4.55242919921875,
      "learning_rate": 0.00019996272895729105,
      "loss": 2.2497,
      "step": 12370
    },
    {
      "epoch": 1.489771359807461,
      "grad_norm": 3.550419569015503,
      "learning_rate": 0.00019996262502895089,
      "loss": 2.0019,
      "step": 12380
    },
    {
      "epoch": 1.4909747292418771,
      "grad_norm": 6.1821770668029785,
      "learning_rate": 0.00019996252095594015,
      "loss": 1.9442,
      "step": 12390
    },
    {
      "epoch": 1.4921780986762936,
      "grad_norm": 4.020054817199707,
      "learning_rate": 0.00019996241673825898,
      "loss": 1.9674,
      "step": 12400
    },
    {
      "epoch": 1.49338146811071,
      "grad_norm": 6.744810581207275,
      "learning_rate": 0.00019996231237590748,
      "loss": 2.0089,
      "step": 12410
    },
    {
      "epoch": 1.4945848375451263,
      "grad_norm": 5.037810325622559,
      "learning_rate": 0.00019996220786888588,
      "loss": 1.956,
      "step": 12420
    },
    {
      "epoch": 1.4957882069795427,
      "grad_norm": 4.0546040534973145,
      "learning_rate": 0.00019996210321719424,
      "loss": 2.0901,
      "step": 12430
    },
    {
      "epoch": 1.4969915764139592,
      "grad_norm": 6.947364330291748,
      "learning_rate": 0.0001999619984208328,
      "loss": 1.9892,
      "step": 12440
    },
    {
      "epoch": 1.4981949458483754,
      "grad_norm": 5.34039831161499,
      "learning_rate": 0.00019996189347980164,
      "loss": 2.0415,
      "step": 12450
    },
    {
      "epoch": 1.4993983152827919,
      "grad_norm": 6.241419315338135,
      "learning_rate": 0.00019996178839410092,
      "loss": 1.9568,
      "step": 12460
    },
    {
      "epoch": 1.5006016847172083,
      "grad_norm": 4.005120754241943,
      "learning_rate": 0.00019996168316373085,
      "loss": 1.7521,
      "step": 12470
    },
    {
      "epoch": 1.5018050541516246,
      "grad_norm": 3.5194106101989746,
      "learning_rate": 0.0001999615777886915,
      "loss": 2.0599,
      "step": 12480
    },
    {
      "epoch": 1.5030084235860408,
      "grad_norm": 4.650184154510498,
      "learning_rate": 0.0001999614722689831,
      "loss": 2.0227,
      "step": 12490
    },
    {
      "epoch": 1.5042117930204573,
      "grad_norm": 4.675848007202148,
      "learning_rate": 0.0001999613666046058,
      "loss": 1.9764,
      "step": 12500
    },
    {
      "epoch": 1.5054151624548737,
      "grad_norm": 8.725130081176758,
      "learning_rate": 0.00019996126079555965,
      "loss": 2.1664,
      "step": 12510
    },
    {
      "epoch": 1.50661853188929,
      "grad_norm": 4.552273750305176,
      "learning_rate": 0.00019996115484184493,
      "loss": 1.8285,
      "step": 12520
    },
    {
      "epoch": 1.5078219013237064,
      "grad_norm": 3.9225306510925293,
      "learning_rate": 0.00019996104874346173,
      "loss": 1.884,
      "step": 12530
    },
    {
      "epoch": 1.5090252707581229,
      "grad_norm": 4.873247146606445,
      "learning_rate": 0.0001999609425004102,
      "loss": 2.0491,
      "step": 12540
    },
    {
      "epoch": 1.510228640192539,
      "grad_norm": 4.000176906585693,
      "learning_rate": 0.0001999608361126905,
      "loss": 1.8773,
      "step": 12550
    },
    {
      "epoch": 1.5114320096269553,
      "grad_norm": 6.219770908355713,
      "learning_rate": 0.00019996072958030283,
      "loss": 1.9446,
      "step": 12560
    },
    {
      "epoch": 1.5126353790613718,
      "grad_norm": 5.555866718292236,
      "learning_rate": 0.0001999606229032473,
      "loss": 1.8992,
      "step": 12570
    },
    {
      "epoch": 1.5138387484957883,
      "grad_norm": 4.3533220291137695,
      "learning_rate": 0.000199960516081524,
      "loss": 2.0626,
      "step": 12580
    },
    {
      "epoch": 1.5150421179302045,
      "grad_norm": 5.74338960647583,
      "learning_rate": 0.00019996040911513324,
      "loss": 1.8156,
      "step": 12590
    },
    {
      "epoch": 1.516245487364621,
      "grad_norm": 4.544538974761963,
      "learning_rate": 0.00019996030200407504,
      "loss": 1.8866,
      "step": 12600
    },
    {
      "epoch": 1.5174488567990374,
      "grad_norm": 7.063827991485596,
      "learning_rate": 0.00019996019474834964,
      "loss": 2.1647,
      "step": 12610
    },
    {
      "epoch": 1.5186522262334536,
      "grad_norm": 6.014230728149414,
      "learning_rate": 0.0001999600873479571,
      "loss": 1.9253,
      "step": 12620
    },
    {
      "epoch": 1.5198555956678699,
      "grad_norm": 4.151496410369873,
      "learning_rate": 0.00019995997980289768,
      "loss": 2.1063,
      "step": 12630
    },
    {
      "epoch": 1.5210589651022866,
      "grad_norm": 5.740432262420654,
      "learning_rate": 0.0001999598721131715,
      "loss": 1.9949,
      "step": 12640
    },
    {
      "epoch": 1.5222623345367028,
      "grad_norm": 5.5747575759887695,
      "learning_rate": 0.00019995976427877868,
      "loss": 2.0398,
      "step": 12650
    },
    {
      "epoch": 1.523465703971119,
      "grad_norm": 6.39547872543335,
      "learning_rate": 0.0001999596562997194,
      "loss": 2.1979,
      "step": 12660
    },
    {
      "epoch": 1.5246690734055355,
      "grad_norm": 5.9210734367370605,
      "learning_rate": 0.00019995954817599382,
      "loss": 2.1166,
      "step": 12670
    },
    {
      "epoch": 1.525872442839952,
      "grad_norm": 4.516720771789551,
      "learning_rate": 0.0001999594399076021,
      "loss": 1.9733,
      "step": 12680
    },
    {
      "epoch": 1.5270758122743682,
      "grad_norm": 6.413763999938965,
      "learning_rate": 0.0001999593314945444,
      "loss": 1.9363,
      "step": 12690
    },
    {
      "epoch": 1.5282791817087846,
      "grad_norm": 5.059680461883545,
      "learning_rate": 0.00019995922293682085,
      "loss": 2.1826,
      "step": 12700
    },
    {
      "epoch": 1.529482551143201,
      "grad_norm": 6.301856517791748,
      "learning_rate": 0.00019995911423443163,
      "loss": 2.1322,
      "step": 12710
    },
    {
      "epoch": 1.5306859205776173,
      "grad_norm": 4.676280498504639,
      "learning_rate": 0.00019995900538737692,
      "loss": 1.7236,
      "step": 12720
    },
    {
      "epoch": 1.5318892900120336,
      "grad_norm": 3.2905354499816895,
      "learning_rate": 0.0001999588963956568,
      "loss": 2.035,
      "step": 12730
    },
    {
      "epoch": 1.53309265944645,
      "grad_norm": 6.076344966888428,
      "learning_rate": 0.0001999587872592715,
      "loss": 1.8391,
      "step": 12740
    },
    {
      "epoch": 1.5342960288808665,
      "grad_norm": 3.6273529529571533,
      "learning_rate": 0.00019995867797822117,
      "loss": 1.7275,
      "step": 12750
    },
    {
      "epoch": 1.5354993983152827,
      "grad_norm": 6.787859916687012,
      "learning_rate": 0.00019995856855250594,
      "loss": 1.9494,
      "step": 12760
    },
    {
      "epoch": 1.5367027677496992,
      "grad_norm": 4.670495986938477,
      "learning_rate": 0.00019995845898212599,
      "loss": 1.9612,
      "step": 12770
    },
    {
      "epoch": 1.5379061371841156,
      "grad_norm": 3.980814218521118,
      "learning_rate": 0.00019995834926708148,
      "loss": 2.186,
      "step": 12780
    },
    {
      "epoch": 1.5391095066185319,
      "grad_norm": 5.668554306030273,
      "learning_rate": 0.00019995823940737256,
      "loss": 1.7496,
      "step": 12790
    },
    {
      "epoch": 1.5403128760529483,
      "grad_norm": 4.963620662689209,
      "learning_rate": 0.00019995812940299933,
      "loss": 2.0943,
      "step": 12800
    },
    {
      "epoch": 1.5415162454873648,
      "grad_norm": 6.285371780395508,
      "learning_rate": 0.00019995801925396206,
      "loss": 1.9992,
      "step": 12810
    },
    {
      "epoch": 1.542719614921781,
      "grad_norm": 5.616092681884766,
      "learning_rate": 0.00019995790896026088,
      "loss": 1.9886,
      "step": 12820
    },
    {
      "epoch": 1.5439229843561972,
      "grad_norm": 3.815455675125122,
      "learning_rate": 0.0001999577985218959,
      "loss": 2.0842,
      "step": 12830
    },
    {
      "epoch": 1.5451263537906137,
      "grad_norm": 4.692776203155518,
      "learning_rate": 0.00019995768793886729,
      "loss": 1.8275,
      "step": 12840
    },
    {
      "epoch": 1.5463297232250302,
      "grad_norm": 4.133147239685059,
      "learning_rate": 0.00019995757721117524,
      "loss": 1.8246,
      "step": 12850
    },
    {
      "epoch": 1.5475330926594464,
      "grad_norm": 9.024523735046387,
      "learning_rate": 0.00019995746633881992,
      "loss": 1.9761,
      "step": 12860
    },
    {
      "epoch": 1.5487364620938628,
      "grad_norm": 4.864393711090088,
      "learning_rate": 0.00019995735532180142,
      "loss": 1.9203,
      "step": 12870
    },
    {
      "epoch": 1.5499398315282793,
      "grad_norm": 3.4315037727355957,
      "learning_rate": 0.00019995724416011998,
      "loss": 2.3091,
      "step": 12880
    },
    {
      "epoch": 1.5511432009626955,
      "grad_norm": 6.324875354766846,
      "learning_rate": 0.0001999571328537757,
      "loss": 2.1473,
      "step": 12890
    },
    {
      "epoch": 1.5523465703971118,
      "grad_norm": 4.298543930053711,
      "learning_rate": 0.0001999570214027688,
      "loss": 1.9465,
      "step": 12900
    },
    {
      "epoch": 1.5535499398315282,
      "grad_norm": 6.035646915435791,
      "learning_rate": 0.00019995690980709942,
      "loss": 2.0224,
      "step": 12910
    },
    {
      "epoch": 1.5547533092659447,
      "grad_norm": 5.38960599899292,
      "learning_rate": 0.00019995679806676767,
      "loss": 2.0193,
      "step": 12920
    },
    {
      "epoch": 1.555956678700361,
      "grad_norm": 4.020500659942627,
      "learning_rate": 0.0001999566861817738,
      "loss": 1.9763,
      "step": 12930
    },
    {
      "epoch": 1.5571600481347774,
      "grad_norm": 5.028992176055908,
      "learning_rate": 0.00019995657415211793,
      "loss": 2.0487,
      "step": 12940
    },
    {
      "epoch": 1.5583634175691938,
      "grad_norm": 4.500495910644531,
      "learning_rate": 0.0001999564619778002,
      "loss": 1.9695,
      "step": 12950
    },
    {
      "epoch": 1.55956678700361,
      "grad_norm": 8.437773704528809,
      "learning_rate": 0.00019995634965882077,
      "loss": 2.2308,
      "step": 12960
    },
    {
      "epoch": 1.5607701564380265,
      "grad_norm": 5.600293159484863,
      "learning_rate": 0.00019995623719517983,
      "loss": 2.1186,
      "step": 12970
    },
    {
      "epoch": 1.561973525872443,
      "grad_norm": 4.004968643188477,
      "learning_rate": 0.00019995612458687757,
      "loss": 1.9492,
      "step": 12980
    },
    {
      "epoch": 1.5631768953068592,
      "grad_norm": 6.176262378692627,
      "learning_rate": 0.0001999560118339141,
      "loss": 1.9266,
      "step": 12990
    },
    {
      "epoch": 1.5643802647412755,
      "grad_norm": 6.919417858123779,
      "learning_rate": 0.00019995589893628962,
      "loss": 1.9525,
      "step": 13000
    },
    {
      "epoch": 1.565583634175692,
      "grad_norm": 5.738315105438232,
      "learning_rate": 0.00019995578589400423,
      "loss": 1.9941,
      "step": 13010
    },
    {
      "epoch": 1.5667870036101084,
      "grad_norm": 4.617683410644531,
      "learning_rate": 0.00019995567270705817,
      "loss": 1.8318,
      "step": 13020
    },
    {
      "epoch": 1.5679903730445246,
      "grad_norm": 3.524646043777466,
      "learning_rate": 0.00019995555937545156,
      "loss": 2.0668,
      "step": 13030
    },
    {
      "epoch": 1.569193742478941,
      "grad_norm": 5.225677967071533,
      "learning_rate": 0.00019995544589918458,
      "loss": 2.0323,
      "step": 13040
    },
    {
      "epoch": 1.5703971119133575,
      "grad_norm": 4.304656028747559,
      "learning_rate": 0.00019995533227825743,
      "loss": 2.0233,
      "step": 13050
    },
    {
      "epoch": 1.5716004813477737,
      "grad_norm": 6.721856117248535,
      "learning_rate": 0.00019995521851267018,
      "loss": 2.0863,
      "step": 13060
    },
    {
      "epoch": 1.57280385078219,
      "grad_norm": 5.3623857498168945,
      "learning_rate": 0.00019995510460242308,
      "loss": 1.8831,
      "step": 13070
    },
    {
      "epoch": 1.5740072202166067,
      "grad_norm": 4.196499824523926,
      "learning_rate": 0.00019995499054751626,
      "loss": 2.0412,
      "step": 13080
    },
    {
      "epoch": 1.575210589651023,
      "grad_norm": 4.95134973526001,
      "learning_rate": 0.0001999548763479499,
      "loss": 1.7917,
      "step": 13090
    },
    {
      "epoch": 1.5764139590854391,
      "grad_norm": 4.267271041870117,
      "learning_rate": 0.00019995476200372413,
      "loss": 1.9908,
      "step": 13100
    },
    {
      "epoch": 1.5776173285198556,
      "grad_norm": 5.7748517990112305,
      "learning_rate": 0.00019995464751483915,
      "loss": 2.0942,
      "step": 13110
    },
    {
      "epoch": 1.578820697954272,
      "grad_norm": 5.3401689529418945,
      "learning_rate": 0.00019995453288129515,
      "loss": 1.9139,
      "step": 13120
    },
    {
      "epoch": 1.5800240673886883,
      "grad_norm": 3.759310483932495,
      "learning_rate": 0.00019995441810309223,
      "loss": 2.064,
      "step": 13130
    },
    {
      "epoch": 1.5812274368231047,
      "grad_norm": 5.477263450622559,
      "learning_rate": 0.00019995430318023059,
      "loss": 2.0441,
      "step": 13140
    },
    {
      "epoch": 1.5824308062575212,
      "grad_norm": 4.138892650604248,
      "learning_rate": 0.0001999541881127104,
      "loss": 2.0506,
      "step": 13150
    },
    {
      "epoch": 1.5836341756919374,
      "grad_norm": 5.960071563720703,
      "learning_rate": 0.00019995407290053185,
      "loss": 2.2549,
      "step": 13160
    },
    {
      "epoch": 1.5848375451263537,
      "grad_norm": 5.113036632537842,
      "learning_rate": 0.00019995395754369503,
      "loss": 1.9663,
      "step": 13170
    },
    {
      "epoch": 1.5860409145607701,
      "grad_norm": 4.506590843200684,
      "learning_rate": 0.0001999538420422002,
      "loss": 2.0485,
      "step": 13180
    },
    {
      "epoch": 1.5872442839951866,
      "grad_norm": 5.975399971008301,
      "learning_rate": 0.00019995372639604745,
      "loss": 1.9692,
      "step": 13190
    },
    {
      "epoch": 1.5884476534296028,
      "grad_norm": 5.632997989654541,
      "learning_rate": 0.000199953610605237,
      "loss": 1.7883,
      "step": 13200
    },
    {
      "epoch": 1.5896510228640193,
      "grad_norm": 6.302413463592529,
      "learning_rate": 0.000199953494669769,
      "loss": 1.9853,
      "step": 13210
    },
    {
      "epoch": 1.5908543922984357,
      "grad_norm": 4.656396389007568,
      "learning_rate": 0.0001999533785896436,
      "loss": 2.1029,
      "step": 13220
    },
    {
      "epoch": 1.592057761732852,
      "grad_norm": 3.7483510971069336,
      "learning_rate": 0.00019995326236486096,
      "loss": 1.9259,
      "step": 13230
    },
    {
      "epoch": 1.5932611311672682,
      "grad_norm": 6.738104820251465,
      "learning_rate": 0.00019995314599542132,
      "loss": 2.1684,
      "step": 13240
    },
    {
      "epoch": 1.5944645006016849,
      "grad_norm": 5.493054389953613,
      "learning_rate": 0.00019995302948132477,
      "loss": 2.0286,
      "step": 13250
    },
    {
      "epoch": 1.595667870036101,
      "grad_norm": 6.3818511962890625,
      "learning_rate": 0.00019995291282257153,
      "loss": 2.0926,
      "step": 13260
    },
    {
      "epoch": 1.5968712394705173,
      "grad_norm": 4.904103755950928,
      "learning_rate": 0.00019995279601916176,
      "loss": 2.24,
      "step": 13270
    },
    {
      "epoch": 1.5980746089049338,
      "grad_norm": 3.644404888153076,
      "learning_rate": 0.00019995267907109556,
      "loss": 1.9631,
      "step": 13280
    },
    {
      "epoch": 1.5992779783393503,
      "grad_norm": 5.314022541046143,
      "learning_rate": 0.0001999525619783732,
      "loss": 1.8594,
      "step": 13290
    },
    {
      "epoch": 1.6004813477737665,
      "grad_norm": 4.290368556976318,
      "learning_rate": 0.0001999524447409948,
      "loss": 2.0327,
      "step": 13300
    },
    {
      "epoch": 1.601684717208183,
      "grad_norm": 6.445858478546143,
      "learning_rate": 0.00019995232735896054,
      "loss": 1.9905,
      "step": 13310
    },
    {
      "epoch": 1.6028880866425994,
      "grad_norm": 5.4139275550842285,
      "learning_rate": 0.00019995220983227055,
      "loss": 1.8597,
      "step": 13320
    },
    {
      "epoch": 1.6040914560770156,
      "grad_norm": 4.069512367248535,
      "learning_rate": 0.00019995209216092506,
      "loss": 1.9423,
      "step": 13330
    },
    {
      "epoch": 1.6052948255114319,
      "grad_norm": 5.475891590118408,
      "learning_rate": 0.0001999519743449242,
      "loss": 1.9228,
      "step": 13340
    },
    {
      "epoch": 1.6064981949458483,
      "grad_norm": 4.515195846557617,
      "learning_rate": 0.00019995185638426818,
      "loss": 2.0322,
      "step": 13350
    },
    {
      "epoch": 1.6077015643802648,
      "grad_norm": 7.007673740386963,
      "learning_rate": 0.00019995173827895714,
      "loss": 2.0488,
      "step": 13360
    },
    {
      "epoch": 1.608904933814681,
      "grad_norm": 6.145302772521973,
      "learning_rate": 0.00019995162002899125,
      "loss": 2.0093,
      "step": 13370
    },
    {
      "epoch": 1.6101083032490975,
      "grad_norm": 4.393493175506592,
      "learning_rate": 0.0001999515016343707,
      "loss": 1.9531,
      "step": 13380
    },
    {
      "epoch": 1.611311672683514,
      "grad_norm": 7.322495937347412,
      "learning_rate": 0.00019995138309509566,
      "loss": 1.9249,
      "step": 13390
    },
    {
      "epoch": 1.6125150421179302,
      "grad_norm": 4.038941860198975,
      "learning_rate": 0.00019995126441116624,
      "loss": 1.9711,
      "step": 13400
    },
    {
      "epoch": 1.6137184115523464,
      "grad_norm": 6.373983860015869,
      "learning_rate": 0.00019995114558258272,
      "loss": 2.1252,
      "step": 13410
    },
    {
      "epoch": 1.614921780986763,
      "grad_norm": 5.309967517852783,
      "learning_rate": 0.0001999510266093452,
      "loss": 1.9081,
      "step": 13420
    },
    {
      "epoch": 1.6161251504211793,
      "grad_norm": 3.5125749111175537,
      "learning_rate": 0.00019995090749145384,
      "loss": 1.9977,
      "step": 13430
    },
    {
      "epoch": 1.6173285198555956,
      "grad_norm": 5.910234451293945,
      "learning_rate": 0.00019995078822890888,
      "loss": 2.0796,
      "step": 13440
    },
    {
      "epoch": 1.618531889290012,
      "grad_norm": 4.721339702606201,
      "learning_rate": 0.0001999506688217104,
      "loss": 2.0782,
      "step": 13450
    },
    {
      "epoch": 1.6197352587244285,
      "grad_norm": 6.251277923583984,
      "learning_rate": 0.0001999505492698587,
      "loss": 2.1032,
      "step": 13460
    },
    {
      "epoch": 1.6209386281588447,
      "grad_norm": 5.622414588928223,
      "learning_rate": 0.0001999504295733538,
      "loss": 2.0356,
      "step": 13470
    },
    {
      "epoch": 1.6221419975932612,
      "grad_norm": 3.497246265411377,
      "learning_rate": 0.000199950309732196,
      "loss": 2.0234,
      "step": 13480
    },
    {
      "epoch": 1.6233453670276776,
      "grad_norm": 5.098435878753662,
      "learning_rate": 0.0001999501897463854,
      "loss": 1.8807,
      "step": 13490
    },
    {
      "epoch": 1.6245487364620939,
      "grad_norm": 4.744088649749756,
      "learning_rate": 0.00019995006961592222,
      "loss": 2.1631,
      "step": 13500
    },
    {
      "epoch": 1.62575210589651,
      "grad_norm": 5.996286392211914,
      "learning_rate": 0.00019994994934080662,
      "loss": 1.9999,
      "step": 13510
    },
    {
      "epoch": 1.6269554753309265,
      "grad_norm": 4.346140384674072,
      "learning_rate": 0.00019994982892103874,
      "loss": 1.7685,
      "step": 13520
    },
    {
      "epoch": 1.628158844765343,
      "grad_norm": 3.85640811920166,
      "learning_rate": 0.00019994970835661877,
      "loss": 2.0004,
      "step": 13530
    },
    {
      "epoch": 1.6293622141997592,
      "grad_norm": 5.55426549911499,
      "learning_rate": 0.0001999495876475469,
      "loss": 2.1797,
      "step": 13540
    },
    {
      "epoch": 1.6305655836341757,
      "grad_norm": 4.2621941566467285,
      "learning_rate": 0.00019994946679382337,
      "loss": 2.1064,
      "step": 13550
    },
    {
      "epoch": 1.6317689530685922,
      "grad_norm": 6.668655872344971,
      "learning_rate": 0.00019994934579544823,
      "loss": 2.1616,
      "step": 13560
    },
    {
      "epoch": 1.6329723225030084,
      "grad_norm": 3.979400634765625,
      "learning_rate": 0.00019994922465242169,
      "loss": 1.8361,
      "step": 13570
    },
    {
      "epoch": 1.6341756919374246,
      "grad_norm": 3.248394250869751,
      "learning_rate": 0.00019994910336474395,
      "loss": 2.0907,
      "step": 13580
    },
    {
      "epoch": 1.6353790613718413,
      "grad_norm": 5.4946112632751465,
      "learning_rate": 0.00019994898193241524,
      "loss": 2.2238,
      "step": 13590
    },
    {
      "epoch": 1.6365824308062575,
      "grad_norm": 4.9663405418396,
      "learning_rate": 0.00019994886035543558,
      "loss": 2.0246,
      "step": 13600
    },
    {
      "epoch": 1.6377858002406738,
      "grad_norm": 8.015204429626465,
      "learning_rate": 0.0001999487386338053,
      "loss": 1.9332,
      "step": 13610
    },
    {
      "epoch": 1.6389891696750902,
      "grad_norm": 5.355557441711426,
      "learning_rate": 0.00019994861676752454,
      "loss": 1.8407,
      "step": 13620
    },
    {
      "epoch": 1.6401925391095067,
      "grad_norm": 3.132603883743286,
      "learning_rate": 0.00019994849475659342,
      "loss": 2.0967,
      "step": 13630
    },
    {
      "epoch": 1.641395908543923,
      "grad_norm": 5.896000385284424,
      "learning_rate": 0.00019994837260101217,
      "loss": 1.9129,
      "step": 13640
    },
    {
      "epoch": 1.6425992779783394,
      "grad_norm": 4.218536376953125,
      "learning_rate": 0.00019994825030078092,
      "loss": 1.9994,
      "step": 13650
    },
    {
      "epoch": 1.6438026474127558,
      "grad_norm": 6.520483016967773,
      "learning_rate": 0.0001999481278558999,
      "loss": 1.9769,
      "step": 13660
    },
    {
      "epoch": 1.645006016847172,
      "grad_norm": 5.227589130401611,
      "learning_rate": 0.00019994800526636924,
      "loss": 2.0373,
      "step": 13670
    },
    {
      "epoch": 1.6462093862815883,
      "grad_norm": 3.4769139289855957,
      "learning_rate": 0.00019994788253218916,
      "loss": 2.1869,
      "step": 13680
    },
    {
      "epoch": 1.6474127557160048,
      "grad_norm": 5.577840805053711,
      "learning_rate": 0.00019994775965335984,
      "loss": 1.9615,
      "step": 13690
    },
    {
      "epoch": 1.6486161251504212,
      "grad_norm": 4.6957855224609375,
      "learning_rate": 0.0001999476366298814,
      "loss": 2.0406,
      "step": 13700
    },
    {
      "epoch": 1.6498194945848375,
      "grad_norm": 6.101114273071289,
      "learning_rate": 0.00019994751346175406,
      "loss": 2.021,
      "step": 13710
    },
    {
      "epoch": 1.651022864019254,
      "grad_norm": 4.463509559631348,
      "learning_rate": 0.000199947390148978,
      "loss": 2.0715,
      "step": 13720
    },
    {
      "epoch": 1.6522262334536704,
      "grad_norm": 3.30945086479187,
      "learning_rate": 0.0001999472666915534,
      "loss": 2.0682,
      "step": 13730
    },
    {
      "epoch": 1.6534296028880866,
      "grad_norm": 5.090394020080566,
      "learning_rate": 0.00019994714308948038,
      "loss": 2.0736,
      "step": 13740
    },
    {
      "epoch": 1.654632972322503,
      "grad_norm": 4.4032883644104,
      "learning_rate": 0.00019994701934275923,
      "loss": 2.0155,
      "step": 13750
    },
    {
      "epoch": 1.6558363417569195,
      "grad_norm": 7.14220666885376,
      "learning_rate": 0.00019994689545139004,
      "loss": 2.0574,
      "step": 13760
    },
    {
      "epoch": 1.6570397111913358,
      "grad_norm": 4.546938896179199,
      "learning_rate": 0.000199946771415373,
      "loss": 1.8947,
      "step": 13770
    },
    {
      "epoch": 1.658243080625752,
      "grad_norm": 3.6042399406433105,
      "learning_rate": 0.00019994664723470832,
      "loss": 2.0706,
      "step": 13780
    },
    {
      "epoch": 1.6594464500601684,
      "grad_norm": 6.472089767456055,
      "learning_rate": 0.00019994652290939618,
      "loss": 1.9499,
      "step": 13790
    },
    {
      "epoch": 1.660649819494585,
      "grad_norm": 4.142209053039551,
      "learning_rate": 0.00019994639843943672,
      "loss": 1.9042,
      "step": 13800
    },
    {
      "epoch": 1.6618531889290011,
      "grad_norm": 7.9362688064575195,
      "learning_rate": 0.00019994627382483016,
      "loss": 2.3313,
      "step": 13810
    },
    {
      "epoch": 1.6630565583634176,
      "grad_norm": 4.758325576782227,
      "learning_rate": 0.00019994614906557664,
      "loss": 1.9777,
      "step": 13820
    },
    {
      "epoch": 1.664259927797834,
      "grad_norm": 3.6581249237060547,
      "learning_rate": 0.0001999460241616764,
      "loss": 2.0724,
      "step": 13830
    },
    {
      "epoch": 1.6654632972322503,
      "grad_norm": 5.850959300994873,
      "learning_rate": 0.0001999458991131296,
      "loss": 2.0206,
      "step": 13840
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 4.039100170135498,
      "learning_rate": 0.00019994577391993637,
      "loss": 2.0497,
      "step": 13850
    },
    {
      "epoch": 1.6678700361010832,
      "grad_norm": 7.437899112701416,
      "learning_rate": 0.00019994564858209695,
      "loss": 2.1651,
      "step": 13860
    },
    {
      "epoch": 1.6690734055354994,
      "grad_norm": 5.1110358238220215,
      "learning_rate": 0.0001999455230996115,
      "loss": 1.8606,
      "step": 13870
    },
    {
      "epoch": 1.6702767749699157,
      "grad_norm": 4.0577192306518555,
      "learning_rate": 0.00019994539747248018,
      "loss": 1.9556,
      "step": 13880
    },
    {
      "epoch": 1.6714801444043321,
      "grad_norm": 6.085712432861328,
      "learning_rate": 0.00019994527170070323,
      "loss": 1.8232,
      "step": 13890
    },
    {
      "epoch": 1.6726835138387486,
      "grad_norm": 4.046009540557861,
      "learning_rate": 0.00019994514578428076,
      "loss": 1.9657,
      "step": 13900
    },
    {
      "epoch": 1.6738868832731648,
      "grad_norm": 6.810433864593506,
      "learning_rate": 0.000199945019723213,
      "loss": 1.9914,
      "step": 13910
    },
    {
      "epoch": 1.6750902527075813,
      "grad_norm": 4.578562259674072,
      "learning_rate": 0.00019994489351750014,
      "loss": 2.0339,
      "step": 13920
    },
    {
      "epoch": 1.6762936221419977,
      "grad_norm": 3.8101818561553955,
      "learning_rate": 0.00019994476716714233,
      "loss": 2.0032,
      "step": 13930
    },
    {
      "epoch": 1.677496991576414,
      "grad_norm": 6.676901817321777,
      "learning_rate": 0.0001999446406721398,
      "loss": 2.0849,
      "step": 13940
    },
    {
      "epoch": 1.6787003610108302,
      "grad_norm": 4.303239822387695,
      "learning_rate": 0.00019994451403249265,
      "loss": 2.0743,
      "step": 13950
    },
    {
      "epoch": 1.6799037304452467,
      "grad_norm": 7.639218330383301,
      "learning_rate": 0.0001999443872482011,
      "loss": 2.1194,
      "step": 13960
    },
    {
      "epoch": 1.6811070998796631,
      "grad_norm": 4.64432954788208,
      "learning_rate": 0.00019994426031926538,
      "loss": 1.8633,
      "step": 13970
    },
    {
      "epoch": 1.6823104693140793,
      "grad_norm": 3.69960618019104,
      "learning_rate": 0.00019994413324568564,
      "loss": 2.2756,
      "step": 13980
    },
    {
      "epoch": 1.6835138387484958,
      "grad_norm": 5.044643878936768,
      "learning_rate": 0.00019994400602746205,
      "loss": 1.7677,
      "step": 13990
    },
    {
      "epoch": 1.6847172081829123,
      "grad_norm": 4.38433313369751,
      "learning_rate": 0.00019994387866459484,
      "loss": 1.9455,
      "step": 14000
    },
    {
      "epoch": 1.6859205776173285,
      "grad_norm": 10.714807510375977,
      "learning_rate": 0.00019994375115708413,
      "loss": 2.2971,
      "step": 14010
    },
    {
      "epoch": 1.6871239470517447,
      "grad_norm": 6.168661117553711,
      "learning_rate": 0.00019994362350493015,
      "loss": 2.0647,
      "step": 14020
    },
    {
      "epoch": 1.6883273164861614,
      "grad_norm": 3.642040729522705,
      "learning_rate": 0.00019994349570813308,
      "loss": 2.1859,
      "step": 14030
    },
    {
      "epoch": 1.6895306859205776,
      "grad_norm": 7.047633171081543,
      "learning_rate": 0.00019994336776669306,
      "loss": 2.0542,
      "step": 14040
    },
    {
      "epoch": 1.6907340553549939,
      "grad_norm": 3.594160318374634,
      "learning_rate": 0.00019994323968061033,
      "loss": 2.0876,
      "step": 14050
    },
    {
      "epoch": 1.6919374247894103,
      "grad_norm": 5.530572414398193,
      "learning_rate": 0.00019994311144988508,
      "loss": 2.0243,
      "step": 14060
    },
    {
      "epoch": 1.6931407942238268,
      "grad_norm": 5.095473289489746,
      "learning_rate": 0.00019994298307451745,
      "loss": 1.9181,
      "step": 14070
    },
    {
      "epoch": 1.694344163658243,
      "grad_norm": 4.656373500823975,
      "learning_rate": 0.00019994285455450763,
      "loss": 2.2075,
      "step": 14080
    },
    {
      "epoch": 1.6955475330926595,
      "grad_norm": 5.468774795532227,
      "learning_rate": 0.00019994272588985586,
      "loss": 2.036,
      "step": 14090
    },
    {
      "epoch": 1.696750902527076,
      "grad_norm": 4.336451530456543,
      "learning_rate": 0.00019994259708056226,
      "loss": 1.9637,
      "step": 14100
    },
    {
      "epoch": 1.6979542719614922,
      "grad_norm": 6.688697338104248,
      "learning_rate": 0.00019994246812662708,
      "loss": 1.9577,
      "step": 14110
    },
    {
      "epoch": 1.6991576413959084,
      "grad_norm": 4.899430274963379,
      "learning_rate": 0.00019994233902805047,
      "loss": 2.029,
      "step": 14120
    },
    {
      "epoch": 1.7003610108303249,
      "grad_norm": 4.238423824310303,
      "learning_rate": 0.0001999422097848326,
      "loss": 2.1059,
      "step": 14130
    },
    {
      "epoch": 1.7015643802647413,
      "grad_norm": 5.7632527351379395,
      "learning_rate": 0.00019994208039697368,
      "loss": 1.8448,
      "step": 14140
    },
    {
      "epoch": 1.7027677496991576,
      "grad_norm": 4.8242950439453125,
      "learning_rate": 0.00019994195086447389,
      "loss": 1.9235,
      "step": 14150
    },
    {
      "epoch": 1.703971119133574,
      "grad_norm": 6.783804416656494,
      "learning_rate": 0.00019994182118733344,
      "loss": 1.9539,
      "step": 14160
    },
    {
      "epoch": 1.7051744885679905,
      "grad_norm": 4.87070369720459,
      "learning_rate": 0.0001999416913655525,
      "loss": 2.0481,
      "step": 14170
    },
    {
      "epoch": 1.7063778580024067,
      "grad_norm": 4.1888041496276855,
      "learning_rate": 0.00019994156139913126,
      "loss": 2.0329,
      "step": 14180
    },
    {
      "epoch": 1.707581227436823,
      "grad_norm": 5.123957633972168,
      "learning_rate": 0.0001999414312880699,
      "loss": 1.882,
      "step": 14190
    },
    {
      "epoch": 1.7087845968712396,
      "grad_norm": 4.196894645690918,
      "learning_rate": 0.0001999413010323686,
      "loss": 2.0405,
      "step": 14200
    },
    {
      "epoch": 1.7099879663056559,
      "grad_norm": 5.899720668792725,
      "learning_rate": 0.00019994117063202758,
      "loss": 2.1224,
      "step": 14210
    },
    {
      "epoch": 1.711191335740072,
      "grad_norm": 4.157967567443848,
      "learning_rate": 0.000199941040087047,
      "loss": 1.7508,
      "step": 14220
    },
    {
      "epoch": 1.7123947051744886,
      "grad_norm": 3.6508965492248535,
      "learning_rate": 0.00019994090939742706,
      "loss": 1.9189,
      "step": 14230
    },
    {
      "epoch": 1.713598074608905,
      "grad_norm": 6.262744903564453,
      "learning_rate": 0.00019994077856316794,
      "loss": 1.8024,
      "step": 14240
    },
    {
      "epoch": 1.7148014440433212,
      "grad_norm": 4.435568809509277,
      "learning_rate": 0.00019994064758426986,
      "loss": 2.1275,
      "step": 14250
    },
    {
      "epoch": 1.7160048134777377,
      "grad_norm": 6.094752788543701,
      "learning_rate": 0.000199940516460733,
      "loss": 1.8911,
      "step": 14260
    },
    {
      "epoch": 1.7172081829121542,
      "grad_norm": 5.009769916534424,
      "learning_rate": 0.00019994038519255752,
      "loss": 1.9552,
      "step": 14270
    },
    {
      "epoch": 1.7184115523465704,
      "grad_norm": 3.6759421825408936,
      "learning_rate": 0.00019994025377974363,
      "loss": 2.107,
      "step": 14280
    },
    {
      "epoch": 1.7196149217809866,
      "grad_norm": 6.70467472076416,
      "learning_rate": 0.0001999401222222915,
      "loss": 1.9142,
      "step": 14290
    },
    {
      "epoch": 1.720818291215403,
      "grad_norm": 5.994391918182373,
      "learning_rate": 0.0001999399905202014,
      "loss": 2.1916,
      "step": 14300
    },
    {
      "epoch": 1.7220216606498195,
      "grad_norm": 6.835086822509766,
      "learning_rate": 0.00019993985867347341,
      "loss": 1.7309,
      "step": 14310
    },
    {
      "epoch": 1.7232250300842358,
      "grad_norm": 4.927305698394775,
      "learning_rate": 0.00019993972668210774,
      "loss": 2.0401,
      "step": 14320
    },
    {
      "epoch": 1.7244283995186522,
      "grad_norm": 4.515219688415527,
      "learning_rate": 0.00019993959454610467,
      "loss": 2.1645,
      "step": 14330
    },
    {
      "epoch": 1.7256317689530687,
      "grad_norm": 5.741243362426758,
      "learning_rate": 0.0001999394622654643,
      "loss": 2.144,
      "step": 14340
    },
    {
      "epoch": 1.726835138387485,
      "grad_norm": 4.268014907836914,
      "learning_rate": 0.00019993932984018687,
      "loss": 2.1394,
      "step": 14350
    },
    {
      "epoch": 1.7280385078219012,
      "grad_norm": 7.619650840759277,
      "learning_rate": 0.00019993919727027252,
      "loss": 1.7946,
      "step": 14360
    },
    {
      "epoch": 1.7292418772563178,
      "grad_norm": 4.427943229675293,
      "learning_rate": 0.00019993906455572152,
      "loss": 2.0447,
      "step": 14370
    },
    {
      "epoch": 1.730445246690734,
      "grad_norm": 3.5642616748809814,
      "learning_rate": 0.000199938931696534,
      "loss": 2.0119,
      "step": 14380
    },
    {
      "epoch": 1.7316486161251503,
      "grad_norm": 5.555940628051758,
      "learning_rate": 0.0001999387986927102,
      "loss": 2.1435,
      "step": 14390
    },
    {
      "epoch": 1.7328519855595668,
      "grad_norm": 4.9382147789001465,
      "learning_rate": 0.00019993866554425025,
      "loss": 1.9292,
      "step": 14400
    },
    {
      "epoch": 1.7340553549939832,
      "grad_norm": 6.797666072845459,
      "learning_rate": 0.0001999385322511544,
      "loss": 1.8982,
      "step": 14410
    },
    {
      "epoch": 1.7352587244283995,
      "grad_norm": 5.267373561859131,
      "learning_rate": 0.00019993839881342278,
      "loss": 1.9265,
      "step": 14420
    },
    {
      "epoch": 1.736462093862816,
      "grad_norm": 4.399893760681152,
      "learning_rate": 0.00019993826523105566,
      "loss": 1.9926,
      "step": 14430
    },
    {
      "epoch": 1.7376654632972324,
      "grad_norm": 4.862719535827637,
      "learning_rate": 0.00019993813150405318,
      "loss": 2.0627,
      "step": 14440
    },
    {
      "epoch": 1.7388688327316486,
      "grad_norm": 5.0265278816223145,
      "learning_rate": 0.00019993799763241555,
      "loss": 1.987,
      "step": 14450
    },
    {
      "epoch": 1.7400722021660648,
      "grad_norm": 6.209497451782227,
      "learning_rate": 0.000199937863616143,
      "loss": 1.945,
      "step": 14460
    },
    {
      "epoch": 1.7412755716004813,
      "grad_norm": 4.86743688583374,
      "learning_rate": 0.00019993772945523563,
      "loss": 1.8645,
      "step": 14470
    },
    {
      "epoch": 1.7424789410348978,
      "grad_norm": 4.553318977355957,
      "learning_rate": 0.00019993759514969374,
      "loss": 1.8207,
      "step": 14480
    },
    {
      "epoch": 1.743682310469314,
      "grad_norm": 5.645698070526123,
      "learning_rate": 0.00019993746069951746,
      "loss": 2.0576,
      "step": 14490
    },
    {
      "epoch": 1.7448856799037304,
      "grad_norm": 4.6090617179870605,
      "learning_rate": 0.00019993732610470698,
      "loss": 1.9278,
      "step": 14500
    },
    {
      "epoch": 1.746089049338147,
      "grad_norm": 7.098291397094727,
      "learning_rate": 0.00019993719136526252,
      "loss": 2.002,
      "step": 14510
    },
    {
      "epoch": 1.7472924187725631,
      "grad_norm": 4.5874409675598145,
      "learning_rate": 0.0001999370564811843,
      "loss": 2.0125,
      "step": 14520
    },
    {
      "epoch": 1.7484957882069796,
      "grad_norm": 4.147941589355469,
      "learning_rate": 0.00019993692145247243,
      "loss": 1.9628,
      "step": 14530
    },
    {
      "epoch": 1.749699157641396,
      "grad_norm": 5.5968170166015625,
      "learning_rate": 0.0001999367862791272,
      "loss": 1.7204,
      "step": 14540
    },
    {
      "epoch": 1.7509025270758123,
      "grad_norm": 5.026246547698975,
      "learning_rate": 0.00019993665096114879,
      "loss": 1.7494,
      "step": 14550
    },
    {
      "epoch": 1.7521058965102285,
      "grad_norm": 7.101372241973877,
      "learning_rate": 0.00019993651549853736,
      "loss": 2.1529,
      "step": 14560
    },
    {
      "epoch": 1.753309265944645,
      "grad_norm": 4.936278820037842,
      "learning_rate": 0.0001999363798912931,
      "loss": 1.8869,
      "step": 14570
    },
    {
      "epoch": 1.7545126353790614,
      "grad_norm": 3.160592555999756,
      "learning_rate": 0.00019993624413941623,
      "loss": 2.0949,
      "step": 14580
    },
    {
      "epoch": 1.7557160048134777,
      "grad_norm": 5.4424285888671875,
      "learning_rate": 0.00019993610824290694,
      "loss": 1.7147,
      "step": 14590
    },
    {
      "epoch": 1.7569193742478941,
      "grad_norm": 3.808987855911255,
      "learning_rate": 0.00019993597220176542,
      "loss": 2.1881,
      "step": 14600
    },
    {
      "epoch": 1.7581227436823106,
      "grad_norm": 5.4764018058776855,
      "learning_rate": 0.00019993583601599188,
      "loss": 2.0252,
      "step": 14610
    },
    {
      "epoch": 1.7593261131167268,
      "grad_norm": 4.963686466217041,
      "learning_rate": 0.00019993569968558653,
      "loss": 2.238,
      "step": 14620
    },
    {
      "epoch": 1.760529482551143,
      "grad_norm": 3.916469097137451,
      "learning_rate": 0.00019993556321054953,
      "loss": 1.908,
      "step": 14630
    },
    {
      "epoch": 1.7617328519855595,
      "grad_norm": 5.4917144775390625,
      "learning_rate": 0.00019993542659088113,
      "loss": 1.879,
      "step": 14640
    },
    {
      "epoch": 1.762936221419976,
      "grad_norm": 3.6770243644714355,
      "learning_rate": 0.00019993528982658145,
      "loss": 2.0372,
      "step": 14650
    },
    {
      "epoch": 1.7641395908543922,
      "grad_norm": 6.837653636932373,
      "learning_rate": 0.00019993515291765075,
      "loss": 2.0467,
      "step": 14660
    },
    {
      "epoch": 1.7653429602888087,
      "grad_norm": 5.158958435058594,
      "learning_rate": 0.0001999350158640892,
      "loss": 1.8737,
      "step": 14670
    },
    {
      "epoch": 1.7665463297232251,
      "grad_norm": 3.207634210586548,
      "learning_rate": 0.00019993487866589704,
      "loss": 1.9669,
      "step": 14680
    },
    {
      "epoch": 1.7677496991576414,
      "grad_norm": 4.8469696044921875,
      "learning_rate": 0.0001999347413230744,
      "loss": 1.8349,
      "step": 14690
    },
    {
      "epoch": 1.7689530685920578,
      "grad_norm": 4.001832008361816,
      "learning_rate": 0.00019993460383562153,
      "loss": 1.9146,
      "step": 14700
    },
    {
      "epoch": 1.7701564380264743,
      "grad_norm": 8.350502014160156,
      "learning_rate": 0.0001999344662035386,
      "loss": 2.0797,
      "step": 14710
    },
    {
      "epoch": 1.7713598074608905,
      "grad_norm": 4.965343475341797,
      "learning_rate": 0.00019993432842682586,
      "loss": 1.9329,
      "step": 14720
    },
    {
      "epoch": 1.7725631768953067,
      "grad_norm": 3.8536767959594727,
      "learning_rate": 0.00019993419050548346,
      "loss": 1.9399,
      "step": 14730
    },
    {
      "epoch": 1.7737665463297232,
      "grad_norm": 5.0429368019104,
      "learning_rate": 0.0001999340524395116,
      "loss": 2.0099,
      "step": 14740
    },
    {
      "epoch": 1.7749699157641396,
      "grad_norm": 3.9784657955169678,
      "learning_rate": 0.0001999339142289105,
      "loss": 2.0581,
      "step": 14750
    },
    {
      "epoch": 1.7761732851985559,
      "grad_norm": 7.18906831741333,
      "learning_rate": 0.0001999337758736804,
      "loss": 1.9906,
      "step": 14760
    },
    {
      "epoch": 1.7773766546329723,
      "grad_norm": 4.139045715332031,
      "learning_rate": 0.00019993363737382138,
      "loss": 1.989,
      "step": 14770
    },
    {
      "epoch": 1.7785800240673888,
      "grad_norm": 3.125943422317505,
      "learning_rate": 0.00019993349872933375,
      "loss": 2.062,
      "step": 14780
    },
    {
      "epoch": 1.779783393501805,
      "grad_norm": 6.117612838745117,
      "learning_rate": 0.00019993335994021764,
      "loss": 1.8821,
      "step": 14790
    },
    {
      "epoch": 1.7809867629362213,
      "grad_norm": 4.846273899078369,
      "learning_rate": 0.00019993322100647333,
      "loss": 1.8729,
      "step": 14800
    },
    {
      "epoch": 1.782190132370638,
      "grad_norm": 6.4829583168029785,
      "learning_rate": 0.00019993308192810095,
      "loss": 1.9971,
      "step": 14810
    },
    {
      "epoch": 1.7833935018050542,
      "grad_norm": 4.51899528503418,
      "learning_rate": 0.00019993294270510075,
      "loss": 2.0308,
      "step": 14820
    },
    {
      "epoch": 1.7845968712394704,
      "grad_norm": 3.4743294715881348,
      "learning_rate": 0.0001999328033374729,
      "loss": 2.0067,
      "step": 14830
    },
    {
      "epoch": 1.7858002406738869,
      "grad_norm": 6.282670021057129,
      "learning_rate": 0.0001999326638252176,
      "loss": 1.9238,
      "step": 14840
    },
    {
      "epoch": 1.7870036101083033,
      "grad_norm": 4.519623279571533,
      "learning_rate": 0.0001999325241683351,
      "loss": 1.9772,
      "step": 14850
    },
    {
      "epoch": 1.7882069795427196,
      "grad_norm": 6.4352216720581055,
      "learning_rate": 0.0001999323843668255,
      "loss": 2.0207,
      "step": 14860
    },
    {
      "epoch": 1.789410348977136,
      "grad_norm": 4.4146599769592285,
      "learning_rate": 0.00019993224442068912,
      "loss": 1.8675,
      "step": 14870
    },
    {
      "epoch": 1.7906137184115525,
      "grad_norm": 3.8463425636291504,
      "learning_rate": 0.0001999321043299261,
      "loss": 2.0264,
      "step": 14880
    },
    {
      "epoch": 1.7918170878459687,
      "grad_norm": 5.0575175285339355,
      "learning_rate": 0.00019993196409453663,
      "loss": 2.2029,
      "step": 14890
    },
    {
      "epoch": 1.793020457280385,
      "grad_norm": 4.583847522735596,
      "learning_rate": 0.00019993182371452094,
      "loss": 2.0102,
      "step": 14900
    },
    {
      "epoch": 1.7942238267148014,
      "grad_norm": 6.8686418533325195,
      "learning_rate": 0.00019993168318987924,
      "loss": 1.9933,
      "step": 14910
    },
    {
      "epoch": 1.7954271961492179,
      "grad_norm": 5.599619388580322,
      "learning_rate": 0.00019993154252061172,
      "loss": 2.1391,
      "step": 14920
    },
    {
      "epoch": 1.796630565583634,
      "grad_norm": 3.494934320449829,
      "learning_rate": 0.0001999314017067186,
      "loss": 1.9257,
      "step": 14930
    },
    {
      "epoch": 1.7978339350180506,
      "grad_norm": 5.740988254547119,
      "learning_rate": 0.00019993126074820006,
      "loss": 2.1163,
      "step": 14940
    },
    {
      "epoch": 1.799037304452467,
      "grad_norm": 4.562415599822998,
      "learning_rate": 0.00019993111964505632,
      "loss": 1.9872,
      "step": 14950
    },
    {
      "epoch": 1.8002406738868832,
      "grad_norm": 5.620573043823242,
      "learning_rate": 0.00019993097839728755,
      "loss": 2.0738,
      "step": 14960
    },
    {
      "epoch": 1.8014440433212995,
      "grad_norm": 5.37310266494751,
      "learning_rate": 0.00019993083700489402,
      "loss": 2.1425,
      "step": 14970
    },
    {
      "epoch": 1.8026474127557162,
      "grad_norm": 3.5825746059417725,
      "learning_rate": 0.00019993069546787585,
      "loss": 1.9019,
      "step": 14980
    },
    {
      "epoch": 1.8038507821901324,
      "grad_norm": 5.194967746734619,
      "learning_rate": 0.00019993055378623333,
      "loss": 1.8986,
      "step": 14990
    },
    {
      "epoch": 1.8050541516245486,
      "grad_norm": 4.382756233215332,
      "learning_rate": 0.00019993041195996662,
      "loss": 1.8099,
      "step": 15000
    },
    {
      "epoch": 1.806257521058965,
      "grad_norm": 6.057338237762451,
      "learning_rate": 0.00019993026998907593,
      "loss": 2.2388,
      "step": 15010
    },
    {
      "epoch": 1.8074608904933815,
      "grad_norm": 5.524228572845459,
      "learning_rate": 0.00019993012787356148,
      "loss": 1.9931,
      "step": 15020
    },
    {
      "epoch": 1.8086642599277978,
      "grad_norm": 3.8152716159820557,
      "learning_rate": 0.00019992998561342343,
      "loss": 1.8704,
      "step": 15030
    },
    {
      "epoch": 1.8098676293622142,
      "grad_norm": 6.073026180267334,
      "learning_rate": 0.00019992984320866205,
      "loss": 1.9418,
      "step": 15040
    },
    {
      "epoch": 1.8110709987966307,
      "grad_norm": 4.161746025085449,
      "learning_rate": 0.00019992970065927748,
      "loss": 1.9705,
      "step": 15050
    },
    {
      "epoch": 1.812274368231047,
      "grad_norm": 7.0932440757751465,
      "learning_rate": 0.00019992955796526996,
      "loss": 1.9387,
      "step": 15060
    },
    {
      "epoch": 1.8134777376654632,
      "grad_norm": 5.212820053100586,
      "learning_rate": 0.00019992941512663974,
      "loss": 1.8921,
      "step": 15070
    },
    {
      "epoch": 1.8146811070998796,
      "grad_norm": 3.6544785499572754,
      "learning_rate": 0.00019992927214338696,
      "loss": 2.0141,
      "step": 15080
    },
    {
      "epoch": 1.815884476534296,
      "grad_norm": 5.903501987457275,
      "learning_rate": 0.00019992912901551187,
      "loss": 2.358,
      "step": 15090
    },
    {
      "epoch": 1.8170878459687123,
      "grad_norm": 3.894439220428467,
      "learning_rate": 0.00019992898574301465,
      "loss": 2.089,
      "step": 15100
    },
    {
      "epoch": 1.8182912154031288,
      "grad_norm": 6.723239898681641,
      "learning_rate": 0.00019992884232589552,
      "loss": 1.8207,
      "step": 15110
    },
    {
      "epoch": 1.8194945848375452,
      "grad_norm": 4.931623935699463,
      "learning_rate": 0.00019992869876415465,
      "loss": 1.7152,
      "step": 15120
    },
    {
      "epoch": 1.8206979542719615,
      "grad_norm": 3.3028016090393066,
      "learning_rate": 0.0001999285550577923,
      "loss": 1.9668,
      "step": 15130
    },
    {
      "epoch": 1.8219013237063777,
      "grad_norm": 5.859135150909424,
      "learning_rate": 0.00019992841120680866,
      "loss": 1.9609,
      "step": 15140
    },
    {
      "epoch": 1.8231046931407944,
      "grad_norm": 3.4727377891540527,
      "learning_rate": 0.00019992826721120394,
      "loss": 1.8742,
      "step": 15150
    },
    {
      "epoch": 1.8243080625752106,
      "grad_norm": 6.720226764678955,
      "learning_rate": 0.00019992812307097836,
      "loss": 1.9918,
      "step": 15160
    },
    {
      "epoch": 1.8255114320096268,
      "grad_norm": 3.607022762298584,
      "learning_rate": 0.0001999279787861321,
      "loss": 1.953,
      "step": 15170
    },
    {
      "epoch": 1.8267148014440433,
      "grad_norm": 3.5477395057678223,
      "learning_rate": 0.00019992783435666538,
      "loss": 1.9944,
      "step": 15180
    },
    {
      "epoch": 1.8279181708784598,
      "grad_norm": 5.895172119140625,
      "learning_rate": 0.00019992768978257844,
      "loss": 2.0263,
      "step": 15190
    },
    {
      "epoch": 1.829121540312876,
      "grad_norm": 3.352827310562134,
      "learning_rate": 0.00019992754506387142,
      "loss": 1.8496,
      "step": 15200
    },
    {
      "epoch": 1.8303249097472925,
      "grad_norm": 7.611035346984863,
      "learning_rate": 0.00019992740020054458,
      "loss": 1.9174,
      "step": 15210
    },
    {
      "epoch": 1.831528279181709,
      "grad_norm": 5.316261291503906,
      "learning_rate": 0.00019992725519259814,
      "loss": 1.8641,
      "step": 15220
    },
    {
      "epoch": 1.8327316486161251,
      "grad_norm": 4.250819206237793,
      "learning_rate": 0.00019992711004003231,
      "loss": 2.0508,
      "step": 15230
    },
    {
      "epoch": 1.8339350180505414,
      "grad_norm": 6.519930839538574,
      "learning_rate": 0.00019992696474284722,
      "loss": 1.91,
      "step": 15240
    },
    {
      "epoch": 1.8351383874849578,
      "grad_norm": 4.351858615875244,
      "learning_rate": 0.0001999268193010432,
      "loss": 2.0671,
      "step": 15250
    },
    {
      "epoch": 1.8363417569193743,
      "grad_norm": 6.781875133514404,
      "learning_rate": 0.00019992667371462037,
      "loss": 2.0352,
      "step": 15260
    },
    {
      "epoch": 1.8375451263537905,
      "grad_norm": 4.305140972137451,
      "learning_rate": 0.00019992652798357896,
      "loss": 2.0264,
      "step": 15270
    },
    {
      "epoch": 1.838748495788207,
      "grad_norm": 3.690814733505249,
      "learning_rate": 0.00019992638210791924,
      "loss": 2.0293,
      "step": 15280
    },
    {
      "epoch": 1.8399518652226234,
      "grad_norm": 6.05497407913208,
      "learning_rate": 0.0001999262360876413,
      "loss": 2.1248,
      "step": 15290
    },
    {
      "epoch": 1.8411552346570397,
      "grad_norm": 4.777842998504639,
      "learning_rate": 0.00019992608992274548,
      "loss": 2.2184,
      "step": 15300
    },
    {
      "epoch": 1.8423586040914561,
      "grad_norm": 6.349704742431641,
      "learning_rate": 0.00019992594361323193,
      "loss": 1.9533,
      "step": 15310
    },
    {
      "epoch": 1.8435619735258726,
      "grad_norm": 4.9056878089904785,
      "learning_rate": 0.00019992579715910088,
      "loss": 2.0726,
      "step": 15320
    },
    {
      "epoch": 1.8447653429602888,
      "grad_norm": 2.93888521194458,
      "learning_rate": 0.00019992565056035252,
      "loss": 1.8859,
      "step": 15330
    },
    {
      "epoch": 1.845968712394705,
      "grad_norm": 7.04545259475708,
      "learning_rate": 0.00019992550381698707,
      "loss": 1.9061,
      "step": 15340
    },
    {
      "epoch": 1.8471720818291215,
      "grad_norm": 4.191628932952881,
      "learning_rate": 0.0001999253569290047,
      "loss": 2.1132,
      "step": 15350
    },
    {
      "epoch": 1.848375451263538,
      "grad_norm": 6.260605812072754,
      "learning_rate": 0.00019992520989640572,
      "loss": 1.9419,
      "step": 15360
    },
    {
      "epoch": 1.8495788206979542,
      "grad_norm": 4.712516784667969,
      "learning_rate": 0.0001999250627191903,
      "loss": 2.1588,
      "step": 15370
    },
    {
      "epoch": 1.8507821901323707,
      "grad_norm": 4.36837100982666,
      "learning_rate": 0.00019992491539735861,
      "loss": 2.0079,
      "step": 15380
    },
    {
      "epoch": 1.8519855595667871,
      "grad_norm": 5.028352737426758,
      "learning_rate": 0.00019992476793091092,
      "loss": 1.8677,
      "step": 15390
    },
    {
      "epoch": 1.8531889290012034,
      "grad_norm": 3.766176223754883,
      "learning_rate": 0.00019992462031984741,
      "loss": 1.927,
      "step": 15400
    },
    {
      "epoch": 1.8543922984356196,
      "grad_norm": 7.136031627655029,
      "learning_rate": 0.00019992447256416833,
      "loss": 2.0741,
      "step": 15410
    },
    {
      "epoch": 1.855595667870036,
      "grad_norm": 4.481396198272705,
      "learning_rate": 0.00019992432466387383,
      "loss": 1.9438,
      "step": 15420
    },
    {
      "epoch": 1.8567990373044525,
      "grad_norm": 3.7558610439300537,
      "learning_rate": 0.00019992417661896417,
      "loss": 2.1443,
      "step": 15430
    },
    {
      "epoch": 1.8580024067388687,
      "grad_norm": 5.567291736602783,
      "learning_rate": 0.00019992402842943955,
      "loss": 2.1461,
      "step": 15440
    },
    {
      "epoch": 1.8592057761732852,
      "grad_norm": 4.288766384124756,
      "learning_rate": 0.00019992388009530023,
      "loss": 2.009,
      "step": 15450
    },
    {
      "epoch": 1.8604091456077017,
      "grad_norm": 6.941226482391357,
      "learning_rate": 0.00019992373161654633,
      "loss": 2.1498,
      "step": 15460
    },
    {
      "epoch": 1.8616125150421179,
      "grad_norm": 5.285544395446777,
      "learning_rate": 0.00019992358299317816,
      "loss": 1.892,
      "step": 15470
    },
    {
      "epoch": 1.8628158844765343,
      "grad_norm": 3.7738306522369385,
      "learning_rate": 0.00019992343422519588,
      "loss": 2.0769,
      "step": 15480
    },
    {
      "epoch": 1.8640192539109508,
      "grad_norm": 6.734550476074219,
      "learning_rate": 0.0001999232853125997,
      "loss": 1.9419,
      "step": 15490
    },
    {
      "epoch": 1.865222623345367,
      "grad_norm": 4.281076431274414,
      "learning_rate": 0.0001999231362553899,
      "loss": 2.044,
      "step": 15500
    },
    {
      "epoch": 1.8664259927797833,
      "grad_norm": 6.944756507873535,
      "learning_rate": 0.00019992298705356665,
      "loss": 2.109,
      "step": 15510
    },
    {
      "epoch": 1.8676293622141997,
      "grad_norm": 4.5112810134887695,
      "learning_rate": 0.00019992283770713015,
      "loss": 1.9144,
      "step": 15520
    },
    {
      "epoch": 1.8688327316486162,
      "grad_norm": 3.437556505203247,
      "learning_rate": 0.0001999226882160806,
      "loss": 1.9841,
      "step": 15530
    },
    {
      "epoch": 1.8700361010830324,
      "grad_norm": 4.6591105461120605,
      "learning_rate": 0.0001999225385804183,
      "loss": 2.0429,
      "step": 15540
    },
    {
      "epoch": 1.8712394705174489,
      "grad_norm": 4.675044536590576,
      "learning_rate": 0.0001999223888001434,
      "loss": 2.2725,
      "step": 15550
    },
    {
      "epoch": 1.8724428399518653,
      "grad_norm": 6.455400466918945,
      "learning_rate": 0.0001999222388752561,
      "loss": 2.1985,
      "step": 15560
    },
    {
      "epoch": 1.8736462093862816,
      "grad_norm": 5.524036884307861,
      "learning_rate": 0.0001999220888057567,
      "loss": 1.6719,
      "step": 15570
    },
    {
      "epoch": 1.8748495788206978,
      "grad_norm": 3.953951597213745,
      "learning_rate": 0.00019992193859164534,
      "loss": 1.8878,
      "step": 15580
    },
    {
      "epoch": 1.8760529482551145,
      "grad_norm": 5.688681125640869,
      "learning_rate": 0.00019992178823292228,
      "loss": 1.8412,
      "step": 15590
    },
    {
      "epoch": 1.8772563176895307,
      "grad_norm": 3.426849126815796,
      "learning_rate": 0.0001999216377295877,
      "loss": 1.8212,
      "step": 15600
    },
    {
      "epoch": 1.878459687123947,
      "grad_norm": 6.213525295257568,
      "learning_rate": 0.00019992148708164188,
      "loss": 2.0282,
      "step": 15610
    },
    {
      "epoch": 1.8796630565583634,
      "grad_norm": 4.974131107330322,
      "learning_rate": 0.00019992133628908497,
      "loss": 1.7227,
      "step": 15620
    },
    {
      "epoch": 1.8808664259927799,
      "grad_norm": 4.237664699554443,
      "learning_rate": 0.0001999211853519172,
      "loss": 1.8608,
      "step": 15630
    },
    {
      "epoch": 1.882069795427196,
      "grad_norm": 5.078277111053467,
      "learning_rate": 0.00019992103427013885,
      "loss": 2.0934,
      "step": 15640
    },
    {
      "epoch": 1.8832731648616126,
      "grad_norm": 4.74167013168335,
      "learning_rate": 0.00019992088304375004,
      "loss": 1.9874,
      "step": 15650
    },
    {
      "epoch": 1.884476534296029,
      "grad_norm": 6.7923126220703125,
      "learning_rate": 0.00019992073167275108,
      "loss": 1.9907,
      "step": 15660
    },
    {
      "epoch": 1.8856799037304453,
      "grad_norm": 4.451918125152588,
      "learning_rate": 0.00019992058015714216,
      "loss": 1.9799,
      "step": 15670
    },
    {
      "epoch": 1.8868832731648615,
      "grad_norm": 3.4242498874664307,
      "learning_rate": 0.00019992042849692346,
      "loss": 1.8887,
      "step": 15680
    },
    {
      "epoch": 1.888086642599278,
      "grad_norm": 5.157835006713867,
      "learning_rate": 0.00019992027669209526,
      "loss": 1.9297,
      "step": 15690
    },
    {
      "epoch": 1.8892900120336944,
      "grad_norm": 5.290318012237549,
      "learning_rate": 0.00019992012474265774,
      "loss": 2.0429,
      "step": 15700
    },
    {
      "epoch": 1.8904933814681106,
      "grad_norm": 7.171584606170654,
      "learning_rate": 0.00019991997264861113,
      "loss": 2.1705,
      "step": 15710
    },
    {
      "epoch": 1.891696750902527,
      "grad_norm": 4.1063714027404785,
      "learning_rate": 0.00019991982040995564,
      "loss": 1.8669,
      "step": 15720
    },
    {
      "epoch": 1.8929001203369435,
      "grad_norm": 3.4416663646698,
      "learning_rate": 0.0001999196680266915,
      "loss": 1.9693,
      "step": 15730
    },
    {
      "epoch": 1.8941034897713598,
      "grad_norm": 6.243912220001221,
      "learning_rate": 0.00019991951549881893,
      "loss": 1.9294,
      "step": 15740
    },
    {
      "epoch": 1.895306859205776,
      "grad_norm": 5.662498950958252,
      "learning_rate": 0.00019991936282633818,
      "loss": 1.7883,
      "step": 15750
    },
    {
      "epoch": 1.8965102286401927,
      "grad_norm": 5.44572639465332,
      "learning_rate": 0.00019991921000924942,
      "loss": 2.029,
      "step": 15760
    },
    {
      "epoch": 1.897713598074609,
      "grad_norm": 6.009549617767334,
      "learning_rate": 0.0001999190570475529,
      "loss": 1.8292,
      "step": 15770
    },
    {
      "epoch": 1.8989169675090252,
      "grad_norm": 3.629751205444336,
      "learning_rate": 0.00019991890394124883,
      "loss": 1.965,
      "step": 15780
    },
    {
      "epoch": 1.9001203369434416,
      "grad_norm": 5.768898010253906,
      "learning_rate": 0.00019991875069033744,
      "loss": 1.8766,
      "step": 15790
    },
    {
      "epoch": 1.901323706377858,
      "grad_norm": 4.1167120933532715,
      "learning_rate": 0.00019991859729481893,
      "loss": 2.1118,
      "step": 15800
    },
    {
      "epoch": 1.9025270758122743,
      "grad_norm": 7.279045581817627,
      "learning_rate": 0.00019991844375469357,
      "loss": 1.837,
      "step": 15810
    },
    {
      "epoch": 1.9037304452466908,
      "grad_norm": 5.072184085845947,
      "learning_rate": 0.00019991829006996153,
      "loss": 1.9979,
      "step": 15820
    },
    {
      "epoch": 1.9049338146811072,
      "grad_norm": 4.018231391906738,
      "learning_rate": 0.00019991813624062305,
      "loss": 1.8404,
      "step": 15830
    },
    {
      "epoch": 1.9061371841155235,
      "grad_norm": 6.246421813964844,
      "learning_rate": 0.0001999179822666784,
      "loss": 2.1459,
      "step": 15840
    },
    {
      "epoch": 1.9073405535499397,
      "grad_norm": 5.623615741729736,
      "learning_rate": 0.00019991782814812771,
      "loss": 1.92,
      "step": 15850
    },
    {
      "epoch": 1.9085439229843562,
      "grad_norm": 6.654271602630615,
      "learning_rate": 0.0001999176738849713,
      "loss": 2.153,
      "step": 15860
    },
    {
      "epoch": 1.9097472924187726,
      "grad_norm": 4.772064685821533,
      "learning_rate": 0.00019991751947720933,
      "loss": 1.8769,
      "step": 15870
    },
    {
      "epoch": 1.9109506618531888,
      "grad_norm": 3.7780356407165527,
      "learning_rate": 0.00019991736492484203,
      "loss": 1.9606,
      "step": 15880
    },
    {
      "epoch": 1.9121540312876053,
      "grad_norm": 5.867314338684082,
      "learning_rate": 0.00019991721022786965,
      "loss": 2.1057,
      "step": 15890
    },
    {
      "epoch": 1.9133574007220218,
      "grad_norm": 3.425398111343384,
      "learning_rate": 0.00019991705538629237,
      "loss": 2.0502,
      "step": 15900
    },
    {
      "epoch": 1.914560770156438,
      "grad_norm": 6.75115442276001,
      "learning_rate": 0.00019991690040011047,
      "loss": 2.102,
      "step": 15910
    },
    {
      "epoch": 1.9157641395908542,
      "grad_norm": 4.971395492553711,
      "learning_rate": 0.0001999167452693241,
      "loss": 1.8051,
      "step": 15920
    },
    {
      "epoch": 1.916967509025271,
      "grad_norm": 3.4800236225128174,
      "learning_rate": 0.00019991658999393358,
      "loss": 1.9978,
      "step": 15930
    },
    {
      "epoch": 1.9181708784596871,
      "grad_norm": 5.5732855796813965,
      "learning_rate": 0.0001999164345739391,
      "loss": 1.9867,
      "step": 15940
    },
    {
      "epoch": 1.9193742478941034,
      "grad_norm": 3.989680051803589,
      "learning_rate": 0.00019991627900934082,
      "loss": 1.9552,
      "step": 15950
    },
    {
      "epoch": 1.9205776173285198,
      "grad_norm": 6.894227981567383,
      "learning_rate": 0.00019991612330013905,
      "loss": 2.0555,
      "step": 15960
    },
    {
      "epoch": 1.9217809867629363,
      "grad_norm": 4.548664569854736,
      "learning_rate": 0.00019991596744633395,
      "loss": 1.9647,
      "step": 15970
    },
    {
      "epoch": 1.9229843561973525,
      "grad_norm": 3.2041382789611816,
      "learning_rate": 0.0001999158114479258,
      "loss": 1.8206,
      "step": 15980
    },
    {
      "epoch": 1.924187725631769,
      "grad_norm": 5.597415447235107,
      "learning_rate": 0.0001999156553049148,
      "loss": 2.1227,
      "step": 15990
    },
    {
      "epoch": 1.9253910950661854,
      "grad_norm": 4.656121730804443,
      "learning_rate": 0.0001999154990173012,
      "loss": 2.095,
      "step": 16000
    },
    {
      "epoch": 1.9265944645006017,
      "grad_norm": 6.465431213378906,
      "learning_rate": 0.00019991534258508516,
      "loss": 1.9937,
      "step": 16010
    },
    {
      "epoch": 1.927797833935018,
      "grad_norm": 5.056748867034912,
      "learning_rate": 0.000199915186008267,
      "loss": 1.854,
      "step": 16020
    },
    {
      "epoch": 1.9290012033694344,
      "grad_norm": 3.3747501373291016,
      "learning_rate": 0.00019991502928684685,
      "loss": 2.0646,
      "step": 16030
    },
    {
      "epoch": 1.9302045728038508,
      "grad_norm": 5.906112194061279,
      "learning_rate": 0.00019991487242082502,
      "loss": 2.1587,
      "step": 16040
    },
    {
      "epoch": 1.931407942238267,
      "grad_norm": 3.7816739082336426,
      "learning_rate": 0.00019991471541020168,
      "loss": 1.8309,
      "step": 16050
    },
    {
      "epoch": 1.9326113116726835,
      "grad_norm": 5.560314655303955,
      "learning_rate": 0.00019991455825497706,
      "loss": 1.8769,
      "step": 16060
    },
    {
      "epoch": 1.9338146811071,
      "grad_norm": 5.638073444366455,
      "learning_rate": 0.00019991440095515147,
      "loss": 1.8768,
      "step": 16070
    },
    {
      "epoch": 1.9350180505415162,
      "grad_norm": 3.7518393993377686,
      "learning_rate": 0.000199914243510725,
      "loss": 2.0219,
      "step": 16080
    },
    {
      "epoch": 1.9362214199759324,
      "grad_norm": 5.50349235534668,
      "learning_rate": 0.000199914085921698,
      "loss": 1.8097,
      "step": 16090
    },
    {
      "epoch": 1.9374247894103491,
      "grad_norm": 4.016904354095459,
      "learning_rate": 0.00019991392818807064,
      "loss": 1.9738,
      "step": 16100
    },
    {
      "epoch": 1.9386281588447654,
      "grad_norm": 6.328729629516602,
      "learning_rate": 0.00019991377030984314,
      "loss": 1.9604,
      "step": 16110
    },
    {
      "epoch": 1.9398315282791816,
      "grad_norm": 5.4566168785095215,
      "learning_rate": 0.00019991361228701576,
      "loss": 2.1133,
      "step": 16120
    },
    {
      "epoch": 1.941034897713598,
      "grad_norm": 4.5634284019470215,
      "learning_rate": 0.00019991345411958872,
      "loss": 2.1536,
      "step": 16130
    },
    {
      "epoch": 1.9422382671480145,
      "grad_norm": 5.645349979400635,
      "learning_rate": 0.00019991329580756225,
      "loss": 1.9691,
      "step": 16140
    },
    {
      "epoch": 1.9434416365824307,
      "grad_norm": 4.688941955566406,
      "learning_rate": 0.00019991313735093655,
      "loss": 1.926,
      "step": 16150
    },
    {
      "epoch": 1.9446450060168472,
      "grad_norm": 6.3975043296813965,
      "learning_rate": 0.0001999129787497119,
      "loss": 1.9637,
      "step": 16160
    },
    {
      "epoch": 1.9458483754512637,
      "grad_norm": 4.604057312011719,
      "learning_rate": 0.0001999128200038885,
      "loss": 1.8218,
      "step": 16170
    },
    {
      "epoch": 1.94705174488568,
      "grad_norm": 3.644859552383423,
      "learning_rate": 0.00019991266111346655,
      "loss": 1.7935,
      "step": 16180
    },
    {
      "epoch": 1.9482551143200961,
      "grad_norm": 4.452095031738281,
      "learning_rate": 0.00019991250207844634,
      "loss": 1.8601,
      "step": 16190
    },
    {
      "epoch": 1.9494584837545126,
      "grad_norm": 4.731088161468506,
      "learning_rate": 0.00019991234289882804,
      "loss": 2.1167,
      "step": 16200
    },
    {
      "epoch": 1.950661853188929,
      "grad_norm": 7.040933132171631,
      "learning_rate": 0.00019991218357461195,
      "loss": 2.0416,
      "step": 16210
    },
    {
      "epoch": 1.9518652226233453,
      "grad_norm": 6.245818138122559,
      "learning_rate": 0.00019991202410579826,
      "loss": 1.96,
      "step": 16220
    },
    {
      "epoch": 1.9530685920577617,
      "grad_norm": 3.276776075363159,
      "learning_rate": 0.0001999118644923872,
      "loss": 2.1505,
      "step": 16230
    },
    {
      "epoch": 1.9542719614921782,
      "grad_norm": 6.008669853210449,
      "learning_rate": 0.00019991170473437897,
      "loss": 2.0642,
      "step": 16240
    },
    {
      "epoch": 1.9554753309265944,
      "grad_norm": 3.994748592376709,
      "learning_rate": 0.00019991154483177387,
      "loss": 1.776,
      "step": 16250
    },
    {
      "epoch": 1.9566787003610109,
      "grad_norm": 6.522171497344971,
      "learning_rate": 0.0001999113847845721,
      "loss": 2.0933,
      "step": 16260
    },
    {
      "epoch": 1.9578820697954273,
      "grad_norm": 4.953426837921143,
      "learning_rate": 0.00019991122459277384,
      "loss": 1.8951,
      "step": 16270
    },
    {
      "epoch": 1.9590854392298436,
      "grad_norm": 3.7762913703918457,
      "learning_rate": 0.00019991106425637942,
      "loss": 2.1214,
      "step": 16280
    },
    {
      "epoch": 1.9602888086642598,
      "grad_norm": 4.925233840942383,
      "learning_rate": 0.000199910903775389,
      "loss": 1.9764,
      "step": 16290
    },
    {
      "epoch": 1.9614921780986763,
      "grad_norm": 3.9478254318237305,
      "learning_rate": 0.00019991074314980287,
      "loss": 2.0233,
      "step": 16300
    },
    {
      "epoch": 1.9626955475330927,
      "grad_norm": 6.4193806648254395,
      "learning_rate": 0.0001999105823796212,
      "loss": 2.092,
      "step": 16310
    },
    {
      "epoch": 1.963898916967509,
      "grad_norm": 5.982722759246826,
      "learning_rate": 0.00019991042146484425,
      "loss": 1.7777,
      "step": 16320
    },
    {
      "epoch": 1.9651022864019254,
      "grad_norm": 3.0726537704467773,
      "learning_rate": 0.00019991026040547225,
      "loss": 2.0554,
      "step": 16330
    },
    {
      "epoch": 1.9663056558363419,
      "grad_norm": 4.883552551269531,
      "learning_rate": 0.00019991009920150546,
      "loss": 1.9911,
      "step": 16340
    },
    {
      "epoch": 1.967509025270758,
      "grad_norm": 4.612387180328369,
      "learning_rate": 0.00019990993785294408,
      "loss": 2.0314,
      "step": 16350
    },
    {
      "epoch": 1.9687123947051743,
      "grad_norm": 6.7942328453063965,
      "learning_rate": 0.00019990977635978835,
      "loss": 1.7717,
      "step": 16360
    },
    {
      "epoch": 1.969915764139591,
      "grad_norm": 4.324357986450195,
      "learning_rate": 0.00019990961472203848,
      "loss": 1.9512,
      "step": 16370
    },
    {
      "epoch": 1.9711191335740073,
      "grad_norm": 3.6651668548583984,
      "learning_rate": 0.00019990945293969477,
      "loss": 2.0161,
      "step": 16380
    },
    {
      "epoch": 1.9723225030084235,
      "grad_norm": 5.03709077835083,
      "learning_rate": 0.0001999092910127574,
      "loss": 2.1105,
      "step": 16390
    },
    {
      "epoch": 1.97352587244284,
      "grad_norm": 4.680255889892578,
      "learning_rate": 0.00019990912894122663,
      "loss": 1.9587,
      "step": 16400
    },
    {
      "epoch": 1.9747292418772564,
      "grad_norm": 7.243875026702881,
      "learning_rate": 0.00019990896672510266,
      "loss": 2.0094,
      "step": 16410
    },
    {
      "epoch": 1.9759326113116726,
      "grad_norm": 4.893266677856445,
      "learning_rate": 0.00019990880436438575,
      "loss": 1.9973,
      "step": 16420
    },
    {
      "epoch": 1.977135980746089,
      "grad_norm": 3.392618417739868,
      "learning_rate": 0.00019990864185907614,
      "loss": 2.0151,
      "step": 16430
    },
    {
      "epoch": 1.9783393501805056,
      "grad_norm": 4.7665696144104,
      "learning_rate": 0.00019990847920917405,
      "loss": 1.6015,
      "step": 16440
    },
    {
      "epoch": 1.9795427196149218,
      "grad_norm": 4.276864528656006,
      "learning_rate": 0.00019990831641467972,
      "loss": 1.9985,
      "step": 16450
    },
    {
      "epoch": 1.980746089049338,
      "grad_norm": 8.40011215209961,
      "learning_rate": 0.00019990815347559342,
      "loss": 2.0585,
      "step": 16460
    },
    {
      "epoch": 1.9819494584837545,
      "grad_norm": 5.415799617767334,
      "learning_rate": 0.00019990799039191535,
      "loss": 1.9161,
      "step": 16470
    },
    {
      "epoch": 1.983152827918171,
      "grad_norm": 3.8374929428100586,
      "learning_rate": 0.00019990782716364575,
      "loss": 2.0741,
      "step": 16480
    },
    {
      "epoch": 1.9843561973525872,
      "grad_norm": 5.7505083084106445,
      "learning_rate": 0.00019990766379078485,
      "loss": 1.941,
      "step": 16490
    },
    {
      "epoch": 1.9855595667870036,
      "grad_norm": 4.585755348205566,
      "learning_rate": 0.00019990750027333286,
      "loss": 1.572,
      "step": 16500
    },
    {
      "epoch": 1.98676293622142,
      "grad_norm": 6.762844562530518,
      "learning_rate": 0.00019990733661129012,
      "loss": 2.2276,
      "step": 16510
    },
    {
      "epoch": 1.9879663056558363,
      "grad_norm": 4.439902305603027,
      "learning_rate": 0.00019990717280465674,
      "loss": 2.1657,
      "step": 16520
    },
    {
      "epoch": 1.9891696750902526,
      "grad_norm": 3.378763437271118,
      "learning_rate": 0.00019990700885343302,
      "loss": 2.13,
      "step": 16530
    },
    {
      "epoch": 1.9903730445246692,
      "grad_norm": 5.796677589416504,
      "learning_rate": 0.00019990684475761924,
      "loss": 1.8933,
      "step": 16540
    },
    {
      "epoch": 1.9915764139590855,
      "grad_norm": 4.522659778594971,
      "learning_rate": 0.00019990668051721554,
      "loss": 1.9627,
      "step": 16550
    },
    {
      "epoch": 1.9927797833935017,
      "grad_norm": 6.611879825592041,
      "learning_rate": 0.00019990651613222224,
      "loss": 1.9534,
      "step": 16560
    },
    {
      "epoch": 1.9939831528279182,
      "grad_norm": 4.917909622192383,
      "learning_rate": 0.00019990635160263952,
      "loss": 1.864,
      "step": 16570
    },
    {
      "epoch": 1.9951865222623346,
      "grad_norm": 3.834904193878174,
      "learning_rate": 0.00019990618692846765,
      "loss": 2.1645,
      "step": 16580
    },
    {
      "epoch": 1.9963898916967509,
      "grad_norm": 7.771172046661377,
      "learning_rate": 0.00019990602210970688,
      "loss": 2.0834,
      "step": 16590
    },
    {
      "epoch": 1.9975932611311673,
      "grad_norm": 3.9011645317077637,
      "learning_rate": 0.00019990585714635742,
      "loss": 2.2294,
      "step": 16600
    },
    {
      "epoch": 1.9987966305655838,
      "grad_norm": 8.939842224121094,
      "learning_rate": 0.00019990569203841952,
      "loss": 1.7705,
      "step": 16610
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.323520183563232,
      "learning_rate": 0.0001999055267858934,
      "loss": 1.7649,
      "step": 16620
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.9142165184020996,
      "eval_runtime": 118.7366,
      "eval_samples_per_second": 62.213,
      "eval_steps_per_second": 7.782,
      "step": 16620
    },
    {
      "epoch": 2.0012033694344162,
      "grad_norm": 5.1999969482421875,
      "learning_rate": 0.0001999053613887793,
      "loss": 1.9435,
      "step": 16630
    },
    {
      "epoch": 2.002406738868833,
      "grad_norm": 3.153641700744629,
      "learning_rate": 0.00019990519584707755,
      "loss": 1.6858,
      "step": 16640
    },
    {
      "epoch": 2.003610108303249,
      "grad_norm": 5.022435665130615,
      "learning_rate": 0.00019990503016078823,
      "loss": 1.8025,
      "step": 16650
    },
    {
      "epoch": 2.0048134777376654,
      "grad_norm": 4.489079475402832,
      "learning_rate": 0.00019990486432991172,
      "loss": 1.8494,
      "step": 16660
    },
    {
      "epoch": 2.0060168471720816,
      "grad_norm": 6.600327968597412,
      "learning_rate": 0.00019990469835444818,
      "loss": 1.8307,
      "step": 16670
    },
    {
      "epoch": 2.0072202166064983,
      "grad_norm": 6.060215950012207,
      "learning_rate": 0.0001999045322343979,
      "loss": 1.8325,
      "step": 16680
    },
    {
      "epoch": 2.0084235860409145,
      "grad_norm": 3.5549871921539307,
      "learning_rate": 0.00019990436596976108,
      "loss": 1.8593,
      "step": 16690
    },
    {
      "epoch": 2.0096269554753308,
      "grad_norm": 5.249209880828857,
      "learning_rate": 0.00019990419956053798,
      "loss": 1.8023,
      "step": 16700
    },
    {
      "epoch": 2.0108303249097474,
      "grad_norm": 3.7713396549224854,
      "learning_rate": 0.00019990403300672885,
      "loss": 1.6932,
      "step": 16710
    },
    {
      "epoch": 2.0120336943441637,
      "grad_norm": 6.833237171173096,
      "learning_rate": 0.0001999038663083339,
      "loss": 1.8434,
      "step": 16720
    },
    {
      "epoch": 2.01323706377858,
      "grad_norm": 4.462503433227539,
      "learning_rate": 0.0001999036994653534,
      "loss": 1.8469,
      "step": 16730
    },
    {
      "epoch": 2.0144404332129966,
      "grad_norm": 4.416675567626953,
      "learning_rate": 0.00019990353247778756,
      "loss": 1.8928,
      "step": 16740
    },
    {
      "epoch": 2.015643802647413,
      "grad_norm": 5.76512336730957,
      "learning_rate": 0.00019990336534563666,
      "loss": 1.7162,
      "step": 16750
    },
    {
      "epoch": 2.016847172081829,
      "grad_norm": 4.121241569519043,
      "learning_rate": 0.0001999031980689009,
      "loss": 2.0468,
      "step": 16760
    },
    {
      "epoch": 2.0180505415162453,
      "grad_norm": 6.77939510345459,
      "learning_rate": 0.0001999030306475806,
      "loss": 2.0418,
      "step": 16770
    },
    {
      "epoch": 2.019253910950662,
      "grad_norm": 5.205681800842285,
      "learning_rate": 0.00019990286308167594,
      "loss": 1.8573,
      "step": 16780
    },
    {
      "epoch": 2.020457280385078,
      "grad_norm": 3.1619973182678223,
      "learning_rate": 0.00019990269537118716,
      "loss": 1.9974,
      "step": 16790
    },
    {
      "epoch": 2.0216606498194944,
      "grad_norm": 5.370748043060303,
      "learning_rate": 0.0001999025275161145,
      "loss": 1.9107,
      "step": 16800
    },
    {
      "epoch": 2.022864019253911,
      "grad_norm": 4.123819351196289,
      "learning_rate": 0.00019990235951645823,
      "loss": 2.043,
      "step": 16810
    },
    {
      "epoch": 2.0240673886883274,
      "grad_norm": 6.360437393188477,
      "learning_rate": 0.00019990219137221857,
      "loss": 1.9069,
      "step": 16820
    },
    {
      "epoch": 2.0252707581227436,
      "grad_norm": 5.279680252075195,
      "learning_rate": 0.00019990202308339577,
      "loss": 1.6415,
      "step": 16830
    },
    {
      "epoch": 2.02647412755716,
      "grad_norm": 3.7279410362243652,
      "learning_rate": 0.0001999018546499901,
      "loss": 1.855,
      "step": 16840
    },
    {
      "epoch": 2.0276774969915765,
      "grad_norm": 5.45247745513916,
      "learning_rate": 0.00019990168607200178,
      "loss": 2.0367,
      "step": 16850
    },
    {
      "epoch": 2.0288808664259927,
      "grad_norm": 4.370023727416992,
      "learning_rate": 0.00019990151734943103,
      "loss": 1.8561,
      "step": 16860
    },
    {
      "epoch": 2.030084235860409,
      "grad_norm": 7.897660255432129,
      "learning_rate": 0.00019990134848227813,
      "loss": 1.8043,
      "step": 16870
    },
    {
      "epoch": 2.0312876052948257,
      "grad_norm": 4.6693010330200195,
      "learning_rate": 0.00019990117947054335,
      "loss": 1.7646,
      "step": 16880
    },
    {
      "epoch": 2.032490974729242,
      "grad_norm": 3.690429210662842,
      "learning_rate": 0.00019990101031422684,
      "loss": 1.9676,
      "step": 16890
    },
    {
      "epoch": 2.033694344163658,
      "grad_norm": 6.373749256134033,
      "learning_rate": 0.00019990084101332895,
      "loss": 1.8346,
      "step": 16900
    },
    {
      "epoch": 2.034897713598075,
      "grad_norm": 4.280575752258301,
      "learning_rate": 0.00019990067156784984,
      "loss": 1.7286,
      "step": 16910
    },
    {
      "epoch": 2.036101083032491,
      "grad_norm": 6.801653861999512,
      "learning_rate": 0.00019990050197778983,
      "loss": 2.0454,
      "step": 16920
    },
    {
      "epoch": 2.0373044524669073,
      "grad_norm": 5.937229633331299,
      "learning_rate": 0.0001999003322431491,
      "loss": 1.6404,
      "step": 16930
    },
    {
      "epoch": 2.0385078219013235,
      "grad_norm": 4.118158340454102,
      "learning_rate": 0.00019990016236392792,
      "loss": 1.7363,
      "step": 16940
    },
    {
      "epoch": 2.03971119133574,
      "grad_norm": 6.1648850440979,
      "learning_rate": 0.00019989999234012657,
      "loss": 1.8067,
      "step": 16950
    },
    {
      "epoch": 2.0409145607701564,
      "grad_norm": 4.802121639251709,
      "learning_rate": 0.00019989982217174525,
      "loss": 1.926,
      "step": 16960
    },
    {
      "epoch": 2.0421179302045727,
      "grad_norm": 6.523443698883057,
      "learning_rate": 0.00019989965185878423,
      "loss": 1.6961,
      "step": 16970
    },
    {
      "epoch": 2.0433212996389893,
      "grad_norm": 5.312000274658203,
      "learning_rate": 0.00019989948140124376,
      "loss": 1.8128,
      "step": 16980
    },
    {
      "epoch": 2.0445246690734056,
      "grad_norm": 5.419356822967529,
      "learning_rate": 0.00019989931079912406,
      "loss": 1.841,
      "step": 16990
    },
    {
      "epoch": 2.045728038507822,
      "grad_norm": 4.926950931549072,
      "learning_rate": 0.00019989914005242538,
      "loss": 1.7705,
      "step": 17000
    },
    {
      "epoch": 2.046931407942238,
      "grad_norm": 4.7852654457092285,
      "learning_rate": 0.000199898969161148,
      "loss": 1.6895,
      "step": 17010
    },
    {
      "epoch": 2.0481347773766547,
      "grad_norm": 6.510125160217285,
      "learning_rate": 0.00019989879812529212,
      "loss": 1.8186,
      "step": 17020
    },
    {
      "epoch": 2.049338146811071,
      "grad_norm": 5.870050430297852,
      "learning_rate": 0.00019989862694485806,
      "loss": 1.6761,
      "step": 17030
    },
    {
      "epoch": 2.050541516245487,
      "grad_norm": 4.159613609313965,
      "learning_rate": 0.000199898455619846,
      "loss": 1.797,
      "step": 17040
    },
    {
      "epoch": 2.051744885679904,
      "grad_norm": 5.965736389160156,
      "learning_rate": 0.00019989828415025622,
      "loss": 1.8211,
      "step": 17050
    },
    {
      "epoch": 2.05294825511432,
      "grad_norm": 5.167622089385986,
      "learning_rate": 0.00019989811253608896,
      "loss": 1.7205,
      "step": 17060
    },
    {
      "epoch": 2.0541516245487363,
      "grad_norm": 5.409761905670166,
      "learning_rate": 0.00019989794077734441,
      "loss": 1.7689,
      "step": 17070
    },
    {
      "epoch": 2.055354993983153,
      "grad_norm": 5.324963092803955,
      "learning_rate": 0.00019989776887402295,
      "loss": 1.7323,
      "step": 17080
    },
    {
      "epoch": 2.0565583634175693,
      "grad_norm": 4.008548736572266,
      "learning_rate": 0.00019989759682612472,
      "loss": 1.6856,
      "step": 17090
    },
    {
      "epoch": 2.0577617328519855,
      "grad_norm": 6.131975173950195,
      "learning_rate": 0.00019989742463365003,
      "loss": 1.8537,
      "step": 17100
    },
    {
      "epoch": 2.0589651022864017,
      "grad_norm": 3.6276772022247314,
      "learning_rate": 0.00019989725229659905,
      "loss": 1.5171,
      "step": 17110
    },
    {
      "epoch": 2.0601684717208184,
      "grad_norm": 7.486250400543213,
      "learning_rate": 0.0001998970798149721,
      "loss": 1.7464,
      "step": 17120
    },
    {
      "epoch": 2.0613718411552346,
      "grad_norm": 5.081457138061523,
      "learning_rate": 0.00019989690718876944,
      "loss": 1.8673,
      "step": 17130
    },
    {
      "epoch": 2.062575210589651,
      "grad_norm": 4.898556232452393,
      "learning_rate": 0.0001998967344179913,
      "loss": 1.8692,
      "step": 17140
    },
    {
      "epoch": 2.0637785800240676,
      "grad_norm": 7.015048980712891,
      "learning_rate": 0.0001998965615026379,
      "loss": 1.9418,
      "step": 17150
    },
    {
      "epoch": 2.064981949458484,
      "grad_norm": 4.729976177215576,
      "learning_rate": 0.0001998963884427095,
      "loss": 2.0248,
      "step": 17160
    },
    {
      "epoch": 2.0661853188929,
      "grad_norm": 7.050203800201416,
      "learning_rate": 0.00019989621523820635,
      "loss": 1.7396,
      "step": 17170
    },
    {
      "epoch": 2.0673886883273163,
      "grad_norm": 5.0453948974609375,
      "learning_rate": 0.00019989604188912875,
      "loss": 1.8594,
      "step": 17180
    },
    {
      "epoch": 2.068592057761733,
      "grad_norm": 3.3906519412994385,
      "learning_rate": 0.0001998958683954769,
      "loss": 1.9991,
      "step": 17190
    },
    {
      "epoch": 2.069795427196149,
      "grad_norm": 5.707901477813721,
      "learning_rate": 0.00019989569475725106,
      "loss": 1.8237,
      "step": 17200
    },
    {
      "epoch": 2.0709987966305654,
      "grad_norm": 4.8201470375061035,
      "learning_rate": 0.0001998955209744515,
      "loss": 1.6789,
      "step": 17210
    },
    {
      "epoch": 2.072202166064982,
      "grad_norm": 7.725179672241211,
      "learning_rate": 0.00019989534704707845,
      "loss": 1.7809,
      "step": 17220
    },
    {
      "epoch": 2.0734055354993983,
      "grad_norm": 5.349000930786133,
      "learning_rate": 0.00019989517297513217,
      "loss": 1.8191,
      "step": 17230
    },
    {
      "epoch": 2.0746089049338146,
      "grad_norm": 3.1116085052490234,
      "learning_rate": 0.00019989499875861292,
      "loss": 1.7864,
      "step": 17240
    },
    {
      "epoch": 2.0758122743682312,
      "grad_norm": 5.141925811767578,
      "learning_rate": 0.00019989482439752092,
      "loss": 1.6877,
      "step": 17250
    },
    {
      "epoch": 2.0770156438026475,
      "grad_norm": 3.531400680541992,
      "learning_rate": 0.00019989464989185647,
      "loss": 1.7856,
      "step": 17260
    },
    {
      "epoch": 2.0782190132370637,
      "grad_norm": 6.522139072418213,
      "learning_rate": 0.00019989447524161976,
      "loss": 1.8581,
      "step": 17270
    },
    {
      "epoch": 2.07942238267148,
      "grad_norm": 5.104706764221191,
      "learning_rate": 0.00019989430044681114,
      "loss": 1.766,
      "step": 17280
    },
    {
      "epoch": 2.0806257521058966,
      "grad_norm": 3.9711599349975586,
      "learning_rate": 0.00019989412550743075,
      "loss": 1.8496,
      "step": 17290
    },
    {
      "epoch": 2.081829121540313,
      "grad_norm": 5.783777236938477,
      "learning_rate": 0.00019989395042347894,
      "loss": 1.7734,
      "step": 17300
    },
    {
      "epoch": 2.083032490974729,
      "grad_norm": 4.533494472503662,
      "learning_rate": 0.0001998937751949559,
      "loss": 1.828,
      "step": 17310
    },
    {
      "epoch": 2.0842358604091458,
      "grad_norm": 7.173120975494385,
      "learning_rate": 0.0001998935998218619,
      "loss": 2.0683,
      "step": 17320
    },
    {
      "epoch": 2.085439229843562,
      "grad_norm": 5.991465091705322,
      "learning_rate": 0.00019989342430419722,
      "loss": 1.5061,
      "step": 17330
    },
    {
      "epoch": 2.0866425992779782,
      "grad_norm": 3.940812110900879,
      "learning_rate": 0.00019989324864196206,
      "loss": 1.9932,
      "step": 17340
    },
    {
      "epoch": 2.087845968712395,
      "grad_norm": 5.346851348876953,
      "learning_rate": 0.00019989307283515672,
      "loss": 1.7388,
      "step": 17350
    },
    {
      "epoch": 2.089049338146811,
      "grad_norm": 5.48711633682251,
      "learning_rate": 0.00019989289688378143,
      "loss": 1.7923,
      "step": 17360
    },
    {
      "epoch": 2.0902527075812274,
      "grad_norm": 6.613363265991211,
      "learning_rate": 0.0001998927207878365,
      "loss": 1.8068,
      "step": 17370
    },
    {
      "epoch": 2.0914560770156436,
      "grad_norm": 4.155871868133545,
      "learning_rate": 0.0001998925445473221,
      "loss": 1.7835,
      "step": 17380
    },
    {
      "epoch": 2.0926594464500603,
      "grad_norm": 3.4781439304351807,
      "learning_rate": 0.00019989236816223853,
      "loss": 1.8167,
      "step": 17390
    },
    {
      "epoch": 2.0938628158844765,
      "grad_norm": 6.0336432456970215,
      "learning_rate": 0.00019989219163258604,
      "loss": 1.7585,
      "step": 17400
    },
    {
      "epoch": 2.0950661853188928,
      "grad_norm": 4.189842224121094,
      "learning_rate": 0.00019989201495836489,
      "loss": 1.676,
      "step": 17410
    },
    {
      "epoch": 2.0962695547533094,
      "grad_norm": 7.258090019226074,
      "learning_rate": 0.00019989183813957532,
      "loss": 1.8626,
      "step": 17420
    },
    {
      "epoch": 2.0974729241877257,
      "grad_norm": 5.412322044372559,
      "learning_rate": 0.00019989166117621762,
      "loss": 1.615,
      "step": 17430
    },
    {
      "epoch": 2.098676293622142,
      "grad_norm": 4.196578502655029,
      "learning_rate": 0.00019989148406829202,
      "loss": 1.7245,
      "step": 17440
    },
    {
      "epoch": 2.099879663056558,
      "grad_norm": 6.083682537078857,
      "learning_rate": 0.00019989130681579876,
      "loss": 1.872,
      "step": 17450
    },
    {
      "epoch": 2.101083032490975,
      "grad_norm": 4.3092732429504395,
      "learning_rate": 0.00019989112941873812,
      "loss": 1.863,
      "step": 17460
    },
    {
      "epoch": 2.102286401925391,
      "grad_norm": 6.418511867523193,
      "learning_rate": 0.00019989095187711038,
      "loss": 1.8506,
      "step": 17470
    },
    {
      "epoch": 2.1034897713598073,
      "grad_norm": 4.709066867828369,
      "learning_rate": 0.0001998907741909158,
      "loss": 1.8186,
      "step": 17480
    },
    {
      "epoch": 2.104693140794224,
      "grad_norm": 4.116019248962402,
      "learning_rate": 0.00019989059636015452,
      "loss": 1.8143,
      "step": 17490
    },
    {
      "epoch": 2.10589651022864,
      "grad_norm": 7.0124430656433105,
      "learning_rate": 0.0001998904183848269,
      "loss": 1.9522,
      "step": 17500
    },
    {
      "epoch": 2.1070998796630565,
      "grad_norm": 4.389957427978516,
      "learning_rate": 0.00019989024026493323,
      "loss": 1.9003,
      "step": 17510
    },
    {
      "epoch": 2.108303249097473,
      "grad_norm": 7.033724308013916,
      "learning_rate": 0.00019989006200047372,
      "loss": 1.6658,
      "step": 17520
    },
    {
      "epoch": 2.1095066185318894,
      "grad_norm": 4.837581157684326,
      "learning_rate": 0.0001998898835914486,
      "loss": 1.7855,
      "step": 17530
    },
    {
      "epoch": 2.1107099879663056,
      "grad_norm": 3.7902698516845703,
      "learning_rate": 0.00019988970503785817,
      "loss": 1.7216,
      "step": 17540
    },
    {
      "epoch": 2.111913357400722,
      "grad_norm": 7.119017124176025,
      "learning_rate": 0.00019988952633970267,
      "loss": 1.744,
      "step": 17550
    },
    {
      "epoch": 2.1131167268351385,
      "grad_norm": 4.171570777893066,
      "learning_rate": 0.00019988934749698233,
      "loss": 1.7703,
      "step": 17560
    },
    {
      "epoch": 2.1143200962695547,
      "grad_norm": 5.869996070861816,
      "learning_rate": 0.0001998891685096975,
      "loss": 1.7115,
      "step": 17570
    },
    {
      "epoch": 2.115523465703971,
      "grad_norm": 5.453281402587891,
      "learning_rate": 0.00019988898937784835,
      "loss": 1.5491,
      "step": 17580
    },
    {
      "epoch": 2.1167268351383877,
      "grad_norm": 3.6909027099609375,
      "learning_rate": 0.00019988881010143516,
      "loss": 1.9484,
      "step": 17590
    },
    {
      "epoch": 2.117930204572804,
      "grad_norm": 7.22687292098999,
      "learning_rate": 0.00019988863068045825,
      "loss": 1.7824,
      "step": 17600
    },
    {
      "epoch": 2.11913357400722,
      "grad_norm": 4.892089366912842,
      "learning_rate": 0.00019988845111491778,
      "loss": 1.6621,
      "step": 17610
    },
    {
      "epoch": 2.1203369434416364,
      "grad_norm": 6.53886079788208,
      "learning_rate": 0.00019988827140481408,
      "loss": 1.8786,
      "step": 17620
    },
    {
      "epoch": 2.121540312876053,
      "grad_norm": 5.709807395935059,
      "learning_rate": 0.00019988809155014738,
      "loss": 1.6452,
      "step": 17630
    },
    {
      "epoch": 2.1227436823104693,
      "grad_norm": 3.1700313091278076,
      "learning_rate": 0.00019988791155091797,
      "loss": 1.718,
      "step": 17640
    },
    {
      "epoch": 2.1239470517448855,
      "grad_norm": 5.557100296020508,
      "learning_rate": 0.00019988773140712607,
      "loss": 1.5421,
      "step": 17650
    },
    {
      "epoch": 2.125150421179302,
      "grad_norm": 4.083098888397217,
      "learning_rate": 0.00019988755111877198,
      "loss": 1.5452,
      "step": 17660
    },
    {
      "epoch": 2.1263537906137184,
      "grad_norm": 5.638491630554199,
      "learning_rate": 0.0001998873706858559,
      "loss": 1.7364,
      "step": 17670
    },
    {
      "epoch": 2.1275571600481347,
      "grad_norm": 5.190164089202881,
      "learning_rate": 0.00019988719010837817,
      "loss": 1.9735,
      "step": 17680
    },
    {
      "epoch": 2.1287605294825513,
      "grad_norm": 3.7146685123443604,
      "learning_rate": 0.00019988700938633903,
      "loss": 2.1372,
      "step": 17690
    },
    {
      "epoch": 2.1299638989169676,
      "grad_norm": 6.454514980316162,
      "learning_rate": 0.00019988682851973871,
      "loss": 2.0919,
      "step": 17700
    },
    {
      "epoch": 2.131167268351384,
      "grad_norm": 4.221381187438965,
      "learning_rate": 0.00019988664750857747,
      "loss": 1.7471,
      "step": 17710
    },
    {
      "epoch": 2.1323706377858,
      "grad_norm": 5.4702887535095215,
      "learning_rate": 0.0001998864663528556,
      "loss": 2.0442,
      "step": 17720
    },
    {
      "epoch": 2.1335740072202167,
      "grad_norm": 4.777248382568359,
      "learning_rate": 0.00019988628505257335,
      "loss": 2.0535,
      "step": 17730
    },
    {
      "epoch": 2.134777376654633,
      "grad_norm": 3.488917827606201,
      "learning_rate": 0.000199886103607731,
      "loss": 1.7097,
      "step": 17740
    },
    {
      "epoch": 2.135980746089049,
      "grad_norm": 5.957403182983398,
      "learning_rate": 0.00019988592201832875,
      "loss": 1.6971,
      "step": 17750
    },
    {
      "epoch": 2.137184115523466,
      "grad_norm": 4.465520858764648,
      "learning_rate": 0.00019988574028436697,
      "loss": 1.9346,
      "step": 17760
    },
    {
      "epoch": 2.138387484957882,
      "grad_norm": 6.462950706481934,
      "learning_rate": 0.00019988555840584583,
      "loss": 1.7709,
      "step": 17770
    },
    {
      "epoch": 2.1395908543922983,
      "grad_norm": 4.549928665161133,
      "learning_rate": 0.00019988537638276563,
      "loss": 1.9055,
      "step": 17780
    },
    {
      "epoch": 2.140794223826715,
      "grad_norm": 3.398759365081787,
      "learning_rate": 0.00019988519421512663,
      "loss": 1.6768,
      "step": 17790
    },
    {
      "epoch": 2.1419975932611313,
      "grad_norm": 4.999843597412109,
      "learning_rate": 0.0001998850119029291,
      "loss": 1.997,
      "step": 17800
    },
    {
      "epoch": 2.1432009626955475,
      "grad_norm": 4.756877899169922,
      "learning_rate": 0.0001998848294461733,
      "loss": 1.69,
      "step": 17810
    },
    {
      "epoch": 2.1444043321299637,
      "grad_norm": 5.443968296051025,
      "learning_rate": 0.00019988464684485947,
      "loss": 1.926,
      "step": 17820
    },
    {
      "epoch": 2.1456077015643804,
      "grad_norm": 5.2449421882629395,
      "learning_rate": 0.00019988446409898792,
      "loss": 1.8667,
      "step": 17830
    },
    {
      "epoch": 2.1468110709987966,
      "grad_norm": 4.510014533996582,
      "learning_rate": 0.00019988428120855888,
      "loss": 1.7223,
      "step": 17840
    },
    {
      "epoch": 2.148014440433213,
      "grad_norm": 6.149260997772217,
      "learning_rate": 0.0001998840981735726,
      "loss": 1.9277,
      "step": 17850
    },
    {
      "epoch": 2.1492178098676296,
      "grad_norm": 4.433861255645752,
      "learning_rate": 0.00019988391499402942,
      "loss": 1.765,
      "step": 17860
    },
    {
      "epoch": 2.150421179302046,
      "grad_norm": 6.538389682769775,
      "learning_rate": 0.00019988373166992954,
      "loss": 1.9675,
      "step": 17870
    },
    {
      "epoch": 2.151624548736462,
      "grad_norm": 3.9230477809906006,
      "learning_rate": 0.00019988354820127323,
      "loss": 1.6862,
      "step": 17880
    },
    {
      "epoch": 2.1528279181708783,
      "grad_norm": 3.716688632965088,
      "learning_rate": 0.00019988336458806073,
      "loss": 1.6858,
      "step": 17890
    },
    {
      "epoch": 2.154031287605295,
      "grad_norm": 5.82971715927124,
      "learning_rate": 0.00019988318083029237,
      "loss": 1.7986,
      "step": 17900
    },
    {
      "epoch": 2.155234657039711,
      "grad_norm": 4.714937686920166,
      "learning_rate": 0.0001998829969279684,
      "loss": 1.6655,
      "step": 17910
    },
    {
      "epoch": 2.1564380264741274,
      "grad_norm": 7.878839015960693,
      "learning_rate": 0.00019988281288108904,
      "loss": 1.6974,
      "step": 17920
    },
    {
      "epoch": 2.157641395908544,
      "grad_norm": 5.158947944641113,
      "learning_rate": 0.00019988262868965464,
      "loss": 1.965,
      "step": 17930
    },
    {
      "epoch": 2.1588447653429603,
      "grad_norm": 3.980877161026001,
      "learning_rate": 0.00019988244435366538,
      "loss": 1.9019,
      "step": 17940
    },
    {
      "epoch": 2.1600481347773766,
      "grad_norm": 5.513377666473389,
      "learning_rate": 0.00019988225987312158,
      "loss": 1.8467,
      "step": 17950
    },
    {
      "epoch": 2.1612515042117932,
      "grad_norm": 4.000430107116699,
      "learning_rate": 0.0001998820752480235,
      "loss": 1.9303,
      "step": 17960
    },
    {
      "epoch": 2.1624548736462095,
      "grad_norm": 7.605454444885254,
      "learning_rate": 0.00019988189047837136,
      "loss": 1.9393,
      "step": 17970
    },
    {
      "epoch": 2.1636582430806257,
      "grad_norm": 5.936460971832275,
      "learning_rate": 0.00019988170556416548,
      "loss": 1.7185,
      "step": 17980
    },
    {
      "epoch": 2.164861612515042,
      "grad_norm": 3.905332565307617,
      "learning_rate": 0.00019988152050540612,
      "loss": 1.9451,
      "step": 17990
    },
    {
      "epoch": 2.1660649819494586,
      "grad_norm": 5.122852325439453,
      "learning_rate": 0.00019988133530209355,
      "loss": 1.6526,
      "step": 18000
    },
    {
      "epoch": 2.167268351383875,
      "grad_norm": 5.5804667472839355,
      "learning_rate": 0.000199881149954228,
      "loss": 1.8796,
      "step": 18010
    },
    {
      "epoch": 2.168471720818291,
      "grad_norm": 7.340632438659668,
      "learning_rate": 0.00019988096446180978,
      "loss": 1.6315,
      "step": 18020
    },
    {
      "epoch": 2.1696750902527078,
      "grad_norm": 5.029829502105713,
      "learning_rate": 0.00019988077882483914,
      "loss": 1.535,
      "step": 18030
    },
    {
      "epoch": 2.170878459687124,
      "grad_norm": 3.562346935272217,
      "learning_rate": 0.00019988059304331635,
      "loss": 1.8802,
      "step": 18040
    },
    {
      "epoch": 2.1720818291215402,
      "grad_norm": 5.396111965179443,
      "learning_rate": 0.00019988040711724168,
      "loss": 1.8595,
      "step": 18050
    },
    {
      "epoch": 2.1732851985559565,
      "grad_norm": 3.975900173187256,
      "learning_rate": 0.00019988022104661544,
      "loss": 1.8975,
      "step": 18060
    },
    {
      "epoch": 2.174488567990373,
      "grad_norm": 6.670739650726318,
      "learning_rate": 0.0001998800348314378,
      "loss": 1.9552,
      "step": 18070
    },
    {
      "epoch": 2.1756919374247894,
      "grad_norm": 5.525782108306885,
      "learning_rate": 0.00019987984847170913,
      "loss": 1.7285,
      "step": 18080
    },
    {
      "epoch": 2.1768953068592056,
      "grad_norm": 3.629214286804199,
      "learning_rate": 0.00019987966196742963,
      "loss": 1.8154,
      "step": 18090
    },
    {
      "epoch": 2.1780986762936223,
      "grad_norm": 5.791319847106934,
      "learning_rate": 0.00019987947531859964,
      "loss": 1.8833,
      "step": 18100
    },
    {
      "epoch": 2.1793020457280385,
      "grad_norm": 5.108678817749023,
      "learning_rate": 0.00019987928852521937,
      "loss": 1.9112,
      "step": 18110
    },
    {
      "epoch": 2.1805054151624548,
      "grad_norm": 7.058563709259033,
      "learning_rate": 0.00019987910158728912,
      "loss": 1.8469,
      "step": 18120
    },
    {
      "epoch": 2.1817087845968715,
      "grad_norm": 5.198415756225586,
      "learning_rate": 0.00019987891450480913,
      "loss": 1.622,
      "step": 18130
    },
    {
      "epoch": 2.1829121540312877,
      "grad_norm": 3.8137941360473633,
      "learning_rate": 0.0001998787272777797,
      "loss": 2.0485,
      "step": 18140
    },
    {
      "epoch": 2.184115523465704,
      "grad_norm": 6.22781229019165,
      "learning_rate": 0.00019987853990620109,
      "loss": 1.7697,
      "step": 18150
    },
    {
      "epoch": 2.18531889290012,
      "grad_norm": 4.489179611206055,
      "learning_rate": 0.00019987835239007357,
      "loss": 1.9367,
      "step": 18160
    },
    {
      "epoch": 2.186522262334537,
      "grad_norm": 7.405348300933838,
      "learning_rate": 0.00019987816472939745,
      "loss": 1.9643,
      "step": 18170
    },
    {
      "epoch": 2.187725631768953,
      "grad_norm": 5.456291198730469,
      "learning_rate": 0.00019987797692417293,
      "loss": 1.6383,
      "step": 18180
    },
    {
      "epoch": 2.1889290012033693,
      "grad_norm": 4.349714756011963,
      "learning_rate": 0.00019987778897440032,
      "loss": 1.8251,
      "step": 18190
    },
    {
      "epoch": 2.190132370637786,
      "grad_norm": 4.806549072265625,
      "learning_rate": 0.00019987760088007993,
      "loss": 1.6985,
      "step": 18200
    },
    {
      "epoch": 2.191335740072202,
      "grad_norm": 4.846847057342529,
      "learning_rate": 0.00019987741264121194,
      "loss": 1.663,
      "step": 18210
    },
    {
      "epoch": 2.1925391095066185,
      "grad_norm": 6.967729568481445,
      "learning_rate": 0.00019987722425779668,
      "loss": 2.0329,
      "step": 18220
    },
    {
      "epoch": 2.1937424789410347,
      "grad_norm": 4.565564155578613,
      "learning_rate": 0.00019987703572983445,
      "loss": 1.6841,
      "step": 18230
    },
    {
      "epoch": 2.1949458483754514,
      "grad_norm": 4.039554119110107,
      "learning_rate": 0.00019987684705732547,
      "loss": 1.8231,
      "step": 18240
    },
    {
      "epoch": 2.1961492178098676,
      "grad_norm": 5.261109828948975,
      "learning_rate": 0.00019987665824027005,
      "loss": 1.7919,
      "step": 18250
    },
    {
      "epoch": 2.197352587244284,
      "grad_norm": 4.446812152862549,
      "learning_rate": 0.00019987646927866842,
      "loss": 1.9482,
      "step": 18260
    },
    {
      "epoch": 2.1985559566787005,
      "grad_norm": 6.604482173919678,
      "learning_rate": 0.0001998762801725209,
      "loss": 2.0712,
      "step": 18270
    },
    {
      "epoch": 2.1997593261131168,
      "grad_norm": 5.471787452697754,
      "learning_rate": 0.00019987609092182772,
      "loss": 1.7306,
      "step": 18280
    },
    {
      "epoch": 2.200962695547533,
      "grad_norm": 4.047811508178711,
      "learning_rate": 0.00019987590152658921,
      "loss": 1.7998,
      "step": 18290
    },
    {
      "epoch": 2.2021660649819497,
      "grad_norm": 4.471895217895508,
      "learning_rate": 0.00019987571198680558,
      "loss": 1.5925,
      "step": 18300
    },
    {
      "epoch": 2.203369434416366,
      "grad_norm": 3.748727798461914,
      "learning_rate": 0.00019987552230247717,
      "loss": 1.7194,
      "step": 18310
    },
    {
      "epoch": 2.204572803850782,
      "grad_norm": 7.51377534866333,
      "learning_rate": 0.0001998753324736042,
      "loss": 2.0262,
      "step": 18320
    },
    {
      "epoch": 2.2057761732851984,
      "grad_norm": 5.626355171203613,
      "learning_rate": 0.00019987514250018696,
      "loss": 1.9133,
      "step": 18330
    },
    {
      "epoch": 2.206979542719615,
      "grad_norm": 3.5132031440734863,
      "learning_rate": 0.00019987495238222577,
      "loss": 2.0141,
      "step": 18340
    },
    {
      "epoch": 2.2081829121540313,
      "grad_norm": 5.36952018737793,
      "learning_rate": 0.00019987476211972082,
      "loss": 1.6841,
      "step": 18350
    },
    {
      "epoch": 2.2093862815884475,
      "grad_norm": 4.150166034698486,
      "learning_rate": 0.00019987457171267246,
      "loss": 1.7867,
      "step": 18360
    },
    {
      "epoch": 2.210589651022864,
      "grad_norm": 6.524456024169922,
      "learning_rate": 0.0001998743811610809,
      "loss": 2.0483,
      "step": 18370
    },
    {
      "epoch": 2.2117930204572804,
      "grad_norm": 5.273902416229248,
      "learning_rate": 0.0001998741904649465,
      "loss": 1.8468,
      "step": 18380
    },
    {
      "epoch": 2.2129963898916967,
      "grad_norm": 3.610470771789551,
      "learning_rate": 0.00019987399962426947,
      "loss": 1.8327,
      "step": 18390
    },
    {
      "epoch": 2.214199759326113,
      "grad_norm": 6.863369464874268,
      "learning_rate": 0.0001998738086390501,
      "loss": 1.7468,
      "step": 18400
    },
    {
      "epoch": 2.2154031287605296,
      "grad_norm": 3.936666250228882,
      "learning_rate": 0.00019987361750928865,
      "loss": 1.7581,
      "step": 18410
    },
    {
      "epoch": 2.216606498194946,
      "grad_norm": 7.433270454406738,
      "learning_rate": 0.00019987342623498545,
      "loss": 1.9996,
      "step": 18420
    },
    {
      "epoch": 2.217809867629362,
      "grad_norm": 5.657661437988281,
      "learning_rate": 0.00019987323481614072,
      "loss": 1.7346,
      "step": 18430
    },
    {
      "epoch": 2.2190132370637787,
      "grad_norm": 3.727800130844116,
      "learning_rate": 0.00019987304325275477,
      "loss": 1.7012,
      "step": 18440
    },
    {
      "epoch": 2.220216606498195,
      "grad_norm": 6.024631977081299,
      "learning_rate": 0.00019987285154482787,
      "loss": 1.7962,
      "step": 18450
    },
    {
      "epoch": 2.221419975932611,
      "grad_norm": 7.292879104614258,
      "learning_rate": 0.0001998726596923603,
      "loss": 2.1374,
      "step": 18460
    },
    {
      "epoch": 2.222623345367028,
      "grad_norm": 6.066665172576904,
      "learning_rate": 0.00019987246769535234,
      "loss": 2.0177,
      "step": 18470
    },
    {
      "epoch": 2.223826714801444,
      "grad_norm": 4.70356559753418,
      "learning_rate": 0.00019987227555380425,
      "loss": 1.9923,
      "step": 18480
    },
    {
      "epoch": 2.2250300842358604,
      "grad_norm": 3.302060604095459,
      "learning_rate": 0.00019987208326771634,
      "loss": 1.9954,
      "step": 18490
    },
    {
      "epoch": 2.2262334536702766,
      "grad_norm": 5.334634304046631,
      "learning_rate": 0.00019987189083708884,
      "loss": 1.8809,
      "step": 18500
    },
    {
      "epoch": 2.2274368231046933,
      "grad_norm": 4.367860794067383,
      "learning_rate": 0.00019987169826192206,
      "loss": 1.9273,
      "step": 18510
    },
    {
      "epoch": 2.2286401925391095,
      "grad_norm": 5.571081161499023,
      "learning_rate": 0.00019987150554221628,
      "loss": 1.6665,
      "step": 18520
    },
    {
      "epoch": 2.2298435619735257,
      "grad_norm": 4.982118606567383,
      "learning_rate": 0.0001998713126779718,
      "loss": 1.7887,
      "step": 18530
    },
    {
      "epoch": 2.2310469314079424,
      "grad_norm": 4.198168754577637,
      "learning_rate": 0.00019987111966918882,
      "loss": 1.7969,
      "step": 18540
    },
    {
      "epoch": 2.2322503008423586,
      "grad_norm": 5.355137825012207,
      "learning_rate": 0.0001998709265158677,
      "loss": 1.7528,
      "step": 18550
    },
    {
      "epoch": 2.233453670276775,
      "grad_norm": 4.837794303894043,
      "learning_rate": 0.0001998707332180087,
      "loss": 1.7877,
      "step": 18560
    },
    {
      "epoch": 2.234657039711191,
      "grad_norm": 7.143967628479004,
      "learning_rate": 0.0001998705397756121,
      "loss": 1.9044,
      "step": 18570
    },
    {
      "epoch": 2.235860409145608,
      "grad_norm": 5.132889747619629,
      "learning_rate": 0.00019987034618867812,
      "loss": 1.6545,
      "step": 18580
    },
    {
      "epoch": 2.237063778580024,
      "grad_norm": 3.638214588165283,
      "learning_rate": 0.00019987015245720715,
      "loss": 1.9747,
      "step": 18590
    },
    {
      "epoch": 2.2382671480144403,
      "grad_norm": 5.9062604904174805,
      "learning_rate": 0.0001998699585811994,
      "loss": 2.0217,
      "step": 18600
    },
    {
      "epoch": 2.239470517448857,
      "grad_norm": 4.019762992858887,
      "learning_rate": 0.00019986976456065514,
      "loss": 1.9784,
      "step": 18610
    },
    {
      "epoch": 2.240673886883273,
      "grad_norm": 7.049264907836914,
      "learning_rate": 0.0001998695703955747,
      "loss": 1.855,
      "step": 18620
    },
    {
      "epoch": 2.2418772563176894,
      "grad_norm": 4.697753429412842,
      "learning_rate": 0.00019986937608595834,
      "loss": 1.8122,
      "step": 18630
    },
    {
      "epoch": 2.243080625752106,
      "grad_norm": 3.323150157928467,
      "learning_rate": 0.0001998691816318063,
      "loss": 1.8771,
      "step": 18640
    },
    {
      "epoch": 2.2442839951865223,
      "grad_norm": 4.828363418579102,
      "learning_rate": 0.0001998689870331189,
      "loss": 1.7434,
      "step": 18650
    },
    {
      "epoch": 2.2454873646209386,
      "grad_norm": 4.217282772064209,
      "learning_rate": 0.00019986879228989644,
      "loss": 1.8604,
      "step": 18660
    },
    {
      "epoch": 2.246690734055355,
      "grad_norm": 7.136775493621826,
      "learning_rate": 0.00019986859740213917,
      "loss": 2.0133,
      "step": 18670
    },
    {
      "epoch": 2.2478941034897715,
      "grad_norm": 5.266910552978516,
      "learning_rate": 0.0001998684023698474,
      "loss": 1.6699,
      "step": 18680
    },
    {
      "epoch": 2.2490974729241877,
      "grad_norm": 3.553563117980957,
      "learning_rate": 0.00019986820719302138,
      "loss": 2.0129,
      "step": 18690
    },
    {
      "epoch": 2.250300842358604,
      "grad_norm": 5.233100414276123,
      "learning_rate": 0.0001998680118716614,
      "loss": 1.8662,
      "step": 18700
    },
    {
      "epoch": 2.2515042117930206,
      "grad_norm": 4.113808631896973,
      "learning_rate": 0.00019986781640576778,
      "loss": 1.7728,
      "step": 18710
    },
    {
      "epoch": 2.252707581227437,
      "grad_norm": 7.668570518493652,
      "learning_rate": 0.00019986762079534073,
      "loss": 2.1581,
      "step": 18720
    },
    {
      "epoch": 2.253910950661853,
      "grad_norm": 4.77482795715332,
      "learning_rate": 0.0001998674250403806,
      "loss": 1.8397,
      "step": 18730
    },
    {
      "epoch": 2.2551143200962693,
      "grad_norm": 3.207192897796631,
      "learning_rate": 0.00019986722914088766,
      "loss": 2.2075,
      "step": 18740
    },
    {
      "epoch": 2.256317689530686,
      "grad_norm": 5.926999568939209,
      "learning_rate": 0.00019986703309686217,
      "loss": 1.8827,
      "step": 18750
    },
    {
      "epoch": 2.2575210589651022,
      "grad_norm": 4.553532600402832,
      "learning_rate": 0.00019986683690830442,
      "loss": 1.8602,
      "step": 18760
    },
    {
      "epoch": 2.2587244283995185,
      "grad_norm": 7.679495811462402,
      "learning_rate": 0.0001998666405752147,
      "loss": 1.7327,
      "step": 18770
    },
    {
      "epoch": 2.259927797833935,
      "grad_norm": 5.038778781890869,
      "learning_rate": 0.00019986644409759332,
      "loss": 1.7501,
      "step": 18780
    },
    {
      "epoch": 2.2611311672683514,
      "grad_norm": 2.6412038803100586,
      "learning_rate": 0.00019986624747544052,
      "loss": 1.9424,
      "step": 18790
    },
    {
      "epoch": 2.2623345367027676,
      "grad_norm": 6.090629577636719,
      "learning_rate": 0.00019986605070875662,
      "loss": 2.0169,
      "step": 18800
    },
    {
      "epoch": 2.2635379061371843,
      "grad_norm": 4.529531478881836,
      "learning_rate": 0.0001998658537975419,
      "loss": 1.7735,
      "step": 18810
    },
    {
      "epoch": 2.2647412755716005,
      "grad_norm": 7.689573764801025,
      "learning_rate": 0.00019986565674179655,
      "loss": 1.8555,
      "step": 18820
    },
    {
      "epoch": 2.2659446450060168,
      "grad_norm": 5.21312141418457,
      "learning_rate": 0.00019986545954152102,
      "loss": 1.8347,
      "step": 18830
    },
    {
      "epoch": 2.2671480144404335,
      "grad_norm": 3.2846319675445557,
      "learning_rate": 0.0001998652621967155,
      "loss": 1.7728,
      "step": 18840
    },
    {
      "epoch": 2.2683513838748497,
      "grad_norm": 5.718740940093994,
      "learning_rate": 0.00019986506470738027,
      "loss": 1.7359,
      "step": 18850
    },
    {
      "epoch": 2.269554753309266,
      "grad_norm": 4.876806259155273,
      "learning_rate": 0.00019986486707351567,
      "loss": 1.6682,
      "step": 18860
    },
    {
      "epoch": 2.270758122743682,
      "grad_norm": 7.878754138946533,
      "learning_rate": 0.0001998646692951219,
      "loss": 1.7444,
      "step": 18870
    },
    {
      "epoch": 2.271961492178099,
      "grad_norm": 5.099309921264648,
      "learning_rate": 0.0001998644713721993,
      "loss": 1.9449,
      "step": 18880
    },
    {
      "epoch": 2.273164861612515,
      "grad_norm": 3.5428872108459473,
      "learning_rate": 0.00019986427330474823,
      "loss": 1.7586,
      "step": 18890
    },
    {
      "epoch": 2.2743682310469313,
      "grad_norm": 6.376409530639648,
      "learning_rate": 0.00019986407509276885,
      "loss": 1.848,
      "step": 18900
    },
    {
      "epoch": 2.2755716004813475,
      "grad_norm": 5.111598491668701,
      "learning_rate": 0.0001998638767362615,
      "loss": 1.952,
      "step": 18910
    },
    {
      "epoch": 2.2767749699157642,
      "grad_norm": 6.774966239929199,
      "learning_rate": 0.00019986367823522644,
      "loss": 1.6374,
      "step": 18920
    },
    {
      "epoch": 2.2779783393501805,
      "grad_norm": 5.007774829864502,
      "learning_rate": 0.00019986347958966404,
      "loss": 2.0364,
      "step": 18930
    },
    {
      "epoch": 2.2791817087845967,
      "grad_norm": 3.2465569972991943,
      "learning_rate": 0.00019986328079957448,
      "loss": 1.8738,
      "step": 18940
    },
    {
      "epoch": 2.2803850782190134,
      "grad_norm": 5.423501491546631,
      "learning_rate": 0.0001998630818649581,
      "loss": 1.6515,
      "step": 18950
    },
    {
      "epoch": 2.2815884476534296,
      "grad_norm": 3.999821186065674,
      "learning_rate": 0.0001998628827858152,
      "loss": 1.654,
      "step": 18960
    },
    {
      "epoch": 2.282791817087846,
      "grad_norm": 5.91510009765625,
      "learning_rate": 0.00019986268356214605,
      "loss": 1.8803,
      "step": 18970
    },
    {
      "epoch": 2.2839951865222625,
      "grad_norm": 5.5415215492248535,
      "learning_rate": 0.00019986248419395094,
      "loss": 1.8058,
      "step": 18980
    },
    {
      "epoch": 2.2851985559566788,
      "grad_norm": 3.26278018951416,
      "learning_rate": 0.00019986228468123017,
      "loss": 1.8605,
      "step": 18990
    },
    {
      "epoch": 2.286401925391095,
      "grad_norm": 5.6961894035339355,
      "learning_rate": 0.00019986208502398403,
      "loss": 1.8349,
      "step": 19000
    },
    {
      "epoch": 2.2876052948255117,
      "grad_norm": 5.6630730628967285,
      "learning_rate": 0.00019986188522221277,
      "loss": 1.8018,
      "step": 19010
    },
    {
      "epoch": 2.288808664259928,
      "grad_norm": 7.9914398193359375,
      "learning_rate": 0.0001998616852759167,
      "loss": 1.9708,
      "step": 19020
    },
    {
      "epoch": 2.290012033694344,
      "grad_norm": 4.706212997436523,
      "learning_rate": 0.00019986148518509614,
      "loss": 1.8582,
      "step": 19030
    },
    {
      "epoch": 2.2912154031287604,
      "grad_norm": 4.37938117980957,
      "learning_rate": 0.00019986128494975134,
      "loss": 2.2603,
      "step": 19040
    },
    {
      "epoch": 2.292418772563177,
      "grad_norm": 6.98267936706543,
      "learning_rate": 0.0001998610845698826,
      "loss": 1.9635,
      "step": 19050
    },
    {
      "epoch": 2.2936221419975933,
      "grad_norm": 3.565274715423584,
      "learning_rate": 0.00019986088404549027,
      "loss": 1.752,
      "step": 19060
    },
    {
      "epoch": 2.2948255114320095,
      "grad_norm": 6.4155192375183105,
      "learning_rate": 0.00019986068337657454,
      "loss": 1.8159,
      "step": 19070
    },
    {
      "epoch": 2.2960288808664258,
      "grad_norm": 4.892825603485107,
      "learning_rate": 0.00019986048256313575,
      "loss": 1.7489,
      "step": 19080
    },
    {
      "epoch": 2.2972322503008424,
      "grad_norm": 3.5658504962921143,
      "learning_rate": 0.00019986028160517418,
      "loss": 1.8876,
      "step": 19090
    },
    {
      "epoch": 2.2984356197352587,
      "grad_norm": 5.374711990356445,
      "learning_rate": 0.00019986008050269013,
      "loss": 1.8162,
      "step": 19100
    },
    {
      "epoch": 2.299638989169675,
      "grad_norm": 4.758020401000977,
      "learning_rate": 0.0001998598792556839,
      "loss": 1.5871,
      "step": 19110
    },
    {
      "epoch": 2.3008423586040916,
      "grad_norm": 6.21620512008667,
      "learning_rate": 0.00019985967786415577,
      "loss": 1.7226,
      "step": 19120
    },
    {
      "epoch": 2.302045728038508,
      "grad_norm": 5.469278812408447,
      "learning_rate": 0.000199859476328106,
      "loss": 2.0724,
      "step": 19130
    },
    {
      "epoch": 2.303249097472924,
      "grad_norm": 4.54677152633667,
      "learning_rate": 0.00019985927464753495,
      "loss": 1.9097,
      "step": 19140
    },
    {
      "epoch": 2.3044524669073407,
      "grad_norm": 5.128119468688965,
      "learning_rate": 0.00019985907282244289,
      "loss": 1.6518,
      "step": 19150
    },
    {
      "epoch": 2.305655836341757,
      "grad_norm": 4.618437767028809,
      "learning_rate": 0.00019985887085283008,
      "loss": 2.0012,
      "step": 19160
    },
    {
      "epoch": 2.306859205776173,
      "grad_norm": 6.141031742095947,
      "learning_rate": 0.00019985866873869682,
      "loss": 2.0247,
      "step": 19170
    },
    {
      "epoch": 2.30806257521059,
      "grad_norm": 4.4620537757873535,
      "learning_rate": 0.0001998584664800434,
      "loss": 2.0114,
      "step": 19180
    },
    {
      "epoch": 2.309265944645006,
      "grad_norm": 4.38834810256958,
      "learning_rate": 0.00019985826407687015,
      "loss": 2.009,
      "step": 19190
    },
    {
      "epoch": 2.3104693140794224,
      "grad_norm": 5.883998870849609,
      "learning_rate": 0.0001998580615291773,
      "loss": 1.6037,
      "step": 19200
    },
    {
      "epoch": 2.3116726835138386,
      "grad_norm": 4.275137424468994,
      "learning_rate": 0.00019985785883696523,
      "loss": 2.0258,
      "step": 19210
    },
    {
      "epoch": 2.3128760529482553,
      "grad_norm": 6.975653648376465,
      "learning_rate": 0.00019985765600023415,
      "loss": 1.6874,
      "step": 19220
    },
    {
      "epoch": 2.3140794223826715,
      "grad_norm": 4.13192081451416,
      "learning_rate": 0.00019985745301898442,
      "loss": 1.786,
      "step": 19230
    },
    {
      "epoch": 2.3152827918170877,
      "grad_norm": 4.088593006134033,
      "learning_rate": 0.00019985724989321626,
      "loss": 1.9076,
      "step": 19240
    },
    {
      "epoch": 2.316486161251504,
      "grad_norm": 6.3527092933654785,
      "learning_rate": 0.00019985704662293005,
      "loss": 1.8687,
      "step": 19250
    },
    {
      "epoch": 2.3176895306859207,
      "grad_norm": 5.279321670532227,
      "learning_rate": 0.000199856843208126,
      "loss": 1.9265,
      "step": 19260
    },
    {
      "epoch": 2.318892900120337,
      "grad_norm": 7.881455421447754,
      "learning_rate": 0.00019985663964880446,
      "loss": 1.876,
      "step": 19270
    },
    {
      "epoch": 2.320096269554753,
      "grad_norm": 4.470962047576904,
      "learning_rate": 0.00019985643594496572,
      "loss": 1.5564,
      "step": 19280
    },
    {
      "epoch": 2.32129963898917,
      "grad_norm": 3.5736684799194336,
      "learning_rate": 0.00019985623209661007,
      "loss": 1.6959,
      "step": 19290
    },
    {
      "epoch": 2.322503008423586,
      "grad_norm": 5.825920104980469,
      "learning_rate": 0.00019985602810373775,
      "loss": 1.9138,
      "step": 19300
    },
    {
      "epoch": 2.3237063778580023,
      "grad_norm": 4.798040390014648,
      "learning_rate": 0.00019985582396634915,
      "loss": 1.9029,
      "step": 19310
    },
    {
      "epoch": 2.324909747292419,
      "grad_norm": 6.569250583648682,
      "learning_rate": 0.0001998556196844445,
      "loss": 1.9013,
      "step": 19320
    },
    {
      "epoch": 2.326113116726835,
      "grad_norm": 5.732389450073242,
      "learning_rate": 0.00019985541525802412,
      "loss": 1.6371,
      "step": 19330
    },
    {
      "epoch": 2.3273164861612514,
      "grad_norm": 3.9813873767852783,
      "learning_rate": 0.0001998552106870883,
      "loss": 1.9835,
      "step": 19340
    },
    {
      "epoch": 2.328519855595668,
      "grad_norm": 6.8331804275512695,
      "learning_rate": 0.00019985500597163733,
      "loss": 2.1205,
      "step": 19350
    },
    {
      "epoch": 2.3297232250300843,
      "grad_norm": 4.132762908935547,
      "learning_rate": 0.0001998548011116715,
      "loss": 1.8935,
      "step": 19360
    },
    {
      "epoch": 2.3309265944645006,
      "grad_norm": 7.562819957733154,
      "learning_rate": 0.00019985459610719116,
      "loss": 1.8519,
      "step": 19370
    },
    {
      "epoch": 2.332129963898917,
      "grad_norm": 5.643316268920898,
      "learning_rate": 0.00019985439095819655,
      "loss": 1.9336,
      "step": 19380
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 3.6668975353240967,
      "learning_rate": 0.00019985418566468796,
      "loss": 1.8849,
      "step": 19390
    },
    {
      "epoch": 2.3345367027677497,
      "grad_norm": 6.986433029174805,
      "learning_rate": 0.00019985398022666573,
      "loss": 1.836,
      "step": 19400
    },
    {
      "epoch": 2.335740072202166,
      "grad_norm": 4.0480241775512695,
      "learning_rate": 0.00019985377464413014,
      "loss": 1.9559,
      "step": 19410
    },
    {
      "epoch": 2.336943441636582,
      "grad_norm": 6.787583351135254,
      "learning_rate": 0.00019985356891708147,
      "loss": 2.1796,
      "step": 19420
    },
    {
      "epoch": 2.338146811070999,
      "grad_norm": 5.128816604614258,
      "learning_rate": 0.00019985336304552008,
      "loss": 1.6992,
      "step": 19430
    },
    {
      "epoch": 2.339350180505415,
      "grad_norm": 3.483121871948242,
      "learning_rate": 0.00019985315702944617,
      "loss": 1.9576,
      "step": 19440
    },
    {
      "epoch": 2.3405535499398313,
      "grad_norm": 4.72206449508667,
      "learning_rate": 0.0001998529508688601,
      "loss": 2.0157,
      "step": 19450
    },
    {
      "epoch": 2.341756919374248,
      "grad_norm": 3.925147533416748,
      "learning_rate": 0.00019985274456376216,
      "loss": 1.7235,
      "step": 19460
    },
    {
      "epoch": 2.3429602888086642,
      "grad_norm": 7.588326930999756,
      "learning_rate": 0.00019985253811415264,
      "loss": 1.8368,
      "step": 19470
    },
    {
      "epoch": 2.3441636582430805,
      "grad_norm": 5.377657890319824,
      "learning_rate": 0.00019985233152003184,
      "loss": 1.9796,
      "step": 19480
    },
    {
      "epoch": 2.345367027677497,
      "grad_norm": 4.09199333190918,
      "learning_rate": 0.00019985212478140007,
      "loss": 1.8423,
      "step": 19490
    },
    {
      "epoch": 2.3465703971119134,
      "grad_norm": 5.174166679382324,
      "learning_rate": 0.00019985191789825764,
      "loss": 1.7323,
      "step": 19500
    },
    {
      "epoch": 2.3477737665463296,
      "grad_norm": 4.120990753173828,
      "learning_rate": 0.0001998517108706048,
      "loss": 1.8738,
      "step": 19510
    },
    {
      "epoch": 2.3489771359807463,
      "grad_norm": 7.727570533752441,
      "learning_rate": 0.0001998515036984419,
      "loss": 1.6302,
      "step": 19520
    },
    {
      "epoch": 2.3501805054151625,
      "grad_norm": 4.989604949951172,
      "learning_rate": 0.00019985129638176922,
      "loss": 1.8801,
      "step": 19530
    },
    {
      "epoch": 2.351383874849579,
      "grad_norm": 3.2973055839538574,
      "learning_rate": 0.0001998510889205871,
      "loss": 2.0017,
      "step": 19540
    },
    {
      "epoch": 2.352587244283995,
      "grad_norm": 5.632620334625244,
      "learning_rate": 0.00019985088131489574,
      "loss": 1.6783,
      "step": 19550
    },
    {
      "epoch": 2.3537906137184117,
      "grad_norm": 4.893600940704346,
      "learning_rate": 0.00019985067356469553,
      "loss": 1.7866,
      "step": 19560
    },
    {
      "epoch": 2.354993983152828,
      "grad_norm": 8.183561325073242,
      "learning_rate": 0.00019985046566998676,
      "loss": 1.9867,
      "step": 19570
    },
    {
      "epoch": 2.356197352587244,
      "grad_norm": 5.605899333953857,
      "learning_rate": 0.00019985025763076967,
      "loss": 1.9443,
      "step": 19580
    },
    {
      "epoch": 2.357400722021661,
      "grad_norm": 3.518667221069336,
      "learning_rate": 0.00019985004944704464,
      "loss": 1.8668,
      "step": 19590
    },
    {
      "epoch": 2.358604091456077,
      "grad_norm": 5.827317714691162,
      "learning_rate": 0.00019984984111881193,
      "loss": 1.9045,
      "step": 19600
    },
    {
      "epoch": 2.3598074608904933,
      "grad_norm": 4.613023281097412,
      "learning_rate": 0.00019984963264607183,
      "loss": 1.9533,
      "step": 19610
    },
    {
      "epoch": 2.3610108303249095,
      "grad_norm": 6.259252071380615,
      "learning_rate": 0.00019984942402882468,
      "loss": 1.8545,
      "step": 19620
    },
    {
      "epoch": 2.3622141997593262,
      "grad_norm": 6.066045761108398,
      "learning_rate": 0.00019984921526707073,
      "loss": 1.9019,
      "step": 19630
    },
    {
      "epoch": 2.3634175691937425,
      "grad_norm": 3.8455309867858887,
      "learning_rate": 0.00019984900636081036,
      "loss": 1.8717,
      "step": 19640
    },
    {
      "epoch": 2.3646209386281587,
      "grad_norm": 6.248837471008301,
      "learning_rate": 0.0001998487973100438,
      "loss": 1.7347,
      "step": 19650
    },
    {
      "epoch": 2.3658243080625754,
      "grad_norm": 4.124472618103027,
      "learning_rate": 0.00019984858811477138,
      "loss": 1.7833,
      "step": 19660
    },
    {
      "epoch": 2.3670276774969916,
      "grad_norm": 6.054342269897461,
      "learning_rate": 0.00019984837877499341,
      "loss": 1.8304,
      "step": 19670
    },
    {
      "epoch": 2.368231046931408,
      "grad_norm": 4.665483474731445,
      "learning_rate": 0.00019984816929071017,
      "loss": 1.965,
      "step": 19680
    },
    {
      "epoch": 2.3694344163658245,
      "grad_norm": 3.366262912750244,
      "learning_rate": 0.000199847959661922,
      "loss": 1.8008,
      "step": 19690
    },
    {
      "epoch": 2.3706377858002408,
      "grad_norm": 6.070033550262451,
      "learning_rate": 0.00019984774988862915,
      "loss": 1.727,
      "step": 19700
    },
    {
      "epoch": 2.371841155234657,
      "grad_norm": 3.947265148162842,
      "learning_rate": 0.00019984753997083198,
      "loss": 1.6893,
      "step": 19710
    },
    {
      "epoch": 2.3730445246690732,
      "grad_norm": 6.37141752243042,
      "learning_rate": 0.00019984732990853074,
      "loss": 1.85,
      "step": 19720
    },
    {
      "epoch": 2.37424789410349,
      "grad_norm": 4.367625713348389,
      "learning_rate": 0.0001998471197017258,
      "loss": 1.8522,
      "step": 19730
    },
    {
      "epoch": 2.375451263537906,
      "grad_norm": 3.7760021686553955,
      "learning_rate": 0.0001998469093504174,
      "loss": 1.9469,
      "step": 19740
    },
    {
      "epoch": 2.3766546329723224,
      "grad_norm": 6.074053764343262,
      "learning_rate": 0.00019984669885460588,
      "loss": 1.8709,
      "step": 19750
    },
    {
      "epoch": 2.377858002406739,
      "grad_norm": 4.137975692749023,
      "learning_rate": 0.00019984648821429153,
      "loss": 1.8361,
      "step": 19760
    },
    {
      "epoch": 2.3790613718411553,
      "grad_norm": 8.683006286621094,
      "learning_rate": 0.00019984627742947465,
      "loss": 2.0524,
      "step": 19770
    },
    {
      "epoch": 2.3802647412755715,
      "grad_norm": 4.150774002075195,
      "learning_rate": 0.00019984606650015559,
      "loss": 1.6457,
      "step": 19780
    },
    {
      "epoch": 2.3814681107099878,
      "grad_norm": 4.112648010253906,
      "learning_rate": 0.0001998458554263346,
      "loss": 1.7112,
      "step": 19790
    },
    {
      "epoch": 2.3826714801444044,
      "grad_norm": 6.109230041503906,
      "learning_rate": 0.000199845644208012,
      "loss": 2.1384,
      "step": 19800
    },
    {
      "epoch": 2.3838748495788207,
      "grad_norm": 4.682973384857178,
      "learning_rate": 0.0001998454328451881,
      "loss": 1.7411,
      "step": 19810
    },
    {
      "epoch": 2.385078219013237,
      "grad_norm": 7.387558460235596,
      "learning_rate": 0.00019984522133786322,
      "loss": 1.8546,
      "step": 19820
    },
    {
      "epoch": 2.3862815884476536,
      "grad_norm": 4.582014083862305,
      "learning_rate": 0.00019984500968603763,
      "loss": 1.6319,
      "step": 19830
    },
    {
      "epoch": 2.38748495788207,
      "grad_norm": 4.170936107635498,
      "learning_rate": 0.00019984479788971168,
      "loss": 1.8747,
      "step": 19840
    },
    {
      "epoch": 2.388688327316486,
      "grad_norm": 6.265957355499268,
      "learning_rate": 0.00019984458594888568,
      "loss": 1.7328,
      "step": 19850
    },
    {
      "epoch": 2.3898916967509027,
      "grad_norm": 4.465024471282959,
      "learning_rate": 0.00019984437386355987,
      "loss": 1.8335,
      "step": 19860
    },
    {
      "epoch": 2.391095066185319,
      "grad_norm": 7.820645332336426,
      "learning_rate": 0.00019984416163373463,
      "loss": 1.9983,
      "step": 19870
    },
    {
      "epoch": 2.392298435619735,
      "grad_norm": 4.321065902709961,
      "learning_rate": 0.0001998439492594102,
      "loss": 1.8749,
      "step": 19880
    },
    {
      "epoch": 2.3935018050541514,
      "grad_norm": 3.951925039291382,
      "learning_rate": 0.00019984373674058697,
      "loss": 1.9719,
      "step": 19890
    },
    {
      "epoch": 2.394705174488568,
      "grad_norm": 6.700965404510498,
      "learning_rate": 0.00019984352407726518,
      "loss": 1.9855,
      "step": 19900
    },
    {
      "epoch": 2.3959085439229844,
      "grad_norm": 4.175863742828369,
      "learning_rate": 0.00019984331126944515,
      "loss": 1.7449,
      "step": 19910
    },
    {
      "epoch": 2.3971119133574006,
      "grad_norm": 8.502548217773438,
      "learning_rate": 0.00019984309831712723,
      "loss": 1.894,
      "step": 19920
    },
    {
      "epoch": 2.3983152827918173,
      "grad_norm": 5.1616387367248535,
      "learning_rate": 0.00019984288522031167,
      "loss": 1.9793,
      "step": 19930
    },
    {
      "epoch": 2.3995186522262335,
      "grad_norm": 3.631253957748413,
      "learning_rate": 0.0001998426719789988,
      "loss": 2.0157,
      "step": 19940
    },
    {
      "epoch": 2.4007220216606497,
      "grad_norm": 6.3541035652160645,
      "learning_rate": 0.00019984245859318897,
      "loss": 1.9913,
      "step": 19950
    },
    {
      "epoch": 2.401925391095066,
      "grad_norm": 4.3100104331970215,
      "learning_rate": 0.0001998422450628824,
      "loss": 1.7548,
      "step": 19960
    },
    {
      "epoch": 2.4031287605294827,
      "grad_norm": 6.630500316619873,
      "learning_rate": 0.0001998420313880795,
      "loss": 2.2605,
      "step": 19970
    },
    {
      "epoch": 2.404332129963899,
      "grad_norm": 4.805111408233643,
      "learning_rate": 0.0001998418175687805,
      "loss": 1.9942,
      "step": 19980
    },
    {
      "epoch": 2.405535499398315,
      "grad_norm": 3.27608585357666,
      "learning_rate": 0.00019984160360498574,
      "loss": 1.8665,
      "step": 19990
    },
    {
      "epoch": 2.406738868832732,
      "grad_norm": 5.916595458984375,
      "learning_rate": 0.00019984138949669556,
      "loss": 1.8347,
      "step": 20000
    },
    {
      "epoch": 2.407942238267148,
      "grad_norm": 4.1156907081604,
      "learning_rate": 0.0001998411752439102,
      "loss": 1.8613,
      "step": 20010
    },
    {
      "epoch": 2.4091456077015643,
      "grad_norm": 6.155092239379883,
      "learning_rate": 0.00019984096084663003,
      "loss": 1.9602,
      "step": 20020
    },
    {
      "epoch": 2.410348977135981,
      "grad_norm": 5.840784072875977,
      "learning_rate": 0.00019984074630485533,
      "loss": 1.9026,
      "step": 20030
    },
    {
      "epoch": 2.411552346570397,
      "grad_norm": 3.539097309112549,
      "learning_rate": 0.00019984053161858643,
      "loss": 2.1181,
      "step": 20040
    },
    {
      "epoch": 2.4127557160048134,
      "grad_norm": 4.8869709968566895,
      "learning_rate": 0.00019984031678782363,
      "loss": 1.7118,
      "step": 20050
    },
    {
      "epoch": 2.4139590854392297,
      "grad_norm": 4.145012378692627,
      "learning_rate": 0.00019984010181256726,
      "loss": 1.8493,
      "step": 20060
    },
    {
      "epoch": 2.4151624548736463,
      "grad_norm": 7.2109456062316895,
      "learning_rate": 0.00019983988669281758,
      "loss": 1.9388,
      "step": 20070
    },
    {
      "epoch": 2.4163658243080626,
      "grad_norm": 5.20852518081665,
      "learning_rate": 0.00019983967142857492,
      "loss": 1.6899,
      "step": 20080
    },
    {
      "epoch": 2.417569193742479,
      "grad_norm": 3.464789628982544,
      "learning_rate": 0.00019983945601983964,
      "loss": 1.7821,
      "step": 20090
    },
    {
      "epoch": 2.4187725631768955,
      "grad_norm": 6.397153854370117,
      "learning_rate": 0.00019983924046661203,
      "loss": 1.8121,
      "step": 20100
    },
    {
      "epoch": 2.4199759326113117,
      "grad_norm": 4.136017322540283,
      "learning_rate": 0.00019983902476889236,
      "loss": 1.642,
      "step": 20110
    },
    {
      "epoch": 2.421179302045728,
      "grad_norm": 7.516378879547119,
      "learning_rate": 0.000199838808926681,
      "loss": 2.1426,
      "step": 20120
    },
    {
      "epoch": 2.422382671480144,
      "grad_norm": 5.808988571166992,
      "learning_rate": 0.0001998385929399782,
      "loss": 1.7826,
      "step": 20130
    },
    {
      "epoch": 2.423586040914561,
      "grad_norm": 3.586364507675171,
      "learning_rate": 0.00019983837680878432,
      "loss": 1.9311,
      "step": 20140
    },
    {
      "epoch": 2.424789410348977,
      "grad_norm": 6.14862060546875,
      "learning_rate": 0.00019983816053309965,
      "loss": 1.7535,
      "step": 20150
    },
    {
      "epoch": 2.4259927797833933,
      "grad_norm": 4.461113452911377,
      "learning_rate": 0.00019983794411292452,
      "loss": 1.6474,
      "step": 20160
    },
    {
      "epoch": 2.42719614921781,
      "grad_norm": 7.186124324798584,
      "learning_rate": 0.00019983772754825923,
      "loss": 1.8949,
      "step": 20170
    },
    {
      "epoch": 2.4283995186522263,
      "grad_norm": 4.948721885681152,
      "learning_rate": 0.00019983751083910414,
      "loss": 1.7561,
      "step": 20180
    },
    {
      "epoch": 2.4296028880866425,
      "grad_norm": 3.0712690353393555,
      "learning_rate": 0.0001998372939854595,
      "loss": 1.9438,
      "step": 20190
    },
    {
      "epoch": 2.430806257521059,
      "grad_norm": 5.280097484588623,
      "learning_rate": 0.0001998370769873256,
      "loss": 1.9033,
      "step": 20200
    },
    {
      "epoch": 2.4320096269554754,
      "grad_norm": 4.612265586853027,
      "learning_rate": 0.00019983685984470287,
      "loss": 1.8435,
      "step": 20210
    },
    {
      "epoch": 2.4332129963898916,
      "grad_norm": 7.592869758605957,
      "learning_rate": 0.0001998366425575915,
      "loss": 2.0325,
      "step": 20220
    },
    {
      "epoch": 2.434416365824308,
      "grad_norm": 4.777297496795654,
      "learning_rate": 0.00019983642512599187,
      "loss": 1.8841,
      "step": 20230
    },
    {
      "epoch": 2.4356197352587245,
      "grad_norm": 3.448274850845337,
      "learning_rate": 0.00019983620754990432,
      "loss": 1.9001,
      "step": 20240
    },
    {
      "epoch": 2.436823104693141,
      "grad_norm": 4.815512657165527,
      "learning_rate": 0.0001998359898293291,
      "loss": 1.9369,
      "step": 20250
    },
    {
      "epoch": 2.438026474127557,
      "grad_norm": 4.012821674346924,
      "learning_rate": 0.00019983577196426656,
      "loss": 1.9488,
      "step": 20260
    },
    {
      "epoch": 2.4392298435619737,
      "grad_norm": 7.183079242706299,
      "learning_rate": 0.00019983555395471702,
      "loss": 1.8742,
      "step": 20270
    },
    {
      "epoch": 2.44043321299639,
      "grad_norm": 6.024168968200684,
      "learning_rate": 0.00019983533580068076,
      "loss": 1.6017,
      "step": 20280
    },
    {
      "epoch": 2.441636582430806,
      "grad_norm": 3.5001578330993652,
      "learning_rate": 0.00019983511750215814,
      "loss": 1.8542,
      "step": 20290
    },
    {
      "epoch": 2.4428399518652224,
      "grad_norm": 6.278378963470459,
      "learning_rate": 0.00019983489905914945,
      "loss": 1.6961,
      "step": 20300
    },
    {
      "epoch": 2.444043321299639,
      "grad_norm": 4.57430362701416,
      "learning_rate": 0.00019983468047165501,
      "loss": 1.8088,
      "step": 20310
    },
    {
      "epoch": 2.4452466907340553,
      "grad_norm": 7.231283187866211,
      "learning_rate": 0.00019983446173967514,
      "loss": 2.0128,
      "step": 20320
    },
    {
      "epoch": 2.4464500601684716,
      "grad_norm": 4.400961875915527,
      "learning_rate": 0.00019983424286321014,
      "loss": 1.8736,
      "step": 20330
    },
    {
      "epoch": 2.4476534296028882,
      "grad_norm": 3.3811419010162354,
      "learning_rate": 0.00019983402384226039,
      "loss": 1.7218,
      "step": 20340
    },
    {
      "epoch": 2.4488567990373045,
      "grad_norm": 5.952632427215576,
      "learning_rate": 0.00019983380467682613,
      "loss": 1.7915,
      "step": 20350
    },
    {
      "epoch": 2.4500601684717207,
      "grad_norm": 4.2375288009643555,
      "learning_rate": 0.0001998335853669077,
      "loss": 1.799,
      "step": 20360
    },
    {
      "epoch": 2.4512635379061374,
      "grad_norm": 7.192270278930664,
      "learning_rate": 0.00019983336591250544,
      "loss": 1.9928,
      "step": 20370
    },
    {
      "epoch": 2.4524669073405536,
      "grad_norm": 4.667410373687744,
      "learning_rate": 0.00019983314631361965,
      "loss": 1.7461,
      "step": 20380
    },
    {
      "epoch": 2.45367027677497,
      "grad_norm": 3.813703775405884,
      "learning_rate": 0.00019983292657025064,
      "loss": 1.8673,
      "step": 20390
    },
    {
      "epoch": 2.4548736462093865,
      "grad_norm": 5.364323616027832,
      "learning_rate": 0.0001998327066823987,
      "loss": 1.7289,
      "step": 20400
    },
    {
      "epoch": 2.4560770156438028,
      "grad_norm": 4.271958827972412,
      "learning_rate": 0.00019983248665006423,
      "loss": 1.9626,
      "step": 20410
    },
    {
      "epoch": 2.457280385078219,
      "grad_norm": 5.91215181350708,
      "learning_rate": 0.00019983226647324751,
      "loss": 1.8125,
      "step": 20420
    },
    {
      "epoch": 2.4584837545126352,
      "grad_norm": 5.356571674346924,
      "learning_rate": 0.00019983204615194885,
      "loss": 1.7691,
      "step": 20430
    },
    {
      "epoch": 2.459687123947052,
      "grad_norm": 3.500491142272949,
      "learning_rate": 0.00019983182568616856,
      "loss": 1.8946,
      "step": 20440
    },
    {
      "epoch": 2.460890493381468,
      "grad_norm": 5.471607685089111,
      "learning_rate": 0.00019983160507590697,
      "loss": 1.7181,
      "step": 20450
    },
    {
      "epoch": 2.4620938628158844,
      "grad_norm": 6.24301815032959,
      "learning_rate": 0.0001998313843211644,
      "loss": 1.7694,
      "step": 20460
    },
    {
      "epoch": 2.4632972322503006,
      "grad_norm": 6.391057968139648,
      "learning_rate": 0.00019983116342194116,
      "loss": 2.1287,
      "step": 20470
    },
    {
      "epoch": 2.4645006016847173,
      "grad_norm": 5.084754467010498,
      "learning_rate": 0.00019983094237823758,
      "loss": 1.5653,
      "step": 20480
    },
    {
      "epoch": 2.4657039711191335,
      "grad_norm": 4.073486804962158,
      "learning_rate": 0.00019983072119005398,
      "loss": 1.944,
      "step": 20490
    },
    {
      "epoch": 2.4669073405535498,
      "grad_norm": 5.489533424377441,
      "learning_rate": 0.00019983049985739074,
      "loss": 1.8211,
      "step": 20500
    },
    {
      "epoch": 2.4681107099879664,
      "grad_norm": 5.330563545227051,
      "learning_rate": 0.00019983027838024807,
      "loss": 2.0054,
      "step": 20510
    },
    {
      "epoch": 2.4693140794223827,
      "grad_norm": 6.520805358886719,
      "learning_rate": 0.00019983005675862633,
      "loss": 1.8473,
      "step": 20520
    },
    {
      "epoch": 2.470517448856799,
      "grad_norm": 5.546237468719482,
      "learning_rate": 0.00019982983499252583,
      "loss": 2.0259,
      "step": 20530
    },
    {
      "epoch": 2.4717208182912156,
      "grad_norm": 4.316344738006592,
      "learning_rate": 0.00019982961308194695,
      "loss": 1.7913,
      "step": 20540
    },
    {
      "epoch": 2.472924187725632,
      "grad_norm": 5.053488731384277,
      "learning_rate": 0.00019982939102688999,
      "loss": 1.7793,
      "step": 20550
    },
    {
      "epoch": 2.474127557160048,
      "grad_norm": 4.186037063598633,
      "learning_rate": 0.0001998291688273552,
      "loss": 1.9729,
      "step": 20560
    },
    {
      "epoch": 2.4753309265944647,
      "grad_norm": 6.9018635749816895,
      "learning_rate": 0.000199828946483343,
      "loss": 1.6633,
      "step": 20570
    },
    {
      "epoch": 2.476534296028881,
      "grad_norm": 5.570452690124512,
      "learning_rate": 0.0001998287239948536,
      "loss": 1.6677,
      "step": 20580
    },
    {
      "epoch": 2.477737665463297,
      "grad_norm": 3.0675888061523438,
      "learning_rate": 0.00019982850136188746,
      "loss": 2.1564,
      "step": 20590
    },
    {
      "epoch": 2.4789410348977134,
      "grad_norm": 5.23740816116333,
      "learning_rate": 0.0001998282785844448,
      "loss": 1.8553,
      "step": 20600
    },
    {
      "epoch": 2.48014440433213,
      "grad_norm": 4.496595859527588,
      "learning_rate": 0.000199828055662526,
      "loss": 1.6143,
      "step": 20610
    },
    {
      "epoch": 2.4813477737665464,
      "grad_norm": 6.116359710693359,
      "learning_rate": 0.00019982783259613136,
      "loss": 2.0443,
      "step": 20620
    },
    {
      "epoch": 2.4825511432009626,
      "grad_norm": 4.956220626831055,
      "learning_rate": 0.00019982760938526116,
      "loss": 1.7014,
      "step": 20630
    },
    {
      "epoch": 2.483754512635379,
      "grad_norm": 3.1296913623809814,
      "learning_rate": 0.0001998273860299158,
      "loss": 1.7959,
      "step": 20640
    },
    {
      "epoch": 2.4849578820697955,
      "grad_norm": 5.865634441375732,
      "learning_rate": 0.00019982716253009555,
      "loss": 1.874,
      "step": 20650
    },
    {
      "epoch": 2.4861612515042117,
      "grad_norm": 4.335112571716309,
      "learning_rate": 0.00019982693888580074,
      "loss": 1.9807,
      "step": 20660
    },
    {
      "epoch": 2.487364620938628,
      "grad_norm": 8.73818302154541,
      "learning_rate": 0.0001998267150970317,
      "loss": 1.8346,
      "step": 20670
    },
    {
      "epoch": 2.4885679903730447,
      "grad_norm": 4.9280171394348145,
      "learning_rate": 0.00019982649116378877,
      "loss": 1.8952,
      "step": 20680
    },
    {
      "epoch": 2.489771359807461,
      "grad_norm": 3.2900710105895996,
      "learning_rate": 0.00019982626708607226,
      "loss": 1.8782,
      "step": 20690
    },
    {
      "epoch": 2.490974729241877,
      "grad_norm": 5.722655773162842,
      "learning_rate": 0.0001998260428638825,
      "loss": 1.883,
      "step": 20700
    },
    {
      "epoch": 2.492178098676294,
      "grad_norm": 4.088940620422363,
      "learning_rate": 0.0001998258184972198,
      "loss": 1.8307,
      "step": 20710
    },
    {
      "epoch": 2.49338146811071,
      "grad_norm": 6.719072341918945,
      "learning_rate": 0.0001998255939860845,
      "loss": 1.901,
      "step": 20720
    },
    {
      "epoch": 2.4945848375451263,
      "grad_norm": 5.0046868324279785,
      "learning_rate": 0.00019982536933047692,
      "loss": 1.9503,
      "step": 20730
    },
    {
      "epoch": 2.495788206979543,
      "grad_norm": 3.331597328186035,
      "learning_rate": 0.0001998251445303974,
      "loss": 1.6925,
      "step": 20740
    },
    {
      "epoch": 2.496991576413959,
      "grad_norm": 6.52798318862915,
      "learning_rate": 0.00019982491958584625,
      "loss": 1.7648,
      "step": 20750
    },
    {
      "epoch": 2.4981949458483754,
      "grad_norm": 4.069162845611572,
      "learning_rate": 0.00019982469449682374,
      "loss": 1.918,
      "step": 20760
    },
    {
      "epoch": 2.4993983152827917,
      "grad_norm": 6.859929084777832,
      "learning_rate": 0.0001998244692633303,
      "loss": 2.1157,
      "step": 20770
    },
    {
      "epoch": 2.5006016847172083,
      "grad_norm": 5.4571075439453125,
      "learning_rate": 0.00019982424388536617,
      "loss": 1.7556,
      "step": 20780
    },
    {
      "epoch": 2.5018050541516246,
      "grad_norm": 2.8539035320281982,
      "learning_rate": 0.00019982401836293176,
      "loss": 1.8848,
      "step": 20790
    },
    {
      "epoch": 2.503008423586041,
      "grad_norm": 5.45432710647583,
      "learning_rate": 0.00019982379269602732,
      "loss": 1.8332,
      "step": 20800
    },
    {
      "epoch": 2.504211793020457,
      "grad_norm": 4.622279644012451,
      "learning_rate": 0.00019982356688465322,
      "loss": 1.9187,
      "step": 20810
    },
    {
      "epoch": 2.5054151624548737,
      "grad_norm": 6.064669609069824,
      "learning_rate": 0.00019982334092880977,
      "loss": 1.9692,
      "step": 20820
    },
    {
      "epoch": 2.50661853188929,
      "grad_norm": 5.854981422424316,
      "learning_rate": 0.00019982311482849728,
      "loss": 1.7879,
      "step": 20830
    },
    {
      "epoch": 2.507821901323706,
      "grad_norm": 3.0214333534240723,
      "learning_rate": 0.00019982288858371613,
      "loss": 1.9418,
      "step": 20840
    },
    {
      "epoch": 2.509025270758123,
      "grad_norm": 7.712893009185791,
      "learning_rate": 0.0001998226621944666,
      "loss": 1.6371,
      "step": 20850
    },
    {
      "epoch": 2.510228640192539,
      "grad_norm": 4.892581462860107,
      "learning_rate": 0.00019982243566074904,
      "loss": 1.8865,
      "step": 20860
    },
    {
      "epoch": 2.5114320096269553,
      "grad_norm": 8.19919204711914,
      "learning_rate": 0.00019982220898256378,
      "loss": 1.7838,
      "step": 20870
    },
    {
      "epoch": 2.512635379061372,
      "grad_norm": 5.1628098487854,
      "learning_rate": 0.00019982198215991114,
      "loss": 1.932,
      "step": 20880
    },
    {
      "epoch": 2.5138387484957883,
      "grad_norm": 4.552578926086426,
      "learning_rate": 0.0001998217551927914,
      "loss": 1.7088,
      "step": 20890
    },
    {
      "epoch": 2.5150421179302045,
      "grad_norm": 5.526708602905273,
      "learning_rate": 0.00019982152808120497,
      "loss": 1.8121,
      "step": 20900
    },
    {
      "epoch": 2.516245487364621,
      "grad_norm": 4.256577968597412,
      "learning_rate": 0.00019982130082515213,
      "loss": 1.7975,
      "step": 20910
    },
    {
      "epoch": 2.5174488567990374,
      "grad_norm": 7.442561149597168,
      "learning_rate": 0.00019982107342463322,
      "loss": 1.9656,
      "step": 20920
    },
    {
      "epoch": 2.5186522262334536,
      "grad_norm": 4.96824836730957,
      "learning_rate": 0.00019982084587964857,
      "loss": 1.6666,
      "step": 20930
    },
    {
      "epoch": 2.51985559566787,
      "grad_norm": 3.7357962131500244,
      "learning_rate": 0.00019982061819019852,
      "loss": 1.9951,
      "step": 20940
    },
    {
      "epoch": 2.5210589651022866,
      "grad_norm": 5.7628865242004395,
      "learning_rate": 0.00019982039035628342,
      "loss": 1.7586,
      "step": 20950
    },
    {
      "epoch": 2.522262334536703,
      "grad_norm": 4.2057695388793945,
      "learning_rate": 0.00019982016237790354,
      "loss": 2.0508,
      "step": 20960
    },
    {
      "epoch": 2.523465703971119,
      "grad_norm": 6.62906551361084,
      "learning_rate": 0.00019981993425505925,
      "loss": 1.8817,
      "step": 20970
    },
    {
      "epoch": 2.5246690734055353,
      "grad_norm": 4.822028636932373,
      "learning_rate": 0.00019981970598775085,
      "loss": 1.9177,
      "step": 20980
    },
    {
      "epoch": 2.525872442839952,
      "grad_norm": 3.578030586242676,
      "learning_rate": 0.0001998194775759787,
      "loss": 1.9915,
      "step": 20990
    },
    {
      "epoch": 2.527075812274368,
      "grad_norm": 6.2930498123168945,
      "learning_rate": 0.00019981924901974315,
      "loss": 1.9624,
      "step": 21000
    },
    {
      "epoch": 2.5282791817087844,
      "grad_norm": 3.8709285259246826,
      "learning_rate": 0.00019981902031904447,
      "loss": 1.8362,
      "step": 21010
    },
    {
      "epoch": 2.529482551143201,
      "grad_norm": 7.663669109344482,
      "learning_rate": 0.00019981879147388303,
      "loss": 1.9585,
      "step": 21020
    },
    {
      "epoch": 2.5306859205776173,
      "grad_norm": 4.16254997253418,
      "learning_rate": 0.00019981856248425914,
      "loss": 1.5374,
      "step": 21030
    },
    {
      "epoch": 2.5318892900120336,
      "grad_norm": 4.396316051483154,
      "learning_rate": 0.00019981833335017317,
      "loss": 2.0933,
      "step": 21040
    },
    {
      "epoch": 2.5330926594464502,
      "grad_norm": 5.005481719970703,
      "learning_rate": 0.0001998181040716254,
      "loss": 1.894,
      "step": 21050
    },
    {
      "epoch": 2.5342960288808665,
      "grad_norm": 4.025967121124268,
      "learning_rate": 0.00019981787464861623,
      "loss": 1.817,
      "step": 21060
    },
    {
      "epoch": 2.5354993983152827,
      "grad_norm": 6.891594886779785,
      "learning_rate": 0.00019981764508114595,
      "loss": 1.7859,
      "step": 21070
    },
    {
      "epoch": 2.5367027677496994,
      "grad_norm": 6.6164374351501465,
      "learning_rate": 0.00019981741536921485,
      "loss": 1.8623,
      "step": 21080
    },
    {
      "epoch": 2.5379061371841156,
      "grad_norm": 3.5325465202331543,
      "learning_rate": 0.00019981718551282334,
      "loss": 1.8581,
      "step": 21090
    },
    {
      "epoch": 2.539109506618532,
      "grad_norm": 5.685319423675537,
      "learning_rate": 0.00019981695551197167,
      "loss": 1.7876,
      "step": 21100
    },
    {
      "epoch": 2.5403128760529485,
      "grad_norm": 3.816117286682129,
      "learning_rate": 0.0001998167253666603,
      "loss": 1.9363,
      "step": 21110
    },
    {
      "epoch": 2.5415162454873648,
      "grad_norm": 6.195153713226318,
      "learning_rate": 0.0001998164950768894,
      "loss": 1.9416,
      "step": 21120
    },
    {
      "epoch": 2.542719614921781,
      "grad_norm": 4.386119365692139,
      "learning_rate": 0.00019981626464265944,
      "loss": 1.7258,
      "step": 21130
    },
    {
      "epoch": 2.5439229843561972,
      "grad_norm": 3.807783603668213,
      "learning_rate": 0.0001998160340639707,
      "loss": 2.0765,
      "step": 21140
    },
    {
      "epoch": 2.5451263537906135,
      "grad_norm": 6.045656204223633,
      "learning_rate": 0.0001998158033408235,
      "loss": 2.1099,
      "step": 21150
    },
    {
      "epoch": 2.54632972322503,
      "grad_norm": 4.081486701965332,
      "learning_rate": 0.00019981557247321821,
      "loss": 1.7787,
      "step": 21160
    },
    {
      "epoch": 2.5475330926594464,
      "grad_norm": 7.350472927093506,
      "learning_rate": 0.00019981534146115514,
      "loss": 2.0342,
      "step": 21170
    },
    {
      "epoch": 2.5487364620938626,
      "grad_norm": 4.729325294494629,
      "learning_rate": 0.0001998151103046346,
      "loss": 1.8585,
      "step": 21180
    },
    {
      "epoch": 2.5499398315282793,
      "grad_norm": 4.003896713256836,
      "learning_rate": 0.00019981487900365695,
      "loss": 2.1392,
      "step": 21190
    },
    {
      "epoch": 2.5511432009626955,
      "grad_norm": 5.162550449371338,
      "learning_rate": 0.00019981464755822255,
      "loss": 1.7903,
      "step": 21200
    },
    {
      "epoch": 2.5523465703971118,
      "grad_norm": 5.016550064086914,
      "learning_rate": 0.00019981441596833173,
      "loss": 1.7447,
      "step": 21210
    },
    {
      "epoch": 2.5535499398315284,
      "grad_norm": 6.916841983795166,
      "learning_rate": 0.00019981418423398477,
      "loss": 1.8037,
      "step": 21220
    },
    {
      "epoch": 2.5547533092659447,
      "grad_norm": 4.650218486785889,
      "learning_rate": 0.00019981395235518207,
      "loss": 1.8007,
      "step": 21230
    },
    {
      "epoch": 2.555956678700361,
      "grad_norm": 3.214818000793457,
      "learning_rate": 0.0001998137203319239,
      "loss": 1.8741,
      "step": 21240
    },
    {
      "epoch": 2.5571600481347776,
      "grad_norm": 5.459062576293945,
      "learning_rate": 0.00019981348816421065,
      "loss": 1.7321,
      "step": 21250
    },
    {
      "epoch": 2.558363417569194,
      "grad_norm": 3.7758872509002686,
      "learning_rate": 0.00019981325585204264,
      "loss": 1.7992,
      "step": 21260
    },
    {
      "epoch": 2.55956678700361,
      "grad_norm": 7.222395896911621,
      "learning_rate": 0.0001998130233954202,
      "loss": 1.6814,
      "step": 21270
    },
    {
      "epoch": 2.5607701564380267,
      "grad_norm": 4.6296515464782715,
      "learning_rate": 0.00019981279079434372,
      "loss": 1.8424,
      "step": 21280
    },
    {
      "epoch": 2.561973525872443,
      "grad_norm": 3.446894407272339,
      "learning_rate": 0.00019981255804881343,
      "loss": 1.9884,
      "step": 21290
    },
    {
      "epoch": 2.563176895306859,
      "grad_norm": 5.880203723907471,
      "learning_rate": 0.00019981232515882974,
      "loss": 1.793,
      "step": 21300
    },
    {
      "epoch": 2.5643802647412755,
      "grad_norm": 4.337691783905029,
      "learning_rate": 0.00019981209212439297,
      "loss": 1.7987,
      "step": 21310
    },
    {
      "epoch": 2.5655836341756917,
      "grad_norm": 5.445236682891846,
      "learning_rate": 0.0001998118589455035,
      "loss": 1.8639,
      "step": 21320
    },
    {
      "epoch": 2.5667870036101084,
      "grad_norm": 5.637457847595215,
      "learning_rate": 0.0001998116256221616,
      "loss": 1.7117,
      "step": 21330
    },
    {
      "epoch": 2.5679903730445246,
      "grad_norm": 4.634398460388184,
      "learning_rate": 0.00019981139215436762,
      "loss": 1.9762,
      "step": 21340
    },
    {
      "epoch": 2.569193742478941,
      "grad_norm": 6.838907241821289,
      "learning_rate": 0.00019981115854212194,
      "loss": 2.0023,
      "step": 21350
    },
    {
      "epoch": 2.5703971119133575,
      "grad_norm": 4.272093296051025,
      "learning_rate": 0.00019981092478542483,
      "loss": 1.8791,
      "step": 21360
    },
    {
      "epoch": 2.5716004813477737,
      "grad_norm": 5.1662750244140625,
      "learning_rate": 0.0001998106908842767,
      "loss": 1.815,
      "step": 21370
    },
    {
      "epoch": 2.57280385078219,
      "grad_norm": 4.997259616851807,
      "learning_rate": 0.00019981045683867786,
      "loss": 1.7007,
      "step": 21380
    },
    {
      "epoch": 2.5740072202166067,
      "grad_norm": 3.608267068862915,
      "learning_rate": 0.00019981022264862862,
      "loss": 1.9887,
      "step": 21390
    },
    {
      "epoch": 2.575210589651023,
      "grad_norm": 5.867269515991211,
      "learning_rate": 0.00019980998831412939,
      "loss": 1.825,
      "step": 21400
    },
    {
      "epoch": 2.576413959085439,
      "grad_norm": 4.047500133514404,
      "learning_rate": 0.0001998097538351804,
      "loss": 1.7734,
      "step": 21410
    },
    {
      "epoch": 2.577617328519856,
      "grad_norm": 6.287036418914795,
      "learning_rate": 0.0001998095192117821,
      "loss": 1.7597,
      "step": 21420
    },
    {
      "epoch": 2.578820697954272,
      "grad_norm": 5.959181308746338,
      "learning_rate": 0.0001998092844439348,
      "loss": 2.1468,
      "step": 21430
    },
    {
      "epoch": 2.5800240673886883,
      "grad_norm": 3.160041332244873,
      "learning_rate": 0.00019980904953163878,
      "loss": 1.7567,
      "step": 21440
    },
    {
      "epoch": 2.581227436823105,
      "grad_norm": 5.435869216918945,
      "learning_rate": 0.00019980881447489445,
      "loss": 1.9945,
      "step": 21450
    },
    {
      "epoch": 2.582430806257521,
      "grad_norm": 3.9015581607818604,
      "learning_rate": 0.0001998085792737021,
      "loss": 1.7607,
      "step": 21460
    },
    {
      "epoch": 2.5836341756919374,
      "grad_norm": 6.834968090057373,
      "learning_rate": 0.00019980834392806213,
      "loss": 1.8431,
      "step": 21470
    },
    {
      "epoch": 2.5848375451263537,
      "grad_norm": 5.225518226623535,
      "learning_rate": 0.00019980810843797485,
      "loss": 1.7017,
      "step": 21480
    },
    {
      "epoch": 2.58604091456077,
      "grad_norm": 4.1456379890441895,
      "learning_rate": 0.00019980787280344056,
      "loss": 1.8974,
      "step": 21490
    },
    {
      "epoch": 2.5872442839951866,
      "grad_norm": 5.812695026397705,
      "learning_rate": 0.00019980763702445964,
      "loss": 1.9485,
      "step": 21500
    },
    {
      "epoch": 2.588447653429603,
      "grad_norm": 4.893465995788574,
      "learning_rate": 0.0001998074011010324,
      "loss": 1.8701,
      "step": 21510
    },
    {
      "epoch": 2.589651022864019,
      "grad_norm": 6.693197250366211,
      "learning_rate": 0.0001998071650331593,
      "loss": 1.8423,
      "step": 21520
    },
    {
      "epoch": 2.5908543922984357,
      "grad_norm": 4.072821140289307,
      "learning_rate": 0.00019980692882084052,
      "loss": 1.6027,
      "step": 21530
    },
    {
      "epoch": 2.592057761732852,
      "grad_norm": 3.2784483432769775,
      "learning_rate": 0.0001998066924640765,
      "loss": 1.7356,
      "step": 21540
    },
    {
      "epoch": 2.593261131167268,
      "grad_norm": 5.643657207489014,
      "learning_rate": 0.00019980645596286752,
      "loss": 1.7846,
      "step": 21550
    },
    {
      "epoch": 2.594464500601685,
      "grad_norm": 4.14614725112915,
      "learning_rate": 0.00019980621931721398,
      "loss": 2.0322,
      "step": 21560
    },
    {
      "epoch": 2.595667870036101,
      "grad_norm": 6.763193130493164,
      "learning_rate": 0.0001998059825271162,
      "loss": 1.7348,
      "step": 21570
    },
    {
      "epoch": 2.5968712394705173,
      "grad_norm": 4.553989410400391,
      "learning_rate": 0.0001998057455925745,
      "loss": 1.7151,
      "step": 21580
    },
    {
      "epoch": 2.598074608904934,
      "grad_norm": 3.643972158432007,
      "learning_rate": 0.00019980550851358928,
      "loss": 1.8768,
      "step": 21590
    },
    {
      "epoch": 2.5992779783393503,
      "grad_norm": 6.013065338134766,
      "learning_rate": 0.00019980527129016083,
      "loss": 1.6403,
      "step": 21600
    },
    {
      "epoch": 2.6004813477737665,
      "grad_norm": 4.082161903381348,
      "learning_rate": 0.0001998050339222895,
      "loss": 1.7879,
      "step": 21610
    },
    {
      "epoch": 2.601684717208183,
      "grad_norm": 6.888028621673584,
      "learning_rate": 0.00019980479640997566,
      "loss": 1.8847,
      "step": 21620
    },
    {
      "epoch": 2.6028880866425994,
      "grad_norm": 4.781867504119873,
      "learning_rate": 0.00019980455875321965,
      "loss": 1.9966,
      "step": 21630
    },
    {
      "epoch": 2.6040914560770156,
      "grad_norm": 3.677938938140869,
      "learning_rate": 0.00019980432095202178,
      "loss": 1.8977,
      "step": 21640
    },
    {
      "epoch": 2.605294825511432,
      "grad_norm": 6.58203649520874,
      "learning_rate": 0.00019980408300638243,
      "loss": 1.8308,
      "step": 21650
    },
    {
      "epoch": 2.606498194945848,
      "grad_norm": 4.445764541625977,
      "learning_rate": 0.00019980384491630193,
      "loss": 2.0844,
      "step": 21660
    },
    {
      "epoch": 2.607701564380265,
      "grad_norm": 7.188562393188477,
      "learning_rate": 0.00019980360668178063,
      "loss": 1.7353,
      "step": 21670
    },
    {
      "epoch": 2.608904933814681,
      "grad_norm": 6.179540157318115,
      "learning_rate": 0.00019980336830281885,
      "loss": 1.9086,
      "step": 21680
    },
    {
      "epoch": 2.6101083032490973,
      "grad_norm": 3.520310401916504,
      "learning_rate": 0.00019980312977941697,
      "loss": 1.8717,
      "step": 21690
    },
    {
      "epoch": 2.611311672683514,
      "grad_norm": 6.002782344818115,
      "learning_rate": 0.00019980289111157533,
      "loss": 1.8961,
      "step": 21700
    },
    {
      "epoch": 2.61251504211793,
      "grad_norm": 4.381802082061768,
      "learning_rate": 0.00019980265229929426,
      "loss": 1.9205,
      "step": 21710
    },
    {
      "epoch": 2.6137184115523464,
      "grad_norm": 6.384566307067871,
      "learning_rate": 0.00019980241334257407,
      "loss": 1.7532,
      "step": 21720
    },
    {
      "epoch": 2.614921780986763,
      "grad_norm": 5.342832565307617,
      "learning_rate": 0.0001998021742414152,
      "loss": 1.9215,
      "step": 21730
    },
    {
      "epoch": 2.6161251504211793,
      "grad_norm": 3.5546505451202393,
      "learning_rate": 0.00019980193499581793,
      "loss": 1.9366,
      "step": 21740
    },
    {
      "epoch": 2.6173285198555956,
      "grad_norm": 6.061792850494385,
      "learning_rate": 0.0001998016956057826,
      "loss": 1.836,
      "step": 21750
    },
    {
      "epoch": 2.6185318892900122,
      "grad_norm": 3.879345655441284,
      "learning_rate": 0.0001998014560713096,
      "loss": 1.7855,
      "step": 21760
    },
    {
      "epoch": 2.6197352587244285,
      "grad_norm": 6.151535987854004,
      "learning_rate": 0.00019980121639239927,
      "loss": 1.8757,
      "step": 21770
    },
    {
      "epoch": 2.6209386281588447,
      "grad_norm": 5.681988716125488,
      "learning_rate": 0.0001998009765690519,
      "loss": 1.8547,
      "step": 21780
    },
    {
      "epoch": 2.6221419975932614,
      "grad_norm": 3.828183650970459,
      "learning_rate": 0.00019980073660126791,
      "loss": 1.8725,
      "step": 21790
    },
    {
      "epoch": 2.6233453670276776,
      "grad_norm": 5.166325569152832,
      "learning_rate": 0.0001998004964890476,
      "loss": 1.7821,
      "step": 21800
    },
    {
      "epoch": 2.624548736462094,
      "grad_norm": 5.182612419128418,
      "learning_rate": 0.00019980025623239132,
      "loss": 2.1479,
      "step": 21810
    },
    {
      "epoch": 2.62575210589651,
      "grad_norm": 6.3542094230651855,
      "learning_rate": 0.00019980001583129944,
      "loss": 1.5588,
      "step": 21820
    },
    {
      "epoch": 2.6269554753309263,
      "grad_norm": 5.207058429718018,
      "learning_rate": 0.00019979977528577228,
      "loss": 1.8539,
      "step": 21830
    },
    {
      "epoch": 2.628158844765343,
      "grad_norm": 3.4649484157562256,
      "learning_rate": 0.00019979953459581024,
      "loss": 1.7373,
      "step": 21840
    },
    {
      "epoch": 2.6293622141997592,
      "grad_norm": 5.4307780265808105,
      "learning_rate": 0.0001997992937614136,
      "loss": 1.8633,
      "step": 21850
    },
    {
      "epoch": 2.6305655836341755,
      "grad_norm": 3.1341311931610107,
      "learning_rate": 0.00019979905278258276,
      "loss": 1.8146,
      "step": 21860
    },
    {
      "epoch": 2.631768953068592,
      "grad_norm": 6.513323783874512,
      "learning_rate": 0.00019979881165931804,
      "loss": 1.9327,
      "step": 21870
    },
    {
      "epoch": 2.6329723225030084,
      "grad_norm": 4.858741283416748,
      "learning_rate": 0.00019979857039161981,
      "loss": 1.9581,
      "step": 21880
    },
    {
      "epoch": 2.6341756919374246,
      "grad_norm": 3.1470422744750977,
      "learning_rate": 0.00019979832897948843,
      "loss": 1.8119,
      "step": 21890
    },
    {
      "epoch": 2.6353790613718413,
      "grad_norm": 5.330724239349365,
      "learning_rate": 0.00019979808742292418,
      "loss": 1.7647,
      "step": 21900
    },
    {
      "epoch": 2.6365824308062575,
      "grad_norm": 3.593313217163086,
      "learning_rate": 0.00019979784572192748,
      "loss": 1.93,
      "step": 21910
    },
    {
      "epoch": 2.6377858002406738,
      "grad_norm": 7.264196395874023,
      "learning_rate": 0.00019979760387649868,
      "loss": 1.9577,
      "step": 21920
    },
    {
      "epoch": 2.6389891696750905,
      "grad_norm": 4.571249961853027,
      "learning_rate": 0.0001997973618866381,
      "loss": 1.8587,
      "step": 21930
    },
    {
      "epoch": 2.6401925391095067,
      "grad_norm": 4.024495601654053,
      "learning_rate": 0.00019979711975234607,
      "loss": 1.9326,
      "step": 21940
    },
    {
      "epoch": 2.641395908543923,
      "grad_norm": 6.552509307861328,
      "learning_rate": 0.000199796877473623,
      "loss": 1.8718,
      "step": 21950
    },
    {
      "epoch": 2.6425992779783396,
      "grad_norm": 4.4985527992248535,
      "learning_rate": 0.00019979663505046918,
      "loss": 1.8838,
      "step": 21960
    },
    {
      "epoch": 2.643802647412756,
      "grad_norm": 7.27703332901001,
      "learning_rate": 0.000199796392482885,
      "loss": 1.8949,
      "step": 21970
    },
    {
      "epoch": 2.645006016847172,
      "grad_norm": 3.9912779331207275,
      "learning_rate": 0.00019979614977087083,
      "loss": 1.93,
      "step": 21980
    },
    {
      "epoch": 2.6462093862815883,
      "grad_norm": 4.768033504486084,
      "learning_rate": 0.00019979590691442695,
      "loss": 1.7158,
      "step": 21990
    },
    {
      "epoch": 2.6474127557160045,
      "grad_norm": 7.420052528381348,
      "learning_rate": 0.00019979566391355378,
      "loss": 1.7915,
      "step": 22000
    },
    {
      "epoch": 2.648616125150421,
      "grad_norm": 4.023570537567139,
      "learning_rate": 0.00019979542076825163,
      "loss": 1.9327,
      "step": 22010
    },
    {
      "epoch": 2.6498194945848375,
      "grad_norm": 7.8548197746276855,
      "learning_rate": 0.00019979517747852087,
      "loss": 1.82,
      "step": 22020
    },
    {
      "epoch": 2.6510228640192537,
      "grad_norm": 5.435407638549805,
      "learning_rate": 0.00019979493404436187,
      "loss": 1.9069,
      "step": 22030
    },
    {
      "epoch": 2.6522262334536704,
      "grad_norm": 3.096574544906616,
      "learning_rate": 0.00019979469046577496,
      "loss": 1.967,
      "step": 22040
    },
    {
      "epoch": 2.6534296028880866,
      "grad_norm": 5.799994468688965,
      "learning_rate": 0.00019979444674276048,
      "loss": 1.8177,
      "step": 22050
    },
    {
      "epoch": 2.654632972322503,
      "grad_norm": 4.348621368408203,
      "learning_rate": 0.00019979420287531878,
      "loss": 1.7023,
      "step": 22060
    },
    {
      "epoch": 2.6558363417569195,
      "grad_norm": 6.873908996582031,
      "learning_rate": 0.00019979395886345026,
      "loss": 1.9659,
      "step": 22070
    },
    {
      "epoch": 2.6570397111913358,
      "grad_norm": 4.480973243713379,
      "learning_rate": 0.00019979371470715522,
      "loss": 1.7301,
      "step": 22080
    },
    {
      "epoch": 2.658243080625752,
      "grad_norm": 3.9883265495300293,
      "learning_rate": 0.00019979347040643405,
      "loss": 1.9525,
      "step": 22090
    },
    {
      "epoch": 2.6594464500601687,
      "grad_norm": 5.971280574798584,
      "learning_rate": 0.0001997932259612871,
      "loss": 1.7962,
      "step": 22100
    },
    {
      "epoch": 2.660649819494585,
      "grad_norm": 4.024284839630127,
      "learning_rate": 0.0001997929813717147,
      "loss": 1.6596,
      "step": 22110
    },
    {
      "epoch": 2.661853188929001,
      "grad_norm": 6.47664213180542,
      "learning_rate": 0.0001997927366377172,
      "loss": 1.7912,
      "step": 22120
    },
    {
      "epoch": 2.663056558363418,
      "grad_norm": 4.878775119781494,
      "learning_rate": 0.000199792491759295,
      "loss": 2.0033,
      "step": 22130
    },
    {
      "epoch": 2.664259927797834,
      "grad_norm": 3.3809902667999268,
      "learning_rate": 0.00019979224673644843,
      "loss": 1.9003,
      "step": 22140
    },
    {
      "epoch": 2.6654632972322503,
      "grad_norm": 5.8103346824646,
      "learning_rate": 0.00019979200156917783,
      "loss": 1.8983,
      "step": 22150
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 4.957248687744141,
      "learning_rate": 0.00019979175625748356,
      "loss": 1.9426,
      "step": 22160
    },
    {
      "epoch": 2.667870036101083,
      "grad_norm": 7.3740105628967285,
      "learning_rate": 0.00019979151080136597,
      "loss": 1.7852,
      "step": 22170
    },
    {
      "epoch": 2.6690734055354994,
      "grad_norm": 5.305781364440918,
      "learning_rate": 0.00019979126520082544,
      "loss": 1.967,
      "step": 22180
    },
    {
      "epoch": 2.6702767749699157,
      "grad_norm": 3.9674527645111084,
      "learning_rate": 0.00019979101945586235,
      "loss": 1.9288,
      "step": 22190
    },
    {
      "epoch": 2.671480144404332,
      "grad_norm": 5.577676773071289,
      "learning_rate": 0.00019979077356647697,
      "loss": 1.8592,
      "step": 22200
    },
    {
      "epoch": 2.6726835138387486,
      "grad_norm": 4.325507640838623,
      "learning_rate": 0.0001997905275326697,
      "loss": 1.6857,
      "step": 22210
    },
    {
      "epoch": 2.673886883273165,
      "grad_norm": 8.045052528381348,
      "learning_rate": 0.00019979028135444092,
      "loss": 1.9639,
      "step": 22220
    },
    {
      "epoch": 2.675090252707581,
      "grad_norm": 5.293313026428223,
      "learning_rate": 0.00019979003503179095,
      "loss": 1.891,
      "step": 22230
    },
    {
      "epoch": 2.6762936221419977,
      "grad_norm": 3.4179043769836426,
      "learning_rate": 0.0001997897885647202,
      "loss": 2.0617,
      "step": 22240
    },
    {
      "epoch": 2.677496991576414,
      "grad_norm": 5.773406982421875,
      "learning_rate": 0.00019978954195322895,
      "loss": 1.9614,
      "step": 22250
    },
    {
      "epoch": 2.67870036101083,
      "grad_norm": 4.094402313232422,
      "learning_rate": 0.00019978929519731762,
      "loss": 1.8016,
      "step": 22260
    },
    {
      "epoch": 2.679903730445247,
      "grad_norm": 7.269993305206299,
      "learning_rate": 0.0001997890482969865,
      "loss": 2.0542,
      "step": 22270
    },
    {
      "epoch": 2.681107099879663,
      "grad_norm": 4.553739070892334,
      "learning_rate": 0.00019978880125223603,
      "loss": 1.6936,
      "step": 22280
    },
    {
      "epoch": 2.6823104693140793,
      "grad_norm": 4.25542688369751,
      "learning_rate": 0.00019978855406306654,
      "loss": 2.0244,
      "step": 22290
    },
    {
      "epoch": 2.683513838748496,
      "grad_norm": 6.054823398590088,
      "learning_rate": 0.00019978830672947836,
      "loss": 1.836,
      "step": 22300
    },
    {
      "epoch": 2.6847172081829123,
      "grad_norm": 5.341670513153076,
      "learning_rate": 0.00019978805925147182,
      "loss": 1.7344,
      "step": 22310
    },
    {
      "epoch": 2.6859205776173285,
      "grad_norm": 6.641788959503174,
      "learning_rate": 0.00019978781162904735,
      "loss": 2.0983,
      "step": 22320
    },
    {
      "epoch": 2.6871239470517447,
      "grad_norm": 5.529277324676514,
      "learning_rate": 0.00019978756386220532,
      "loss": 1.8698,
      "step": 22330
    },
    {
      "epoch": 2.6883273164861614,
      "grad_norm": 3.4033820629119873,
      "learning_rate": 0.000199787315950946,
      "loss": 1.8614,
      "step": 22340
    },
    {
      "epoch": 2.6895306859205776,
      "grad_norm": 5.466459274291992,
      "learning_rate": 0.0001997870678952698,
      "loss": 1.6805,
      "step": 22350
    },
    {
      "epoch": 2.690734055354994,
      "grad_norm": 4.521859645843506,
      "learning_rate": 0.0001997868196951771,
      "loss": 1.6645,
      "step": 22360
    },
    {
      "epoch": 2.69193742478941,
      "grad_norm": 7.471880912780762,
      "learning_rate": 0.00019978657135066822,
      "loss": 1.8694,
      "step": 22370
    },
    {
      "epoch": 2.693140794223827,
      "grad_norm": 4.243253707885742,
      "learning_rate": 0.00019978632286174355,
      "loss": 2.0404,
      "step": 22380
    },
    {
      "epoch": 2.694344163658243,
      "grad_norm": 3.6581270694732666,
      "learning_rate": 0.0001997860742284034,
      "loss": 1.9062,
      "step": 22390
    },
    {
      "epoch": 2.6955475330926593,
      "grad_norm": 5.81765604019165,
      "learning_rate": 0.0001997858254506482,
      "loss": 1.7131,
      "step": 22400
    },
    {
      "epoch": 2.696750902527076,
      "grad_norm": 4.9746904373168945,
      "learning_rate": 0.00019978557652847826,
      "loss": 1.8687,
      "step": 22410
    },
    {
      "epoch": 2.697954271961492,
      "grad_norm": 5.633240699768066,
      "learning_rate": 0.00019978532746189393,
      "loss": 1.6438,
      "step": 22420
    },
    {
      "epoch": 2.6991576413959084,
      "grad_norm": 5.067403793334961,
      "learning_rate": 0.00019978507825089566,
      "loss": 1.854,
      "step": 22430
    },
    {
      "epoch": 2.700361010830325,
      "grad_norm": 3.9688496589660645,
      "learning_rate": 0.0001997848288954837,
      "loss": 1.7494,
      "step": 22440
    },
    {
      "epoch": 2.7015643802647413,
      "grad_norm": 5.424824237823486,
      "learning_rate": 0.00019978457939565846,
      "loss": 1.9343,
      "step": 22450
    },
    {
      "epoch": 2.7027677496991576,
      "grad_norm": 3.9777631759643555,
      "learning_rate": 0.00019978432975142029,
      "loss": 1.7792,
      "step": 22460
    },
    {
      "epoch": 2.7039711191335742,
      "grad_norm": 7.992976665496826,
      "learning_rate": 0.00019978407996276958,
      "loss": 2.0647,
      "step": 22470
    },
    {
      "epoch": 2.7051744885679905,
      "grad_norm": 4.650917053222656,
      "learning_rate": 0.00019978383002970665,
      "loss": 1.9787,
      "step": 22480
    },
    {
      "epoch": 2.7063778580024067,
      "grad_norm": 3.4096484184265137,
      "learning_rate": 0.0001997835799522319,
      "loss": 1.921,
      "step": 22490
    },
    {
      "epoch": 2.707581227436823,
      "grad_norm": 5.971583843231201,
      "learning_rate": 0.00019978332973034566,
      "loss": 1.8383,
      "step": 22500
    },
    {
      "epoch": 2.7087845968712396,
      "grad_norm": 4.310146331787109,
      "learning_rate": 0.0001997830793640483,
      "loss": 1.8006,
      "step": 22510
    },
    {
      "epoch": 2.709987966305656,
      "grad_norm": 6.434731960296631,
      "learning_rate": 0.00019978282885334018,
      "loss": 2.0012,
      "step": 22520
    },
    {
      "epoch": 2.711191335740072,
      "grad_norm": 4.9365129470825195,
      "learning_rate": 0.00019978257819822172,
      "loss": 1.8275,
      "step": 22530
    },
    {
      "epoch": 2.7123947051744883,
      "grad_norm": 3.4394466876983643,
      "learning_rate": 0.0001997823273986932,
      "loss": 1.87,
      "step": 22540
    },
    {
      "epoch": 2.713598074608905,
      "grad_norm": 5.213423728942871,
      "learning_rate": 0.000199782076454755,
      "loss": 1.803,
      "step": 22550
    },
    {
      "epoch": 2.7148014440433212,
      "grad_norm": 4.815708637237549,
      "learning_rate": 0.00019978182536640752,
      "loss": 1.8732,
      "step": 22560
    },
    {
      "epoch": 2.7160048134777375,
      "grad_norm": 6.802679061889648,
      "learning_rate": 0.0001997815741336511,
      "loss": 1.7972,
      "step": 22570
    },
    {
      "epoch": 2.717208182912154,
      "grad_norm": 4.965303421020508,
      "learning_rate": 0.00019978132275648615,
      "loss": 1.8636,
      "step": 22580
    },
    {
      "epoch": 2.7184115523465704,
      "grad_norm": 3.497936248779297,
      "learning_rate": 0.00019978107123491294,
      "loss": 1.997,
      "step": 22590
    },
    {
      "epoch": 2.7196149217809866,
      "grad_norm": 6.207457065582275,
      "learning_rate": 0.0001997808195689319,
      "loss": 1.5876,
      "step": 22600
    },
    {
      "epoch": 2.7208182912154033,
      "grad_norm": 4.572378158569336,
      "learning_rate": 0.00019978056775854337,
      "loss": 1.9223,
      "step": 22610
    },
    {
      "epoch": 2.7220216606498195,
      "grad_norm": 7.0065202713012695,
      "learning_rate": 0.00019978031580374772,
      "loss": 1.8662,
      "step": 22620
    },
    {
      "epoch": 2.7232250300842358,
      "grad_norm": 4.983758926391602,
      "learning_rate": 0.00019978006370454534,
      "loss": 1.7349,
      "step": 22630
    },
    {
      "epoch": 2.7244283995186525,
      "grad_norm": 3.7315967082977295,
      "learning_rate": 0.00019977981146093656,
      "loss": 1.8489,
      "step": 22640
    },
    {
      "epoch": 2.7256317689530687,
      "grad_norm": 5.7065839767456055,
      "learning_rate": 0.00019977955907292177,
      "loss": 1.7187,
      "step": 22650
    },
    {
      "epoch": 2.726835138387485,
      "grad_norm": 3.9562723636627197,
      "learning_rate": 0.0001997793065405013,
      "loss": 1.7005,
      "step": 22660
    },
    {
      "epoch": 2.728038507821901,
      "grad_norm": 6.072809219360352,
      "learning_rate": 0.00019977905386367554,
      "loss": 1.9656,
      "step": 22670
    },
    {
      "epoch": 2.729241877256318,
      "grad_norm": 4.331212520599365,
      "learning_rate": 0.0001997788010424449,
      "loss": 1.963,
      "step": 22680
    },
    {
      "epoch": 2.730445246690734,
      "grad_norm": 3.5483672618865967,
      "learning_rate": 0.00019977854807680964,
      "loss": 1.788,
      "step": 22690
    },
    {
      "epoch": 2.7316486161251503,
      "grad_norm": 5.671530246734619,
      "learning_rate": 0.0001997782949667702,
      "loss": 1.7867,
      "step": 22700
    },
    {
      "epoch": 2.7328519855595665,
      "grad_norm": 4.266991138458252,
      "learning_rate": 0.00019977804171232694,
      "loss": 1.804,
      "step": 22710
    },
    {
      "epoch": 2.7340553549939832,
      "grad_norm": 7.12966251373291,
      "learning_rate": 0.00019977778831348024,
      "loss": 1.9685,
      "step": 22720
    },
    {
      "epoch": 2.7352587244283995,
      "grad_norm": 5.654016494750977,
      "learning_rate": 0.00019977753477023043,
      "loss": 1.9926,
      "step": 22730
    },
    {
      "epoch": 2.7364620938628157,
      "grad_norm": 3.3825466632843018,
      "learning_rate": 0.0001997772810825779,
      "loss": 1.8915,
      "step": 22740
    },
    {
      "epoch": 2.7376654632972324,
      "grad_norm": 5.568051815032959,
      "learning_rate": 0.000199777027250523,
      "loss": 1.8189,
      "step": 22750
    },
    {
      "epoch": 2.7388688327316486,
      "grad_norm": 4.8424201011657715,
      "learning_rate": 0.0001997767732740661,
      "loss": 1.8559,
      "step": 22760
    },
    {
      "epoch": 2.740072202166065,
      "grad_norm": 8.739863395690918,
      "learning_rate": 0.0001997765191532076,
      "loss": 1.8546,
      "step": 22770
    },
    {
      "epoch": 2.7412755716004815,
      "grad_norm": 4.288040637969971,
      "learning_rate": 0.00019977626488794782,
      "loss": 1.5549,
      "step": 22780
    },
    {
      "epoch": 2.7424789410348978,
      "grad_norm": 3.2434864044189453,
      "learning_rate": 0.00019977601047828716,
      "loss": 1.98,
      "step": 22790
    },
    {
      "epoch": 2.743682310469314,
      "grad_norm": 4.835328102111816,
      "learning_rate": 0.00019977575592422598,
      "loss": 1.9714,
      "step": 22800
    },
    {
      "epoch": 2.7448856799037307,
      "grad_norm": 5.448550224304199,
      "learning_rate": 0.00019977550122576463,
      "loss": 1.8661,
      "step": 22810
    },
    {
      "epoch": 2.746089049338147,
      "grad_norm": 6.611584186553955,
      "learning_rate": 0.0001997752463829035,
      "loss": 2.0355,
      "step": 22820
    },
    {
      "epoch": 2.747292418772563,
      "grad_norm": 5.065571308135986,
      "learning_rate": 0.00019977499139564296,
      "loss": 1.6624,
      "step": 22830
    },
    {
      "epoch": 2.74849578820698,
      "grad_norm": 3.28558349609375,
      "learning_rate": 0.0001997747362639834,
      "loss": 2.0575,
      "step": 22840
    },
    {
      "epoch": 2.749699157641396,
      "grad_norm": 6.408359527587891,
      "learning_rate": 0.00019977448098792514,
      "loss": 1.8269,
      "step": 22850
    },
    {
      "epoch": 2.7509025270758123,
      "grad_norm": 4.822616100311279,
      "learning_rate": 0.00019977422556746857,
      "loss": 1.797,
      "step": 22860
    },
    {
      "epoch": 2.7521058965102285,
      "grad_norm": 7.014891147613525,
      "learning_rate": 0.00019977397000261408,
      "loss": 2.1429,
      "step": 22870
    },
    {
      "epoch": 2.7533092659446448,
      "grad_norm": 5.4905171394348145,
      "learning_rate": 0.000199773714293362,
      "loss": 1.7902,
      "step": 22880
    },
    {
      "epoch": 2.7545126353790614,
      "grad_norm": 3.7834863662719727,
      "learning_rate": 0.00019977345843971275,
      "loss": 1.8674,
      "step": 22890
    },
    {
      "epoch": 2.7557160048134777,
      "grad_norm": 5.6181535720825195,
      "learning_rate": 0.00019977320244166663,
      "loss": 1.8222,
      "step": 22900
    },
    {
      "epoch": 2.756919374247894,
      "grad_norm": 4.6425395011901855,
      "learning_rate": 0.00019977294629922406,
      "loss": 2.0321,
      "step": 22910
    },
    {
      "epoch": 2.7581227436823106,
      "grad_norm": 6.405762672424316,
      "learning_rate": 0.00019977269001238544,
      "loss": 1.8953,
      "step": 22920
    },
    {
      "epoch": 2.759326113116727,
      "grad_norm": 5.6902313232421875,
      "learning_rate": 0.0001997724335811511,
      "loss": 1.6785,
      "step": 22930
    },
    {
      "epoch": 2.760529482551143,
      "grad_norm": 3.4790284633636475,
      "learning_rate": 0.00019977217700552138,
      "loss": 1.9017,
      "step": 22940
    },
    {
      "epoch": 2.7617328519855597,
      "grad_norm": 6.034868240356445,
      "learning_rate": 0.00019977192028549672,
      "loss": 1.9224,
      "step": 22950
    },
    {
      "epoch": 2.762936221419976,
      "grad_norm": 4.756815433502197,
      "learning_rate": 0.00019977166342107741,
      "loss": 1.913,
      "step": 22960
    },
    {
      "epoch": 2.764139590854392,
      "grad_norm": 6.949899673461914,
      "learning_rate": 0.00019977140641226393,
      "loss": 1.7652,
      "step": 22970
    },
    {
      "epoch": 2.765342960288809,
      "grad_norm": 4.67728853225708,
      "learning_rate": 0.00019977114925905657,
      "loss": 1.9011,
      "step": 22980
    },
    {
      "epoch": 2.766546329723225,
      "grad_norm": 3.5102603435516357,
      "learning_rate": 0.00019977089196145573,
      "loss": 2.1067,
      "step": 22990
    },
    {
      "epoch": 2.7677496991576414,
      "grad_norm": 5.74012565612793,
      "learning_rate": 0.00019977063451946176,
      "loss": 2.0833,
      "step": 23000
    },
    {
      "epoch": 2.768953068592058,
      "grad_norm": 4.5373616218566895,
      "learning_rate": 0.00019977037693307508,
      "loss": 1.8623,
      "step": 23010
    },
    {
      "epoch": 2.7701564380264743,
      "grad_norm": 6.993628025054932,
      "learning_rate": 0.000199770119202296,
      "loss": 1.7051,
      "step": 23020
    },
    {
      "epoch": 2.7713598074608905,
      "grad_norm": 4.627161979675293,
      "learning_rate": 0.00019976986132712492,
      "loss": 1.8854,
      "step": 23030
    },
    {
      "epoch": 2.7725631768953067,
      "grad_norm": 4.533265590667725,
      "learning_rate": 0.00019976960330756222,
      "loss": 2.2145,
      "step": 23040
    },
    {
      "epoch": 2.773766546329723,
      "grad_norm": 5.27393102645874,
      "learning_rate": 0.0001997693451436083,
      "loss": 1.7967,
      "step": 23050
    },
    {
      "epoch": 2.7749699157641396,
      "grad_norm": 4.08444881439209,
      "learning_rate": 0.00019976908683526349,
      "loss": 2.0044,
      "step": 23060
    },
    {
      "epoch": 2.776173285198556,
      "grad_norm": 6.545950889587402,
      "learning_rate": 0.00019976882838252817,
      "loss": 1.9537,
      "step": 23070
    },
    {
      "epoch": 2.777376654632972,
      "grad_norm": 4.16444730758667,
      "learning_rate": 0.00019976856978540274,
      "loss": 1.7258,
      "step": 23080
    },
    {
      "epoch": 2.778580024067389,
      "grad_norm": 5.483246803283691,
      "learning_rate": 0.00019976831104388755,
      "loss": 1.8828,
      "step": 23090
    },
    {
      "epoch": 2.779783393501805,
      "grad_norm": 6.4330153465271,
      "learning_rate": 0.000199768052157983,
      "loss": 1.8548,
      "step": 23100
    },
    {
      "epoch": 2.7809867629362213,
      "grad_norm": 5.590136528015137,
      "learning_rate": 0.00019976779312768942,
      "loss": 1.7924,
      "step": 23110
    },
    {
      "epoch": 2.782190132370638,
      "grad_norm": 6.044424533843994,
      "learning_rate": 0.00019976753395300722,
      "loss": 1.9499,
      "step": 23120
    },
    {
      "epoch": 2.783393501805054,
      "grad_norm": 5.545258045196533,
      "learning_rate": 0.00019976727463393677,
      "loss": 1.7807,
      "step": 23130
    },
    {
      "epoch": 2.7845968712394704,
      "grad_norm": 3.408629894256592,
      "learning_rate": 0.00019976701517047843,
      "loss": 2.015,
      "step": 23140
    },
    {
      "epoch": 2.785800240673887,
      "grad_norm": 6.710221767425537,
      "learning_rate": 0.0001997667555626326,
      "loss": 1.7425,
      "step": 23150
    },
    {
      "epoch": 2.7870036101083033,
      "grad_norm": 4.039109230041504,
      "learning_rate": 0.00019976649581039964,
      "loss": 1.9283,
      "step": 23160
    },
    {
      "epoch": 2.7882069795427196,
      "grad_norm": 7.614040851593018,
      "learning_rate": 0.00019976623591377992,
      "loss": 1.803,
      "step": 23170
    },
    {
      "epoch": 2.7894103489771362,
      "grad_norm": 4.319331169128418,
      "learning_rate": 0.00019976597587277383,
      "loss": 1.6085,
      "step": 23180
    },
    {
      "epoch": 2.7906137184115525,
      "grad_norm": 3.3719117641448975,
      "learning_rate": 0.00019976571568738175,
      "loss": 1.9033,
      "step": 23190
    },
    {
      "epoch": 2.7918170878459687,
      "grad_norm": 5.799076557159424,
      "learning_rate": 0.00019976545535760405,
      "loss": 2.0389,
      "step": 23200
    },
    {
      "epoch": 2.793020457280385,
      "grad_norm": 3.9176876544952393,
      "learning_rate": 0.00019976519488344109,
      "loss": 1.9371,
      "step": 23210
    },
    {
      "epoch": 2.794223826714801,
      "grad_norm": 6.319335460662842,
      "learning_rate": 0.0001997649342648933,
      "loss": 2.0688,
      "step": 23220
    },
    {
      "epoch": 2.795427196149218,
      "grad_norm": 4.826409339904785,
      "learning_rate": 0.00019976467350196097,
      "loss": 1.9013,
      "step": 23230
    },
    {
      "epoch": 2.796630565583634,
      "grad_norm": 3.234429359436035,
      "learning_rate": 0.00019976441259464455,
      "loss": 1.9304,
      "step": 23240
    },
    {
      "epoch": 2.7978339350180503,
      "grad_norm": 5.633996486663818,
      "learning_rate": 0.00019976415154294438,
      "loss": 1.9082,
      "step": 23250
    },
    {
      "epoch": 2.799037304452467,
      "grad_norm": 6.302307605743408,
      "learning_rate": 0.00019976389034686085,
      "loss": 1.775,
      "step": 23260
    },
    {
      "epoch": 2.8002406738868832,
      "grad_norm": 6.776801109313965,
      "learning_rate": 0.00019976362900639435,
      "loss": 1.9379,
      "step": 23270
    },
    {
      "epoch": 2.8014440433212995,
      "grad_norm": 4.77065896987915,
      "learning_rate": 0.00019976336752154525,
      "loss": 1.7112,
      "step": 23280
    },
    {
      "epoch": 2.802647412755716,
      "grad_norm": 3.3480606079101562,
      "learning_rate": 0.00019976310589231392,
      "loss": 1.6538,
      "step": 23290
    },
    {
      "epoch": 2.8038507821901324,
      "grad_norm": 6.364193439483643,
      "learning_rate": 0.00019976284411870078,
      "loss": 1.9962,
      "step": 23300
    },
    {
      "epoch": 2.8050541516245486,
      "grad_norm": 3.677845001220703,
      "learning_rate": 0.0001997625822007061,
      "loss": 1.7511,
      "step": 23310
    },
    {
      "epoch": 2.8062575210589653,
      "grad_norm": 7.029861927032471,
      "learning_rate": 0.00019976232013833037,
      "loss": 1.8961,
      "step": 23320
    },
    {
      "epoch": 2.8074608904933815,
      "grad_norm": 4.60750150680542,
      "learning_rate": 0.00019976205793157396,
      "loss": 1.7058,
      "step": 23330
    },
    {
      "epoch": 2.808664259927798,
      "grad_norm": 4.142286777496338,
      "learning_rate": 0.0001997617955804372,
      "loss": 1.9601,
      "step": 23340
    },
    {
      "epoch": 2.8098676293622145,
      "grad_norm": 5.287271499633789,
      "learning_rate": 0.00019976153308492049,
      "loss": 1.9171,
      "step": 23350
    },
    {
      "epoch": 2.8110709987966307,
      "grad_norm": 4.549359321594238,
      "learning_rate": 0.00019976127044502422,
      "loss": 1.8671,
      "step": 23360
    },
    {
      "epoch": 2.812274368231047,
      "grad_norm": 6.705315589904785,
      "learning_rate": 0.00019976100766074873,
      "loss": 1.9865,
      "step": 23370
    },
    {
      "epoch": 2.813477737665463,
      "grad_norm": 4.726195335388184,
      "learning_rate": 0.0001997607447320945,
      "loss": 1.7513,
      "step": 23380
    },
    {
      "epoch": 2.8146811070998794,
      "grad_norm": 3.288010597229004,
      "learning_rate": 0.00019976048165906178,
      "loss": 1.8895,
      "step": 23390
    },
    {
      "epoch": 2.815884476534296,
      "grad_norm": 5.803439617156982,
      "learning_rate": 0.00019976021844165102,
      "loss": 2.0504,
      "step": 23400
    },
    {
      "epoch": 2.8170878459687123,
      "grad_norm": 5.1353535652160645,
      "learning_rate": 0.0001997599550798626,
      "loss": 1.9217,
      "step": 23410
    },
    {
      "epoch": 2.8182912154031285,
      "grad_norm": 6.596933364868164,
      "learning_rate": 0.00019975969157369692,
      "loss": 2.0369,
      "step": 23420
    },
    {
      "epoch": 2.8194945848375452,
      "grad_norm": 6.023066520690918,
      "learning_rate": 0.00019975942792315433,
      "loss": 1.9239,
      "step": 23430
    },
    {
      "epoch": 2.8206979542719615,
      "grad_norm": 3.893421173095703,
      "learning_rate": 0.00019975916412823518,
      "loss": 1.864,
      "step": 23440
    },
    {
      "epoch": 2.8219013237063777,
      "grad_norm": 6.207060813903809,
      "learning_rate": 0.00019975890018893993,
      "loss": 1.8543,
      "step": 23450
    },
    {
      "epoch": 2.8231046931407944,
      "grad_norm": 4.030089378356934,
      "learning_rate": 0.00019975863610526888,
      "loss": 2.0821,
      "step": 23460
    },
    {
      "epoch": 2.8243080625752106,
      "grad_norm": 7.685181617736816,
      "learning_rate": 0.0001997583718772225,
      "loss": 1.8729,
      "step": 23470
    },
    {
      "epoch": 2.825511432009627,
      "grad_norm": 4.690788269042969,
      "learning_rate": 0.00019975810750480112,
      "loss": 1.8349,
      "step": 23480
    },
    {
      "epoch": 2.8267148014440435,
      "grad_norm": 3.888273239135742,
      "learning_rate": 0.00019975784298800512,
      "loss": 1.9653,
      "step": 23490
    },
    {
      "epoch": 2.8279181708784598,
      "grad_norm": 5.698587894439697,
      "learning_rate": 0.00019975757832683489,
      "loss": 1.8071,
      "step": 23500
    },
    {
      "epoch": 2.829121540312876,
      "grad_norm": 4.445411205291748,
      "learning_rate": 0.0001997573135212908,
      "loss": 1.9974,
      "step": 23510
    },
    {
      "epoch": 2.8303249097472927,
      "grad_norm": 6.7462897300720215,
      "learning_rate": 0.00019975704857137325,
      "loss": 2.0135,
      "step": 23520
    },
    {
      "epoch": 2.831528279181709,
      "grad_norm": 4.6754536628723145,
      "learning_rate": 0.00019975678347708265,
      "loss": 1.6235,
      "step": 23530
    },
    {
      "epoch": 2.832731648616125,
      "grad_norm": 4.018809795379639,
      "learning_rate": 0.00019975651823841932,
      "loss": 1.7181,
      "step": 23540
    },
    {
      "epoch": 2.8339350180505414,
      "grad_norm": 6.560571193695068,
      "learning_rate": 0.0001997562528553837,
      "loss": 1.9334,
      "step": 23550
    },
    {
      "epoch": 2.8351383874849576,
      "grad_norm": 4.695711135864258,
      "learning_rate": 0.00019975598732797613,
      "loss": 1.7037,
      "step": 23560
    },
    {
      "epoch": 2.8363417569193743,
      "grad_norm": 5.652125835418701,
      "learning_rate": 0.00019975572165619704,
      "loss": 1.8348,
      "step": 23570
    },
    {
      "epoch": 2.8375451263537905,
      "grad_norm": 4.512273788452148,
      "learning_rate": 0.0001997554558400468,
      "loss": 1.7516,
      "step": 23580
    },
    {
      "epoch": 2.8387484957882068,
      "grad_norm": 3.05820894241333,
      "learning_rate": 0.00019975518987952576,
      "loss": 1.9117,
      "step": 23590
    },
    {
      "epoch": 2.8399518652226234,
      "grad_norm": 5.962392330169678,
      "learning_rate": 0.00019975492377463435,
      "loss": 1.6103,
      "step": 23600
    },
    {
      "epoch": 2.8411552346570397,
      "grad_norm": 5.1483049392700195,
      "learning_rate": 0.00019975465752537295,
      "loss": 1.9393,
      "step": 23610
    },
    {
      "epoch": 2.842358604091456,
      "grad_norm": 7.242753028869629,
      "learning_rate": 0.00019975439113174192,
      "loss": 1.9032,
      "step": 23620
    },
    {
      "epoch": 2.8435619735258726,
      "grad_norm": 5.350884437561035,
      "learning_rate": 0.00019975412459374164,
      "loss": 1.9788,
      "step": 23630
    },
    {
      "epoch": 2.844765342960289,
      "grad_norm": 3.617100715637207,
      "learning_rate": 0.0001997538579113725,
      "loss": 1.986,
      "step": 23640
    },
    {
      "epoch": 2.845968712394705,
      "grad_norm": 6.817924499511719,
      "learning_rate": 0.00019975359108463496,
      "loss": 1.9646,
      "step": 23650
    },
    {
      "epoch": 2.8471720818291217,
      "grad_norm": 5.259153366088867,
      "learning_rate": 0.0001997533241135293,
      "loss": 1.837,
      "step": 23660
    },
    {
      "epoch": 2.848375451263538,
      "grad_norm": 5.788957595825195,
      "learning_rate": 0.00019975305699805596,
      "loss": 1.7797,
      "step": 23670
    },
    {
      "epoch": 2.849578820697954,
      "grad_norm": 4.100399494171143,
      "learning_rate": 0.00019975278973821532,
      "loss": 1.723,
      "step": 23680
    },
    {
      "epoch": 2.850782190132371,
      "grad_norm": 3.2705655097961426,
      "learning_rate": 0.00019975252233400776,
      "loss": 1.6919,
      "step": 23690
    },
    {
      "epoch": 2.851985559566787,
      "grad_norm": 5.190719127655029,
      "learning_rate": 0.00019975225478543367,
      "loss": 1.7586,
      "step": 23700
    },
    {
      "epoch": 2.8531889290012034,
      "grad_norm": 4.4668731689453125,
      "learning_rate": 0.00019975198709249346,
      "loss": 1.6683,
      "step": 23710
    },
    {
      "epoch": 2.8543922984356196,
      "grad_norm": 5.66727352142334,
      "learning_rate": 0.00019975171925518747,
      "loss": 2.0834,
      "step": 23720
    },
    {
      "epoch": 2.855595667870036,
      "grad_norm": 4.898777008056641,
      "learning_rate": 0.00019975145127351613,
      "loss": 1.7352,
      "step": 23730
    },
    {
      "epoch": 2.8567990373044525,
      "grad_norm": 3.783613920211792,
      "learning_rate": 0.0001997511831474798,
      "loss": 1.827,
      "step": 23740
    },
    {
      "epoch": 2.8580024067388687,
      "grad_norm": 9.265079498291016,
      "learning_rate": 0.0001997509148770789,
      "loss": 1.8773,
      "step": 23750
    },
    {
      "epoch": 2.859205776173285,
      "grad_norm": 4.308231830596924,
      "learning_rate": 0.00019975064646231378,
      "loss": 1.86,
      "step": 23760
    },
    {
      "epoch": 2.8604091456077017,
      "grad_norm": 5.7617411613464355,
      "learning_rate": 0.00019975037790318488,
      "loss": 2.0498,
      "step": 23770
    },
    {
      "epoch": 2.861612515042118,
      "grad_norm": 6.328868389129639,
      "learning_rate": 0.00019975010919969253,
      "loss": 1.7433,
      "step": 23780
    },
    {
      "epoch": 2.862815884476534,
      "grad_norm": 3.3558099269866943,
      "learning_rate": 0.00019974984035183715,
      "loss": 1.9166,
      "step": 23790
    },
    {
      "epoch": 2.864019253910951,
      "grad_norm": 5.800656318664551,
      "learning_rate": 0.0001997495713596191,
      "loss": 2.0589,
      "step": 23800
    },
    {
      "epoch": 2.865222623345367,
      "grad_norm": 4.649380207061768,
      "learning_rate": 0.00019974930222303882,
      "loss": 1.8411,
      "step": 23810
    },
    {
      "epoch": 2.8664259927797833,
      "grad_norm": 6.917948246002197,
      "learning_rate": 0.00019974903294209666,
      "loss": 2.0134,
      "step": 23820
    },
    {
      "epoch": 2.8676293622142,
      "grad_norm": 4.925341606140137,
      "learning_rate": 0.00019974876351679303,
      "loss": 1.6361,
      "step": 23830
    },
    {
      "epoch": 2.868832731648616,
      "grad_norm": 4.769173622131348,
      "learning_rate": 0.00019974849394712832,
      "loss": 2.1269,
      "step": 23840
    },
    {
      "epoch": 2.8700361010830324,
      "grad_norm": 5.753753662109375,
      "learning_rate": 0.0001997482242331029,
      "loss": 1.9573,
      "step": 23850
    },
    {
      "epoch": 2.871239470517449,
      "grad_norm": 3.9171459674835205,
      "learning_rate": 0.00019974795437471715,
      "loss": 1.8201,
      "step": 23860
    },
    {
      "epoch": 2.8724428399518653,
      "grad_norm": 7.415256977081299,
      "learning_rate": 0.00019974768437197154,
      "loss": 2.1456,
      "step": 23870
    },
    {
      "epoch": 2.8736462093862816,
      "grad_norm": 5.117340087890625,
      "learning_rate": 0.00019974741422486636,
      "loss": 1.6893,
      "step": 23880
    },
    {
      "epoch": 2.874849578820698,
      "grad_norm": 3.578383207321167,
      "learning_rate": 0.00019974714393340205,
      "loss": 1.9563,
      "step": 23890
    },
    {
      "epoch": 2.8760529482551145,
      "grad_norm": 5.163645267486572,
      "learning_rate": 0.00019974687349757898,
      "loss": 1.83,
      "step": 23900
    },
    {
      "epoch": 2.8772563176895307,
      "grad_norm": 4.585817813873291,
      "learning_rate": 0.0001997466029173976,
      "loss": 1.9772,
      "step": 23910
    },
    {
      "epoch": 2.878459687123947,
      "grad_norm": 7.4253830909729,
      "learning_rate": 0.00019974633219285823,
      "loss": 1.9644,
      "step": 23920
    },
    {
      "epoch": 2.879663056558363,
      "grad_norm": 4.518739700317383,
      "learning_rate": 0.00019974606132396128,
      "loss": 1.839,
      "step": 23930
    },
    {
      "epoch": 2.88086642599278,
      "grad_norm": 4.008555889129639,
      "learning_rate": 0.00019974579031070718,
      "loss": 1.9355,
      "step": 23940
    },
    {
      "epoch": 2.882069795427196,
      "grad_norm": 5.686634063720703,
      "learning_rate": 0.0001997455191530963,
      "loss": 1.7624,
      "step": 23950
    },
    {
      "epoch": 2.8832731648616123,
      "grad_norm": 4.48956823348999,
      "learning_rate": 0.000199745247851129,
      "loss": 1.805,
      "step": 23960
    },
    {
      "epoch": 2.884476534296029,
      "grad_norm": 7.608232498168945,
      "learning_rate": 0.0001997449764048057,
      "loss": 2.0757,
      "step": 23970
    },
    {
      "epoch": 2.8856799037304453,
      "grad_norm": 5.478265285491943,
      "learning_rate": 0.0001997447048141268,
      "loss": 2.0261,
      "step": 23980
    },
    {
      "epoch": 2.8868832731648615,
      "grad_norm": 3.0724799633026123,
      "learning_rate": 0.0001997444330790927,
      "loss": 1.9523,
      "step": 23990
    },
    {
      "epoch": 2.888086642599278,
      "grad_norm": 6.806112289428711,
      "learning_rate": 0.00019974416119970378,
      "loss": 1.8005,
      "step": 24000
    },
    {
      "epoch": 2.8892900120336944,
      "grad_norm": 4.565250873565674,
      "learning_rate": 0.00019974388917596043,
      "loss": 1.8808,
      "step": 24010
    },
    {
      "epoch": 2.8904933814681106,
      "grad_norm": 5.74095344543457,
      "learning_rate": 0.00019974361700786304,
      "loss": 2.1453,
      "step": 24020
    },
    {
      "epoch": 2.8916967509025273,
      "grad_norm": 4.752941131591797,
      "learning_rate": 0.000199743344695412,
      "loss": 1.6872,
      "step": 24030
    },
    {
      "epoch": 2.8929001203369435,
      "grad_norm": 3.5324058532714844,
      "learning_rate": 0.00019974307223860773,
      "loss": 1.9335,
      "step": 24040
    },
    {
      "epoch": 2.89410348977136,
      "grad_norm": 5.88004732131958,
      "learning_rate": 0.00019974279963745061,
      "loss": 1.7958,
      "step": 24050
    },
    {
      "epoch": 2.895306859205776,
      "grad_norm": 4.072971820831299,
      "learning_rate": 0.000199742526891941,
      "loss": 2.0524,
      "step": 24060
    },
    {
      "epoch": 2.8965102286401927,
      "grad_norm": 8.117522239685059,
      "learning_rate": 0.00019974225400207937,
      "loss": 1.7339,
      "step": 24070
    },
    {
      "epoch": 2.897713598074609,
      "grad_norm": 4.792110443115234,
      "learning_rate": 0.00019974198096786607,
      "loss": 1.7673,
      "step": 24080
    },
    {
      "epoch": 2.898916967509025,
      "grad_norm": 3.5243046283721924,
      "learning_rate": 0.00019974170778930148,
      "loss": 1.6451,
      "step": 24090
    },
    {
      "epoch": 2.9001203369434414,
      "grad_norm": 6.484212398529053,
      "learning_rate": 0.00019974143446638602,
      "loss": 1.9361,
      "step": 24100
    },
    {
      "epoch": 2.901323706377858,
      "grad_norm": 4.859592437744141,
      "learning_rate": 0.0001997411609991201,
      "loss": 1.9698,
      "step": 24110
    },
    {
      "epoch": 2.9025270758122743,
      "grad_norm": 6.178374290466309,
      "learning_rate": 0.00019974088738750406,
      "loss": 1.7964,
      "step": 24120
    },
    {
      "epoch": 2.9037304452466906,
      "grad_norm": 5.469675064086914,
      "learning_rate": 0.00019974061363153833,
      "loss": 1.7654,
      "step": 24130
    },
    {
      "epoch": 2.9049338146811072,
      "grad_norm": 3.926400661468506,
      "learning_rate": 0.00019974033973122333,
      "loss": 2.025,
      "step": 24140
    },
    {
      "epoch": 2.9061371841155235,
      "grad_norm": 5.796122074127197,
      "learning_rate": 0.00019974006568655942,
      "loss": 1.7356,
      "step": 24150
    },
    {
      "epoch": 2.9073405535499397,
      "grad_norm": 3.973008871078491,
      "learning_rate": 0.000199739791497547,
      "loss": 1.7903,
      "step": 24160
    },
    {
      "epoch": 2.9085439229843564,
      "grad_norm": 6.530457496643066,
      "learning_rate": 0.0001997395171641865,
      "loss": 1.7029,
      "step": 24170
    },
    {
      "epoch": 2.9097472924187726,
      "grad_norm": 4.4459614753723145,
      "learning_rate": 0.00019973924268647827,
      "loss": 1.8556,
      "step": 24180
    },
    {
      "epoch": 2.910950661853189,
      "grad_norm": 3.5894134044647217,
      "learning_rate": 0.00019973896806442276,
      "loss": 1.9814,
      "step": 24190
    },
    {
      "epoch": 2.9121540312876055,
      "grad_norm": 5.986927032470703,
      "learning_rate": 0.00019973869329802034,
      "loss": 1.9392,
      "step": 24200
    },
    {
      "epoch": 2.9133574007220218,
      "grad_norm": 4.745354175567627,
      "learning_rate": 0.0001997384183872714,
      "loss": 1.9679,
      "step": 24210
    },
    {
      "epoch": 2.914560770156438,
      "grad_norm": 5.20849084854126,
      "learning_rate": 0.00019973814333217632,
      "loss": 1.8934,
      "step": 24220
    },
    {
      "epoch": 2.9157641395908542,
      "grad_norm": 5.41616153717041,
      "learning_rate": 0.00019973786813273553,
      "loss": 1.8729,
      "step": 24230
    },
    {
      "epoch": 2.916967509025271,
      "grad_norm": 3.55181884765625,
      "learning_rate": 0.0001997375927889494,
      "loss": 1.6742,
      "step": 24240
    },
    {
      "epoch": 2.918170878459687,
      "grad_norm": 6.453026294708252,
      "learning_rate": 0.0001997373173008184,
      "loss": 2.0943,
      "step": 24250
    },
    {
      "epoch": 2.9193742478941034,
      "grad_norm": 4.038445472717285,
      "learning_rate": 0.00019973704166834284,
      "loss": 1.8318,
      "step": 24260
    },
    {
      "epoch": 2.9205776173285196,
      "grad_norm": 7.606594562530518,
      "learning_rate": 0.00019973676589152316,
      "loss": 1.8344,
      "step": 24270
    },
    {
      "epoch": 2.9217809867629363,
      "grad_norm": 5.004695892333984,
      "learning_rate": 0.00019973648997035978,
      "loss": 1.7604,
      "step": 24280
    },
    {
      "epoch": 2.9229843561973525,
      "grad_norm": 3.423807144165039,
      "learning_rate": 0.00019973621390485303,
      "loss": 1.9921,
      "step": 24290
    },
    {
      "epoch": 2.9241877256317688,
      "grad_norm": 5.876102924346924,
      "learning_rate": 0.00019973593769500342,
      "loss": 1.8028,
      "step": 24300
    },
    {
      "epoch": 2.9253910950661854,
      "grad_norm": 5.172757625579834,
      "learning_rate": 0.00019973566134081123,
      "loss": 1.8126,
      "step": 24310
    },
    {
      "epoch": 2.9265944645006017,
      "grad_norm": 6.53843355178833,
      "learning_rate": 0.00019973538484227695,
      "loss": 1.7146,
      "step": 24320
    },
    {
      "epoch": 2.927797833935018,
      "grad_norm": 5.280789375305176,
      "learning_rate": 0.00019973510819940092,
      "loss": 1.8282,
      "step": 24330
    },
    {
      "epoch": 2.9290012033694346,
      "grad_norm": 2.917238473892212,
      "learning_rate": 0.00019973483141218356,
      "loss": 1.9238,
      "step": 24340
    },
    {
      "epoch": 2.930204572803851,
      "grad_norm": 6.673867225646973,
      "learning_rate": 0.00019973455448062528,
      "loss": 1.8187,
      "step": 24350
    },
    {
      "epoch": 2.931407942238267,
      "grad_norm": 4.365293979644775,
      "learning_rate": 0.0001997342774047265,
      "loss": 1.961,
      "step": 24360
    },
    {
      "epoch": 2.9326113116726837,
      "grad_norm": 7.935892105102539,
      "learning_rate": 0.00019973400018448757,
      "loss": 1.9479,
      "step": 24370
    },
    {
      "epoch": 2.9338146811071,
      "grad_norm": 4.404233455657959,
      "learning_rate": 0.00019973372281990894,
      "loss": 1.8156,
      "step": 24380
    },
    {
      "epoch": 2.935018050541516,
      "grad_norm": 3.4874532222747803,
      "learning_rate": 0.00019973344531099097,
      "loss": 2.0095,
      "step": 24390
    },
    {
      "epoch": 2.9362214199759324,
      "grad_norm": 5.68506383895874,
      "learning_rate": 0.00019973316765773408,
      "loss": 1.892,
      "step": 24400
    },
    {
      "epoch": 2.937424789410349,
      "grad_norm": 4.7850260734558105,
      "learning_rate": 0.00019973288986013866,
      "loss": 1.8653,
      "step": 24410
    },
    {
      "epoch": 2.9386281588447654,
      "grad_norm": 7.241191864013672,
      "learning_rate": 0.00019973261191820517,
      "loss": 1.9725,
      "step": 24420
    },
    {
      "epoch": 2.9398315282791816,
      "grad_norm": 5.28766393661499,
      "learning_rate": 0.00019973233383193396,
      "loss": 1.9111,
      "step": 24430
    },
    {
      "epoch": 2.941034897713598,
      "grad_norm": 3.953686475753784,
      "learning_rate": 0.0001997320556013254,
      "loss": 1.7484,
      "step": 24440
    },
    {
      "epoch": 2.9422382671480145,
      "grad_norm": 5.54633092880249,
      "learning_rate": 0.00019973177722637996,
      "loss": 1.8454,
      "step": 24450
    },
    {
      "epoch": 2.9434416365824307,
      "grad_norm": 4.033097267150879,
      "learning_rate": 0.00019973149870709803,
      "loss": 1.9392,
      "step": 24460
    },
    {
      "epoch": 2.944645006016847,
      "grad_norm": 6.493199348449707,
      "learning_rate": 0.00019973122004347997,
      "loss": 2.0715,
      "step": 24470
    },
    {
      "epoch": 2.9458483754512637,
      "grad_norm": 4.184035301208496,
      "learning_rate": 0.0001997309412355262,
      "loss": 1.8558,
      "step": 24480
    },
    {
      "epoch": 2.94705174488568,
      "grad_norm": 4.929866313934326,
      "learning_rate": 0.0001997306622832372,
      "loss": 1.8668,
      "step": 24490
    },
    {
      "epoch": 2.948255114320096,
      "grad_norm": 5.194972515106201,
      "learning_rate": 0.00019973038318661326,
      "loss": 1.9602,
      "step": 24500
    },
    {
      "epoch": 2.949458483754513,
      "grad_norm": 4.429738521575928,
      "learning_rate": 0.0001997301039456548,
      "loss": 1.8157,
      "step": 24510
    },
    {
      "epoch": 2.950661853188929,
      "grad_norm": 7.0151519775390625,
      "learning_rate": 0.00019972982456036232,
      "loss": 2.0124,
      "step": 24520
    },
    {
      "epoch": 2.9518652226233453,
      "grad_norm": 5.25596284866333,
      "learning_rate": 0.00019972954503073614,
      "loss": 1.9175,
      "step": 24530
    },
    {
      "epoch": 2.953068592057762,
      "grad_norm": 3.3625776767730713,
      "learning_rate": 0.00019972926535677667,
      "loss": 1.8197,
      "step": 24540
    },
    {
      "epoch": 2.954271961492178,
      "grad_norm": 5.699072360992432,
      "learning_rate": 0.00019972898553848433,
      "loss": 1.6902,
      "step": 24550
    },
    {
      "epoch": 2.9554753309265944,
      "grad_norm": 4.332197666168213,
      "learning_rate": 0.00019972870557585958,
      "loss": 1.7647,
      "step": 24560
    },
    {
      "epoch": 2.956678700361011,
      "grad_norm": 5.7750020027160645,
      "learning_rate": 0.00019972842546890269,
      "loss": 1.8058,
      "step": 24570
    },
    {
      "epoch": 2.9578820697954273,
      "grad_norm": 5.301845073699951,
      "learning_rate": 0.0001997281452176142,
      "loss": 1.8142,
      "step": 24580
    },
    {
      "epoch": 2.9590854392298436,
      "grad_norm": 3.2149767875671387,
      "learning_rate": 0.00019972786482199445,
      "loss": 1.8397,
      "step": 24590
    },
    {
      "epoch": 2.96028880866426,
      "grad_norm": 5.64735221862793,
      "learning_rate": 0.00019972758428204386,
      "loss": 1.744,
      "step": 24600
    },
    {
      "epoch": 2.961492178098676,
      "grad_norm": 3.4880824089050293,
      "learning_rate": 0.00019972730359776284,
      "loss": 1.8447,
      "step": 24610
    },
    {
      "epoch": 2.9626955475330927,
      "grad_norm": 6.680085182189941,
      "learning_rate": 0.00019972702276915176,
      "loss": 1.8838,
      "step": 24620
    },
    {
      "epoch": 2.963898916967509,
      "grad_norm": 5.064449787139893,
      "learning_rate": 0.00019972674179621108,
      "loss": 1.6851,
      "step": 24630
    },
    {
      "epoch": 2.965102286401925,
      "grad_norm": 3.714988946914673,
      "learning_rate": 0.00019972646067894115,
      "loss": 1.8391,
      "step": 24640
    },
    {
      "epoch": 2.966305655836342,
      "grad_norm": 6.406152248382568,
      "learning_rate": 0.00019972617941734245,
      "loss": 1.8827,
      "step": 24650
    },
    {
      "epoch": 2.967509025270758,
      "grad_norm": 4.480008125305176,
      "learning_rate": 0.00019972589801141535,
      "loss": 1.9606,
      "step": 24660
    },
    {
      "epoch": 2.9687123947051743,
      "grad_norm": 6.958746433258057,
      "learning_rate": 0.00019972561646116024,
      "loss": 2.0608,
      "step": 24670
    },
    {
      "epoch": 2.969915764139591,
      "grad_norm": 5.884395599365234,
      "learning_rate": 0.00019972533476657757,
      "loss": 1.7973,
      "step": 24680
    },
    {
      "epoch": 2.9711191335740073,
      "grad_norm": 4.368142127990723,
      "learning_rate": 0.00019972505292766766,
      "loss": 1.7885,
      "step": 24690
    },
    {
      "epoch": 2.9723225030084235,
      "grad_norm": 5.5579705238342285,
      "learning_rate": 0.00019972477094443102,
      "loss": 2.0045,
      "step": 24700
    },
    {
      "epoch": 2.97352587244284,
      "grad_norm": 4.20939826965332,
      "learning_rate": 0.00019972448881686798,
      "loss": 2.0522,
      "step": 24710
    },
    {
      "epoch": 2.9747292418772564,
      "grad_norm": 7.418403148651123,
      "learning_rate": 0.000199724206544979,
      "loss": 2.0363,
      "step": 24720
    },
    {
      "epoch": 2.9759326113116726,
      "grad_norm": 6.376554489135742,
      "learning_rate": 0.00019972392412876452,
      "loss": 1.9427,
      "step": 24730
    },
    {
      "epoch": 2.9771359807460893,
      "grad_norm": 3.917717933654785,
      "learning_rate": 0.00019972364156822483,
      "loss": 1.7764,
      "step": 24740
    },
    {
      "epoch": 2.9783393501805056,
      "grad_norm": 5.7326836585998535,
      "learning_rate": 0.00019972335886336043,
      "loss": 1.9515,
      "step": 24750
    },
    {
      "epoch": 2.979542719614922,
      "grad_norm": 3.9300084114074707,
      "learning_rate": 0.00019972307601417172,
      "loss": 1.9282,
      "step": 24760
    },
    {
      "epoch": 2.980746089049338,
      "grad_norm": 6.65960168838501,
      "learning_rate": 0.0001997227930206591,
      "loss": 1.9123,
      "step": 24770
    },
    {
      "epoch": 2.9819494584837543,
      "grad_norm": 5.051340579986572,
      "learning_rate": 0.00019972250988282297,
      "loss": 1.5143,
      "step": 24780
    },
    {
      "epoch": 2.983152827918171,
      "grad_norm": 3.710218906402588,
      "learning_rate": 0.00019972222660066377,
      "loss": 1.9307,
      "step": 24790
    },
    {
      "epoch": 2.984356197352587,
      "grad_norm": 5.473617076873779,
      "learning_rate": 0.00019972194317418186,
      "loss": 1.639,
      "step": 24800
    },
    {
      "epoch": 2.9855595667870034,
      "grad_norm": 3.464711904525757,
      "learning_rate": 0.00019972165960337768,
      "loss": 1.9858,
      "step": 24810
    },
    {
      "epoch": 2.98676293622142,
      "grad_norm": 7.262167453765869,
      "learning_rate": 0.00019972137588825163,
      "loss": 1.8256,
      "step": 24820
    },
    {
      "epoch": 2.9879663056558363,
      "grad_norm": 3.993345260620117,
      "learning_rate": 0.00019972109202880413,
      "loss": 1.4988,
      "step": 24830
    },
    {
      "epoch": 2.9891696750902526,
      "grad_norm": 3.6429708003997803,
      "learning_rate": 0.00019972080802503563,
      "loss": 1.8395,
      "step": 24840
    },
    {
      "epoch": 2.9903730445246692,
      "grad_norm": 6.208446979522705,
      "learning_rate": 0.00019972052387694645,
      "loss": 1.7543,
      "step": 24850
    },
    {
      "epoch": 2.9915764139590855,
      "grad_norm": 5.225845813751221,
      "learning_rate": 0.0001997202395845371,
      "loss": 1.827,
      "step": 24860
    },
    {
      "epoch": 2.9927797833935017,
      "grad_norm": 6.085515975952148,
      "learning_rate": 0.0001997199551478079,
      "loss": 1.921,
      "step": 24870
    },
    {
      "epoch": 2.9939831528279184,
      "grad_norm": 5.2673869132995605,
      "learning_rate": 0.00019971967056675932,
      "loss": 1.8041,
      "step": 24880
    },
    {
      "epoch": 2.9951865222623346,
      "grad_norm": 3.902085781097412,
      "learning_rate": 0.00019971938584139177,
      "loss": 2.168,
      "step": 24890
    },
    {
      "epoch": 2.996389891696751,
      "grad_norm": 5.707598686218262,
      "learning_rate": 0.00019971910097170564,
      "loss": 1.9218,
      "step": 24900
    },
    {
      "epoch": 2.9975932611311675,
      "grad_norm": 4.32947301864624,
      "learning_rate": 0.00019971881595770135,
      "loss": 1.9337,
      "step": 24910
    },
    {
      "epoch": 2.9987966305655838,
      "grad_norm": 6.908236503601074,
      "learning_rate": 0.0001997185307993793,
      "loss": 1.9814,
      "step": 24920
    },
    {
      "epoch": 3.0,
      "grad_norm": 6.15101957321167,
      "learning_rate": 0.00019971824549673992,
      "loss": 1.9834,
      "step": 24930
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.8243311643600464,
      "eval_runtime": 119.5201,
      "eval_samples_per_second": 61.806,
      "eval_steps_per_second": 7.731,
      "step": 24930
    },
    {
      "epoch": 3.0012033694344162,
      "grad_norm": 5.2833147048950195,
      "learning_rate": 0.00019971796004978365,
      "loss": 1.7781,
      "step": 24940
    },
    {
      "epoch": 3.002406738868833,
      "grad_norm": 3.5444066524505615,
      "learning_rate": 0.00019971767445851085,
      "loss": 1.8386,
      "step": 24950
    },
    {
      "epoch": 3.003610108303249,
      "grad_norm": 5.110971927642822,
      "learning_rate": 0.00019971738872292193,
      "loss": 1.843,
      "step": 24960
    },
    {
      "epoch": 3.0048134777376654,
      "grad_norm": 3.896223545074463,
      "learning_rate": 0.00019971710284301736,
      "loss": 1.579,
      "step": 24970
    },
    {
      "epoch": 3.0060168471720816,
      "grad_norm": 5.891197204589844,
      "learning_rate": 0.0001997168168187975,
      "loss": 1.912,
      "step": 24980
    },
    {
      "epoch": 3.0072202166064983,
      "grad_norm": 4.517705917358398,
      "learning_rate": 0.0001997165306502628,
      "loss": 1.5632,
      "step": 24990
    },
    {
      "epoch": 3.0084235860409145,
      "grad_norm": 3.676591157913208,
      "learning_rate": 0.0001997162443374137,
      "loss": 1.6171,
      "step": 25000
    },
    {
      "epoch": 3.0096269554753308,
      "grad_norm": 5.9294610023498535,
      "learning_rate": 0.00019971595788025054,
      "loss": 2.0759,
      "step": 25010
    },
    {
      "epoch": 3.0108303249097474,
      "grad_norm": 4.007317066192627,
      "learning_rate": 0.00019971567127877375,
      "loss": 1.6752,
      "step": 25020
    },
    {
      "epoch": 3.0120336943441637,
      "grad_norm": 6.3257646560668945,
      "learning_rate": 0.00019971538453298376,
      "loss": 1.6235,
      "step": 25030
    },
    {
      "epoch": 3.01323706377858,
      "grad_norm": 4.893363952636719,
      "learning_rate": 0.00019971509764288103,
      "loss": 1.7263,
      "step": 25040
    },
    {
      "epoch": 3.0144404332129966,
      "grad_norm": 3.142510414123535,
      "learning_rate": 0.00019971481060846594,
      "loss": 1.6694,
      "step": 25050
    },
    {
      "epoch": 3.015643802647413,
      "grad_norm": 5.823273658752441,
      "learning_rate": 0.00019971452342973888,
      "loss": 1.7035,
      "step": 25060
    },
    {
      "epoch": 3.016847172081829,
      "grad_norm": 3.8851661682128906,
      "learning_rate": 0.00019971423610670028,
      "loss": 1.6738,
      "step": 25070
    },
    {
      "epoch": 3.0180505415162453,
      "grad_norm": 6.197877407073975,
      "learning_rate": 0.00019971394863935055,
      "loss": 1.5583,
      "step": 25080
    },
    {
      "epoch": 3.019253910950662,
      "grad_norm": 5.340153217315674,
      "learning_rate": 0.00019971366102769012,
      "loss": 1.48,
      "step": 25090
    },
    {
      "epoch": 3.020457280385078,
      "grad_norm": 3.749476909637451,
      "learning_rate": 0.00019971337327171943,
      "loss": 1.5379,
      "step": 25100
    },
    {
      "epoch": 3.0216606498194944,
      "grad_norm": 4.925633430480957,
      "learning_rate": 0.00019971308537143886,
      "loss": 1.5849,
      "step": 25110
    },
    {
      "epoch": 3.022864019253911,
      "grad_norm": 4.040213108062744,
      "learning_rate": 0.00019971279732684883,
      "loss": 1.5022,
      "step": 25120
    },
    {
      "epoch": 3.0240673886883274,
      "grad_norm": 7.1975297927856445,
      "learning_rate": 0.0001997125091379498,
      "loss": 1.7892,
      "step": 25130
    },
    {
      "epoch": 3.0252707581227436,
      "grad_norm": 4.804437637329102,
      "learning_rate": 0.0001997122208047421,
      "loss": 1.6166,
      "step": 25140
    },
    {
      "epoch": 3.02647412755716,
      "grad_norm": 4.398497104644775,
      "learning_rate": 0.00019971193232722623,
      "loss": 1.7888,
      "step": 25150
    },
    {
      "epoch": 3.0276774969915765,
      "grad_norm": 6.2576375007629395,
      "learning_rate": 0.00019971164370540255,
      "loss": 1.6956,
      "step": 25160
    },
    {
      "epoch": 3.0288808664259927,
      "grad_norm": 4.663084030151367,
      "learning_rate": 0.00019971135493927154,
      "loss": 1.6047,
      "step": 25170
    },
    {
      "epoch": 3.030084235860409,
      "grad_norm": 6.627993106842041,
      "learning_rate": 0.00019971106602883354,
      "loss": 1.6708,
      "step": 25180
    },
    {
      "epoch": 3.0312876052948257,
      "grad_norm": 4.21818733215332,
      "learning_rate": 0.00019971077697408904,
      "loss": 1.6286,
      "step": 25190
    },
    {
      "epoch": 3.032490974729242,
      "grad_norm": 4.1206135749816895,
      "learning_rate": 0.00019971048777503843,
      "loss": 1.7903,
      "step": 25200
    },
    {
      "epoch": 3.033694344163658,
      "grad_norm": 6.09546422958374,
      "learning_rate": 0.0001997101984316821,
      "loss": 1.7904,
      "step": 25210
    },
    {
      "epoch": 3.034897713598075,
      "grad_norm": 4.578518390655518,
      "learning_rate": 0.0001997099089440205,
      "loss": 1.6353,
      "step": 25220
    },
    {
      "epoch": 3.036101083032491,
      "grad_norm": 6.658674240112305,
      "learning_rate": 0.00019970961931205404,
      "loss": 1.7366,
      "step": 25230
    },
    {
      "epoch": 3.0373044524669073,
      "grad_norm": 5.015432834625244,
      "learning_rate": 0.0001997093295357832,
      "loss": 1.6397,
      "step": 25240
    },
    {
      "epoch": 3.0385078219013235,
      "grad_norm": 4.643667697906494,
      "learning_rate": 0.00019970903961520828,
      "loss": 1.5767,
      "step": 25250
    },
    {
      "epoch": 3.03971119133574,
      "grad_norm": 6.313589572906494,
      "learning_rate": 0.00019970874955032978,
      "loss": 1.6188,
      "step": 25260
    },
    {
      "epoch": 3.0409145607701564,
      "grad_norm": 3.5082130432128906,
      "learning_rate": 0.0001997084593411481,
      "loss": 1.7944,
      "step": 25270
    },
    {
      "epoch": 3.0421179302045727,
      "grad_norm": 7.634500026702881,
      "learning_rate": 0.0001997081689876637,
      "loss": 1.7638,
      "step": 25280
    },
    {
      "epoch": 3.0433212996389893,
      "grad_norm": 5.156243324279785,
      "learning_rate": 0.00019970787848987688,
      "loss": 1.75,
      "step": 25290
    },
    {
      "epoch": 3.0445246690734056,
      "grad_norm": 3.355189800262451,
      "learning_rate": 0.00019970758784778822,
      "loss": 1.662,
      "step": 25300
    },
    {
      "epoch": 3.045728038507822,
      "grad_norm": 5.785003185272217,
      "learning_rate": 0.00019970729706139803,
      "loss": 1.5954,
      "step": 25310
    },
    {
      "epoch": 3.046931407942238,
      "grad_norm": 4.599120616912842,
      "learning_rate": 0.00019970700613070676,
      "loss": 1.6176,
      "step": 25320
    },
    {
      "epoch": 3.0481347773766547,
      "grad_norm": 6.337950706481934,
      "learning_rate": 0.00019970671505571485,
      "loss": 1.5627,
      "step": 25330
    },
    {
      "epoch": 3.049338146811071,
      "grad_norm": 5.127763271331787,
      "learning_rate": 0.0001997064238364227,
      "loss": 1.6074,
      "step": 25340
    },
    {
      "epoch": 3.050541516245487,
      "grad_norm": 3.8004400730133057,
      "learning_rate": 0.00019970613247283073,
      "loss": 1.6712,
      "step": 25350
    },
    {
      "epoch": 3.051744885679904,
      "grad_norm": 6.495005130767822,
      "learning_rate": 0.0001997058409649394,
      "loss": 1.6494,
      "step": 25360
    },
    {
      "epoch": 3.05294825511432,
      "grad_norm": 3.9298477172851562,
      "learning_rate": 0.00019970554931274907,
      "loss": 1.5972,
      "step": 25370
    },
    {
      "epoch": 3.0541516245487363,
      "grad_norm": 6.946953773498535,
      "learning_rate": 0.00019970525751626021,
      "loss": 1.7613,
      "step": 25380
    },
    {
      "epoch": 3.055354993983153,
      "grad_norm": 4.861685276031494,
      "learning_rate": 0.0001997049655754732,
      "loss": 1.7822,
      "step": 25390
    },
    {
      "epoch": 3.0565583634175693,
      "grad_norm": 3.3610851764678955,
      "learning_rate": 0.0001997046734903885,
      "loss": 1.6958,
      "step": 25400
    },
    {
      "epoch": 3.0577617328519855,
      "grad_norm": 4.182857990264893,
      "learning_rate": 0.00019970438126100655,
      "loss": 1.6831,
      "step": 25410
    },
    {
      "epoch": 3.0589651022864017,
      "grad_norm": 4.551860809326172,
      "learning_rate": 0.0001997040888873277,
      "loss": 1.7159,
      "step": 25420
    },
    {
      "epoch": 3.0601684717208184,
      "grad_norm": 7.149112224578857,
      "learning_rate": 0.00019970379636935246,
      "loss": 1.5726,
      "step": 25430
    },
    {
      "epoch": 3.0613718411552346,
      "grad_norm": 5.928719997406006,
      "learning_rate": 0.0001997035037070812,
      "loss": 1.7758,
      "step": 25440
    },
    {
      "epoch": 3.062575210589651,
      "grad_norm": 3.2937114238739014,
      "learning_rate": 0.00019970321090051433,
      "loss": 1.5158,
      "step": 25450
    },
    {
      "epoch": 3.0637785800240676,
      "grad_norm": 5.856424808502197,
      "learning_rate": 0.00019970291794965233,
      "loss": 1.6228,
      "step": 25460
    },
    {
      "epoch": 3.064981949458484,
      "grad_norm": 4.503580093383789,
      "learning_rate": 0.00019970262485449557,
      "loss": 1.7698,
      "step": 25470
    },
    {
      "epoch": 3.0661853188929,
      "grad_norm": 7.763538360595703,
      "learning_rate": 0.0001997023316150445,
      "loss": 1.7132,
      "step": 25480
    },
    {
      "epoch": 3.0673886883273163,
      "grad_norm": 4.79904317855835,
      "learning_rate": 0.00019970203823129953,
      "loss": 1.6942,
      "step": 25490
    },
    {
      "epoch": 3.068592057761733,
      "grad_norm": 3.115844488143921,
      "learning_rate": 0.0001997017447032611,
      "loss": 1.587,
      "step": 25500
    },
    {
      "epoch": 3.069795427196149,
      "grad_norm": 6.3336944580078125,
      "learning_rate": 0.00019970145103092966,
      "loss": 1.7281,
      "step": 25510
    },
    {
      "epoch": 3.0709987966305654,
      "grad_norm": 3.8646318912506104,
      "learning_rate": 0.00019970115721430558,
      "loss": 1.6836,
      "step": 25520
    },
    {
      "epoch": 3.072202166064982,
      "grad_norm": 6.718202114105225,
      "learning_rate": 0.0001997008632533893,
      "loss": 1.6268,
      "step": 25530
    },
    {
      "epoch": 3.0734055354993983,
      "grad_norm": 4.689975738525391,
      "learning_rate": 0.00019970056914818128,
      "loss": 1.7044,
      "step": 25540
    },
    {
      "epoch": 3.0746089049338146,
      "grad_norm": 4.25233793258667,
      "learning_rate": 0.00019970027489868192,
      "loss": 1.6907,
      "step": 25550
    },
    {
      "epoch": 3.0758122743682312,
      "grad_norm": 7.111978530883789,
      "learning_rate": 0.00019969998050489162,
      "loss": 1.7677,
      "step": 25560
    },
    {
      "epoch": 3.0770156438026475,
      "grad_norm": 4.082468509674072,
      "learning_rate": 0.00019969968596681085,
      "loss": 1.7532,
      "step": 25570
    },
    {
      "epoch": 3.0782190132370637,
      "grad_norm": 6.769084930419922,
      "learning_rate": 0.00019969939128444,
      "loss": 1.5383,
      "step": 25580
    },
    {
      "epoch": 3.07942238267148,
      "grad_norm": 5.465221881866455,
      "learning_rate": 0.00019969909645777952,
      "loss": 1.4176,
      "step": 25590
    },
    {
      "epoch": 3.0806257521058966,
      "grad_norm": 4.117949962615967,
      "learning_rate": 0.00019969880148682986,
      "loss": 1.8286,
      "step": 25600
    },
    {
      "epoch": 3.081829121540313,
      "grad_norm": 4.822063446044922,
      "learning_rate": 0.0001996985063715914,
      "loss": 1.7715,
      "step": 25610
    },
    {
      "epoch": 3.083032490974729,
      "grad_norm": 4.072781562805176,
      "learning_rate": 0.00019969821111206462,
      "loss": 1.8508,
      "step": 25620
    },
    {
      "epoch": 3.0842358604091458,
      "grad_norm": 7.279651165008545,
      "learning_rate": 0.00019969791570824988,
      "loss": 1.7172,
      "step": 25630
    },
    {
      "epoch": 3.085439229843562,
      "grad_norm": 4.747053146362305,
      "learning_rate": 0.00019969762016014764,
      "loss": 1.7487,
      "step": 25640
    },
    {
      "epoch": 3.0866425992779782,
      "grad_norm": 3.4259531497955322,
      "learning_rate": 0.00019969732446775833,
      "loss": 1.6637,
      "step": 25650
    },
    {
      "epoch": 3.087845968712395,
      "grad_norm": 6.953102111816406,
      "learning_rate": 0.00019969702863108238,
      "loss": 1.6367,
      "step": 25660
    },
    {
      "epoch": 3.089049338146811,
      "grad_norm": 4.110683917999268,
      "learning_rate": 0.00019969673265012023,
      "loss": 1.695,
      "step": 25670
    },
    {
      "epoch": 3.0902527075812274,
      "grad_norm": 7.62086820602417,
      "learning_rate": 0.00019969643652487226,
      "loss": 1.7798,
      "step": 25680
    },
    {
      "epoch": 3.0914560770156436,
      "grad_norm": 4.716604709625244,
      "learning_rate": 0.00019969614025533897,
      "loss": 1.8736,
      "step": 25690
    },
    {
      "epoch": 3.0926594464500603,
      "grad_norm": 3.269044876098633,
      "learning_rate": 0.00019969584384152073,
      "loss": 1.9083,
      "step": 25700
    },
    {
      "epoch": 3.0938628158844765,
      "grad_norm": 7.067758083343506,
      "learning_rate": 0.000199695547283418,
      "loss": 1.9578,
      "step": 25710
    },
    {
      "epoch": 3.0950661853188928,
      "grad_norm": 4.73967170715332,
      "learning_rate": 0.0001996952505810312,
      "loss": 1.7247,
      "step": 25720
    },
    {
      "epoch": 3.0962695547533094,
      "grad_norm": 7.015458106994629,
      "learning_rate": 0.00019969495373436074,
      "loss": 1.5704,
      "step": 25730
    },
    {
      "epoch": 3.0974729241877257,
      "grad_norm": 4.821101188659668,
      "learning_rate": 0.00019969465674340707,
      "loss": 1.8243,
      "step": 25740
    },
    {
      "epoch": 3.098676293622142,
      "grad_norm": 4.758875846862793,
      "learning_rate": 0.00019969435960817062,
      "loss": 1.6514,
      "step": 25750
    },
    {
      "epoch": 3.099879663056558,
      "grad_norm": 5.373219966888428,
      "learning_rate": 0.00019969406232865183,
      "loss": 1.6879,
      "step": 25760
    },
    {
      "epoch": 3.101083032490975,
      "grad_norm": 4.321028709411621,
      "learning_rate": 0.00019969376490485116,
      "loss": 1.6593,
      "step": 25770
    },
    {
      "epoch": 3.102286401925391,
      "grad_norm": 6.139832973480225,
      "learning_rate": 0.00019969346733676892,
      "loss": 1.8688,
      "step": 25780
    },
    {
      "epoch": 3.1034897713598073,
      "grad_norm": 5.178373336791992,
      "learning_rate": 0.00019969316962440564,
      "loss": 1.7524,
      "step": 25790
    },
    {
      "epoch": 3.104693140794224,
      "grad_norm": 4.018850326538086,
      "learning_rate": 0.00019969287176776175,
      "loss": 1.8681,
      "step": 25800
    },
    {
      "epoch": 3.10589651022864,
      "grad_norm": 6.357236862182617,
      "learning_rate": 0.00019969257376683766,
      "loss": 1.7481,
      "step": 25810
    },
    {
      "epoch": 3.1070998796630565,
      "grad_norm": 4.7395243644714355,
      "learning_rate": 0.0001996922756216338,
      "loss": 1.8221,
      "step": 25820
    },
    {
      "epoch": 3.108303249097473,
      "grad_norm": 7.194039821624756,
      "learning_rate": 0.0001996919773321506,
      "loss": 1.9351,
      "step": 25830
    },
    {
      "epoch": 3.1095066185318894,
      "grad_norm": 5.321618556976318,
      "learning_rate": 0.00019969167889838848,
      "loss": 1.7784,
      "step": 25840
    },
    {
      "epoch": 3.1107099879663056,
      "grad_norm": 3.5895071029663086,
      "learning_rate": 0.0001996913803203479,
      "loss": 1.5046,
      "step": 25850
    },
    {
      "epoch": 3.111913357400722,
      "grad_norm": 6.43202018737793,
      "learning_rate": 0.0001996910815980293,
      "loss": 1.573,
      "step": 25860
    },
    {
      "epoch": 3.1131167268351385,
      "grad_norm": 3.8133111000061035,
      "learning_rate": 0.00019969078273143306,
      "loss": 1.8075,
      "step": 25870
    },
    {
      "epoch": 3.1143200962695547,
      "grad_norm": 6.6105194091796875,
      "learning_rate": 0.00019969048372055968,
      "loss": 1.8469,
      "step": 25880
    },
    {
      "epoch": 3.115523465703971,
      "grad_norm": 5.352079391479492,
      "learning_rate": 0.0001996901845654095,
      "loss": 1.7894,
      "step": 25890
    },
    {
      "epoch": 3.1167268351383877,
      "grad_norm": 4.465076923370361,
      "learning_rate": 0.00019968988526598307,
      "loss": 2.003,
      "step": 25900
    },
    {
      "epoch": 3.117930204572804,
      "grad_norm": 5.748092174530029,
      "learning_rate": 0.00019968958582228073,
      "loss": 1.6668,
      "step": 25910
    },
    {
      "epoch": 3.11913357400722,
      "grad_norm": 4.764005184173584,
      "learning_rate": 0.00019968928623430296,
      "loss": 1.7606,
      "step": 25920
    },
    {
      "epoch": 3.1203369434416364,
      "grad_norm": 7.332487106323242,
      "learning_rate": 0.0001996889865020502,
      "loss": 1.5667,
      "step": 25930
    },
    {
      "epoch": 3.121540312876053,
      "grad_norm": 4.77926778793335,
      "learning_rate": 0.00019968868662552283,
      "loss": 1.8115,
      "step": 25940
    },
    {
      "epoch": 3.1227436823104693,
      "grad_norm": 3.5133652687072754,
      "learning_rate": 0.0001996883866047213,
      "loss": 1.8562,
      "step": 25950
    },
    {
      "epoch": 3.1239470517448855,
      "grad_norm": 6.457540035247803,
      "learning_rate": 0.0001996880864396461,
      "loss": 1.523,
      "step": 25960
    },
    {
      "epoch": 3.125150421179302,
      "grad_norm": 3.8357396125793457,
      "learning_rate": 0.00019968778613029762,
      "loss": 1.4882,
      "step": 25970
    },
    {
      "epoch": 3.1263537906137184,
      "grad_norm": 6.633978843688965,
      "learning_rate": 0.0001996874856766763,
      "loss": 1.632,
      "step": 25980
    },
    {
      "epoch": 3.1275571600481347,
      "grad_norm": 5.3305792808532715,
      "learning_rate": 0.00019968718507878256,
      "loss": 1.6133,
      "step": 25990
    },
    {
      "epoch": 3.1287605294825513,
      "grad_norm": 3.532611608505249,
      "learning_rate": 0.00019968688433661687,
      "loss": 1.5845,
      "step": 26000
    },
    {
      "epoch": 3.1299638989169676,
      "grad_norm": 6.723804473876953,
      "learning_rate": 0.0001996865834501797,
      "loss": 1.609,
      "step": 26010
    },
    {
      "epoch": 3.131167268351384,
      "grad_norm": 5.00460147857666,
      "learning_rate": 0.00019968628241947133,
      "loss": 1.6016,
      "step": 26020
    },
    {
      "epoch": 3.1323706377858,
      "grad_norm": 7.103140354156494,
      "learning_rate": 0.00019968598124449235,
      "loss": 1.9265,
      "step": 26030
    },
    {
      "epoch": 3.1335740072202167,
      "grad_norm": 5.142113208770752,
      "learning_rate": 0.00019968567992524315,
      "loss": 1.7629,
      "step": 26040
    },
    {
      "epoch": 3.134777376654633,
      "grad_norm": 4.044442653656006,
      "learning_rate": 0.00019968537846172413,
      "loss": 1.8743,
      "step": 26050
    },
    {
      "epoch": 3.135980746089049,
      "grad_norm": 4.6342949867248535,
      "learning_rate": 0.00019968507685393577,
      "loss": 1.7952,
      "step": 26060
    },
    {
      "epoch": 3.137184115523466,
      "grad_norm": 4.4024786949157715,
      "learning_rate": 0.00019968477510187848,
      "loss": 1.7952,
      "step": 26070
    },
    {
      "epoch": 3.138387484957882,
      "grad_norm": 6.709743976593018,
      "learning_rate": 0.00019968447320555275,
      "loss": 1.8065,
      "step": 26080
    },
    {
      "epoch": 3.1395908543922983,
      "grad_norm": 5.030803680419922,
      "learning_rate": 0.00019968417116495894,
      "loss": 1.5954,
      "step": 26090
    },
    {
      "epoch": 3.140794223826715,
      "grad_norm": 3.494459629058838,
      "learning_rate": 0.00019968386898009755,
      "loss": 1.7154,
      "step": 26100
    },
    {
      "epoch": 3.1419975932611313,
      "grad_norm": 5.076336860656738,
      "learning_rate": 0.00019968356665096896,
      "loss": 1.7257,
      "step": 26110
    },
    {
      "epoch": 3.1432009626955475,
      "grad_norm": 4.133296012878418,
      "learning_rate": 0.00019968326417757365,
      "loss": 1.5624,
      "step": 26120
    },
    {
      "epoch": 3.1444043321299637,
      "grad_norm": 6.244276523590088,
      "learning_rate": 0.00019968296155991203,
      "loss": 1.603,
      "step": 26130
    },
    {
      "epoch": 3.1456077015643804,
      "grad_norm": 4.406643390655518,
      "learning_rate": 0.00019968265879798458,
      "loss": 1.4633,
      "step": 26140
    },
    {
      "epoch": 3.1468110709987966,
      "grad_norm": 3.8273041248321533,
      "learning_rate": 0.0001996823558917917,
      "loss": 1.8377,
      "step": 26150
    },
    {
      "epoch": 3.148014440433213,
      "grad_norm": 6.194026470184326,
      "learning_rate": 0.00019968205284133386,
      "loss": 1.5792,
      "step": 26160
    },
    {
      "epoch": 3.1492178098676296,
      "grad_norm": 4.927667617797852,
      "learning_rate": 0.00019968174964661147,
      "loss": 1.6037,
      "step": 26170
    },
    {
      "epoch": 3.150421179302046,
      "grad_norm": 7.056860446929932,
      "learning_rate": 0.00019968144630762495,
      "loss": 1.9054,
      "step": 26180
    },
    {
      "epoch": 3.151624548736462,
      "grad_norm": 4.70307731628418,
      "learning_rate": 0.00019968114282437478,
      "loss": 1.5009,
      "step": 26190
    },
    {
      "epoch": 3.1528279181708783,
      "grad_norm": 3.752289295196533,
      "learning_rate": 0.00019968083919686137,
      "loss": 1.8344,
      "step": 26200
    },
    {
      "epoch": 3.154031287605295,
      "grad_norm": 5.970653057098389,
      "learning_rate": 0.00019968053542508522,
      "loss": 1.7478,
      "step": 26210
    },
    {
      "epoch": 3.155234657039711,
      "grad_norm": 4.398066520690918,
      "learning_rate": 0.0001996802315090467,
      "loss": 1.8283,
      "step": 26220
    },
    {
      "epoch": 3.1564380264741274,
      "grad_norm": 7.221063137054443,
      "learning_rate": 0.0001996799274487463,
      "loss": 1.7113,
      "step": 26230
    },
    {
      "epoch": 3.157641395908544,
      "grad_norm": 4.993231773376465,
      "learning_rate": 0.0001996796232441844,
      "loss": 1.8439,
      "step": 26240
    },
    {
      "epoch": 3.1588447653429603,
      "grad_norm": 4.482081413269043,
      "learning_rate": 0.00019967931889536148,
      "loss": 1.7393,
      "step": 26250
    },
    {
      "epoch": 3.1600481347773766,
      "grad_norm": 6.5946221351623535,
      "learning_rate": 0.00019967901440227798,
      "loss": 1.7978,
      "step": 26260
    },
    {
      "epoch": 3.1612515042117932,
      "grad_norm": 5.204263687133789,
      "learning_rate": 0.00019967870976493433,
      "loss": 1.6644,
      "step": 26270
    },
    {
      "epoch": 3.1624548736462095,
      "grad_norm": 6.908845901489258,
      "learning_rate": 0.000199678404983331,
      "loss": 1.8345,
      "step": 26280
    },
    {
      "epoch": 3.1636582430806257,
      "grad_norm": 5.161684989929199,
      "learning_rate": 0.0001996781000574684,
      "loss": 1.5343,
      "step": 26290
    },
    {
      "epoch": 3.164861612515042,
      "grad_norm": 3.900618553161621,
      "learning_rate": 0.00019967779498734697,
      "loss": 1.7535,
      "step": 26300
    },
    {
      "epoch": 3.1660649819494586,
      "grad_norm": 5.767590522766113,
      "learning_rate": 0.00019967748977296713,
      "loss": 1.6541,
      "step": 26310
    },
    {
      "epoch": 3.167268351383875,
      "grad_norm": 3.6571078300476074,
      "learning_rate": 0.0001996771844143294,
      "loss": 1.6133,
      "step": 26320
    },
    {
      "epoch": 3.168471720818291,
      "grad_norm": 7.802633285522461,
      "learning_rate": 0.00019967687891143417,
      "loss": 1.5927,
      "step": 26330
    },
    {
      "epoch": 3.1696750902527078,
      "grad_norm": 5.249373435974121,
      "learning_rate": 0.0001996765732642819,
      "loss": 1.5744,
      "step": 26340
    },
    {
      "epoch": 3.170878459687124,
      "grad_norm": 4.153386116027832,
      "learning_rate": 0.00019967626747287298,
      "loss": 1.8146,
      "step": 26350
    },
    {
      "epoch": 3.1720818291215402,
      "grad_norm": 6.111057758331299,
      "learning_rate": 0.0001996759615372079,
      "loss": 1.7322,
      "step": 26360
    },
    {
      "epoch": 3.1732851985559565,
      "grad_norm": 4.793424129486084,
      "learning_rate": 0.0001996756554572871,
      "loss": 1.8235,
      "step": 26370
    },
    {
      "epoch": 3.174488567990373,
      "grad_norm": 6.269195556640625,
      "learning_rate": 0.00019967534923311103,
      "loss": 1.8598,
      "step": 26380
    },
    {
      "epoch": 3.1756919374247894,
      "grad_norm": 4.551454544067383,
      "learning_rate": 0.00019967504286468008,
      "loss": 1.7414,
      "step": 26390
    },
    {
      "epoch": 3.1768953068592056,
      "grad_norm": 2.942845106124878,
      "learning_rate": 0.00019967473635199479,
      "loss": 1.8039,
      "step": 26400
    },
    {
      "epoch": 3.1780986762936223,
      "grad_norm": 5.321710109710693,
      "learning_rate": 0.00019967442969505552,
      "loss": 1.7472,
      "step": 26410
    },
    {
      "epoch": 3.1793020457280385,
      "grad_norm": 4.249934673309326,
      "learning_rate": 0.00019967412289386273,
      "loss": 1.8743,
      "step": 26420
    },
    {
      "epoch": 3.1805054151624548,
      "grad_norm": 6.844578266143799,
      "learning_rate": 0.0001996738159484169,
      "loss": 1.8235,
      "step": 26430
    },
    {
      "epoch": 3.1817087845968715,
      "grad_norm": 4.887540340423584,
      "learning_rate": 0.00019967350885871844,
      "loss": 1.886,
      "step": 26440
    },
    {
      "epoch": 3.1829121540312877,
      "grad_norm": 4.189870357513428,
      "learning_rate": 0.0001996732016247678,
      "loss": 1.8164,
      "step": 26450
    },
    {
      "epoch": 3.184115523465704,
      "grad_norm": 5.997005939483643,
      "learning_rate": 0.00019967289424656544,
      "loss": 1.7669,
      "step": 26460
    },
    {
      "epoch": 3.18531889290012,
      "grad_norm": 5.44852876663208,
      "learning_rate": 0.00019967258672411178,
      "loss": 1.6729,
      "step": 26470
    },
    {
      "epoch": 3.186522262334537,
      "grad_norm": 7.447607040405273,
      "learning_rate": 0.00019967227905740728,
      "loss": 1.8408,
      "step": 26480
    },
    {
      "epoch": 3.187725631768953,
      "grad_norm": 5.315221309661865,
      "learning_rate": 0.0001996719712464524,
      "loss": 1.6769,
      "step": 26490
    },
    {
      "epoch": 3.1889290012033693,
      "grad_norm": 5.13824987411499,
      "learning_rate": 0.00019967166329124756,
      "loss": 1.5989,
      "step": 26500
    },
    {
      "epoch": 3.190132370637786,
      "grad_norm": 6.396768569946289,
      "learning_rate": 0.00019967135519179322,
      "loss": 1.5882,
      "step": 26510
    },
    {
      "epoch": 3.191335740072202,
      "grad_norm": 4.440583229064941,
      "learning_rate": 0.0001996710469480898,
      "loss": 1.657,
      "step": 26520
    },
    {
      "epoch": 3.1925391095066185,
      "grad_norm": 6.554316997528076,
      "learning_rate": 0.0001996707385601378,
      "loss": 1.9095,
      "step": 26530
    },
    {
      "epoch": 3.1937424789410347,
      "grad_norm": 4.493051528930664,
      "learning_rate": 0.0001996704300279376,
      "loss": 1.7128,
      "step": 26540
    },
    {
      "epoch": 3.1949458483754514,
      "grad_norm": 4.213508129119873,
      "learning_rate": 0.0001996701213514897,
      "loss": 1.7449,
      "step": 26550
    },
    {
      "epoch": 3.1961492178098676,
      "grad_norm": 6.3384833335876465,
      "learning_rate": 0.0001996698125307945,
      "loss": 1.5784,
      "step": 26560
    },
    {
      "epoch": 3.197352587244284,
      "grad_norm": 4.621504783630371,
      "learning_rate": 0.0001996695035658525,
      "loss": 1.5697,
      "step": 26570
    },
    {
      "epoch": 3.1985559566787005,
      "grad_norm": 6.80503511428833,
      "learning_rate": 0.00019966919445666412,
      "loss": 1.877,
      "step": 26580
    },
    {
      "epoch": 3.1997593261131168,
      "grad_norm": 5.692870616912842,
      "learning_rate": 0.0001996688852032298,
      "loss": 1.8058,
      "step": 26590
    },
    {
      "epoch": 3.200962695547533,
      "grad_norm": 3.311182737350464,
      "learning_rate": 0.00019966857580555,
      "loss": 1.8477,
      "step": 26600
    },
    {
      "epoch": 3.2021660649819497,
      "grad_norm": 5.926977634429932,
      "learning_rate": 0.00019966826626362515,
      "loss": 1.6738,
      "step": 26610
    },
    {
      "epoch": 3.203369434416366,
      "grad_norm": 3.704402208328247,
      "learning_rate": 0.00019966795657745573,
      "loss": 1.6778,
      "step": 26620
    },
    {
      "epoch": 3.204572803850782,
      "grad_norm": 8.318079948425293,
      "learning_rate": 0.00019966764674704215,
      "loss": 1.9835,
      "step": 26630
    },
    {
      "epoch": 3.2057761732851984,
      "grad_norm": 5.058340549468994,
      "learning_rate": 0.0001996673367723849,
      "loss": 1.6039,
      "step": 26640
    },
    {
      "epoch": 3.206979542719615,
      "grad_norm": 4.320777893066406,
      "learning_rate": 0.0001996670266534844,
      "loss": 1.7686,
      "step": 26650
    },
    {
      "epoch": 3.2081829121540313,
      "grad_norm": 5.931827068328857,
      "learning_rate": 0.0001996667163903411,
      "loss": 1.8661,
      "step": 26660
    },
    {
      "epoch": 3.2093862815884475,
      "grad_norm": 4.6182050704956055,
      "learning_rate": 0.00019966640598295546,
      "loss": 1.8591,
      "step": 26670
    },
    {
      "epoch": 3.210589651022864,
      "grad_norm": 7.831111431121826,
      "learning_rate": 0.00019966609543132792,
      "loss": 1.816,
      "step": 26680
    },
    {
      "epoch": 3.2117930204572804,
      "grad_norm": 5.658038139343262,
      "learning_rate": 0.00019966578473545893,
      "loss": 1.5925,
      "step": 26690
    },
    {
      "epoch": 3.2129963898916967,
      "grad_norm": 3.685478448867798,
      "learning_rate": 0.00019966547389534895,
      "loss": 1.739,
      "step": 26700
    },
    {
      "epoch": 3.214199759326113,
      "grad_norm": 5.0785956382751465,
      "learning_rate": 0.0001996651629109984,
      "loss": 1.5135,
      "step": 26710
    },
    {
      "epoch": 3.2154031287605296,
      "grad_norm": 4.372645854949951,
      "learning_rate": 0.0001996648517824078,
      "loss": 1.7063,
      "step": 26720
    },
    {
      "epoch": 3.216606498194946,
      "grad_norm": 7.1049628257751465,
      "learning_rate": 0.00019966454050957753,
      "loss": 1.5647,
      "step": 26730
    },
    {
      "epoch": 3.217809867629362,
      "grad_norm": 4.710076808929443,
      "learning_rate": 0.00019966422909250807,
      "loss": 1.6602,
      "step": 26740
    },
    {
      "epoch": 3.2190132370637787,
      "grad_norm": 4.299315929412842,
      "learning_rate": 0.00019966391753119985,
      "loss": 1.7935,
      "step": 26750
    },
    {
      "epoch": 3.220216606498195,
      "grad_norm": 7.561284065246582,
      "learning_rate": 0.0001996636058256533,
      "loss": 1.6962,
      "step": 26760
    },
    {
      "epoch": 3.221419975932611,
      "grad_norm": 4.574403285980225,
      "learning_rate": 0.00019966329397586895,
      "loss": 1.925,
      "step": 26770
    },
    {
      "epoch": 3.222623345367028,
      "grad_norm": 7.978187561035156,
      "learning_rate": 0.00019966298198184724,
      "loss": 1.7282,
      "step": 26780
    },
    {
      "epoch": 3.223826714801444,
      "grad_norm": 4.1126627922058105,
      "learning_rate": 0.00019966266984358853,
      "loss": 1.4681,
      "step": 26790
    },
    {
      "epoch": 3.2250300842358604,
      "grad_norm": 4.2365617752075195,
      "learning_rate": 0.00019966235756109334,
      "loss": 1.6335,
      "step": 26800
    },
    {
      "epoch": 3.2262334536702766,
      "grad_norm": 5.861513137817383,
      "learning_rate": 0.00019966204513436212,
      "loss": 1.7006,
      "step": 26810
    },
    {
      "epoch": 3.2274368231046933,
      "grad_norm": 4.411993980407715,
      "learning_rate": 0.00019966173256339532,
      "loss": 1.8046,
      "step": 26820
    },
    {
      "epoch": 3.2286401925391095,
      "grad_norm": 8.419879913330078,
      "learning_rate": 0.0001996614198481934,
      "loss": 1.833,
      "step": 26830
    },
    {
      "epoch": 3.2298435619735257,
      "grad_norm": 4.881294250488281,
      "learning_rate": 0.0001996611069887568,
      "loss": 1.774,
      "step": 26840
    },
    {
      "epoch": 3.2310469314079424,
      "grad_norm": 3.6665778160095215,
      "learning_rate": 0.00019966079398508595,
      "loss": 1.7573,
      "step": 26850
    },
    {
      "epoch": 3.2322503008423586,
      "grad_norm": 6.3349528312683105,
      "learning_rate": 0.00019966048083718132,
      "loss": 1.8857,
      "step": 26860
    },
    {
      "epoch": 3.233453670276775,
      "grad_norm": 4.5454511642456055,
      "learning_rate": 0.0001996601675450434,
      "loss": 1.8435,
      "step": 26870
    },
    {
      "epoch": 3.234657039711191,
      "grad_norm": 7.010152339935303,
      "learning_rate": 0.00019965985410867258,
      "loss": 1.7535,
      "step": 26880
    },
    {
      "epoch": 3.235860409145608,
      "grad_norm": 5.6580071449279785,
      "learning_rate": 0.00019965954052806936,
      "loss": 1.6308,
      "step": 26890
    },
    {
      "epoch": 3.237063778580024,
      "grad_norm": 3.9693939685821533,
      "learning_rate": 0.0001996592268032342,
      "loss": 1.9586,
      "step": 26900
    },
    {
      "epoch": 3.2382671480144403,
      "grad_norm": 5.903779029846191,
      "learning_rate": 0.00019965891293416752,
      "loss": 1.8053,
      "step": 26910
    },
    {
      "epoch": 3.239470517448857,
      "grad_norm": 4.855864524841309,
      "learning_rate": 0.00019965859892086979,
      "loss": 1.8226,
      "step": 26920
    },
    {
      "epoch": 3.240673886883273,
      "grad_norm": 6.977510452270508,
      "learning_rate": 0.00019965828476334143,
      "loss": 1.666,
      "step": 26930
    },
    {
      "epoch": 3.2418772563176894,
      "grad_norm": 5.6782612800598145,
      "learning_rate": 0.00019965797046158298,
      "loss": 1.7855,
      "step": 26940
    },
    {
      "epoch": 3.243080625752106,
      "grad_norm": 3.3649098873138428,
      "learning_rate": 0.0001996576560155948,
      "loss": 1.8999,
      "step": 26950
    },
    {
      "epoch": 3.2442839951865223,
      "grad_norm": 6.129069805145264,
      "learning_rate": 0.00019965734142537742,
      "loss": 1.8824,
      "step": 26960
    },
    {
      "epoch": 3.2454873646209386,
      "grad_norm": 4.5676984786987305,
      "learning_rate": 0.00019965702669093123,
      "loss": 1.7751,
      "step": 26970
    },
    {
      "epoch": 3.246690734055355,
      "grad_norm": 6.600085735321045,
      "learning_rate": 0.00019965671181225675,
      "loss": 1.8884,
      "step": 26980
    },
    {
      "epoch": 3.2478941034897715,
      "grad_norm": 5.7984395027160645,
      "learning_rate": 0.00019965639678935437,
      "loss": 1.6576,
      "step": 26990
    },
    {
      "epoch": 3.2490974729241877,
      "grad_norm": 3.6901161670684814,
      "learning_rate": 0.0001996560816222246,
      "loss": 1.9137,
      "step": 27000
    },
    {
      "epoch": 3.250300842358604,
      "grad_norm": 6.228165149688721,
      "learning_rate": 0.00019965576631086787,
      "loss": 1.8833,
      "step": 27010
    },
    {
      "epoch": 3.2515042117930206,
      "grad_norm": 3.7460849285125732,
      "learning_rate": 0.00019965545085528464,
      "loss": 1.6934,
      "step": 27020
    },
    {
      "epoch": 3.252707581227437,
      "grad_norm": 6.766097068786621,
      "learning_rate": 0.00019965513525547536,
      "loss": 1.8627,
      "step": 27030
    },
    {
      "epoch": 3.253910950661853,
      "grad_norm": 5.582569599151611,
      "learning_rate": 0.00019965481951144052,
      "loss": 1.7107,
      "step": 27040
    },
    {
      "epoch": 3.2551143200962693,
      "grad_norm": 3.7459499835968018,
      "learning_rate": 0.00019965450362318055,
      "loss": 1.9077,
      "step": 27050
    },
    {
      "epoch": 3.256317689530686,
      "grad_norm": 5.6414031982421875,
      "learning_rate": 0.00019965418759069586,
      "loss": 1.6465,
      "step": 27060
    },
    {
      "epoch": 3.2575210589651022,
      "grad_norm": 4.878627777099609,
      "learning_rate": 0.00019965387141398702,
      "loss": 1.7643,
      "step": 27070
    },
    {
      "epoch": 3.2587244283995185,
      "grad_norm": 6.926384449005127,
      "learning_rate": 0.00019965355509305442,
      "loss": 1.925,
      "step": 27080
    },
    {
      "epoch": 3.259927797833935,
      "grad_norm": 5.348418712615967,
      "learning_rate": 0.00019965323862789847,
      "loss": 1.55,
      "step": 27090
    },
    {
      "epoch": 3.2611311672683514,
      "grad_norm": 4.101518154144287,
      "learning_rate": 0.0001996529220185197,
      "loss": 1.6806,
      "step": 27100
    },
    {
      "epoch": 3.2623345367027676,
      "grad_norm": 6.153325080871582,
      "learning_rate": 0.00019965260526491857,
      "loss": 1.8083,
      "step": 27110
    },
    {
      "epoch": 3.2635379061371843,
      "grad_norm": 4.059078216552734,
      "learning_rate": 0.00019965228836709553,
      "loss": 1.6787,
      "step": 27120
    },
    {
      "epoch": 3.2647412755716005,
      "grad_norm": 6.101953983306885,
      "learning_rate": 0.000199651971325051,
      "loss": 1.6913,
      "step": 27130
    },
    {
      "epoch": 3.2659446450060168,
      "grad_norm": 4.951807498931885,
      "learning_rate": 0.00019965165413878547,
      "loss": 1.5629,
      "step": 27140
    },
    {
      "epoch": 3.2671480144404335,
      "grad_norm": 4.110108375549316,
      "learning_rate": 0.00019965133680829937,
      "loss": 1.9686,
      "step": 27150
    },
    {
      "epoch": 3.2683513838748497,
      "grad_norm": 6.574784278869629,
      "learning_rate": 0.0001996510193335932,
      "loss": 1.7583,
      "step": 27160
    },
    {
      "epoch": 3.269554753309266,
      "grad_norm": 4.252917766571045,
      "learning_rate": 0.0001996507017146674,
      "loss": 1.6502,
      "step": 27170
    },
    {
      "epoch": 3.270758122743682,
      "grad_norm": 6.667531490325928,
      "learning_rate": 0.00019965038395152246,
      "loss": 1.7054,
      "step": 27180
    },
    {
      "epoch": 3.271961492178099,
      "grad_norm": 5.201156139373779,
      "learning_rate": 0.00019965006604415878,
      "loss": 1.6824,
      "step": 27190
    },
    {
      "epoch": 3.273164861612515,
      "grad_norm": 3.730102062225342,
      "learning_rate": 0.00019964974799257685,
      "loss": 1.6422,
      "step": 27200
    },
    {
      "epoch": 3.2743682310469313,
      "grad_norm": 5.933589935302734,
      "learning_rate": 0.00019964942979677715,
      "loss": 1.6453,
      "step": 27210
    },
    {
      "epoch": 3.2755716004813475,
      "grad_norm": 4.717698097229004,
      "learning_rate": 0.00019964911145676012,
      "loss": 1.5944,
      "step": 27220
    },
    {
      "epoch": 3.2767749699157642,
      "grad_norm": 6.802528381347656,
      "learning_rate": 0.00019964879297252618,
      "loss": 2.118,
      "step": 27230
    },
    {
      "epoch": 3.2779783393501805,
      "grad_norm": 4.5299506187438965,
      "learning_rate": 0.0001996484743440759,
      "loss": 1.9077,
      "step": 27240
    },
    {
      "epoch": 3.2791817087845967,
      "grad_norm": 4.457561016082764,
      "learning_rate": 0.00019964815557140965,
      "loss": 1.813,
      "step": 27250
    },
    {
      "epoch": 3.2803850782190134,
      "grad_norm": 6.079787731170654,
      "learning_rate": 0.0001996478366545279,
      "loss": 1.7428,
      "step": 27260
    },
    {
      "epoch": 3.2815884476534296,
      "grad_norm": 4.322761535644531,
      "learning_rate": 0.00019964751759343115,
      "loss": 1.5646,
      "step": 27270
    },
    {
      "epoch": 3.282791817087846,
      "grad_norm": 6.566214084625244,
      "learning_rate": 0.00019964719838811982,
      "loss": 1.8035,
      "step": 27280
    },
    {
      "epoch": 3.2839951865222625,
      "grad_norm": 5.867137908935547,
      "learning_rate": 0.0001996468790385944,
      "loss": 1.7421,
      "step": 27290
    },
    {
      "epoch": 3.2851985559566788,
      "grad_norm": 3.765441417694092,
      "learning_rate": 0.00019964655954485535,
      "loss": 1.8022,
      "step": 27300
    },
    {
      "epoch": 3.286401925391095,
      "grad_norm": 6.182241916656494,
      "learning_rate": 0.00019964623990690312,
      "loss": 1.7031,
      "step": 27310
    },
    {
      "epoch": 3.2876052948255117,
      "grad_norm": 4.88283634185791,
      "learning_rate": 0.00019964592012473818,
      "loss": 1.5069,
      "step": 27320
    },
    {
      "epoch": 3.288808664259928,
      "grad_norm": 6.951658725738525,
      "learning_rate": 0.000199645600198361,
      "loss": 1.9607,
      "step": 27330
    },
    {
      "epoch": 3.290012033694344,
      "grad_norm": 5.33024787902832,
      "learning_rate": 0.00019964528012777203,
      "loss": 1.8222,
      "step": 27340
    },
    {
      "epoch": 3.2912154031287604,
      "grad_norm": 4.643368244171143,
      "learning_rate": 0.00019964495991297172,
      "loss": 1.6245,
      "step": 27350
    },
    {
      "epoch": 3.292418772563177,
      "grad_norm": 5.759073257446289,
      "learning_rate": 0.0001996446395539606,
      "loss": 1.7253,
      "step": 27360
    },
    {
      "epoch": 3.2936221419975933,
      "grad_norm": 4.402641773223877,
      "learning_rate": 0.00019964431905073905,
      "loss": 1.7703,
      "step": 27370
    },
    {
      "epoch": 3.2948255114320095,
      "grad_norm": 5.987547874450684,
      "learning_rate": 0.00019964399840330756,
      "loss": 1.8539,
      "step": 27380
    },
    {
      "epoch": 3.2960288808664258,
      "grad_norm": 5.379052639007568,
      "learning_rate": 0.00019964367761166662,
      "loss": 1.7614,
      "step": 27390
    },
    {
      "epoch": 3.2972322503008424,
      "grad_norm": 3.9488117694854736,
      "learning_rate": 0.00019964335667581668,
      "loss": 1.6899,
      "step": 27400
    },
    {
      "epoch": 3.2984356197352587,
      "grad_norm": 5.289420127868652,
      "learning_rate": 0.0001996430355957582,
      "loss": 1.6323,
      "step": 27410
    },
    {
      "epoch": 3.299638989169675,
      "grad_norm": 4.561107635498047,
      "learning_rate": 0.00019964271437149163,
      "loss": 1.605,
      "step": 27420
    },
    {
      "epoch": 3.3008423586040916,
      "grad_norm": 7.212870121002197,
      "learning_rate": 0.00019964239300301743,
      "loss": 1.6594,
      "step": 27430
    },
    {
      "epoch": 3.302045728038508,
      "grad_norm": 5.4261040687561035,
      "learning_rate": 0.00019964207149033612,
      "loss": 1.562,
      "step": 27440
    },
    {
      "epoch": 3.303249097472924,
      "grad_norm": 3.1123127937316895,
      "learning_rate": 0.00019964174983344815,
      "loss": 1.8834,
      "step": 27450
    },
    {
      "epoch": 3.3044524669073407,
      "grad_norm": 6.524343013763428,
      "learning_rate": 0.00019964142803235394,
      "loss": 1.6465,
      "step": 27460
    },
    {
      "epoch": 3.305655836341757,
      "grad_norm": 4.955365180969238,
      "learning_rate": 0.00019964110608705399,
      "loss": 1.7944,
      "step": 27470
    },
    {
      "epoch": 3.306859205776173,
      "grad_norm": 7.838707447052002,
      "learning_rate": 0.00019964078399754875,
      "loss": 1.9876,
      "step": 27480
    },
    {
      "epoch": 3.30806257521059,
      "grad_norm": 4.1487555503845215,
      "learning_rate": 0.00019964046176383873,
      "loss": 1.6249,
      "step": 27490
    },
    {
      "epoch": 3.309265944645006,
      "grad_norm": 3.6820261478424072,
      "learning_rate": 0.00019964013938592433,
      "loss": 1.773,
      "step": 27500
    },
    {
      "epoch": 3.3104693140794224,
      "grad_norm": 4.7243194580078125,
      "learning_rate": 0.00019963981686380605,
      "loss": 1.6834,
      "step": 27510
    },
    {
      "epoch": 3.3116726835138386,
      "grad_norm": 4.824052333831787,
      "learning_rate": 0.00019963949419748436,
      "loss": 1.7135,
      "step": 27520
    },
    {
      "epoch": 3.3128760529482553,
      "grad_norm": 7.697936058044434,
      "learning_rate": 0.0001996391713869597,
      "loss": 1.7224,
      "step": 27530
    },
    {
      "epoch": 3.3140794223826715,
      "grad_norm": 6.1690850257873535,
      "learning_rate": 0.0001996388484322326,
      "loss": 1.728,
      "step": 27540
    },
    {
      "epoch": 3.3152827918170877,
      "grad_norm": 4.328515529632568,
      "learning_rate": 0.00019963852533330344,
      "loss": 1.8722,
      "step": 27550
    },
    {
      "epoch": 3.316486161251504,
      "grad_norm": 6.749997615814209,
      "learning_rate": 0.00019963820209017278,
      "loss": 1.7789,
      "step": 27560
    },
    {
      "epoch": 3.3176895306859207,
      "grad_norm": 5.340113162994385,
      "learning_rate": 0.000199637878702841,
      "loss": 1.9433,
      "step": 27570
    },
    {
      "epoch": 3.318892900120337,
      "grad_norm": 7.050694465637207,
      "learning_rate": 0.00019963755517130865,
      "loss": 1.7184,
      "step": 27580
    },
    {
      "epoch": 3.320096269554753,
      "grad_norm": 4.519673824310303,
      "learning_rate": 0.0001996372314955761,
      "loss": 1.8321,
      "step": 27590
    },
    {
      "epoch": 3.32129963898917,
      "grad_norm": 3.892613649368286,
      "learning_rate": 0.00019963690767564393,
      "loss": 1.87,
      "step": 27600
    },
    {
      "epoch": 3.322503008423586,
      "grad_norm": 5.246593952178955,
      "learning_rate": 0.00019963658371151255,
      "loss": 1.6093,
      "step": 27610
    },
    {
      "epoch": 3.3237063778580023,
      "grad_norm": 4.326037883758545,
      "learning_rate": 0.00019963625960318243,
      "loss": 1.6772,
      "step": 27620
    },
    {
      "epoch": 3.324909747292419,
      "grad_norm": 6.822893142700195,
      "learning_rate": 0.000199635935350654,
      "loss": 1.9091,
      "step": 27630
    },
    {
      "epoch": 3.326113116726835,
      "grad_norm": 5.5352253913879395,
      "learning_rate": 0.0001996356109539278,
      "loss": 1.5584,
      "step": 27640
    },
    {
      "epoch": 3.3273164861612514,
      "grad_norm": 4.748112201690674,
      "learning_rate": 0.0001996352864130043,
      "loss": 1.7727,
      "step": 27650
    },
    {
      "epoch": 3.328519855595668,
      "grad_norm": 6.762351036071777,
      "learning_rate": 0.00019963496172788393,
      "loss": 1.8328,
      "step": 27660
    },
    {
      "epoch": 3.3297232250300843,
      "grad_norm": 3.778073787689209,
      "learning_rate": 0.0001996346368985672,
      "loss": 1.7227,
      "step": 27670
    },
    {
      "epoch": 3.3309265944645006,
      "grad_norm": 6.242730617523193,
      "learning_rate": 0.0001996343119250545,
      "loss": 1.7442,
      "step": 27680
    },
    {
      "epoch": 3.332129963898917,
      "grad_norm": 5.287921905517578,
      "learning_rate": 0.00019963398680734634,
      "loss": 1.7142,
      "step": 27690
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 4.432364463806152,
      "learning_rate": 0.00019963366154544324,
      "loss": 1.7966,
      "step": 27700
    },
    {
      "epoch": 3.3345367027677497,
      "grad_norm": 5.745725154876709,
      "learning_rate": 0.0001996333361393456,
      "loss": 1.7246,
      "step": 27710
    },
    {
      "epoch": 3.335740072202166,
      "grad_norm": 4.296098232269287,
      "learning_rate": 0.00019963301058905397,
      "loss": 1.6841,
      "step": 27720
    },
    {
      "epoch": 3.336943441636582,
      "grad_norm": 7.974771976470947,
      "learning_rate": 0.00019963268489456875,
      "loss": 1.723,
      "step": 27730
    },
    {
      "epoch": 3.338146811070999,
      "grad_norm": 4.985914707183838,
      "learning_rate": 0.00019963235905589045,
      "loss": 1.7322,
      "step": 27740
    },
    {
      "epoch": 3.339350180505415,
      "grad_norm": 4.049668788909912,
      "learning_rate": 0.0001996320330730195,
      "loss": 1.881,
      "step": 27750
    },
    {
      "epoch": 3.3405535499398313,
      "grad_norm": 6.071661472320557,
      "learning_rate": 0.00019963170694595642,
      "loss": 1.9796,
      "step": 27760
    },
    {
      "epoch": 3.341756919374248,
      "grad_norm": 4.73226261138916,
      "learning_rate": 0.00019963138067470167,
      "loss": 1.898,
      "step": 27770
    },
    {
      "epoch": 3.3429602888086642,
      "grad_norm": 6.587105751037598,
      "learning_rate": 0.00019963105425925567,
      "loss": 1.8603,
      "step": 27780
    },
    {
      "epoch": 3.3441636582430805,
      "grad_norm": 5.582319259643555,
      "learning_rate": 0.000199630727699619,
      "loss": 1.832,
      "step": 27790
    },
    {
      "epoch": 3.345367027677497,
      "grad_norm": 4.537518501281738,
      "learning_rate": 0.00019963040099579203,
      "loss": 1.7574,
      "step": 27800
    },
    {
      "epoch": 3.3465703971119134,
      "grad_norm": 6.402120113372803,
      "learning_rate": 0.00019963007414777525,
      "loss": 1.8413,
      "step": 27810
    },
    {
      "epoch": 3.3477737665463296,
      "grad_norm": 4.006775379180908,
      "learning_rate": 0.00019962974715556918,
      "loss": 1.8227,
      "step": 27820
    },
    {
      "epoch": 3.3489771359807463,
      "grad_norm": 7.730136394500732,
      "learning_rate": 0.00019962942001917428,
      "loss": 1.9199,
      "step": 27830
    },
    {
      "epoch": 3.3501805054151625,
      "grad_norm": 4.88864803314209,
      "learning_rate": 0.000199629092738591,
      "loss": 1.7746,
      "step": 27840
    },
    {
      "epoch": 3.351383874849579,
      "grad_norm": 4.529849529266357,
      "learning_rate": 0.00019962876531381984,
      "loss": 1.7188,
      "step": 27850
    },
    {
      "epoch": 3.352587244283995,
      "grad_norm": 6.139866352081299,
      "learning_rate": 0.00019962843774486123,
      "loss": 1.7252,
      "step": 27860
    },
    {
      "epoch": 3.3537906137184117,
      "grad_norm": 5.090317249298096,
      "learning_rate": 0.0001996281100317157,
      "loss": 1.573,
      "step": 27870
    },
    {
      "epoch": 3.354993983152828,
      "grad_norm": 7.82069730758667,
      "learning_rate": 0.00019962778217438366,
      "loss": 1.7227,
      "step": 27880
    },
    {
      "epoch": 3.356197352587244,
      "grad_norm": 6.4739813804626465,
      "learning_rate": 0.00019962745417286568,
      "loss": 1.7838,
      "step": 27890
    },
    {
      "epoch": 3.357400722021661,
      "grad_norm": 4.006805896759033,
      "learning_rate": 0.00019962712602716214,
      "loss": 1.8307,
      "step": 27900
    },
    {
      "epoch": 3.358604091456077,
      "grad_norm": 5.227746486663818,
      "learning_rate": 0.00019962679773727354,
      "loss": 1.8486,
      "step": 27910
    },
    {
      "epoch": 3.3598074608904933,
      "grad_norm": 4.264801025390625,
      "learning_rate": 0.0001996264693032004,
      "loss": 1.8223,
      "step": 27920
    },
    {
      "epoch": 3.3610108303249095,
      "grad_norm": 8.17780876159668,
      "learning_rate": 0.00019962614072494312,
      "loss": 1.8582,
      "step": 27930
    },
    {
      "epoch": 3.3622141997593262,
      "grad_norm": 4.795365333557129,
      "learning_rate": 0.00019962581200250225,
      "loss": 1.7779,
      "step": 27940
    },
    {
      "epoch": 3.3634175691937425,
      "grad_norm": 4.197148323059082,
      "learning_rate": 0.00019962548313587824,
      "loss": 1.8181,
      "step": 27950
    },
    {
      "epoch": 3.3646209386281587,
      "grad_norm": 5.381394386291504,
      "learning_rate": 0.00019962515412507155,
      "loss": 1.7951,
      "step": 27960
    },
    {
      "epoch": 3.3658243080625754,
      "grad_norm": 4.752811431884766,
      "learning_rate": 0.00019962482497008264,
      "loss": 1.7309,
      "step": 27970
    },
    {
      "epoch": 3.3670276774969916,
      "grad_norm": 6.7400407791137695,
      "learning_rate": 0.00019962449567091203,
      "loss": 1.6858,
      "step": 27980
    },
    {
      "epoch": 3.368231046931408,
      "grad_norm": 5.006594181060791,
      "learning_rate": 0.0001996241662275602,
      "loss": 1.8097,
      "step": 27990
    },
    {
      "epoch": 3.3694344163658245,
      "grad_norm": 3.310363292694092,
      "learning_rate": 0.00019962383664002758,
      "loss": 1.6574,
      "step": 28000
    },
    {
      "epoch": 3.3706377858002408,
      "grad_norm": 7.438520431518555,
      "learning_rate": 0.00019962350690831468,
      "loss": 1.6039,
      "step": 28010
    },
    {
      "epoch": 3.371841155234657,
      "grad_norm": 4.770726680755615,
      "learning_rate": 0.00019962317703242196,
      "loss": 1.8445,
      "step": 28020
    },
    {
      "epoch": 3.3730445246690732,
      "grad_norm": 7.135133266448975,
      "learning_rate": 0.00019962284701234993,
      "loss": 1.7904,
      "step": 28030
    },
    {
      "epoch": 3.37424789410349,
      "grad_norm": 4.703583240509033,
      "learning_rate": 0.000199622516848099,
      "loss": 1.6413,
      "step": 28040
    },
    {
      "epoch": 3.375451263537906,
      "grad_norm": 3.2322211265563965,
      "learning_rate": 0.00019962218653966972,
      "loss": 1.779,
      "step": 28050
    },
    {
      "epoch": 3.3766546329723224,
      "grad_norm": 7.324565410614014,
      "learning_rate": 0.00019962185608706255,
      "loss": 1.7111,
      "step": 28060
    },
    {
      "epoch": 3.377858002406739,
      "grad_norm": 5.59927225112915,
      "learning_rate": 0.00019962152549027795,
      "loss": 1.8828,
      "step": 28070
    },
    {
      "epoch": 3.3790613718411553,
      "grad_norm": 9.892529487609863,
      "learning_rate": 0.00019962119474931642,
      "loss": 1.7942,
      "step": 28080
    },
    {
      "epoch": 3.3802647412755715,
      "grad_norm": 5.770024299621582,
      "learning_rate": 0.0001996208638641784,
      "loss": 1.8067,
      "step": 28090
    },
    {
      "epoch": 3.3814681107099878,
      "grad_norm": 3.9775357246398926,
      "learning_rate": 0.0001996205328348644,
      "loss": 1.6265,
      "step": 28100
    },
    {
      "epoch": 3.3826714801444044,
      "grad_norm": 5.100301265716553,
      "learning_rate": 0.0001996202016613749,
      "loss": 2.0109,
      "step": 28110
    },
    {
      "epoch": 3.3838748495788207,
      "grad_norm": 4.932925701141357,
      "learning_rate": 0.00019961987034371038,
      "loss": 1.6067,
      "step": 28120
    },
    {
      "epoch": 3.385078219013237,
      "grad_norm": 6.15216064453125,
      "learning_rate": 0.0001996195388818713,
      "loss": 1.7458,
      "step": 28130
    },
    {
      "epoch": 3.3862815884476536,
      "grad_norm": 4.748860836029053,
      "learning_rate": 0.00019961920727585815,
      "loss": 1.7924,
      "step": 28140
    },
    {
      "epoch": 3.38748495788207,
      "grad_norm": 3.8551394939422607,
      "learning_rate": 0.00019961887552567142,
      "loss": 1.9233,
      "step": 28150
    },
    {
      "epoch": 3.388688327316486,
      "grad_norm": 5.9154534339904785,
      "learning_rate": 0.00019961854363131158,
      "loss": 1.7453,
      "step": 28160
    },
    {
      "epoch": 3.3898916967509027,
      "grad_norm": 4.891790390014648,
      "learning_rate": 0.00019961821159277912,
      "loss": 1.9713,
      "step": 28170
    },
    {
      "epoch": 3.391095066185319,
      "grad_norm": 7.742791652679443,
      "learning_rate": 0.0001996178794100745,
      "loss": 1.8933,
      "step": 28180
    },
    {
      "epoch": 3.392298435619735,
      "grad_norm": 4.845676422119141,
      "learning_rate": 0.00019961754708319818,
      "loss": 1.7511,
      "step": 28190
    },
    {
      "epoch": 3.3935018050541514,
      "grad_norm": 3.511476755142212,
      "learning_rate": 0.00019961721461215075,
      "loss": 1.7201,
      "step": 28200
    },
    {
      "epoch": 3.394705174488568,
      "grad_norm": 7.694055557250977,
      "learning_rate": 0.00019961688199693258,
      "loss": 1.7666,
      "step": 28210
    },
    {
      "epoch": 3.3959085439229844,
      "grad_norm": 4.1438164710998535,
      "learning_rate": 0.00019961654923754418,
      "loss": 1.5868,
      "step": 28220
    },
    {
      "epoch": 3.3971119133574006,
      "grad_norm": 7.416901588439941,
      "learning_rate": 0.00019961621633398605,
      "loss": 1.892,
      "step": 28230
    },
    {
      "epoch": 3.3983152827918173,
      "grad_norm": 5.698300361633301,
      "learning_rate": 0.00019961588328625863,
      "loss": 1.7046,
      "step": 28240
    },
    {
      "epoch": 3.3995186522262335,
      "grad_norm": 4.730085849761963,
      "learning_rate": 0.00019961555009436247,
      "loss": 1.8509,
      "step": 28250
    },
    {
      "epoch": 3.4007220216606497,
      "grad_norm": 7.788651943206787,
      "learning_rate": 0.000199615216758298,
      "loss": 1.7517,
      "step": 28260
    },
    {
      "epoch": 3.401925391095066,
      "grad_norm": 4.645796775817871,
      "learning_rate": 0.0001996148832780657,
      "loss": 1.8645,
      "step": 28270
    },
    {
      "epoch": 3.4031287605294827,
      "grad_norm": 6.676630973815918,
      "learning_rate": 0.00019961454965366612,
      "loss": 2.0621,
      "step": 28280
    },
    {
      "epoch": 3.404332129963899,
      "grad_norm": 4.801552772521973,
      "learning_rate": 0.00019961421588509965,
      "loss": 1.6914,
      "step": 28290
    },
    {
      "epoch": 3.405535499398315,
      "grad_norm": 3.3285067081451416,
      "learning_rate": 0.00019961388197236684,
      "loss": 1.8584,
      "step": 28300
    },
    {
      "epoch": 3.406738868832732,
      "grad_norm": 4.599587917327881,
      "learning_rate": 0.00019961354791546814,
      "loss": 1.788,
      "step": 28310
    },
    {
      "epoch": 3.407942238267148,
      "grad_norm": 5.92603874206543,
      "learning_rate": 0.00019961321371440404,
      "loss": 1.8698,
      "step": 28320
    },
    {
      "epoch": 3.4091456077015643,
      "grad_norm": 7.2303147315979,
      "learning_rate": 0.00019961287936917502,
      "loss": 1.8969,
      "step": 28330
    },
    {
      "epoch": 3.410348977135981,
      "grad_norm": 5.587429523468018,
      "learning_rate": 0.00019961254487978157,
      "loss": 1.7323,
      "step": 28340
    },
    {
      "epoch": 3.411552346570397,
      "grad_norm": 4.460331916809082,
      "learning_rate": 0.00019961221024622418,
      "loss": 1.8817,
      "step": 28350
    },
    {
      "epoch": 3.4127557160048134,
      "grad_norm": 6.216007232666016,
      "learning_rate": 0.00019961187546850334,
      "loss": 1.9292,
      "step": 28360
    },
    {
      "epoch": 3.4139590854392297,
      "grad_norm": 3.744415283203125,
      "learning_rate": 0.0001996115405466195,
      "loss": 1.7755,
      "step": 28370
    },
    {
      "epoch": 3.4151624548736463,
      "grad_norm": 7.921230792999268,
      "learning_rate": 0.00019961120548057316,
      "loss": 1.8186,
      "step": 28380
    },
    {
      "epoch": 3.4163658243080626,
      "grad_norm": 5.462467670440674,
      "learning_rate": 0.00019961087027036485,
      "loss": 1.787,
      "step": 28390
    },
    {
      "epoch": 3.417569193742479,
      "grad_norm": 4.054779052734375,
      "learning_rate": 0.00019961053491599498,
      "loss": 1.8242,
      "step": 28400
    },
    {
      "epoch": 3.4187725631768955,
      "grad_norm": 5.601592540740967,
      "learning_rate": 0.00019961019941746412,
      "loss": 1.9354,
      "step": 28410
    },
    {
      "epoch": 3.4199759326113117,
      "grad_norm": 4.777987480163574,
      "learning_rate": 0.00019960986377477266,
      "loss": 1.8717,
      "step": 28420
    },
    {
      "epoch": 3.421179302045728,
      "grad_norm": 7.961142063140869,
      "learning_rate": 0.00019960952798792113,
      "loss": 1.7858,
      "step": 28430
    },
    {
      "epoch": 3.422382671480144,
      "grad_norm": 4.813782691955566,
      "learning_rate": 0.00019960919205691007,
      "loss": 1.8777,
      "step": 28440
    },
    {
      "epoch": 3.423586040914561,
      "grad_norm": 3.7830121517181396,
      "learning_rate": 0.00019960885598173988,
      "loss": 1.7667,
      "step": 28450
    },
    {
      "epoch": 3.424789410348977,
      "grad_norm": 7.098151206970215,
      "learning_rate": 0.00019960851976241108,
      "loss": 1.8785,
      "step": 28460
    },
    {
      "epoch": 3.4259927797833933,
      "grad_norm": 3.914496660232544,
      "learning_rate": 0.0001996081833989242,
      "loss": 1.6083,
      "step": 28470
    },
    {
      "epoch": 3.42719614921781,
      "grad_norm": 6.490516662597656,
      "learning_rate": 0.00019960784689127965,
      "loss": 1.8888,
      "step": 28480
    },
    {
      "epoch": 3.4283995186522263,
      "grad_norm": 5.446444034576416,
      "learning_rate": 0.00019960751023947798,
      "loss": 1.6972,
      "step": 28490
    },
    {
      "epoch": 3.4296028880866425,
      "grad_norm": 4.287990570068359,
      "learning_rate": 0.0001996071734435196,
      "loss": 1.7736,
      "step": 28500
    },
    {
      "epoch": 3.430806257521059,
      "grad_norm": 5.645087242126465,
      "learning_rate": 0.00019960683650340508,
      "loss": 1.7633,
      "step": 28510
    },
    {
      "epoch": 3.4320096269554754,
      "grad_norm": 5.143428325653076,
      "learning_rate": 0.00019960649941913489,
      "loss": 1.6987,
      "step": 28520
    },
    {
      "epoch": 3.4332129963898916,
      "grad_norm": 6.8210859298706055,
      "learning_rate": 0.0001996061621907095,
      "loss": 1.8576,
      "step": 28530
    },
    {
      "epoch": 3.434416365824308,
      "grad_norm": 5.176815986633301,
      "learning_rate": 0.00019960582481812938,
      "loss": 1.6781,
      "step": 28540
    },
    {
      "epoch": 3.4356197352587245,
      "grad_norm": 3.825925588607788,
      "learning_rate": 0.00019960548730139506,
      "loss": 1.8528,
      "step": 28550
    },
    {
      "epoch": 3.436823104693141,
      "grad_norm": 5.517031192779541,
      "learning_rate": 0.000199605149640507,
      "loss": 1.5867,
      "step": 28560
    },
    {
      "epoch": 3.438026474127557,
      "grad_norm": 4.681079864501953,
      "learning_rate": 0.0001996048118354657,
      "loss": 1.6254,
      "step": 28570
    },
    {
      "epoch": 3.4392298435619737,
      "grad_norm": 8.500269889831543,
      "learning_rate": 0.00019960447388627162,
      "loss": 1.868,
      "step": 28580
    },
    {
      "epoch": 3.44043321299639,
      "grad_norm": 4.874539375305176,
      "learning_rate": 0.0001996041357929253,
      "loss": 1.7744,
      "step": 28590
    },
    {
      "epoch": 3.441636582430806,
      "grad_norm": 3.704907178878784,
      "learning_rate": 0.00019960379755542718,
      "loss": 1.8901,
      "step": 28600
    },
    {
      "epoch": 3.4428399518652224,
      "grad_norm": 8.360466003417969,
      "learning_rate": 0.0001996034591737778,
      "loss": 1.9189,
      "step": 28610
    },
    {
      "epoch": 3.444043321299639,
      "grad_norm": 4.338244915008545,
      "learning_rate": 0.0001996031206479776,
      "loss": 1.6906,
      "step": 28620
    },
    {
      "epoch": 3.4452466907340553,
      "grad_norm": 7.269326210021973,
      "learning_rate": 0.00019960278197802712,
      "loss": 1.9446,
      "step": 28630
    },
    {
      "epoch": 3.4464500601684716,
      "grad_norm": 5.376789093017578,
      "learning_rate": 0.0001996024431639268,
      "loss": 1.8607,
      "step": 28640
    },
    {
      "epoch": 3.4476534296028882,
      "grad_norm": 4.045756816864014,
      "learning_rate": 0.00019960210420567719,
      "loss": 1.7736,
      "step": 28650
    },
    {
      "epoch": 3.4488567990373045,
      "grad_norm": 6.43265962600708,
      "learning_rate": 0.0001996017651032787,
      "loss": 1.7796,
      "step": 28660
    },
    {
      "epoch": 3.4500601684717207,
      "grad_norm": 4.935984134674072,
      "learning_rate": 0.0001996014258567319,
      "loss": 1.8473,
      "step": 28670
    },
    {
      "epoch": 3.4512635379061374,
      "grad_norm": 5.484255313873291,
      "learning_rate": 0.00019960108646603723,
      "loss": 1.7414,
      "step": 28680
    },
    {
      "epoch": 3.4524669073405536,
      "grad_norm": 4.610409736633301,
      "learning_rate": 0.0001996007469311952,
      "loss": 1.8739,
      "step": 28690
    },
    {
      "epoch": 3.45367027677497,
      "grad_norm": 5.166720867156982,
      "learning_rate": 0.00019960040725220627,
      "loss": 1.7138,
      "step": 28700
    },
    {
      "epoch": 3.4548736462093865,
      "grad_norm": 6.070875644683838,
      "learning_rate": 0.000199600067429071,
      "loss": 1.9382,
      "step": 28710
    },
    {
      "epoch": 3.4560770156438028,
      "grad_norm": 5.008320331573486,
      "learning_rate": 0.00019959972746178982,
      "loss": 1.6997,
      "step": 28720
    },
    {
      "epoch": 3.457280385078219,
      "grad_norm": 5.522164344787598,
      "learning_rate": 0.00019959938735036325,
      "loss": 1.7085,
      "step": 28730
    },
    {
      "epoch": 3.4584837545126352,
      "grad_norm": 4.9618916511535645,
      "learning_rate": 0.0001995990470947918,
      "loss": 1.7234,
      "step": 28740
    },
    {
      "epoch": 3.459687123947052,
      "grad_norm": 3.4637858867645264,
      "learning_rate": 0.0001995987066950759,
      "loss": 1.805,
      "step": 28750
    },
    {
      "epoch": 3.460890493381468,
      "grad_norm": 6.136050701141357,
      "learning_rate": 0.00019959836615121609,
      "loss": 2.0882,
      "step": 28760
    },
    {
      "epoch": 3.4620938628158844,
      "grad_norm": 4.310469150543213,
      "learning_rate": 0.00019959802546321286,
      "loss": 1.853,
      "step": 28770
    },
    {
      "epoch": 3.4632972322503006,
      "grad_norm": 6.130943298339844,
      "learning_rate": 0.00019959768463106672,
      "loss": 1.5557,
      "step": 28780
    },
    {
      "epoch": 3.4645006016847173,
      "grad_norm": 4.595365524291992,
      "learning_rate": 0.00019959734365477814,
      "loss": 1.7351,
      "step": 28790
    },
    {
      "epoch": 3.4657039711191335,
      "grad_norm": 4.329321384429932,
      "learning_rate": 0.00019959700253434758,
      "loss": 1.67,
      "step": 28800
    },
    {
      "epoch": 3.4669073405535498,
      "grad_norm": 6.03887939453125,
      "learning_rate": 0.00019959666126977558,
      "loss": 1.628,
      "step": 28810
    },
    {
      "epoch": 3.4681107099879664,
      "grad_norm": 4.7212677001953125,
      "learning_rate": 0.00019959631986106262,
      "loss": 1.8574,
      "step": 28820
    },
    {
      "epoch": 3.4693140794223827,
      "grad_norm": 9.018230438232422,
      "learning_rate": 0.00019959597830820919,
      "loss": 1.8657,
      "step": 28830
    },
    {
      "epoch": 3.470517448856799,
      "grad_norm": 4.717865467071533,
      "learning_rate": 0.0001995956366112158,
      "loss": 1.8667,
      "step": 28840
    },
    {
      "epoch": 3.4717208182912156,
      "grad_norm": 3.674168586730957,
      "learning_rate": 0.00019959529477008294,
      "loss": 1.7367,
      "step": 28850
    },
    {
      "epoch": 3.472924187725632,
      "grad_norm": 5.015222549438477,
      "learning_rate": 0.0001995949527848111,
      "loss": 1.7573,
      "step": 28860
    },
    {
      "epoch": 3.474127557160048,
      "grad_norm": 5.016926288604736,
      "learning_rate": 0.00019959461065540074,
      "loss": 1.7272,
      "step": 28870
    },
    {
      "epoch": 3.4753309265944647,
      "grad_norm": 6.085020542144775,
      "learning_rate": 0.00019959426838185245,
      "loss": 1.8565,
      "step": 28880
    },
    {
      "epoch": 3.476534296028881,
      "grad_norm": 5.02958345413208,
      "learning_rate": 0.00019959392596416661,
      "loss": 1.8379,
      "step": 28890
    },
    {
      "epoch": 3.477737665463297,
      "grad_norm": 3.3129117488861084,
      "learning_rate": 0.0001995935834023438,
      "loss": 1.7124,
      "step": 28900
    },
    {
      "epoch": 3.4789410348977134,
      "grad_norm": 6.748370170593262,
      "learning_rate": 0.00019959324069638448,
      "loss": 1.7598,
      "step": 28910
    },
    {
      "epoch": 3.48014440433213,
      "grad_norm": 4.98140287399292,
      "learning_rate": 0.00019959289784628913,
      "loss": 1.6866,
      "step": 28920
    },
    {
      "epoch": 3.4813477737665464,
      "grad_norm": 8.136743545532227,
      "learning_rate": 0.0001995925548520583,
      "loss": 1.6813,
      "step": 28930
    },
    {
      "epoch": 3.4825511432009626,
      "grad_norm": 5.873473167419434,
      "learning_rate": 0.00019959221171369246,
      "loss": 1.7566,
      "step": 28940
    },
    {
      "epoch": 3.483754512635379,
      "grad_norm": 3.825920343399048,
      "learning_rate": 0.00019959186843119205,
      "loss": 1.708,
      "step": 28950
    },
    {
      "epoch": 3.4849578820697955,
      "grad_norm": 5.885858535766602,
      "learning_rate": 0.00019959152500455767,
      "loss": 1.702,
      "step": 28960
    },
    {
      "epoch": 3.4861612515042117,
      "grad_norm": 5.659242153167725,
      "learning_rate": 0.00019959118143378975,
      "loss": 1.6004,
      "step": 28970
    },
    {
      "epoch": 3.487364620938628,
      "grad_norm": 8.56484317779541,
      "learning_rate": 0.00019959083771888878,
      "loss": 1.8466,
      "step": 28980
    },
    {
      "epoch": 3.4885679903730447,
      "grad_norm": 5.031448841094971,
      "learning_rate": 0.0001995904938598553,
      "loss": 1.6605,
      "step": 28990
    },
    {
      "epoch": 3.489771359807461,
      "grad_norm": 3.6741833686828613,
      "learning_rate": 0.0001995901498566898,
      "loss": 2.004,
      "step": 29000
    },
    {
      "epoch": 3.490974729241877,
      "grad_norm": 5.330018997192383,
      "learning_rate": 0.00019958980570939275,
      "loss": 1.8313,
      "step": 29010
    },
    {
      "epoch": 3.492178098676294,
      "grad_norm": 5.911968231201172,
      "learning_rate": 0.00019958946141796467,
      "loss": 1.7885,
      "step": 29020
    },
    {
      "epoch": 3.49338146811071,
      "grad_norm": 7.138945579528809,
      "learning_rate": 0.00019958911698240604,
      "loss": 1.7559,
      "step": 29030
    },
    {
      "epoch": 3.4945848375451263,
      "grad_norm": 4.7268967628479,
      "learning_rate": 0.00019958877240271736,
      "loss": 1.6986,
      "step": 29040
    },
    {
      "epoch": 3.495788206979543,
      "grad_norm": 3.9076035022735596,
      "learning_rate": 0.00019958842767889916,
      "loss": 1.8307,
      "step": 29050
    },
    {
      "epoch": 3.496991576413959,
      "grad_norm": 6.532130718231201,
      "learning_rate": 0.0001995880828109519,
      "loss": 1.8026,
      "step": 29060
    },
    {
      "epoch": 3.4981949458483754,
      "grad_norm": 3.987971544265747,
      "learning_rate": 0.00019958773779887612,
      "loss": 1.6147,
      "step": 29070
    },
    {
      "epoch": 3.4993983152827917,
      "grad_norm": 7.269247531890869,
      "learning_rate": 0.00019958739264267227,
      "loss": 1.8936,
      "step": 29080
    },
    {
      "epoch": 3.5006016847172083,
      "grad_norm": 5.082418441772461,
      "learning_rate": 0.00019958704734234092,
      "loss": 1.8007,
      "step": 29090
    },
    {
      "epoch": 3.5018050541516246,
      "grad_norm": 5.6328535079956055,
      "learning_rate": 0.00019958670189788248,
      "loss": 1.8599,
      "step": 29100
    },
    {
      "epoch": 3.503008423586041,
      "grad_norm": 5.192950248718262,
      "learning_rate": 0.0001995863563092975,
      "loss": 1.9142,
      "step": 29110
    },
    {
      "epoch": 3.504211793020457,
      "grad_norm": 4.52918815612793,
      "learning_rate": 0.00019958601057658648,
      "loss": 1.6936,
      "step": 29120
    },
    {
      "epoch": 3.5054151624548737,
      "grad_norm": 8.50179386138916,
      "learning_rate": 0.00019958566469974992,
      "loss": 1.7705,
      "step": 29130
    },
    {
      "epoch": 3.50661853188929,
      "grad_norm": 4.8946213722229,
      "learning_rate": 0.00019958531867878834,
      "loss": 1.7076,
      "step": 29140
    },
    {
      "epoch": 3.507821901323706,
      "grad_norm": 3.867093324661255,
      "learning_rate": 0.00019958497251370218,
      "loss": 1.9257,
      "step": 29150
    },
    {
      "epoch": 3.509025270758123,
      "grad_norm": 5.413902759552002,
      "learning_rate": 0.00019958462620449198,
      "loss": 1.5908,
      "step": 29160
    },
    {
      "epoch": 3.510228640192539,
      "grad_norm": 4.138807773590088,
      "learning_rate": 0.00019958427975115827,
      "loss": 1.9396,
      "step": 29170
    },
    {
      "epoch": 3.5114320096269553,
      "grad_norm": 6.781573295593262,
      "learning_rate": 0.0001995839331537015,
      "loss": 1.5823,
      "step": 29180
    },
    {
      "epoch": 3.512635379061372,
      "grad_norm": 6.4733500480651855,
      "learning_rate": 0.00019958358641212218,
      "loss": 1.8299,
      "step": 29190
    },
    {
      "epoch": 3.5138387484957883,
      "grad_norm": 3.607250928878784,
      "learning_rate": 0.00019958323952642084,
      "loss": 1.7479,
      "step": 29200
    },
    {
      "epoch": 3.5150421179302045,
      "grad_norm": 7.795487880706787,
      "learning_rate": 0.00019958289249659796,
      "loss": 1.6971,
      "step": 29210
    },
    {
      "epoch": 3.516245487364621,
      "grad_norm": 4.45479679107666,
      "learning_rate": 0.00019958254532265405,
      "loss": 1.6372,
      "step": 29220
    },
    {
      "epoch": 3.5174488567990374,
      "grad_norm": 7.399784564971924,
      "learning_rate": 0.00019958219800458963,
      "loss": 1.9275,
      "step": 29230
    },
    {
      "epoch": 3.5186522262334536,
      "grad_norm": 5.853253364562988,
      "learning_rate": 0.00019958185054240514,
      "loss": 1.8425,
      "step": 29240
    },
    {
      "epoch": 3.51985559566787,
      "grad_norm": 3.379256248474121,
      "learning_rate": 0.00019958150293610118,
      "loss": 1.6625,
      "step": 29250
    },
    {
      "epoch": 3.5210589651022866,
      "grad_norm": 5.286121845245361,
      "learning_rate": 0.00019958115518567814,
      "loss": 1.6161,
      "step": 29260
    },
    {
      "epoch": 3.522262334536703,
      "grad_norm": 4.732770919799805,
      "learning_rate": 0.00019958080729113665,
      "loss": 1.7531,
      "step": 29270
    },
    {
      "epoch": 3.523465703971119,
      "grad_norm": 6.605644226074219,
      "learning_rate": 0.00019958045925247707,
      "loss": 1.8319,
      "step": 29280
    },
    {
      "epoch": 3.5246690734055353,
      "grad_norm": 6.168133735656738,
      "learning_rate": 0.00019958011106970005,
      "loss": 1.8519,
      "step": 29290
    },
    {
      "epoch": 3.525872442839952,
      "grad_norm": 3.8957347869873047,
      "learning_rate": 0.00019957976274280598,
      "loss": 1.825,
      "step": 29300
    },
    {
      "epoch": 3.527075812274368,
      "grad_norm": 5.978893280029297,
      "learning_rate": 0.00019957941427179544,
      "loss": 1.7849,
      "step": 29310
    },
    {
      "epoch": 3.5282791817087844,
      "grad_norm": 4.665668487548828,
      "learning_rate": 0.00019957906565666887,
      "loss": 1.6495,
      "step": 29320
    },
    {
      "epoch": 3.529482551143201,
      "grad_norm": 8.054835319519043,
      "learning_rate": 0.00019957871689742684,
      "loss": 1.8495,
      "step": 29330
    },
    {
      "epoch": 3.5306859205776173,
      "grad_norm": 5.350250244140625,
      "learning_rate": 0.0001995783679940698,
      "loss": 1.7535,
      "step": 29340
    },
    {
      "epoch": 3.5318892900120336,
      "grad_norm": 3.2606489658355713,
      "learning_rate": 0.00019957801894659826,
      "loss": 1.9664,
      "step": 29350
    },
    {
      "epoch": 3.5330926594464502,
      "grad_norm": 7.022886276245117,
      "learning_rate": 0.00019957766975501278,
      "loss": 1.7611,
      "step": 29360
    },
    {
      "epoch": 3.5342960288808665,
      "grad_norm": 4.701299667358398,
      "learning_rate": 0.0001995773204193138,
      "loss": 1.7028,
      "step": 29370
    },
    {
      "epoch": 3.5354993983152827,
      "grad_norm": 7.65661096572876,
      "learning_rate": 0.00019957697093950184,
      "loss": 1.8365,
      "step": 29380
    },
    {
      "epoch": 3.5367027677496994,
      "grad_norm": 5.906957626342773,
      "learning_rate": 0.00019957662131557747,
      "loss": 1.7081,
      "step": 29390
    },
    {
      "epoch": 3.5379061371841156,
      "grad_norm": 3.471609592437744,
      "learning_rate": 0.0001995762715475411,
      "loss": 1.9223,
      "step": 29400
    },
    {
      "epoch": 3.539109506618532,
      "grad_norm": 6.040035247802734,
      "learning_rate": 0.0001995759216353933,
      "loss": 1.9884,
      "step": 29410
    },
    {
      "epoch": 3.5403128760529485,
      "grad_norm": 4.192653179168701,
      "learning_rate": 0.00019957557157913454,
      "loss": 1.9343,
      "step": 29420
    },
    {
      "epoch": 3.5415162454873648,
      "grad_norm": 7.143955707550049,
      "learning_rate": 0.00019957522137876535,
      "loss": 1.8719,
      "step": 29430
    },
    {
      "epoch": 3.542719614921781,
      "grad_norm": 5.790853023529053,
      "learning_rate": 0.00019957487103428623,
      "loss": 1.7103,
      "step": 29440
    },
    {
      "epoch": 3.5439229843561972,
      "grad_norm": 3.4837894439697266,
      "learning_rate": 0.00019957452054569772,
      "loss": 1.9471,
      "step": 29450
    },
    {
      "epoch": 3.5451263537906135,
      "grad_norm": 6.555818557739258,
      "learning_rate": 0.00019957416991300025,
      "loss": 1.8474,
      "step": 29460
    },
    {
      "epoch": 3.54632972322503,
      "grad_norm": 4.449881553649902,
      "learning_rate": 0.0001995738191361944,
      "loss": 1.9303,
      "step": 29470
    },
    {
      "epoch": 3.5475330926594464,
      "grad_norm": 6.920135021209717,
      "learning_rate": 0.00019957346821528063,
      "loss": 1.7256,
      "step": 29480
    },
    {
      "epoch": 3.5487364620938626,
      "grad_norm": 4.741700172424316,
      "learning_rate": 0.00019957311715025945,
      "loss": 1.7866,
      "step": 29490
    },
    {
      "epoch": 3.5499398315282793,
      "grad_norm": 3.9805476665496826,
      "learning_rate": 0.00019957276594113142,
      "loss": 1.7484,
      "step": 29500
    },
    {
      "epoch": 3.5511432009626955,
      "grad_norm": 6.7731170654296875,
      "learning_rate": 0.00019957241458789702,
      "loss": 1.9032,
      "step": 29510
    },
    {
      "epoch": 3.5523465703971118,
      "grad_norm": 5.218935012817383,
      "learning_rate": 0.00019957206309055672,
      "loss": 1.8034,
      "step": 29520
    },
    {
      "epoch": 3.5535499398315284,
      "grad_norm": 7.347984313964844,
      "learning_rate": 0.00019957171144911108,
      "loss": 1.9651,
      "step": 29530
    },
    {
      "epoch": 3.5547533092659447,
      "grad_norm": 5.54328727722168,
      "learning_rate": 0.0001995713596635606,
      "loss": 1.5477,
      "step": 29540
    },
    {
      "epoch": 3.555956678700361,
      "grad_norm": 3.764723539352417,
      "learning_rate": 0.00019957100773390578,
      "loss": 1.7769,
      "step": 29550
    },
    {
      "epoch": 3.5571600481347776,
      "grad_norm": 6.686862945556641,
      "learning_rate": 0.0001995706556601471,
      "loss": 1.8664,
      "step": 29560
    },
    {
      "epoch": 3.558363417569194,
      "grad_norm": 4.710370063781738,
      "learning_rate": 0.00019957030344228512,
      "loss": 1.7083,
      "step": 29570
    },
    {
      "epoch": 3.55956678700361,
      "grad_norm": 7.936213493347168,
      "learning_rate": 0.00019956995108032032,
      "loss": 1.825,
      "step": 29580
    },
    {
      "epoch": 3.5607701564380267,
      "grad_norm": 4.937076091766357,
      "learning_rate": 0.0001995695985742532,
      "loss": 1.7293,
      "step": 29590
    },
    {
      "epoch": 3.561973525872443,
      "grad_norm": 4.393558502197266,
      "learning_rate": 0.00019956924592408432,
      "loss": 1.8649,
      "step": 29600
    },
    {
      "epoch": 3.563176895306859,
      "grad_norm": 5.727424144744873,
      "learning_rate": 0.00019956889312981418,
      "loss": 1.733,
      "step": 29610
    },
    {
      "epoch": 3.5643802647412755,
      "grad_norm": 5.007108688354492,
      "learning_rate": 0.0001995685401914432,
      "loss": 2.0541,
      "step": 29620
    },
    {
      "epoch": 3.5655836341756917,
      "grad_norm": 7.5373969078063965,
      "learning_rate": 0.00019956818710897202,
      "loss": 1.9797,
      "step": 29630
    },
    {
      "epoch": 3.5667870036101084,
      "grad_norm": 5.30107307434082,
      "learning_rate": 0.00019956783388240106,
      "loss": 1.8703,
      "step": 29640
    },
    {
      "epoch": 3.5679903730445246,
      "grad_norm": 4.112225532531738,
      "learning_rate": 0.00019956748051173088,
      "loss": 1.8334,
      "step": 29650
    },
    {
      "epoch": 3.569193742478941,
      "grad_norm": 5.8449602127075195,
      "learning_rate": 0.00019956712699696193,
      "loss": 1.8945,
      "step": 29660
    },
    {
      "epoch": 3.5703971119133575,
      "grad_norm": 4.755184173583984,
      "learning_rate": 0.00019956677333809482,
      "loss": 1.9676,
      "step": 29670
    },
    {
      "epoch": 3.5716004813477737,
      "grad_norm": 7.043939113616943,
      "learning_rate": 0.00019956641953512998,
      "loss": 1.8171,
      "step": 29680
    },
    {
      "epoch": 3.57280385078219,
      "grad_norm": 5.145519733428955,
      "learning_rate": 0.00019956606558806794,
      "loss": 1.805,
      "step": 29690
    },
    {
      "epoch": 3.5740072202166067,
      "grad_norm": 4.182318210601807,
      "learning_rate": 0.00019956571149690923,
      "loss": 1.81,
      "step": 29700
    },
    {
      "epoch": 3.575210589651023,
      "grad_norm": 5.767271518707275,
      "learning_rate": 0.00019956535726165436,
      "loss": 1.765,
      "step": 29710
    },
    {
      "epoch": 3.576413959085439,
      "grad_norm": 4.717093467712402,
      "learning_rate": 0.00019956500288230382,
      "loss": 1.9329,
      "step": 29720
    },
    {
      "epoch": 3.577617328519856,
      "grad_norm": 6.811041831970215,
      "learning_rate": 0.00019956464835885814,
      "loss": 1.6949,
      "step": 29730
    },
    {
      "epoch": 3.578820697954272,
      "grad_norm": 6.049142837524414,
      "learning_rate": 0.00019956429369131785,
      "loss": 1.7861,
      "step": 29740
    },
    {
      "epoch": 3.5800240673886883,
      "grad_norm": 3.6417489051818848,
      "learning_rate": 0.00019956393887968346,
      "loss": 1.9231,
      "step": 29750
    },
    {
      "epoch": 3.581227436823105,
      "grad_norm": 5.419465065002441,
      "learning_rate": 0.00019956358392395545,
      "loss": 1.7896,
      "step": 29760
    },
    {
      "epoch": 3.582430806257521,
      "grad_norm": 4.61770486831665,
      "learning_rate": 0.00019956322882413435,
      "loss": 1.8287,
      "step": 29770
    },
    {
      "epoch": 3.5836341756919374,
      "grad_norm": 7.098788738250732,
      "learning_rate": 0.00019956287358022066,
      "loss": 1.9132,
      "step": 29780
    },
    {
      "epoch": 3.5848375451263537,
      "grad_norm": 5.410339832305908,
      "learning_rate": 0.00019956251819221494,
      "loss": 1.6021,
      "step": 29790
    },
    {
      "epoch": 3.58604091456077,
      "grad_norm": 3.4524776935577393,
      "learning_rate": 0.00019956216266011766,
      "loss": 1.7949,
      "step": 29800
    },
    {
      "epoch": 3.5872442839951866,
      "grad_norm": 5.46954870223999,
      "learning_rate": 0.00019956180698392934,
      "loss": 1.9543,
      "step": 29810
    },
    {
      "epoch": 3.588447653429603,
      "grad_norm": 5.387228488922119,
      "learning_rate": 0.00019956145116365053,
      "loss": 1.9277,
      "step": 29820
    },
    {
      "epoch": 3.589651022864019,
      "grad_norm": 9.30466079711914,
      "learning_rate": 0.0001995610951992817,
      "loss": 1.8404,
      "step": 29830
    },
    {
      "epoch": 3.5908543922984357,
      "grad_norm": 7.823917865753174,
      "learning_rate": 0.0001995607390908234,
      "loss": 1.8102,
      "step": 29840
    },
    {
      "epoch": 3.592057761732852,
      "grad_norm": 3.8454854488372803,
      "learning_rate": 0.00019956038283827613,
      "loss": 1.9939,
      "step": 29850
    },
    {
      "epoch": 3.593261131167268,
      "grad_norm": 6.494553089141846,
      "learning_rate": 0.00019956002644164035,
      "loss": 2.0794,
      "step": 29860
    },
    {
      "epoch": 3.594464500601685,
      "grad_norm": 3.75325345993042,
      "learning_rate": 0.0001995596699009167,
      "loss": 1.6462,
      "step": 29870
    },
    {
      "epoch": 3.595667870036101,
      "grad_norm": 9.10042953491211,
      "learning_rate": 0.0001995593132161056,
      "loss": 1.748,
      "step": 29880
    },
    {
      "epoch": 3.5968712394705173,
      "grad_norm": 5.1772308349609375,
      "learning_rate": 0.00019955895638720762,
      "loss": 1.5885,
      "step": 29890
    },
    {
      "epoch": 3.598074608904934,
      "grad_norm": 3.6802306175231934,
      "learning_rate": 0.0001995585994142232,
      "loss": 1.9793,
      "step": 29900
    },
    {
      "epoch": 3.5992779783393503,
      "grad_norm": 5.864447116851807,
      "learning_rate": 0.00019955824229715293,
      "loss": 1.8902,
      "step": 29910
    },
    {
      "epoch": 3.6004813477737665,
      "grad_norm": 5.455096244812012,
      "learning_rate": 0.00019955788503599732,
      "loss": 1.9929,
      "step": 29920
    },
    {
      "epoch": 3.601684717208183,
      "grad_norm": 8.159369468688965,
      "learning_rate": 0.00019955752763075685,
      "loss": 1.8808,
      "step": 29930
    },
    {
      "epoch": 3.6028880866425994,
      "grad_norm": 4.807581901550293,
      "learning_rate": 0.00019955717008143204,
      "loss": 1.7032,
      "step": 29940
    },
    {
      "epoch": 3.6040914560770156,
      "grad_norm": 4.044142723083496,
      "learning_rate": 0.00019955681238802345,
      "loss": 1.8653,
      "step": 29950
    },
    {
      "epoch": 3.605294825511432,
      "grad_norm": 6.339328765869141,
      "learning_rate": 0.0001995564545505316,
      "loss": 1.859,
      "step": 29960
    },
    {
      "epoch": 3.606498194945848,
      "grad_norm": 4.411223888397217,
      "learning_rate": 0.00019955609656895694,
      "loss": 1.7288,
      "step": 29970
    },
    {
      "epoch": 3.607701564380265,
      "grad_norm": 7.663251876831055,
      "learning_rate": 0.00019955573844330005,
      "loss": 1.7244,
      "step": 29980
    },
    {
      "epoch": 3.608904933814681,
      "grad_norm": 5.699988842010498,
      "learning_rate": 0.0001995553801735614,
      "loss": 1.6705,
      "step": 29990
    },
    {
      "epoch": 3.6101083032490973,
      "grad_norm": 4.149991989135742,
      "learning_rate": 0.00019955502175974157,
      "loss": 1.8152,
      "step": 30000
    },
    {
      "epoch": 3.611311672683514,
      "grad_norm": 6.543876647949219,
      "learning_rate": 0.00019955466320184103,
      "loss": 1.7618,
      "step": 30010
    },
    {
      "epoch": 3.61251504211793,
      "grad_norm": 4.540354251861572,
      "learning_rate": 0.0001995543044998603,
      "loss": 1.9189,
      "step": 30020
    },
    {
      "epoch": 3.6137184115523464,
      "grad_norm": 8.781855583190918,
      "learning_rate": 0.00019955394565379991,
      "loss": 1.7893,
      "step": 30030
    },
    {
      "epoch": 3.614921780986763,
      "grad_norm": 4.790679454803467,
      "learning_rate": 0.0001995535866636604,
      "loss": 1.7217,
      "step": 30040
    },
    {
      "epoch": 3.6161251504211793,
      "grad_norm": 4.0239577293396,
      "learning_rate": 0.00019955322752944227,
      "loss": 1.8524,
      "step": 30050
    },
    {
      "epoch": 3.6173285198555956,
      "grad_norm": 5.290555000305176,
      "learning_rate": 0.00019955286825114606,
      "loss": 1.616,
      "step": 30060
    },
    {
      "epoch": 3.6185318892900122,
      "grad_norm": 4.982362747192383,
      "learning_rate": 0.00019955250882877222,
      "loss": 1.7095,
      "step": 30070
    },
    {
      "epoch": 3.6197352587244285,
      "grad_norm": 6.728891849517822,
      "learning_rate": 0.00019955214926232136,
      "loss": 1.4373,
      "step": 30080
    },
    {
      "epoch": 3.6209386281588447,
      "grad_norm": 4.579793453216553,
      "learning_rate": 0.00019955178955179394,
      "loss": 1.962,
      "step": 30090
    },
    {
      "epoch": 3.6221419975932614,
      "grad_norm": 4.157637596130371,
      "learning_rate": 0.00019955142969719052,
      "loss": 1.8267,
      "step": 30100
    },
    {
      "epoch": 3.6233453670276776,
      "grad_norm": 5.86622953414917,
      "learning_rate": 0.00019955106969851157,
      "loss": 1.6419,
      "step": 30110
    },
    {
      "epoch": 3.624548736462094,
      "grad_norm": 4.5456862449646,
      "learning_rate": 0.0001995507095557577,
      "loss": 1.7266,
      "step": 30120
    },
    {
      "epoch": 3.62575210589651,
      "grad_norm": 9.584850311279297,
      "learning_rate": 0.00019955034926892932,
      "loss": 1.8276,
      "step": 30130
    },
    {
      "epoch": 3.6269554753309263,
      "grad_norm": 5.99807596206665,
      "learning_rate": 0.00019954998883802704,
      "loss": 1.7605,
      "step": 30140
    },
    {
      "epoch": 3.628158844765343,
      "grad_norm": 4.4949421882629395,
      "learning_rate": 0.00019954962826305135,
      "loss": 1.9129,
      "step": 30150
    },
    {
      "epoch": 3.6293622141997592,
      "grad_norm": 6.591169834136963,
      "learning_rate": 0.00019954926754400275,
      "loss": 1.8907,
      "step": 30160
    },
    {
      "epoch": 3.6305655836341755,
      "grad_norm": 5.168471813201904,
      "learning_rate": 0.00019954890668088176,
      "loss": 1.6874,
      "step": 30170
    },
    {
      "epoch": 3.631768953068592,
      "grad_norm": 6.996469974517822,
      "learning_rate": 0.00019954854567368898,
      "loss": 1.7469,
      "step": 30180
    },
    {
      "epoch": 3.6329723225030084,
      "grad_norm": 5.0457048416137695,
      "learning_rate": 0.00019954818452242484,
      "loss": 1.7064,
      "step": 30190
    },
    {
      "epoch": 3.6341756919374246,
      "grad_norm": 4.4163055419921875,
      "learning_rate": 0.0001995478232270899,
      "loss": 1.7586,
      "step": 30200
    },
    {
      "epoch": 3.6353790613718413,
      "grad_norm": 7.42832612991333,
      "learning_rate": 0.0001995474617876847,
      "loss": 1.8413,
      "step": 30210
    },
    {
      "epoch": 3.6365824308062575,
      "grad_norm": 4.629641056060791,
      "learning_rate": 0.00019954710020420974,
      "loss": 1.7846,
      "step": 30220
    },
    {
      "epoch": 3.6377858002406738,
      "grad_norm": 7.234807968139648,
      "learning_rate": 0.00019954673847666555,
      "loss": 1.7984,
      "step": 30230
    },
    {
      "epoch": 3.6389891696750905,
      "grad_norm": 5.725372314453125,
      "learning_rate": 0.00019954637660505263,
      "loss": 1.7409,
      "step": 30240
    },
    {
      "epoch": 3.6401925391095067,
      "grad_norm": 3.353891134262085,
      "learning_rate": 0.00019954601458937156,
      "loss": 1.5833,
      "step": 30250
    },
    {
      "epoch": 3.641395908543923,
      "grad_norm": 6.6229753494262695,
      "learning_rate": 0.0001995456524296228,
      "loss": 1.6474,
      "step": 30260
    },
    {
      "epoch": 3.6425992779783396,
      "grad_norm": 4.420374393463135,
      "learning_rate": 0.00019954529012580692,
      "loss": 1.7998,
      "step": 30270
    },
    {
      "epoch": 3.643802647412756,
      "grad_norm": 7.485829830169678,
      "learning_rate": 0.00019954492767792443,
      "loss": 1.8961,
      "step": 30280
    },
    {
      "epoch": 3.645006016847172,
      "grad_norm": 4.7207465171813965,
      "learning_rate": 0.00019954456508597585,
      "loss": 1.4601,
      "step": 30290
    },
    {
      "epoch": 3.6462093862815883,
      "grad_norm": 4.8101606369018555,
      "learning_rate": 0.00019954420234996172,
      "loss": 1.8445,
      "step": 30300
    },
    {
      "epoch": 3.6474127557160045,
      "grad_norm": 5.383038520812988,
      "learning_rate": 0.00019954383946988252,
      "loss": 1.91,
      "step": 30310
    },
    {
      "epoch": 3.648616125150421,
      "grad_norm": 4.799562454223633,
      "learning_rate": 0.00019954347644573884,
      "loss": 1.8472,
      "step": 30320
    },
    {
      "epoch": 3.6498194945848375,
      "grad_norm": 8.128114700317383,
      "learning_rate": 0.00019954311327753115,
      "loss": 1.7467,
      "step": 30330
    },
    {
      "epoch": 3.6510228640192537,
      "grad_norm": 5.222129821777344,
      "learning_rate": 0.00019954274996526001,
      "loss": 1.6014,
      "step": 30340
    },
    {
      "epoch": 3.6522262334536704,
      "grad_norm": 4.164952278137207,
      "learning_rate": 0.00019954238650892592,
      "loss": 1.7273,
      "step": 30350
    },
    {
      "epoch": 3.6534296028880866,
      "grad_norm": 6.184064865112305,
      "learning_rate": 0.00019954202290852947,
      "loss": 2.0273,
      "step": 30360
    },
    {
      "epoch": 3.654632972322503,
      "grad_norm": 4.962084770202637,
      "learning_rate": 0.00019954165916407108,
      "loss": 1.6934,
      "step": 30370
    },
    {
      "epoch": 3.6558363417569195,
      "grad_norm": 7.644177436828613,
      "learning_rate": 0.00019954129527555134,
      "loss": 1.8554,
      "step": 30380
    },
    {
      "epoch": 3.6570397111913358,
      "grad_norm": 5.701297283172607,
      "learning_rate": 0.0001995409312429708,
      "loss": 1.7282,
      "step": 30390
    },
    {
      "epoch": 3.658243080625752,
      "grad_norm": 3.3246676921844482,
      "learning_rate": 0.00019954056706632993,
      "loss": 2.1334,
      "step": 30400
    },
    {
      "epoch": 3.6594464500601687,
      "grad_norm": 7.3835859298706055,
      "learning_rate": 0.0001995402027456293,
      "loss": 1.8398,
      "step": 30410
    },
    {
      "epoch": 3.660649819494585,
      "grad_norm": 4.011591911315918,
      "learning_rate": 0.0001995398382808694,
      "loss": 1.6743,
      "step": 30420
    },
    {
      "epoch": 3.661853188929001,
      "grad_norm": 5.709569454193115,
      "learning_rate": 0.0001995394736720508,
      "loss": 1.7296,
      "step": 30430
    },
    {
      "epoch": 3.663056558363418,
      "grad_norm": 4.6262712478637695,
      "learning_rate": 0.000199539108919174,
      "loss": 1.8377,
      "step": 30440
    },
    {
      "epoch": 3.664259927797834,
      "grad_norm": 4.076691627502441,
      "learning_rate": 0.00019953874402223954,
      "loss": 1.6846,
      "step": 30450
    },
    {
      "epoch": 3.6654632972322503,
      "grad_norm": 5.821899890899658,
      "learning_rate": 0.0001995383789812479,
      "loss": 2.0932,
      "step": 30460
    },
    {
      "epoch": 3.6666666666666665,
      "grad_norm": 4.39022159576416,
      "learning_rate": 0.0001995380137961997,
      "loss": 1.7241,
      "step": 30470
    },
    {
      "epoch": 3.667870036101083,
      "grad_norm": 6.937319755554199,
      "learning_rate": 0.0001995376484670954,
      "loss": 1.8848,
      "step": 30480
    },
    {
      "epoch": 3.6690734055354994,
      "grad_norm": 6.681645393371582,
      "learning_rate": 0.00019953728299393556,
      "loss": 1.7457,
      "step": 30490
    },
    {
      "epoch": 3.6702767749699157,
      "grad_norm": 4.0746750831604,
      "learning_rate": 0.0001995369173767207,
      "loss": 1.7131,
      "step": 30500
    },
    {
      "epoch": 3.671480144404332,
      "grad_norm": 5.25900411605835,
      "learning_rate": 0.0001995365516154513,
      "loss": 1.8158,
      "step": 30510
    },
    {
      "epoch": 3.6726835138387486,
      "grad_norm": 4.574855804443359,
      "learning_rate": 0.000199536185710128,
      "loss": 1.7533,
      "step": 30520
    },
    {
      "epoch": 3.673886883273165,
      "grad_norm": 7.876123905181885,
      "learning_rate": 0.0001995358196607512,
      "loss": 1.7022,
      "step": 30530
    },
    {
      "epoch": 3.675090252707581,
      "grad_norm": 6.505460739135742,
      "learning_rate": 0.00019953545346732153,
      "loss": 1.7364,
      "step": 30540
    },
    {
      "epoch": 3.6762936221419977,
      "grad_norm": 3.9832680225372314,
      "learning_rate": 0.0001995350871298395,
      "loss": 1.7994,
      "step": 30550
    },
    {
      "epoch": 3.677496991576414,
      "grad_norm": 6.498528480529785,
      "learning_rate": 0.00019953472064830561,
      "loss": 1.6826,
      "step": 30560
    },
    {
      "epoch": 3.67870036101083,
      "grad_norm": 5.466087341308594,
      "learning_rate": 0.00019953435402272042,
      "loss": 1.6335,
      "step": 30570
    },
    {
      "epoch": 3.679903730445247,
      "grad_norm": 8.307892799377441,
      "learning_rate": 0.0001995339872530844,
      "loss": 1.7924,
      "step": 30580
    },
    {
      "epoch": 3.681107099879663,
      "grad_norm": 5.839925289154053,
      "learning_rate": 0.00019953362033939818,
      "loss": 1.6798,
      "step": 30590
    },
    {
      "epoch": 3.6823104693140793,
      "grad_norm": 3.620140314102173,
      "learning_rate": 0.0001995332532816622,
      "loss": 1.7933,
      "step": 30600
    },
    {
      "epoch": 3.683513838748496,
      "grad_norm": 7.285762310028076,
      "learning_rate": 0.00019953288607987705,
      "loss": 1.7459,
      "step": 30610
    },
    {
      "epoch": 3.6847172081829123,
      "grad_norm": 3.719926357269287,
      "learning_rate": 0.0001995325187340432,
      "loss": 1.7365,
      "step": 30620
    },
    {
      "epoch": 3.6859205776173285,
      "grad_norm": 9.426250457763672,
      "learning_rate": 0.00019953215124416125,
      "loss": 2.0286,
      "step": 30630
    },
    {
      "epoch": 3.6871239470517447,
      "grad_norm": 5.778400897979736,
      "learning_rate": 0.0001995317836102317,
      "loss": 1.9127,
      "step": 30640
    },
    {
      "epoch": 3.6883273164861614,
      "grad_norm": 3.7114202976226807,
      "learning_rate": 0.0001995314158322551,
      "loss": 1.6488,
      "step": 30650
    },
    {
      "epoch": 3.6895306859205776,
      "grad_norm": 5.503793239593506,
      "learning_rate": 0.00019953104791023196,
      "loss": 1.6922,
      "step": 30660
    },
    {
      "epoch": 3.690734055354994,
      "grad_norm": 5.489040374755859,
      "learning_rate": 0.00019953067984416283,
      "loss": 1.787,
      "step": 30670
    },
    {
      "epoch": 3.69193742478941,
      "grad_norm": 7.977235317230225,
      "learning_rate": 0.0001995303116340482,
      "loss": 2.02,
      "step": 30680
    },
    {
      "epoch": 3.693140794223827,
      "grad_norm": 5.784267425537109,
      "learning_rate": 0.00019952994327988868,
      "loss": 1.9409,
      "step": 30690
    },
    {
      "epoch": 3.694344163658243,
      "grad_norm": 3.913222551345825,
      "learning_rate": 0.00019952957478168472,
      "loss": 1.8691,
      "step": 30700
    },
    {
      "epoch": 3.6955475330926593,
      "grad_norm": 8.454694747924805,
      "learning_rate": 0.00019952920613943692,
      "loss": 1.7532,
      "step": 30710
    },
    {
      "epoch": 3.696750902527076,
      "grad_norm": 5.654797554016113,
      "learning_rate": 0.00019952883735314576,
      "loss": 1.7633,
      "step": 30720
    },
    {
      "epoch": 3.697954271961492,
      "grad_norm": 10.2830171585083,
      "learning_rate": 0.0001995284684228118,
      "loss": 1.9116,
      "step": 30730
    },
    {
      "epoch": 3.6991576413959084,
      "grad_norm": 8.618124961853027,
      "learning_rate": 0.0001995280993484356,
      "loss": 1.8392,
      "step": 30740
    },
    {
      "epoch": 3.700361010830325,
      "grad_norm": 4.4848480224609375,
      "learning_rate": 0.00019952773013001766,
      "loss": 1.8059,
      "step": 30750
    },
    {
      "epoch": 3.7015643802647413,
      "grad_norm": 7.338475227355957,
      "learning_rate": 0.0001995273607675585,
      "loss": 1.9288,
      "step": 30760
    },
    {
      "epoch": 3.7027677496991576,
      "grad_norm": 5.886021614074707,
      "learning_rate": 0.0001995269912610587,
      "loss": 1.8206,
      "step": 30770
    },
    {
      "epoch": 3.7039711191335742,
      "grad_norm": 7.312557697296143,
      "learning_rate": 0.00019952662161051876,
      "loss": 2.0556,
      "step": 30780
    },
    {
      "epoch": 3.7051744885679905,
      "grad_norm": 5.307499885559082,
      "learning_rate": 0.00019952625181593922,
      "loss": 1.8527,
      "step": 30790
    },
    {
      "epoch": 3.7063778580024067,
      "grad_norm": 4.026417255401611,
      "learning_rate": 0.00019952588187732064,
      "loss": 1.6672,
      "step": 30800
    },
    {
      "epoch": 3.707581227436823,
      "grad_norm": 8.716653823852539,
      "learning_rate": 0.00019952551179466352,
      "loss": 1.7917,
      "step": 30810
    },
    {
      "epoch": 3.7087845968712396,
      "grad_norm": 4.396418571472168,
      "learning_rate": 0.00019952514156796843,
      "loss": 2.0722,
      "step": 30820
    },
    {
      "epoch": 3.709987966305656,
      "grad_norm": 8.340319633483887,
      "learning_rate": 0.00019952477119723587,
      "loss": 1.9754,
      "step": 30830
    },
    {
      "epoch": 3.711191335740072,
      "grad_norm": 6.628535747528076,
      "learning_rate": 0.0001995244006824664,
      "loss": 1.8696,
      "step": 30840
    },
    {
      "epoch": 3.7123947051744883,
      "grad_norm": 3.324138879776001,
      "learning_rate": 0.00019952403002366056,
      "loss": 1.6401,
      "step": 30850
    },
    {
      "epoch": 3.713598074608905,
      "grad_norm": 7.406821250915527,
      "learning_rate": 0.00019952365922081886,
      "loss": 1.7917,
      "step": 30860
    },
    {
      "epoch": 3.7148014440433212,
      "grad_norm": 4.478430271148682,
      "learning_rate": 0.00019952328827394186,
      "loss": 1.7915,
      "step": 30870
    },
    {
      "epoch": 3.7160048134777375,
      "grad_norm": 7.9550089836120605,
      "learning_rate": 0.00019952291718303012,
      "loss": 1.9478,
      "step": 30880
    },
    {
      "epoch": 3.717208182912154,
      "grad_norm": 6.367208957672119,
      "learning_rate": 0.0001995225459480841,
      "loss": 1.8459,
      "step": 30890
    },
    {
      "epoch": 3.7184115523465704,
      "grad_norm": 4.044086933135986,
      "learning_rate": 0.00019952217456910442,
      "loss": 1.8967,
      "step": 30900
    },
    {
      "epoch": 3.7196149217809866,
      "grad_norm": 6.0452375411987305,
      "learning_rate": 0.00019952180304609157,
      "loss": 2.1581,
      "step": 30910
    },
    {
      "epoch": 3.7208182912154033,
      "grad_norm": 4.80734395980835,
      "learning_rate": 0.0001995214313790461,
      "loss": 1.827,
      "step": 30920
    },
    {
      "epoch": 3.7220216606498195,
      "grad_norm": 8.116764068603516,
      "learning_rate": 0.00019952105956796853,
      "loss": 1.9225,
      "step": 30930
    },
    {
      "epoch": 3.7232250300842358,
      "grad_norm": 5.5072526931762695,
      "learning_rate": 0.00019952068761285943,
      "loss": 1.7358,
      "step": 30940
    },
    {
      "epoch": 3.7244283995186525,
      "grad_norm": 3.0707244873046875,
      "learning_rate": 0.00019952031551371933,
      "loss": 1.9061,
      "step": 30950
    },
    {
      "epoch": 3.7256317689530687,
      "grad_norm": 5.494694709777832,
      "learning_rate": 0.00019951994327054875,
      "loss": 1.9599,
      "step": 30960
    },
    {
      "epoch": 3.726835138387485,
      "grad_norm": 4.31592321395874,
      "learning_rate": 0.00019951957088334824,
      "loss": 1.8695,
      "step": 30970
    },
    {
      "epoch": 3.728038507821901,
      "grad_norm": 7.593077182769775,
      "learning_rate": 0.00019951919835211837,
      "loss": 2.0167,
      "step": 30980
    },
    {
      "epoch": 3.729241877256318,
      "grad_norm": 5.866640567779541,
      "learning_rate": 0.00019951882567685963,
      "loss": 1.8846,
      "step": 30990
    },
    {
      "epoch": 3.730445246690734,
      "grad_norm": 5.204739570617676,
      "learning_rate": 0.00019951845285757255,
      "loss": 1.7558,
      "step": 31000
    },
    {
      "epoch": 3.7316486161251503,
      "grad_norm": 6.713199615478516,
      "learning_rate": 0.00019951807989425775,
      "loss": 1.8385,
      "step": 31010
    },
    {
      "epoch": 3.7328519855595665,
      "grad_norm": 4.8455939292907715,
      "learning_rate": 0.0001995177067869157,
      "loss": 2.009,
      "step": 31020
    },
    {
      "epoch": 3.7340553549939832,
      "grad_norm": 8.778068542480469,
      "learning_rate": 0.00019951733353554697,
      "loss": 1.8584,
      "step": 31030
    },
    {
      "epoch": 3.7352587244283995,
      "grad_norm": 7.0389180183410645,
      "learning_rate": 0.00019951696014015204,
      "loss": 1.7476,
      "step": 31040
    },
    {
      "epoch": 3.7364620938628157,
      "grad_norm": 3.5129287242889404,
      "learning_rate": 0.00019951658660073154,
      "loss": 1.8695,
      "step": 31050
    },
    {
      "epoch": 3.7376654632972324,
      "grad_norm": 7.025370121002197,
      "learning_rate": 0.00019951621291728597,
      "loss": 1.8708,
      "step": 31060
    },
    {
      "epoch": 3.7388688327316486,
      "grad_norm": 5.776852607727051,
      "learning_rate": 0.00019951583908981585,
      "loss": 1.7144,
      "step": 31070
    },
    {
      "epoch": 3.740072202166065,
      "grad_norm": 8.288241386413574,
      "learning_rate": 0.00019951546511832175,
      "loss": 1.8474,
      "step": 31080
    },
    {
      "epoch": 3.7412755716004815,
      "grad_norm": 5.879149436950684,
      "learning_rate": 0.0001995150910028042,
      "loss": 1.7656,
      "step": 31090
    },
    {
      "epoch": 3.7424789410348978,
      "grad_norm": 3.3691365718841553,
      "learning_rate": 0.00019951471674326375,
      "loss": 1.7449,
      "step": 31100
    },
    {
      "epoch": 3.743682310469314,
      "grad_norm": 7.728209972381592,
      "learning_rate": 0.0001995143423397009,
      "loss": 1.868,
      "step": 31110
    },
    {
      "epoch": 3.7448856799037307,
      "grad_norm": 4.389208793640137,
      "learning_rate": 0.0001995139677921163,
      "loss": 1.7713,
      "step": 31120
    },
    {
      "epoch": 3.746089049338147,
      "grad_norm": 8.326992988586426,
      "learning_rate": 0.00019951359310051037,
      "loss": 2.0934,
      "step": 31130
    },
    {
      "epoch": 3.747292418772563,
      "grad_norm": 5.7215657234191895,
      "learning_rate": 0.0001995132182648837,
      "loss": 1.6712,
      "step": 31140
    },
    {
      "epoch": 3.74849578820698,
      "grad_norm": 4.79075813293457,
      "learning_rate": 0.00019951284328523686,
      "loss": 1.7695,
      "step": 31150
    },
    {
      "epoch": 3.749699157641396,
      "grad_norm": 6.803378105163574,
      "learning_rate": 0.00019951246816157035,
      "loss": 2.0639,
      "step": 31160
    },
    {
      "epoch": 3.7509025270758123,
      "grad_norm": 7.559670925140381,
      "learning_rate": 0.00019951209289388472,
      "loss": 1.8505,
      "step": 31170
    },
    {
      "epoch": 3.7521058965102285,
      "grad_norm": 7.970536231994629,
      "learning_rate": 0.00019951171748218053,
      "loss": 1.7489,
      "step": 31180
    },
    {
      "epoch": 3.7533092659446448,
      "grad_norm": 5.386301040649414,
      "learning_rate": 0.00019951134192645832,
      "loss": 1.8248,
      "step": 31190
    },
    {
      "epoch": 3.7545126353790614,
      "grad_norm": 4.138087749481201,
      "learning_rate": 0.00019951096622671864,
      "loss": 1.8383,
      "step": 31200
    },
    {
      "epoch": 3.7557160048134777,
      "grad_norm": 6.622817516326904,
      "learning_rate": 0.00019951059038296202,
      "loss": 1.9262,
      "step": 31210
    },
    {
      "epoch": 3.756919374247894,
      "grad_norm": 4.338325023651123,
      "learning_rate": 0.00019951021439518899,
      "loss": 1.5994,
      "step": 31220
    },
    {
      "epoch": 3.7581227436823106,
      "grad_norm": 7.29240608215332,
      "learning_rate": 0.00019950983826340013,
      "loss": 1.7926,
      "step": 31230
    },
    {
      "epoch": 3.759326113116727,
      "grad_norm": 5.991332054138184,
      "learning_rate": 0.00019950946198759594,
      "loss": 1.7988,
      "step": 31240
    },
    {
      "epoch": 3.760529482551143,
      "grad_norm": 3.630941152572632,
      "learning_rate": 0.00019950908556777704,
      "loss": 1.8122,
      "step": 31250
    },
    {
      "epoch": 3.7617328519855597,
      "grad_norm": 6.232128620147705,
      "learning_rate": 0.0001995087090039439,
      "loss": 2.0054,
      "step": 31260
    },
    {
      "epoch": 3.762936221419976,
      "grad_norm": 4.277485370635986,
      "learning_rate": 0.00019950833229609707,
      "loss": 1.8732,
      "step": 31270
    },
    {
      "epoch": 3.764139590854392,
      "grad_norm": 7.671878814697266,
      "learning_rate": 0.00019950795544423713,
      "loss": 1.8111,
      "step": 31280
    },
    {
      "epoch": 3.765342960288809,
      "grad_norm": 5.963305950164795,
      "learning_rate": 0.00019950757844836462,
      "loss": 1.7443,
      "step": 31290
    },
    {
      "epoch": 3.766546329723225,
      "grad_norm": 3.8424763679504395,
      "learning_rate": 0.00019950720130848003,
      "loss": 1.8689,
      "step": 31300
    },
    {
      "epoch": 3.7677496991576414,
      "grad_norm": 7.679384708404541,
      "learning_rate": 0.000199506824024584,
      "loss": 1.4353,
      "step": 31310
    },
    {
      "epoch": 3.768953068592058,
      "grad_norm": 4.970163822174072,
      "learning_rate": 0.00019950644659667702,
      "loss": 1.7932,
      "step": 31320
    },
    {
      "epoch": 3.7701564380264743,
      "grad_norm": 7.572193145751953,
      "learning_rate": 0.00019950606902475966,
      "loss": 2.0589,
      "step": 31330
    },
    {
      "epoch": 3.7713598074608905,
      "grad_norm": 5.745700836181641,
      "learning_rate": 0.0001995056913088324,
      "loss": 1.8469,
      "step": 31340
    },
    {
      "epoch": 3.7725631768953067,
      "grad_norm": 4.959266185760498,
      "learning_rate": 0.00019950531344889588,
      "loss": 1.7171,
      "step": 31350
    },
    {
      "epoch": 3.773766546329723,
      "grad_norm": 6.040401935577393,
      "learning_rate": 0.00019950493544495062,
      "loss": 1.8227,
      "step": 31360
    },
    {
      "epoch": 3.7749699157641396,
      "grad_norm": 5.258152008056641,
      "learning_rate": 0.0001995045572969971,
      "loss": 1.8235,
      "step": 31370
    },
    {
      "epoch": 3.776173285198556,
      "grad_norm": 7.423575401306152,
      "learning_rate": 0.00019950417900503595,
      "loss": 1.9273,
      "step": 31380
    },
    {
      "epoch": 3.777376654632972,
      "grad_norm": 6.263308048248291,
      "learning_rate": 0.00019950380056906768,
      "loss": 1.8643,
      "step": 31390
    },
    {
      "epoch": 3.778580024067389,
      "grad_norm": 4.115621089935303,
      "learning_rate": 0.00019950342198909283,
      "loss": 1.8648,
      "step": 31400
    },
    {
      "epoch": 3.779783393501805,
      "grad_norm": 6.392474174499512,
      "learning_rate": 0.00019950304326511197,
      "loss": 1.6777,
      "step": 31410
    },
    {
      "epoch": 3.7809867629362213,
      "grad_norm": 6.091777324676514,
      "learning_rate": 0.00019950266439712567,
      "loss": 1.8839,
      "step": 31420
    },
    {
      "epoch": 3.782190132370638,
      "grad_norm": 7.052060127258301,
      "learning_rate": 0.00019950228538513442,
      "loss": 1.8583,
      "step": 31430
    },
    {
      "epoch": 3.783393501805054,
      "grad_norm": 5.742568016052246,
      "learning_rate": 0.0001995019062291388,
      "loss": 1.8722,
      "step": 31440
    },
    {
      "epoch": 3.7845968712394704,
      "grad_norm": 3.930863380432129,
      "learning_rate": 0.00019950152692913936,
      "loss": 1.8836,
      "step": 31450
    },
    {
      "epoch": 3.785800240673887,
      "grad_norm": 9.043641090393066,
      "learning_rate": 0.00019950114748513664,
      "loss": 1.7099,
      "step": 31460
    },
    {
      "epoch": 3.7870036101083033,
      "grad_norm": 6.021909713745117,
      "learning_rate": 0.0001995007678971312,
      "loss": 1.8497,
      "step": 31470
    },
    {
      "epoch": 3.7882069795427196,
      "grad_norm": 8.166851043701172,
      "learning_rate": 0.00019950038816512358,
      "loss": 1.8112,
      "step": 31480
    },
    {
      "epoch": 3.7894103489771362,
      "grad_norm": 5.161540985107422,
      "learning_rate": 0.0001995000082891143,
      "loss": 1.5989,
      "step": 31490
    },
    {
      "epoch": 3.7906137184115525,
      "grad_norm": 4.264873027801514,
      "learning_rate": 0.000199499628269104,
      "loss": 1.8338,
      "step": 31500
    },
    {
      "epoch": 3.7918170878459687,
      "grad_norm": 7.2630181312561035,
      "learning_rate": 0.00019949924810509315,
      "loss": 1.6914,
      "step": 31510
    },
    {
      "epoch": 3.793020457280385,
      "grad_norm": 5.075514316558838,
      "learning_rate": 0.0001994988677970823,
      "loss": 1.8512,
      "step": 31520
    },
    {
      "epoch": 3.794223826714801,
      "grad_norm": 10.236590385437012,
      "learning_rate": 0.00019949848734507203,
      "loss": 1.8832,
      "step": 31530
    },
    {
      "epoch": 3.795427196149218,
      "grad_norm": 5.291720390319824,
      "learning_rate": 0.0001994981067490629,
      "loss": 1.7179,
      "step": 31540
    },
    {
      "epoch": 3.796630565583634,
      "grad_norm": 3.2387924194335938,
      "learning_rate": 0.00019949772600905545,
      "loss": 1.8682,
      "step": 31550
    },
    {
      "epoch": 3.7978339350180503,
      "grad_norm": 6.901940822601318,
      "learning_rate": 0.0001994973451250502,
      "loss": 1.9452,
      "step": 31560
    },
    {
      "epoch": 3.799037304452467,
      "grad_norm": 5.423299312591553,
      "learning_rate": 0.00019949696409704774,
      "loss": 1.9803,
      "step": 31570
    },
    {
      "epoch": 3.8002406738868832,
      "grad_norm": 8.527501106262207,
      "learning_rate": 0.0001994965829250486,
      "loss": 2.028,
      "step": 31580
    },
    {
      "epoch": 3.8014440433212995,
      "grad_norm": 5.666082859039307,
      "learning_rate": 0.00019949620160905336,
      "loss": 1.9733,
      "step": 31590
    },
    {
      "epoch": 3.802647412755716,
      "grad_norm": 3.896209239959717,
      "learning_rate": 0.00019949582014906251,
      "loss": 1.7397,
      "step": 31600
    },
    {
      "epoch": 3.8038507821901324,
      "grad_norm": 5.688864231109619,
      "learning_rate": 0.0001994954385450767,
      "loss": 1.999,
      "step": 31610
    },
    {
      "epoch": 3.8050541516245486,
      "grad_norm": 4.99901819229126,
      "learning_rate": 0.0001994950567970964,
      "loss": 1.8133,
      "step": 31620
    },
    {
      "epoch": 3.8062575210589653,
      "grad_norm": 9.179929733276367,
      "learning_rate": 0.00019949467490512217,
      "loss": 1.9697,
      "step": 31630
    },
    {
      "epoch": 3.8074608904933815,
      "grad_norm": 5.502438545227051,
      "learning_rate": 0.00019949429286915458,
      "loss": 1.7058,
      "step": 31640
    },
    {
      "epoch": 3.808664259927798,
      "grad_norm": 4.436057090759277,
      "learning_rate": 0.00019949391068919422,
      "loss": 1.6759,
      "step": 31650
    },
    {
      "epoch": 3.8098676293622145,
      "grad_norm": 5.866961479187012,
      "learning_rate": 0.00019949352836524158,
      "loss": 1.963,
      "step": 31660
    },
    {
      "epoch": 3.8110709987966307,
      "grad_norm": 5.318753719329834,
      "learning_rate": 0.00019949314589729725,
      "loss": 1.8176,
      "step": 31670
    },
    {
      "epoch": 3.812274368231047,
      "grad_norm": 8.733996391296387,
      "learning_rate": 0.00019949276328536176,
      "loss": 1.7677,
      "step": 31680
    },
    {
      "epoch": 3.813477737665463,
      "grad_norm": 4.764383792877197,
      "learning_rate": 0.00019949238052943571,
      "loss": 1.7102,
      "step": 31690
    },
    {
      "epoch": 3.8146811070998794,
      "grad_norm": 4.100124359130859,
      "learning_rate": 0.0001994919976295196,
      "loss": 1.9146,
      "step": 31700
    },
    {
      "epoch": 3.815884476534296,
      "grad_norm": 5.41672945022583,
      "learning_rate": 0.000199491614585614,
      "loss": 1.7753,
      "step": 31710
    },
    {
      "epoch": 3.8170878459687123,
      "grad_norm": 4.436059951782227,
      "learning_rate": 0.00019949123139771947,
      "loss": 1.6387,
      "step": 31720
    },
    {
      "epoch": 3.8182912154031285,
      "grad_norm": 7.260838031768799,
      "learning_rate": 0.00019949084806583658,
      "loss": 1.8845,
      "step": 31730
    },
    {
      "epoch": 3.8194945848375452,
      "grad_norm": 5.721841812133789,
      "learning_rate": 0.00019949046458996587,
      "loss": 1.7196,
      "step": 31740
    },
    {
      "epoch": 3.8206979542719615,
      "grad_norm": 3.769033908843994,
      "learning_rate": 0.00019949008097010787,
      "loss": 1.8924,
      "step": 31750
    },
    {
      "epoch": 3.8219013237063777,
      "grad_norm": 7.916036605834961,
      "learning_rate": 0.00019948969720626317,
      "loss": 1.8181,
      "step": 31760
    },
    {
      "epoch": 3.8231046931407944,
      "grad_norm": 5.001548767089844,
      "learning_rate": 0.00019948931329843232,
      "loss": 1.8001,
      "step": 31770
    },
    {
      "epoch": 3.8243080625752106,
      "grad_norm": 7.248650074005127,
      "learning_rate": 0.00019948892924661585,
      "loss": 1.9166,
      "step": 31780
    },
    {
      "epoch": 3.825511432009627,
      "grad_norm": 4.9142913818359375,
      "learning_rate": 0.00019948854505081438,
      "loss": 1.779,
      "step": 31790
    },
    {
      "epoch": 3.8267148014440435,
      "grad_norm": 4.243422985076904,
      "learning_rate": 0.00019948816071102839,
      "loss": 1.8412,
      "step": 31800
    },
    {
      "epoch": 3.8279181708784598,
      "grad_norm": 5.534666061401367,
      "learning_rate": 0.0001994877762272585,
      "loss": 1.9305,
      "step": 31810
    },
    {
      "epoch": 3.829121540312876,
      "grad_norm": 5.762303352355957,
      "learning_rate": 0.0001994873915995052,
      "loss": 1.7955,
      "step": 31820
    },
    {
      "epoch": 3.8303249097472927,
      "grad_norm": 8.31263256072998,
      "learning_rate": 0.00019948700682776907,
      "loss": 2.0227,
      "step": 31830
    },
    {
      "epoch": 3.831528279181709,
      "grad_norm": 6.127068996429443,
      "learning_rate": 0.00019948662191205071,
      "loss": 2.045,
      "step": 31840
    },
    {
      "epoch": 3.832731648616125,
      "grad_norm": 4.620586395263672,
      "learning_rate": 0.00019948623685235063,
      "loss": 1.9066,
      "step": 31850
    },
    {
      "epoch": 3.8339350180505414,
      "grad_norm": 6.38811731338501,
      "learning_rate": 0.00019948585164866942,
      "loss": 1.778,
      "step": 31860
    },
    {
      "epoch": 3.8351383874849576,
      "grad_norm": 4.971603870391846,
      "learning_rate": 0.0001994854663010076,
      "loss": 1.789,
      "step": 31870
    },
    {
      "epoch": 3.8363417569193743,
      "grad_norm": 6.25272798538208,
      "learning_rate": 0.0001994850808093658,
      "loss": 2.0515,
      "step": 31880
    },
    {
      "epoch": 3.8375451263537905,
      "grad_norm": 7.055475234985352,
      "learning_rate": 0.00019948469517374446,
      "loss": 1.7579,
      "step": 31890
    },
    {
      "epoch": 3.8387484957882068,
      "grad_norm": 4.127619743347168,
      "learning_rate": 0.00019948430939414422,
      "loss": 1.9073,
      "step": 31900
    },
    {
      "epoch": 3.8399518652226234,
      "grad_norm": 6.342318534851074,
      "learning_rate": 0.00019948392347056567,
      "loss": 1.6187,
      "step": 31910
    },
    {
      "epoch": 3.8411552346570397,
      "grad_norm": 4.762055397033691,
      "learning_rate": 0.00019948353740300925,
      "loss": 1.8488,
      "step": 31920
    },
    {
      "epoch": 3.842358604091456,
      "grad_norm": 7.973830223083496,
      "learning_rate": 0.00019948315119147565,
      "loss": 1.7017,
      "step": 31930
    },
    {
      "epoch": 3.8435619735258726,
      "grad_norm": 5.887857437133789,
      "learning_rate": 0.0001994827648359653,
      "loss": 1.9345,
      "step": 31940
    },
    {
      "epoch": 3.844765342960289,
      "grad_norm": 4.972349166870117,
      "learning_rate": 0.00019948237833647888,
      "loss": 1.9818,
      "step": 31950
    },
    {
      "epoch": 3.845968712394705,
      "grad_norm": 5.498665809631348,
      "learning_rate": 0.0001994819916930169,
      "loss": 1.75,
      "step": 31960
    },
    {
      "epoch": 3.8471720818291217,
      "grad_norm": 4.5480637550354,
      "learning_rate": 0.0001994816049055799,
      "loss": 1.4933,
      "step": 31970
    },
    {
      "epoch": 3.848375451263538,
      "grad_norm": 7.472859859466553,
      "learning_rate": 0.00019948121797416844,
      "loss": 1.9189,
      "step": 31980
    },
    {
      "epoch": 3.849578820697954,
      "grad_norm": 6.643512725830078,
      "learning_rate": 0.00019948083089878313,
      "loss": 2.0863,
      "step": 31990
    },
    {
      "epoch": 3.850782190132371,
      "grad_norm": 4.6981635093688965,
      "learning_rate": 0.00019948044367942448,
      "loss": 2.0633,
      "step": 32000
    },
    {
      "epoch": 3.851985559566787,
      "grad_norm": 6.2199177742004395,
      "learning_rate": 0.00019948005631609307,
      "loss": 1.9438,
      "step": 32010
    },
    {
      "epoch": 3.8531889290012034,
      "grad_norm": 6.112263202667236,
      "learning_rate": 0.00019947966880878948,
      "loss": 1.9433,
      "step": 32020
    },
    {
      "epoch": 3.8543922984356196,
      "grad_norm": 7.201168060302734,
      "learning_rate": 0.00019947928115751423,
      "loss": 1.9611,
      "step": 32030
    },
    {
      "epoch": 3.855595667870036,
      "grad_norm": 5.691432952880859,
      "learning_rate": 0.00019947889336226787,
      "loss": 1.6684,
      "step": 32040
    },
    {
      "epoch": 3.8567990373044525,
      "grad_norm": 4.065805435180664,
      "learning_rate": 0.00019947850542305102,
      "loss": 1.7571,
      "step": 32050
    },
    {
      "epoch": 3.8580024067388687,
      "grad_norm": 6.2115702629089355,
      "learning_rate": 0.0001994781173398642,
      "loss": 1.9781,
      "step": 32060
    },
    {
      "epoch": 3.859205776173285,
      "grad_norm": 4.214781284332275,
      "learning_rate": 0.000199477729112708,
      "loss": 1.8943,
      "step": 32070
    },
    {
      "epoch": 3.8604091456077017,
      "grad_norm": 9.22004222869873,
      "learning_rate": 0.00019947734074158294,
      "loss": 1.7834,
      "step": 32080
    },
    {
      "epoch": 3.861612515042118,
      "grad_norm": 5.543321132659912,
      "learning_rate": 0.00019947695222648966,
      "loss": 1.9621,
      "step": 32090
    },
    {
      "epoch": 3.862815884476534,
      "grad_norm": 4.949936389923096,
      "learning_rate": 0.00019947656356742864,
      "loss": 1.9948,
      "step": 32100
    },
    {
      "epoch": 3.864019253910951,
      "grad_norm": 7.064183235168457,
      "learning_rate": 0.00019947617476440046,
      "loss": 2.0271,
      "step": 32110
    },
    {
      "epoch": 3.865222623345367,
      "grad_norm": 6.2395758628845215,
      "learning_rate": 0.0001994757858174057,
      "loss": 1.789,
      "step": 32120
    },
    {
      "epoch": 3.8664259927797833,
      "grad_norm": 7.911624431610107,
      "learning_rate": 0.00019947539672644492,
      "loss": 1.8299,
      "step": 32130
    },
    {
      "epoch": 3.8676293622142,
      "grad_norm": 5.5708770751953125,
      "learning_rate": 0.00019947500749151867,
      "loss": 1.6924,
      "step": 32140
    },
    {
      "epoch": 3.868832731648616,
      "grad_norm": 3.6271274089813232,
      "learning_rate": 0.00019947461811262756,
      "loss": 1.753,
      "step": 32150
    },
    {
      "epoch": 3.8700361010830324,
      "grad_norm": 5.830916404724121,
      "learning_rate": 0.00019947422858977208,
      "loss": 1.6217,
      "step": 32160
    },
    {
      "epoch": 3.871239470517449,
      "grad_norm": 4.291306972503662,
      "learning_rate": 0.00019947383892295284,
      "loss": 1.7472,
      "step": 32170
    },
    {
      "epoch": 3.8724428399518653,
      "grad_norm": 7.956405162811279,
      "learning_rate": 0.00019947344911217042,
      "loss": 1.6967,
      "step": 32180
    },
    {
      "epoch": 3.8736462093862816,
      "grad_norm": 5.54548454284668,
      "learning_rate": 0.00019947305915742533,
      "loss": 1.8514,
      "step": 32190
    },
    {
      "epoch": 3.874849578820698,
      "grad_norm": 4.318104267120361,
      "learning_rate": 0.00019947266905871817,
      "loss": 1.8138,
      "step": 32200
    },
    {
      "epoch": 3.8760529482551145,
      "grad_norm": 7.319823741912842,
      "learning_rate": 0.0001994722788160495,
      "loss": 1.8414,
      "step": 32210
    },
    {
      "epoch": 3.8772563176895307,
      "grad_norm": 4.983766078948975,
      "learning_rate": 0.0001994718884294199,
      "loss": 1.814,
      "step": 32220
    },
    {
      "epoch": 3.878459687123947,
      "grad_norm": 7.071497440338135,
      "learning_rate": 0.00019947149789882991,
      "loss": 2.0585,
      "step": 32230
    },
    {
      "epoch": 3.879663056558363,
      "grad_norm": 6.389927864074707,
      "learning_rate": 0.00019947110722428007,
      "loss": 1.6976,
      "step": 32240
    },
    {
      "epoch": 3.88086642599278,
      "grad_norm": 3.7990963459014893,
      "learning_rate": 0.00019947071640577104,
      "loss": 1.9019,
      "step": 32250
    },
    {
      "epoch": 3.882069795427196,
      "grad_norm": 6.669363498687744,
      "learning_rate": 0.0001994703254433033,
      "loss": 1.8831,
      "step": 32260
    },
    {
      "epoch": 3.8832731648616123,
      "grad_norm": 3.8320515155792236,
      "learning_rate": 0.00019946993433687742,
      "loss": 1.9731,
      "step": 32270
    },
    {
      "epoch": 3.884476534296029,
      "grad_norm": 8.193338394165039,
      "learning_rate": 0.000199469543086494,
      "loss": 1.8038,
      "step": 32280
    },
    {
      "epoch": 3.8856799037304453,
      "grad_norm": 4.97507905960083,
      "learning_rate": 0.0001994691516921536,
      "loss": 1.7103,
      "step": 32290
    },
    {
      "epoch": 3.8868832731648615,
      "grad_norm": 4.077975749969482,
      "learning_rate": 0.00019946876015385678,
      "loss": 1.7446,
      "step": 32300
    },
    {
      "epoch": 3.888086642599278,
      "grad_norm": 6.96763801574707,
      "learning_rate": 0.0001994683684716041,
      "loss": 1.7259,
      "step": 32310
    },
    {
      "epoch": 3.8892900120336944,
      "grad_norm": 4.485548496246338,
      "learning_rate": 0.00019946797664539614,
      "loss": 1.8251,
      "step": 32320
    },
    {
      "epoch": 3.8904933814681106,
      "grad_norm": 7.93310546875,
      "learning_rate": 0.00019946758467523346,
      "loss": 1.7753,
      "step": 32330
    },
    {
      "epoch": 3.8916967509025273,
      "grad_norm": 5.288084983825684,
      "learning_rate": 0.00019946719256111662,
      "loss": 1.9221,
      "step": 32340
    },
    {
      "epoch": 3.8929001203369435,
      "grad_norm": 4.369314193725586,
      "learning_rate": 0.0001994668003030462,
      "loss": 1.6287,
      "step": 32350
    },
    {
      "epoch": 3.89410348977136,
      "grad_norm": 5.8581390380859375,
      "learning_rate": 0.00019946640790102275,
      "loss": 1.7752,
      "step": 32360
    },
    {
      "epoch": 3.895306859205776,
      "grad_norm": 4.442620754241943,
      "learning_rate": 0.00019946601535504686,
      "loss": 1.8486,
      "step": 32370
    },
    {
      "epoch": 3.8965102286401927,
      "grad_norm": 8.320189476013184,
      "learning_rate": 0.00019946562266511908,
      "loss": 1.9129,
      "step": 32380
    },
    {
      "epoch": 3.897713598074609,
      "grad_norm": 5.990233898162842,
      "learning_rate": 0.00019946522983124002,
      "loss": 1.9838,
      "step": 32390
    },
    {
      "epoch": 3.898916967509025,
      "grad_norm": 5.078760623931885,
      "learning_rate": 0.00019946483685341016,
      "loss": 1.6586,
      "step": 32400
    },
    {
      "epoch": 3.9001203369434414,
      "grad_norm": 6.948725700378418,
      "learning_rate": 0.00019946444373163016,
      "loss": 1.8155,
      "step": 32410
    },
    {
      "epoch": 3.901323706377858,
      "grad_norm": 4.908626079559326,
      "learning_rate": 0.00019946405046590056,
      "loss": 1.7441,
      "step": 32420
    },
    {
      "epoch": 3.9025270758122743,
      "grad_norm": 7.935828685760498,
      "learning_rate": 0.00019946365705622192,
      "loss": 2.0591,
      "step": 32430
    },
    {
      "epoch": 3.9037304452466906,
      "grad_norm": 5.426856994628906,
      "learning_rate": 0.00019946326350259478,
      "loss": 1.7126,
      "step": 32440
    },
    {
      "epoch": 3.9049338146811072,
      "grad_norm": 3.5557751655578613,
      "learning_rate": 0.00019946286980501978,
      "loss": 1.756,
      "step": 32450
    },
    {
      "epoch": 3.9061371841155235,
      "grad_norm": 7.084009647369385,
      "learning_rate": 0.0001994624759634974,
      "loss": 1.8829,
      "step": 32460
    },
    {
      "epoch": 3.9073405535499397,
      "grad_norm": 5.085668563842773,
      "learning_rate": 0.0001994620819780283,
      "loss": 1.7786,
      "step": 32470
    },
    {
      "epoch": 3.9085439229843564,
      "grad_norm": 8.720808982849121,
      "learning_rate": 0.000199461687848613,
      "loss": 1.9302,
      "step": 32480
    },
    {
      "epoch": 3.9097472924187726,
      "grad_norm": 6.374059200286865,
      "learning_rate": 0.0001994612935752521,
      "loss": 1.9572,
      "step": 32490
    },
    {
      "epoch": 3.910950661853189,
      "grad_norm": 4.396307468414307,
      "learning_rate": 0.00019946089915794612,
      "loss": 1.9987,
      "step": 32500
    },
    {
      "epoch": 3.9121540312876055,
      "grad_norm": 7.3977203369140625,
      "learning_rate": 0.0001994605045966957,
      "loss": 1.9049,
      "step": 32510
    },
    {
      "epoch": 3.9133574007220218,
      "grad_norm": 4.642151355743408,
      "learning_rate": 0.00019946010989150135,
      "loss": 1.7011,
      "step": 32520
    },
    {
      "epoch": 3.914560770156438,
      "grad_norm": 7.033011436462402,
      "learning_rate": 0.00019945971504236367,
      "loss": 2.227,
      "step": 32530
    },
    {
      "epoch": 3.9157641395908542,
      "grad_norm": 6.167843341827393,
      "learning_rate": 0.00019945932004928322,
      "loss": 1.8048,
      "step": 32540
    },
    {
      "epoch": 3.916967509025271,
      "grad_norm": 5.016860485076904,
      "learning_rate": 0.0001994589249122606,
      "loss": 1.6254,
      "step": 32550
    },
    {
      "epoch": 3.918170878459687,
      "grad_norm": 6.315822124481201,
      "learning_rate": 0.00019945852963129635,
      "loss": 1.6375,
      "step": 32560
    },
    {
      "epoch": 3.9193742478941034,
      "grad_norm": 4.872376441955566,
      "learning_rate": 0.00019945813420639102,
      "loss": 1.7956,
      "step": 32570
    },
    {
      "epoch": 3.9205776173285196,
      "grad_norm": 7.527656555175781,
      "learning_rate": 0.00019945773863754525,
      "loss": 1.8707,
      "step": 32580
    },
    {
      "epoch": 3.9217809867629363,
      "grad_norm": 4.640732765197754,
      "learning_rate": 0.00019945734292475958,
      "loss": 1.7628,
      "step": 32590
    },
    {
      "epoch": 3.9229843561973525,
      "grad_norm": 3.8397715091705322,
      "learning_rate": 0.00019945694706803455,
      "loss": 1.894,
      "step": 32600
    },
    {
      "epoch": 3.9241877256317688,
      "grad_norm": 6.057290077209473,
      "learning_rate": 0.00019945655106737078,
      "loss": 1.8933,
      "step": 32610
    },
    {
      "epoch": 3.9253910950661854,
      "grad_norm": 4.693942546844482,
      "learning_rate": 0.00019945615492276882,
      "loss": 1.867,
      "step": 32620
    },
    {
      "epoch": 3.9265944645006017,
      "grad_norm": 7.134082794189453,
      "learning_rate": 0.00019945575863422924,
      "loss": 1.6996,
      "step": 32630
    },
    {
      "epoch": 3.927797833935018,
      "grad_norm": 6.2517900466918945,
      "learning_rate": 0.00019945536220175266,
      "loss": 1.9585,
      "step": 32640
    },
    {
      "epoch": 3.9290012033694346,
      "grad_norm": 3.822658061981201,
      "learning_rate": 0.0001994549656253396,
      "loss": 1.7689,
      "step": 32650
    },
    {
      "epoch": 3.930204572803851,
      "grad_norm": 8.290013313293457,
      "learning_rate": 0.0001994545689049906,
      "loss": 1.6267,
      "step": 32660
    },
    {
      "epoch": 3.931407942238267,
      "grad_norm": 5.099239826202393,
      "learning_rate": 0.00019945417204070637,
      "loss": 1.8065,
      "step": 32670
    },
    {
      "epoch": 3.9326113116726837,
      "grad_norm": 7.6024861335754395,
      "learning_rate": 0.00019945377503248731,
      "loss": 1.982,
      "step": 32680
    },
    {
      "epoch": 3.9338146811071,
      "grad_norm": 4.7708659172058105,
      "learning_rate": 0.00019945337788033416,
      "loss": 1.8551,
      "step": 32690
    },
    {
      "epoch": 3.935018050541516,
      "grad_norm": 4.781246185302734,
      "learning_rate": 0.00019945298058424737,
      "loss": 1.9533,
      "step": 32700
    },
    {
      "epoch": 3.9362214199759324,
      "grad_norm": 5.206472396850586,
      "learning_rate": 0.0001994525831442276,
      "loss": 1.8055,
      "step": 32710
    },
    {
      "epoch": 3.937424789410349,
      "grad_norm": 4.313125133514404,
      "learning_rate": 0.00019945218556027537,
      "loss": 1.8589,
      "step": 32720
    },
    {
      "epoch": 3.9386281588447654,
      "grad_norm": 10.546980857849121,
      "learning_rate": 0.00019945178783239124,
      "loss": 1.9307,
      "step": 32730
    },
    {
      "epoch": 3.9398315282791816,
      "grad_norm": 5.398140907287598,
      "learning_rate": 0.00019945138996057586,
      "loss": 1.8189,
      "step": 32740
    },
    {
      "epoch": 3.941034897713598,
      "grad_norm": 4.577977180480957,
      "learning_rate": 0.00019945099194482977,
      "loss": 1.9049,
      "step": 32750
    },
    {
      "epoch": 3.9422382671480145,
      "grad_norm": 6.033322334289551,
      "learning_rate": 0.00019945059378515353,
      "loss": 1.771,
      "step": 32760
    },
    {
      "epoch": 3.9434416365824307,
      "grad_norm": 5.217099189758301,
      "learning_rate": 0.00019945019548154772,
      "loss": 1.7072,
      "step": 32770
    },
    {
      "epoch": 3.944645006016847,
      "grad_norm": 9.646574974060059,
      "learning_rate": 0.0001994497970340129,
      "loss": 1.7588,
      "step": 32780
    },
    {
      "epoch": 3.9458483754512637,
      "grad_norm": 13.237478256225586,
      "learning_rate": 0.0001994493984425497,
      "loss": 1.8727,
      "step": 32790
    },
    {
      "epoch": 3.94705174488568,
      "grad_norm": 4.134167671203613,
      "learning_rate": 0.00019944899970715868,
      "loss": 1.9761,
      "step": 32800
    },
    {
      "epoch": 3.948255114320096,
      "grad_norm": 6.274007797241211,
      "learning_rate": 0.0001994486008278404,
      "loss": 1.9178,
      "step": 32810
    },
    {
      "epoch": 3.949458483754513,
      "grad_norm": 4.518835544586182,
      "learning_rate": 0.00019944820180459543,
      "loss": 1.7662,
      "step": 32820
    },
    {
      "epoch": 3.950661853188929,
      "grad_norm": 8.868928909301758,
      "learning_rate": 0.00019944780263742436,
      "loss": 1.9949,
      "step": 32830
    },
    {
      "epoch": 3.9518652226233453,
      "grad_norm": 5.864382743835449,
      "learning_rate": 0.00019944740332632774,
      "loss": 1.9119,
      "step": 32840
    },
    {
      "epoch": 3.953068592057762,
      "grad_norm": 4.512370586395264,
      "learning_rate": 0.00019944700387130622,
      "loss": 1.7882,
      "step": 32850
    },
    {
      "epoch": 3.954271961492178,
      "grad_norm": 5.2216715812683105,
      "learning_rate": 0.0001994466042723603,
      "loss": 1.9608,
      "step": 32860
    },
    {
      "epoch": 3.9554753309265944,
      "grad_norm": 5.1193461418151855,
      "learning_rate": 0.0001994462045294906,
      "loss": 1.8013,
      "step": 32870
    },
    {
      "epoch": 3.956678700361011,
      "grad_norm": 9.372982025146484,
      "learning_rate": 0.0001994458046426977,
      "loss": 1.7889,
      "step": 32880
    },
    {
      "epoch": 3.9578820697954273,
      "grad_norm": 6.510072231292725,
      "learning_rate": 0.00019944540461198216,
      "loss": 1.6052,
      "step": 32890
    },
    {
      "epoch": 3.9590854392298436,
      "grad_norm": 4.168869495391846,
      "learning_rate": 0.00019944500443734456,
      "loss": 1.6398,
      "step": 32900
    },
    {
      "epoch": 3.96028880866426,
      "grad_norm": 6.725529193878174,
      "learning_rate": 0.00019944460411878552,
      "loss": 1.9307,
      "step": 32910
    },
    {
      "epoch": 3.961492178098676,
      "grad_norm": 8.65323543548584,
      "learning_rate": 0.00019944420365630556,
      "loss": 1.6071,
      "step": 32920
    },
    {
      "epoch": 3.9626955475330927,
      "grad_norm": 6.908558368682861,
      "learning_rate": 0.0001994438030499053,
      "loss": 2.0262,
      "step": 32930
    },
    {
      "epoch": 3.963898916967509,
      "grad_norm": 5.8963398933410645,
      "learning_rate": 0.00019944340229958527,
      "loss": 1.6929,
      "step": 32940
    },
    {
      "epoch": 3.965102286401925,
      "grad_norm": 4.276820659637451,
      "learning_rate": 0.0001994430014053461,
      "loss": 1.9331,
      "step": 32950
    },
    {
      "epoch": 3.966305655836342,
      "grad_norm": 7.858771324157715,
      "learning_rate": 0.00019944260036718837,
      "loss": 1.7647,
      "step": 32960
    },
    {
      "epoch": 3.967509025270758,
      "grad_norm": 5.020040035247803,
      "learning_rate": 0.00019944219918511264,
      "loss": 1.7347,
      "step": 32970
    },
    {
      "epoch": 3.9687123947051743,
      "grad_norm": 8.482172966003418,
      "learning_rate": 0.0001994417978591195,
      "loss": 1.8629,
      "step": 32980
    },
    {
      "epoch": 3.969915764139591,
      "grad_norm": 4.87166166305542,
      "learning_rate": 0.0001994413963892095,
      "loss": 1.8508,
      "step": 32990
    },
    {
      "epoch": 3.9711191335740073,
      "grad_norm": 4.907790184020996,
      "learning_rate": 0.0001994409947753833,
      "loss": 1.8536,
      "step": 33000
    },
    {
      "epoch": 3.9723225030084235,
      "grad_norm": 6.560929298400879,
      "learning_rate": 0.00019944059301764137,
      "loss": 1.9947,
      "step": 33010
    },
    {
      "epoch": 3.97352587244284,
      "grad_norm": 4.404400825500488,
      "learning_rate": 0.0001994401911159844,
      "loss": 1.9147,
      "step": 33020
    },
    {
      "epoch": 3.9747292418772564,
      "grad_norm": 8.753676414489746,
      "learning_rate": 0.0001994397890704129,
      "loss": 1.8392,
      "step": 33030
    },
    {
      "epoch": 3.9759326113116726,
      "grad_norm": 5.625767707824707,
      "learning_rate": 0.00019943938688092748,
      "loss": 1.7824,
      "step": 33040
    },
    {
      "epoch": 3.9771359807460893,
      "grad_norm": 4.419516563415527,
      "learning_rate": 0.0001994389845475287,
      "loss": 1.8224,
      "step": 33050
    },
    {
      "epoch": 3.9783393501805056,
      "grad_norm": 6.591396331787109,
      "learning_rate": 0.00019943858207021716,
      "loss": 1.9277,
      "step": 33060
    },
    {
      "epoch": 3.979542719614922,
      "grad_norm": 5.4560866355896,
      "learning_rate": 0.00019943817944899345,
      "loss": 2.0471,
      "step": 33070
    },
    {
      "epoch": 3.980746089049338,
      "grad_norm": 10.069039344787598,
      "learning_rate": 0.00019943777668385816,
      "loss": 1.9399,
      "step": 33080
    },
    {
      "epoch": 3.9819494584837543,
      "grad_norm": 6.393735408782959,
      "learning_rate": 0.00019943737377481186,
      "loss": 1.9805,
      "step": 33090
    },
    {
      "epoch": 3.983152827918171,
      "grad_norm": 5.489099502563477,
      "learning_rate": 0.0001994369707218551,
      "loss": 1.8353,
      "step": 33100
    },
    {
      "epoch": 3.984356197352587,
      "grad_norm": 6.359148025512695,
      "learning_rate": 0.0001994365675249885,
      "loss": 1.9273,
      "step": 33110
    },
    {
      "epoch": 3.9855595667870034,
      "grad_norm": 4.177426338195801,
      "learning_rate": 0.00019943616418421266,
      "loss": 1.8576,
      "step": 33120
    },
    {
      "epoch": 3.98676293622142,
      "grad_norm": 9.475194931030273,
      "learning_rate": 0.0001994357606995281,
      "loss": 1.9668,
      "step": 33130
    },
    {
      "epoch": 3.9879663056558363,
      "grad_norm": 8.060712814331055,
      "learning_rate": 0.00019943535707093549,
      "loss": 1.894,
      "step": 33140
    },
    {
      "epoch": 3.9891696750902526,
      "grad_norm": 3.6547257900238037,
      "learning_rate": 0.00019943495329843532,
      "loss": 1.9538,
      "step": 33150
    },
    {
      "epoch": 3.9903730445246692,
      "grad_norm": 7.021988391876221,
      "learning_rate": 0.00019943454938202825,
      "loss": 1.8346,
      "step": 33160
    },
    {
      "epoch": 3.9915764139590855,
      "grad_norm": 5.056446552276611,
      "learning_rate": 0.00019943414532171488,
      "loss": 1.8549,
      "step": 33170
    },
    {
      "epoch": 3.9927797833935017,
      "grad_norm": 10.918617248535156,
      "learning_rate": 0.0001994337411174957,
      "loss": 1.8216,
      "step": 33180
    },
    {
      "epoch": 3.9939831528279184,
      "grad_norm": 5.906192302703857,
      "learning_rate": 0.00019943333676937134,
      "loss": 1.8366,
      "step": 33190
    },
    {
      "epoch": 3.9951865222623346,
      "grad_norm": 3.4675395488739014,
      "learning_rate": 0.0001994329322773424,
      "loss": 1.9512,
      "step": 33200
    },
    {
      "epoch": 3.996389891696751,
      "grad_norm": 6.853846549987793,
      "learning_rate": 0.00019943252764140947,
      "loss": 1.8738,
      "step": 33210
    },
    {
      "epoch": 3.9975932611311675,
      "grad_norm": 4.8922648429870605,
      "learning_rate": 0.0001994321228615731,
      "loss": 1.7699,
      "step": 33220
    },
    {
      "epoch": 3.9987966305655838,
      "grad_norm": 6.658156871795654,
      "learning_rate": 0.00019943171793783396,
      "loss": 1.9007,
      "step": 33230
    },
    {
      "epoch": 4.0,
      "grad_norm": 6.3400421142578125,
      "learning_rate": 0.00019943131287019253,
      "loss": 1.5921,
      "step": 33240
    },
    {
      "epoch": 4.0,
      "eval_loss": 1.8273708820343018,
      "eval_runtime": 120.2867,
      "eval_samples_per_second": 61.412,
      "eval_steps_per_second": 7.682,
      "step": 33240
    },
    {
      "epoch": 4.001203369434417,
      "grad_norm": 6.284422874450684,
      "learning_rate": 0.00019943090765864945,
      "loss": 1.7291,
      "step": 33250
    },
    {
      "epoch": 4.0024067388688325,
      "grad_norm": 4.046846389770508,
      "learning_rate": 0.00019943050230320528,
      "loss": 1.8244,
      "step": 33260
    },
    {
      "epoch": 4.003610108303249,
      "grad_norm": 6.676296710968018,
      "learning_rate": 0.00019943009680386062,
      "loss": 1.7329,
      "step": 33270
    },
    {
      "epoch": 4.004813477737666,
      "grad_norm": 4.912373065948486,
      "learning_rate": 0.0001994296911606161,
      "loss": 1.4706,
      "step": 33280
    },
    {
      "epoch": 4.006016847172082,
      "grad_norm": 7.784985542297363,
      "learning_rate": 0.00019942928537347224,
      "loss": 1.6247,
      "step": 33290
    },
    {
      "epoch": 4.007220216606498,
      "grad_norm": 6.531484603881836,
      "learning_rate": 0.00019942887944242966,
      "loss": 1.6704,
      "step": 33300
    },
    {
      "epoch": 4.008423586040915,
      "grad_norm": 4.292130947113037,
      "learning_rate": 0.00019942847336748895,
      "loss": 1.5558,
      "step": 33310
    },
    {
      "epoch": 4.009626955475331,
      "grad_norm": 6.882789134979248,
      "learning_rate": 0.0001994280671486507,
      "loss": 1.9312,
      "step": 33320
    },
    {
      "epoch": 4.0108303249097474,
      "grad_norm": 5.988173961639404,
      "learning_rate": 0.00019942766078591547,
      "loss": 1.6523,
      "step": 33330
    },
    {
      "epoch": 4.012033694344163,
      "grad_norm": 11.171236038208008,
      "learning_rate": 0.0001994272542792839,
      "loss": 1.7468,
      "step": 33340
    },
    {
      "epoch": 4.01323706377858,
      "grad_norm": 4.555514812469482,
      "learning_rate": 0.0001994268476287565,
      "loss": 1.5814,
      "step": 33350
    },
    {
      "epoch": 4.014440433212997,
      "grad_norm": 4.666054725646973,
      "learning_rate": 0.0001994264408343339,
      "loss": 1.5923,
      "step": 33360
    },
    {
      "epoch": 4.015643802647412,
      "grad_norm": 6.363561630249023,
      "learning_rate": 0.00019942603389601671,
      "loss": 1.9146,
      "step": 33370
    },
    {
      "epoch": 4.016847172081829,
      "grad_norm": 5.004948139190674,
      "learning_rate": 0.00019942562681380554,
      "loss": 1.7598,
      "step": 33380
    },
    {
      "epoch": 4.018050541516246,
      "grad_norm": 9.599754333496094,
      "learning_rate": 0.0001994252195877009,
      "loss": 1.8804,
      "step": 33390
    },
    {
      "epoch": 4.0192539109506615,
      "grad_norm": 5.644397735595703,
      "learning_rate": 0.0001994248122177034,
      "loss": 1.6345,
      "step": 33400
    },
    {
      "epoch": 4.020457280385078,
      "grad_norm": 7.430975437164307,
      "learning_rate": 0.00019942440470381372,
      "loss": 1.9448,
      "step": 33410
    },
    {
      "epoch": 4.021660649819495,
      "grad_norm": 10.672667503356934,
      "learning_rate": 0.00019942399704603233,
      "loss": 1.8085,
      "step": 33420
    },
    {
      "epoch": 4.022864019253911,
      "grad_norm": 5.232245922088623,
      "learning_rate": 0.00019942358924435985,
      "loss": 1.5639,
      "step": 33430
    },
    {
      "epoch": 4.024067388688327,
      "grad_norm": 7.8057355880737305,
      "learning_rate": 0.00019942318129879694,
      "loss": 1.7672,
      "step": 33440
    },
    {
      "epoch": 4.025270758122744,
      "grad_norm": 8.06316089630127,
      "learning_rate": 0.00019942277320934412,
      "loss": 1.616,
      "step": 33450
    },
    {
      "epoch": 4.02647412755716,
      "grad_norm": 7.106293678283691,
      "learning_rate": 0.000199422364976002,
      "loss": 1.698,
      "step": 33460
    },
    {
      "epoch": 4.0276774969915765,
      "grad_norm": 7.68731689453125,
      "learning_rate": 0.00019942195659877114,
      "loss": 1.8209,
      "step": 33470
    },
    {
      "epoch": 4.028880866425993,
      "grad_norm": 5.684768199920654,
      "learning_rate": 0.0001994215480776522,
      "loss": 1.5607,
      "step": 33480
    },
    {
      "epoch": 4.030084235860409,
      "grad_norm": 8.300897598266602,
      "learning_rate": 0.00019942113941264573,
      "loss": 1.6034,
      "step": 33490
    },
    {
      "epoch": 4.031287605294826,
      "grad_norm": 6.523977756500244,
      "learning_rate": 0.0001994207306037523,
      "loss": 1.6377,
      "step": 33500
    },
    {
      "epoch": 4.0324909747292415,
      "grad_norm": 4.503218650817871,
      "learning_rate": 0.00019942032165097255,
      "loss": 1.7049,
      "step": 33510
    },
    {
      "epoch": 4.033694344163658,
      "grad_norm": 10.025880813598633,
      "learning_rate": 0.00019941991255430702,
      "loss": 1.608,
      "step": 33520
    },
    {
      "epoch": 4.034897713598075,
      "grad_norm": 4.286937236785889,
      "learning_rate": 0.0001994195033137564,
      "loss": 1.7023,
      "step": 33530
    },
    {
      "epoch": 4.036101083032491,
      "grad_norm": 9.422300338745117,
      "learning_rate": 0.00019941909392932113,
      "loss": 1.7398,
      "step": 33540
    },
    {
      "epoch": 4.037304452466907,
      "grad_norm": 7.26862907409668,
      "learning_rate": 0.00019941868440100194,
      "loss": 1.8983,
      "step": 33550
    },
    {
      "epoch": 4.038507821901324,
      "grad_norm": 4.133803367614746,
      "learning_rate": 0.00019941827472879934,
      "loss": 1.4949,
      "step": 33560
    },
    {
      "epoch": 4.03971119133574,
      "grad_norm": 7.196639537811279,
      "learning_rate": 0.00019941786491271395,
      "loss": 1.7609,
      "step": 33570
    },
    {
      "epoch": 4.040914560770156,
      "grad_norm": 4.263759613037109,
      "learning_rate": 0.00019941745495274637,
      "loss": 1.4943,
      "step": 33580
    },
    {
      "epoch": 4.042117930204573,
      "grad_norm": 8.759932518005371,
      "learning_rate": 0.0001994170448488972,
      "loss": 1.7811,
      "step": 33590
    },
    {
      "epoch": 4.043321299638989,
      "grad_norm": 4.935215473175049,
      "learning_rate": 0.00019941663460116703,
      "loss": 1.698,
      "step": 33600
    },
    {
      "epoch": 4.044524669073406,
      "grad_norm": 5.62227725982666,
      "learning_rate": 0.00019941622420955644,
      "loss": 1.5783,
      "step": 33610
    },
    {
      "epoch": 4.045728038507822,
      "grad_norm": 5.987107276916504,
      "learning_rate": 0.000199415813674066,
      "loss": 1.7969,
      "step": 33620
    },
    {
      "epoch": 4.046931407942238,
      "grad_norm": 6.0525360107421875,
      "learning_rate": 0.00019941540299469633,
      "loss": 1.7639,
      "step": 33630
    },
    {
      "epoch": 4.048134777376655,
      "grad_norm": 9.477378845214844,
      "learning_rate": 0.00019941499217144805,
      "loss": 1.5933,
      "step": 33640
    },
    {
      "epoch": 4.049338146811071,
      "grad_norm": 6.63507080078125,
      "learning_rate": 0.00019941458120432173,
      "loss": 1.4568,
      "step": 33650
    },
    {
      "epoch": 4.050541516245487,
      "grad_norm": 4.016391754150391,
      "learning_rate": 0.00019941417009331796,
      "loss": 1.6234,
      "step": 33660
    },
    {
      "epoch": 4.051744885679904,
      "grad_norm": 8.432229995727539,
      "learning_rate": 0.00019941375883843736,
      "loss": 1.5321,
      "step": 33670
    },
    {
      "epoch": 4.05294825511432,
      "grad_norm": 6.387386322021484,
      "learning_rate": 0.00019941334743968047,
      "loss": 1.6318,
      "step": 33680
    },
    {
      "epoch": 4.054151624548736,
      "grad_norm": 9.522067070007324,
      "learning_rate": 0.00019941293589704798,
      "loss": 1.8896,
      "step": 33690
    },
    {
      "epoch": 4.055354993983153,
      "grad_norm": 6.084255695343018,
      "learning_rate": 0.00019941252421054037,
      "loss": 1.6872,
      "step": 33700
    },
    {
      "epoch": 4.056558363417569,
      "grad_norm": 4.2610955238342285,
      "learning_rate": 0.00019941211238015831,
      "loss": 1.513,
      "step": 33710
    },
    {
      "epoch": 4.0577617328519855,
      "grad_norm": 6.165771961212158,
      "learning_rate": 0.0001994117004059024,
      "loss": 1.8259,
      "step": 33720
    },
    {
      "epoch": 4.058965102286402,
      "grad_norm": 6.202225685119629,
      "learning_rate": 0.0001994112882877732,
      "loss": 1.3925,
      "step": 33730
    },
    {
      "epoch": 4.060168471720818,
      "grad_norm": 6.500694751739502,
      "learning_rate": 0.0001994108760257713,
      "loss": 1.6529,
      "step": 33740
    },
    {
      "epoch": 4.061371841155235,
      "grad_norm": 4.840877532958984,
      "learning_rate": 0.00019941046361989732,
      "loss": 1.6744,
      "step": 33750
    },
    {
      "epoch": 4.062575210589651,
      "grad_norm": 6.332531452178955,
      "learning_rate": 0.0001994100510701519,
      "loss": 1.9234,
      "step": 33760
    },
    {
      "epoch": 4.063778580024067,
      "grad_norm": 5.797792911529541,
      "learning_rate": 0.00019940963837653555,
      "loss": 1.5065,
      "step": 33770
    },
    {
      "epoch": 4.064981949458484,
      "grad_norm": 5.591522693634033,
      "learning_rate": 0.00019940922553904895,
      "loss": 1.6815,
      "step": 33780
    },
    {
      "epoch": 4.0661853188929005,
      "grad_norm": 9.232203483581543,
      "learning_rate": 0.00019940881255769264,
      "loss": 1.9034,
      "step": 33790
    },
    {
      "epoch": 4.067388688327316,
      "grad_norm": 7.283100128173828,
      "learning_rate": 0.0001994083994324672,
      "loss": 1.5855,
      "step": 33800
    },
    {
      "epoch": 4.068592057761733,
      "grad_norm": 3.7003464698791504,
      "learning_rate": 0.00019940798616337332,
      "loss": 1.6843,
      "step": 33810
    },
    {
      "epoch": 4.06979542719615,
      "grad_norm": 5.79300594329834,
      "learning_rate": 0.0001994075727504115,
      "loss": 1.8349,
      "step": 33820
    },
    {
      "epoch": 4.070998796630565,
      "grad_norm": 5.004201889038086,
      "learning_rate": 0.00019940715919358239,
      "loss": 1.8159,
      "step": 33830
    },
    {
      "epoch": 4.072202166064982,
      "grad_norm": 6.606564521789551,
      "learning_rate": 0.0001994067454928866,
      "loss": 1.8115,
      "step": 33840
    },
    {
      "epoch": 4.073405535499398,
      "grad_norm": 4.895559310913086,
      "learning_rate": 0.00019940633164832464,
      "loss": 1.8092,
      "step": 33850
    },
    {
      "epoch": 4.074608904933815,
      "grad_norm": 3.183698892593384,
      "learning_rate": 0.00019940591765989725,
      "loss": 1.7114,
      "step": 33860
    },
    {
      "epoch": 4.075812274368231,
      "grad_norm": 6.283841609954834,
      "learning_rate": 0.0001994055035276049,
      "loss": 1.5288,
      "step": 33870
    },
    {
      "epoch": 4.077015643802647,
      "grad_norm": 3.9874267578125,
      "learning_rate": 0.0001994050892514483,
      "loss": 1.8218,
      "step": 33880
    },
    {
      "epoch": 4.078219013237064,
      "grad_norm": 8.361854553222656,
      "learning_rate": 0.00019940467483142794,
      "loss": 1.7652,
      "step": 33890
    },
    {
      "epoch": 4.07942238267148,
      "grad_norm": 6.726755142211914,
      "learning_rate": 0.0001994042602675445,
      "loss": 1.6765,
      "step": 33900
    },
    {
      "epoch": 4.080625752105896,
      "grad_norm": 4.079782009124756,
      "learning_rate": 0.00019940384555979857,
      "loss": 1.7714,
      "step": 33910
    },
    {
      "epoch": 4.081829121540313,
      "grad_norm": 7.918765544891357,
      "learning_rate": 0.00019940343070819073,
      "loss": 1.7044,
      "step": 33920
    },
    {
      "epoch": 4.0830324909747295,
      "grad_norm": 5.073817253112793,
      "learning_rate": 0.00019940301571272156,
      "loss": 1.8873,
      "step": 33930
    },
    {
      "epoch": 4.084235860409145,
      "grad_norm": 7.854415416717529,
      "learning_rate": 0.0001994026005733917,
      "loss": 1.8234,
      "step": 33940
    },
    {
      "epoch": 4.085439229843562,
      "grad_norm": 9.638435363769531,
      "learning_rate": 0.00019940218529020172,
      "loss": 1.8032,
      "step": 33950
    },
    {
      "epoch": 4.086642599277979,
      "grad_norm": 5.930091857910156,
      "learning_rate": 0.00019940176986315224,
      "loss": 1.7172,
      "step": 33960
    },
    {
      "epoch": 4.0878459687123945,
      "grad_norm": 7.531289100646973,
      "learning_rate": 0.00019940135429224386,
      "loss": 1.9137,
      "step": 33970
    },
    {
      "epoch": 4.089049338146811,
      "grad_norm": 5.307153224945068,
      "learning_rate": 0.00019940093857747718,
      "loss": 1.832,
      "step": 33980
    },
    {
      "epoch": 4.090252707581228,
      "grad_norm": 7.0909905433654785,
      "learning_rate": 0.00019940052271885282,
      "loss": 1.6379,
      "step": 33990
    },
    {
      "epoch": 4.091456077015644,
      "grad_norm": 5.340932846069336,
      "learning_rate": 0.00019940010671637133,
      "loss": 1.6154,
      "step": 34000
    },
    {
      "epoch": 4.09265944645006,
      "grad_norm": 5.438321113586426,
      "learning_rate": 0.00019939969057003335,
      "loss": 1.9888,
      "step": 34010
    },
    {
      "epoch": 4.093862815884476,
      "grad_norm": 5.239195346832275,
      "learning_rate": 0.0001993992742798395,
      "loss": 1.4993,
      "step": 34020
    },
    {
      "epoch": 4.095066185318893,
      "grad_norm": 4.424549102783203,
      "learning_rate": 0.00019939885784579032,
      "loss": 1.7697,
      "step": 34030
    },
    {
      "epoch": 4.0962695547533094,
      "grad_norm": 8.237585067749023,
      "learning_rate": 0.00019939844126788648,
      "loss": 1.8262,
      "step": 34040
    },
    {
      "epoch": 4.097472924187725,
      "grad_norm": 6.078985214233398,
      "learning_rate": 0.00019939802454612858,
      "loss": 1.6547,
      "step": 34050
    },
    {
      "epoch": 4.098676293622142,
      "grad_norm": 4.0947394371032715,
      "learning_rate": 0.00019939760768051713,
      "loss": 1.769,
      "step": 34060
    },
    {
      "epoch": 4.099879663056559,
      "grad_norm": 7.274417877197266,
      "learning_rate": 0.00019939719067105287,
      "loss": 1.663,
      "step": 34070
    },
    {
      "epoch": 4.101083032490974,
      "grad_norm": 4.327084064483643,
      "learning_rate": 0.00019939677351773629,
      "loss": 1.7496,
      "step": 34080
    },
    {
      "epoch": 4.102286401925391,
      "grad_norm": 8.57921314239502,
      "learning_rate": 0.00019939635622056806,
      "loss": 1.6565,
      "step": 34090
    },
    {
      "epoch": 4.103489771359808,
      "grad_norm": 5.8311028480529785,
      "learning_rate": 0.00019939593877954875,
      "loss": 1.6551,
      "step": 34100
    },
    {
      "epoch": 4.1046931407942235,
      "grad_norm": 4.009025573730469,
      "learning_rate": 0.00019939552119467897,
      "loss": 1.8955,
      "step": 34110
    },
    {
      "epoch": 4.10589651022864,
      "grad_norm": 6.618770599365234,
      "learning_rate": 0.00019939510346595935,
      "loss": 1.5865,
      "step": 34120
    },
    {
      "epoch": 4.107099879663057,
      "grad_norm": 3.555020332336426,
      "learning_rate": 0.00019939468559339047,
      "loss": 1.6269,
      "step": 34130
    },
    {
      "epoch": 4.108303249097473,
      "grad_norm": 7.425025463104248,
      "learning_rate": 0.00019939426757697293,
      "loss": 1.7053,
      "step": 34140
    },
    {
      "epoch": 4.109506618531889,
      "grad_norm": 6.488142967224121,
      "learning_rate": 0.00019939384941670735,
      "loss": 1.8724,
      "step": 34150
    },
    {
      "epoch": 4.110709987966306,
      "grad_norm": 3.57918643951416,
      "learning_rate": 0.00019939343111259435,
      "loss": 1.6016,
      "step": 34160
    },
    {
      "epoch": 4.111913357400722,
      "grad_norm": 6.3030924797058105,
      "learning_rate": 0.00019939301266463447,
      "loss": 1.5915,
      "step": 34170
    },
    {
      "epoch": 4.1131167268351385,
      "grad_norm": 4.9697418212890625,
      "learning_rate": 0.00019939259407282842,
      "loss": 1.6447,
      "step": 34180
    },
    {
      "epoch": 4.114320096269555,
      "grad_norm": 7.776681900024414,
      "learning_rate": 0.00019939217533717672,
      "loss": 1.8343,
      "step": 34190
    },
    {
      "epoch": 4.115523465703971,
      "grad_norm": 4.591990947723389,
      "learning_rate": 0.00019939175645768,
      "loss": 1.4935,
      "step": 34200
    },
    {
      "epoch": 4.116726835138388,
      "grad_norm": 4.370880603790283,
      "learning_rate": 0.00019939133743433887,
      "loss": 1.8305,
      "step": 34210
    },
    {
      "epoch": 4.1179302045728035,
      "grad_norm": 6.70311975479126,
      "learning_rate": 0.00019939091826715394,
      "loss": 1.5805,
      "step": 34220
    },
    {
      "epoch": 4.11913357400722,
      "grad_norm": 5.470571994781494,
      "learning_rate": 0.0001993904989561258,
      "loss": 1.6937,
      "step": 34230
    },
    {
      "epoch": 4.120336943441637,
      "grad_norm": 7.2369160652160645,
      "learning_rate": 0.00019939007950125508,
      "loss": 1.8691,
      "step": 34240
    },
    {
      "epoch": 4.121540312876053,
      "grad_norm": 5.476625919342041,
      "learning_rate": 0.0001993896599025424,
      "loss": 1.7346,
      "step": 34250
    },
    {
      "epoch": 4.122743682310469,
      "grad_norm": 3.4230034351348877,
      "learning_rate": 0.0001993892401599883,
      "loss": 2.0286,
      "step": 34260
    },
    {
      "epoch": 4.123947051744886,
      "grad_norm": 6.428805351257324,
      "learning_rate": 0.00019938882027359346,
      "loss": 1.7459,
      "step": 34270
    },
    {
      "epoch": 4.125150421179302,
      "grad_norm": 5.378191947937012,
      "learning_rate": 0.00019938840024335844,
      "loss": 1.7601,
      "step": 34280
    },
    {
      "epoch": 4.126353790613718,
      "grad_norm": 11.22222900390625,
      "learning_rate": 0.00019938798006928385,
      "loss": 1.7118,
      "step": 34290
    },
    {
      "epoch": 4.127557160048135,
      "grad_norm": 21.66954803466797,
      "learning_rate": 0.00019938755975137034,
      "loss": 1.7575,
      "step": 34300
    },
    {
      "epoch": 4.128760529482551,
      "grad_norm": 4.3829026222229,
      "learning_rate": 0.0001993871392896185,
      "loss": 1.903,
      "step": 34310
    },
    {
      "epoch": 4.129963898916968,
      "grad_norm": 6.230332374572754,
      "learning_rate": 0.00019938671868402895,
      "loss": 1.6153,
      "step": 34320
    },
    {
      "epoch": 4.131167268351384,
      "grad_norm": 4.583105564117432,
      "learning_rate": 0.00019938629793460225,
      "loss": 1.7303,
      "step": 34330
    },
    {
      "epoch": 4.1323706377858,
      "grad_norm": 7.515890598297119,
      "learning_rate": 0.00019938587704133905,
      "loss": 1.6951,
      "step": 34340
    },
    {
      "epoch": 4.133574007220217,
      "grad_norm": 4.609417915344238,
      "learning_rate": 0.00019938545600423993,
      "loss": 1.6442,
      "step": 34350
    },
    {
      "epoch": 4.1347773766546325,
      "grad_norm": 4.643413066864014,
      "learning_rate": 0.00019938503482330554,
      "loss": 1.6237,
      "step": 34360
    },
    {
      "epoch": 4.135980746089049,
      "grad_norm": 5.234893798828125,
      "learning_rate": 0.00019938461349853644,
      "loss": 1.7295,
      "step": 34370
    },
    {
      "epoch": 4.137184115523466,
      "grad_norm": 4.667137145996094,
      "learning_rate": 0.00019938419202993328,
      "loss": 1.6193,
      "step": 34380
    },
    {
      "epoch": 4.138387484957882,
      "grad_norm": 9.850410461425781,
      "learning_rate": 0.00019938377041749664,
      "loss": 1.787,
      "step": 34390
    },
    {
      "epoch": 4.139590854392298,
      "grad_norm": 4.949896335601807,
      "learning_rate": 0.0001993833486612272,
      "loss": 1.5774,
      "step": 34400
    },
    {
      "epoch": 4.140794223826715,
      "grad_norm": 4.389410495758057,
      "learning_rate": 0.00019938292676112548,
      "loss": 1.9383,
      "step": 34410
    },
    {
      "epoch": 4.141997593261131,
      "grad_norm": 6.364817142486572,
      "learning_rate": 0.00019938250471719212,
      "loss": 1.7452,
      "step": 34420
    },
    {
      "epoch": 4.1432009626955475,
      "grad_norm": 5.512551307678223,
      "learning_rate": 0.00019938208252942777,
      "loss": 1.8827,
      "step": 34430
    },
    {
      "epoch": 4.144404332129964,
      "grad_norm": 7.060027599334717,
      "learning_rate": 0.00019938166019783297,
      "loss": 1.6919,
      "step": 34440
    },
    {
      "epoch": 4.14560770156438,
      "grad_norm": 10.936914443969727,
      "learning_rate": 0.00019938123772240843,
      "loss": 1.6053,
      "step": 34450
    },
    {
      "epoch": 4.146811070998797,
      "grad_norm": 5.33294677734375,
      "learning_rate": 0.00019938081510315466,
      "loss": 1.8714,
      "step": 34460
    },
    {
      "epoch": 4.148014440433213,
      "grad_norm": 7.198515892028809,
      "learning_rate": 0.0001993803923400723,
      "loss": 1.8982,
      "step": 34470
    },
    {
      "epoch": 4.149217809867629,
      "grad_norm": 5.331077575683594,
      "learning_rate": 0.000199379969433162,
      "loss": 1.5221,
      "step": 34480
    },
    {
      "epoch": 4.150421179302046,
      "grad_norm": 9.042343139648438,
      "learning_rate": 0.00019937954638242435,
      "loss": 1.8686,
      "step": 34490
    },
    {
      "epoch": 4.1516245487364625,
      "grad_norm": 5.276646137237549,
      "learning_rate": 0.00019937912318785994,
      "loss": 1.5271,
      "step": 34500
    },
    {
      "epoch": 4.152827918170878,
      "grad_norm": 3.9702565670013428,
      "learning_rate": 0.00019937869984946942,
      "loss": 1.644,
      "step": 34510
    },
    {
      "epoch": 4.154031287605295,
      "grad_norm": 7.433538436889648,
      "learning_rate": 0.00019937827636725338,
      "loss": 1.6832,
      "step": 34520
    },
    {
      "epoch": 4.155234657039712,
      "grad_norm": 7.242001533508301,
      "learning_rate": 0.00019937785274121245,
      "loss": 1.6214,
      "step": 34530
    },
    {
      "epoch": 4.156438026474127,
      "grad_norm": 8.90034294128418,
      "learning_rate": 0.00019937742897134723,
      "loss": 1.6325,
      "step": 34540
    },
    {
      "epoch": 4.157641395908544,
      "grad_norm": 6.180896759033203,
      "learning_rate": 0.00019937700505765832,
      "loss": 1.7317,
      "step": 34550
    },
    {
      "epoch": 4.15884476534296,
      "grad_norm": 5.105179786682129,
      "learning_rate": 0.00019937658100014635,
      "loss": 1.8098,
      "step": 34560
    },
    {
      "epoch": 4.160048134777377,
      "grad_norm": 8.221943855285645,
      "learning_rate": 0.00019937615679881194,
      "loss": 1.7314,
      "step": 34570
    },
    {
      "epoch": 4.161251504211793,
      "grad_norm": 5.7328877449035645,
      "learning_rate": 0.0001993757324536557,
      "loss": 1.7452,
      "step": 34580
    },
    {
      "epoch": 4.162454873646209,
      "grad_norm": 12.69345760345459,
      "learning_rate": 0.00019937530796467824,
      "loss": 1.7706,
      "step": 34590
    },
    {
      "epoch": 4.163658243080626,
      "grad_norm": 6.551113128662109,
      "learning_rate": 0.00019937488333188018,
      "loss": 1.844,
      "step": 34600
    },
    {
      "epoch": 4.164861612515042,
      "grad_norm": 3.889042615890503,
      "learning_rate": 0.0001993744585552621,
      "loss": 1.6707,
      "step": 34610
    },
    {
      "epoch": 4.166064981949458,
      "grad_norm": 6.84438419342041,
      "learning_rate": 0.00019937403363482465,
      "loss": 1.6215,
      "step": 34620
    },
    {
      "epoch": 4.167268351383875,
      "grad_norm": 4.140308380126953,
      "learning_rate": 0.00019937360857056845,
      "loss": 1.8281,
      "step": 34630
    },
    {
      "epoch": 4.1684717208182915,
      "grad_norm": 9.280033111572266,
      "learning_rate": 0.0001993731833624941,
      "loss": 1.8555,
      "step": 34640
    },
    {
      "epoch": 4.169675090252707,
      "grad_norm": 5.941489219665527,
      "learning_rate": 0.00019937275801060225,
      "loss": 1.6224,
      "step": 34650
    },
    {
      "epoch": 4.170878459687124,
      "grad_norm": 3.7432074546813965,
      "learning_rate": 0.00019937233251489346,
      "loss": 1.7454,
      "step": 34660
    },
    {
      "epoch": 4.172081829121541,
      "grad_norm": 6.81998872756958,
      "learning_rate": 0.00019937190687536837,
      "loss": 1.5517,
      "step": 34670
    },
    {
      "epoch": 4.1732851985559565,
      "grad_norm": 6.194040775299072,
      "learning_rate": 0.0001993714810920276,
      "loss": 1.8116,
      "step": 34680
    },
    {
      "epoch": 4.174488567990373,
      "grad_norm": 9.579302787780762,
      "learning_rate": 0.00019937105516487175,
      "loss": 1.9583,
      "step": 34690
    },
    {
      "epoch": 4.17569193742479,
      "grad_norm": 6.3991522789001465,
      "learning_rate": 0.00019937062909390144,
      "loss": 1.7154,
      "step": 34700
    },
    {
      "epoch": 4.176895306859206,
      "grad_norm": 4.774811744689941,
      "learning_rate": 0.00019937020287911735,
      "loss": 1.8765,
      "step": 34710
    },
    {
      "epoch": 4.178098676293622,
      "grad_norm": 7.51531982421875,
      "learning_rate": 0.00019936977652051997,
      "loss": 1.7964,
      "step": 34720
    },
    {
      "epoch": 4.179302045728038,
      "grad_norm": 6.015110492706299,
      "learning_rate": 0.00019936935001811003,
      "loss": 1.6662,
      "step": 34730
    },
    {
      "epoch": 4.180505415162455,
      "grad_norm": 7.868860721588135,
      "learning_rate": 0.0001993689233718881,
      "loss": 1.9115,
      "step": 34740
    },
    {
      "epoch": 4.1817087845968715,
      "grad_norm": 5.770344257354736,
      "learning_rate": 0.00019936849658185483,
      "loss": 1.9144,
      "step": 34750
    },
    {
      "epoch": 4.182912154031287,
      "grad_norm": 4.498798370361328,
      "learning_rate": 0.00019936806964801078,
      "loss": 1.9393,
      "step": 34760
    },
    {
      "epoch": 4.184115523465704,
      "grad_norm": 7.16002893447876,
      "learning_rate": 0.00019936764257035661,
      "loss": 1.6565,
      "step": 34770
    },
    {
      "epoch": 4.185318892900121,
      "grad_norm": 5.324651718139648,
      "learning_rate": 0.00019936721534889292,
      "loss": 1.7747,
      "step": 34780
    },
    {
      "epoch": 4.186522262334536,
      "grad_norm": 8.296109199523926,
      "learning_rate": 0.00019936678798362033,
      "loss": 2.0047,
      "step": 34790
    },
    {
      "epoch": 4.187725631768953,
      "grad_norm": 7.719017028808594,
      "learning_rate": 0.0001993663604745395,
      "loss": 1.6695,
      "step": 34800
    },
    {
      "epoch": 4.18892900120337,
      "grad_norm": 6.086599349975586,
      "learning_rate": 0.000199365932821651,
      "loss": 1.8302,
      "step": 34810
    },
    {
      "epoch": 4.1901323706377855,
      "grad_norm": 7.158595561981201,
      "learning_rate": 0.00019936550502495543,
      "loss": 1.9011,
      "step": 34820
    },
    {
      "epoch": 4.191335740072202,
      "grad_norm": 5.211520195007324,
      "learning_rate": 0.00019936507708445347,
      "loss": 1.732,
      "step": 34830
    },
    {
      "epoch": 4.192539109506619,
      "grad_norm": 7.698371887207031,
      "learning_rate": 0.00019936464900014572,
      "loss": 1.9333,
      "step": 34840
    },
    {
      "epoch": 4.193742478941035,
      "grad_norm": 6.089698314666748,
      "learning_rate": 0.00019936422077203276,
      "loss": 1.8643,
      "step": 34850
    },
    {
      "epoch": 4.194945848375451,
      "grad_norm": 5.273772716522217,
      "learning_rate": 0.00019936379240011526,
      "loss": 1.8143,
      "step": 34860
    },
    {
      "epoch": 4.196149217809868,
      "grad_norm": 6.72157621383667,
      "learning_rate": 0.00019936336388439382,
      "loss": 1.6795,
      "step": 34870
    },
    {
      "epoch": 4.197352587244284,
      "grad_norm": 5.646254062652588,
      "learning_rate": 0.00019936293522486906,
      "loss": 1.8294,
      "step": 34880
    },
    {
      "epoch": 4.1985559566787005,
      "grad_norm": 9.14498519897461,
      "learning_rate": 0.00019936250642154163,
      "loss": 1.8915,
      "step": 34890
    },
    {
      "epoch": 4.199759326113116,
      "grad_norm": 5.023692607879639,
      "learning_rate": 0.00019936207747441208,
      "loss": 1.6755,
      "step": 34900
    },
    {
      "epoch": 4.200962695547533,
      "grad_norm": 4.519577980041504,
      "learning_rate": 0.00019936164838348108,
      "loss": 1.6403,
      "step": 34910
    },
    {
      "epoch": 4.20216606498195,
      "grad_norm": 7.0812482833862305,
      "learning_rate": 0.00019936121914874923,
      "loss": 1.6059,
      "step": 34920
    },
    {
      "epoch": 4.2033694344163655,
      "grad_norm": 5.809424877166748,
      "learning_rate": 0.0001993607897702172,
      "loss": 1.7562,
      "step": 34930
    },
    {
      "epoch": 4.204572803850782,
      "grad_norm": 12.280158042907715,
      "learning_rate": 0.00019936036024788555,
      "loss": 1.6649,
      "step": 34940
    },
    {
      "epoch": 4.205776173285199,
      "grad_norm": 7.3353657722473145,
      "learning_rate": 0.00019935993058175494,
      "loss": 1.7374,
      "step": 34950
    },
    {
      "epoch": 4.206979542719615,
      "grad_norm": 5.186786651611328,
      "learning_rate": 0.00019935950077182597,
      "loss": 2.0647,
      "step": 34960
    },
    {
      "epoch": 4.208182912154031,
      "grad_norm": 6.030091762542725,
      "learning_rate": 0.00019935907081809927,
      "loss": 1.4583,
      "step": 34970
    },
    {
      "epoch": 4.209386281588448,
      "grad_norm": 5.420872688293457,
      "learning_rate": 0.00019935864072057546,
      "loss": 1.8147,
      "step": 34980
    },
    {
      "epoch": 4.210589651022864,
      "grad_norm": 11.73915958404541,
      "learning_rate": 0.0001993582104792552,
      "loss": 1.8084,
      "step": 34990
    },
    {
      "epoch": 4.21179302045728,
      "grad_norm": 6.029935359954834,
      "learning_rate": 0.00019935778009413905,
      "loss": 1.8731,
      "step": 35000
    },
    {
      "epoch": 4.212996389891697,
      "grad_norm": 6.571990013122559,
      "learning_rate": 0.00019935734956522767,
      "loss": 1.7011,
      "step": 35010
    },
    {
      "epoch": 4.214199759326113,
      "grad_norm": 9.478644371032715,
      "learning_rate": 0.00019935691889252165,
      "loss": 1.78,
      "step": 35020
    },
    {
      "epoch": 4.21540312876053,
      "grad_norm": 5.048871994018555,
      "learning_rate": 0.00019935648807602166,
      "loss": 1.6787,
      "step": 35030
    },
    {
      "epoch": 4.216606498194946,
      "grad_norm": 15.470974922180176,
      "learning_rate": 0.00019935605711572828,
      "loss": 1.8724,
      "step": 35040
    },
    {
      "epoch": 4.217809867629362,
      "grad_norm": 6.290297985076904,
      "learning_rate": 0.00019935562601164223,
      "loss": 1.8578,
      "step": 35050
    },
    {
      "epoch": 4.219013237063779,
      "grad_norm": 4.576003551483154,
      "learning_rate": 0.00019935519476376397,
      "loss": 1.9149,
      "step": 35060
    },
    {
      "epoch": 4.2202166064981945,
      "grad_norm": 6.172026634216309,
      "learning_rate": 0.00019935476337209424,
      "loss": 1.7833,
      "step": 35070
    },
    {
      "epoch": 4.221419975932611,
      "grad_norm": 4.892474174499512,
      "learning_rate": 0.00019935433183663367,
      "loss": 1.7075,
      "step": 35080
    },
    {
      "epoch": 4.222623345367028,
      "grad_norm": 13.470946311950684,
      "learning_rate": 0.0001993539001573828,
      "loss": 2.0174,
      "step": 35090
    },
    {
      "epoch": 4.223826714801444,
      "grad_norm": 5.695079326629639,
      "learning_rate": 0.00019935346833434235,
      "loss": 1.5365,
      "step": 35100
    },
    {
      "epoch": 4.22503008423586,
      "grad_norm": 3.428987503051758,
      "learning_rate": 0.00019935303636751288,
      "loss": 1.7887,
      "step": 35110
    },
    {
      "epoch": 4.226233453670277,
      "grad_norm": 6.879944801330566,
      "learning_rate": 0.00019935260425689503,
      "loss": 1.8709,
      "step": 35120
    },
    {
      "epoch": 4.227436823104693,
      "grad_norm": 7.473792552947998,
      "learning_rate": 0.00019935217200248945,
      "loss": 1.806,
      "step": 35130
    },
    {
      "epoch": 4.2286401925391095,
      "grad_norm": 6.771570205688477,
      "learning_rate": 0.00019935173960429673,
      "loss": 1.8248,
      "step": 35140
    },
    {
      "epoch": 4.229843561973526,
      "grad_norm": 5.999017238616943,
      "learning_rate": 0.00019935130706231756,
      "loss": 1.6427,
      "step": 35150
    },
    {
      "epoch": 4.231046931407942,
      "grad_norm": 4.775561332702637,
      "learning_rate": 0.00019935087437655247,
      "loss": 1.8977,
      "step": 35160
    },
    {
      "epoch": 4.232250300842359,
      "grad_norm": 7.566339015960693,
      "learning_rate": 0.00019935044154700213,
      "loss": 1.7255,
      "step": 35170
    },
    {
      "epoch": 4.233453670276775,
      "grad_norm": 5.545777320861816,
      "learning_rate": 0.00019935000857366721,
      "loss": 1.5561,
      "step": 35180
    },
    {
      "epoch": 4.234657039711191,
      "grad_norm": 7.945019245147705,
      "learning_rate": 0.00019934957545654828,
      "loss": 1.846,
      "step": 35190
    },
    {
      "epoch": 4.235860409145608,
      "grad_norm": 7.428624153137207,
      "learning_rate": 0.00019934914219564597,
      "loss": 1.9585,
      "step": 35200
    },
    {
      "epoch": 4.2370637785800245,
      "grad_norm": 5.946137428283691,
      "learning_rate": 0.00019934870879096094,
      "loss": 1.7993,
      "step": 35210
    },
    {
      "epoch": 4.23826714801444,
      "grad_norm": 5.7132039070129395,
      "learning_rate": 0.00019934827524249383,
      "loss": 1.7853,
      "step": 35220
    },
    {
      "epoch": 4.239470517448857,
      "grad_norm": 5.234225273132324,
      "learning_rate": 0.0001993478415502452,
      "loss": 1.735,
      "step": 35230
    },
    {
      "epoch": 4.240673886883273,
      "grad_norm": 7.272106170654297,
      "learning_rate": 0.00019934740771421572,
      "loss": 2.1715,
      "step": 35240
    },
    {
      "epoch": 4.241877256317689,
      "grad_norm": 5.432416915893555,
      "learning_rate": 0.000199346973734406,
      "loss": 1.7854,
      "step": 35250
    },
    {
      "epoch": 4.243080625752106,
      "grad_norm": 3.8165786266326904,
      "learning_rate": 0.0001993465396108167,
      "loss": 2.0691,
      "step": 35260
    },
    {
      "epoch": 4.244283995186522,
      "grad_norm": 6.915083885192871,
      "learning_rate": 0.00019934610534344844,
      "loss": 1.6476,
      "step": 35270
    },
    {
      "epoch": 4.245487364620939,
      "grad_norm": 6.73002815246582,
      "learning_rate": 0.00019934567093230182,
      "loss": 1.9744,
      "step": 35280
    },
    {
      "epoch": 4.246690734055355,
      "grad_norm": 10.002744674682617,
      "learning_rate": 0.00019934523637737747,
      "loss": 2.1234,
      "step": 35290
    },
    {
      "epoch": 4.247894103489771,
      "grad_norm": 5.658259391784668,
      "learning_rate": 0.0001993448016786761,
      "loss": 1.6212,
      "step": 35300
    },
    {
      "epoch": 4.249097472924188,
      "grad_norm": 5.008306503295898,
      "learning_rate": 0.0001993443668361982,
      "loss": 1.8222,
      "step": 35310
    },
    {
      "epoch": 4.250300842358604,
      "grad_norm": 7.853512287139893,
      "learning_rate": 0.0001993439318499445,
      "loss": 1.6649,
      "step": 35320
    },
    {
      "epoch": 4.25150421179302,
      "grad_norm": 6.960813999176025,
      "learning_rate": 0.00019934349671991563,
      "loss": 1.7147,
      "step": 35330
    },
    {
      "epoch": 4.252707581227437,
      "grad_norm": 9.682868003845215,
      "learning_rate": 0.00019934306144611216,
      "loss": 1.7582,
      "step": 35340
    },
    {
      "epoch": 4.2539109506618535,
      "grad_norm": 6.54448938369751,
      "learning_rate": 0.00019934262602853476,
      "loss": 1.7474,
      "step": 35350
    },
    {
      "epoch": 4.255114320096269,
      "grad_norm": 4.013662338256836,
      "learning_rate": 0.00019934219046718405,
      "loss": 1.8111,
      "step": 35360
    },
    {
      "epoch": 4.256317689530686,
      "grad_norm": 7.412820339202881,
      "learning_rate": 0.00019934175476206067,
      "loss": 1.8062,
      "step": 35370
    },
    {
      "epoch": 4.257521058965103,
      "grad_norm": 5.398044109344482,
      "learning_rate": 0.00019934131891316526,
      "loss": 1.989,
      "step": 35380
    },
    {
      "epoch": 4.2587244283995185,
      "grad_norm": 21.605239868164062,
      "learning_rate": 0.0001993408829204984,
      "loss": 1.9033,
      "step": 35390
    },
    {
      "epoch": 4.259927797833935,
      "grad_norm": 9.701786041259766,
      "learning_rate": 0.00019934044678406078,
      "loss": 1.7739,
      "step": 35400
    },
    {
      "epoch": 4.261131167268351,
      "grad_norm": 4.230984210968018,
      "learning_rate": 0.00019934001050385302,
      "loss": 1.6612,
      "step": 35410
    },
    {
      "epoch": 4.262334536702768,
      "grad_norm": 6.254148483276367,
      "learning_rate": 0.00019933957407987572,
      "loss": 1.5458,
      "step": 35420
    },
    {
      "epoch": 4.263537906137184,
      "grad_norm": 3.7436411380767822,
      "learning_rate": 0.00019933913751212954,
      "loss": 1.9107,
      "step": 35430
    },
    {
      "epoch": 4.2647412755716,
      "grad_norm": 8.439335823059082,
      "learning_rate": 0.00019933870080061507,
      "loss": 1.9602,
      "step": 35440
    },
    {
      "epoch": 4.265944645006017,
      "grad_norm": 6.327473163604736,
      "learning_rate": 0.000199338263945333,
      "loss": 1.6398,
      "step": 35450
    },
    {
      "epoch": 4.2671480144404335,
      "grad_norm": 3.532308578491211,
      "learning_rate": 0.00019933782694628394,
      "loss": 1.9751,
      "step": 35460
    },
    {
      "epoch": 4.268351383874849,
      "grad_norm": 11.994088172912598,
      "learning_rate": 0.0001993373898034685,
      "loss": 1.7779,
      "step": 35470
    },
    {
      "epoch": 4.269554753309266,
      "grad_norm": 4.058684349060059,
      "learning_rate": 0.00019933695251688735,
      "loss": 1.8762,
      "step": 35480
    },
    {
      "epoch": 4.270758122743683,
      "grad_norm": 10.060203552246094,
      "learning_rate": 0.0001993365150865411,
      "loss": 1.6106,
      "step": 35490
    },
    {
      "epoch": 4.271961492178098,
      "grad_norm": 4.538617134094238,
      "learning_rate": 0.00019933607751243038,
      "loss": 1.4892,
      "step": 35500
    },
    {
      "epoch": 4.273164861612515,
      "grad_norm": 5.000890254974365,
      "learning_rate": 0.00019933563979455582,
      "loss": 1.73,
      "step": 35510
    },
    {
      "epoch": 4.274368231046932,
      "grad_norm": 6.338278770446777,
      "learning_rate": 0.00019933520193291812,
      "loss": 1.7313,
      "step": 35520
    },
    {
      "epoch": 4.2755716004813475,
      "grad_norm": 4.403758525848389,
      "learning_rate": 0.0001993347639275178,
      "loss": 1.7735,
      "step": 35530
    },
    {
      "epoch": 4.276774969915764,
      "grad_norm": 8.195080757141113,
      "learning_rate": 0.00019933432577835559,
      "loss": 1.7672,
      "step": 35540
    },
    {
      "epoch": 4.277978339350181,
      "grad_norm": 5.781729221343994,
      "learning_rate": 0.00019933388748543204,
      "loss": 1.6646,
      "step": 35550
    },
    {
      "epoch": 4.279181708784597,
      "grad_norm": 3.9311957359313965,
      "learning_rate": 0.00019933344904874786,
      "loss": 1.707,
      "step": 35560
    },
    {
      "epoch": 4.280385078219013,
      "grad_norm": 6.707176685333252,
      "learning_rate": 0.00019933301046830363,
      "loss": 1.8034,
      "step": 35570
    },
    {
      "epoch": 4.28158844765343,
      "grad_norm": 5.253597259521484,
      "learning_rate": 0.00019933257174410004,
      "loss": 1.6822,
      "step": 35580
    },
    {
      "epoch": 4.282791817087846,
      "grad_norm": 7.279968738555908,
      "learning_rate": 0.00019933213287613767,
      "loss": 1.7282,
      "step": 35590
    },
    {
      "epoch": 4.2839951865222625,
      "grad_norm": 5.6604390144348145,
      "learning_rate": 0.00019933169386441722,
      "loss": 1.7609,
      "step": 35600
    },
    {
      "epoch": 4.285198555956678,
      "grad_norm": 3.6584994792938232,
      "learning_rate": 0.00019933125470893926,
      "loss": 1.7549,
      "step": 35610
    },
    {
      "epoch": 4.286401925391095,
      "grad_norm": 5.592550754547119,
      "learning_rate": 0.00019933081540970444,
      "loss": 1.449,
      "step": 35620
    },
    {
      "epoch": 4.287605294825512,
      "grad_norm": 4.747725963592529,
      "learning_rate": 0.00019933037596671341,
      "loss": 1.8285,
      "step": 35630
    },
    {
      "epoch": 4.2888086642599275,
      "grad_norm": 8.285834312438965,
      "learning_rate": 0.00019932993637996683,
      "loss": 1.986,
      "step": 35640
    },
    {
      "epoch": 4.290012033694344,
      "grad_norm": 6.873829364776611,
      "learning_rate": 0.00019932949664946525,
      "loss": 1.6899,
      "step": 35650
    },
    {
      "epoch": 4.291215403128761,
      "grad_norm": 4.187936305999756,
      "learning_rate": 0.00019932905677520941,
      "loss": 1.7355,
      "step": 35660
    },
    {
      "epoch": 4.292418772563177,
      "grad_norm": 5.897096157073975,
      "learning_rate": 0.0001993286167571999,
      "loss": 1.8627,
      "step": 35670
    },
    {
      "epoch": 4.293622141997593,
      "grad_norm": 4.470292568206787,
      "learning_rate": 0.00019932817659543736,
      "loss": 1.6871,
      "step": 35680
    },
    {
      "epoch": 4.29482551143201,
      "grad_norm": 11.189642906188965,
      "learning_rate": 0.0001993277362899224,
      "loss": 1.7162,
      "step": 35690
    },
    {
      "epoch": 4.296028880866426,
      "grad_norm": 5.609859943389893,
      "learning_rate": 0.0001993272958406557,
      "loss": 1.844,
      "step": 35700
    },
    {
      "epoch": 4.297232250300842,
      "grad_norm": 4.639491558074951,
      "learning_rate": 0.00019932685524763787,
      "loss": 1.8034,
      "step": 35710
    },
    {
      "epoch": 4.298435619735259,
      "grad_norm": 6.2805352210998535,
      "learning_rate": 0.00019932641451086955,
      "loss": 1.6152,
      "step": 35720
    },
    {
      "epoch": 4.299638989169675,
      "grad_norm": 5.58298397064209,
      "learning_rate": 0.0001993259736303514,
      "loss": 1.8664,
      "step": 35730
    },
    {
      "epoch": 4.300842358604092,
      "grad_norm": 7.744947910308838,
      "learning_rate": 0.00019932553260608405,
      "loss": 1.6432,
      "step": 35740
    },
    {
      "epoch": 4.302045728038507,
      "grad_norm": 5.18134069442749,
      "learning_rate": 0.00019932509143806812,
      "loss": 1.8218,
      "step": 35750
    },
    {
      "epoch": 4.303249097472924,
      "grad_norm": 4.261021614074707,
      "learning_rate": 0.0001993246501263043,
      "loss": 1.947,
      "step": 35760
    },
    {
      "epoch": 4.304452466907341,
      "grad_norm": 13.76388168334961,
      "learning_rate": 0.00019932420867079312,
      "loss": 1.7549,
      "step": 35770
    },
    {
      "epoch": 4.3056558363417565,
      "grad_norm": 4.586688041687012,
      "learning_rate": 0.00019932376707153533,
      "loss": 1.8343,
      "step": 35780
    },
    {
      "epoch": 4.306859205776173,
      "grad_norm": 8.211957931518555,
      "learning_rate": 0.0001993233253285315,
      "loss": 1.8815,
      "step": 35790
    },
    {
      "epoch": 4.30806257521059,
      "grad_norm": 6.673455715179443,
      "learning_rate": 0.0001993228834417823,
      "loss": 1.7281,
      "step": 35800
    },
    {
      "epoch": 4.309265944645006,
      "grad_norm": 5.128195285797119,
      "learning_rate": 0.0001993224414112884,
      "loss": 1.7434,
      "step": 35810
    },
    {
      "epoch": 4.310469314079422,
      "grad_norm": 15.036243438720703,
      "learning_rate": 0.0001993219992370504,
      "loss": 1.6523,
      "step": 35820
    },
    {
      "epoch": 4.311672683513839,
      "grad_norm": 5.425032138824463,
      "learning_rate": 0.0001993215569190689,
      "loss": 1.8124,
      "step": 35830
    },
    {
      "epoch": 4.312876052948255,
      "grad_norm": 8.083535194396973,
      "learning_rate": 0.00019932111445734461,
      "loss": 1.9315,
      "step": 35840
    },
    {
      "epoch": 4.3140794223826715,
      "grad_norm": 6.526995658874512,
      "learning_rate": 0.00019932067185187818,
      "loss": 1.857,
      "step": 35850
    },
    {
      "epoch": 4.315282791817088,
      "grad_norm": 3.849255323410034,
      "learning_rate": 0.00019932022910267017,
      "loss": 1.7268,
      "step": 35860
    },
    {
      "epoch": 4.316486161251504,
      "grad_norm": 7.1441450119018555,
      "learning_rate": 0.00019931978620972126,
      "loss": 1.7334,
      "step": 35870
    },
    {
      "epoch": 4.317689530685921,
      "grad_norm": 5.974358558654785,
      "learning_rate": 0.00019931934317303217,
      "loss": 1.827,
      "step": 35880
    },
    {
      "epoch": 4.318892900120337,
      "grad_norm": 9.186847686767578,
      "learning_rate": 0.0001993188999926034,
      "loss": 1.8247,
      "step": 35890
    },
    {
      "epoch": 4.320096269554753,
      "grad_norm": 5.240936756134033,
      "learning_rate": 0.00019931845666843568,
      "loss": 1.6278,
      "step": 35900
    },
    {
      "epoch": 4.32129963898917,
      "grad_norm": 4.584122180938721,
      "learning_rate": 0.00019931801320052962,
      "loss": 1.8463,
      "step": 35910
    },
    {
      "epoch": 4.3225030084235865,
      "grad_norm": 15.641439437866211,
      "learning_rate": 0.0001993175695888859,
      "loss": 1.7596,
      "step": 35920
    },
    {
      "epoch": 4.323706377858002,
      "grad_norm": 5.413671016693115,
      "learning_rate": 0.00019931712583350512,
      "loss": 1.8015,
      "step": 35930
    },
    {
      "epoch": 4.324909747292419,
      "grad_norm": 9.118616104125977,
      "learning_rate": 0.00019931668193438795,
      "loss": 1.7768,
      "step": 35940
    },
    {
      "epoch": 4.326113116726835,
      "grad_norm": 5.554944038391113,
      "learning_rate": 0.00019931623789153503,
      "loss": 1.7296,
      "step": 35950
    },
    {
      "epoch": 4.327316486161251,
      "grad_norm": 3.5678491592407227,
      "learning_rate": 0.00019931579370494697,
      "loss": 1.9177,
      "step": 35960
    },
    {
      "epoch": 4.328519855595668,
      "grad_norm": 7.3495869636535645,
      "learning_rate": 0.00019931534937462443,
      "loss": 1.7876,
      "step": 35970
    },
    {
      "epoch": 4.329723225030084,
      "grad_norm": 4.591203212738037,
      "learning_rate": 0.00019931490490056807,
      "loss": 1.9446,
      "step": 35980
    },
    {
      "epoch": 4.330926594464501,
      "grad_norm": 9.50680160522461,
      "learning_rate": 0.00019931446028277852,
      "loss": 2.0639,
      "step": 35990
    },
    {
      "epoch": 4.332129963898917,
      "grad_norm": 5.03529167175293,
      "learning_rate": 0.00019931401552125645,
      "loss": 1.9546,
      "step": 36000
    },
    {
      "epoch": 4.333333333333333,
      "grad_norm": 5.492129325866699,
      "learning_rate": 0.00019931357061600245,
      "loss": 1.8102,
      "step": 36010
    },
    {
      "epoch": 4.33453670276775,
      "grad_norm": 6.360865592956543,
      "learning_rate": 0.00019931312556701721,
      "loss": 1.7551,
      "step": 36020
    },
    {
      "epoch": 4.335740072202166,
      "grad_norm": 4.650359630584717,
      "learning_rate": 0.00019931268037430132,
      "loss": 1.7854,
      "step": 36030
    },
    {
      "epoch": 4.336943441636582,
      "grad_norm": 10.43637752532959,
      "learning_rate": 0.00019931223503785552,
      "loss": 1.9163,
      "step": 36040
    },
    {
      "epoch": 4.338146811070999,
      "grad_norm": 8.118700981140137,
      "learning_rate": 0.00019931178955768036,
      "loss": 1.9775,
      "step": 36050
    },
    {
      "epoch": 4.3393501805054155,
      "grad_norm": 9.109759330749512,
      "learning_rate": 0.00019931134393377656,
      "loss": 1.8744,
      "step": 36060
    },
    {
      "epoch": 4.340553549939831,
      "grad_norm": 7.326494216918945,
      "learning_rate": 0.0001993108981661447,
      "loss": 1.9723,
      "step": 36070
    },
    {
      "epoch": 4.341756919374248,
      "grad_norm": 6.213112831115723,
      "learning_rate": 0.00019931045225478542,
      "loss": 1.7883,
      "step": 36080
    },
    {
      "epoch": 4.342960288808664,
      "grad_norm": 6.885368347167969,
      "learning_rate": 0.00019931000619969943,
      "loss": 1.9124,
      "step": 36090
    },
    {
      "epoch": 4.3441636582430805,
      "grad_norm": 5.544731616973877,
      "learning_rate": 0.00019930956000088735,
      "loss": 1.7551,
      "step": 36100
    },
    {
      "epoch": 4.345367027677497,
      "grad_norm": 3.515679359436035,
      "learning_rate": 0.00019930911365834982,
      "loss": 1.8179,
      "step": 36110
    },
    {
      "epoch": 4.346570397111913,
      "grad_norm": 6.745066165924072,
      "learning_rate": 0.00019930866717208744,
      "loss": 1.9082,
      "step": 36120
    },
    {
      "epoch": 4.34777376654633,
      "grad_norm": 4.27610969543457,
      "learning_rate": 0.00019930822054210095,
      "loss": 1.8479,
      "step": 36130
    },
    {
      "epoch": 4.348977135980746,
      "grad_norm": 6.727752685546875,
      "learning_rate": 0.0001993077737683909,
      "loss": 1.5559,
      "step": 36140
    },
    {
      "epoch": 4.350180505415162,
      "grad_norm": 5.7208027839660645,
      "learning_rate": 0.00019930732685095802,
      "loss": 1.7292,
      "step": 36150
    },
    {
      "epoch": 4.351383874849579,
      "grad_norm": 4.691730976104736,
      "learning_rate": 0.00019930687978980292,
      "loss": 1.8678,
      "step": 36160
    },
    {
      "epoch": 4.3525872442839955,
      "grad_norm": 6.359548568725586,
      "learning_rate": 0.00019930643258492622,
      "loss": 1.931,
      "step": 36170
    },
    {
      "epoch": 4.353790613718411,
      "grad_norm": 6.833027362823486,
      "learning_rate": 0.00019930598523632863,
      "loss": 1.7092,
      "step": 36180
    },
    {
      "epoch": 4.354993983152828,
      "grad_norm": 7.761911869049072,
      "learning_rate": 0.00019930553774401074,
      "loss": 1.7483,
      "step": 36190
    },
    {
      "epoch": 4.356197352587245,
      "grad_norm": 5.729844093322754,
      "learning_rate": 0.0001993050901079732,
      "loss": 1.7814,
      "step": 36200
    },
    {
      "epoch": 4.35740072202166,
      "grad_norm": 4.560784339904785,
      "learning_rate": 0.0001993046423282167,
      "loss": 1.7795,
      "step": 36210
    },
    {
      "epoch": 4.358604091456077,
      "grad_norm": 7.4706244468688965,
      "learning_rate": 0.00019930419440474187,
      "loss": 1.6964,
      "step": 36220
    },
    {
      "epoch": 4.359807460890494,
      "grad_norm": 11.191123962402344,
      "learning_rate": 0.00019930374633754934,
      "loss": 1.6069,
      "step": 36230
    },
    {
      "epoch": 4.3610108303249095,
      "grad_norm": 7.026401996612549,
      "learning_rate": 0.0001993032981266398,
      "loss": 1.7645,
      "step": 36240
    },
    {
      "epoch": 4.362214199759326,
      "grad_norm": 7.390120983123779,
      "learning_rate": 0.00019930284977201382,
      "loss": 1.7388,
      "step": 36250
    },
    {
      "epoch": 4.363417569193743,
      "grad_norm": 4.75221061706543,
      "learning_rate": 0.00019930240127367214,
      "loss": 1.6189,
      "step": 36260
    },
    {
      "epoch": 4.364620938628159,
      "grad_norm": 8.297945022583008,
      "learning_rate": 0.00019930195263161536,
      "loss": 1.8325,
      "step": 36270
    },
    {
      "epoch": 4.365824308062575,
      "grad_norm": 5.3272600173950195,
      "learning_rate": 0.00019930150384584412,
      "loss": 1.8726,
      "step": 36280
    },
    {
      "epoch": 4.367027677496991,
      "grad_norm": 7.332561492919922,
      "learning_rate": 0.00019930105491635912,
      "loss": 1.8565,
      "step": 36290
    },
    {
      "epoch": 4.368231046931408,
      "grad_norm": 5.097908020019531,
      "learning_rate": 0.00019930060584316095,
      "loss": 1.6533,
      "step": 36300
    },
    {
      "epoch": 4.3694344163658245,
      "grad_norm": 6.119940280914307,
      "learning_rate": 0.00019930015662625026,
      "loss": 2.0051,
      "step": 36310
    },
    {
      "epoch": 4.37063778580024,
      "grad_norm": 6.721015930175781,
      "learning_rate": 0.0001992997072656278,
      "loss": 1.7732,
      "step": 36320
    },
    {
      "epoch": 4.371841155234657,
      "grad_norm": 5.655241966247559,
      "learning_rate": 0.0001992992577612941,
      "loss": 1.7963,
      "step": 36330
    },
    {
      "epoch": 4.373044524669074,
      "grad_norm": 7.532368183135986,
      "learning_rate": 0.00019929880811324986,
      "loss": 2.1093,
      "step": 36340
    },
    {
      "epoch": 4.3742478941034895,
      "grad_norm": 7.15523624420166,
      "learning_rate": 0.00019929835832149575,
      "loss": 1.7397,
      "step": 36350
    },
    {
      "epoch": 4.375451263537906,
      "grad_norm": 5.214695930480957,
      "learning_rate": 0.00019929790838603238,
      "loss": 1.732,
      "step": 36360
    },
    {
      "epoch": 4.376654632972323,
      "grad_norm": 6.590765953063965,
      "learning_rate": 0.00019929745830686045,
      "loss": 1.8307,
      "step": 36370
    },
    {
      "epoch": 4.377858002406739,
      "grad_norm": 7.025494575500488,
      "learning_rate": 0.00019929700808398053,
      "loss": 1.955,
      "step": 36380
    },
    {
      "epoch": 4.379061371841155,
      "grad_norm": 8.426220893859863,
      "learning_rate": 0.00019929655771739336,
      "loss": 1.9521,
      "step": 36390
    },
    {
      "epoch": 4.380264741275572,
      "grad_norm": 6.4095048904418945,
      "learning_rate": 0.00019929610720709955,
      "loss": 1.682,
      "step": 36400
    },
    {
      "epoch": 4.381468110709988,
      "grad_norm": 3.4742069244384766,
      "learning_rate": 0.00019929565655309975,
      "loss": 1.7903,
      "step": 36410
    },
    {
      "epoch": 4.382671480144404,
      "grad_norm": 7.236454010009766,
      "learning_rate": 0.00019929520575539462,
      "loss": 1.752,
      "step": 36420
    },
    {
      "epoch": 4.38387484957882,
      "grad_norm": 6.319152355194092,
      "learning_rate": 0.00019929475481398484,
      "loss": 1.9301,
      "step": 36430
    },
    {
      "epoch": 4.385078219013237,
      "grad_norm": 9.045572280883789,
      "learning_rate": 0.000199294303728871,
      "loss": 2.134,
      "step": 36440
    },
    {
      "epoch": 4.386281588447654,
      "grad_norm": 5.281524658203125,
      "learning_rate": 0.0001992938525000538,
      "loss": 1.6693,
      "step": 36450
    },
    {
      "epoch": 4.387484957882069,
      "grad_norm": 7.266637802124023,
      "learning_rate": 0.00019929340112753385,
      "loss": 1.9864,
      "step": 36460
    },
    {
      "epoch": 4.388688327316486,
      "grad_norm": 5.887492656707764,
      "learning_rate": 0.0001992929496113119,
      "loss": 1.9689,
      "step": 36470
    },
    {
      "epoch": 4.389891696750903,
      "grad_norm": 6.876590728759766,
      "learning_rate": 0.00019929249795138846,
      "loss": 1.757,
      "step": 36480
    },
    {
      "epoch": 4.3910950661853185,
      "grad_norm": 7.290596008300781,
      "learning_rate": 0.00019929204614776432,
      "loss": 1.8331,
      "step": 36490
    },
    {
      "epoch": 4.392298435619735,
      "grad_norm": 6.67741060256958,
      "learning_rate": 0.00019929159420044003,
      "loss": 1.689,
      "step": 36500
    },
    {
      "epoch": 4.393501805054152,
      "grad_norm": 3.9267070293426514,
      "learning_rate": 0.00019929114210941634,
      "loss": 1.9893,
      "step": 36510
    },
    {
      "epoch": 4.394705174488568,
      "grad_norm": 18.108749389648438,
      "learning_rate": 0.00019929068987469383,
      "loss": 1.6952,
      "step": 36520
    },
    {
      "epoch": 4.395908543922984,
      "grad_norm": 5.565483093261719,
      "learning_rate": 0.00019929023749627316,
      "loss": 1.8553,
      "step": 36530
    },
    {
      "epoch": 4.397111913357401,
      "grad_norm": 7.668908596038818,
      "learning_rate": 0.00019928978497415502,
      "loss": 1.9182,
      "step": 36540
    },
    {
      "epoch": 4.398315282791817,
      "grad_norm": 5.820605754852295,
      "learning_rate": 0.00019928933230834006,
      "loss": 1.7105,
      "step": 36550
    },
    {
      "epoch": 4.3995186522262335,
      "grad_norm": 4.420670986175537,
      "learning_rate": 0.00019928887949882887,
      "loss": 1.8847,
      "step": 36560
    },
    {
      "epoch": 4.40072202166065,
      "grad_norm": 18.569854736328125,
      "learning_rate": 0.00019928842654562222,
      "loss": 1.881,
      "step": 36570
    },
    {
      "epoch": 4.401925391095066,
      "grad_norm": 9.885371208190918,
      "learning_rate": 0.00019928797344872067,
      "loss": 2.0918,
      "step": 36580
    },
    {
      "epoch": 4.403128760529483,
      "grad_norm": 19.252552032470703,
      "learning_rate": 0.00019928752020812493,
      "loss": 1.9647,
      "step": 36590
    },
    {
      "epoch": 4.404332129963899,
      "grad_norm": 11.048675537109375,
      "learning_rate": 0.00019928706682383561,
      "loss": 1.7183,
      "step": 36600
    },
    {
      "epoch": 4.405535499398315,
      "grad_norm": 8.772344589233398,
      "learning_rate": 0.00019928661329585344,
      "loss": 2.0971,
      "step": 36610
    },
    {
      "epoch": 4.406738868832732,
      "grad_norm": 7.278609752655029,
      "learning_rate": 0.000199286159624179,
      "loss": 1.8773,
      "step": 36620
    },
    {
      "epoch": 4.4079422382671485,
      "grad_norm": 4.808311939239502,
      "learning_rate": 0.00019928570580881295,
      "loss": 1.9369,
      "step": 36630
    },
    {
      "epoch": 4.409145607701564,
      "grad_norm": 12.39403247833252,
      "learning_rate": 0.00019928525184975598,
      "loss": 1.6563,
      "step": 36640
    },
    {
      "epoch": 4.410348977135981,
      "grad_norm": 5.27571439743042,
      "learning_rate": 0.00019928479774700874,
      "loss": 1.8594,
      "step": 36650
    },
    {
      "epoch": 4.411552346570397,
      "grad_norm": 7.558592796325684,
      "learning_rate": 0.00019928434350057188,
      "loss": 1.8163,
      "step": 36660
    },
    {
      "epoch": 4.412755716004813,
      "grad_norm": 8.557501792907715,
      "learning_rate": 0.00019928388911044607,
      "loss": 1.9418,
      "step": 36670
    },
    {
      "epoch": 4.41395908543923,
      "grad_norm": 6.074558258056641,
      "learning_rate": 0.00019928343457663195,
      "loss": 2.0838,
      "step": 36680
    },
    {
      "epoch": 4.415162454873646,
      "grad_norm": 11.97942066192627,
      "learning_rate": 0.00019928297989913022,
      "loss": 1.9122,
      "step": 36690
    },
    {
      "epoch": 4.416365824308063,
      "grad_norm": 6.416935443878174,
      "learning_rate": 0.00019928252507794147,
      "loss": 1.756,
      "step": 36700
    },
    {
      "epoch": 4.417569193742479,
      "grad_norm": 6.5072221755981445,
      "learning_rate": 0.0001992820701130664,
      "loss": 2.0011,
      "step": 36710
    },
    {
      "epoch": 4.418772563176895,
      "grad_norm": 24.140560150146484,
      "learning_rate": 0.00019928161500450567,
      "loss": 2.118,
      "step": 36720
    },
    {
      "epoch": 4.419975932611312,
      "grad_norm": 6.745471000671387,
      "learning_rate": 0.00019928115975225997,
      "loss": 1.5276,
      "step": 36730
    },
    {
      "epoch": 4.421179302045728,
      "grad_norm": 15.678532600402832,
      "learning_rate": 0.00019928070435632986,
      "loss": 1.8905,
      "step": 36740
    },
    {
      "epoch": 4.422382671480144,
      "grad_norm": 6.229355335235596,
      "learning_rate": 0.00019928024881671607,
      "loss": 1.8864,
      "step": 36750
    },
    {
      "epoch": 4.423586040914561,
      "grad_norm": 4.903979301452637,
      "learning_rate": 0.00019927979313341928,
      "loss": 2.031,
      "step": 36760
    },
    {
      "epoch": 4.4247894103489775,
      "grad_norm": 11.69352912902832,
      "learning_rate": 0.00019927933730644012,
      "loss": 1.8753,
      "step": 36770
    },
    {
      "epoch": 4.425992779783393,
      "grad_norm": 8.421667098999023,
      "learning_rate": 0.00019927888133577918,
      "loss": 1.8974,
      "step": 36780
    },
    {
      "epoch": 4.42719614921781,
      "grad_norm": 9.678336143493652,
      "learning_rate": 0.00019927842522143724,
      "loss": 2.2166,
      "step": 36790
    },
    {
      "epoch": 4.428399518652226,
      "grad_norm": 11.020577430725098,
      "learning_rate": 0.0001992779689634149,
      "loss": 2.0467,
      "step": 36800
    },
    {
      "epoch": 4.4296028880866425,
      "grad_norm": 8.437112808227539,
      "learning_rate": 0.00019927751256171284,
      "loss": 2.0048,
      "step": 36810
    },
    {
      "epoch": 4.430806257521059,
      "grad_norm": 10.001437187194824,
      "learning_rate": 0.0001992770560163317,
      "loss": 1.8926,
      "step": 36820
    },
    {
      "epoch": 4.432009626955475,
      "grad_norm": 6.060479640960693,
      "learning_rate": 0.00019927659932727213,
      "loss": 1.6919,
      "step": 36830
    },
    {
      "epoch": 4.433212996389892,
      "grad_norm": 7.661810874938965,
      "learning_rate": 0.00019927614249453486,
      "loss": 2.0966,
      "step": 36840
    },
    {
      "epoch": 4.434416365824308,
      "grad_norm": 6.644181251525879,
      "learning_rate": 0.00019927568551812046,
      "loss": 1.8284,
      "step": 36850
    },
    {
      "epoch": 4.435619735258724,
      "grad_norm": 4.807556629180908,
      "learning_rate": 0.00019927522839802963,
      "loss": 1.8337,
      "step": 36860
    },
    {
      "epoch": 4.436823104693141,
      "grad_norm": 8.891944885253906,
      "learning_rate": 0.00019927477113426304,
      "loss": 2.0124,
      "step": 36870
    },
    {
      "epoch": 4.4380264741275575,
      "grad_norm": 21.717252731323242,
      "learning_rate": 0.00019927431372682137,
      "loss": 1.847,
      "step": 36880
    },
    {
      "epoch": 4.439229843561973,
      "grad_norm": 7.565113067626953,
      "learning_rate": 0.00019927385617570523,
      "loss": 1.9442,
      "step": 36890
    },
    {
      "epoch": 4.44043321299639,
      "grad_norm": 5.530343055725098,
      "learning_rate": 0.00019927339848091534,
      "loss": 1.8794,
      "step": 36900
    },
    {
      "epoch": 4.441636582430807,
      "grad_norm": 5.1620635986328125,
      "learning_rate": 0.0001992729406424523,
      "loss": 1.9209,
      "step": 36910
    },
    {
      "epoch": 4.442839951865222,
      "grad_norm": 33.19838333129883,
      "learning_rate": 0.0001992724826603168,
      "loss": 1.8482,
      "step": 36920
    },
    {
      "epoch": 4.444043321299639,
      "grad_norm": 11.720840454101562,
      "learning_rate": 0.00019927202453450953,
      "loss": 1.952,
      "step": 36930
    },
    {
      "epoch": 4.445246690734056,
      "grad_norm": 8.067503929138184,
      "learning_rate": 0.00019927156626503115,
      "loss": 1.9282,
      "step": 36940
    },
    {
      "epoch": 4.4464500601684716,
      "grad_norm": 6.697203159332275,
      "learning_rate": 0.00019927110785188226,
      "loss": 1.8177,
      "step": 36950
    },
    {
      "epoch": 4.447653429602888,
      "grad_norm": 5.20427131652832,
      "learning_rate": 0.0001992706492950636,
      "loss": 1.9685,
      "step": 36960
    },
    {
      "epoch": 4.448856799037305,
      "grad_norm": 7.286891460418701,
      "learning_rate": 0.0001992701905945758,
      "loss": 1.9866,
      "step": 36970
    },
    {
      "epoch": 4.450060168471721,
      "grad_norm": 6.762622833251953,
      "learning_rate": 0.0001992697317504195,
      "loss": 1.9211,
      "step": 36980
    },
    {
      "epoch": 4.451263537906137,
      "grad_norm": 10.337411880493164,
      "learning_rate": 0.0001992692727625954,
      "loss": 2.1011,
      "step": 36990
    },
    {
      "epoch": 4.452466907340553,
      "grad_norm": 9.371976852416992,
      "learning_rate": 0.00019926881363110417,
      "loss": 1.8377,
      "step": 37000
    },
    {
      "epoch": 4.45367027677497,
      "grad_norm": 6.596066474914551,
      "learning_rate": 0.00019926835435594645,
      "loss": 2.0199,
      "step": 37010
    },
    {
      "epoch": 4.4548736462093865,
      "grad_norm": 11.738470077514648,
      "learning_rate": 0.0001992678949371229,
      "loss": 2.0109,
      "step": 37020
    },
    {
      "epoch": 4.456077015643802,
      "grad_norm": 8.737201690673828,
      "learning_rate": 0.00019926743537463423,
      "loss": 2.0224,
      "step": 37030
    },
    {
      "epoch": 4.457280385078219,
      "grad_norm": 8.715923309326172,
      "learning_rate": 0.00019926697566848104,
      "loss": 2.058,
      "step": 37040
    },
    {
      "epoch": 4.458483754512636,
      "grad_norm": 19.24488067626953,
      "learning_rate": 0.000199266515818664,
      "loss": 1.927,
      "step": 37050
    },
    {
      "epoch": 4.4596871239470515,
      "grad_norm": 7.10844087600708,
      "learning_rate": 0.00019926605582518386,
      "loss": 1.9459,
      "step": 37060
    },
    {
      "epoch": 4.460890493381468,
      "grad_norm": 7.398124694824219,
      "learning_rate": 0.0001992655956880412,
      "loss": 1.8815,
      "step": 37070
    },
    {
      "epoch": 4.462093862815885,
      "grad_norm": 5.712508201599121,
      "learning_rate": 0.00019926513540723676,
      "loss": 2.0228,
      "step": 37080
    },
    {
      "epoch": 4.463297232250301,
      "grad_norm": 14.620765686035156,
      "learning_rate": 0.00019926467498277111,
      "loss": 1.7733,
      "step": 37090
    },
    {
      "epoch": 4.464500601684717,
      "grad_norm": 6.207784175872803,
      "learning_rate": 0.00019926421441464497,
      "loss": 1.8915,
      "step": 37100
    },
    {
      "epoch": 4.465703971119134,
      "grad_norm": 5.447126388549805,
      "learning_rate": 0.00019926375370285903,
      "loss": 2.0654,
      "step": 37110
    },
    {
      "epoch": 4.46690734055355,
      "grad_norm": 25.362512588500977,
      "learning_rate": 0.0001992632928474139,
      "loss": 1.9623,
      "step": 37120
    },
    {
      "epoch": 4.4681107099879664,
      "grad_norm": 5.799778938293457,
      "learning_rate": 0.0001992628318483103,
      "loss": 1.993,
      "step": 37130
    },
    {
      "epoch": 4.469314079422382,
      "grad_norm": 8.49531078338623,
      "learning_rate": 0.0001992623707055489,
      "loss": 1.9349,
      "step": 37140
    },
    {
      "epoch": 4.470517448856799,
      "grad_norm": 12.713287353515625,
      "learning_rate": 0.0001992619094191303,
      "loss": 1.9627,
      "step": 37150
    },
    {
      "epoch": 4.471720818291216,
      "grad_norm": 6.367859363555908,
      "learning_rate": 0.00019926144798905522,
      "loss": 1.888,
      "step": 37160
    },
    {
      "epoch": 4.472924187725631,
      "grad_norm": 8.403925895690918,
      "learning_rate": 0.0001992609864153243,
      "loss": 1.9791,
      "step": 37170
    },
    {
      "epoch": 4.474127557160048,
      "grad_norm": 9.251457214355469,
      "learning_rate": 0.00019926052469793827,
      "loss": 1.8487,
      "step": 37180
    },
    {
      "epoch": 4.475330926594465,
      "grad_norm": 12.146878242492676,
      "learning_rate": 0.00019926006283689773,
      "loss": 1.9635,
      "step": 37190
    },
    {
      "epoch": 4.4765342960288805,
      "grad_norm": 6.026196002960205,
      "learning_rate": 0.00019925960083220335,
      "loss": 1.6327,
      "step": 37200
    },
    {
      "epoch": 4.477737665463297,
      "grad_norm": 7.230823993682861,
      "learning_rate": 0.00019925913868385587,
      "loss": 1.9194,
      "step": 37210
    },
    {
      "epoch": 4.478941034897714,
      "grad_norm": 6.682619571685791,
      "learning_rate": 0.0001992586763918559,
      "loss": 1.9382,
      "step": 37220
    },
    {
      "epoch": 4.48014440433213,
      "grad_norm": 7.898355484008789,
      "learning_rate": 0.0001992582139562041,
      "loss": 2.1301,
      "step": 37230
    },
    {
      "epoch": 4.481347773766546,
      "grad_norm": 16.16246223449707,
      "learning_rate": 0.00019925775137690118,
      "loss": 1.9163,
      "step": 37240
    },
    {
      "epoch": 4.482551143200963,
      "grad_norm": 8.502813339233398,
      "learning_rate": 0.00019925728865394774,
      "loss": 1.8153,
      "step": 37250
    },
    {
      "epoch": 4.483754512635379,
      "grad_norm": 6.817388534545898,
      "learning_rate": 0.00019925682578734452,
      "loss": 2.0369,
      "step": 37260
    },
    {
      "epoch": 4.4849578820697955,
      "grad_norm": 7.099721431732178,
      "learning_rate": 0.0001992563627770922,
      "loss": 1.9944,
      "step": 37270
    },
    {
      "epoch": 4.486161251504212,
      "grad_norm": 4.773013114929199,
      "learning_rate": 0.0001992558996231914,
      "loss": 1.8088,
      "step": 37280
    },
    {
      "epoch": 4.487364620938628,
      "grad_norm": 16.34334945678711,
      "learning_rate": 0.0001992554363256428,
      "loss": 1.9865,
      "step": 37290
    },
    {
      "epoch": 4.488567990373045,
      "grad_norm": 7.298641681671143,
      "learning_rate": 0.00019925497288444705,
      "loss": 2.083,
      "step": 37300
    },
    {
      "epoch": 4.489771359807461,
      "grad_norm": 5.195982933044434,
      "learning_rate": 0.0001992545092996049,
      "loss": 2.1238,
      "step": 37310
    },
    {
      "epoch": 4.490974729241877,
      "grad_norm": 8.375102996826172,
      "learning_rate": 0.00019925404557111692,
      "loss": 1.9985,
      "step": 37320
    },
    {
      "epoch": 4.492178098676294,
      "grad_norm": 6.089810371398926,
      "learning_rate": 0.0001992535816989839,
      "loss": 2.0661,
      "step": 37330
    },
    {
      "epoch": 4.49338146811071,
      "grad_norm": 7.642695903778076,
      "learning_rate": 0.00019925311768320637,
      "loss": 1.8794,
      "step": 37340
    },
    {
      "epoch": 4.494584837545126,
      "grad_norm": 6.981215953826904,
      "learning_rate": 0.00019925265352378513,
      "loss": 1.8602,
      "step": 37350
    },
    {
      "epoch": 4.495788206979543,
      "grad_norm": 4.471064567565918,
      "learning_rate": 0.00019925218922072077,
      "loss": 1.947,
      "step": 37360
    },
    {
      "epoch": 4.496991576413959,
      "grad_norm": 6.985415458679199,
      "learning_rate": 0.000199251724774014,
      "loss": 1.8577,
      "step": 37370
    },
    {
      "epoch": 4.498194945848375,
      "grad_norm": 7.6307549476623535,
      "learning_rate": 0.00019925126018366545,
      "loss": 1.8119,
      "step": 37380
    },
    {
      "epoch": 4.499398315282792,
      "grad_norm": 13.285627365112305,
      "learning_rate": 0.00019925079544967585,
      "loss": 1.8662,
      "step": 37390
    },
    {
      "epoch": 4.500601684717208,
      "grad_norm": 6.505211353302002,
      "learning_rate": 0.00019925033057204583,
      "loss": 1.9064,
      "step": 37400
    },
    {
      "epoch": 4.501805054151625,
      "grad_norm": 9.097518920898438,
      "learning_rate": 0.0001992498655507761,
      "loss": 1.8405,
      "step": 37410
    },
    {
      "epoch": 4.503008423586041,
      "grad_norm": 7.4252424240112305,
      "learning_rate": 0.00019924940038586729,
      "loss": 2.1371,
      "step": 37420
    },
    {
      "epoch": 4.504211793020457,
      "grad_norm": 7.798403739929199,
      "learning_rate": 0.00019924893507732009,
      "loss": 1.9207,
      "step": 37430
    },
    {
      "epoch": 4.505415162454874,
      "grad_norm": 7.5146164894104,
      "learning_rate": 0.00019924846962513518,
      "loss": 2.0236,
      "step": 37440
    },
    {
      "epoch": 4.5066185318892895,
      "grad_norm": 6.379922866821289,
      "learning_rate": 0.00019924800402931325,
      "loss": 1.869,
      "step": 37450
    },
    {
      "epoch": 4.507821901323706,
      "grad_norm": 6.786048412322998,
      "learning_rate": 0.0001992475382898549,
      "loss": 2.0661,
      "step": 37460
    },
    {
      "epoch": 4.509025270758123,
      "grad_norm": 8.681519508361816,
      "learning_rate": 0.00019924707240676092,
      "loss": 1.8231,
      "step": 37470
    },
    {
      "epoch": 4.510228640192539,
      "grad_norm": 4.627808570861816,
      "learning_rate": 0.00019924660638003192,
      "loss": 1.9974,
      "step": 37480
    },
    {
      "epoch": 4.511432009626955,
      "grad_norm": 8.16715145111084,
      "learning_rate": 0.00019924614020966854,
      "loss": 2.0107,
      "step": 37490
    },
    {
      "epoch": 4.512635379061372,
      "grad_norm": 7.240932941436768,
      "learning_rate": 0.00019924567389567152,
      "loss": 1.9919,
      "step": 37500
    },
    {
      "epoch": 4.513838748495788,
      "grad_norm": 4.046568870544434,
      "learning_rate": 0.00019924520743804148,
      "loss": 1.8022,
      "step": 37510
    },
    {
      "epoch": 4.5150421179302045,
      "grad_norm": 8.555015563964844,
      "learning_rate": 0.00019924474083677916,
      "loss": 1.8356,
      "step": 37520
    },
    {
      "epoch": 4.516245487364621,
      "grad_norm": 5.146360397338867,
      "learning_rate": 0.00019924427409188518,
      "loss": 1.9993,
      "step": 37530
    },
    {
      "epoch": 4.517448856799037,
      "grad_norm": 11.656143188476562,
      "learning_rate": 0.00019924380720336021,
      "loss": 2.1149,
      "step": 37540
    },
    {
      "epoch": 4.518652226233454,
      "grad_norm": 12.554299354553223,
      "learning_rate": 0.000199243340171205,
      "loss": 1.801,
      "step": 37550
    },
    {
      "epoch": 4.51985559566787,
      "grad_norm": 7.014607906341553,
      "learning_rate": 0.00019924287299542012,
      "loss": 1.7003,
      "step": 37560
    },
    {
      "epoch": 4.521058965102286,
      "grad_norm": 7.310118675231934,
      "learning_rate": 0.00019924240567600635,
      "loss": 1.9469,
      "step": 37570
    },
    {
      "epoch": 4.522262334536703,
      "grad_norm": 8.424588203430176,
      "learning_rate": 0.00019924193821296425,
      "loss": 1.9928,
      "step": 37580
    },
    {
      "epoch": 4.5234657039711195,
      "grad_norm": 7.748364448547363,
      "learning_rate": 0.00019924147060629464,
      "loss": 2.1891,
      "step": 37590
    },
    {
      "epoch": 4.524669073405535,
      "grad_norm": 7.200401782989502,
      "learning_rate": 0.00019924100285599808,
      "loss": 1.7539,
      "step": 37600
    },
    {
      "epoch": 4.525872442839952,
      "grad_norm": 12.594355583190918,
      "learning_rate": 0.00019924053496207529,
      "loss": 1.9731,
      "step": 37610
    },
    {
      "epoch": 4.527075812274369,
      "grad_norm": 10.205979347229004,
      "learning_rate": 0.00019924006692452695,
      "loss": 1.6851,
      "step": 37620
    },
    {
      "epoch": 4.528279181708784,
      "grad_norm": 6.131598949432373,
      "learning_rate": 0.0001992395987433537,
      "loss": 1.7934,
      "step": 37630
    },
    {
      "epoch": 4.529482551143201,
      "grad_norm": 6.139014720916748,
      "learning_rate": 0.0001992391304185563,
      "loss": 1.8358,
      "step": 37640
    },
    {
      "epoch": 4.530685920577618,
      "grad_norm": 8.550439834594727,
      "learning_rate": 0.00019923866195013535,
      "loss": 1.9257,
      "step": 37650
    },
    {
      "epoch": 4.5318892900120336,
      "grad_norm": 4.324347972869873,
      "learning_rate": 0.00019923819333809154,
      "loss": 1.9961,
      "step": 37660
    },
    {
      "epoch": 4.53309265944645,
      "grad_norm": 8.383240699768066,
      "learning_rate": 0.0001992377245824256,
      "loss": 1.8115,
      "step": 37670
    },
    {
      "epoch": 4.534296028880867,
      "grad_norm": 6.141906261444092,
      "learning_rate": 0.00019923725568313816,
      "loss": 2.1539,
      "step": 37680
    },
    {
      "epoch": 4.535499398315283,
      "grad_norm": 7.333021640777588,
      "learning_rate": 0.00019923678664022989,
      "loss": 2.143,
      "step": 37690
    },
    {
      "epoch": 4.536702767749699,
      "grad_norm": 5.740495204925537,
      "learning_rate": 0.0001992363174537015,
      "loss": 1.9818,
      "step": 37700
    },
    {
      "epoch": 4.537906137184115,
      "grad_norm": 6.611884117126465,
      "learning_rate": 0.00019923584812355367,
      "loss": 2.067,
      "step": 37710
    },
    {
      "epoch": 4.539109506618532,
      "grad_norm": 5.949376583099365,
      "learning_rate": 0.00019923537864978706,
      "loss": 1.825,
      "step": 37720
    },
    {
      "epoch": 4.5403128760529485,
      "grad_norm": 15.834717750549316,
      "learning_rate": 0.00019923490903240234,
      "loss": 1.7445,
      "step": 37730
    },
    {
      "epoch": 4.541516245487364,
      "grad_norm": 10.180578231811523,
      "learning_rate": 0.00019923443927140022,
      "loss": 2.1226,
      "step": 37740
    },
    {
      "epoch": 4.542719614921781,
      "grad_norm": 9.784039497375488,
      "learning_rate": 0.0001992339693667814,
      "loss": 1.6301,
      "step": 37750
    },
    {
      "epoch": 4.543922984356198,
      "grad_norm": 4.267033100128174,
      "learning_rate": 0.00019923349931854647,
      "loss": 1.7469,
      "step": 37760
    },
    {
      "epoch": 4.5451263537906135,
      "grad_norm": 8.615670204162598,
      "learning_rate": 0.00019923302912669617,
      "loss": 2.0589,
      "step": 37770
    },
    {
      "epoch": 4.54632972322503,
      "grad_norm": 4.914862632751465,
      "learning_rate": 0.0001992325587912312,
      "loss": 2.0424,
      "step": 37780
    },
    {
      "epoch": 4.547533092659447,
      "grad_norm": 7.973679065704346,
      "learning_rate": 0.00019923208831215222,
      "loss": 1.9179,
      "step": 37790
    },
    {
      "epoch": 4.548736462093863,
      "grad_norm": 5.358487129211426,
      "learning_rate": 0.0001992316176894599,
      "loss": 1.8285,
      "step": 37800
    },
    {
      "epoch": 4.549939831528279,
      "grad_norm": 4.255433082580566,
      "learning_rate": 0.00019923114692315492,
      "loss": 2.0838,
      "step": 37810
    },
    {
      "epoch": 4.551143200962695,
      "grad_norm": 9.114831924438477,
      "learning_rate": 0.00019923067601323798,
      "loss": 1.904,
      "step": 37820
    },
    {
      "epoch": 4.552346570397112,
      "grad_norm": 5.057202339172363,
      "learning_rate": 0.00019923020495970977,
      "loss": 1.9548,
      "step": 37830
    },
    {
      "epoch": 4.5535499398315284,
      "grad_norm": 11.94949722290039,
      "learning_rate": 0.00019922973376257095,
      "loss": 2.0864,
      "step": 37840
    },
    {
      "epoch": 4.554753309265944,
      "grad_norm": 6.582622528076172,
      "learning_rate": 0.0001992292624218222,
      "loss": 2.0782,
      "step": 37850
    },
    {
      "epoch": 4.555956678700361,
      "grad_norm": 5.160717964172363,
      "learning_rate": 0.0001992287909374642,
      "loss": 1.9943,
      "step": 37860
    },
    {
      "epoch": 4.557160048134778,
      "grad_norm": 8.315377235412598,
      "learning_rate": 0.00019922831930949762,
      "loss": 1.8425,
      "step": 37870
    },
    {
      "epoch": 4.558363417569193,
      "grad_norm": 6.430552959442139,
      "learning_rate": 0.0001992278475379232,
      "loss": 1.8823,
      "step": 37880
    },
    {
      "epoch": 4.55956678700361,
      "grad_norm": 10.638143539428711,
      "learning_rate": 0.00019922737562274158,
      "loss": 2.0202,
      "step": 37890
    },
    {
      "epoch": 4.560770156438027,
      "grad_norm": 7.622200965881348,
      "learning_rate": 0.00019922690356395344,
      "loss": 1.7269,
      "step": 37900
    },
    {
      "epoch": 4.5619735258724425,
      "grad_norm": 5.80261754989624,
      "learning_rate": 0.0001992264313615595,
      "loss": 1.9202,
      "step": 37910
    },
    {
      "epoch": 4.563176895306859,
      "grad_norm": 7.089993953704834,
      "learning_rate": 0.00019922595901556038,
      "loss": 1.9562,
      "step": 37920
    },
    {
      "epoch": 4.564380264741276,
      "grad_norm": 7.092113018035889,
      "learning_rate": 0.00019922548652595681,
      "loss": 2.0324,
      "step": 37930
    },
    {
      "epoch": 4.565583634175692,
      "grad_norm": 16.311973571777344,
      "learning_rate": 0.00019922501389274945,
      "loss": 1.856,
      "step": 37940
    },
    {
      "epoch": 4.566787003610108,
      "grad_norm": 9.96202278137207,
      "learning_rate": 0.00019922454111593902,
      "loss": 1.986,
      "step": 37950
    },
    {
      "epoch": 4.567990373044525,
      "grad_norm": 5.622683048248291,
      "learning_rate": 0.00019922406819552617,
      "loss": 1.7719,
      "step": 37960
    },
    {
      "epoch": 4.569193742478941,
      "grad_norm": 6.116504192352295,
      "learning_rate": 0.00019922359513151162,
      "loss": 1.9339,
      "step": 37970
    },
    {
      "epoch": 4.5703971119133575,
      "grad_norm": 10.435121536254883,
      "learning_rate": 0.00019922312192389597,
      "loss": 1.9793,
      "step": 37980
    },
    {
      "epoch": 4.571600481347774,
      "grad_norm": 8.617453575134277,
      "learning_rate": 0.00019922264857268003,
      "loss": 1.8636,
      "step": 37990
    },
    {
      "epoch": 4.57280385078219,
      "grad_norm": 10.9584379196167,
      "learning_rate": 0.00019922217507786438,
      "loss": 1.8736,
      "step": 38000
    },
    {
      "epoch": 4.574007220216607,
      "grad_norm": 14.49172306060791,
      "learning_rate": 0.00019922170143944976,
      "loss": 1.7835,
      "step": 38010
    },
    {
      "epoch": 4.575210589651023,
      "grad_norm": 7.032172203063965,
      "learning_rate": 0.00019922122765743682,
      "loss": 1.9243,
      "step": 38020
    },
    {
      "epoch": 4.576413959085439,
      "grad_norm": 8.402040481567383,
      "learning_rate": 0.00019922075373182628,
      "loss": 1.9136,
      "step": 38030
    },
    {
      "epoch": 4.577617328519856,
      "grad_norm": 8.604158401489258,
      "learning_rate": 0.00019922027966261883,
      "loss": 2.0815,
      "step": 38040
    },
    {
      "epoch": 4.578820697954272,
      "grad_norm": 5.890580177307129,
      "learning_rate": 0.00019921980544981515,
      "loss": 2.0205,
      "step": 38050
    },
    {
      "epoch": 4.580024067388688,
      "grad_norm": 11.066864967346191,
      "learning_rate": 0.00019921933109341589,
      "loss": 1.8682,
      "step": 38060
    },
    {
      "epoch": 4.581227436823105,
      "grad_norm": 8.122532844543457,
      "learning_rate": 0.00019921885659342174,
      "loss": 2.0424,
      "step": 38070
    },
    {
      "epoch": 4.582430806257521,
      "grad_norm": 8.963947296142578,
      "learning_rate": 0.00019921838194983344,
      "loss": 1.7892,
      "step": 38080
    },
    {
      "epoch": 4.583634175691937,
      "grad_norm": 6.899969577789307,
      "learning_rate": 0.00019921790716265165,
      "loss": 2.0055,
      "step": 38090
    },
    {
      "epoch": 4.584837545126354,
      "grad_norm": 8.297788619995117,
      "learning_rate": 0.00019921743223187704,
      "loss": 1.945,
      "step": 38100
    },
    {
      "epoch": 4.58604091456077,
      "grad_norm": 36.4146842956543,
      "learning_rate": 0.00019921695715751029,
      "loss": 2.0239,
      "step": 38110
    },
    {
      "epoch": 4.587244283995187,
      "grad_norm": 6.881565570831299,
      "learning_rate": 0.00019921648193955215,
      "loss": 2.0318,
      "step": 38120
    },
    {
      "epoch": 4.588447653429603,
      "grad_norm": 6.308768272399902,
      "learning_rate": 0.00019921600657800322,
      "loss": 1.8649,
      "step": 38130
    },
    {
      "epoch": 4.589651022864019,
      "grad_norm": 9.94020938873291,
      "learning_rate": 0.00019921553107286427,
      "loss": 1.809,
      "step": 38140
    },
    {
      "epoch": 4.590854392298436,
      "grad_norm": 6.146694660186768,
      "learning_rate": 0.0001992150554241359,
      "loss": 1.9957,
      "step": 38150
    },
    {
      "epoch": 4.5920577617328515,
      "grad_norm": 8.65699291229248,
      "learning_rate": 0.0001992145796318189,
      "loss": 1.8722,
      "step": 38160
    },
    {
      "epoch": 4.593261131167268,
      "grad_norm": 8.434221267700195,
      "learning_rate": 0.0001992141036959139,
      "loss": 2.0671,
      "step": 38170
    },
    {
      "epoch": 4.594464500601685,
      "grad_norm": 13.314875602722168,
      "learning_rate": 0.0001992136276164216,
      "loss": 1.8473,
      "step": 38180
    },
    {
      "epoch": 4.595667870036101,
      "grad_norm": 9.386907577514648,
      "learning_rate": 0.00019921315139334266,
      "loss": 2.1,
      "step": 38190
    },
    {
      "epoch": 4.596871239470517,
      "grad_norm": 5.242396354675293,
      "learning_rate": 0.00019921267502667783,
      "loss": 1.9444,
      "step": 38200
    },
    {
      "epoch": 4.598074608904934,
      "grad_norm": 4.145884990692139,
      "learning_rate": 0.00019921219851642773,
      "loss": 1.947,
      "step": 38210
    },
    {
      "epoch": 4.59927797833935,
      "grad_norm": 6.597233772277832,
      "learning_rate": 0.0001992117218625931,
      "loss": 1.8639,
      "step": 38220
    },
    {
      "epoch": 4.6004813477737665,
      "grad_norm": 5.497170448303223,
      "learning_rate": 0.00019921124506517462,
      "loss": 1.9504,
      "step": 38230
    },
    {
      "epoch": 4.601684717208183,
      "grad_norm": 7.455959796905518,
      "learning_rate": 0.00019921076812417298,
      "loss": 2.058,
      "step": 38240
    },
    {
      "epoch": 4.602888086642599,
      "grad_norm": 5.906527996063232,
      "learning_rate": 0.00019921029103958882,
      "loss": 1.7151,
      "step": 38250
    },
    {
      "epoch": 4.604091456077016,
      "grad_norm": 4.899425983428955,
      "learning_rate": 0.00019920981381142292,
      "loss": 2.1783,
      "step": 38260
    },
    {
      "epoch": 4.605294825511432,
      "grad_norm": 7.4382100105285645,
      "learning_rate": 0.00019920933643967592,
      "loss": 1.9446,
      "step": 38270
    },
    {
      "epoch": 4.606498194945848,
      "grad_norm": 5.704405784606934,
      "learning_rate": 0.0001992088589243485,
      "loss": 1.9866,
      "step": 38280
    },
    {
      "epoch": 4.607701564380265,
      "grad_norm": 18.863250732421875,
      "learning_rate": 0.00019920838126544137,
      "loss": 1.9098,
      "step": 38290
    },
    {
      "epoch": 4.6089049338146815,
      "grad_norm": 6.587116241455078,
      "learning_rate": 0.0001992079034629552,
      "loss": 1.8921,
      "step": 38300
    },
    {
      "epoch": 4.610108303249097,
      "grad_norm": 5.117888450622559,
      "learning_rate": 0.00019920742551689073,
      "loss": 2.1543,
      "step": 38310
    },
    {
      "epoch": 4.611311672683514,
      "grad_norm": 9.064175605773926,
      "learning_rate": 0.0001992069474272486,
      "loss": 1.9865,
      "step": 38320
    },
    {
      "epoch": 4.612515042117931,
      "grad_norm": 5.347578525543213,
      "learning_rate": 0.00019920646919402958,
      "loss": 2.0254,
      "step": 38330
    },
    {
      "epoch": 4.613718411552346,
      "grad_norm": 7.340710639953613,
      "learning_rate": 0.00019920599081723426,
      "loss": 2.1486,
      "step": 38340
    },
    {
      "epoch": 4.614921780986763,
      "grad_norm": 6.411147594451904,
      "learning_rate": 0.00019920551229686337,
      "loss": 2.0321,
      "step": 38350
    },
    {
      "epoch": 4.61612515042118,
      "grad_norm": 8.462505340576172,
      "learning_rate": 0.0001992050336329176,
      "loss": 2.2851,
      "step": 38360
    },
    {
      "epoch": 4.617328519855596,
      "grad_norm": 14.334405899047852,
      "learning_rate": 0.0001992045548253977,
      "loss": 2.1559,
      "step": 38370
    },
    {
      "epoch": 4.618531889290012,
      "grad_norm": 5.688232898712158,
      "learning_rate": 0.0001992040758743043,
      "loss": 2.036,
      "step": 38380
    },
    {
      "epoch": 4.619735258724428,
      "grad_norm": 7.32138204574585,
      "learning_rate": 0.0001992035967796381,
      "loss": 2.0482,
      "step": 38390
    },
    {
      "epoch": 4.620938628158845,
      "grad_norm": 9.042555809020996,
      "learning_rate": 0.0001992031175413998,
      "loss": 1.8062,
      "step": 38400
    },
    {
      "epoch": 4.622141997593261,
      "grad_norm": 5.123806953430176,
      "learning_rate": 0.00019920263815959008,
      "loss": 2.0217,
      "step": 38410
    },
    {
      "epoch": 4.623345367027677,
      "grad_norm": 7.053982734680176,
      "learning_rate": 0.00019920215863420969,
      "loss": 2.0229,
      "step": 38420
    },
    {
      "epoch": 4.624548736462094,
      "grad_norm": 8.836836814880371,
      "learning_rate": 0.00019920167896525927,
      "loss": 1.9295,
      "step": 38430
    },
    {
      "epoch": 4.6257521058965105,
      "grad_norm": 7.976781368255615,
      "learning_rate": 0.00019920119915273953,
      "loss": 1.9316,
      "step": 38440
    },
    {
      "epoch": 4.626955475330926,
      "grad_norm": 6.514922142028809,
      "learning_rate": 0.00019920071919665114,
      "loss": 1.9108,
      "step": 38450
    },
    {
      "epoch": 4.628158844765343,
      "grad_norm": 5.870746612548828,
      "learning_rate": 0.00019920023909699482,
      "loss": 1.8907,
      "step": 38460
    },
    {
      "epoch": 4.62936221419976,
      "grad_norm": 7.147825717926025,
      "learning_rate": 0.0001991997588537713,
      "loss": 2.0451,
      "step": 38470
    },
    {
      "epoch": 4.6305655836341755,
      "grad_norm": 15.836055755615234,
      "learning_rate": 0.00019919927846698122,
      "loss": 1.8581,
      "step": 38480
    },
    {
      "epoch": 4.631768953068592,
      "grad_norm": 8.948719024658203,
      "learning_rate": 0.00019919879793662526,
      "loss": 1.9407,
      "step": 38490
    },
    {
      "epoch": 4.632972322503008,
      "grad_norm": 13.253113746643066,
      "learning_rate": 0.00019919831726270419,
      "loss": 2.1827,
      "step": 38500
    },
    {
      "epoch": 4.634175691937425,
      "grad_norm": 5.759546756744385,
      "learning_rate": 0.00019919783644521864,
      "loss": 1.9114,
      "step": 38510
    },
    {
      "epoch": 4.635379061371841,
      "grad_norm": 9.337198257446289,
      "learning_rate": 0.00019919735548416933,
      "loss": 1.9372,
      "step": 38520
    },
    {
      "epoch": 4.636582430806257,
      "grad_norm": 9.921058654785156,
      "learning_rate": 0.00019919687437955697,
      "loss": 1.9241,
      "step": 38530
    },
    {
      "epoch": 4.637785800240674,
      "grad_norm": 10.45384407043457,
      "learning_rate": 0.00019919639313138222,
      "loss": 2.0231,
      "step": 38540
    },
    {
      "epoch": 4.6389891696750905,
      "grad_norm": 8.505849838256836,
      "learning_rate": 0.00019919591173964582,
      "loss": 1.7725,
      "step": 38550
    },
    {
      "epoch": 4.640192539109506,
      "grad_norm": 7.61210823059082,
      "learning_rate": 0.00019919543020434843,
      "loss": 1.8658,
      "step": 38560
    },
    {
      "epoch": 4.641395908543923,
      "grad_norm": 6.48394250869751,
      "learning_rate": 0.0001991949485254908,
      "loss": 1.9097,
      "step": 38570
    },
    {
      "epoch": 4.64259927797834,
      "grad_norm": 7.609060287475586,
      "learning_rate": 0.00019919446670307354,
      "loss": 1.8554,
      "step": 38580
    },
    {
      "epoch": 4.643802647412755,
      "grad_norm": 8.147257804870605,
      "learning_rate": 0.00019919398473709744,
      "loss": 1.9777,
      "step": 38590
    },
    {
      "epoch": 4.645006016847172,
      "grad_norm": 10.474865913391113,
      "learning_rate": 0.00019919350262756312,
      "loss": 1.768,
      "step": 38600
    },
    {
      "epoch": 4.646209386281589,
      "grad_norm": 5.379399299621582,
      "learning_rate": 0.0001991930203744713,
      "loss": 2.0159,
      "step": 38610
    },
    {
      "epoch": 4.6474127557160045,
      "grad_norm": 6.453421592712402,
      "learning_rate": 0.00019919253797782274,
      "loss": 1.8079,
      "step": 38620
    },
    {
      "epoch": 4.648616125150421,
      "grad_norm": 4.304332733154297,
      "learning_rate": 0.00019919205543761806,
      "loss": 2.0221,
      "step": 38630
    },
    {
      "epoch": 4.649819494584838,
      "grad_norm": 11.97105884552002,
      "learning_rate": 0.00019919157275385797,
      "loss": 1.8953,
      "step": 38640
    },
    {
      "epoch": 4.651022864019254,
      "grad_norm": 6.9134297370910645,
      "learning_rate": 0.0001991910899265432,
      "loss": 2.0701,
      "step": 38650
    },
    {
      "epoch": 4.65222623345367,
      "grad_norm": 5.144277572631836,
      "learning_rate": 0.0001991906069556744,
      "loss": 2.0459,
      "step": 38660
    },
    {
      "epoch": 4.653429602888087,
      "grad_norm": 13.8529634475708,
      "learning_rate": 0.00019919012384125235,
      "loss": 2.0014,
      "step": 38670
    },
    {
      "epoch": 4.654632972322503,
      "grad_norm": 4.78397274017334,
      "learning_rate": 0.0001991896405832777,
      "loss": 2.0482,
      "step": 38680
    },
    {
      "epoch": 4.6558363417569195,
      "grad_norm": 12.549687385559082,
      "learning_rate": 0.00019918915718175113,
      "loss": 1.9872,
      "step": 38690
    },
    {
      "epoch": 4.657039711191336,
      "grad_norm": 5.540886402130127,
      "learning_rate": 0.00019918867363667337,
      "loss": 2.0857,
      "step": 38700
    },
    {
      "epoch": 4.658243080625752,
      "grad_norm": 12.373787879943848,
      "learning_rate": 0.0001991881899480451,
      "loss": 2.0306,
      "step": 38710
    },
    {
      "epoch": 4.659446450060169,
      "grad_norm": 11.565423965454102,
      "learning_rate": 0.00019918770611586703,
      "loss": 1.8649,
      "step": 38720
    },
    {
      "epoch": 4.6606498194945845,
      "grad_norm": 5.739513397216797,
      "learning_rate": 0.0001991872221401399,
      "loss": 2.0518,
      "step": 38730
    },
    {
      "epoch": 4.661853188929001,
      "grad_norm": 15.935868263244629,
      "learning_rate": 0.00019918673802086435,
      "loss": 2.1491,
      "step": 38740
    },
    {
      "epoch": 4.663056558363418,
      "grad_norm": 5.813896656036377,
      "learning_rate": 0.00019918625375804107,
      "loss": 1.8102,
      "step": 38750
    },
    {
      "epoch": 4.664259927797834,
      "grad_norm": 8.6066312789917,
      "learning_rate": 0.0001991857693516708,
      "loss": 2.1178,
      "step": 38760
    },
    {
      "epoch": 4.66546329723225,
      "grad_norm": 13.907723426818848,
      "learning_rate": 0.00019918528480175428,
      "loss": 1.9506,
      "step": 38770
    },
    {
      "epoch": 4.666666666666667,
      "grad_norm": 4.396481513977051,
      "learning_rate": 0.0001991848001082921,
      "loss": 1.8585,
      "step": 38780
    },
    {
      "epoch": 4.667870036101083,
      "grad_norm": 12.915420532226562,
      "learning_rate": 0.00019918431527128508,
      "loss": 1.9045,
      "step": 38790
    },
    {
      "epoch": 4.669073405535499,
      "grad_norm": 7.082880973815918,
      "learning_rate": 0.00019918383029073384,
      "loss": 1.8275,
      "step": 38800
    },
    {
      "epoch": 4.670276774969916,
      "grad_norm": 8.76838207244873,
      "learning_rate": 0.00019918334516663913,
      "loss": 1.9593,
      "step": 38810
    },
    {
      "epoch": 4.671480144404332,
      "grad_norm": 7.5392231941223145,
      "learning_rate": 0.00019918285989900162,
      "loss": 2.0901,
      "step": 38820
    },
    {
      "epoch": 4.672683513838749,
      "grad_norm": 9.134536743164062,
      "learning_rate": 0.00019918237448782202,
      "loss": 1.8903,
      "step": 38830
    },
    {
      "epoch": 4.673886883273164,
      "grad_norm": 10.274060249328613,
      "learning_rate": 0.000199181888933101,
      "loss": 1.8829,
      "step": 38840
    },
    {
      "epoch": 4.675090252707581,
      "grad_norm": 6.194350719451904,
      "learning_rate": 0.00019918140323483936,
      "loss": 1.9095,
      "step": 38850
    },
    {
      "epoch": 4.676293622141998,
      "grad_norm": 5.424354553222656,
      "learning_rate": 0.00019918091739303773,
      "loss": 1.8485,
      "step": 38860
    },
    {
      "epoch": 4.6774969915764135,
      "grad_norm": 13.536675453186035,
      "learning_rate": 0.0001991804314076968,
      "loss": 1.8535,
      "step": 38870
    },
    {
      "epoch": 4.67870036101083,
      "grad_norm": 5.463991641998291,
      "learning_rate": 0.00019917994527881734,
      "loss": 1.7198,
      "step": 38880
    },
    {
      "epoch": 4.679903730445247,
      "grad_norm": 8.189309120178223,
      "learning_rate": 0.00019917945900639999,
      "loss": 2.0346,
      "step": 38890
    },
    {
      "epoch": 4.681107099879663,
      "grad_norm": 6.468610763549805,
      "learning_rate": 0.00019917897259044546,
      "loss": 1.8921,
      "step": 38900
    },
    {
      "epoch": 4.682310469314079,
      "grad_norm": 3.821913242340088,
      "learning_rate": 0.0001991784860309545,
      "loss": 1.9713,
      "step": 38910
    },
    {
      "epoch": 4.683513838748496,
      "grad_norm": 5.608883857727051,
      "learning_rate": 0.00019917799932792776,
      "loss": 2.2028,
      "step": 38920
    },
    {
      "epoch": 4.684717208182912,
      "grad_norm": 21.48900604248047,
      "learning_rate": 0.00019917751248136597,
      "loss": 1.8266,
      "step": 38930
    },
    {
      "epoch": 4.6859205776173285,
      "grad_norm": 8.551660537719727,
      "learning_rate": 0.00019917702549126983,
      "loss": 1.8584,
      "step": 38940
    },
    {
      "epoch": 4.687123947051745,
      "grad_norm": 6.141820907592773,
      "learning_rate": 0.0001991765383576401,
      "loss": 1.8196,
      "step": 38950
    },
    {
      "epoch": 4.688327316486161,
      "grad_norm": 6.991450309753418,
      "learning_rate": 0.00019917605108047738,
      "loss": 1.8017,
      "step": 38960
    },
    {
      "epoch": 4.689530685920578,
      "grad_norm": 8.62569808959961,
      "learning_rate": 0.00019917556365978243,
      "loss": 1.9349,
      "step": 38970
    },
    {
      "epoch": 4.690734055354994,
      "grad_norm": 5.824770927429199,
      "learning_rate": 0.00019917507609555597,
      "loss": 1.9892,
      "step": 38980
    },
    {
      "epoch": 4.69193742478941,
      "grad_norm": 10.973548889160156,
      "learning_rate": 0.00019917458838779868,
      "loss": 2.1287,
      "step": 38990
    },
    {
      "epoch": 4.693140794223827,
      "grad_norm": 8.601175308227539,
      "learning_rate": 0.00019917410053651126,
      "loss": 1.9457,
      "step": 39000
    },
    {
      "epoch": 4.6943441636582435,
      "grad_norm": 6.339250564575195,
      "learning_rate": 0.00019917361254169448,
      "loss": 1.9717,
      "step": 39010
    },
    {
      "epoch": 4.695547533092659,
      "grad_norm": 10.602438926696777,
      "learning_rate": 0.00019917312440334898,
      "loss": 1.9831,
      "step": 39020
    },
    {
      "epoch": 4.696750902527076,
      "grad_norm": 4.155811786651611,
      "learning_rate": 0.00019917263612147546,
      "loss": 1.7227,
      "step": 39030
    },
    {
      "epoch": 4.697954271961493,
      "grad_norm": 9.204297065734863,
      "learning_rate": 0.00019917214769607467,
      "loss": 1.9866,
      "step": 39040
    },
    {
      "epoch": 4.699157641395908,
      "grad_norm": 10.78105640411377,
      "learning_rate": 0.00019917165912714725,
      "loss": 2.1162,
      "step": 39050
    },
    {
      "epoch": 4.700361010830325,
      "grad_norm": 5.022622585296631,
      "learning_rate": 0.00019917117041469402,
      "loss": 1.9013,
      "step": 39060
    },
    {
      "epoch": 4.701564380264742,
      "grad_norm": 8.880290031433105,
      "learning_rate": 0.00019917068155871559,
      "loss": 1.7069,
      "step": 39070
    },
    {
      "epoch": 4.702767749699158,
      "grad_norm": 6.2118377685546875,
      "learning_rate": 0.0001991701925592127,
      "loss": 1.9699,
      "step": 39080
    },
    {
      "epoch": 4.703971119133574,
      "grad_norm": 7.420884609222412,
      "learning_rate": 0.00019916970341618607,
      "loss": 2.1992,
      "step": 39090
    },
    {
      "epoch": 4.70517448856799,
      "grad_norm": 6.072713375091553,
      "learning_rate": 0.0001991692141296364,
      "loss": 1.9586,
      "step": 39100
    },
    {
      "epoch": 4.706377858002407,
      "grad_norm": 4.617374897003174,
      "learning_rate": 0.00019916872469956438,
      "loss": 1.9509,
      "step": 39110
    },
    {
      "epoch": 4.707581227436823,
      "grad_norm": 5.9688334465026855,
      "learning_rate": 0.00019916823512597072,
      "loss": 2.0487,
      "step": 39120
    },
    {
      "epoch": 4.708784596871239,
      "grad_norm": 9.01282787322998,
      "learning_rate": 0.00019916774540885615,
      "loss": 1.9915,
      "step": 39130
    },
    {
      "epoch": 4.709987966305656,
      "grad_norm": 8.382208824157715,
      "learning_rate": 0.00019916725554822138,
      "loss": 1.943,
      "step": 39140
    },
    {
      "epoch": 4.7111913357400725,
      "grad_norm": 6.859579563140869,
      "learning_rate": 0.00019916676554406707,
      "loss": 1.8657,
      "step": 39150
    },
    {
      "epoch": 4.712394705174488,
      "grad_norm": 10.082171440124512,
      "learning_rate": 0.000199166275396394,
      "loss": 1.9387,
      "step": 39160
    },
    {
      "epoch": 4.713598074608905,
      "grad_norm": 5.954122066497803,
      "learning_rate": 0.00019916578510520286,
      "loss": 1.9521,
      "step": 39170
    },
    {
      "epoch": 4.714801444043322,
      "grad_norm": 8.008260726928711,
      "learning_rate": 0.0001991652946704943,
      "loss": 2.1352,
      "step": 39180
    },
    {
      "epoch": 4.7160048134777375,
      "grad_norm": 9.707684516906738,
      "learning_rate": 0.0001991648040922691,
      "loss": 1.7714,
      "step": 39190
    },
    {
      "epoch": 4.717208182912154,
      "grad_norm": 5.941680908203125,
      "learning_rate": 0.00019916431337052798,
      "loss": 2.0328,
      "step": 39200
    },
    {
      "epoch": 4.71841155234657,
      "grad_norm": 4.283074378967285,
      "learning_rate": 0.00019916382250527158,
      "loss": 2.0465,
      "step": 39210
    },
    {
      "epoch": 4.719614921780987,
      "grad_norm": 5.88446569442749,
      "learning_rate": 0.00019916333149650064,
      "loss": 2.009,
      "step": 39220
    },
    {
      "epoch": 4.720818291215403,
      "grad_norm": 14.850969314575195,
      "learning_rate": 0.0001991628403442159,
      "loss": 2.07,
      "step": 39230
    },
    {
      "epoch": 4.722021660649819,
      "grad_norm": 7.3254804611206055,
      "learning_rate": 0.000199162349048418,
      "loss": 1.8446,
      "step": 39240
    },
    {
      "epoch": 4.723225030084236,
      "grad_norm": 10.557574272155762,
      "learning_rate": 0.00019916185760910775,
      "loss": 2.0767,
      "step": 39250
    },
    {
      "epoch": 4.7244283995186525,
      "grad_norm": 6.993002891540527,
      "learning_rate": 0.0001991613660262858,
      "loss": 2.117,
      "step": 39260
    },
    {
      "epoch": 4.725631768953068,
      "grad_norm": 16.8297176361084,
      "learning_rate": 0.00019916087429995287,
      "loss": 2.1953,
      "step": 39270
    },
    {
      "epoch": 4.726835138387485,
      "grad_norm": 5.601232528686523,
      "learning_rate": 0.00019916038243010967,
      "loss": 1.9631,
      "step": 39280
    },
    {
      "epoch": 4.728038507821902,
      "grad_norm": 7.897584438323975,
      "learning_rate": 0.00019915989041675692,
      "loss": 2.111,
      "step": 39290
    },
    {
      "epoch": 4.729241877256317,
      "grad_norm": 16.734622955322266,
      "learning_rate": 0.0001991593982598953,
      "loss": 2.097,
      "step": 39300
    },
    {
      "epoch": 4.730445246690734,
      "grad_norm": 10.241059303283691,
      "learning_rate": 0.00019915890595952559,
      "loss": 2.124,
      "step": 39310
    },
    {
      "epoch": 4.731648616125151,
      "grad_norm": 8.126265525817871,
      "learning_rate": 0.00019915841351564843,
      "loss": 1.8868,
      "step": 39320
    },
    {
      "epoch": 4.7328519855595665,
      "grad_norm": 8.909639358520508,
      "learning_rate": 0.00019915792092826457,
      "loss": 1.9609,
      "step": 39330
    },
    {
      "epoch": 4.734055354993983,
      "grad_norm": 8.8421049118042,
      "learning_rate": 0.0001991574281973747,
      "loss": 2.2227,
      "step": 39340
    },
    {
      "epoch": 4.7352587244284,
      "grad_norm": 10.465092658996582,
      "learning_rate": 0.00019915693532297958,
      "loss": 1.8981,
      "step": 39350
    },
    {
      "epoch": 4.736462093862816,
      "grad_norm": 6.22850227355957,
      "learning_rate": 0.00019915644230507988,
      "loss": 1.9493,
      "step": 39360
    },
    {
      "epoch": 4.737665463297232,
      "grad_norm": 7.862131118774414,
      "learning_rate": 0.00019915594914367635,
      "loss": 1.8518,
      "step": 39370
    },
    {
      "epoch": 4.738868832731649,
      "grad_norm": 5.224724769592285,
      "learning_rate": 0.00019915545583876966,
      "loss": 2.1733,
      "step": 39380
    },
    {
      "epoch": 4.740072202166065,
      "grad_norm": 7.3520917892456055,
      "learning_rate": 0.00019915496239036054,
      "loss": 1.781,
      "step": 39390
    },
    {
      "epoch": 4.7412755716004815,
      "grad_norm": 13.000665664672852,
      "learning_rate": 0.0001991544687984497,
      "loss": 1.8869,
      "step": 39400
    },
    {
      "epoch": 4.742478941034898,
      "grad_norm": 6.38198184967041,
      "learning_rate": 0.0001991539750630379,
      "loss": 1.9835,
      "step": 39410
    },
    {
      "epoch": 4.743682310469314,
      "grad_norm": 8.315906524658203,
      "learning_rate": 0.00019915348118412578,
      "loss": 1.8297,
      "step": 39420
    },
    {
      "epoch": 4.744885679903731,
      "grad_norm": 8.645343780517578,
      "learning_rate": 0.0001991529871617141,
      "loss": 1.8701,
      "step": 39430
    },
    {
      "epoch": 4.7460890493381465,
      "grad_norm": 7.214653491973877,
      "learning_rate": 0.0001991524929958036,
      "loss": 2.0204,
      "step": 39440
    },
    {
      "epoch": 4.747292418772563,
      "grad_norm": 6.126239776611328,
      "learning_rate": 0.00019915199868639493,
      "loss": 2.0778,
      "step": 39450
    },
    {
      "epoch": 4.74849578820698,
      "grad_norm": 6.310642242431641,
      "learning_rate": 0.00019915150423348882,
      "loss": 2.0158,
      "step": 39460
    },
    {
      "epoch": 4.749699157641396,
      "grad_norm": 6.208978652954102,
      "learning_rate": 0.00019915100963708606,
      "loss": 1.801,
      "step": 39470
    },
    {
      "epoch": 4.750902527075812,
      "grad_norm": 4.734537601470947,
      "learning_rate": 0.00019915051489718727,
      "loss": 1.965,
      "step": 39480
    },
    {
      "epoch": 4.752105896510229,
      "grad_norm": 7.569093227386475,
      "learning_rate": 0.0001991500200137932,
      "loss": 2.1077,
      "step": 39490
    },
    {
      "epoch": 4.753309265944645,
      "grad_norm": 12.631699562072754,
      "learning_rate": 0.0001991495249869046,
      "loss": 1.9071,
      "step": 39500
    },
    {
      "epoch": 4.754512635379061,
      "grad_norm": 5.180163383483887,
      "learning_rate": 0.00019914902981652214,
      "loss": 2.1518,
      "step": 39510
    },
    {
      "epoch": 4.755716004813478,
      "grad_norm": 14.124245643615723,
      "learning_rate": 0.00019914853450264655,
      "loss": 2.0155,
      "step": 39520
    },
    {
      "epoch": 4.756919374247894,
      "grad_norm": 4.914310932159424,
      "learning_rate": 0.00019914803904527858,
      "loss": 2.0019,
      "step": 39530
    },
    {
      "epoch": 4.758122743682311,
      "grad_norm": 8.773962020874023,
      "learning_rate": 0.0001991475434444189,
      "loss": 2.1427,
      "step": 39540
    },
    {
      "epoch": 4.759326113116726,
      "grad_norm": 6.844643592834473,
      "learning_rate": 0.00019914704770006826,
      "loss": 1.9901,
      "step": 39550
    },
    {
      "epoch": 4.760529482551143,
      "grad_norm": 5.214836597442627,
      "learning_rate": 0.00019914655181222735,
      "loss": 2.2465,
      "step": 39560
    },
    {
      "epoch": 4.76173285198556,
      "grad_norm": 7.27862024307251,
      "learning_rate": 0.0001991460557808969,
      "loss": 2.0193,
      "step": 39570
    },
    {
      "epoch": 4.7629362214199755,
      "grad_norm": 6.350466728210449,
      "learning_rate": 0.00019914555960607761,
      "loss": 1.9277,
      "step": 39580
    },
    {
      "epoch": 4.764139590854392,
      "grad_norm": 8.133624076843262,
      "learning_rate": 0.00019914506328777023,
      "loss": 1.8919,
      "step": 39590
    },
    {
      "epoch": 4.765342960288809,
      "grad_norm": 5.769651412963867,
      "learning_rate": 0.00019914456682597548,
      "loss": 1.8878,
      "step": 39600
    },
    {
      "epoch": 4.766546329723225,
      "grad_norm": 7.046748161315918,
      "learning_rate": 0.00019914407022069406,
      "loss": 2.029,
      "step": 39610
    },
    {
      "epoch": 4.767749699157641,
      "grad_norm": 14.47966194152832,
      "learning_rate": 0.0001991435734719267,
      "loss": 2.1774,
      "step": 39620
    },
    {
      "epoch": 4.768953068592058,
      "grad_norm": 7.438046455383301,
      "learning_rate": 0.0001991430765796741,
      "loss": 2.0883,
      "step": 39630
    },
    {
      "epoch": 4.770156438026474,
      "grad_norm": 8.632655143737793,
      "learning_rate": 0.00019914257954393695,
      "loss": 2.1434,
      "step": 39640
    },
    {
      "epoch": 4.7713598074608905,
      "grad_norm": 5.609837055206299,
      "learning_rate": 0.00019914208236471607,
      "loss": 1.8814,
      "step": 39650
    },
    {
      "epoch": 4.772563176895307,
      "grad_norm": 4.805448055267334,
      "learning_rate": 0.0001991415850420121,
      "loss": 2.1829,
      "step": 39660
    },
    {
      "epoch": 4.773766546329723,
      "grad_norm": 11.303045272827148,
      "learning_rate": 0.00019914108757582578,
      "loss": 1.9826,
      "step": 39670
    },
    {
      "epoch": 4.77496991576414,
      "grad_norm": 8.081250190734863,
      "learning_rate": 0.00019914058996615784,
      "loss": 2.1111,
      "step": 39680
    },
    {
      "epoch": 4.776173285198556,
      "grad_norm": 12.083354949951172,
      "learning_rate": 0.00019914009221300898,
      "loss": 1.934,
      "step": 39690
    },
    {
      "epoch": 4.777376654632972,
      "grad_norm": 6.836496829986572,
      "learning_rate": 0.0001991395943163799,
      "loss": 1.9859,
      "step": 39700
    },
    {
      "epoch": 4.778580024067389,
      "grad_norm": 5.122470855712891,
      "learning_rate": 0.00019913909627627138,
      "loss": 2.0142,
      "step": 39710
    },
    {
      "epoch": 4.7797833935018055,
      "grad_norm": 7.628266334533691,
      "learning_rate": 0.00019913859809268412,
      "loss": 1.9407,
      "step": 39720
    },
    {
      "epoch": 4.780986762936221,
      "grad_norm": 11.423534393310547,
      "learning_rate": 0.00019913809976561882,
      "loss": 2.0385,
      "step": 39730
    },
    {
      "epoch": 4.782190132370638,
      "grad_norm": 15.587311744689941,
      "learning_rate": 0.0001991376012950762,
      "loss": 2.1869,
      "step": 39740
    },
    {
      "epoch": 4.783393501805055,
      "grad_norm": 5.586488246917725,
      "learning_rate": 0.00019913710268105702,
      "loss": 1.9371,
      "step": 39750
    },
    {
      "epoch": 4.78459687123947,
      "grad_norm": 10.33002758026123,
      "learning_rate": 0.00019913660392356194,
      "loss": 2.0048,
      "step": 39760
    },
    {
      "epoch": 4.785800240673887,
      "grad_norm": 13.47502613067627,
      "learning_rate": 0.00019913610502259176,
      "loss": 1.9103,
      "step": 39770
    },
    {
      "epoch": 4.787003610108303,
      "grad_norm": 11.487744331359863,
      "learning_rate": 0.00019913560597814716,
      "loss": 1.9125,
      "step": 39780
    },
    {
      "epoch": 4.78820697954272,
      "grad_norm": 20.302907943725586,
      "learning_rate": 0.0001991351067902288,
      "loss": 2.0052,
      "step": 39790
    },
    {
      "epoch": 4.789410348977136,
      "grad_norm": 7.953364849090576,
      "learning_rate": 0.00019913460745883755,
      "loss": 1.8549,
      "step": 39800
    },
    {
      "epoch": 4.790613718411552,
      "grad_norm": 6.035819053649902,
      "learning_rate": 0.000199134107983974,
      "loss": 2.0331,
      "step": 39810
    },
    {
      "epoch": 4.791817087845969,
      "grad_norm": 13.217386245727539,
      "learning_rate": 0.00019913360836563892,
      "loss": 1.9468,
      "step": 39820
    },
    {
      "epoch": 4.793020457280385,
      "grad_norm": 6.257696628570557,
      "learning_rate": 0.00019913310860383306,
      "loss": 2.05,
      "step": 39830
    },
    {
      "epoch": 4.794223826714801,
      "grad_norm": 8.752163887023926,
      "learning_rate": 0.00019913260869855708,
      "loss": 2.0791,
      "step": 39840
    },
    {
      "epoch": 4.795427196149218,
      "grad_norm": 8.14365291595459,
      "learning_rate": 0.00019913210864981175,
      "loss": 2.0613,
      "step": 39850
    },
    {
      "epoch": 4.7966305655836345,
      "grad_norm": 4.922191619873047,
      "learning_rate": 0.0001991316084575978,
      "loss": 2.0093,
      "step": 39860
    },
    {
      "epoch": 4.79783393501805,
      "grad_norm": 8.692961692810059,
      "learning_rate": 0.00019913110812191592,
      "loss": 2.2555,
      "step": 39870
    },
    {
      "epoch": 4.799037304452467,
      "grad_norm": 5.166471481323242,
      "learning_rate": 0.00019913060764276686,
      "loss": 1.9626,
      "step": 39880
    },
    {
      "epoch": 4.800240673886883,
      "grad_norm": 9.108636856079102,
      "learning_rate": 0.00019913010702015137,
      "loss": 1.9603,
      "step": 39890
    },
    {
      "epoch": 4.8014440433212995,
      "grad_norm": 10.726107597351074,
      "learning_rate": 0.0001991296062540701,
      "loss": 1.9157,
      "step": 39900
    },
    {
      "epoch": 4.802647412755716,
      "grad_norm": 6.966701984405518,
      "learning_rate": 0.0001991291053445238,
      "loss": 1.919,
      "step": 39910
    },
    {
      "epoch": 4.803850782190132,
      "grad_norm": 28.718181610107422,
      "learning_rate": 0.00019912860429151325,
      "loss": 1.9522,
      "step": 39920
    },
    {
      "epoch": 4.805054151624549,
      "grad_norm": 4.801002502441406,
      "learning_rate": 0.00019912810309503912,
      "loss": 1.91,
      "step": 39930
    },
    {
      "epoch": 4.806257521058965,
      "grad_norm": 10.285552978515625,
      "learning_rate": 0.00019912760175510213,
      "loss": 2.0466,
      "step": 39940
    },
    {
      "epoch": 4.807460890493381,
      "grad_norm": 6.975771427154541,
      "learning_rate": 0.00019912710027170306,
      "loss": 1.7739,
      "step": 39950
    },
    {
      "epoch": 4.808664259927798,
      "grad_norm": 5.462837219238281,
      "learning_rate": 0.00019912659864484256,
      "loss": 1.9376,
      "step": 39960
    },
    {
      "epoch": 4.8098676293622145,
      "grad_norm": 6.584494113922119,
      "learning_rate": 0.00019912609687452144,
      "loss": 2.0732,
      "step": 39970
    },
    {
      "epoch": 4.81107099879663,
      "grad_norm": 9.6201171875,
      "learning_rate": 0.00019912559496074037,
      "loss": 1.9854,
      "step": 39980
    },
    {
      "epoch": 4.812274368231047,
      "grad_norm": 10.575060844421387,
      "learning_rate": 0.0001991250929035001,
      "loss": 1.964,
      "step": 39990
    },
    {
      "epoch": 4.813477737665464,
      "grad_norm": 6.967158317565918,
      "learning_rate": 0.00019912459070280133,
      "loss": 1.8992,
      "step": 40000
    },
    {
      "epoch": 4.814681107099879,
      "grad_norm": 5.697733402252197,
      "learning_rate": 0.00019912408835864482,
      "loss": 1.96,
      "step": 40010
    },
    {
      "epoch": 4.815884476534296,
      "grad_norm": 8.761931419372559,
      "learning_rate": 0.0001991235858710313,
      "loss": 2.086,
      "step": 40020
    },
    {
      "epoch": 4.817087845968713,
      "grad_norm": 8.758502006530762,
      "learning_rate": 0.00019912308323996142,
      "loss": 1.8458,
      "step": 40030
    },
    {
      "epoch": 4.8182912154031285,
      "grad_norm": 13.713561058044434,
      "learning_rate": 0.000199122580465436,
      "loss": 2.0011,
      "step": 40040
    },
    {
      "epoch": 4.819494584837545,
      "grad_norm": 5.9470534324646,
      "learning_rate": 0.00019912207754745574,
      "loss": 1.8996,
      "step": 40050
    },
    {
      "epoch": 4.820697954271962,
      "grad_norm": 8.278471946716309,
      "learning_rate": 0.00019912157448602136,
      "loss": 1.9504,
      "step": 40060
    },
    {
      "epoch": 4.821901323706378,
      "grad_norm": 7.7857346534729,
      "learning_rate": 0.00019912107128113358,
      "loss": 1.8783,
      "step": 40070
    },
    {
      "epoch": 4.823104693140794,
      "grad_norm": 5.615408420562744,
      "learning_rate": 0.00019912056793279314,
      "loss": 2.1982,
      "step": 40080
    },
    {
      "epoch": 4.824308062575211,
      "grad_norm": 9.775245666503906,
      "learning_rate": 0.00019912006444100077,
      "loss": 2.0322,
      "step": 40090
    },
    {
      "epoch": 4.825511432009627,
      "grad_norm": 5.027626037597656,
      "learning_rate": 0.00019911956080575718,
      "loss": 1.9425,
      "step": 40100
    },
    {
      "epoch": 4.8267148014440435,
      "grad_norm": 5.311755180358887,
      "learning_rate": 0.00019911905702706313,
      "loss": 2.1683,
      "step": 40110
    },
    {
      "epoch": 4.827918170878459,
      "grad_norm": 6.125672817230225,
      "learning_rate": 0.00019911855310491934,
      "loss": 1.982,
      "step": 40120
    },
    {
      "epoch": 4.829121540312876,
      "grad_norm": 8.904145240783691,
      "learning_rate": 0.00019911804903932653,
      "loss": 1.9801,
      "step": 40130
    },
    {
      "epoch": 4.830324909747293,
      "grad_norm": 7.311500072479248,
      "learning_rate": 0.00019911754483028542,
      "loss": 2.0804,
      "step": 40140
    },
    {
      "epoch": 4.8315282791817085,
      "grad_norm": 6.035048961639404,
      "learning_rate": 0.00019911704047779674,
      "loss": 2.0655,
      "step": 40150
    },
    {
      "epoch": 4.832731648616125,
      "grad_norm": 7.495687484741211,
      "learning_rate": 0.00019911653598186124,
      "loss": 2.0579,
      "step": 40160
    },
    {
      "epoch": 4.833935018050542,
      "grad_norm": 8.822660446166992,
      "learning_rate": 0.00019911603134247966,
      "loss": 1.9424,
      "step": 40170
    },
    {
      "epoch": 4.835138387484958,
      "grad_norm": 5.27180290222168,
      "learning_rate": 0.00019911552655965269,
      "loss": 2.0385,
      "step": 40180
    },
    {
      "epoch": 4.836341756919374,
      "grad_norm": 8.535886764526367,
      "learning_rate": 0.00019911502163338107,
      "loss": 1.9898,
      "step": 40190
    },
    {
      "epoch": 4.837545126353791,
      "grad_norm": 7.249156951904297,
      "learning_rate": 0.00019911451656366559,
      "loss": 1.6896,
      "step": 40200
    },
    {
      "epoch": 4.838748495788207,
      "grad_norm": 6.474113941192627,
      "learning_rate": 0.0001991140113505069,
      "loss": 2.0244,
      "step": 40210
    },
    {
      "epoch": 4.839951865222623,
      "grad_norm": 6.112491607666016,
      "learning_rate": 0.00019911350599390577,
      "loss": 2.0344,
      "step": 40220
    },
    {
      "epoch": 4.841155234657039,
      "grad_norm": 4.396851062774658,
      "learning_rate": 0.00019911300049386293,
      "loss": 1.8916,
      "step": 40230
    },
    {
      "epoch": 4.842358604091456,
      "grad_norm": 7.9455342292785645,
      "learning_rate": 0.0001991124948503791,
      "loss": 2.2096,
      "step": 40240
    },
    {
      "epoch": 4.843561973525873,
      "grad_norm": 6.555493354797363,
      "learning_rate": 0.00019911198906345502,
      "loss": 1.8789,
      "step": 40250
    },
    {
      "epoch": 4.844765342960288,
      "grad_norm": 4.341710567474365,
      "learning_rate": 0.00019911148313309145,
      "loss": 2.1128,
      "step": 40260
    },
    {
      "epoch": 4.845968712394705,
      "grad_norm": 7.658411502838135,
      "learning_rate": 0.00019911097705928906,
      "loss": 1.7827,
      "step": 40270
    },
    {
      "epoch": 4.847172081829122,
      "grad_norm": 6.984907150268555,
      "learning_rate": 0.00019911047084204863,
      "loss": 1.9475,
      "step": 40280
    },
    {
      "epoch": 4.8483754512635375,
      "grad_norm": 8.7039155960083,
      "learning_rate": 0.00019910996448137087,
      "loss": 2.2006,
      "step": 40290
    },
    {
      "epoch": 4.849578820697954,
      "grad_norm": 6.686063289642334,
      "learning_rate": 0.00019910945797725654,
      "loss": 2.1521,
      "step": 40300
    },
    {
      "epoch": 4.850782190132371,
      "grad_norm": 4.801222801208496,
      "learning_rate": 0.00019910895132970634,
      "loss": 2.0962,
      "step": 40310
    },
    {
      "epoch": 4.851985559566787,
      "grad_norm": 10.462652206420898,
      "learning_rate": 0.00019910844453872102,
      "loss": 2.0403,
      "step": 40320
    },
    {
      "epoch": 4.853188929001203,
      "grad_norm": 8.48914909362793,
      "learning_rate": 0.00019910793760430134,
      "loss": 1.9896,
      "step": 40330
    },
    {
      "epoch": 4.85439229843562,
      "grad_norm": 8.941189765930176,
      "learning_rate": 0.00019910743052644795,
      "loss": 2.2719,
      "step": 40340
    },
    {
      "epoch": 4.855595667870036,
      "grad_norm": 6.038933753967285,
      "learning_rate": 0.00019910692330516168,
      "loss": 1.8343,
      "step": 40350
    },
    {
      "epoch": 4.8567990373044525,
      "grad_norm": 5.75877046585083,
      "learning_rate": 0.00019910641594044322,
      "loss": 1.86,
      "step": 40360
    },
    {
      "epoch": 4.858002406738869,
      "grad_norm": 6.664186000823975,
      "learning_rate": 0.0001991059084322933,
      "loss": 2.0437,
      "step": 40370
    },
    {
      "epoch": 4.859205776173285,
      "grad_norm": 5.103994369506836,
      "learning_rate": 0.00019910540078071264,
      "loss": 1.9842,
      "step": 40380
    },
    {
      "epoch": 4.860409145607702,
      "grad_norm": 8.606149673461914,
      "learning_rate": 0.00019910489298570204,
      "loss": 2.1473,
      "step": 40390
    },
    {
      "epoch": 4.861612515042118,
      "grad_norm": 5.822866439819336,
      "learning_rate": 0.00019910438504726216,
      "loss": 1.7962,
      "step": 40400
    },
    {
      "epoch": 4.862815884476534,
      "grad_norm": 4.304443836212158,
      "learning_rate": 0.00019910387696539376,
      "loss": 1.8914,
      "step": 40410
    },
    {
      "epoch": 4.864019253910951,
      "grad_norm": 6.2228593826293945,
      "learning_rate": 0.0001991033687400976,
      "loss": 1.9029,
      "step": 40420
    },
    {
      "epoch": 4.8652226233453675,
      "grad_norm": 6.532971382141113,
      "learning_rate": 0.0001991028603713744,
      "loss": 2.2502,
      "step": 40430
    },
    {
      "epoch": 4.866425992779783,
      "grad_norm": 7.718857765197754,
      "learning_rate": 0.0001991023518592249,
      "loss": 2.063,
      "step": 40440
    },
    {
      "epoch": 4.8676293622142,
      "grad_norm": 4.780737400054932,
      "learning_rate": 0.0001991018432036498,
      "loss": 2.0256,
      "step": 40450
    },
    {
      "epoch": 4.868832731648616,
      "grad_norm": 4.792891979217529,
      "learning_rate": 0.00019910133440464992,
      "loss": 1.886,
      "step": 40460
    },
    {
      "epoch": 4.870036101083032,
      "grad_norm": 8.56484317779541,
      "learning_rate": 0.00019910082546222588,
      "loss": 2.0183,
      "step": 40470
    },
    {
      "epoch": 4.871239470517449,
      "grad_norm": 5.709110736846924,
      "learning_rate": 0.0001991003163763785,
      "loss": 1.916,
      "step": 40480
    },
    {
      "epoch": 4.872442839951865,
      "grad_norm": 10.676185607910156,
      "learning_rate": 0.0001990998071471085,
      "loss": 1.8557,
      "step": 40490
    },
    {
      "epoch": 4.873646209386282,
      "grad_norm": 8.244970321655273,
      "learning_rate": 0.00019909929777441661,
      "loss": 2.0351,
      "step": 40500
    },
    {
      "epoch": 4.874849578820698,
      "grad_norm": 5.744730472564697,
      "learning_rate": 0.00019909878825830358,
      "loss": 1.9496,
      "step": 40510
    },
    {
      "epoch": 4.876052948255114,
      "grad_norm": 9.303240776062012,
      "learning_rate": 0.00019909827859877015,
      "loss": 1.9293,
      "step": 40520
    },
    {
      "epoch": 4.877256317689531,
      "grad_norm": 8.385213851928711,
      "learning_rate": 0.00019909776879581701,
      "loss": 1.8325,
      "step": 40530
    },
    {
      "epoch": 4.878459687123947,
      "grad_norm": 9.659674644470215,
      "learning_rate": 0.00019909725884944494,
      "loss": 2.056,
      "step": 40540
    },
    {
      "epoch": 4.879663056558363,
      "grad_norm": 7.0810089111328125,
      "learning_rate": 0.00019909674875965468,
      "loss": 1.8832,
      "step": 40550
    },
    {
      "epoch": 4.88086642599278,
      "grad_norm": 4.187409400939941,
      "learning_rate": 0.00019909623852644695,
      "loss": 1.8321,
      "step": 40560
    },
    {
      "epoch": 4.882069795427196,
      "grad_norm": 7.634274005889893,
      "learning_rate": 0.00019909572814982252,
      "loss": 1.959,
      "step": 40570
    },
    {
      "epoch": 4.883273164861612,
      "grad_norm": 5.2782745361328125,
      "learning_rate": 0.00019909521762978208,
      "loss": 2.0625,
      "step": 40580
    },
    {
      "epoch": 4.884476534296029,
      "grad_norm": 7.979857444763184,
      "learning_rate": 0.00019909470696632638,
      "loss": 2.0411,
      "step": 40590
    },
    {
      "epoch": 4.885679903730445,
      "grad_norm": 4.926553249359131,
      "learning_rate": 0.00019909419615945622,
      "loss": 1.9154,
      "step": 40600
    },
    {
      "epoch": 4.8868832731648615,
      "grad_norm": 8.534509658813477,
      "learning_rate": 0.00019909368520917224,
      "loss": 2.051,
      "step": 40610
    },
    {
      "epoch": 4.888086642599278,
      "grad_norm": 10.87283992767334,
      "learning_rate": 0.00019909317411547525,
      "loss": 1.9651,
      "step": 40620
    },
    {
      "epoch": 4.889290012033694,
      "grad_norm": 10.513968467712402,
      "learning_rate": 0.000199092662878366,
      "loss": 2.195,
      "step": 40630
    },
    {
      "epoch": 4.890493381468111,
      "grad_norm": 7.238644123077393,
      "learning_rate": 0.00019909215149784516,
      "loss": 2.0604,
      "step": 40640
    },
    {
      "epoch": 4.891696750902527,
      "grad_norm": 8.014089584350586,
      "learning_rate": 0.00019909163997391355,
      "loss": 1.7281,
      "step": 40650
    },
    {
      "epoch": 4.892900120336943,
      "grad_norm": 12.775314331054688,
      "learning_rate": 0.00019909112830657186,
      "loss": 1.9691,
      "step": 40660
    },
    {
      "epoch": 4.89410348977136,
      "grad_norm": 7.021448135375977,
      "learning_rate": 0.00019909061649582083,
      "loss": 1.9499,
      "step": 40670
    },
    {
      "epoch": 4.8953068592057765,
      "grad_norm": 6.132105350494385,
      "learning_rate": 0.0001990901045416612,
      "loss": 2.0732,
      "step": 40680
    },
    {
      "epoch": 4.896510228640192,
      "grad_norm": 17.236846923828125,
      "learning_rate": 0.0001990895924440938,
      "loss": 1.8141,
      "step": 40690
    },
    {
      "epoch": 4.897713598074609,
      "grad_norm": 8.221872329711914,
      "learning_rate": 0.00019908908020311922,
      "loss": 1.8637,
      "step": 40700
    },
    {
      "epoch": 4.898916967509026,
      "grad_norm": 4.113970756530762,
      "learning_rate": 0.00019908856781873828,
      "loss": 2.1051,
      "step": 40710
    },
    {
      "epoch": 4.900120336943441,
      "grad_norm": 9.410972595214844,
      "learning_rate": 0.00019908805529095174,
      "loss": 1.9666,
      "step": 40720
    },
    {
      "epoch": 4.901323706377858,
      "grad_norm": 5.489748477935791,
      "learning_rate": 0.0001990875426197603,
      "loss": 1.8994,
      "step": 40730
    },
    {
      "epoch": 4.902527075812275,
      "grad_norm": 9.212996482849121,
      "learning_rate": 0.00019908702980516477,
      "loss": 2.1255,
      "step": 40740
    },
    {
      "epoch": 4.9037304452466906,
      "grad_norm": 11.50767707824707,
      "learning_rate": 0.0001990865168471658,
      "loss": 1.996,
      "step": 40750
    },
    {
      "epoch": 4.904933814681107,
      "grad_norm": 4.1979475021362305,
      "learning_rate": 0.00019908600374576415,
      "loss": 1.9806,
      "step": 40760
    },
    {
      "epoch": 4.906137184115524,
      "grad_norm": 5.681236743927002,
      "learning_rate": 0.00019908549050096067,
      "loss": 1.9222,
      "step": 40770
    },
    {
      "epoch": 4.90734055354994,
      "grad_norm": 6.431387901306152,
      "learning_rate": 0.00019908497711275595,
      "loss": 1.7968,
      "step": 40780
    },
    {
      "epoch": 4.908543922984356,
      "grad_norm": 8.566272735595703,
      "learning_rate": 0.00019908446358115085,
      "loss": 2.0927,
      "step": 40790
    },
    {
      "epoch": 4.909747292418773,
      "grad_norm": 5.922589302062988,
      "learning_rate": 0.00019908394990614606,
      "loss": 2.1326,
      "step": 40800
    },
    {
      "epoch": 4.910950661853189,
      "grad_norm": 6.080644607543945,
      "learning_rate": 0.00019908343608774232,
      "loss": 2.1035,
      "step": 40810
    },
    {
      "epoch": 4.9121540312876055,
      "grad_norm": 6.855172634124756,
      "learning_rate": 0.0001990829221259404,
      "loss": 1.8538,
      "step": 40820
    },
    {
      "epoch": 4.913357400722021,
      "grad_norm": 12.062527656555176,
      "learning_rate": 0.00019908240802074098,
      "loss": 1.9515,
      "step": 40830
    },
    {
      "epoch": 4.914560770156438,
      "grad_norm": 8.997509956359863,
      "learning_rate": 0.0001990818937721449,
      "loss": 1.8753,
      "step": 40840
    },
    {
      "epoch": 4.915764139590855,
      "grad_norm": 5.290858268737793,
      "learning_rate": 0.00019908137938015282,
      "loss": 1.8015,
      "step": 40850
    },
    {
      "epoch": 4.9169675090252705,
      "grad_norm": 5.533619403839111,
      "learning_rate": 0.00019908086484476557,
      "loss": 2.2231,
      "step": 40860
    },
    {
      "epoch": 4.918170878459687,
      "grad_norm": 6.96120548248291,
      "learning_rate": 0.00019908035016598383,
      "loss": 1.8157,
      "step": 40870
    },
    {
      "epoch": 4.919374247894104,
      "grad_norm": 11.576964378356934,
      "learning_rate": 0.00019907983534380838,
      "loss": 2.0325,
      "step": 40880
    },
    {
      "epoch": 4.92057761732852,
      "grad_norm": 10.48487663269043,
      "learning_rate": 0.0001990793203782399,
      "loss": 2.2996,
      "step": 40890
    },
    {
      "epoch": 4.921780986762936,
      "grad_norm": 6.319563388824463,
      "learning_rate": 0.0001990788052692792,
      "loss": 1.8925,
      "step": 40900
    },
    {
      "epoch": 4.922984356197353,
      "grad_norm": 18.32505989074707,
      "learning_rate": 0.00019907829001692703,
      "loss": 2.0216,
      "step": 40910
    },
    {
      "epoch": 4.924187725631769,
      "grad_norm": 5.896711826324463,
      "learning_rate": 0.00019907777462118407,
      "loss": 1.9168,
      "step": 40920
    },
    {
      "epoch": 4.925391095066185,
      "grad_norm": 10.592049598693848,
      "learning_rate": 0.00019907725908205112,
      "loss": 1.8219,
      "step": 40930
    },
    {
      "epoch": 4.926594464500601,
      "grad_norm": 17.71052360534668,
      "learning_rate": 0.00019907674339952894,
      "loss": 2.0158,
      "step": 40940
    },
    {
      "epoch": 4.927797833935018,
      "grad_norm": 6.570394992828369,
      "learning_rate": 0.00019907622757361822,
      "loss": 2.038,
      "step": 40950
    },
    {
      "epoch": 4.929001203369435,
      "grad_norm": 7.129170894622803,
      "learning_rate": 0.00019907571160431977,
      "loss": 2.0614,
      "step": 40960
    },
    {
      "epoch": 4.93020457280385,
      "grad_norm": 5.932984828948975,
      "learning_rate": 0.0001990751954916343,
      "loss": 2.1398,
      "step": 40970
    },
    {
      "epoch": 4.931407942238267,
      "grad_norm": 5.241306781768799,
      "learning_rate": 0.00019907467923556253,
      "loss": 2.0994,
      "step": 40980
    },
    {
      "epoch": 4.932611311672684,
      "grad_norm": 8.4271821975708,
      "learning_rate": 0.00019907416283610526,
      "loss": 2.0999,
      "step": 40990
    },
    {
      "epoch": 4.9338146811070995,
      "grad_norm": 20.040708541870117,
      "learning_rate": 0.0001990736462932632,
      "loss": 2.1312,
      "step": 41000
    },
    {
      "epoch": 4.935018050541516,
      "grad_norm": 4.027037143707275,
      "learning_rate": 0.00019907312960703715,
      "loss": 1.7862,
      "step": 41010
    },
    {
      "epoch": 4.936221419975933,
      "grad_norm": 7.310624122619629,
      "learning_rate": 0.0001990726127774278,
      "loss": 1.8866,
      "step": 41020
    },
    {
      "epoch": 4.937424789410349,
      "grad_norm": 18.408870697021484,
      "learning_rate": 0.0001990720958044359,
      "loss": 1.8095,
      "step": 41030
    },
    {
      "epoch": 4.938628158844765,
      "grad_norm": 8.137670516967773,
      "learning_rate": 0.00019907157868806225,
      "loss": 1.9862,
      "step": 41040
    },
    {
      "epoch": 4.939831528279182,
      "grad_norm": 7.439457416534424,
      "learning_rate": 0.00019907106142830756,
      "loss": 2.1416,
      "step": 41050
    },
    {
      "epoch": 4.941034897713598,
      "grad_norm": 10.49796199798584,
      "learning_rate": 0.00019907054402517254,
      "loss": 2.075,
      "step": 41060
    },
    {
      "epoch": 4.9422382671480145,
      "grad_norm": 10.617677688598633,
      "learning_rate": 0.00019907002647865806,
      "loss": 2.0176,
      "step": 41070
    },
    {
      "epoch": 4.943441636582431,
      "grad_norm": 10.57815933227539,
      "learning_rate": 0.00019906950878876474,
      "loss": 1.8455,
      "step": 41080
    },
    {
      "epoch": 4.944645006016847,
      "grad_norm": 11.920673370361328,
      "learning_rate": 0.0001990689909554934,
      "loss": 1.879,
      "step": 41090
    },
    {
      "epoch": 4.945848375451264,
      "grad_norm": 16.81657600402832,
      "learning_rate": 0.00019906847297884476,
      "loss": 1.935,
      "step": 41100
    },
    {
      "epoch": 4.94705174488568,
      "grad_norm": 5.1014323234558105,
      "learning_rate": 0.00019906795485881957,
      "loss": 1.8974,
      "step": 41110
    },
    {
      "epoch": 4.948255114320096,
      "grad_norm": 7.044492244720459,
      "learning_rate": 0.0001990674365954186,
      "loss": 1.8451,
      "step": 41120
    },
    {
      "epoch": 4.949458483754513,
      "grad_norm": 11.909997940063477,
      "learning_rate": 0.0001990669181886426,
      "loss": 2.1763,
      "step": 41130
    },
    {
      "epoch": 4.9506618531889295,
      "grad_norm": 7.357934951782227,
      "learning_rate": 0.00019906639963849232,
      "loss": 2.0923,
      "step": 41140
    },
    {
      "epoch": 4.951865222623345,
      "grad_norm": 9.347342491149902,
      "learning_rate": 0.00019906588094496848,
      "loss": 2.0536,
      "step": 41150
    },
    {
      "epoch": 4.953068592057762,
      "grad_norm": 8.337024688720703,
      "learning_rate": 0.00019906536210807186,
      "loss": 2.1872,
      "step": 41160
    },
    {
      "epoch": 4.954271961492178,
      "grad_norm": 8.844291687011719,
      "learning_rate": 0.0001990648431278032,
      "loss": 1.9118,
      "step": 41170
    },
    {
      "epoch": 4.955475330926594,
      "grad_norm": 5.540093898773193,
      "learning_rate": 0.00019906432400416325,
      "loss": 2.1319,
      "step": 41180
    },
    {
      "epoch": 4.956678700361011,
      "grad_norm": 18.86222267150879,
      "learning_rate": 0.00019906380473715276,
      "loss": 2.0275,
      "step": 41190
    },
    {
      "epoch": 4.957882069795427,
      "grad_norm": 6.521647930145264,
      "learning_rate": 0.00019906328532677246,
      "loss": 1.885,
      "step": 41200
    },
    {
      "epoch": 4.959085439229844,
      "grad_norm": 6.3068437576293945,
      "learning_rate": 0.0001990627657730232,
      "loss": 2.165,
      "step": 41210
    },
    {
      "epoch": 4.96028880866426,
      "grad_norm": 12.06706714630127,
      "learning_rate": 0.0001990622460759056,
      "loss": 2.2039,
      "step": 41220
    },
    {
      "epoch": 4.961492178098676,
      "grad_norm": 8.219376564025879,
      "learning_rate": 0.00019906172623542048,
      "loss": 2.0524,
      "step": 41230
    },
    {
      "epoch": 4.962695547533093,
      "grad_norm": 7.559345245361328,
      "learning_rate": 0.00019906120625156864,
      "loss": 1.7926,
      "step": 41240
    },
    {
      "epoch": 4.963898916967509,
      "grad_norm": 9.300811767578125,
      "learning_rate": 0.00019906068612435073,
      "loss": 1.985,
      "step": 41250
    },
    {
      "epoch": 4.965102286401925,
      "grad_norm": 5.420833110809326,
      "learning_rate": 0.00019906016585376755,
      "loss": 2.1751,
      "step": 41260
    },
    {
      "epoch": 4.966305655836342,
      "grad_norm": 9.66323471069336,
      "learning_rate": 0.00019905964543981988,
      "loss": 2.0754,
      "step": 41270
    },
    {
      "epoch": 4.967509025270758,
      "grad_norm": 7.217255592346191,
      "learning_rate": 0.0001990591248825084,
      "loss": 2.0565,
      "step": 41280
    },
    {
      "epoch": 4.968712394705174,
      "grad_norm": 8.194851875305176,
      "learning_rate": 0.00019905860418183396,
      "loss": 1.9816,
      "step": 41290
    },
    {
      "epoch": 4.969915764139591,
      "grad_norm": 5.018765926361084,
      "learning_rate": 0.0001990580833377972,
      "loss": 2.1554,
      "step": 41300
    },
    {
      "epoch": 4.971119133574007,
      "grad_norm": 5.87050724029541,
      "learning_rate": 0.000199057562350399,
      "loss": 2.1388,
      "step": 41310
    },
    {
      "epoch": 4.9723225030084235,
      "grad_norm": 8.783880233764648,
      "learning_rate": 0.00019905704121964002,
      "loss": 1.9649,
      "step": 41320
    },
    {
      "epoch": 4.97352587244284,
      "grad_norm": 7.442572116851807,
      "learning_rate": 0.00019905651994552105,
      "loss": 2.0547,
      "step": 41330
    },
    {
      "epoch": 4.974729241877256,
      "grad_norm": 20.498619079589844,
      "learning_rate": 0.00019905599852804284,
      "loss": 2.2141,
      "step": 41340
    },
    {
      "epoch": 4.975932611311673,
      "grad_norm": 30.7430362701416,
      "learning_rate": 0.00019905547696720614,
      "loss": 2.3052,
      "step": 41350
    },
    {
      "epoch": 4.977135980746089,
      "grad_norm": 6.986667156219482,
      "learning_rate": 0.00019905495526301173,
      "loss": 2.0343,
      "step": 41360
    },
    {
      "epoch": 4.978339350180505,
      "grad_norm": 8.35204792022705,
      "learning_rate": 0.0001990544334154603,
      "loss": 2.0233,
      "step": 41370
    },
    {
      "epoch": 4.979542719614922,
      "grad_norm": 4.920589447021484,
      "learning_rate": 0.0001990539114245527,
      "loss": 1.9176,
      "step": 41380
    },
    {
      "epoch": 4.9807460890493385,
      "grad_norm": 13.045265197753906,
      "learning_rate": 0.00019905338929028957,
      "loss": 2.0661,
      "step": 41390
    },
    {
      "epoch": 4.981949458483754,
      "grad_norm": 8.587635040283203,
      "learning_rate": 0.0001990528670126718,
      "loss": 2.1709,
      "step": 41400
    },
    {
      "epoch": 4.983152827918171,
      "grad_norm": 9.253944396972656,
      "learning_rate": 0.00019905234459170003,
      "loss": 1.8809,
      "step": 41410
    },
    {
      "epoch": 4.984356197352588,
      "grad_norm": 11.270646095275879,
      "learning_rate": 0.0001990518220273751,
      "loss": 2.0994,
      "step": 41420
    },
    {
      "epoch": 4.985559566787003,
      "grad_norm": 10.309446334838867,
      "learning_rate": 0.0001990512993196977,
      "loss": 2.0687,
      "step": 41430
    },
    {
      "epoch": 4.98676293622142,
      "grad_norm": 64.89546966552734,
      "learning_rate": 0.00019905077646866862,
      "loss": 2.2817,
      "step": 41440
    },
    {
      "epoch": 4.987966305655837,
      "grad_norm": 259.8202209472656,
      "learning_rate": 0.00019905025347428864,
      "loss": 2.873,
      "step": 41450
    },
    {
      "epoch": 4.9891696750902526,
      "grad_norm": 181.38555908203125,
      "learning_rate": 0.00019904973033655843,
      "loss": 4.1569,
      "step": 41460
    },
    {
      "epoch": 4.990373044524669,
      "grad_norm": 199.8432159423828,
      "learning_rate": 0.00019904920705547885,
      "loss": 3.7375,
      "step": 41470
    },
    {
      "epoch": 4.991576413959086,
      "grad_norm": 152.0066680908203,
      "learning_rate": 0.0001990486836310506,
      "loss": 4.5042,
      "step": 41480
    },
    {
      "epoch": 4.992779783393502,
      "grad_norm": 89.15633392333984,
      "learning_rate": 0.00019904816006327444,
      "loss": 3.5483,
      "step": 41490
    },
    {
      "epoch": 4.993983152827918,
      "grad_norm": 111.64482879638672,
      "learning_rate": 0.00019904763635215117,
      "loss": 3.1912,
      "step": 41500
    },
    {
      "epoch": 4.995186522262334,
      "grad_norm": 97.60633087158203,
      "learning_rate": 0.00019904711249768147,
      "loss": 3.2804,
      "step": 41510
    },
    {
      "epoch": 4.996389891696751,
      "grad_norm": 699.0576171875,
      "learning_rate": 0.0001990465884998662,
      "loss": 5.176,
      "step": 41520
    },
    {
      "epoch": 4.9975932611311675,
      "grad_norm": 52.12234878540039,
      "learning_rate": 0.000199046064358706,
      "loss": 7.5459,
      "step": 41530
    },
    {
      "epoch": 4.998796630565583,
      "grad_norm": 246.80284118652344,
      "learning_rate": 0.00019904554007420174,
      "loss": 7.5954,
      "step": 41540
    },
    {
      "epoch": 5.0,
      "grad_norm": 46.441307067871094,
      "learning_rate": 0.0001990450156463541,
      "loss": 7.6823,
      "step": 41550
    },
    {
      "epoch": 5.0,
      "eval_loss": 6.835190296173096,
      "eval_runtime": 119.0343,
      "eval_samples_per_second": 62.058,
      "eval_steps_per_second": 7.762,
      "step": 41550
    },
    {
      "epoch": 5.001203369434417,
      "grad_norm": 188.66549682617188,
      "learning_rate": 0.00019904449107516388,
      "loss": 6.5096,
      "step": 41560
    },
    {
      "epoch": 5.0024067388688325,
      "grad_norm": 26.782245635986328,
      "learning_rate": 0.00019904396636063183,
      "loss": 6.2541,
      "step": 41570
    },
    {
      "epoch": 5.003610108303249,
      "grad_norm": 90.96510314941406,
      "learning_rate": 0.0001990434415027587,
      "loss": 5.5302,
      "step": 41580
    },
    {
      "epoch": 5.004813477737666,
      "grad_norm": 107.69341278076172,
      "learning_rate": 0.00019904291650154528,
      "loss": 5.2539,
      "step": 41590
    },
    {
      "epoch": 5.006016847172082,
      "grad_norm": 110.74876403808594,
      "learning_rate": 0.0001990423913569923,
      "loss": 6.1839,
      "step": 41600
    },
    {
      "epoch": 5.007220216606498,
      "grad_norm": 84.80052185058594,
      "learning_rate": 0.0001990418660691005,
      "loss": 5.2663,
      "step": 41610
    },
    {
      "epoch": 5.008423586040915,
      "grad_norm": 163.00718688964844,
      "learning_rate": 0.0001990413406378707,
      "loss": 5.6696,
      "step": 41620
    },
    {
      "epoch": 5.009626955475331,
      "grad_norm": 52.932132720947266,
      "learning_rate": 0.0001990408150633036,
      "loss": 6.2381,
      "step": 41630
    },
    {
      "epoch": 5.0108303249097474,
      "grad_norm": 172.20726013183594,
      "learning_rate": 0.00019904028934540003,
      "loss": 6.0336,
      "step": 41640
    },
    {
      "epoch": 5.012033694344163,
      "grad_norm": 148.40406799316406,
      "learning_rate": 0.00019903976348416066,
      "loss": 7.1318,
      "step": 41650
    },
    {
      "epoch": 5.01323706377858,
      "grad_norm": 58.285980224609375,
      "learning_rate": 0.00019903923747958632,
      "loss": 7.3503,
      "step": 41660
    },
    {
      "epoch": 5.014440433212997,
      "grad_norm": 9.146730422973633,
      "learning_rate": 0.00019903871133167774,
      "loss": 7.6519,
      "step": 41670
    },
    {
      "epoch": 5.015643802647412,
      "grad_norm": 390.4382019042969,
      "learning_rate": 0.00019903818504043573,
      "loss": 8.1563,
      "step": 41680
    },
    {
      "epoch": 5.016847172081829,
      "grad_norm": 405.6904296875,
      "learning_rate": 0.00019903765860586098,
      "loss": 18.0557,
      "step": 41690
    },
    {
      "epoch": 5.018050541516246,
      "grad_norm": 552.901611328125,
      "learning_rate": 0.0001990371320279543,
      "loss": 8.5311,
      "step": 41700
    },
    {
      "epoch": 5.0192539109506615,
      "grad_norm": 14.789624214172363,
      "learning_rate": 0.00019903660530671644,
      "loss": 8.0326,
      "step": 41710
    },
    {
      "epoch": 5.020457280385078,
      "grad_norm": 77.91343688964844,
      "learning_rate": 0.00019903607844214816,
      "loss": 7.5159,
      "step": 41720
    },
    {
      "epoch": 5.021660649819495,
      "grad_norm": 217.33673095703125,
      "learning_rate": 0.0001990355514342502,
      "loss": 8.4779,
      "step": 41730
    },
    {
      "epoch": 5.022864019253911,
      "grad_norm": 24.981643676757812,
      "learning_rate": 0.00019903502428302337,
      "loss": 7.9218,
      "step": 41740
    },
    {
      "epoch": 5.024067388688327,
      "grad_norm": 19.880374908447266,
      "learning_rate": 0.00019903449698846842,
      "loss": 8.2566,
      "step": 41750
    },
    {
      "epoch": 5.025270758122744,
      "grad_norm": 9.139537811279297,
      "learning_rate": 0.00019903396955058607,
      "loss": 7.8532,
      "step": 41760
    },
    {
      "epoch": 5.02647412755716,
      "grad_norm": 31.869863510131836,
      "learning_rate": 0.00019903344196937713,
      "loss": 7.8622,
      "step": 41770
    },
    {
      "epoch": 5.0276774969915765,
      "grad_norm": 20.091535568237305,
      "learning_rate": 0.00019903291424484235,
      "loss": 7.8264,
      "step": 41780
    },
    {
      "epoch": 5.028880866425993,
      "grad_norm": 33.73579788208008,
      "learning_rate": 0.0001990323863769825,
      "loss": 7.6968,
      "step": 41790
    },
    {
      "epoch": 5.030084235860409,
      "grad_norm": 208.64324951171875,
      "learning_rate": 0.00019903185836579832,
      "loss": 8.0207,
      "step": 41800
    },
    {
      "epoch": 5.031287605294826,
      "grad_norm": 8.617533683776855,
      "learning_rate": 0.0001990313302112906,
      "loss": 8.0253,
      "step": 41810
    },
    {
      "epoch": 5.0324909747292415,
      "grad_norm": 53.43470764160156,
      "learning_rate": 0.00019903080191346008,
      "loss": 7.7598,
      "step": 41820
    },
    {
      "epoch": 5.033694344163658,
      "grad_norm": 7.555762767791748,
      "learning_rate": 0.00019903027347230757,
      "loss": 7.6914,
      "step": 41830
    },
    {
      "epoch": 5.034897713598075,
      "grad_norm": 10.855253219604492,
      "learning_rate": 0.0001990297448878338,
      "loss": 7.8153,
      "step": 41840
    },
    {
      "epoch": 5.036101083032491,
      "grad_norm": 15.845646858215332,
      "learning_rate": 0.0001990292161600395,
      "loss": 7.7012,
      "step": 41850
    },
    {
      "epoch": 5.037304452466907,
      "grad_norm": 8.219528198242188,
      "learning_rate": 0.00019902868728892553,
      "loss": 7.6016,
      "step": 41860
    },
    {
      "epoch": 5.038507821901324,
      "grad_norm": 16.66413116455078,
      "learning_rate": 0.00019902815827449257,
      "loss": 7.435,
      "step": 41870
    },
    {
      "epoch": 5.03971119133574,
      "grad_norm": 40.152835845947266,
      "learning_rate": 0.0001990276291167414,
      "loss": 7.9836,
      "step": 41880
    },
    {
      "epoch": 5.040914560770156,
      "grad_norm": 16.253328323364258,
      "learning_rate": 0.00019902709981567282,
      "loss": 7.6251,
      "step": 41890
    },
    {
      "epoch": 5.042117930204573,
      "grad_norm": 16.718053817749023,
      "learning_rate": 0.0001990265703712876,
      "loss": 7.5786,
      "step": 41900
    },
    {
      "epoch": 5.043321299638989,
      "grad_norm": 27.236248016357422,
      "learning_rate": 0.00019902604078358643,
      "loss": 7.4927,
      "step": 41910
    },
    {
      "epoch": 5.044524669073406,
      "grad_norm": 9.109397888183594,
      "learning_rate": 0.0001990255110525702,
      "loss": 7.4796,
      "step": 41920
    },
    {
      "epoch": 5.045728038507822,
      "grad_norm": 30.82512855529785,
      "learning_rate": 0.00019902498117823955,
      "loss": 7.5358,
      "step": 41930
    },
    {
      "epoch": 5.046931407942238,
      "grad_norm": 19.24618148803711,
      "learning_rate": 0.00019902445116059533,
      "loss": 7.5374,
      "step": 41940
    },
    {
      "epoch": 5.048134777376655,
      "grad_norm": 8.26912784576416,
      "learning_rate": 0.00019902392099963826,
      "loss": 7.5509,
      "step": 41950
    },
    {
      "epoch": 5.049338146811071,
      "grad_norm": 15.063525199890137,
      "learning_rate": 0.00019902339069536917,
      "loss": 7.505,
      "step": 41960
    },
    {
      "epoch": 5.050541516245487,
      "grad_norm": 6.537913799285889,
      "learning_rate": 0.00019902286024778873,
      "loss": 7.538,
      "step": 41970
    },
    {
      "epoch": 5.051744885679904,
      "grad_norm": 13.247450828552246,
      "learning_rate": 0.0001990223296568978,
      "loss": 7.4489,
      "step": 41980
    },
    {
      "epoch": 5.05294825511432,
      "grad_norm": 88.28585815429688,
      "learning_rate": 0.0001990217989226971,
      "loss": 7.6089,
      "step": 41990
    },
    {
      "epoch": 5.054151624548736,
      "grad_norm": 12.32447624206543,
      "learning_rate": 0.0001990212680451874,
      "loss": 7.5718,
      "step": 42000
    },
    {
      "epoch": 5.055354993983153,
      "grad_norm": 22.057758331298828,
      "learning_rate": 0.00019902073702436952,
      "loss": 7.6186,
      "step": 42010
    },
    {
      "epoch": 5.056558363417569,
      "grad_norm": 12.481575965881348,
      "learning_rate": 0.00019902020586024414,
      "loss": 7.5671,
      "step": 42020
    },
    {
      "epoch": 5.0577617328519855,
      "grad_norm": 16.605792999267578,
      "learning_rate": 0.00019901967455281209,
      "loss": 7.4652,
      "step": 42030
    },
    {
      "epoch": 5.058965102286402,
      "grad_norm": 43.19899368286133,
      "learning_rate": 0.00019901914310207412,
      "loss": 7.4846,
      "step": 42040
    },
    {
      "epoch": 5.060168471720818,
      "grad_norm": 13.148046493530273,
      "learning_rate": 0.000199018611508031,
      "loss": 7.4514,
      "step": 42050
    },
    {
      "epoch": 5.061371841155235,
      "grad_norm": 64.03712463378906,
      "learning_rate": 0.00019901807977068352,
      "loss": 7.6252,
      "step": 42060
    },
    {
      "epoch": 5.062575210589651,
      "grad_norm": 25.0727481842041,
      "learning_rate": 0.00019901754789003242,
      "loss": 7.7015,
      "step": 42070
    },
    {
      "epoch": 5.063778580024067,
      "grad_norm": 18.119224548339844,
      "learning_rate": 0.00019901701586607848,
      "loss": 7.7114,
      "step": 42080
    },
    {
      "epoch": 5.064981949458484,
      "grad_norm": 7.71126127243042,
      "learning_rate": 0.00019901648369882247,
      "loss": 7.6586,
      "step": 42090
    },
    {
      "epoch": 5.0661853188929005,
      "grad_norm": 13.048935890197754,
      "learning_rate": 0.00019901595138826516,
      "loss": 7.9082,
      "step": 42100
    },
    {
      "epoch": 5.067388688327316,
      "grad_norm": 5.803314685821533,
      "learning_rate": 0.00019901541893440735,
      "loss": 7.5679,
      "step": 42110
    },
    {
      "epoch": 5.068592057761733,
      "grad_norm": 4.404675006866455,
      "learning_rate": 0.00019901488633724976,
      "loss": 7.6752,
      "step": 42120
    },
    {
      "epoch": 5.06979542719615,
      "grad_norm": 22.034263610839844,
      "learning_rate": 0.00019901435359679316,
      "loss": 7.8712,
      "step": 42130
    },
    {
      "epoch": 5.070998796630565,
      "grad_norm": 8.816366195678711,
      "learning_rate": 0.00019901382071303838,
      "loss": 7.7475,
      "step": 42140
    },
    {
      "epoch": 5.072202166064982,
      "grad_norm": 5.255917072296143,
      "learning_rate": 0.00019901328768598612,
      "loss": 7.6526,
      "step": 42150
    },
    {
      "epoch": 5.073405535499398,
      "grad_norm": 7.4424543380737305,
      "learning_rate": 0.00019901275451563724,
      "loss": 7.5585,
      "step": 42160
    },
    {
      "epoch": 5.074608904933815,
      "grad_norm": 5.376596450805664,
      "learning_rate": 0.00019901222120199244,
      "loss": 7.4951,
      "step": 42170
    },
    {
      "epoch": 5.075812274368231,
      "grad_norm": 9.995231628417969,
      "learning_rate": 0.00019901168774505247,
      "loss": 7.271,
      "step": 42180
    },
    {
      "epoch": 5.077015643802647,
      "grad_norm": 9.695402145385742,
      "learning_rate": 0.00019901115414481816,
      "loss": 7.312,
      "step": 42190
    },
    {
      "epoch": 5.078219013237064,
      "grad_norm": 14.717586517333984,
      "learning_rate": 0.0001990106204012903,
      "loss": 7.44,
      "step": 42200
    },
    {
      "epoch": 5.07942238267148,
      "grad_norm": 8.688631057739258,
      "learning_rate": 0.00019901008651446959,
      "loss": 7.2563,
      "step": 42210
    },
    {
      "epoch": 5.080625752105896,
      "grad_norm": 6.643031120300293,
      "learning_rate": 0.0001990095524843569,
      "loss": 7.275,
      "step": 42220
    },
    {
      "epoch": 5.081829121540313,
      "grad_norm": 12.19149398803711,
      "learning_rate": 0.00019900901831095287,
      "loss": 7.2457,
      "step": 42230
    },
    {
      "epoch": 5.0830324909747295,
      "grad_norm": 9.875896453857422,
      "learning_rate": 0.00019900848399425838,
      "loss": 7.2449,
      "step": 42240
    },
    {
      "epoch": 5.084235860409145,
      "grad_norm": 8.756649017333984,
      "learning_rate": 0.00019900794953427414,
      "loss": 7.1868,
      "step": 42250
    },
    {
      "epoch": 5.085439229843562,
      "grad_norm": 45.32009506225586,
      "learning_rate": 0.000199007414931001,
      "loss": 7.1647,
      "step": 42260
    },
    {
      "epoch": 5.086642599277979,
      "grad_norm": 10.563359260559082,
      "learning_rate": 0.00019900688018443968,
      "loss": 7.2773,
      "step": 42270
    },
    {
      "epoch": 5.0878459687123945,
      "grad_norm": 5.818143367767334,
      "learning_rate": 0.0001990063452945909,
      "loss": 7.0704,
      "step": 42280
    },
    {
      "epoch": 5.089049338146811,
      "grad_norm": 6.811473369598389,
      "learning_rate": 0.00019900581026145554,
      "loss": 6.9841,
      "step": 42290
    },
    {
      "epoch": 5.090252707581228,
      "grad_norm": 8.492681503295898,
      "learning_rate": 0.00019900527508503432,
      "loss": 7.0224,
      "step": 42300
    },
    {
      "epoch": 5.091456077015644,
      "grad_norm": 6.135493278503418,
      "learning_rate": 0.00019900473976532806,
      "loss": 6.8552,
      "step": 42310
    },
    {
      "epoch": 5.09265944645006,
      "grad_norm": 31.769546508789062,
      "learning_rate": 0.00019900420430233747,
      "loss": 7.4176,
      "step": 42320
    },
    {
      "epoch": 5.093862815884476,
      "grad_norm": 15.368035316467285,
      "learning_rate": 0.00019900366869606332,
      "loss": 7.6599,
      "step": 42330
    },
    {
      "epoch": 5.095066185318893,
      "grad_norm": 20.432085037231445,
      "learning_rate": 0.00019900313294650648,
      "loss": 7.6455,
      "step": 42340
    },
    {
      "epoch": 5.0962695547533094,
      "grad_norm": 8.058140754699707,
      "learning_rate": 0.00019900259705366762,
      "loss": 7.5279,
      "step": 42350
    },
    {
      "epoch": 5.097472924187725,
      "grad_norm": 14.024980545043945,
      "learning_rate": 0.00019900206101754756,
      "loss": 7.2253,
      "step": 42360
    },
    {
      "epoch": 5.098676293622142,
      "grad_norm": 4.082953453063965,
      "learning_rate": 0.0001990015248381471,
      "loss": 7.2084,
      "step": 42370
    },
    {
      "epoch": 5.099879663056559,
      "grad_norm": 10.778286933898926,
      "learning_rate": 0.00019900098851546698,
      "loss": 7.2656,
      "step": 42380
    },
    {
      "epoch": 5.101083032490974,
      "grad_norm": 4.0808491706848145,
      "learning_rate": 0.000199000452049508,
      "loss": 7.0583,
      "step": 42390
    },
    {
      "epoch": 5.102286401925391,
      "grad_norm": 8.811480522155762,
      "learning_rate": 0.0001989999154402709,
      "loss": 7.0453,
      "step": 42400
    },
    {
      "epoch": 5.103489771359808,
      "grad_norm": 14.220723152160645,
      "learning_rate": 0.00019899937868775647,
      "loss": 6.9582,
      "step": 42410
    },
    {
      "epoch": 5.1046931407942235,
      "grad_norm": 6.13651180267334,
      "learning_rate": 0.00019899884179196554,
      "loss": 6.9618,
      "step": 42420
    },
    {
      "epoch": 5.10589651022864,
      "grad_norm": 13.354302406311035,
      "learning_rate": 0.00019899830475289881,
      "loss": 6.9541,
      "step": 42430
    },
    {
      "epoch": 5.107099879663057,
      "grad_norm": 6.9171576499938965,
      "learning_rate": 0.00019899776757055707,
      "loss": 6.9258,
      "step": 42440
    },
    {
      "epoch": 5.108303249097473,
      "grad_norm": 27.351240158081055,
      "learning_rate": 0.00019899723024494116,
      "loss": 6.8234,
      "step": 42450
    },
    {
      "epoch": 5.109506618531889,
      "grad_norm": 39.086952209472656,
      "learning_rate": 0.0001989966927760518,
      "loss": 7.5022,
      "step": 42460
    },
    {
      "epoch": 5.110709987966306,
      "grad_norm": 6.120758056640625,
      "learning_rate": 0.0001989961551638898,
      "loss": 7.5716,
      "step": 42470
    },
    {
      "epoch": 5.111913357400722,
      "grad_norm": 11.316590309143066,
      "learning_rate": 0.00019899561740845592,
      "loss": 7.2036,
      "step": 42480
    },
    {
      "epoch": 5.1131167268351385,
      "grad_norm": 22.515880584716797,
      "learning_rate": 0.0001989950795097509,
      "loss": 6.9729,
      "step": 42490
    },
    {
      "epoch": 5.114320096269555,
      "grad_norm": 26.785188674926758,
      "learning_rate": 0.00019899454146777562,
      "loss": 7.164,
      "step": 42500
    },
    {
      "epoch": 5.115523465703971,
      "grad_norm": 93.84312438964844,
      "learning_rate": 0.00019899400328253074,
      "loss": 7.1441,
      "step": 42510
    },
    {
      "epoch": 5.116726835138388,
      "grad_norm": 59.40557861328125,
      "learning_rate": 0.00019899346495401714,
      "loss": 7.0329,
      "step": 42520
    },
    {
      "epoch": 5.1179302045728035,
      "grad_norm": 15.077173233032227,
      "learning_rate": 0.00019899292648223554,
      "loss": 7.3618,
      "step": 42530
    },
    {
      "epoch": 5.11913357400722,
      "grad_norm": 14.304474830627441,
      "learning_rate": 0.00019899238786718671,
      "loss": 7.0873,
      "step": 42540
    },
    {
      "epoch": 5.120336943441637,
      "grad_norm": 13.122438430786133,
      "learning_rate": 0.0001989918491088715,
      "loss": 7.2357,
      "step": 42550
    },
    {
      "epoch": 5.121540312876053,
      "grad_norm": 58.743526458740234,
      "learning_rate": 0.00019899131020729063,
      "loss": 7.1096,
      "step": 42560
    },
    {
      "epoch": 5.122743682310469,
      "grad_norm": 11.453507423400879,
      "learning_rate": 0.0001989907711624449,
      "loss": 7.2142,
      "step": 42570
    },
    {
      "epoch": 5.123947051744886,
      "grad_norm": 24.48541259765625,
      "learning_rate": 0.00019899023197433505,
      "loss": 7.0047,
      "step": 42580
    },
    {
      "epoch": 5.125150421179302,
      "grad_norm": 18.91883087158203,
      "learning_rate": 0.00019898969264296192,
      "loss": 6.9835,
      "step": 42590
    },
    {
      "epoch": 5.126353790613718,
      "grad_norm": 24.201261520385742,
      "learning_rate": 0.00019898915316832625,
      "loss": 7.1037,
      "step": 42600
    },
    {
      "epoch": 5.127557160048135,
      "grad_norm": 7.234903812408447,
      "learning_rate": 0.00019898861355042885,
      "loss": 6.9852,
      "step": 42610
    },
    {
      "epoch": 5.128760529482551,
      "grad_norm": 9.424560546875,
      "learning_rate": 0.00019898807378927047,
      "loss": 6.9081,
      "step": 42620
    },
    {
      "epoch": 5.129963898916968,
      "grad_norm": 21.928312301635742,
      "learning_rate": 0.0001989875338848519,
      "loss": 6.9391,
      "step": 42630
    },
    {
      "epoch": 5.131167268351384,
      "grad_norm": 14.327727317810059,
      "learning_rate": 0.000198986993837174,
      "loss": 6.9213,
      "step": 42640
    },
    {
      "epoch": 5.1323706377858,
      "grad_norm": 294.3425598144531,
      "learning_rate": 0.0001989864536462374,
      "loss": 6.962,
      "step": 42650
    },
    {
      "epoch": 5.133574007220217,
      "grad_norm": 30.27318572998047,
      "learning_rate": 0.00019898591331204298,
      "loss": 7.1761,
      "step": 42660
    },
    {
      "epoch": 5.1347773766546325,
      "grad_norm": 13.551473617553711,
      "learning_rate": 0.00019898537283459153,
      "loss": 7.1354,
      "step": 42670
    },
    {
      "epoch": 5.135980746089049,
      "grad_norm": 17.305870056152344,
      "learning_rate": 0.00019898483221388383,
      "loss": 7.0328,
      "step": 42680
    },
    {
      "epoch": 5.137184115523466,
      "grad_norm": 462.5865478515625,
      "learning_rate": 0.00019898429144992057,
      "loss": 6.912,
      "step": 42690
    },
    {
      "epoch": 5.138387484957882,
      "grad_norm": 283.6278076171875,
      "learning_rate": 0.00019898375054270263,
      "loss": 7.2048,
      "step": 42700
    },
    {
      "epoch": 5.139590854392298,
      "grad_norm": 13.909567832946777,
      "learning_rate": 0.00019898320949223076,
      "loss": 7.026,
      "step": 42710
    },
    {
      "epoch": 5.140794223826715,
      "grad_norm": 54.53024673461914,
      "learning_rate": 0.00019898266829850578,
      "loss": 7.0217,
      "step": 42720
    },
    {
      "epoch": 5.141997593261131,
      "grad_norm": 82.68240356445312,
      "learning_rate": 0.0001989821269615284,
      "loss": 7.0224,
      "step": 42730
    },
    {
      "epoch": 5.1432009626955475,
      "grad_norm": 41.48139953613281,
      "learning_rate": 0.0001989815854812995,
      "loss": 6.9344,
      "step": 42740
    },
    {
      "epoch": 5.144404332129964,
      "grad_norm": 21.438369750976562,
      "learning_rate": 0.00019898104385781976,
      "loss": 7.1404,
      "step": 42750
    },
    {
      "epoch": 5.14560770156438,
      "grad_norm": 32.261722564697266,
      "learning_rate": 0.00019898050209109003,
      "loss": 7.0036,
      "step": 42760
    },
    {
      "epoch": 5.146811070998797,
      "grad_norm": 28.524431228637695,
      "learning_rate": 0.00019897996018111106,
      "loss": 7.0866,
      "step": 42770
    },
    {
      "epoch": 5.148014440433213,
      "grad_norm": 82.64258575439453,
      "learning_rate": 0.00019897941812788364,
      "loss": 7.144,
      "step": 42780
    },
    {
      "epoch": 5.149217809867629,
      "grad_norm": 13.357208251953125,
      "learning_rate": 0.00019897887593140861,
      "loss": 7.4019,
      "step": 42790
    },
    {
      "epoch": 5.150421179302046,
      "grad_norm": 6.039828777313232,
      "learning_rate": 0.00019897833359168668,
      "loss": 7.2395,
      "step": 42800
    },
    {
      "epoch": 5.1516245487364625,
      "grad_norm": 21.577228546142578,
      "learning_rate": 0.00019897779110871866,
      "loss": 7.0816,
      "step": 42810
    },
    {
      "epoch": 5.152827918170878,
      "grad_norm": 27.31770133972168,
      "learning_rate": 0.00019897724848250535,
      "loss": 6.9184,
      "step": 42820
    },
    {
      "epoch": 5.154031287605295,
      "grad_norm": 24.126413345336914,
      "learning_rate": 0.00019897670571304753,
      "loss": 6.9494,
      "step": 42830
    },
    {
      "epoch": 5.155234657039712,
      "grad_norm": 17.958532333374023,
      "learning_rate": 0.00019897616280034597,
      "loss": 7.1084,
      "step": 42840
    },
    {
      "epoch": 5.156438026474127,
      "grad_norm": 11.24973201751709,
      "learning_rate": 0.00019897561974440144,
      "loss": 7.0399,
      "step": 42850
    },
    {
      "epoch": 5.157641395908544,
      "grad_norm": 7.521182537078857,
      "learning_rate": 0.0001989750765452148,
      "loss": 6.9667,
      "step": 42860
    },
    {
      "epoch": 5.15884476534296,
      "grad_norm": 7.6414642333984375,
      "learning_rate": 0.00019897453320278676,
      "loss": 6.9056,
      "step": 42870
    },
    {
      "epoch": 5.160048134777377,
      "grad_norm": 4.558192253112793,
      "learning_rate": 0.00019897398971711815,
      "loss": 6.8337,
      "step": 42880
    },
    {
      "epoch": 5.161251504211793,
      "grad_norm": 6.217107772827148,
      "learning_rate": 0.00019897344608820974,
      "loss": 6.8912,
      "step": 42890
    },
    {
      "epoch": 5.162454873646209,
      "grad_norm": 8.388609886169434,
      "learning_rate": 0.0001989729023160623,
      "loss": 6.7866,
      "step": 42900
    },
    {
      "epoch": 5.163658243080626,
      "grad_norm": 4.573419570922852,
      "learning_rate": 0.00019897235840067665,
      "loss": 6.8123,
      "step": 42910
    },
    {
      "epoch": 5.164861612515042,
      "grad_norm": 8.137279510498047,
      "learning_rate": 0.00019897181434205358,
      "loss": 6.7084,
      "step": 42920
    },
    {
      "epoch": 5.166064981949458,
      "grad_norm": 9.95292854309082,
      "learning_rate": 0.00019897127014019384,
      "loss": 6.8599,
      "step": 42930
    },
    {
      "epoch": 5.167268351383875,
      "grad_norm": 18.444242477416992,
      "learning_rate": 0.0001989707257950982,
      "loss": 6.7955,
      "step": 42940
    },
    {
      "epoch": 5.1684717208182915,
      "grad_norm": 13.281579971313477,
      "learning_rate": 0.00019897018130676753,
      "loss": 6.7585,
      "step": 42950
    },
    {
      "epoch": 5.169675090252707,
      "grad_norm": 11.055079460144043,
      "learning_rate": 0.00019896963667520256,
      "loss": 6.8016,
      "step": 42960
    },
    {
      "epoch": 5.170878459687124,
      "grad_norm": 5.975857257843018,
      "learning_rate": 0.0001989690919004041,
      "loss": 6.8301,
      "step": 42970
    },
    {
      "epoch": 5.172081829121541,
      "grad_norm": 8.398676872253418,
      "learning_rate": 0.0001989685469823729,
      "loss": 6.8602,
      "step": 42980
    },
    {
      "epoch": 5.1732851985559565,
      "grad_norm": 14.473787307739258,
      "learning_rate": 0.0001989680019211098,
      "loss": 6.9072,
      "step": 42990
    },
    {
      "epoch": 5.174488567990373,
      "grad_norm": 182.6295928955078,
      "learning_rate": 0.0001989674567166156,
      "loss": 6.7887,
      "step": 43000
    },
    {
      "epoch": 5.17569193742479,
      "grad_norm": 13.410714149475098,
      "learning_rate": 0.00019896691136889102,
      "loss": 6.8109,
      "step": 43010
    },
    {
      "epoch": 5.176895306859206,
      "grad_norm": 33.2347526550293,
      "learning_rate": 0.00019896636587793685,
      "loss": 6.9216,
      "step": 43020
    },
    {
      "epoch": 5.178098676293622,
      "grad_norm": 4.994111061096191,
      "learning_rate": 0.00019896582024375397,
      "loss": 7.3826,
      "step": 43030
    },
    {
      "epoch": 5.179302045728038,
      "grad_norm": 10.641875267028809,
      "learning_rate": 0.0001989652744663431,
      "loss": 7.2316,
      "step": 43040
    },
    {
      "epoch": 5.180505415162455,
      "grad_norm": 50.609764099121094,
      "learning_rate": 0.00019896472854570503,
      "loss": 7.1885,
      "step": 43050
    },
    {
      "epoch": 5.1817087845968715,
      "grad_norm": 16.520709991455078,
      "learning_rate": 0.00019896418248184058,
      "loss": 6.9104,
      "step": 43060
    },
    {
      "epoch": 5.182912154031287,
      "grad_norm": 15.513710021972656,
      "learning_rate": 0.00019896363627475052,
      "loss": 6.9581,
      "step": 43070
    },
    {
      "epoch": 5.184115523465704,
      "grad_norm": 16.634061813354492,
      "learning_rate": 0.00019896308992443565,
      "loss": 6.8967,
      "step": 43080
    },
    {
      "epoch": 5.185318892900121,
      "grad_norm": 12.569376945495605,
      "learning_rate": 0.00019896254343089676,
      "loss": 7.1891,
      "step": 43090
    },
    {
      "epoch": 5.186522262334536,
      "grad_norm": 9.628414154052734,
      "learning_rate": 0.00019896199679413461,
      "loss": 7.0908,
      "step": 43100
    },
    {
      "epoch": 5.187725631768953,
      "grad_norm": 16.586896896362305,
      "learning_rate": 0.00019896145001415004,
      "loss": 6.8478,
      "step": 43110
    },
    {
      "epoch": 5.18892900120337,
      "grad_norm": 19.666330337524414,
      "learning_rate": 0.00019896090309094383,
      "loss": 6.9485,
      "step": 43120
    },
    {
      "epoch": 5.1901323706377855,
      "grad_norm": 27.628435134887695,
      "learning_rate": 0.00019896035602451673,
      "loss": 6.7844,
      "step": 43130
    },
    {
      "epoch": 5.191335740072202,
      "grad_norm": 41.259517669677734,
      "learning_rate": 0.00019895980881486956,
      "loss": 6.8364,
      "step": 43140
    },
    {
      "epoch": 5.192539109506619,
      "grad_norm": 25.46467399597168,
      "learning_rate": 0.00019895926146200313,
      "loss": 6.7461,
      "step": 43150
    },
    {
      "epoch": 5.193742478941035,
      "grad_norm": 19.08307456970215,
      "learning_rate": 0.00019895871396591823,
      "loss": 6.8297,
      "step": 43160
    },
    {
      "epoch": 5.194945848375451,
      "grad_norm": 18.73707389831543,
      "learning_rate": 0.00019895816632661562,
      "loss": 6.8379,
      "step": 43170
    },
    {
      "epoch": 5.196149217809868,
      "grad_norm": 11.898574829101562,
      "learning_rate": 0.00019895761854409614,
      "loss": 6.7958,
      "step": 43180
    },
    {
      "epoch": 5.197352587244284,
      "grad_norm": 10.803080558776855,
      "learning_rate": 0.0001989570706183605,
      "loss": 6.8844,
      "step": 43190
    },
    {
      "epoch": 5.1985559566787005,
      "grad_norm": 7.940894603729248,
      "learning_rate": 0.00019895652254940958,
      "loss": 6.7806,
      "step": 43200
    },
    {
      "epoch": 5.199759326113116,
      "grad_norm": 9.130962371826172,
      "learning_rate": 0.00019895597433724417,
      "loss": 6.7556,
      "step": 43210
    },
    {
      "epoch": 5.200962695547533,
      "grad_norm": 6.850533962249756,
      "learning_rate": 0.000198955425981865,
      "loss": 6.5902,
      "step": 43220
    },
    {
      "epoch": 5.20216606498195,
      "grad_norm": 5.628232002258301,
      "learning_rate": 0.00019895487748327288,
      "loss": 6.812,
      "step": 43230
    },
    {
      "epoch": 5.2033694344163655,
      "grad_norm": 17.266263961791992,
      "learning_rate": 0.00019895432884146864,
      "loss": 6.7475,
      "step": 43240
    },
    {
      "epoch": 5.204572803850782,
      "grad_norm": 20.48459243774414,
      "learning_rate": 0.00019895378005645306,
      "loss": 6.7628,
      "step": 43250
    },
    {
      "epoch": 5.205776173285199,
      "grad_norm": 7.41314697265625,
      "learning_rate": 0.0001989532311282269,
      "loss": 6.6927,
      "step": 43260
    },
    {
      "epoch": 5.206979542719615,
      "grad_norm": 7.002894878387451,
      "learning_rate": 0.00019895268205679103,
      "loss": 6.8254,
      "step": 43270
    },
    {
      "epoch": 5.208182912154031,
      "grad_norm": 11.923335075378418,
      "learning_rate": 0.00019895213284214618,
      "loss": 6.6697,
      "step": 43280
    },
    {
      "epoch": 5.209386281588448,
      "grad_norm": 11.946123123168945,
      "learning_rate": 0.00019895158348429316,
      "loss": 6.7241,
      "step": 43290
    },
    {
      "epoch": 5.210589651022864,
      "grad_norm": 13.587684631347656,
      "learning_rate": 0.00019895103398323279,
      "loss": 6.7099,
      "step": 43300
    },
    {
      "epoch": 5.21179302045728,
      "grad_norm": 7.899662017822266,
      "learning_rate": 0.00019895048433896578,
      "loss": 6.6365,
      "step": 43310
    },
    {
      "epoch": 5.212996389891697,
      "grad_norm": 129.24859619140625,
      "learning_rate": 0.00019894993455149305,
      "loss": 6.8448,
      "step": 43320
    },
    {
      "epoch": 5.214199759326113,
      "grad_norm": 29.2088680267334,
      "learning_rate": 0.0001989493846208153,
      "loss": 6.9342,
      "step": 43330
    },
    {
      "epoch": 5.21540312876053,
      "grad_norm": 15.965337753295898,
      "learning_rate": 0.00019894883454693338,
      "loss": 6.7689,
      "step": 43340
    },
    {
      "epoch": 5.216606498194946,
      "grad_norm": 10.934747695922852,
      "learning_rate": 0.00019894828432984806,
      "loss": 6.7079,
      "step": 43350
    },
    {
      "epoch": 5.217809867629362,
      "grad_norm": 8.470879554748535,
      "learning_rate": 0.00019894773396956017,
      "loss": 6.8349,
      "step": 43360
    },
    {
      "epoch": 5.219013237063779,
      "grad_norm": 21.610458374023438,
      "learning_rate": 0.00019894718346607045,
      "loss": 6.8553,
      "step": 43370
    },
    {
      "epoch": 5.2202166064981945,
      "grad_norm": 7.955434322357178,
      "learning_rate": 0.00019894663281937972,
      "loss": 6.707,
      "step": 43380
    },
    {
      "epoch": 5.221419975932611,
      "grad_norm": 24.681379318237305,
      "learning_rate": 0.0001989460820294888,
      "loss": 6.7217,
      "step": 43390
    },
    {
      "epoch": 5.222623345367028,
      "grad_norm": 42.13639831542969,
      "learning_rate": 0.00019894553109639848,
      "loss": 6.994,
      "step": 43400
    },
    {
      "epoch": 5.223826714801444,
      "grad_norm": 91.69528198242188,
      "learning_rate": 0.00019894498002010952,
      "loss": 7.0314,
      "step": 43410
    },
    {
      "epoch": 5.22503008423586,
      "grad_norm": 13.20634651184082,
      "learning_rate": 0.00019894442880062277,
      "loss": 7.0666,
      "step": 43420
    },
    {
      "epoch": 5.226233453670277,
      "grad_norm": 18.00299072265625,
      "learning_rate": 0.000198943877437939,
      "loss": 6.9771,
      "step": 43430
    },
    {
      "epoch": 5.227436823104693,
      "grad_norm": 9.667646408081055,
      "learning_rate": 0.000198943325932059,
      "loss": 6.9375,
      "step": 43440
    },
    {
      "epoch": 5.2286401925391095,
      "grad_norm": 10.863765716552734,
      "learning_rate": 0.0001989427742829836,
      "loss": 6.7448,
      "step": 43450
    },
    {
      "epoch": 5.229843561973526,
      "grad_norm": 13.755781173706055,
      "learning_rate": 0.00019894222249071354,
      "loss": 6.8791,
      "step": 43460
    },
    {
      "epoch": 5.231046931407942,
      "grad_norm": 6.535577774047852,
      "learning_rate": 0.00019894167055524968,
      "loss": 6.8609,
      "step": 43470
    },
    {
      "epoch": 5.232250300842359,
      "grad_norm": 12.226031303405762,
      "learning_rate": 0.00019894111847659277,
      "loss": 6.8005,
      "step": 43480
    },
    {
      "epoch": 5.233453670276775,
      "grad_norm": 9.054750442504883,
      "learning_rate": 0.00019894056625474366,
      "loss": 6.7917,
      "step": 43490
    },
    {
      "epoch": 5.234657039711191,
      "grad_norm": 6.2871904373168945,
      "learning_rate": 0.00019894001388970312,
      "loss": 6.8346,
      "step": 43500
    },
    {
      "epoch": 5.235860409145608,
      "grad_norm": 7.824136734008789,
      "learning_rate": 0.00019893946138147196,
      "loss": 6.75,
      "step": 43510
    },
    {
      "epoch": 5.2370637785800245,
      "grad_norm": 10.756155967712402,
      "learning_rate": 0.00019893890873005096,
      "loss": 6.6822,
      "step": 43520
    },
    {
      "epoch": 5.23826714801444,
      "grad_norm": 17.44888687133789,
      "learning_rate": 0.00019893835593544094,
      "loss": 6.6595,
      "step": 43530
    },
    {
      "epoch": 5.239470517448857,
      "grad_norm": 13.10835075378418,
      "learning_rate": 0.00019893780299764266,
      "loss": 6.8248,
      "step": 43540
    },
    {
      "epoch": 5.240673886883273,
      "grad_norm": 18.595415115356445,
      "learning_rate": 0.000198937249916657,
      "loss": 6.6696,
      "step": 43550
    },
    {
      "epoch": 5.241877256317689,
      "grad_norm": 6.757090091705322,
      "learning_rate": 0.0001989366966924847,
      "loss": 6.7951,
      "step": 43560
    },
    {
      "epoch": 5.243080625752106,
      "grad_norm": 24.652469635009766,
      "learning_rate": 0.00019893614332512654,
      "loss": 6.6635,
      "step": 43570
    },
    {
      "epoch": 5.244283995186522,
      "grad_norm": 8.90617847442627,
      "learning_rate": 0.00019893558981458337,
      "loss": 6.6297,
      "step": 43580
    },
    {
      "epoch": 5.245487364620939,
      "grad_norm": 9.917733192443848,
      "learning_rate": 0.000198935036160856,
      "loss": 6.6716,
      "step": 43590
    },
    {
      "epoch": 5.246690734055355,
      "grad_norm": 18.178878784179688,
      "learning_rate": 0.00019893448236394518,
      "loss": 6.6969,
      "step": 43600
    },
    {
      "epoch": 5.247894103489771,
      "grad_norm": 47.34915542602539,
      "learning_rate": 0.00019893392842385176,
      "loss": 6.6517,
      "step": 43610
    },
    {
      "epoch": 5.249097472924188,
      "grad_norm": 29.6829833984375,
      "learning_rate": 0.00019893337434057648,
      "loss": 6.5945,
      "step": 43620
    },
    {
      "epoch": 5.250300842358604,
      "grad_norm": 15.29988956451416,
      "learning_rate": 0.0001989328201141202,
      "loss": 6.6756,
      "step": 43630
    },
    {
      "epoch": 5.25150421179302,
      "grad_norm": 10.747675895690918,
      "learning_rate": 0.00019893226574448374,
      "loss": 6.8092,
      "step": 43640
    },
    {
      "epoch": 5.252707581227437,
      "grad_norm": 7.114383697509766,
      "learning_rate": 0.00019893171123166784,
      "loss": 6.7703,
      "step": 43650
    },
    {
      "epoch": 5.2539109506618535,
      "grad_norm": 8.199907302856445,
      "learning_rate": 0.00019893115657567328,
      "loss": 6.6793,
      "step": 43660
    },
    {
      "epoch": 5.255114320096269,
      "grad_norm": 7.84580659866333,
      "learning_rate": 0.000198930601776501,
      "loss": 6.754,
      "step": 43670
    },
    {
      "epoch": 5.256317689530686,
      "grad_norm": 10.778278350830078,
      "learning_rate": 0.00019893004683415165,
      "loss": 6.6663,
      "step": 43680
    },
    {
      "epoch": 5.257521058965103,
      "grad_norm": 12.549016952514648,
      "learning_rate": 0.00019892949174862612,
      "loss": 6.6602,
      "step": 43690
    },
    {
      "epoch": 5.2587244283995185,
      "grad_norm": 7.637269496917725,
      "learning_rate": 0.00019892893651992518,
      "loss": 6.7702,
      "step": 43700
    },
    {
      "epoch": 5.259927797833935,
      "grad_norm": 29.3295841217041,
      "learning_rate": 0.00019892838114804966,
      "loss": 6.7947,
      "step": 43710
    },
    {
      "epoch": 5.261131167268351,
      "grad_norm": 53.47911834716797,
      "learning_rate": 0.00019892782563300034,
      "loss": 6.887,
      "step": 43720
    },
    {
      "epoch": 5.262334536702768,
      "grad_norm": 16.587194442749023,
      "learning_rate": 0.00019892726997477805,
      "loss": 7.2881,
      "step": 43730
    },
    {
      "epoch": 5.263537906137184,
      "grad_norm": 18.45558738708496,
      "learning_rate": 0.00019892671417338353,
      "loss": 7.2121,
      "step": 43740
    },
    {
      "epoch": 5.2647412755716,
      "grad_norm": 86.3219223022461,
      "learning_rate": 0.00019892615822881767,
      "loss": 7.0734,
      "step": 43750
    },
    {
      "epoch": 5.265944645006017,
      "grad_norm": 132.38186645507812,
      "learning_rate": 0.00019892560214108122,
      "loss": 7.3361,
      "step": 43760
    },
    {
      "epoch": 5.2671480144404335,
      "grad_norm": 24.015554428100586,
      "learning_rate": 0.00019892504591017502,
      "loss": 6.9493,
      "step": 43770
    },
    {
      "epoch": 5.268351383874849,
      "grad_norm": 44.07176208496094,
      "learning_rate": 0.00019892448953609982,
      "loss": 7.0007,
      "step": 43780
    },
    {
      "epoch": 5.269554753309266,
      "grad_norm": 69.94461822509766,
      "learning_rate": 0.00019892393301885648,
      "loss": 7.1794,
      "step": 43790
    },
    {
      "epoch": 5.270758122743683,
      "grad_norm": 158.1248321533203,
      "learning_rate": 0.0001989233763584458,
      "loss": 7.2272,
      "step": 43800
    },
    {
      "epoch": 5.271961492178098,
      "grad_norm": 26.2042179107666,
      "learning_rate": 0.00019892281955486853,
      "loss": 7.123,
      "step": 43810
    },
    {
      "epoch": 5.273164861612515,
      "grad_norm": 9.133247375488281,
      "learning_rate": 0.00019892226260812557,
      "loss": 7.2685,
      "step": 43820
    },
    {
      "epoch": 5.274368231046932,
      "grad_norm": 21.261714935302734,
      "learning_rate": 0.00019892170551821762,
      "loss": 7.2168,
      "step": 43830
    },
    {
      "epoch": 5.2755716004813475,
      "grad_norm": 135.9453887939453,
      "learning_rate": 0.00019892114828514557,
      "loss": 7.1195,
      "step": 43840
    },
    {
      "epoch": 5.276774969915764,
      "grad_norm": 27.17777442932129,
      "learning_rate": 0.0001989205909089102,
      "loss": 7.2869,
      "step": 43850
    },
    {
      "epoch": 5.277978339350181,
      "grad_norm": 72.9548110961914,
      "learning_rate": 0.0001989200333895123,
      "loss": 6.9865,
      "step": 43860
    },
    {
      "epoch": 5.279181708784597,
      "grad_norm": 47.44607162475586,
      "learning_rate": 0.00019891947572695266,
      "loss": 7.1993,
      "step": 43870
    },
    {
      "epoch": 5.280385078219013,
      "grad_norm": 30.02273941040039,
      "learning_rate": 0.00019891891792123216,
      "loss": 7.2468,
      "step": 43880
    },
    {
      "epoch": 5.28158844765343,
      "grad_norm": 85.73828125,
      "learning_rate": 0.00019891835997235156,
      "loss": 7.0706,
      "step": 43890
    },
    {
      "epoch": 5.282791817087846,
      "grad_norm": 133.78439331054688,
      "learning_rate": 0.00019891780188031164,
      "loss": 7.1245,
      "step": 43900
    },
    {
      "epoch": 5.2839951865222625,
      "grad_norm": 31.00410270690918,
      "learning_rate": 0.00019891724364511326,
      "loss": 7.2328,
      "step": 43910
    },
    {
      "epoch": 5.285198555956678,
      "grad_norm": 26.69734764099121,
      "learning_rate": 0.00019891668526675723,
      "loss": 7.0371,
      "step": 43920
    },
    {
      "epoch": 5.286401925391095,
      "grad_norm": 23.31447410583496,
      "learning_rate": 0.0001989161267452443,
      "loss": 7.2138,
      "step": 43930
    },
    {
      "epoch": 5.287605294825512,
      "grad_norm": 42.201351165771484,
      "learning_rate": 0.0001989155680805753,
      "loss": 7.1619,
      "step": 43940
    },
    {
      "epoch": 5.2888086642599275,
      "grad_norm": 23.718971252441406,
      "learning_rate": 0.0001989150092727511,
      "loss": 7.0938,
      "step": 43950
    },
    {
      "epoch": 5.290012033694344,
      "grad_norm": 43.38616943359375,
      "learning_rate": 0.00019891445032177243,
      "loss": 7.0059,
      "step": 43960
    },
    {
      "epoch": 5.291215403128761,
      "grad_norm": 37.879234313964844,
      "learning_rate": 0.00019891389122764013,
      "loss": 6.9589,
      "step": 43970
    },
    {
      "epoch": 5.292418772563177,
      "grad_norm": 16.111597061157227,
      "learning_rate": 0.00019891333199035502,
      "loss": 6.936,
      "step": 43980
    },
    {
      "epoch": 5.293622141997593,
      "grad_norm": 63.2092170715332,
      "learning_rate": 0.0001989127726099179,
      "loss": 6.9534,
      "step": 43990
    },
    {
      "epoch": 5.29482551143201,
      "grad_norm": 15.879876136779785,
      "learning_rate": 0.00019891221308632954,
      "loss": 6.9656,
      "step": 44000
    },
    {
      "epoch": 5.296028880866426,
      "grad_norm": 28.71221160888672,
      "learning_rate": 0.0001989116534195908,
      "loss": 7.0509,
      "step": 44010
    },
    {
      "epoch": 5.297232250300842,
      "grad_norm": 108.84678649902344,
      "learning_rate": 0.00019891109360970253,
      "loss": 7.1132,
      "step": 44020
    },
    {
      "epoch": 5.298435619735259,
      "grad_norm": 101.40082550048828,
      "learning_rate": 0.00019891053365666547,
      "loss": 7.069,
      "step": 44030
    },
    {
      "epoch": 5.299638989169675,
      "grad_norm": 140.65460205078125,
      "learning_rate": 0.0001989099735604804,
      "loss": 7.01,
      "step": 44040
    },
    {
      "epoch": 5.300842358604092,
      "grad_norm": 76.26039123535156,
      "learning_rate": 0.0001989094133211482,
      "loss": 6.9123,
      "step": 44050
    },
    {
      "epoch": 5.302045728038507,
      "grad_norm": 92.19999694824219,
      "learning_rate": 0.00019890885293866964,
      "loss": 6.8988,
      "step": 44060
    },
    {
      "epoch": 5.303249097472924,
      "grad_norm": 59.36953353881836,
      "learning_rate": 0.0001989082924130456,
      "loss": 7.0809,
      "step": 44070
    },
    {
      "epoch": 5.304452466907341,
      "grad_norm": 205.3117218017578,
      "learning_rate": 0.0001989077317442768,
      "loss": 7.0318,
      "step": 44080
    },
    {
      "epoch": 5.3056558363417565,
      "grad_norm": 60.94562530517578,
      "learning_rate": 0.00019890717093236414,
      "loss": 6.9655,
      "step": 44090
    },
    {
      "epoch": 5.306859205776173,
      "grad_norm": 82.89745330810547,
      "learning_rate": 0.00019890660997730835,
      "loss": 7.3218,
      "step": 44100
    },
    {
      "epoch": 5.30806257521059,
      "grad_norm": 111.86553192138672,
      "learning_rate": 0.00019890604887911026,
      "loss": 7.1645,
      "step": 44110
    },
    {
      "epoch": 5.309265944645006,
      "grad_norm": 97.93233489990234,
      "learning_rate": 0.00019890548763777072,
      "loss": 7.0804,
      "step": 44120
    },
    {
      "epoch": 5.310469314079422,
      "grad_norm": 27.88705825805664,
      "learning_rate": 0.0001989049262532905,
      "loss": 6.9342,
      "step": 44130
    },
    {
      "epoch": 5.311672683513839,
      "grad_norm": 21.621044158935547,
      "learning_rate": 0.00019890436472567047,
      "loss": 7.0262,
      "step": 44140
    },
    {
      "epoch": 5.312876052948255,
      "grad_norm": 32.33016586303711,
      "learning_rate": 0.0001989038030549114,
      "loss": 6.9299,
      "step": 44150
    },
    {
      "epoch": 5.3140794223826715,
      "grad_norm": 70.36555480957031,
      "learning_rate": 0.00019890324124101407,
      "loss": 7.001,
      "step": 44160
    },
    {
      "epoch": 5.315282791817088,
      "grad_norm": 42.8079719543457,
      "learning_rate": 0.00019890267928397934,
      "loss": 6.8388,
      "step": 44170
    },
    {
      "epoch": 5.316486161251504,
      "grad_norm": 33.48402404785156,
      "learning_rate": 0.00019890211718380805,
      "loss": 6.9514,
      "step": 44180
    },
    {
      "epoch": 5.317689530685921,
      "grad_norm": 14.187588691711426,
      "learning_rate": 0.00019890155494050097,
      "loss": 6.885,
      "step": 44190
    },
    {
      "epoch": 5.318892900120337,
      "grad_norm": 13.851027488708496,
      "learning_rate": 0.0001989009925540589,
      "loss": 7.1243,
      "step": 44200
    },
    {
      "epoch": 5.320096269554753,
      "grad_norm": 5.626720905303955,
      "learning_rate": 0.0001989004300244827,
      "loss": 7.2875,
      "step": 44210
    },
    {
      "epoch": 5.32129963898917,
      "grad_norm": 15.217040061950684,
      "learning_rate": 0.00019889986735177313,
      "loss": 7.4021,
      "step": 44220
    },
    {
      "epoch": 5.3225030084235865,
      "grad_norm": 33.85987091064453,
      "learning_rate": 0.00019889930453593105,
      "loss": 7.3542,
      "step": 44230
    },
    {
      "epoch": 5.323706377858002,
      "grad_norm": 56.41918182373047,
      "learning_rate": 0.00019889874157695724,
      "loss": 7.4216,
      "step": 44240
    },
    {
      "epoch": 5.324909747292419,
      "grad_norm": 37.51495361328125,
      "learning_rate": 0.00019889817847485258,
      "loss": 7.5273,
      "step": 44250
    },
    {
      "epoch": 5.326113116726835,
      "grad_norm": 56.5140266418457,
      "learning_rate": 0.0001988976152296178,
      "loss": 7.4221,
      "step": 44260
    },
    {
      "epoch": 5.327316486161251,
      "grad_norm": 90.66345977783203,
      "learning_rate": 0.00019889705184125375,
      "loss": 7.3977,
      "step": 44270
    },
    {
      "epoch": 5.328519855595668,
      "grad_norm": 12.699202537536621,
      "learning_rate": 0.00019889648830976125,
      "loss": 7.3602,
      "step": 44280
    },
    {
      "epoch": 5.329723225030084,
      "grad_norm": 7.04261589050293,
      "learning_rate": 0.00019889592463514117,
      "loss": 7.3357,
      "step": 44290
    },
    {
      "epoch": 5.330926594464501,
      "grad_norm": 26.140731811523438,
      "learning_rate": 0.0001988953608173942,
      "loss": 7.358,
      "step": 44300
    },
    {
      "epoch": 5.332129963898917,
      "grad_norm": 16.69830322265625,
      "learning_rate": 0.00019889479685652125,
      "loss": 7.4231,
      "step": 44310
    },
    {
      "epoch": 5.333333333333333,
      "grad_norm": 13.26341438293457,
      "learning_rate": 0.00019889423275252314,
      "loss": 7.1647,
      "step": 44320
    },
    {
      "epoch": 5.33453670276775,
      "grad_norm": 78.64390563964844,
      "learning_rate": 0.00019889366850540062,
      "loss": 7.2275,
      "step": 44330
    },
    {
      "epoch": 5.335740072202166,
      "grad_norm": 71.00178527832031,
      "learning_rate": 0.00019889310411515458,
      "loss": 7.1559,
      "step": 44340
    },
    {
      "epoch": 5.336943441636582,
      "grad_norm": 14.184823989868164,
      "learning_rate": 0.00019889253958178577,
      "loss": 7.0014,
      "step": 44350
    },
    {
      "epoch": 5.338146811070999,
      "grad_norm": 12.006002426147461,
      "learning_rate": 0.0001988919749052951,
      "loss": 7.0046,
      "step": 44360
    },
    {
      "epoch": 5.3393501805054155,
      "grad_norm": 18.308382034301758,
      "learning_rate": 0.00019889141008568327,
      "loss": 6.9466,
      "step": 44370
    },
    {
      "epoch": 5.340553549939831,
      "grad_norm": 11.394606590270996,
      "learning_rate": 0.00019889084512295116,
      "loss": 6.8929,
      "step": 44380
    },
    {
      "epoch": 5.341756919374248,
      "grad_norm": 25.32445526123047,
      "learning_rate": 0.00019889028001709958,
      "loss": 6.8786,
      "step": 44390
    },
    {
      "epoch": 5.342960288808664,
      "grad_norm": 35.181915283203125,
      "learning_rate": 0.00019888971476812937,
      "loss": 7.1647,
      "step": 44400
    },
    {
      "epoch": 5.3441636582430805,
      "grad_norm": 30.413166046142578,
      "learning_rate": 0.0001988891493760413,
      "loss": 7.2058,
      "step": 44410
    },
    {
      "epoch": 5.345367027677497,
      "grad_norm": 49.28425979614258,
      "learning_rate": 0.00019888858384083623,
      "loss": 7.1434,
      "step": 44420
    },
    {
      "epoch": 5.346570397111913,
      "grad_norm": 13.06767749786377,
      "learning_rate": 0.00019888801816251496,
      "loss": 7.0828,
      "step": 44430
    },
    {
      "epoch": 5.34777376654633,
      "grad_norm": 17.158748626708984,
      "learning_rate": 0.00019888745234107833,
      "loss": 7.2581,
      "step": 44440
    },
    {
      "epoch": 5.348977135980746,
      "grad_norm": 18.183988571166992,
      "learning_rate": 0.00019888688637652711,
      "loss": 7.2694,
      "step": 44450
    },
    {
      "epoch": 5.350180505415162,
      "grad_norm": 36.71931838989258,
      "learning_rate": 0.0001988863202688622,
      "loss": 7.136,
      "step": 44460
    },
    {
      "epoch": 5.351383874849579,
      "grad_norm": 8.497072219848633,
      "learning_rate": 0.00019888575401808432,
      "loss": 7.0786,
      "step": 44470
    },
    {
      "epoch": 5.3525872442839955,
      "grad_norm": 11.919191360473633,
      "learning_rate": 0.00019888518762419436,
      "loss": 6.9378,
      "step": 44480
    },
    {
      "epoch": 5.353790613718411,
      "grad_norm": 9.162808418273926,
      "learning_rate": 0.00019888462108719313,
      "loss": 7.0053,
      "step": 44490
    },
    {
      "epoch": 5.354993983152828,
      "grad_norm": 15.718364715576172,
      "learning_rate": 0.0001988840544070814,
      "loss": 6.8903,
      "step": 44500
    },
    {
      "epoch": 5.356197352587245,
      "grad_norm": 14.616697311401367,
      "learning_rate": 0.00019888348758386007,
      "loss": 6.9972,
      "step": 44510
    },
    {
      "epoch": 5.35740072202166,
      "grad_norm": 15.98686408996582,
      "learning_rate": 0.0001988829206175299,
      "loss": 6.9958,
      "step": 44520
    },
    {
      "epoch": 5.358604091456077,
      "grad_norm": 31.18436622619629,
      "learning_rate": 0.00019888235350809177,
      "loss": 6.8708,
      "step": 44530
    },
    {
      "epoch": 5.359807460890494,
      "grad_norm": 5.7120361328125,
      "learning_rate": 0.00019888178625554644,
      "loss": 6.8127,
      "step": 44540
    },
    {
      "epoch": 5.3610108303249095,
      "grad_norm": 11.021726608276367,
      "learning_rate": 0.00019888121885989473,
      "loss": 6.8173,
      "step": 44550
    },
    {
      "epoch": 5.362214199759326,
      "grad_norm": 20.15838623046875,
      "learning_rate": 0.0001988806513211375,
      "loss": 6.737,
      "step": 44560
    },
    {
      "epoch": 5.363417569193743,
      "grad_norm": 9.004701614379883,
      "learning_rate": 0.00019888008363927554,
      "loss": 6.9287,
      "step": 44570
    },
    {
      "epoch": 5.364620938628159,
      "grad_norm": 13.968864440917969,
      "learning_rate": 0.0001988795158143097,
      "loss": 6.815,
      "step": 44580
    },
    {
      "epoch": 5.365824308062575,
      "grad_norm": 4.209850788116455,
      "learning_rate": 0.0001988789478462408,
      "loss": 6.713,
      "step": 44590
    },
    {
      "epoch": 5.367027677496991,
      "grad_norm": 8.388750076293945,
      "learning_rate": 0.00019887837973506963,
      "loss": 6.7283,
      "step": 44600
    },
    {
      "epoch": 5.368231046931408,
      "grad_norm": 10.483491897583008,
      "learning_rate": 0.00019887781148079703,
      "loss": 6.7655,
      "step": 44610
    },
    {
      "epoch": 5.3694344163658245,
      "grad_norm": 7.144164562225342,
      "learning_rate": 0.00019887724308342383,
      "loss": 6.7459,
      "step": 44620
    },
    {
      "epoch": 5.37063778580024,
      "grad_norm": 12.199914932250977,
      "learning_rate": 0.00019887667454295086,
      "loss": 6.6878,
      "step": 44630
    },
    {
      "epoch": 5.371841155234657,
      "grad_norm": 5.8632426261901855,
      "learning_rate": 0.00019887610585937892,
      "loss": 6.617,
      "step": 44640
    },
    {
      "epoch": 5.373044524669074,
      "grad_norm": 12.842899322509766,
      "learning_rate": 0.00019887553703270885,
      "loss": 6.7336,
      "step": 44650
    },
    {
      "epoch": 5.3742478941034895,
      "grad_norm": 9.459039688110352,
      "learning_rate": 0.00019887496806294145,
      "loss": 6.7469,
      "step": 44660
    },
    {
      "epoch": 5.375451263537906,
      "grad_norm": 26.74758529663086,
      "learning_rate": 0.00019887439895007754,
      "loss": 6.7122,
      "step": 44670
    },
    {
      "epoch": 5.376654632972323,
      "grad_norm": 7.502065181732178,
      "learning_rate": 0.00019887382969411803,
      "loss": 6.6749,
      "step": 44680
    },
    {
      "epoch": 5.377858002406739,
      "grad_norm": 4.629199504852295,
      "learning_rate": 0.00019887326029506363,
      "loss": 6.7005,
      "step": 44690
    },
    {
      "epoch": 5.379061371841155,
      "grad_norm": 5.4511284828186035,
      "learning_rate": 0.0001988726907529152,
      "loss": 6.6812,
      "step": 44700
    },
    {
      "epoch": 5.380264741275572,
      "grad_norm": 6.781862735748291,
      "learning_rate": 0.0001988721210676736,
      "loss": 6.6241,
      "step": 44710
    },
    {
      "epoch": 5.381468110709988,
      "grad_norm": 15.501797676086426,
      "learning_rate": 0.00019887155123933964,
      "loss": 6.594,
      "step": 44720
    },
    {
      "epoch": 5.382671480144404,
      "grad_norm": 11.129393577575684,
      "learning_rate": 0.00019887098126791412,
      "loss": 6.6791,
      "step": 44730
    },
    {
      "epoch": 5.38387484957882,
      "grad_norm": 15.134982109069824,
      "learning_rate": 0.0001988704111533979,
      "loss": 6.7703,
      "step": 44740
    },
    {
      "epoch": 5.385078219013237,
      "grad_norm": 23.988197326660156,
      "learning_rate": 0.00019886984089579178,
      "loss": 6.6215,
      "step": 44750
    },
    {
      "epoch": 5.386281588447654,
      "grad_norm": 5.169701099395752,
      "learning_rate": 0.00019886927049509657,
      "loss": 6.6566,
      "step": 44760
    },
    {
      "epoch": 5.387484957882069,
      "grad_norm": 75.0279541015625,
      "learning_rate": 0.0001988686999513131,
      "loss": 6.6738,
      "step": 44770
    },
    {
      "epoch": 5.388688327316486,
      "grad_norm": 9.8955717086792,
      "learning_rate": 0.00019886812926444222,
      "loss": 6.5693,
      "step": 44780
    },
    {
      "epoch": 5.389891696750903,
      "grad_norm": 9.198053359985352,
      "learning_rate": 0.0001988675584344848,
      "loss": 6.6713,
      "step": 44790
    },
    {
      "epoch": 5.3910950661853185,
      "grad_norm": 67.88687896728516,
      "learning_rate": 0.00019886698746144154,
      "loss": 6.6615,
      "step": 44800
    },
    {
      "epoch": 5.392298435619735,
      "grad_norm": 85.84774780273438,
      "learning_rate": 0.00019886641634531339,
      "loss": 6.8457,
      "step": 44810
    },
    {
      "epoch": 5.393501805054152,
      "grad_norm": 8.06728744506836,
      "learning_rate": 0.0001988658450861011,
      "loss": 6.7166,
      "step": 44820
    },
    {
      "epoch": 5.394705174488568,
      "grad_norm": 10.970793724060059,
      "learning_rate": 0.00019886527368380555,
      "loss": 6.7579,
      "step": 44830
    },
    {
      "epoch": 5.395908543922984,
      "grad_norm": 8.422438621520996,
      "learning_rate": 0.00019886470213842752,
      "loss": 6.5816,
      "step": 44840
    },
    {
      "epoch": 5.397111913357401,
      "grad_norm": 6.012128829956055,
      "learning_rate": 0.00019886413044996784,
      "loss": 6.6582,
      "step": 44850
    },
    {
      "epoch": 5.398315282791817,
      "grad_norm": 7.007898330688477,
      "learning_rate": 0.0001988635586184274,
      "loss": 6.6602,
      "step": 44860
    },
    {
      "epoch": 5.3995186522262335,
      "grad_norm": 19.37340545654297,
      "learning_rate": 0.00019886298664380695,
      "loss": 6.7178,
      "step": 44870
    },
    {
      "epoch": 5.40072202166065,
      "grad_norm": 12.28745174407959,
      "learning_rate": 0.00019886241452610735,
      "loss": 6.6701,
      "step": 44880
    },
    {
      "epoch": 5.401925391095066,
      "grad_norm": 23.0762996673584,
      "learning_rate": 0.00019886184226532941,
      "loss": 6.5621,
      "step": 44890
    },
    {
      "epoch": 5.403128760529483,
      "grad_norm": 89.00927734375,
      "learning_rate": 0.000198861269861474,
      "loss": 6.7886,
      "step": 44900
    },
    {
      "epoch": 5.404332129963899,
      "grad_norm": 42.586124420166016,
      "learning_rate": 0.00019886069731454194,
      "loss": 6.7432,
      "step": 44910
    },
    {
      "epoch": 5.405535499398315,
      "grad_norm": 10.239490509033203,
      "learning_rate": 0.00019886012462453403,
      "loss": 7.2723,
      "step": 44920
    },
    {
      "epoch": 5.406738868832732,
      "grad_norm": 9.26880168914795,
      "learning_rate": 0.00019885955179145111,
      "loss": 7.4963,
      "step": 44930
    },
    {
      "epoch": 5.4079422382671485,
      "grad_norm": 19.099369049072266,
      "learning_rate": 0.000198858978815294,
      "loss": 7.0627,
      "step": 44940
    },
    {
      "epoch": 5.409145607701564,
      "grad_norm": 15.05772876739502,
      "learning_rate": 0.00019885840569606355,
      "loss": 6.9925,
      "step": 44950
    },
    {
      "epoch": 5.410348977135981,
      "grad_norm": 13.30691909790039,
      "learning_rate": 0.0001988578324337606,
      "loss": 6.7894,
      "step": 44960
    },
    {
      "epoch": 5.411552346570397,
      "grad_norm": 34.78816604614258,
      "learning_rate": 0.0001988572590283859,
      "loss": 6.6909,
      "step": 44970
    },
    {
      "epoch": 5.412755716004813,
      "grad_norm": 14.33220100402832,
      "learning_rate": 0.0001988566854799404,
      "loss": 6.8136,
      "step": 44980
    },
    {
      "epoch": 5.41395908543923,
      "grad_norm": 11.890724182128906,
      "learning_rate": 0.00019885611178842484,
      "loss": 6.785,
      "step": 44990
    },
    {
      "epoch": 5.415162454873646,
      "grad_norm": 7.02847146987915,
      "learning_rate": 0.00019885553795384012,
      "loss": 6.6332,
      "step": 45000
    },
    {
      "epoch": 5.416365824308063,
      "grad_norm": 7.15176248550415,
      "learning_rate": 0.00019885496397618698,
      "loss": 6.6381,
      "step": 45010
    },
    {
      "epoch": 5.417569193742479,
      "grad_norm": 6.993597984313965,
      "learning_rate": 0.00019885438985546631,
      "loss": 6.7677,
      "step": 45020
    },
    {
      "epoch": 5.418772563176895,
      "grad_norm": 5.475958347320557,
      "learning_rate": 0.00019885381559167894,
      "loss": 6.6059,
      "step": 45030
    },
    {
      "epoch": 5.419975932611312,
      "grad_norm": 7.7009596824646,
      "learning_rate": 0.0001988532411848257,
      "loss": 6.5585,
      "step": 45040
    },
    {
      "epoch": 5.421179302045728,
      "grad_norm": 13.65756893157959,
      "learning_rate": 0.0001988526666349074,
      "loss": 6.642,
      "step": 45050
    },
    {
      "epoch": 5.422382671480144,
      "grad_norm": 43.964698791503906,
      "learning_rate": 0.0001988520919419249,
      "loss": 6.5914,
      "step": 45060
    },
    {
      "epoch": 5.423586040914561,
      "grad_norm": 16.52776336669922,
      "learning_rate": 0.000198851517105879,
      "loss": 6.6282,
      "step": 45070
    },
    {
      "epoch": 5.4247894103489775,
      "grad_norm": 7.894533157348633,
      "learning_rate": 0.00019885094212677057,
      "loss": 6.5094,
      "step": 45080
    },
    {
      "epoch": 5.425992779783393,
      "grad_norm": 3.9156060218811035,
      "learning_rate": 0.00019885036700460038,
      "loss": 6.5818,
      "step": 45090
    },
    {
      "epoch": 5.42719614921781,
      "grad_norm": 13.460526466369629,
      "learning_rate": 0.00019884979173936934,
      "loss": 6.5725,
      "step": 45100
    },
    {
      "epoch": 5.428399518652226,
      "grad_norm": 7.966900825500488,
      "learning_rate": 0.00019884921633107823,
      "loss": 6.6553,
      "step": 45110
    },
    {
      "epoch": 5.4296028880866425,
      "grad_norm": 9.780647277832031,
      "learning_rate": 0.00019884864077972792,
      "loss": 6.6097,
      "step": 45120
    },
    {
      "epoch": 5.430806257521059,
      "grad_norm": 6.164175510406494,
      "learning_rate": 0.00019884806508531922,
      "loss": 6.4171,
      "step": 45130
    },
    {
      "epoch": 5.432009626955475,
      "grad_norm": 6.212482929229736,
      "learning_rate": 0.00019884748924785294,
      "loss": 6.5031,
      "step": 45140
    },
    {
      "epoch": 5.433212996389892,
      "grad_norm": 4.868154525756836,
      "learning_rate": 0.00019884691326732995,
      "loss": 6.5662,
      "step": 45150
    },
    {
      "epoch": 5.434416365824308,
      "grad_norm": 12.111865043640137,
      "learning_rate": 0.0001988463371437511,
      "loss": 6.4908,
      "step": 45160
    },
    {
      "epoch": 5.435619735258724,
      "grad_norm": 3.300665855407715,
      "learning_rate": 0.00019884576087711716,
      "loss": 6.3347,
      "step": 45170
    },
    {
      "epoch": 5.436823104693141,
      "grad_norm": 24.62493896484375,
      "learning_rate": 0.000198845184467429,
      "loss": 6.4454,
      "step": 45180
    },
    {
      "epoch": 5.4380264741275575,
      "grad_norm": 3.6107540130615234,
      "learning_rate": 0.00019884460791468747,
      "loss": 6.5296,
      "step": 45190
    },
    {
      "epoch": 5.439229843561973,
      "grad_norm": 5.2915940284729,
      "learning_rate": 0.00019884403121889338,
      "loss": 6.3876,
      "step": 45200
    },
    {
      "epoch": 5.44043321299639,
      "grad_norm": 14.972261428833008,
      "learning_rate": 0.00019884345438004756,
      "loss": 6.5571,
      "step": 45210
    },
    {
      "epoch": 5.441636582430807,
      "grad_norm": 20.201309204101562,
      "learning_rate": 0.0001988428773981509,
      "loss": 6.5529,
      "step": 45220
    },
    {
      "epoch": 5.442839951865222,
      "grad_norm": 6.401875972747803,
      "learning_rate": 0.00019884230027320413,
      "loss": 6.5248,
      "step": 45230
    },
    {
      "epoch": 5.444043321299639,
      "grad_norm": 14.680062294006348,
      "learning_rate": 0.0001988417230052082,
      "loss": 6.4735,
      "step": 45240
    },
    {
      "epoch": 5.445246690734056,
      "grad_norm": 13.246824264526367,
      "learning_rate": 0.0001988411455941639,
      "loss": 6.4265,
      "step": 45250
    },
    {
      "epoch": 5.4464500601684716,
      "grad_norm": 17.640949249267578,
      "learning_rate": 0.000198840568040072,
      "loss": 6.4548,
      "step": 45260
    },
    {
      "epoch": 5.447653429602888,
      "grad_norm": 13.611530303955078,
      "learning_rate": 0.00019883999034293344,
      "loss": 6.4113,
      "step": 45270
    },
    {
      "epoch": 5.448856799037305,
      "grad_norm": 4.311097621917725,
      "learning_rate": 0.00019883941250274903,
      "loss": 6.5528,
      "step": 45280
    },
    {
      "epoch": 5.450060168471721,
      "grad_norm": 5.291025161743164,
      "learning_rate": 0.00019883883451951955,
      "loss": 6.4416,
      "step": 45290
    },
    {
      "epoch": 5.451263537906137,
      "grad_norm": 5.119104385375977,
      "learning_rate": 0.00019883825639324586,
      "loss": 6.4439,
      "step": 45300
    },
    {
      "epoch": 5.452466907340553,
      "grad_norm": 4.42533016204834,
      "learning_rate": 0.00019883767812392886,
      "loss": 6.497,
      "step": 45310
    },
    {
      "epoch": 5.45367027677497,
      "grad_norm": 5.035144805908203,
      "learning_rate": 0.0001988370997115693,
      "loss": 6.386,
      "step": 45320
    },
    {
      "epoch": 5.4548736462093865,
      "grad_norm": 7.269014835357666,
      "learning_rate": 0.00019883652115616806,
      "loss": 6.4401,
      "step": 45330
    },
    {
      "epoch": 5.456077015643802,
      "grad_norm": 7.293306350708008,
      "learning_rate": 0.000198835942457726,
      "loss": 6.5438,
      "step": 45340
    },
    {
      "epoch": 5.457280385078219,
      "grad_norm": 11.02853012084961,
      "learning_rate": 0.0001988353636162439,
      "loss": 6.4328,
      "step": 45350
    },
    {
      "epoch": 5.458483754512636,
      "grad_norm": 10.555423736572266,
      "learning_rate": 0.00019883478463172263,
      "loss": 6.4596,
      "step": 45360
    },
    {
      "epoch": 5.4596871239470515,
      "grad_norm": 14.491849899291992,
      "learning_rate": 0.00019883420550416303,
      "loss": 6.423,
      "step": 45370
    },
    {
      "epoch": 5.460890493381468,
      "grad_norm": 10.816365242004395,
      "learning_rate": 0.00019883362623356594,
      "loss": 6.4074,
      "step": 45380
    },
    {
      "epoch": 5.462093862815885,
      "grad_norm": 3.7623682022094727,
      "learning_rate": 0.0001988330468199322,
      "loss": 6.5103,
      "step": 45390
    },
    {
      "epoch": 5.463297232250301,
      "grad_norm": 4.828091144561768,
      "learning_rate": 0.0001988324672632626,
      "loss": 6.4944,
      "step": 45400
    },
    {
      "epoch": 5.464500601684717,
      "grad_norm": 7.3919525146484375,
      "learning_rate": 0.00019883188756355806,
      "loss": 6.5367,
      "step": 45410
    },
    {
      "epoch": 5.465703971119134,
      "grad_norm": 12.38614559173584,
      "learning_rate": 0.00019883130772081937,
      "loss": 6.4677,
      "step": 45420
    },
    {
      "epoch": 5.46690734055355,
      "grad_norm": 22.366512298583984,
      "learning_rate": 0.00019883072773504737,
      "loss": 6.485,
      "step": 45430
    },
    {
      "epoch": 5.4681107099879664,
      "grad_norm": 5.2813920974731445,
      "learning_rate": 0.0001988301476062429,
      "loss": 6.4001,
      "step": 45440
    },
    {
      "epoch": 5.469314079422382,
      "grad_norm": 6.833184242248535,
      "learning_rate": 0.0001988295673344068,
      "loss": 6.4086,
      "step": 45450
    },
    {
      "epoch": 5.470517448856799,
      "grad_norm": 10.400935173034668,
      "learning_rate": 0.00019882898691953992,
      "loss": 6.4306,
      "step": 45460
    },
    {
      "epoch": 5.471720818291216,
      "grad_norm": 9.844170570373535,
      "learning_rate": 0.00019882840636164313,
      "loss": 6.4404,
      "step": 45470
    },
    {
      "epoch": 5.472924187725631,
      "grad_norm": 4.941709041595459,
      "learning_rate": 0.0001988278256607172,
      "loss": 6.4654,
      "step": 45480
    },
    {
      "epoch": 5.474127557160048,
      "grad_norm": 10.349566459655762,
      "learning_rate": 0.00019882724481676303,
      "loss": 6.4547,
      "step": 45490
    },
    {
      "epoch": 5.475330926594465,
      "grad_norm": 31.910966873168945,
      "learning_rate": 0.0001988266638297814,
      "loss": 6.5674,
      "step": 45500
    },
    {
      "epoch": 5.4765342960288805,
      "grad_norm": 17.867481231689453,
      "learning_rate": 0.00019882608269977323,
      "loss": 6.9163,
      "step": 45510
    },
    {
      "epoch": 5.477737665463297,
      "grad_norm": 6.678061008453369,
      "learning_rate": 0.0001988255014267393,
      "loss": 6.7346,
      "step": 45520
    },
    {
      "epoch": 5.478941034897714,
      "grad_norm": 23.98747444152832,
      "learning_rate": 0.00019882492001068046,
      "loss": 6.627,
      "step": 45530
    },
    {
      "epoch": 5.48014440433213,
      "grad_norm": 7.337192535400391,
      "learning_rate": 0.0001988243384515976,
      "loss": 6.5027,
      "step": 45540
    },
    {
      "epoch": 5.481347773766546,
      "grad_norm": 8.542789459228516,
      "learning_rate": 0.00019882375674949148,
      "loss": 6.427,
      "step": 45550
    },
    {
      "epoch": 5.482551143200963,
      "grad_norm": 7.10697603225708,
      "learning_rate": 0.000198823174904363,
      "loss": 6.4406,
      "step": 45560
    },
    {
      "epoch": 5.483754512635379,
      "grad_norm": 6.624485492706299,
      "learning_rate": 0.00019882259291621298,
      "loss": 6.3173,
      "step": 45570
    },
    {
      "epoch": 5.4849578820697955,
      "grad_norm": 74.85208892822266,
      "learning_rate": 0.0001988220107850423,
      "loss": 6.3787,
      "step": 45580
    },
    {
      "epoch": 5.486161251504212,
      "grad_norm": 15.59478759765625,
      "learning_rate": 0.00019882142851085176,
      "loss": 6.558,
      "step": 45590
    },
    {
      "epoch": 5.487364620938628,
      "grad_norm": 20.38960838317871,
      "learning_rate": 0.00019882084609364217,
      "loss": 6.5889,
      "step": 45600
    },
    {
      "epoch": 5.488567990373045,
      "grad_norm": 8.961256980895996,
      "learning_rate": 0.00019882026353341446,
      "loss": 6.5569,
      "step": 45610
    },
    {
      "epoch": 5.489771359807461,
      "grad_norm": 9.782734870910645,
      "learning_rate": 0.00019881968083016943,
      "loss": 6.48,
      "step": 45620
    },
    {
      "epoch": 5.490974729241877,
      "grad_norm": 16.865764617919922,
      "learning_rate": 0.00019881909798390792,
      "loss": 6.3723,
      "step": 45630
    },
    {
      "epoch": 5.492178098676294,
      "grad_norm": 13.1566162109375,
      "learning_rate": 0.00019881851499463075,
      "loss": 6.5305,
      "step": 45640
    },
    {
      "epoch": 5.49338146811071,
      "grad_norm": 6.949885845184326,
      "learning_rate": 0.00019881793186233884,
      "loss": 6.4292,
      "step": 45650
    },
    {
      "epoch": 5.494584837545126,
      "grad_norm": 6.953880310058594,
      "learning_rate": 0.00019881734858703297,
      "loss": 6.3776,
      "step": 45660
    },
    {
      "epoch": 5.495788206979543,
      "grad_norm": 13.188807487487793,
      "learning_rate": 0.00019881676516871397,
      "loss": 6.4843,
      "step": 45670
    },
    {
      "epoch": 5.496991576413959,
      "grad_norm": 8.368364334106445,
      "learning_rate": 0.00019881618160738277,
      "loss": 6.4333,
      "step": 45680
    },
    {
      "epoch": 5.498194945848375,
      "grad_norm": 8.661624908447266,
      "learning_rate": 0.0001988155979030401,
      "loss": 6.4573,
      "step": 45690
    },
    {
      "epoch": 5.499398315282792,
      "grad_norm": 7.819737434387207,
      "learning_rate": 0.00019881501405568687,
      "loss": 6.4315,
      "step": 45700
    },
    {
      "epoch": 5.500601684717208,
      "grad_norm": 17.464813232421875,
      "learning_rate": 0.00019881443006532394,
      "loss": 6.4108,
      "step": 45710
    },
    {
      "epoch": 5.501805054151625,
      "grad_norm": 5.219578742980957,
      "learning_rate": 0.00019881384593195214,
      "loss": 6.2546,
      "step": 45720
    },
    {
      "epoch": 5.503008423586041,
      "grad_norm": 17.478546142578125,
      "learning_rate": 0.00019881326165557229,
      "loss": 6.3163,
      "step": 45730
    },
    {
      "epoch": 5.504211793020457,
      "grad_norm": 8.404958724975586,
      "learning_rate": 0.00019881267723618527,
      "loss": 6.3914,
      "step": 45740
    },
    {
      "epoch": 5.505415162454874,
      "grad_norm": 13.10112476348877,
      "learning_rate": 0.00019881209267379188,
      "loss": 6.2438,
      "step": 45750
    },
    {
      "epoch": 5.5066185318892895,
      "grad_norm": 7.838147163391113,
      "learning_rate": 0.00019881150796839303,
      "loss": 6.3691,
      "step": 45760
    },
    {
      "epoch": 5.507821901323706,
      "grad_norm": 18.114526748657227,
      "learning_rate": 0.0001988109231199895,
      "loss": 6.3613,
      "step": 45770
    },
    {
      "epoch": 5.509025270758123,
      "grad_norm": 34.72455978393555,
      "learning_rate": 0.00019881033812858218,
      "loss": 6.4407,
      "step": 45780
    },
    {
      "epoch": 5.510228640192539,
      "grad_norm": 15.340691566467285,
      "learning_rate": 0.00019880975299417194,
      "loss": 6.4729,
      "step": 45790
    },
    {
      "epoch": 5.511432009626955,
      "grad_norm": 10.677668571472168,
      "learning_rate": 0.00019880916771675955,
      "loss": 6.5155,
      "step": 45800
    },
    {
      "epoch": 5.512635379061372,
      "grad_norm": 12.764081954956055,
      "learning_rate": 0.0001988085822963459,
      "loss": 6.4162,
      "step": 45810
    },
    {
      "epoch": 5.513838748495788,
      "grad_norm": 7.726113319396973,
      "learning_rate": 0.00019880799673293185,
      "loss": 6.394,
      "step": 45820
    },
    {
      "epoch": 5.5150421179302045,
      "grad_norm": 9.001998901367188,
      "learning_rate": 0.00019880741102651825,
      "loss": 6.2998,
      "step": 45830
    },
    {
      "epoch": 5.516245487364621,
      "grad_norm": 7.153882026672363,
      "learning_rate": 0.0001988068251771059,
      "loss": 6.35,
      "step": 45840
    },
    {
      "epoch": 5.517448856799037,
      "grad_norm": 21.886831283569336,
      "learning_rate": 0.0001988062391846957,
      "loss": 6.3185,
      "step": 45850
    },
    {
      "epoch": 5.518652226233454,
      "grad_norm": 15.019268989562988,
      "learning_rate": 0.00019880565304928845,
      "loss": 6.8131,
      "step": 45860
    },
    {
      "epoch": 5.51985559566787,
      "grad_norm": 8.751230239868164,
      "learning_rate": 0.00019880506677088508,
      "loss": 6.5887,
      "step": 45870
    },
    {
      "epoch": 5.521058965102286,
      "grad_norm": 11.072755813598633,
      "learning_rate": 0.00019880448034948636,
      "loss": 6.432,
      "step": 45880
    },
    {
      "epoch": 5.522262334536703,
      "grad_norm": 28.181812286376953,
      "learning_rate": 0.0001988038937850931,
      "loss": 6.5608,
      "step": 45890
    },
    {
      "epoch": 5.5234657039711195,
      "grad_norm": 14.7490816116333,
      "learning_rate": 0.0001988033070777063,
      "loss": 6.3919,
      "step": 45900
    },
    {
      "epoch": 5.524669073405535,
      "grad_norm": 6.526397228240967,
      "learning_rate": 0.00019880272022732668,
      "loss": 6.4002,
      "step": 45910
    },
    {
      "epoch": 5.525872442839952,
      "grad_norm": 9.350018501281738,
      "learning_rate": 0.00019880213323395513,
      "loss": 6.4645,
      "step": 45920
    },
    {
      "epoch": 5.527075812274369,
      "grad_norm": 6.043956756591797,
      "learning_rate": 0.00019880154609759252,
      "loss": 6.3516,
      "step": 45930
    },
    {
      "epoch": 5.528279181708784,
      "grad_norm": 7.510612487792969,
      "learning_rate": 0.00019880095881823966,
      "loss": 6.4044,
      "step": 45940
    },
    {
      "epoch": 5.529482551143201,
      "grad_norm": 11.649624824523926,
      "learning_rate": 0.00019880037139589741,
      "loss": 6.3728,
      "step": 45950
    },
    {
      "epoch": 5.530685920577618,
      "grad_norm": 4.744845390319824,
      "learning_rate": 0.0001987997838305667,
      "loss": 6.2978,
      "step": 45960
    },
    {
      "epoch": 5.5318892900120336,
      "grad_norm": 13.25458812713623,
      "learning_rate": 0.00019879919612224823,
      "loss": 6.3112,
      "step": 45970
    },
    {
      "epoch": 5.53309265944645,
      "grad_norm": 3.9331250190734863,
      "learning_rate": 0.00019879860827094295,
      "loss": 6.271,
      "step": 45980
    },
    {
      "epoch": 5.534296028880867,
      "grad_norm": 6.454897403717041,
      "learning_rate": 0.0001987980202766517,
      "loss": 6.2212,
      "step": 45990
    },
    {
      "epoch": 5.535499398315283,
      "grad_norm": 7.229743957519531,
      "learning_rate": 0.00019879743213937532,
      "loss": 6.3656,
      "step": 46000
    },
    {
      "epoch": 5.536702767749699,
      "grad_norm": 11.61962890625,
      "learning_rate": 0.00019879684385911466,
      "loss": 6.3485,
      "step": 46010
    },
    {
      "epoch": 5.537906137184115,
      "grad_norm": 6.350016117095947,
      "learning_rate": 0.00019879625543587056,
      "loss": 6.2877,
      "step": 46020
    },
    {
      "epoch": 5.539109506618532,
      "grad_norm": 4.834826946258545,
      "learning_rate": 0.00019879566686964393,
      "loss": 6.093,
      "step": 46030
    },
    {
      "epoch": 5.5403128760529485,
      "grad_norm": 9.270788192749023,
      "learning_rate": 0.00019879507816043554,
      "loss": 6.2919,
      "step": 46040
    },
    {
      "epoch": 5.541516245487364,
      "grad_norm": 11.917572975158691,
      "learning_rate": 0.00019879448930824629,
      "loss": 6.259,
      "step": 46050
    },
    {
      "epoch": 5.542719614921781,
      "grad_norm": 5.778326511383057,
      "learning_rate": 0.00019879390031307703,
      "loss": 6.3041,
      "step": 46060
    },
    {
      "epoch": 5.543922984356198,
      "grad_norm": 11.949620246887207,
      "learning_rate": 0.00019879331117492858,
      "loss": 6.3024,
      "step": 46070
    },
    {
      "epoch": 5.5451263537906135,
      "grad_norm": 8.083394050598145,
      "learning_rate": 0.00019879272189380186,
      "loss": 6.3281,
      "step": 46080
    },
    {
      "epoch": 5.54632972322503,
      "grad_norm": 6.177634239196777,
      "learning_rate": 0.00019879213246969766,
      "loss": 6.4272,
      "step": 46090
    },
    {
      "epoch": 5.547533092659447,
      "grad_norm": 6.1454057693481445,
      "learning_rate": 0.00019879154290261685,
      "loss": 6.2131,
      "step": 46100
    },
    {
      "epoch": 5.548736462093863,
      "grad_norm": 6.939477920532227,
      "learning_rate": 0.0001987909531925603,
      "loss": 6.2413,
      "step": 46110
    },
    {
      "epoch": 5.549939831528279,
      "grad_norm": 4.128360271453857,
      "learning_rate": 0.0001987903633395288,
      "loss": 6.3218,
      "step": 46120
    },
    {
      "epoch": 5.551143200962695,
      "grad_norm": 6.849950790405273,
      "learning_rate": 0.00019878977334352328,
      "loss": 6.3315,
      "step": 46130
    },
    {
      "epoch": 5.552346570397112,
      "grad_norm": 6.320056438446045,
      "learning_rate": 0.0001987891832045446,
      "loss": 6.2866,
      "step": 46140
    },
    {
      "epoch": 5.5535499398315284,
      "grad_norm": 7.691494464874268,
      "learning_rate": 0.00019878859292259357,
      "loss": 6.2612,
      "step": 46150
    },
    {
      "epoch": 5.554753309265944,
      "grad_norm": 6.219451904296875,
      "learning_rate": 0.00019878800249767105,
      "loss": 6.3023,
      "step": 46160
    },
    {
      "epoch": 5.555956678700361,
      "grad_norm": 4.319249153137207,
      "learning_rate": 0.0001987874119297779,
      "loss": 6.2107,
      "step": 46170
    },
    {
      "epoch": 5.557160048134778,
      "grad_norm": 7.467607021331787,
      "learning_rate": 0.00019878682121891498,
      "loss": 6.3902,
      "step": 46180
    },
    {
      "epoch": 5.558363417569193,
      "grad_norm": 39.97356033325195,
      "learning_rate": 0.0001987862303650831,
      "loss": 6.2473,
      "step": 46190
    },
    {
      "epoch": 5.55956678700361,
      "grad_norm": 11.547635078430176,
      "learning_rate": 0.0001987856393682832,
      "loss": 6.3006,
      "step": 46200
    },
    {
      "epoch": 5.560770156438027,
      "grad_norm": 5.582194805145264,
      "learning_rate": 0.00019878504822851606,
      "loss": 6.2777,
      "step": 46210
    },
    {
      "epoch": 5.5619735258724425,
      "grad_norm": 5.818282127380371,
      "learning_rate": 0.0001987844569457826,
      "loss": 6.1256,
      "step": 46220
    },
    {
      "epoch": 5.563176895306859,
      "grad_norm": 10.998391151428223,
      "learning_rate": 0.00019878386552008365,
      "loss": 6.2739,
      "step": 46230
    },
    {
      "epoch": 5.564380264741276,
      "grad_norm": 6.707700252532959,
      "learning_rate": 0.00019878327395142,
      "loss": 6.2491,
      "step": 46240
    },
    {
      "epoch": 5.565583634175692,
      "grad_norm": 17.052385330200195,
      "learning_rate": 0.0001987826822397926,
      "loss": 6.2402,
      "step": 46250
    },
    {
      "epoch": 5.566787003610108,
      "grad_norm": 9.396872520446777,
      "learning_rate": 0.00019878209038520225,
      "loss": 6.4235,
      "step": 46260
    },
    {
      "epoch": 5.567990373044525,
      "grad_norm": 8.486729621887207,
      "learning_rate": 0.00019878149838764987,
      "loss": 6.3466,
      "step": 46270
    },
    {
      "epoch": 5.569193742478941,
      "grad_norm": 9.036606788635254,
      "learning_rate": 0.00019878090624713624,
      "loss": 6.3939,
      "step": 46280
    },
    {
      "epoch": 5.5703971119133575,
      "grad_norm": 8.542448997497559,
      "learning_rate": 0.00019878031396366224,
      "loss": 6.4378,
      "step": 46290
    },
    {
      "epoch": 5.571600481347774,
      "grad_norm": 13.90946102142334,
      "learning_rate": 0.00019877972153722876,
      "loss": 6.3125,
      "step": 46300
    },
    {
      "epoch": 5.57280385078219,
      "grad_norm": 46.43960189819336,
      "learning_rate": 0.00019877912896783663,
      "loss": 6.3317,
      "step": 46310
    },
    {
      "epoch": 5.574007220216607,
      "grad_norm": 33.592288970947266,
      "learning_rate": 0.00019877853625548673,
      "loss": 6.3732,
      "step": 46320
    },
    {
      "epoch": 5.575210589651023,
      "grad_norm": 105.87428283691406,
      "learning_rate": 0.0001987779434001799,
      "loss": 6.5342,
      "step": 46330
    },
    {
      "epoch": 5.576413959085439,
      "grad_norm": 9.815814971923828,
      "learning_rate": 0.00019877735040191695,
      "loss": 6.5159,
      "step": 46340
    },
    {
      "epoch": 5.577617328519856,
      "grad_norm": 24.510147094726562,
      "learning_rate": 0.00019877675726069882,
      "loss": 6.5053,
      "step": 46350
    },
    {
      "epoch": 5.578820697954272,
      "grad_norm": 23.83465003967285,
      "learning_rate": 0.00019877616397652635,
      "loss": 6.5839,
      "step": 46360
    },
    {
      "epoch": 5.580024067388688,
      "grad_norm": 34.723670959472656,
      "learning_rate": 0.00019877557054940033,
      "loss": 6.843,
      "step": 46370
    },
    {
      "epoch": 5.581227436823105,
      "grad_norm": 16.495534896850586,
      "learning_rate": 0.00019877497697932172,
      "loss": 6.4301,
      "step": 46380
    },
    {
      "epoch": 5.582430806257521,
      "grad_norm": 16.811086654663086,
      "learning_rate": 0.00019877438326629132,
      "loss": 6.5386,
      "step": 46390
    },
    {
      "epoch": 5.583634175691937,
      "grad_norm": 17.09915542602539,
      "learning_rate": 0.00019877378941031003,
      "loss": 6.5213,
      "step": 46400
    },
    {
      "epoch": 5.584837545126354,
      "grad_norm": 7.495748996734619,
      "learning_rate": 0.0001987731954113786,
      "loss": 6.4153,
      "step": 46410
    },
    {
      "epoch": 5.58604091456077,
      "grad_norm": 5.768733978271484,
      "learning_rate": 0.00019877260126949806,
      "loss": 6.3213,
      "step": 46420
    },
    {
      "epoch": 5.587244283995187,
      "grad_norm": 9.990574836730957,
      "learning_rate": 0.0001987720069846691,
      "loss": 6.2859,
      "step": 46430
    },
    {
      "epoch": 5.588447653429603,
      "grad_norm": 6.547935485839844,
      "learning_rate": 0.00019877141255689274,
      "loss": 6.3208,
      "step": 46440
    },
    {
      "epoch": 5.589651022864019,
      "grad_norm": 6.238420009613037,
      "learning_rate": 0.00019877081798616973,
      "loss": 6.2468,
      "step": 46450
    },
    {
      "epoch": 5.590854392298436,
      "grad_norm": 28.083812713623047,
      "learning_rate": 0.00019877022327250093,
      "loss": 6.2582,
      "step": 46460
    },
    {
      "epoch": 5.5920577617328515,
      "grad_norm": 12.323185920715332,
      "learning_rate": 0.00019876962841588726,
      "loss": 6.2916,
      "step": 46470
    },
    {
      "epoch": 5.593261131167268,
      "grad_norm": 9.366296768188477,
      "learning_rate": 0.00019876903341632955,
      "loss": 6.2955,
      "step": 46480
    },
    {
      "epoch": 5.594464500601685,
      "grad_norm": 19.47185707092285,
      "learning_rate": 0.00019876843827382865,
      "loss": 6.3823,
      "step": 46490
    },
    {
      "epoch": 5.595667870036101,
      "grad_norm": 152.86924743652344,
      "learning_rate": 0.00019876784298838545,
      "loss": 6.3114,
      "step": 46500
    },
    {
      "epoch": 5.596871239470517,
      "grad_norm": 26.83968162536621,
      "learning_rate": 0.00019876724756000077,
      "loss": 7.0908,
      "step": 46510
    },
    {
      "epoch": 5.598074608904934,
      "grad_norm": 11.151542663574219,
      "learning_rate": 0.00019876665198867555,
      "loss": 7.0573,
      "step": 46520
    },
    {
      "epoch": 5.59927797833935,
      "grad_norm": 11.737648963928223,
      "learning_rate": 0.00019876605627441057,
      "loss": 6.849,
      "step": 46530
    },
    {
      "epoch": 5.6004813477737665,
      "grad_norm": 27.343643188476562,
      "learning_rate": 0.0001987654604172067,
      "loss": 6.6941,
      "step": 46540
    },
    {
      "epoch": 5.601684717208183,
      "grad_norm": 18.024578094482422,
      "learning_rate": 0.00019876486441706485,
      "loss": 6.7262,
      "step": 46550
    },
    {
      "epoch": 5.602888086642599,
      "grad_norm": 22.21957015991211,
      "learning_rate": 0.00019876426827398585,
      "loss": 6.6684,
      "step": 46560
    },
    {
      "epoch": 5.604091456077016,
      "grad_norm": 38.372962951660156,
      "learning_rate": 0.00019876367198797055,
      "loss": 6.4944,
      "step": 46570
    },
    {
      "epoch": 5.605294825511432,
      "grad_norm": 33.13584899902344,
      "learning_rate": 0.00019876307555901986,
      "loss": 6.4411,
      "step": 46580
    },
    {
      "epoch": 5.606498194945848,
      "grad_norm": 16.798213958740234,
      "learning_rate": 0.0001987624789871346,
      "loss": 6.4011,
      "step": 46590
    },
    {
      "epoch": 5.607701564380265,
      "grad_norm": 28.71711540222168,
      "learning_rate": 0.00019876188227231566,
      "loss": 6.3832,
      "step": 46600
    },
    {
      "epoch": 5.6089049338146815,
      "grad_norm": 38.26422882080078,
      "learning_rate": 0.0001987612854145639,
      "loss": 6.6683,
      "step": 46610
    },
    {
      "epoch": 5.610108303249097,
      "grad_norm": 36.17931365966797,
      "learning_rate": 0.00019876068841388018,
      "loss": 6.7922,
      "step": 46620
    },
    {
      "epoch": 5.611311672683514,
      "grad_norm": 56.7069091796875,
      "learning_rate": 0.00019876009127026533,
      "loss": 6.6921,
      "step": 46630
    },
    {
      "epoch": 5.612515042117931,
      "grad_norm": 57.587615966796875,
      "learning_rate": 0.00019875949398372026,
      "loss": 6.6318,
      "step": 46640
    },
    {
      "epoch": 5.613718411552346,
      "grad_norm": 51.06390380859375,
      "learning_rate": 0.00019875889655424583,
      "loss": 6.7319,
      "step": 46650
    },
    {
      "epoch": 5.614921780986763,
      "grad_norm": 23.352514266967773,
      "learning_rate": 0.0001987582989818429,
      "loss": 6.5627,
      "step": 46660
    },
    {
      "epoch": 5.61612515042118,
      "grad_norm": 42.385128021240234,
      "learning_rate": 0.00019875770126651232,
      "loss": 6.6787,
      "step": 46670
    },
    {
      "epoch": 5.617328519855596,
      "grad_norm": 13.203081130981445,
      "learning_rate": 0.00019875710340825492,
      "loss": 6.5287,
      "step": 46680
    },
    {
      "epoch": 5.618531889290012,
      "grad_norm": 19.75714111328125,
      "learning_rate": 0.00019875650540707168,
      "loss": 6.5904,
      "step": 46690
    },
    {
      "epoch": 5.619735258724428,
      "grad_norm": 50.870399475097656,
      "learning_rate": 0.00019875590726296336,
      "loss": 6.7852,
      "step": 46700
    },
    {
      "epoch": 5.620938628158845,
      "grad_norm": 1110.648681640625,
      "learning_rate": 0.00019875530897593086,
      "loss": 6.8428,
      "step": 46710
    },
    {
      "epoch": 5.622141997593261,
      "grad_norm": 187.42201232910156,
      "learning_rate": 0.00019875471054597505,
      "loss": 7.1965,
      "step": 46720
    },
    {
      "epoch": 5.623345367027677,
      "grad_norm": 11.785093307495117,
      "learning_rate": 0.0001987541119730968,
      "loss": 7.4633,
      "step": 46730
    },
    {
      "epoch": 5.624548736462094,
      "grad_norm": 4.544264793395996,
      "learning_rate": 0.00019875351325729693,
      "loss": 7.4169,
      "step": 46740
    },
    {
      "epoch": 5.6257521058965105,
      "grad_norm": 11.500255584716797,
      "learning_rate": 0.00019875291439857637,
      "loss": 6.9042,
      "step": 46750
    },
    {
      "epoch": 5.626955475330926,
      "grad_norm": 22.129554748535156,
      "learning_rate": 0.00019875231539693598,
      "loss": 6.5528,
      "step": 46760
    },
    {
      "epoch": 5.628158844765343,
      "grad_norm": 13.432845115661621,
      "learning_rate": 0.0001987517162523766,
      "loss": 6.516,
      "step": 46770
    },
    {
      "epoch": 5.62936221419976,
      "grad_norm": 25.117473602294922,
      "learning_rate": 0.0001987511169648991,
      "loss": 6.6237,
      "step": 46780
    },
    {
      "epoch": 5.6305655836341755,
      "grad_norm": 15.665063858032227,
      "learning_rate": 0.00019875051753450435,
      "loss": 6.5602,
      "step": 46790
    },
    {
      "epoch": 5.631768953068592,
      "grad_norm": 7.756671905517578,
      "learning_rate": 0.00019874991796119321,
      "loss": 6.7064,
      "step": 46800
    },
    {
      "epoch": 5.632972322503008,
      "grad_norm": 16.361539840698242,
      "learning_rate": 0.00019874931824496658,
      "loss": 6.7826,
      "step": 46810
    },
    {
      "epoch": 5.634175691937425,
      "grad_norm": 22.501079559326172,
      "learning_rate": 0.0001987487183858253,
      "loss": 6.6066,
      "step": 46820
    },
    {
      "epoch": 5.635379061371841,
      "grad_norm": 87.65595245361328,
      "learning_rate": 0.00019874811838377024,
      "loss": 6.5869,
      "step": 46830
    },
    {
      "epoch": 5.636582430806257,
      "grad_norm": 113.44232177734375,
      "learning_rate": 0.00019874751823880228,
      "loss": 6.6955,
      "step": 46840
    },
    {
      "epoch": 5.637785800240674,
      "grad_norm": 74.71009063720703,
      "learning_rate": 0.00019874691795092225,
      "loss": 6.8955,
      "step": 46850
    },
    {
      "epoch": 5.6389891696750905,
      "grad_norm": 264.922119140625,
      "learning_rate": 0.0001987463175201311,
      "loss": 6.6997,
      "step": 46860
    },
    {
      "epoch": 5.640192539109506,
      "grad_norm": 64.98933410644531,
      "learning_rate": 0.00019874571694642961,
      "loss": 6.9005,
      "step": 46870
    },
    {
      "epoch": 5.641395908543923,
      "grad_norm": 457.2339782714844,
      "learning_rate": 0.0001987451162298187,
      "loss": 7.1698,
      "step": 46880
    },
    {
      "epoch": 5.64259927797834,
      "grad_norm": 226.53030395507812,
      "learning_rate": 0.00019874451537029925,
      "loss": 6.9027,
      "step": 46890
    },
    {
      "epoch": 5.643802647412755,
      "grad_norm": 60.34992218017578,
      "learning_rate": 0.00019874391436787209,
      "loss": 6.8031,
      "step": 46900
    },
    {
      "epoch": 5.645006016847172,
      "grad_norm": 39.77439498901367,
      "learning_rate": 0.00019874331322253808,
      "loss": 6.7717,
      "step": 46910
    },
    {
      "epoch": 5.646209386281589,
      "grad_norm": 878.1384887695312,
      "learning_rate": 0.00019874271193429815,
      "loss": 6.9093,
      "step": 46920
    },
    {
      "epoch": 5.6474127557160045,
      "grad_norm": 34.1122932434082,
      "learning_rate": 0.0001987421105031531,
      "loss": 6.8495,
      "step": 46930
    },
    {
      "epoch": 5.648616125150421,
      "grad_norm": 90.69535827636719,
      "learning_rate": 0.0001987415089291039,
      "loss": 6.8004,
      "step": 46940
    },
    {
      "epoch": 5.649819494584838,
      "grad_norm": 131.42843627929688,
      "learning_rate": 0.00019874090721215133,
      "loss": 7.014,
      "step": 46950
    },
    {
      "epoch": 5.651022864019254,
      "grad_norm": 52.892433166503906,
      "learning_rate": 0.00019874030535229628,
      "loss": 7.0594,
      "step": 46960
    },
    {
      "epoch": 5.65222623345367,
      "grad_norm": 400.1392517089844,
      "learning_rate": 0.00019873970334953963,
      "loss": 6.9954,
      "step": 46970
    },
    {
      "epoch": 5.653429602888087,
      "grad_norm": 106.0921401977539,
      "learning_rate": 0.00019873910120388224,
      "loss": 7.1363,
      "step": 46980
    },
    {
      "epoch": 5.654632972322503,
      "grad_norm": 36.46430206298828,
      "learning_rate": 0.00019873849891532502,
      "loss": 7.3101,
      "step": 46990
    },
    {
      "epoch": 5.6558363417569195,
      "grad_norm": 5.877980709075928,
      "learning_rate": 0.00019873789648386877,
      "loss": 7.4363,
      "step": 47000
    },
    {
      "epoch": 5.657039711191336,
      "grad_norm": 9.77474308013916,
      "learning_rate": 0.00019873729390951445,
      "loss": 7.2457,
      "step": 47010
    },
    {
      "epoch": 5.658243080625752,
      "grad_norm": 24.199386596679688,
      "learning_rate": 0.00019873669119226288,
      "loss": 7.056,
      "step": 47020
    },
    {
      "epoch": 5.659446450060169,
      "grad_norm": 15.595993041992188,
      "learning_rate": 0.00019873608833211493,
      "loss": 6.8808,
      "step": 47030
    },
    {
      "epoch": 5.6606498194945845,
      "grad_norm": 9.290364265441895,
      "learning_rate": 0.0001987354853290715,
      "loss": 6.7269,
      "step": 47040
    },
    {
      "epoch": 5.661853188929001,
      "grad_norm": 28.148937225341797,
      "learning_rate": 0.0001987348821831334,
      "loss": 6.6595,
      "step": 47050
    },
    {
      "epoch": 5.663056558363418,
      "grad_norm": 52.12007522583008,
      "learning_rate": 0.00019873427889430159,
      "loss": 6.5249,
      "step": 47060
    },
    {
      "epoch": 5.664259927797834,
      "grad_norm": 116.76371002197266,
      "learning_rate": 0.0001987336754625769,
      "loss": 6.7813,
      "step": 47070
    },
    {
      "epoch": 5.66546329723225,
      "grad_norm": 57.85456848144531,
      "learning_rate": 0.0001987330718879602,
      "loss": 6.6487,
      "step": 47080
    },
    {
      "epoch": 5.666666666666667,
      "grad_norm": 174.3936004638672,
      "learning_rate": 0.00019873246817045238,
      "loss": 6.6781,
      "step": 47090
    },
    {
      "epoch": 5.667870036101083,
      "grad_norm": 157.48199462890625,
      "learning_rate": 0.00019873186431005428,
      "loss": 6.9296,
      "step": 47100
    },
    {
      "epoch": 5.669073405535499,
      "grad_norm": 82.49263763427734,
      "learning_rate": 0.00019873126030676682,
      "loss": 6.8377,
      "step": 47110
    },
    {
      "epoch": 5.670276774969916,
      "grad_norm": 85.44804382324219,
      "learning_rate": 0.00019873065616059083,
      "loss": 6.9748,
      "step": 47120
    },
    {
      "epoch": 5.671480144404332,
      "grad_norm": 89.8354721069336,
      "learning_rate": 0.0001987300518715272,
      "loss": 6.7323,
      "step": 47130
    },
    {
      "epoch": 5.672683513838749,
      "grad_norm": 54.78339767456055,
      "learning_rate": 0.00019872944743957682,
      "loss": 6.4902,
      "step": 47140
    },
    {
      "epoch": 5.673886883273164,
      "grad_norm": 46.192203521728516,
      "learning_rate": 0.00019872884286474055,
      "loss": 6.5638,
      "step": 47150
    },
    {
      "epoch": 5.675090252707581,
      "grad_norm": 59.51628494262695,
      "learning_rate": 0.0001987282381470193,
      "loss": 6.476,
      "step": 47160
    },
    {
      "epoch": 5.676293622141998,
      "grad_norm": 252.3123321533203,
      "learning_rate": 0.00019872763328641387,
      "loss": 6.6872,
      "step": 47170
    },
    {
      "epoch": 5.6774969915764135,
      "grad_norm": 12.463716506958008,
      "learning_rate": 0.0001987270282829252,
      "loss": 7.0556,
      "step": 47180
    },
    {
      "epoch": 5.67870036101083,
      "grad_norm": 12.086864471435547,
      "learning_rate": 0.00019872642313655415,
      "loss": 6.9903,
      "step": 47190
    },
    {
      "epoch": 5.679903730445247,
      "grad_norm": 23.978946685791016,
      "learning_rate": 0.00019872581784730157,
      "loss": 6.9209,
      "step": 47200
    },
    {
      "epoch": 5.681107099879663,
      "grad_norm": 25.509599685668945,
      "learning_rate": 0.00019872521241516837,
      "loss": 6.7212,
      "step": 47210
    },
    {
      "epoch": 5.682310469314079,
      "grad_norm": 15.048419952392578,
      "learning_rate": 0.00019872460684015541,
      "loss": 6.6035,
      "step": 47220
    },
    {
      "epoch": 5.683513838748496,
      "grad_norm": 31.152725219726562,
      "learning_rate": 0.00019872400112226358,
      "loss": 6.6196,
      "step": 47230
    },
    {
      "epoch": 5.684717208182912,
      "grad_norm": 22.092208862304688,
      "learning_rate": 0.00019872339526149374,
      "loss": 6.8158,
      "step": 47240
    },
    {
      "epoch": 5.6859205776173285,
      "grad_norm": 46.9203987121582,
      "learning_rate": 0.00019872278925784674,
      "loss": 7.0549,
      "step": 47250
    },
    {
      "epoch": 5.687123947051745,
      "grad_norm": 12.98798942565918,
      "learning_rate": 0.00019872218311132355,
      "loss": 7.1165,
      "step": 47260
    },
    {
      "epoch": 5.688327316486161,
      "grad_norm": 7.375025749206543,
      "learning_rate": 0.00019872157682192495,
      "loss": 6.7941,
      "step": 47270
    },
    {
      "epoch": 5.689530685920578,
      "grad_norm": 11.47173023223877,
      "learning_rate": 0.00019872097038965187,
      "loss": 6.5747,
      "step": 47280
    },
    {
      "epoch": 5.690734055354994,
      "grad_norm": 35.984779357910156,
      "learning_rate": 0.00019872036381450516,
      "loss": 6.6348,
      "step": 47290
    },
    {
      "epoch": 5.69193742478941,
      "grad_norm": 9.054459571838379,
      "learning_rate": 0.00019871975709648574,
      "loss": 6.5689,
      "step": 47300
    },
    {
      "epoch": 5.693140794223827,
      "grad_norm": 8.606990814208984,
      "learning_rate": 0.00019871915023559442,
      "loss": 6.5106,
      "step": 47310
    },
    {
      "epoch": 5.6943441636582435,
      "grad_norm": 7.831537246704102,
      "learning_rate": 0.00019871854323183215,
      "loss": 6.5651,
      "step": 47320
    },
    {
      "epoch": 5.695547533092659,
      "grad_norm": 8.106771469116211,
      "learning_rate": 0.00019871793608519975,
      "loss": 6.6654,
      "step": 47330
    },
    {
      "epoch": 5.696750902527076,
      "grad_norm": 11.992792129516602,
      "learning_rate": 0.00019871732879569813,
      "loss": 6.5571,
      "step": 47340
    },
    {
      "epoch": 5.697954271961493,
      "grad_norm": 8.731738090515137,
      "learning_rate": 0.00019871672136332816,
      "loss": 6.496,
      "step": 47350
    },
    {
      "epoch": 5.699157641395908,
      "grad_norm": 14.152222633361816,
      "learning_rate": 0.00019871611378809073,
      "loss": 6.4799,
      "step": 47360
    },
    {
      "epoch": 5.700361010830325,
      "grad_norm": 6.566752910614014,
      "learning_rate": 0.0001987155060699867,
      "loss": 6.5309,
      "step": 47370
    },
    {
      "epoch": 5.701564380264742,
      "grad_norm": 9.424914360046387,
      "learning_rate": 0.00019871489820901697,
      "loss": 6.3605,
      "step": 47380
    },
    {
      "epoch": 5.702767749699158,
      "grad_norm": 14.224370002746582,
      "learning_rate": 0.0001987142902051824,
      "loss": 6.4596,
      "step": 47390
    },
    {
      "epoch": 5.703971119133574,
      "grad_norm": 11.61376953125,
      "learning_rate": 0.00019871368205848388,
      "loss": 6.4664,
      "step": 47400
    },
    {
      "epoch": 5.70517448856799,
      "grad_norm": 26.754058837890625,
      "learning_rate": 0.00019871307376892233,
      "loss": 6.509,
      "step": 47410
    },
    {
      "epoch": 5.706377858002407,
      "grad_norm": 30.677982330322266,
      "learning_rate": 0.00019871246533649853,
      "loss": 6.6542,
      "step": 47420
    },
    {
      "epoch": 5.707581227436823,
      "grad_norm": 64.0774154663086,
      "learning_rate": 0.00019871185676121345,
      "loss": 6.5922,
      "step": 47430
    },
    {
      "epoch": 5.708784596871239,
      "grad_norm": 34.35859680175781,
      "learning_rate": 0.00019871124804306795,
      "loss": 6.6286,
      "step": 47440
    },
    {
      "epoch": 5.709987966305656,
      "grad_norm": 22.575218200683594,
      "learning_rate": 0.0001987106391820629,
      "loss": 6.5684,
      "step": 47450
    },
    {
      "epoch": 5.7111913357400725,
      "grad_norm": 23.261507034301758,
      "learning_rate": 0.0001987100301781992,
      "loss": 6.4519,
      "step": 47460
    },
    {
      "epoch": 5.712394705174488,
      "grad_norm": 31.767271041870117,
      "learning_rate": 0.00019870942103147768,
      "loss": 6.5099,
      "step": 47470
    },
    {
      "epoch": 5.713598074608905,
      "grad_norm": 96.95401763916016,
      "learning_rate": 0.00019870881174189928,
      "loss": 6.5382,
      "step": 47480
    },
    {
      "epoch": 5.714801444043322,
      "grad_norm": 10.955039978027344,
      "learning_rate": 0.00019870820230946486,
      "loss": 6.615,
      "step": 47490
    },
    {
      "epoch": 5.7160048134777375,
      "grad_norm": 21.750797271728516,
      "learning_rate": 0.00019870759273417528,
      "loss": 6.324,
      "step": 47500
    },
    {
      "epoch": 5.717208182912154,
      "grad_norm": 59.43056869506836,
      "learning_rate": 0.00019870698301603144,
      "loss": 6.7201,
      "step": 47510
    },
    {
      "epoch": 5.71841155234657,
      "grad_norm": 36.652042388916016,
      "learning_rate": 0.00019870637315503426,
      "loss": 6.7443,
      "step": 47520
    },
    {
      "epoch": 5.719614921780987,
      "grad_norm": 50.06197738647461,
      "learning_rate": 0.00019870576315118456,
      "loss": 6.6402,
      "step": 47530
    },
    {
      "epoch": 5.720818291215403,
      "grad_norm": 28.871110916137695,
      "learning_rate": 0.00019870515300448327,
      "loss": 6.5999,
      "step": 47540
    },
    {
      "epoch": 5.722021660649819,
      "grad_norm": 7.493163108825684,
      "learning_rate": 0.00019870454271493122,
      "loss": 6.4686,
      "step": 47550
    },
    {
      "epoch": 5.723225030084236,
      "grad_norm": 25.0213565826416,
      "learning_rate": 0.00019870393228252936,
      "loss": 6.5981,
      "step": 47560
    },
    {
      "epoch": 5.7244283995186525,
      "grad_norm": 41.69451141357422,
      "learning_rate": 0.00019870332170727854,
      "loss": 6.5718,
      "step": 47570
    },
    {
      "epoch": 5.725631768953068,
      "grad_norm": 11.689252853393555,
      "learning_rate": 0.00019870271098917965,
      "loss": 6.6932,
      "step": 47580
    },
    {
      "epoch": 5.726835138387485,
      "grad_norm": 17.256437301635742,
      "learning_rate": 0.00019870210012823354,
      "loss": 6.4876,
      "step": 47590
    },
    {
      "epoch": 5.728038507821902,
      "grad_norm": 9.13249683380127,
      "learning_rate": 0.00019870148912444117,
      "loss": 6.4415,
      "step": 47600
    },
    {
      "epoch": 5.729241877256317,
      "grad_norm": 108.54875946044922,
      "learning_rate": 0.0001987008779778033,
      "loss": 6.3982,
      "step": 47610
    },
    {
      "epoch": 5.730445246690734,
      "grad_norm": 20.35406494140625,
      "learning_rate": 0.00019870026668832095,
      "loss": 6.409,
      "step": 47620
    },
    {
      "epoch": 5.731648616125151,
      "grad_norm": 18.480241775512695,
      "learning_rate": 0.00019869965525599493,
      "loss": 6.4184,
      "step": 47630
    },
    {
      "epoch": 5.7328519855595665,
      "grad_norm": 21.412052154541016,
      "learning_rate": 0.00019869904368082614,
      "loss": 6.4053,
      "step": 47640
    },
    {
      "epoch": 5.734055354993983,
      "grad_norm": 90.26350402832031,
      "learning_rate": 0.00019869843196281548,
      "loss": 6.4532,
      "step": 47650
    },
    {
      "epoch": 5.7352587244284,
      "grad_norm": 13.593746185302734,
      "learning_rate": 0.0001986978201019638,
      "loss": 6.5318,
      "step": 47660
    },
    {
      "epoch": 5.736462093862816,
      "grad_norm": 13.714799880981445,
      "learning_rate": 0.00019869720809827202,
      "loss": 6.4598,
      "step": 47670
    },
    {
      "epoch": 5.737665463297232,
      "grad_norm": 21.631376266479492,
      "learning_rate": 0.00019869659595174102,
      "loss": 6.5457,
      "step": 47680
    },
    {
      "epoch": 5.738868832731649,
      "grad_norm": 65.28184509277344,
      "learning_rate": 0.00019869598366237164,
      "loss": 6.4729,
      "step": 47690
    },
    {
      "epoch": 5.740072202166065,
      "grad_norm": 32.66143798828125,
      "learning_rate": 0.00019869537123016483,
      "loss": 6.6336,
      "step": 47700
    },
    {
      "epoch": 5.7412755716004815,
      "grad_norm": 53.060203552246094,
      "learning_rate": 0.00019869475865512146,
      "loss": 6.3433,
      "step": 47710
    },
    {
      "epoch": 5.742478941034898,
      "grad_norm": 42.53267288208008,
      "learning_rate": 0.0001986941459372424,
      "loss": 6.436,
      "step": 47720
    },
    {
      "epoch": 5.743682310469314,
      "grad_norm": 31.560754776000977,
      "learning_rate": 0.00019869353307652853,
      "loss": 6.3752,
      "step": 47730
    },
    {
      "epoch": 5.744885679903731,
      "grad_norm": 25.02374267578125,
      "learning_rate": 0.0001986929200729808,
      "loss": 6.384,
      "step": 47740
    },
    {
      "epoch": 5.7460890493381465,
      "grad_norm": 28.127748489379883,
      "learning_rate": 0.00019869230692659998,
      "loss": 6.3578,
      "step": 47750
    },
    {
      "epoch": 5.747292418772563,
      "grad_norm": 92.90540313720703,
      "learning_rate": 0.00019869169363738706,
      "loss": 6.3625,
      "step": 47760
    },
    {
      "epoch": 5.74849578820698,
      "grad_norm": 48.94447708129883,
      "learning_rate": 0.0001986910802053429,
      "loss": 6.4407,
      "step": 47770
    },
    {
      "epoch": 5.749699157641396,
      "grad_norm": 37.32133865356445,
      "learning_rate": 0.00019869046663046838,
      "loss": 6.3567,
      "step": 47780
    },
    {
      "epoch": 5.750902527075812,
      "grad_norm": 42.19178009033203,
      "learning_rate": 0.00019868985291276436,
      "loss": 6.6661,
      "step": 47790
    },
    {
      "epoch": 5.752105896510229,
      "grad_norm": 136.2880096435547,
      "learning_rate": 0.00019868923905223178,
      "loss": 6.5138,
      "step": 47800
    },
    {
      "epoch": 5.753309265944645,
      "grad_norm": 68.13109588623047,
      "learning_rate": 0.00019868862504887153,
      "loss": 6.3546,
      "step": 47810
    },
    {
      "epoch": 5.754512635379061,
      "grad_norm": 159.12734985351562,
      "learning_rate": 0.00019868801090268444,
      "loss": 6.4623,
      "step": 47820
    },
    {
      "epoch": 5.755716004813478,
      "grad_norm": 179.521240234375,
      "learning_rate": 0.00019868739661367145,
      "loss": 6.7468,
      "step": 47830
    },
    {
      "epoch": 5.756919374247894,
      "grad_norm": 665.4715576171875,
      "learning_rate": 0.0001986867821818334,
      "loss": 6.9277,
      "step": 47840
    },
    {
      "epoch": 5.758122743682311,
      "grad_norm": 59.42085647583008,
      "learning_rate": 0.00019868616760717123,
      "loss": 6.5894,
      "step": 47850
    },
    {
      "epoch": 5.759326113116726,
      "grad_norm": 226.77992248535156,
      "learning_rate": 0.00019868555288968582,
      "loss": 6.5083,
      "step": 47860
    },
    {
      "epoch": 5.760529482551143,
      "grad_norm": 62.35588455200195,
      "learning_rate": 0.00019868493802937805,
      "loss": 6.5179,
      "step": 47870
    },
    {
      "epoch": 5.76173285198556,
      "grad_norm": 129.4388885498047,
      "learning_rate": 0.00019868432302624879,
      "loss": 6.3476,
      "step": 47880
    },
    {
      "epoch": 5.7629362214199755,
      "grad_norm": 76.39713287353516,
      "learning_rate": 0.00019868370788029898,
      "loss": 6.4871,
      "step": 47890
    },
    {
      "epoch": 5.764139590854392,
      "grad_norm": 71.60661315917969,
      "learning_rate": 0.00019868309259152945,
      "loss": 6.5476,
      "step": 47900
    },
    {
      "epoch": 5.765342960288809,
      "grad_norm": 70.02320098876953,
      "learning_rate": 0.00019868247715994114,
      "loss": 6.5925,
      "step": 47910
    },
    {
      "epoch": 5.766546329723225,
      "grad_norm": 32.23582077026367,
      "learning_rate": 0.00019868186158553494,
      "loss": 6.741,
      "step": 47920
    },
    {
      "epoch": 5.767749699157641,
      "grad_norm": 52.133811950683594,
      "learning_rate": 0.0001986812458683117,
      "loss": 6.4957,
      "step": 47930
    },
    {
      "epoch": 5.768953068592058,
      "grad_norm": 33.83395767211914,
      "learning_rate": 0.00019868063000827232,
      "loss": 6.5662,
      "step": 47940
    },
    {
      "epoch": 5.770156438026474,
      "grad_norm": 27.215024948120117,
      "learning_rate": 0.00019868001400541772,
      "loss": 6.5178,
      "step": 47950
    },
    {
      "epoch": 5.7713598074608905,
      "grad_norm": 42.432456970214844,
      "learning_rate": 0.00019867939785974875,
      "loss": 6.411,
      "step": 47960
    },
    {
      "epoch": 5.772563176895307,
      "grad_norm": 61.703800201416016,
      "learning_rate": 0.00019867878157126638,
      "loss": 6.5101,
      "step": 47970
    },
    {
      "epoch": 5.773766546329723,
      "grad_norm": 46.41755294799805,
      "learning_rate": 0.00019867816513997142,
      "loss": 6.3952,
      "step": 47980
    },
    {
      "epoch": 5.77496991576414,
      "grad_norm": 61.51991653442383,
      "learning_rate": 0.00019867754856586477,
      "loss": 6.454,
      "step": 47990
    },
    {
      "epoch": 5.776173285198556,
      "grad_norm": 75.86617279052734,
      "learning_rate": 0.00019867693184894738,
      "loss": 6.5089,
      "step": 48000
    },
    {
      "epoch": 5.777376654632972,
      "grad_norm": 116.02661895751953,
      "learning_rate": 0.0001986763149892201,
      "loss": 6.5398,
      "step": 48010
    },
    {
      "epoch": 5.778580024067389,
      "grad_norm": 35.84797668457031,
      "learning_rate": 0.00019867569798668382,
      "loss": 6.586,
      "step": 48020
    },
    {
      "epoch": 5.7797833935018055,
      "grad_norm": 106.19542694091797,
      "learning_rate": 0.00019867508084133947,
      "loss": 6.8426,
      "step": 48030
    },
    {
      "epoch": 5.780986762936221,
      "grad_norm": 81.0821304321289,
      "learning_rate": 0.0001986744635531879,
      "loss": 6.8483,
      "step": 48040
    },
    {
      "epoch": 5.782190132370638,
      "grad_norm": 59.08488845825195,
      "learning_rate": 0.00019867384612223,
      "loss": 7.0026,
      "step": 48050
    },
    {
      "epoch": 5.783393501805055,
      "grad_norm": 101.26488494873047,
      "learning_rate": 0.0001986732285484667,
      "loss": 6.98,
      "step": 48060
    },
    {
      "epoch": 5.78459687123947,
      "grad_norm": 37.543392181396484,
      "learning_rate": 0.00019867261083189887,
      "loss": 6.7258,
      "step": 48070
    },
    {
      "epoch": 5.785800240673887,
      "grad_norm": 25.11439323425293,
      "learning_rate": 0.0001986719929725274,
      "loss": 6.663,
      "step": 48080
    },
    {
      "epoch": 5.787003610108303,
      "grad_norm": 130.6200714111328,
      "learning_rate": 0.00019867137497035323,
      "loss": 6.7303,
      "step": 48090
    },
    {
      "epoch": 5.78820697954272,
      "grad_norm": 5723.11767578125,
      "learning_rate": 0.0001986707568253772,
      "loss": 7.0327,
      "step": 48100
    },
    {
      "epoch": 5.789410348977136,
      "grad_norm": 14.471563339233398,
      "learning_rate": 0.00019867013853760025,
      "loss": 7.2823,
      "step": 48110
    },
    {
      "epoch": 5.790613718411552,
      "grad_norm": 17.63632583618164,
      "learning_rate": 0.0001986695201070232,
      "loss": 7.1626,
      "step": 48120
    },
    {
      "epoch": 5.791817087845969,
      "grad_norm": 65.20819854736328,
      "learning_rate": 0.00019866890153364701,
      "loss": 6.791,
      "step": 48130
    },
    {
      "epoch": 5.793020457280385,
      "grad_norm": 82.6868896484375,
      "learning_rate": 0.00019866828281747256,
      "loss": 6.8117,
      "step": 48140
    },
    {
      "epoch": 5.794223826714801,
      "grad_norm": 53.12080001831055,
      "learning_rate": 0.00019866766395850073,
      "loss": 6.8404,
      "step": 48150
    },
    {
      "epoch": 5.795427196149218,
      "grad_norm": 72.31776428222656,
      "learning_rate": 0.00019866704495673248,
      "loss": 6.7715,
      "step": 48160
    },
    {
      "epoch": 5.7966305655836345,
      "grad_norm": 86.35231018066406,
      "learning_rate": 0.0001986664258121686,
      "loss": 6.8323,
      "step": 48170
    },
    {
      "epoch": 5.79783393501805,
      "grad_norm": 76.43995666503906,
      "learning_rate": 0.0001986658065248101,
      "loss": 6.8802,
      "step": 48180
    },
    {
      "epoch": 5.799037304452467,
      "grad_norm": 52.722164154052734,
      "learning_rate": 0.00019866518709465777,
      "loss": 6.7984,
      "step": 48190
    },
    {
      "epoch": 5.800240673886883,
      "grad_norm": 39.93399429321289,
      "learning_rate": 0.00019866456752171257,
      "loss": 7.1838,
      "step": 48200
    },
    {
      "epoch": 5.8014440433212995,
      "grad_norm": 9.672844886779785,
      "learning_rate": 0.00019866394780597538,
      "loss": 7.2161,
      "step": 48210
    },
    {
      "epoch": 5.802647412755716,
      "grad_norm": 15.779073715209961,
      "learning_rate": 0.0001986633279474471,
      "loss": 7.0148,
      "step": 48220
    },
    {
      "epoch": 5.803850782190132,
      "grad_norm": 15.048151016235352,
      "learning_rate": 0.00019866270794612862,
      "loss": 6.8137,
      "step": 48230
    },
    {
      "epoch": 5.805054151624549,
      "grad_norm": 7.421164512634277,
      "learning_rate": 0.00019866208780202084,
      "loss": 6.6582,
      "step": 48240
    },
    {
      "epoch": 5.806257521058965,
      "grad_norm": 12.760626792907715,
      "learning_rate": 0.0001986614675151247,
      "loss": 6.5762,
      "step": 48250
    },
    {
      "epoch": 5.807460890493381,
      "grad_norm": 46.091712951660156,
      "learning_rate": 0.000198660847085441,
      "loss": 6.5812,
      "step": 48260
    },
    {
      "epoch": 5.808664259927798,
      "grad_norm": 52.30769348144531,
      "learning_rate": 0.0001986602265129707,
      "loss": 6.7902,
      "step": 48270
    },
    {
      "epoch": 5.8098676293622145,
      "grad_norm": 20.084352493286133,
      "learning_rate": 0.0001986596057977147,
      "loss": 6.9179,
      "step": 48280
    },
    {
      "epoch": 5.81107099879663,
      "grad_norm": 8.5487699508667,
      "learning_rate": 0.00019865898493967392,
      "loss": 7.0288,
      "step": 48290
    },
    {
      "epoch": 5.812274368231047,
      "grad_norm": 53.21567916870117,
      "learning_rate": 0.00019865836393884924,
      "loss": 6.9843,
      "step": 48300
    },
    {
      "epoch": 5.813477737665464,
      "grad_norm": 26.6538028717041,
      "learning_rate": 0.00019865774279524153,
      "loss": 6.8132,
      "step": 48310
    },
    {
      "epoch": 5.814681107099879,
      "grad_norm": 29.152055740356445,
      "learning_rate": 0.0001986571215088517,
      "loss": 6.838,
      "step": 48320
    },
    {
      "epoch": 5.815884476534296,
      "grad_norm": 11.783438682556152,
      "learning_rate": 0.00019865650007968064,
      "loss": 7.1387,
      "step": 48330
    },
    {
      "epoch": 5.817087845968713,
      "grad_norm": 124.16717529296875,
      "learning_rate": 0.00019865587850772932,
      "loss": 6.9774,
      "step": 48340
    },
    {
      "epoch": 5.8182912154031285,
      "grad_norm": 43.25951385498047,
      "learning_rate": 0.00019865525679299854,
      "loss": 7.0962,
      "step": 48350
    },
    {
      "epoch": 5.819494584837545,
      "grad_norm": 28.44440460205078,
      "learning_rate": 0.00019865463493548926,
      "loss": 6.9566,
      "step": 48360
    },
    {
      "epoch": 5.820697954271962,
      "grad_norm": 8.934304237365723,
      "learning_rate": 0.00019865401293520237,
      "loss": 6.7227,
      "step": 48370
    },
    {
      "epoch": 5.821901323706378,
      "grad_norm": 8.736367225646973,
      "learning_rate": 0.00019865339079213877,
      "loss": 6.5421,
      "step": 48380
    },
    {
      "epoch": 5.823104693140794,
      "grad_norm": 13.938597679138184,
      "learning_rate": 0.00019865276850629935,
      "loss": 6.7676,
      "step": 48390
    },
    {
      "epoch": 5.824308062575211,
      "grad_norm": 15.920125007629395,
      "learning_rate": 0.000198652146077685,
      "loss": 6.6048,
      "step": 48400
    },
    {
      "epoch": 5.825511432009627,
      "grad_norm": 39.54819107055664,
      "learning_rate": 0.00019865152350629668,
      "loss": 6.5634,
      "step": 48410
    },
    {
      "epoch": 5.8267148014440435,
      "grad_norm": 360.4687805175781,
      "learning_rate": 0.00019865090079213522,
      "loss": 6.6545,
      "step": 48420
    },
    {
      "epoch": 5.827918170878459,
      "grad_norm": 67.05709838867188,
      "learning_rate": 0.00019865027793520158,
      "loss": 6.8256,
      "step": 48430
    },
    {
      "epoch": 5.829121540312876,
      "grad_norm": 14.328038215637207,
      "learning_rate": 0.00019864965493549658,
      "loss": 6.9099,
      "step": 48440
    },
    {
      "epoch": 5.830324909747293,
      "grad_norm": 25.57486915588379,
      "learning_rate": 0.0001986490317930212,
      "loss": 6.7648,
      "step": 48450
    },
    {
      "epoch": 5.8315282791817085,
      "grad_norm": 17.249244689941406,
      "learning_rate": 0.0001986484085077763,
      "loss": 6.8231,
      "step": 48460
    },
    {
      "epoch": 5.832731648616125,
      "grad_norm": 18.224124908447266,
      "learning_rate": 0.00019864778507976282,
      "loss": 6.8162,
      "step": 48470
    },
    {
      "epoch": 5.833935018050542,
      "grad_norm": 62.611942291259766,
      "learning_rate": 0.00019864716150898163,
      "loss": 6.6653,
      "step": 48480
    },
    {
      "epoch": 5.835138387484958,
      "grad_norm": 53.798423767089844,
      "learning_rate": 0.00019864653779543367,
      "loss": 6.5726,
      "step": 48490
    },
    {
      "epoch": 5.836341756919374,
      "grad_norm": 60.845558166503906,
      "learning_rate": 0.00019864591393911978,
      "loss": 6.535,
      "step": 48500
    },
    {
      "epoch": 5.837545126353791,
      "grad_norm": 317.6461486816406,
      "learning_rate": 0.00019864528994004092,
      "loss": 6.6664,
      "step": 48510
    },
    {
      "epoch": 5.838748495788207,
      "grad_norm": 59.18181610107422,
      "learning_rate": 0.00019864466579819795,
      "loss": 7.3048,
      "step": 48520
    },
    {
      "epoch": 5.839951865222623,
      "grad_norm": 23.103687286376953,
      "learning_rate": 0.00019864404151359182,
      "loss": 7.0544,
      "step": 48530
    },
    {
      "epoch": 5.841155234657039,
      "grad_norm": 13.91545581817627,
      "learning_rate": 0.0001986434170862234,
      "loss": 6.8467,
      "step": 48540
    },
    {
      "epoch": 5.842358604091456,
      "grad_norm": 35.511844635009766,
      "learning_rate": 0.00019864279251609356,
      "loss": 6.8236,
      "step": 48550
    },
    {
      "epoch": 5.843561973525873,
      "grad_norm": 37.969356536865234,
      "learning_rate": 0.00019864216780320329,
      "loss": 6.8571,
      "step": 48560
    },
    {
      "epoch": 5.844765342960288,
      "grad_norm": 19.357770919799805,
      "learning_rate": 0.00019864154294755345,
      "loss": 6.8503,
      "step": 48570
    },
    {
      "epoch": 5.845968712394705,
      "grad_norm": 82.99191284179688,
      "learning_rate": 0.0001986409179491449,
      "loss": 6.8761,
      "step": 48580
    },
    {
      "epoch": 5.847172081829122,
      "grad_norm": 57.941837310791016,
      "learning_rate": 0.0001986402928079786,
      "loss": 6.9824,
      "step": 48590
    },
    {
      "epoch": 5.8483754512635375,
      "grad_norm": 108.87654876708984,
      "learning_rate": 0.00019863966752405547,
      "loss": 6.987,
      "step": 48600
    },
    {
      "epoch": 5.849578820697954,
      "grad_norm": 85.42047882080078,
      "learning_rate": 0.00019863904209737641,
      "loss": 7.0367,
      "step": 48610
    },
    {
      "epoch": 5.850782190132371,
      "grad_norm": 48.762367248535156,
      "learning_rate": 0.00019863841652794227,
      "loss": 6.9113,
      "step": 48620
    },
    {
      "epoch": 5.851985559566787,
      "grad_norm": 65.24446868896484,
      "learning_rate": 0.000198637790815754,
      "loss": 6.9669,
      "step": 48630
    },
    {
      "epoch": 5.853188929001203,
      "grad_norm": 29.887727737426758,
      "learning_rate": 0.00019863716496081247,
      "loss": 7.2278,
      "step": 48640
    },
    {
      "epoch": 5.85439229843562,
      "grad_norm": 43.98811721801758,
      "learning_rate": 0.00019863653896311863,
      "loss": 7.1588,
      "step": 48650
    },
    {
      "epoch": 5.855595667870036,
      "grad_norm": 55.22576904296875,
      "learning_rate": 0.00019863591282267336,
      "loss": 7.1078,
      "step": 48660
    },
    {
      "epoch": 5.8567990373044525,
      "grad_norm": 67.01032257080078,
      "learning_rate": 0.00019863528653947757,
      "loss": 6.8984,
      "step": 48670
    },
    {
      "epoch": 5.858002406738869,
      "grad_norm": 19.69286346435547,
      "learning_rate": 0.00019863466011353214,
      "loss": 6.7313,
      "step": 48680
    },
    {
      "epoch": 5.859205776173285,
      "grad_norm": 31.302631378173828,
      "learning_rate": 0.00019863403354483805,
      "loss": 6.915,
      "step": 48690
    },
    {
      "epoch": 5.860409145607702,
      "grad_norm": 50.290992736816406,
      "learning_rate": 0.00019863340683339616,
      "loss": 6.9377,
      "step": 48700
    },
    {
      "epoch": 5.861612515042118,
      "grad_norm": 15.987544059753418,
      "learning_rate": 0.00019863277997920733,
      "loss": 6.9527,
      "step": 48710
    },
    {
      "epoch": 5.862815884476534,
      "grad_norm": 26.964595794677734,
      "learning_rate": 0.00019863215298227257,
      "loss": 6.7991,
      "step": 48720
    },
    {
      "epoch": 5.864019253910951,
      "grad_norm": 30.559459686279297,
      "learning_rate": 0.0001986315258425927,
      "loss": 6.7525,
      "step": 48730
    },
    {
      "epoch": 5.8652226233453675,
      "grad_norm": 9.94189739227295,
      "learning_rate": 0.00019863089856016867,
      "loss": 6.6273,
      "step": 48740
    },
    {
      "epoch": 5.866425992779783,
      "grad_norm": 20.493711471557617,
      "learning_rate": 0.0001986302711350014,
      "loss": 6.7033,
      "step": 48750
    },
    {
      "epoch": 5.8676293622142,
      "grad_norm": 96.32124328613281,
      "learning_rate": 0.00019862964356709176,
      "loss": 6.6239,
      "step": 48760
    },
    {
      "epoch": 5.868832731648616,
      "grad_norm": 47.75698471069336,
      "learning_rate": 0.00019862901585644064,
      "loss": 6.6847,
      "step": 48770
    },
    {
      "epoch": 5.870036101083032,
      "grad_norm": 218.04502868652344,
      "learning_rate": 0.00019862838800304902,
      "loss": 6.6552,
      "step": 48780
    },
    {
      "epoch": 5.871239470517449,
      "grad_norm": 11.607353210449219,
      "learning_rate": 0.00019862776000691777,
      "loss": 6.8557,
      "step": 48790
    },
    {
      "epoch": 5.872442839951865,
      "grad_norm": 17.712568283081055,
      "learning_rate": 0.0001986271318680478,
      "loss": 6.8654,
      "step": 48800
    },
    {
      "epoch": 5.873646209386282,
      "grad_norm": 7.156992435455322,
      "learning_rate": 0.00019862650358644005,
      "loss": 6.7159,
      "step": 48810
    },
    {
      "epoch": 5.874849578820698,
      "grad_norm": 22.025548934936523,
      "learning_rate": 0.00019862587516209534,
      "loss": 6.6292,
      "step": 48820
    },
    {
      "epoch": 5.876052948255114,
      "grad_norm": 185.76177978515625,
      "learning_rate": 0.00019862524659501466,
      "loss": 6.6061,
      "step": 48830
    },
    {
      "epoch": 5.877256317689531,
      "grad_norm": 24.29646110534668,
      "learning_rate": 0.00019862461788519893,
      "loss": 6.635,
      "step": 48840
    },
    {
      "epoch": 5.878459687123947,
      "grad_norm": 215.8610076904297,
      "learning_rate": 0.00019862398903264898,
      "loss": 6.5595,
      "step": 48850
    },
    {
      "epoch": 5.879663056558363,
      "grad_norm": 96.23988342285156,
      "learning_rate": 0.0001986233600373658,
      "loss": 6.4922,
      "step": 48860
    },
    {
      "epoch": 5.88086642599278,
      "grad_norm": 36.56184768676758,
      "learning_rate": 0.00019862273089935026,
      "loss": 6.6663,
      "step": 48870
    },
    {
      "epoch": 5.882069795427196,
      "grad_norm": 32.58598709106445,
      "learning_rate": 0.00019862210161860329,
      "loss": 6.6444,
      "step": 48880
    },
    {
      "epoch": 5.883273164861612,
      "grad_norm": 17.777141571044922,
      "learning_rate": 0.00019862147219512578,
      "loss": 6.8146,
      "step": 48890
    },
    {
      "epoch": 5.884476534296029,
      "grad_norm": 77.6886978149414,
      "learning_rate": 0.00019862084262891865,
      "loss": 6.6615,
      "step": 48900
    },
    {
      "epoch": 5.885679903730445,
      "grad_norm": 34.8476676940918,
      "learning_rate": 0.0001986202129199828,
      "loss": 6.7911,
      "step": 48910
    },
    {
      "epoch": 5.8868832731648615,
      "grad_norm": 52.74646759033203,
      "learning_rate": 0.00019861958306831916,
      "loss": 6.6983,
      "step": 48920
    },
    {
      "epoch": 5.888086642599278,
      "grad_norm": 50.859527587890625,
      "learning_rate": 0.00019861895307392866,
      "loss": 6.6,
      "step": 48930
    },
    {
      "epoch": 5.889290012033694,
      "grad_norm": 133.7000732421875,
      "learning_rate": 0.00019861832293681215,
      "loss": 6.6405,
      "step": 48940
    },
    {
      "epoch": 5.890493381468111,
      "grad_norm": 37.14678955078125,
      "learning_rate": 0.0001986176926569706,
      "loss": 6.5676,
      "step": 48950
    },
    {
      "epoch": 5.891696750902527,
      "grad_norm": 55.223114013671875,
      "learning_rate": 0.0001986170622344049,
      "loss": 6.6399,
      "step": 48960
    },
    {
      "epoch": 5.892900120336943,
      "grad_norm": 15.49814224243164,
      "learning_rate": 0.00019861643166911596,
      "loss": 6.7189,
      "step": 48970
    },
    {
      "epoch": 5.89410348977136,
      "grad_norm": 8.983789443969727,
      "learning_rate": 0.0001986158009611047,
      "loss": 6.6321,
      "step": 48980
    },
    {
      "epoch": 5.8953068592057765,
      "grad_norm": 10.405715942382812,
      "learning_rate": 0.00019861517011037202,
      "loss": 6.6703,
      "step": 48990
    },
    {
      "epoch": 5.896510228640192,
      "grad_norm": 6.647861957550049,
      "learning_rate": 0.00019861453911691887,
      "loss": 6.6884,
      "step": 49000
    },
    {
      "epoch": 5.897713598074609,
      "grad_norm": 29.058887481689453,
      "learning_rate": 0.00019861390798074608,
      "loss": 6.5258,
      "step": 49010
    },
    {
      "epoch": 5.898916967509026,
      "grad_norm": 22.239561080932617,
      "learning_rate": 0.00019861327670185464,
      "loss": 6.6583,
      "step": 49020
    },
    {
      "epoch": 5.900120336943441,
      "grad_norm": 154.3556671142578,
      "learning_rate": 0.00019861264528024547,
      "loss": 6.6688,
      "step": 49030
    },
    {
      "epoch": 5.901323706377858,
      "grad_norm": 21.16986083984375,
      "learning_rate": 0.00019861201371591943,
      "loss": 6.7152,
      "step": 49040
    },
    {
      "epoch": 5.902527075812275,
      "grad_norm": 90.42473602294922,
      "learning_rate": 0.00019861138200887748,
      "loss": 6.6219,
      "step": 49050
    },
    {
      "epoch": 5.9037304452466906,
      "grad_norm": 17.86294174194336,
      "learning_rate": 0.0001986107501591205,
      "loss": 6.6203,
      "step": 49060
    },
    {
      "epoch": 5.904933814681107,
      "grad_norm": 14.435193061828613,
      "learning_rate": 0.0001986101181666494,
      "loss": 6.5803,
      "step": 49070
    },
    {
      "epoch": 5.906137184115524,
      "grad_norm": 8.926048278808594,
      "learning_rate": 0.00019860948603146514,
      "loss": 6.4409,
      "step": 49080
    },
    {
      "epoch": 5.90734055354994,
      "grad_norm": 43.382713317871094,
      "learning_rate": 0.0001986088537535686,
      "loss": 6.5406,
      "step": 49090
    },
    {
      "epoch": 5.908543922984356,
      "grad_norm": 18.887489318847656,
      "learning_rate": 0.00019860822133296071,
      "loss": 6.5015,
      "step": 49100
    },
    {
      "epoch": 5.909747292418773,
      "grad_norm": 49.34949493408203,
      "learning_rate": 0.00019860758876964237,
      "loss": 6.6104,
      "step": 49110
    },
    {
      "epoch": 5.910950661853189,
      "grad_norm": 13.59429931640625,
      "learning_rate": 0.0001986069560636145,
      "loss": 6.7589,
      "step": 49120
    },
    {
      "epoch": 5.9121540312876055,
      "grad_norm": 33.782684326171875,
      "learning_rate": 0.00019860632321487803,
      "loss": 6.686,
      "step": 49130
    },
    {
      "epoch": 5.913357400722021,
      "grad_norm": 9.934277534484863,
      "learning_rate": 0.00019860569022343386,
      "loss": 6.7796,
      "step": 49140
    },
    {
      "epoch": 5.914560770156438,
      "grad_norm": 27.600479125976562,
      "learning_rate": 0.0001986050570892829,
      "loss": 6.5805,
      "step": 49150
    },
    {
      "epoch": 5.915764139590855,
      "grad_norm": 20.450551986694336,
      "learning_rate": 0.0001986044238124261,
      "loss": 6.6734,
      "step": 49160
    },
    {
      "epoch": 5.9169675090252705,
      "grad_norm": 10.983992576599121,
      "learning_rate": 0.00019860379039286433,
      "loss": 6.5625,
      "step": 49170
    },
    {
      "epoch": 5.918170878459687,
      "grad_norm": 22.772666931152344,
      "learning_rate": 0.00019860315683059853,
      "loss": 6.5695,
      "step": 49180
    },
    {
      "epoch": 5.919374247894104,
      "grad_norm": 88.57467651367188,
      "learning_rate": 0.00019860252312562963,
      "loss": 6.5864,
      "step": 49190
    },
    {
      "epoch": 5.92057761732852,
      "grad_norm": 200.0468292236328,
      "learning_rate": 0.00019860188927795855,
      "loss": 6.6279,
      "step": 49200
    },
    {
      "epoch": 5.921780986762936,
      "grad_norm": 57.41989517211914,
      "learning_rate": 0.00019860125528758617,
      "loss": 6.7301,
      "step": 49210
    },
    {
      "epoch": 5.922984356197353,
      "grad_norm": 29.92462158203125,
      "learning_rate": 0.00019860062115451343,
      "loss": 6.7402,
      "step": 49220
    },
    {
      "epoch": 5.924187725631769,
      "grad_norm": 71.5514144897461,
      "learning_rate": 0.00019859998687874126,
      "loss": 6.6112,
      "step": 49230
    },
    {
      "epoch": 5.925391095066185,
      "grad_norm": 18.483539581298828,
      "learning_rate": 0.00019859935246027054,
      "loss": 6.6185,
      "step": 49240
    },
    {
      "epoch": 5.926594464500601,
      "grad_norm": 11.909351348876953,
      "learning_rate": 0.00019859871789910228,
      "loss": 6.5832,
      "step": 49250
    },
    {
      "epoch": 5.927797833935018,
      "grad_norm": 11.49924373626709,
      "learning_rate": 0.00019859808319523724,
      "loss": 6.6664,
      "step": 49260
    },
    {
      "epoch": 5.929001203369435,
      "grad_norm": 17.225391387939453,
      "learning_rate": 0.0001985974483486765,
      "loss": 6.5571,
      "step": 49270
    },
    {
      "epoch": 5.93020457280385,
      "grad_norm": 39.27605056762695,
      "learning_rate": 0.00019859681335942088,
      "loss": 6.7195,
      "step": 49280
    },
    {
      "epoch": 5.931407942238267,
      "grad_norm": 33.30513381958008,
      "learning_rate": 0.00019859617822747132,
      "loss": 6.7147,
      "step": 49290
    },
    {
      "epoch": 5.932611311672684,
      "grad_norm": 28.122390747070312,
      "learning_rate": 0.00019859554295282878,
      "loss": 6.6049,
      "step": 49300
    },
    {
      "epoch": 5.9338146811070995,
      "grad_norm": 122.17819213867188,
      "learning_rate": 0.00019859490753549412,
      "loss": 6.657,
      "step": 49310
    },
    {
      "epoch": 5.935018050541516,
      "grad_norm": 199.0071563720703,
      "learning_rate": 0.00019859427197546829,
      "loss": 6.8036,
      "step": 49320
    },
    {
      "epoch": 5.936221419975933,
      "grad_norm": 35.690677642822266,
      "learning_rate": 0.0001985936362727522,
      "loss": 6.7938,
      "step": 49330
    },
    {
      "epoch": 5.937424789410349,
      "grad_norm": 43.040279388427734,
      "learning_rate": 0.0001985930004273468,
      "loss": 6.863,
      "step": 49340
    },
    {
      "epoch": 5.938628158844765,
      "grad_norm": 41.64094161987305,
      "learning_rate": 0.00019859236443925298,
      "loss": 6.6813,
      "step": 49350
    },
    {
      "epoch": 5.939831528279182,
      "grad_norm": 66.19989776611328,
      "learning_rate": 0.00019859172830847166,
      "loss": 6.6297,
      "step": 49360
    },
    {
      "epoch": 5.941034897713598,
      "grad_norm": 142.6940460205078,
      "learning_rate": 0.00019859109203500377,
      "loss": 6.6443,
      "step": 49370
    },
    {
      "epoch": 5.9422382671480145,
      "grad_norm": 77.23341369628906,
      "learning_rate": 0.00019859045561885022,
      "loss": 6.6949,
      "step": 49380
    },
    {
      "epoch": 5.943441636582431,
      "grad_norm": 21.141132354736328,
      "learning_rate": 0.00019858981906001194,
      "loss": 6.6877,
      "step": 49390
    },
    {
      "epoch": 5.944645006016847,
      "grad_norm": 14.902994155883789,
      "learning_rate": 0.00019858918235848987,
      "loss": 6.6521,
      "step": 49400
    },
    {
      "epoch": 5.945848375451264,
      "grad_norm": 42.90838623046875,
      "learning_rate": 0.0001985885455142849,
      "loss": 6.5217,
      "step": 49410
    },
    {
      "epoch": 5.94705174488568,
      "grad_norm": 46.79252243041992,
      "learning_rate": 0.00019858790852739797,
      "loss": 6.5498,
      "step": 49420
    },
    {
      "epoch": 5.948255114320096,
      "grad_norm": 200.27420043945312,
      "learning_rate": 0.00019858727139782998,
      "loss": 6.8862,
      "step": 49430
    },
    {
      "epoch": 5.949458483754513,
      "grad_norm": 21.510629653930664,
      "learning_rate": 0.00019858663412558188,
      "loss": 6.9137,
      "step": 49440
    },
    {
      "epoch": 5.9506618531889295,
      "grad_norm": 240.3212432861328,
      "learning_rate": 0.00019858599671065456,
      "loss": 6.655,
      "step": 49450
    },
    {
      "epoch": 5.951865222623345,
      "grad_norm": 53.579219818115234,
      "learning_rate": 0.000198585359153049,
      "loss": 6.7256,
      "step": 49460
    },
    {
      "epoch": 5.953068592057762,
      "grad_norm": 263.5654602050781,
      "learning_rate": 0.00019858472145276607,
      "loss": 6.8495,
      "step": 49470
    },
    {
      "epoch": 5.954271961492178,
      "grad_norm": 56.07474899291992,
      "learning_rate": 0.00019858408360980668,
      "loss": 7.0469,
      "step": 49480
    },
    {
      "epoch": 5.955475330926594,
      "grad_norm": 9.706421852111816,
      "learning_rate": 0.0001985834456241718,
      "loss": 7.0029,
      "step": 49490
    },
    {
      "epoch": 5.956678700361011,
      "grad_norm": 21.957490921020508,
      "learning_rate": 0.00019858280749586235,
      "loss": 6.8408,
      "step": 49500
    },
    {
      "epoch": 5.957882069795427,
      "grad_norm": 28.383865356445312,
      "learning_rate": 0.00019858216922487922,
      "loss": 6.9464,
      "step": 49510
    },
    {
      "epoch": 5.959085439229844,
      "grad_norm": 38.81053161621094,
      "learning_rate": 0.00019858153081122337,
      "loss": 6.9604,
      "step": 49520
    },
    {
      "epoch": 5.96028880866426,
      "grad_norm": 42.365753173828125,
      "learning_rate": 0.00019858089225489569,
      "loss": 6.874,
      "step": 49530
    },
    {
      "epoch": 5.961492178098676,
      "grad_norm": 85.8763656616211,
      "learning_rate": 0.0001985802535558971,
      "loss": 6.7034,
      "step": 49540
    },
    {
      "epoch": 5.962695547533093,
      "grad_norm": 37.90697479248047,
      "learning_rate": 0.0001985796147142286,
      "loss": 6.8336,
      "step": 49550
    },
    {
      "epoch": 5.963898916967509,
      "grad_norm": 43.58564376831055,
      "learning_rate": 0.00019857897572989104,
      "loss": 6.8717,
      "step": 49560
    },
    {
      "epoch": 5.965102286401925,
      "grad_norm": 60.161529541015625,
      "learning_rate": 0.00019857833660288533,
      "loss": 6.8313,
      "step": 49570
    },
    {
      "epoch": 5.966305655836342,
      "grad_norm": 68.186767578125,
      "learning_rate": 0.00019857769733321246,
      "loss": 6.7628,
      "step": 49580
    },
    {
      "epoch": 5.967509025270758,
      "grad_norm": 444.3592529296875,
      "learning_rate": 0.0001985770579208733,
      "loss": 6.7427,
      "step": 49590
    },
    {
      "epoch": 5.968712394705174,
      "grad_norm": 143.87109375,
      "learning_rate": 0.0001985764183658688,
      "loss": 6.8694,
      "step": 49600
    },
    {
      "epoch": 5.969915764139591,
      "grad_norm": 364.8321838378906,
      "learning_rate": 0.00019857577866819992,
      "loss": 6.8645,
      "step": 49610
    },
    {
      "epoch": 5.971119133574007,
      "grad_norm": 589.8414916992188,
      "learning_rate": 0.00019857513882786754,
      "loss": 7.0478,
      "step": 49620
    },
    {
      "epoch": 5.9723225030084235,
      "grad_norm": 1590.6165771484375,
      "learning_rate": 0.00019857449884487258,
      "loss": 7.4665,
      "step": 49630
    },
    {
      "epoch": 5.97352587244284,
      "grad_norm": 58.226348876953125,
      "learning_rate": 0.000198573858719216,
      "loss": 7.8009,
      "step": 49640
    },
    {
      "epoch": 5.974729241877256,
      "grad_norm": 287.2764892578125,
      "learning_rate": 0.00019857321845089868,
      "loss": 7.7155,
      "step": 49650
    },
    {
      "epoch": 5.975932611311673,
      "grad_norm": 37.035423278808594,
      "learning_rate": 0.0001985725780399216,
      "loss": 7.5602,
      "step": 49660
    },
    {
      "epoch": 5.977135980746089,
      "grad_norm": 8.563516616821289,
      "learning_rate": 0.00019857193748628566,
      "loss": 7.5437,
      "step": 49670
    },
    {
      "epoch": 5.978339350180505,
      "grad_norm": 6.727244853973389,
      "learning_rate": 0.00019857129678999181,
      "loss": 7.6304,
      "step": 49680
    },
    {
      "epoch": 5.979542719614922,
      "grad_norm": 17.694631576538086,
      "learning_rate": 0.00019857065595104093,
      "loss": 7.7271,
      "step": 49690
    },
    {
      "epoch": 5.9807460890493385,
      "grad_norm": 17.664844512939453,
      "learning_rate": 0.00019857001496943397,
      "loss": 7.6097,
      "step": 49700
    },
    {
      "epoch": 5.981949458483754,
      "grad_norm": 26.714693069458008,
      "learning_rate": 0.00019856937384517186,
      "loss": 7.4509,
      "step": 49710
    },
    {
      "epoch": 5.983152827918171,
      "grad_norm": 8.991143226623535,
      "learning_rate": 0.00019856873257825556,
      "loss": 7.3315,
      "step": 49720
    },
    {
      "epoch": 5.984356197352588,
      "grad_norm": 6.968205451965332,
      "learning_rate": 0.00019856809116868592,
      "loss": 7.6477,
      "step": 49730
    },
    {
      "epoch": 5.985559566787003,
      "grad_norm": 14.680328369140625,
      "learning_rate": 0.00019856744961646396,
      "loss": 7.6382,
      "step": 49740
    },
    {
      "epoch": 5.98676293622142,
      "grad_norm": 29.047746658325195,
      "learning_rate": 0.00019856680792159056,
      "loss": 7.4855,
      "step": 49750
    },
    {
      "epoch": 5.987966305655837,
      "grad_norm": 5.303946018218994,
      "learning_rate": 0.00019856616608406665,
      "loss": 7.6234,
      "step": 49760
    },
    {
      "epoch": 5.9891696750902526,
      "grad_norm": 3.9366724491119385,
      "learning_rate": 0.00019856552410389315,
      "loss": 7.4582,
      "step": 49770
    },
    {
      "epoch": 5.990373044524669,
      "grad_norm": 5.2910919189453125,
      "learning_rate": 0.000198564881981071,
      "loss": 7.3924,
      "step": 49780
    },
    {
      "epoch": 5.991576413959086,
      "grad_norm": 28.09467124938965,
      "learning_rate": 0.00019856423971560113,
      "loss": 7.3908,
      "step": 49790
    },
    {
      "epoch": 5.992779783393502,
      "grad_norm": 1442.5479736328125,
      "learning_rate": 0.0001985635973074845,
      "loss": 7.6586,
      "step": 49800
    },
    {
      "epoch": 5.993983152827918,
      "grad_norm": 34.08098220825195,
      "learning_rate": 0.000198562954756722,
      "loss": 7.5497,
      "step": 49810
    },
    {
      "epoch": 5.995186522262334,
      "grad_norm": 94.67636108398438,
      "learning_rate": 0.00019856231206331456,
      "loss": 7.4068,
      "step": 49820
    },
    {
      "epoch": 5.996389891696751,
      "grad_norm": 40.775840759277344,
      "learning_rate": 0.0001985616692272631,
      "loss": 7.2559,
      "step": 49830
    },
    {
      "epoch": 5.9975932611311675,
      "grad_norm": 149.15672302246094,
      "learning_rate": 0.0001985610262485686,
      "loss": 7.144,
      "step": 49840
    },
    {
      "epoch": 5.998796630565583,
      "grad_norm": 458.82275390625,
      "learning_rate": 0.00019856038312723192,
      "loss": 8.0931,
      "step": 49850
    },
    {
      "epoch": 6.0,
      "grad_norm": 65.29741668701172,
      "learning_rate": 0.00019855973986325408,
      "loss": 7.7783,
      "step": 49860
    },
    {
      "epoch": 6.0,
      "eval_loss": 8.033663749694824,
      "eval_runtime": 119.4338,
      "eval_samples_per_second": 61.85,
      "eval_steps_per_second": 7.737,
      "step": 49860
    },
    {
      "epoch": 6.001203369434417,
      "grad_norm": 41.65032958984375,
      "learning_rate": 0.00019855909645663595,
      "loss": 7.717,
      "step": 49870
    },
    {
      "epoch": 6.0024067388688325,
      "grad_norm": 168.38162231445312,
      "learning_rate": 0.00019855845290737848,
      "loss": 7.577,
      "step": 49880
    },
    {
      "epoch": 6.003610108303249,
      "grad_norm": 43.669158935546875,
      "learning_rate": 0.0001985578092154826,
      "loss": 7.3283,
      "step": 49890
    },
    {
      "epoch": 6.004813477737666,
      "grad_norm": 128.16262817382812,
      "learning_rate": 0.00019855716538094923,
      "loss": 7.3191,
      "step": 49900
    },
    {
      "epoch": 6.006016847172082,
      "grad_norm": 82.70264434814453,
      "learning_rate": 0.00019855652140377927,
      "loss": 7.3742,
      "step": 49910
    },
    {
      "epoch": 6.007220216606498,
      "grad_norm": 107.43938446044922,
      "learning_rate": 0.00019855587728397374,
      "loss": 7.2678,
      "step": 49920
    },
    {
      "epoch": 6.008423586040915,
      "grad_norm": 199.47250366210938,
      "learning_rate": 0.0001985552330215335,
      "loss": 7.0853,
      "step": 49930
    },
    {
      "epoch": 6.009626955475331,
      "grad_norm": 44.25736618041992,
      "learning_rate": 0.0001985545886164595,
      "loss": 7.1751,
      "step": 49940
    },
    {
      "epoch": 6.0108303249097474,
      "grad_norm": 29.525846481323242,
      "learning_rate": 0.0001985539440687527,
      "loss": 7.16,
      "step": 49950
    },
    {
      "epoch": 6.012033694344163,
      "grad_norm": 35.5792350769043,
      "learning_rate": 0.000198553299378414,
      "loss": 7.0683,
      "step": 49960
    },
    {
      "epoch": 6.01323706377858,
      "grad_norm": 155.37547302246094,
      "learning_rate": 0.00019855265454544433,
      "loss": 7.0179,
      "step": 49970
    },
    {
      "epoch": 6.014440433212997,
      "grad_norm": 24.824987411499023,
      "learning_rate": 0.00019855200956984465,
      "loss": 6.9366,
      "step": 49980
    },
    {
      "epoch": 6.015643802647412,
      "grad_norm": 33.48954772949219,
      "learning_rate": 0.00019855136445161588,
      "loss": 7.1316,
      "step": 49990
    },
    {
      "epoch": 6.016847172081829,
      "grad_norm": 51.579593658447266,
      "learning_rate": 0.00019855071919075895,
      "loss": 6.9995,
      "step": 50000
    },
    {
      "epoch": 6.018050541516246,
      "grad_norm": 21.164052963256836,
      "learning_rate": 0.00019855007378727482,
      "loss": 6.96,
      "step": 50010
    },
    {
      "epoch": 6.0192539109506615,
      "grad_norm": 26.542234420776367,
      "learning_rate": 0.00019854942824116436,
      "loss": 6.9492,
      "step": 50020
    },
    {
      "epoch": 6.020457280385078,
      "grad_norm": 21.792335510253906,
      "learning_rate": 0.0001985487825524286,
      "loss": 6.8295,
      "step": 50030
    },
    {
      "epoch": 6.021660649819495,
      "grad_norm": 83.90550231933594,
      "learning_rate": 0.00019854813672106838,
      "loss": 6.746,
      "step": 50040
    },
    {
      "epoch": 6.022864019253911,
      "grad_norm": 28.177583694458008,
      "learning_rate": 0.0001985474907470847,
      "loss": 6.6812,
      "step": 50050
    },
    {
      "epoch": 6.024067388688327,
      "grad_norm": 100.0069808959961,
      "learning_rate": 0.00019854684463047843,
      "loss": 6.6394,
      "step": 50060
    },
    {
      "epoch": 6.025270758122744,
      "grad_norm": 17.259765625,
      "learning_rate": 0.00019854619837125057,
      "loss": 6.7483,
      "step": 50070
    },
    {
      "epoch": 6.02647412755716,
      "grad_norm": 41.67375564575195,
      "learning_rate": 0.00019854555196940202,
      "loss": 6.8623,
      "step": 50080
    },
    {
      "epoch": 6.0276774969915765,
      "grad_norm": 38.76065444946289,
      "learning_rate": 0.00019854490542493375,
      "loss": 6.8325,
      "step": 50090
    },
    {
      "epoch": 6.028880866425993,
      "grad_norm": 20.695940017700195,
      "learning_rate": 0.00019854425873784667,
      "loss": 7.3002,
      "step": 50100
    },
    {
      "epoch": 6.030084235860409,
      "grad_norm": 34.97782516479492,
      "learning_rate": 0.00019854361190814172,
      "loss": 7.2257,
      "step": 50110
    },
    {
      "epoch": 6.031287605294826,
      "grad_norm": 43.32956314086914,
      "learning_rate": 0.00019854296493581984,
      "loss": 7.3992,
      "step": 50120
    },
    {
      "epoch": 6.0324909747292415,
      "grad_norm": 124.94662475585938,
      "learning_rate": 0.00019854231782088193,
      "loss": 6.9233,
      "step": 50130
    },
    {
      "epoch": 6.033694344163658,
      "grad_norm": 356.6945495605469,
      "learning_rate": 0.00019854167056332895,
      "loss": 7.1386,
      "step": 50140
    },
    {
      "epoch": 6.034897713598075,
      "grad_norm": 1409.9876708984375,
      "learning_rate": 0.00019854102316316187,
      "loss": 7.4419,
      "step": 50150
    },
    {
      "epoch": 6.036101083032491,
      "grad_norm": 456.3817138671875,
      "learning_rate": 0.0001985403756203816,
      "loss": 7.7544,
      "step": 50160
    },
    {
      "epoch": 6.037304452466907,
      "grad_norm": 134.14495849609375,
      "learning_rate": 0.00019853972793498908,
      "loss": 8.5643,
      "step": 50170
    },
    {
      "epoch": 6.038507821901324,
      "grad_norm": 103.23419189453125,
      "learning_rate": 0.00019853908010698522,
      "loss": 7.7766,
      "step": 50180
    },
    {
      "epoch": 6.03971119133574,
      "grad_norm": 145.68394470214844,
      "learning_rate": 0.00019853843213637101,
      "loss": 7.4989,
      "step": 50190
    },
    {
      "epoch": 6.040914560770156,
      "grad_norm": 10.152735710144043,
      "learning_rate": 0.00019853778402314736,
      "loss": 7.5882,
      "step": 50200
    },
    {
      "epoch": 6.042117930204573,
      "grad_norm": 15.388716697692871,
      "learning_rate": 0.00019853713576731518,
      "loss": 7.5852,
      "step": 50210
    },
    {
      "epoch": 6.043321299638989,
      "grad_norm": 10.143814086914062,
      "learning_rate": 0.00019853648736887545,
      "loss": 7.5448,
      "step": 50220
    },
    {
      "epoch": 6.044524669073406,
      "grad_norm": 76.52864837646484,
      "learning_rate": 0.00019853583882782912,
      "loss": 7.5189,
      "step": 50230
    },
    {
      "epoch": 6.045728038507822,
      "grad_norm": 675.9168701171875,
      "learning_rate": 0.00019853519014417708,
      "loss": 7.5494,
      "step": 50240
    },
    {
      "epoch": 6.046931407942238,
      "grad_norm": 22.926326751708984,
      "learning_rate": 0.00019853454131792028,
      "loss": 7.8772,
      "step": 50250
    },
    {
      "epoch": 6.048134777376655,
      "grad_norm": 57.346290588378906,
      "learning_rate": 0.0001985338923490597,
      "loss": 7.6679,
      "step": 50260
    },
    {
      "epoch": 6.049338146811071,
      "grad_norm": 14.811667442321777,
      "learning_rate": 0.00019853324323759622,
      "loss": 7.5137,
      "step": 50270
    },
    {
      "epoch": 6.050541516245487,
      "grad_norm": 4.689547061920166,
      "learning_rate": 0.00019853259398353083,
      "loss": 7.5534,
      "step": 50280
    },
    {
      "epoch": 6.051744885679904,
      "grad_norm": 27.728551864624023,
      "learning_rate": 0.00019853194458686446,
      "loss": 7.3619,
      "step": 50290
    },
    {
      "epoch": 6.05294825511432,
      "grad_norm": 30.655689239501953,
      "learning_rate": 0.000198531295047598,
      "loss": 7.3617,
      "step": 50300
    },
    {
      "epoch": 6.054151624548736,
      "grad_norm": 48.561012268066406,
      "learning_rate": 0.00019853064536573246,
      "loss": 7.4353,
      "step": 50310
    },
    {
      "epoch": 6.055354993983153,
      "grad_norm": 155.85556030273438,
      "learning_rate": 0.00019852999554126878,
      "loss": 7.241,
      "step": 50320
    },
    {
      "epoch": 6.056558363417569,
      "grad_norm": 73.67689514160156,
      "learning_rate": 0.0001985293455742078,
      "loss": 7.2979,
      "step": 50330
    },
    {
      "epoch": 6.0577617328519855,
      "grad_norm": 90.59698486328125,
      "learning_rate": 0.00019852869546455057,
      "loss": 7.9339,
      "step": 50340
    },
    {
      "epoch": 6.058965102286402,
      "grad_norm": 415.32452392578125,
      "learning_rate": 0.000198528045212298,
      "loss": 7.5561,
      "step": 50350
    },
    {
      "epoch": 6.060168471720818,
      "grad_norm": 161.16436767578125,
      "learning_rate": 0.00019852739481745102,
      "loss": 7.2972,
      "step": 50360
    },
    {
      "epoch": 6.061371841155235,
      "grad_norm": 313.7480163574219,
      "learning_rate": 0.00019852674428001054,
      "loss": 7.3981,
      "step": 50370
    },
    {
      "epoch": 6.062575210589651,
      "grad_norm": 105.19052124023438,
      "learning_rate": 0.00019852609359997757,
      "loss": 7.4124,
      "step": 50380
    },
    {
      "epoch": 6.063778580024067,
      "grad_norm": 53.43791198730469,
      "learning_rate": 0.00019852544277735305,
      "loss": 7.2467,
      "step": 50390
    },
    {
      "epoch": 6.064981949458484,
      "grad_norm": 1362.7615966796875,
      "learning_rate": 0.0001985247918121378,
      "loss": 7.4411,
      "step": 50400
    },
    {
      "epoch": 6.0661853188929005,
      "grad_norm": 50.751976013183594,
      "learning_rate": 0.00019852414070433292,
      "loss": 7.5127,
      "step": 50410
    },
    {
      "epoch": 6.067388688327316,
      "grad_norm": 124.1102523803711,
      "learning_rate": 0.00019852348945393925,
      "loss": 7.4167,
      "step": 50420
    },
    {
      "epoch": 6.068592057761733,
      "grad_norm": 63.0703125,
      "learning_rate": 0.00019852283806095778,
      "loss": 7.4491,
      "step": 50430
    },
    {
      "epoch": 6.06979542719615,
      "grad_norm": 9.66748332977295,
      "learning_rate": 0.00019852218652538945,
      "loss": 7.4479,
      "step": 50440
    },
    {
      "epoch": 6.070998796630565,
      "grad_norm": 4.118216514587402,
      "learning_rate": 0.00019852153484723516,
      "loss": 7.5609,
      "step": 50450
    },
    {
      "epoch": 6.072202166064982,
      "grad_norm": 7.636483192443848,
      "learning_rate": 0.0001985208830264959,
      "loss": 7.5435,
      "step": 50460
    },
    {
      "epoch": 6.073405535499398,
      "grad_norm": 10.706920623779297,
      "learning_rate": 0.0001985202310631726,
      "loss": 7.2436,
      "step": 50470
    },
    {
      "epoch": 6.074608904933815,
      "grad_norm": 5.278939247131348,
      "learning_rate": 0.00019851957895726622,
      "loss": 7.2387,
      "step": 50480
    },
    {
      "epoch": 6.075812274368231,
      "grad_norm": 18.34695816040039,
      "learning_rate": 0.00019851892670877765,
      "loss": 7.0852,
      "step": 50490
    },
    {
      "epoch": 6.077015643802647,
      "grad_norm": 266.5223693847656,
      "learning_rate": 0.0001985182743177079,
      "loss": 7.1159,
      "step": 50500
    },
    {
      "epoch": 6.078219013237064,
      "grad_norm": 18.489425659179688,
      "learning_rate": 0.00019851762178405786,
      "loss": 7.0828,
      "step": 50510
    },
    {
      "epoch": 6.07942238267148,
      "grad_norm": 8.026512145996094,
      "learning_rate": 0.00019851696910782853,
      "loss": 6.9848,
      "step": 50520
    },
    {
      "epoch": 6.080625752105896,
      "grad_norm": 7.6296610832214355,
      "learning_rate": 0.0001985163162890208,
      "loss": 6.9406,
      "step": 50530
    },
    {
      "epoch": 6.081829121540313,
      "grad_norm": 8.11501693725586,
      "learning_rate": 0.00019851566332763562,
      "loss": 6.8414,
      "step": 50540
    },
    {
      "epoch": 6.0830324909747295,
      "grad_norm": 15.515129089355469,
      "learning_rate": 0.000198515010223674,
      "loss": 6.9176,
      "step": 50550
    },
    {
      "epoch": 6.084235860409145,
      "grad_norm": 5.048215866088867,
      "learning_rate": 0.0001985143569771368,
      "loss": 6.7867,
      "step": 50560
    },
    {
      "epoch": 6.085439229843562,
      "grad_norm": 12.029462814331055,
      "learning_rate": 0.00019851370358802502,
      "loss": 6.7815,
      "step": 50570
    },
    {
      "epoch": 6.086642599277979,
      "grad_norm": 11.015639305114746,
      "learning_rate": 0.00019851305005633957,
      "loss": 6.6205,
      "step": 50580
    },
    {
      "epoch": 6.0878459687123945,
      "grad_norm": 9.940922737121582,
      "learning_rate": 0.00019851239638208143,
      "loss": 6.6384,
      "step": 50590
    },
    {
      "epoch": 6.089049338146811,
      "grad_norm": 12.857138633728027,
      "learning_rate": 0.00019851174256525154,
      "loss": 6.7911,
      "step": 50600
    },
    {
      "epoch": 6.090252707581228,
      "grad_norm": 14.697220802307129,
      "learning_rate": 0.0001985110886058508,
      "loss": 6.6931,
      "step": 50610
    },
    {
      "epoch": 6.091456077015644,
      "grad_norm": 6.209646224975586,
      "learning_rate": 0.00019851043450388022,
      "loss": 6.722,
      "step": 50620
    },
    {
      "epoch": 6.09265944645006,
      "grad_norm": 4.238023281097412,
      "learning_rate": 0.00019850978025934073,
      "loss": 6.6767,
      "step": 50630
    },
    {
      "epoch": 6.093862815884476,
      "grad_norm": 12.614048957824707,
      "learning_rate": 0.00019850912587223324,
      "loss": 6.633,
      "step": 50640
    },
    {
      "epoch": 6.095066185318893,
      "grad_norm": 6.149070739746094,
      "learning_rate": 0.00019850847134255875,
      "loss": 6.5404,
      "step": 50650
    },
    {
      "epoch": 6.0962695547533094,
      "grad_norm": 7.979625701904297,
      "learning_rate": 0.00019850781667031818,
      "loss": 6.5466,
      "step": 50660
    },
    {
      "epoch": 6.097472924187725,
      "grad_norm": 7.836061477661133,
      "learning_rate": 0.00019850716185551245,
      "loss": 6.4531,
      "step": 50670
    },
    {
      "epoch": 6.098676293622142,
      "grad_norm": 9.258145332336426,
      "learning_rate": 0.00019850650689814256,
      "loss": 6.6148,
      "step": 50680
    },
    {
      "epoch": 6.099879663056559,
      "grad_norm": 7.844076156616211,
      "learning_rate": 0.0001985058517982094,
      "loss": 6.694,
      "step": 50690
    },
    {
      "epoch": 6.101083032490974,
      "grad_norm": 25.579376220703125,
      "learning_rate": 0.000198505196555714,
      "loss": 6.5015,
      "step": 50700
    },
    {
      "epoch": 6.102286401925391,
      "grad_norm": 42.913421630859375,
      "learning_rate": 0.00019850454117065723,
      "loss": 6.5634,
      "step": 50710
    },
    {
      "epoch": 6.103489771359808,
      "grad_norm": 29.641809463500977,
      "learning_rate": 0.00019850388564304007,
      "loss": 6.5993,
      "step": 50720
    },
    {
      "epoch": 6.1046931407942235,
      "grad_norm": 169.4796142578125,
      "learning_rate": 0.00019850322997286347,
      "loss": 6.5776,
      "step": 50730
    },
    {
      "epoch": 6.10589651022864,
      "grad_norm": 235.8825225830078,
      "learning_rate": 0.0001985025741601284,
      "loss": 7.1905,
      "step": 50740
    },
    {
      "epoch": 6.107099879663057,
      "grad_norm": 408.1968078613281,
      "learning_rate": 0.00019850191820483578,
      "loss": 7.678,
      "step": 50750
    },
    {
      "epoch": 6.108303249097473,
      "grad_norm": 86.42868041992188,
      "learning_rate": 0.00019850126210698658,
      "loss": 7.5839,
      "step": 50760
    },
    {
      "epoch": 6.109506618531889,
      "grad_norm": 27.87179183959961,
      "learning_rate": 0.00019850060586658168,
      "loss": 7.1222,
      "step": 50770
    },
    {
      "epoch": 6.110709987966306,
      "grad_norm": 18.759296417236328,
      "learning_rate": 0.0001984999494836221,
      "loss": 6.8401,
      "step": 50780
    },
    {
      "epoch": 6.111913357400722,
      "grad_norm": 23.341739654541016,
      "learning_rate": 0.0001984992929581088,
      "loss": 6.6808,
      "step": 50790
    },
    {
      "epoch": 6.1131167268351385,
      "grad_norm": 12.954388618469238,
      "learning_rate": 0.00019849863629004268,
      "loss": 6.6503,
      "step": 50800
    },
    {
      "epoch": 6.114320096269555,
      "grad_norm": 9.701177597045898,
      "learning_rate": 0.00019849797947942474,
      "loss": 6.613,
      "step": 50810
    },
    {
      "epoch": 6.115523465703971,
      "grad_norm": 8.269774436950684,
      "learning_rate": 0.00019849732252625587,
      "loss": 6.5937,
      "step": 50820
    },
    {
      "epoch": 6.116726835138388,
      "grad_norm": 87.7800521850586,
      "learning_rate": 0.00019849666543053708,
      "loss": 6.5692,
      "step": 50830
    },
    {
      "epoch": 6.1179302045728035,
      "grad_norm": 752.7943115234375,
      "learning_rate": 0.0001984960081922693,
      "loss": 6.763,
      "step": 50840
    },
    {
      "epoch": 6.11913357400722,
      "grad_norm": 102.5551528930664,
      "learning_rate": 0.00019849535081145348,
      "loss": 6.7909,
      "step": 50850
    },
    {
      "epoch": 6.120336943441637,
      "grad_norm": 83.97057342529297,
      "learning_rate": 0.00019849469328809055,
      "loss": 6.5917,
      "step": 50860
    },
    {
      "epoch": 6.121540312876053,
      "grad_norm": 73.70748901367188,
      "learning_rate": 0.00019849403562218152,
      "loss": 6.5882,
      "step": 50870
    },
    {
      "epoch": 6.122743682310469,
      "grad_norm": 50.015472412109375,
      "learning_rate": 0.00019849337781372724,
      "loss": 6.65,
      "step": 50880
    },
    {
      "epoch": 6.123947051744886,
      "grad_norm": 69.71155548095703,
      "learning_rate": 0.0001984927198627288,
      "loss": 6.6931,
      "step": 50890
    },
    {
      "epoch": 6.125150421179302,
      "grad_norm": 150.67527770996094,
      "learning_rate": 0.00019849206176918702,
      "loss": 6.8684,
      "step": 50900
    },
    {
      "epoch": 6.126353790613718,
      "grad_norm": 216.06704711914062,
      "learning_rate": 0.00019849140353310292,
      "loss": 7.6943,
      "step": 50910
    },
    {
      "epoch": 6.127557160048135,
      "grad_norm": 6.53950309753418,
      "learning_rate": 0.00019849074515447746,
      "loss": 7.7072,
      "step": 50920
    },
    {
      "epoch": 6.128760529482551,
      "grad_norm": 6.397497653961182,
      "learning_rate": 0.00019849008663331156,
      "loss": 7.4985,
      "step": 50930
    },
    {
      "epoch": 6.129963898916968,
      "grad_norm": 15.183296203613281,
      "learning_rate": 0.00019848942796960622,
      "loss": 7.2878,
      "step": 50940
    },
    {
      "epoch": 6.131167268351384,
      "grad_norm": 3.0410451889038086,
      "learning_rate": 0.0001984887691633623,
      "loss": 7.2639,
      "step": 50950
    },
    {
      "epoch": 6.1323706377858,
      "grad_norm": 1423.0872802734375,
      "learning_rate": 0.00019848811021458087,
      "loss": 7.554,
      "step": 50960
    },
    {
      "epoch": 6.133574007220217,
      "grad_norm": 202.90269470214844,
      "learning_rate": 0.00019848745112326278,
      "loss": 7.8729,
      "step": 50970
    },
    {
      "epoch": 6.1347773766546325,
      "grad_norm": 72.13835144042969,
      "learning_rate": 0.00019848679188940908,
      "loss": 7.7909,
      "step": 50980
    },
    {
      "epoch": 6.135980746089049,
      "grad_norm": 779.5035400390625,
      "learning_rate": 0.00019848613251302065,
      "loss": 7.6156,
      "step": 50990
    },
    {
      "epoch": 6.137184115523466,
      "grad_norm": 31.454639434814453,
      "learning_rate": 0.00019848547299409849,
      "loss": 7.65,
      "step": 51000
    },
    {
      "epoch": 6.138387484957882,
      "grad_norm": 47.65021896362305,
      "learning_rate": 0.00019848481333264352,
      "loss": 7.5042,
      "step": 51010
    },
    {
      "epoch": 6.139590854392298,
      "grad_norm": 10.336441993713379,
      "learning_rate": 0.0001984841535286567,
      "loss": 7.4727,
      "step": 51020
    },
    {
      "epoch": 6.140794223826715,
      "grad_norm": 4.083159446716309,
      "learning_rate": 0.00019848349358213906,
      "loss": 7.2821,
      "step": 51030
    },
    {
      "epoch": 6.141997593261131,
      "grad_norm": 11.308076858520508,
      "learning_rate": 0.0001984828334930914,
      "loss": 7.1853,
      "step": 51040
    },
    {
      "epoch": 6.1432009626955475,
      "grad_norm": 18.26197624206543,
      "learning_rate": 0.00019848217326151482,
      "loss": 7.2388,
      "step": 51050
    },
    {
      "epoch": 6.144404332129964,
      "grad_norm": 2207.310791015625,
      "learning_rate": 0.0001984815128874102,
      "loss": 7.2516,
      "step": 51060
    },
    {
      "epoch": 6.14560770156438,
      "grad_norm": 49.99858856201172,
      "learning_rate": 0.00019848085237077854,
      "loss": 7.4258,
      "step": 51070
    },
    {
      "epoch": 6.146811070998797,
      "grad_norm": 151.1298828125,
      "learning_rate": 0.00019848019171162074,
      "loss": 7.5135,
      "step": 51080
    },
    {
      "epoch": 6.148014440433213,
      "grad_norm": 11.325151443481445,
      "learning_rate": 0.00019847953090993778,
      "loss": 7.8672,
      "step": 51090
    },
    {
      "epoch": 6.149217809867629,
      "grad_norm": 9.93419075012207,
      "learning_rate": 0.00019847886996573067,
      "loss": 7.4075,
      "step": 51100
    },
    {
      "epoch": 6.150421179302046,
      "grad_norm": 11.671231269836426,
      "learning_rate": 0.0001984782088790003,
      "loss": 7.2521,
      "step": 51110
    },
    {
      "epoch": 6.1516245487364625,
      "grad_norm": 8.38558292388916,
      "learning_rate": 0.00019847754764974764,
      "loss": 7.0416,
      "step": 51120
    },
    {
      "epoch": 6.152827918170878,
      "grad_norm": 10.745943069458008,
      "learning_rate": 0.00019847688627797367,
      "loss": 7.0191,
      "step": 51130
    },
    {
      "epoch": 6.154031287605295,
      "grad_norm": 9.660490989685059,
      "learning_rate": 0.00019847622476367932,
      "loss": 6.8097,
      "step": 51140
    },
    {
      "epoch": 6.155234657039712,
      "grad_norm": 7.7755045890808105,
      "learning_rate": 0.00019847556310686557,
      "loss": 6.7182,
      "step": 51150
    },
    {
      "epoch": 6.156438026474127,
      "grad_norm": 374.6153869628906,
      "learning_rate": 0.00019847490130753338,
      "loss": 6.8375,
      "step": 51160
    },
    {
      "epoch": 6.157641395908544,
      "grad_norm": 253.876953125,
      "learning_rate": 0.00019847423936568366,
      "loss": 7.1427,
      "step": 51170
    },
    {
      "epoch": 6.15884476534296,
      "grad_norm": 111.31069946289062,
      "learning_rate": 0.0001984735772813174,
      "loss": 7.0288,
      "step": 51180
    },
    {
      "epoch": 6.160048134777377,
      "grad_norm": 57.93216323852539,
      "learning_rate": 0.0001984729150544356,
      "loss": 7.2709,
      "step": 51190
    },
    {
      "epoch": 6.161251504211793,
      "grad_norm": 9.835779190063477,
      "learning_rate": 0.00019847225268503915,
      "loss": 7.0617,
      "step": 51200
    },
    {
      "epoch": 6.162454873646209,
      "grad_norm": 49.00759506225586,
      "learning_rate": 0.00019847159017312907,
      "loss": 6.8884,
      "step": 51210
    },
    {
      "epoch": 6.163658243080626,
      "grad_norm": 98.64542388916016,
      "learning_rate": 0.00019847092751870625,
      "loss": 7.3326,
      "step": 51220
    },
    {
      "epoch": 6.164861612515042,
      "grad_norm": 27.575584411621094,
      "learning_rate": 0.00019847026472177173,
      "loss": 7.4802,
      "step": 51230
    },
    {
      "epoch": 6.166064981949458,
      "grad_norm": 10.02219295501709,
      "learning_rate": 0.0001984696017823264,
      "loss": 7.3218,
      "step": 51240
    },
    {
      "epoch": 6.167268351383875,
      "grad_norm": 18.55671501159668,
      "learning_rate": 0.00019846893870037122,
      "loss": 7.2338,
      "step": 51250
    },
    {
      "epoch": 6.1684717208182915,
      "grad_norm": 14.191424369812012,
      "learning_rate": 0.0001984682754759072,
      "loss": 7.009,
      "step": 51260
    },
    {
      "epoch": 6.169675090252707,
      "grad_norm": 8.056644439697266,
      "learning_rate": 0.00019846761210893528,
      "loss": 6.9959,
      "step": 51270
    },
    {
      "epoch": 6.170878459687124,
      "grad_norm": 54.10923767089844,
      "learning_rate": 0.0001984669485994564,
      "loss": 6.816,
      "step": 51280
    },
    {
      "epoch": 6.172081829121541,
      "grad_norm": 29.607463836669922,
      "learning_rate": 0.00019846628494747155,
      "loss": 6.8708,
      "step": 51290
    },
    {
      "epoch": 6.1732851985559565,
      "grad_norm": 137.29750061035156,
      "learning_rate": 0.00019846562115298166,
      "loss": 6.7867,
      "step": 51300
    },
    {
      "epoch": 6.174488567990373,
      "grad_norm": 175.81932067871094,
      "learning_rate": 0.0001984649572159877,
      "loss": 7.2438,
      "step": 51310
    },
    {
      "epoch": 6.17569193742479,
      "grad_norm": 91.62136840820312,
      "learning_rate": 0.00019846429313649067,
      "loss": 7.5339,
      "step": 51320
    },
    {
      "epoch": 6.176895306859206,
      "grad_norm": 129.83792114257812,
      "learning_rate": 0.0001984636289144915,
      "loss": 7.4754,
      "step": 51330
    },
    {
      "epoch": 6.178098676293622,
      "grad_norm": 205.80770874023438,
      "learning_rate": 0.0001984629645499911,
      "loss": 7.3861,
      "step": 51340
    },
    {
      "epoch": 6.179302045728038,
      "grad_norm": 145.01470947265625,
      "learning_rate": 0.0001984623000429905,
      "loss": 7.3334,
      "step": 51350
    },
    {
      "epoch": 6.180505415162455,
      "grad_norm": 780.2528686523438,
      "learning_rate": 0.00019846163539349065,
      "loss": 7.5486,
      "step": 51360
    },
    {
      "epoch": 6.1817087845968715,
      "grad_norm": 12.88818359375,
      "learning_rate": 0.0001984609706014925,
      "loss": 8.5478,
      "step": 51370
    },
    {
      "epoch": 6.182912154031287,
      "grad_norm": 8.204535484313965,
      "learning_rate": 0.00019846030566699702,
      "loss": 7.3626,
      "step": 51380
    },
    {
      "epoch": 6.184115523465704,
      "grad_norm": 50.635101318359375,
      "learning_rate": 0.00019845964059000517,
      "loss": 7.0688,
      "step": 51390
    },
    {
      "epoch": 6.185318892900121,
      "grad_norm": 84.38154602050781,
      "learning_rate": 0.00019845897537051793,
      "loss": 7.188,
      "step": 51400
    },
    {
      "epoch": 6.186522262334536,
      "grad_norm": 31.963542938232422,
      "learning_rate": 0.00019845831000853622,
      "loss": 7.0432,
      "step": 51410
    },
    {
      "epoch": 6.187725631768953,
      "grad_norm": 14.625631332397461,
      "learning_rate": 0.00019845764450406104,
      "loss": 7.028,
      "step": 51420
    },
    {
      "epoch": 6.18892900120337,
      "grad_norm": 19.176286697387695,
      "learning_rate": 0.00019845697885709332,
      "loss": 7.1465,
      "step": 51430
    },
    {
      "epoch": 6.1901323706377855,
      "grad_norm": 198.12120056152344,
      "learning_rate": 0.00019845631306763406,
      "loss": 7.3459,
      "step": 51440
    },
    {
      "epoch": 6.191335740072202,
      "grad_norm": 8.463104248046875,
      "learning_rate": 0.00019845564713568417,
      "loss": 7.2657,
      "step": 51450
    },
    {
      "epoch": 6.192539109506619,
      "grad_norm": 28.150205612182617,
      "learning_rate": 0.0001984549810612447,
      "loss": 7.0837,
      "step": 51460
    },
    {
      "epoch": 6.193742478941035,
      "grad_norm": 53.76273727416992,
      "learning_rate": 0.00019845431484431653,
      "loss": 7.1347,
      "step": 51470
    },
    {
      "epoch": 6.194945848375451,
      "grad_norm": 21.740747451782227,
      "learning_rate": 0.0001984536484849007,
      "loss": 7.0647,
      "step": 51480
    },
    {
      "epoch": 6.196149217809868,
      "grad_norm": 328.05584716796875,
      "learning_rate": 0.00019845298198299812,
      "loss": 7.0085,
      "step": 51490
    },
    {
      "epoch": 6.197352587244284,
      "grad_norm": 184.19383239746094,
      "learning_rate": 0.00019845231533860972,
      "loss": 7.113,
      "step": 51500
    },
    {
      "epoch": 6.1985559566787005,
      "grad_norm": 62.600467681884766,
      "learning_rate": 0.00019845164855173658,
      "loss": 7.0953,
      "step": 51510
    },
    {
      "epoch": 6.199759326113116,
      "grad_norm": 14.029105186462402,
      "learning_rate": 0.00019845098162237954,
      "loss": 7.2352,
      "step": 51520
    },
    {
      "epoch": 6.200962695547533,
      "grad_norm": 25.74992561340332,
      "learning_rate": 0.00019845031455053964,
      "loss": 7.0544,
      "step": 51530
    },
    {
      "epoch": 6.20216606498195,
      "grad_norm": 59.89943313598633,
      "learning_rate": 0.00019844964733621784,
      "loss": 7.0761,
      "step": 51540
    },
    {
      "epoch": 6.2033694344163655,
      "grad_norm": 33.79304504394531,
      "learning_rate": 0.0001984489799794151,
      "loss": 7.1421,
      "step": 51550
    },
    {
      "epoch": 6.204572803850782,
      "grad_norm": 53.30209732055664,
      "learning_rate": 0.00019844831248013237,
      "loss": 7.0327,
      "step": 51560
    },
    {
      "epoch": 6.205776173285199,
      "grad_norm": 52.294410705566406,
      "learning_rate": 0.0001984476448383706,
      "loss": 7.0361,
      "step": 51570
    },
    {
      "epoch": 6.206979542719615,
      "grad_norm": 38.50944137573242,
      "learning_rate": 0.00019844697705413084,
      "loss": 7.0574,
      "step": 51580
    },
    {
      "epoch": 6.208182912154031,
      "grad_norm": 103.11620330810547,
      "learning_rate": 0.00019844630912741398,
      "loss": 7.074,
      "step": 51590
    },
    {
      "epoch": 6.209386281588448,
      "grad_norm": 30.791276931762695,
      "learning_rate": 0.000198445641058221,
      "loss": 7.0055,
      "step": 51600
    },
    {
      "epoch": 6.210589651022864,
      "grad_norm": 54.68841552734375,
      "learning_rate": 0.00019844497284655287,
      "loss": 6.9821,
      "step": 51610
    },
    {
      "epoch": 6.21179302045728,
      "grad_norm": 62.289852142333984,
      "learning_rate": 0.00019844430449241052,
      "loss": 6.8426,
      "step": 51620
    },
    {
      "epoch": 6.212996389891697,
      "grad_norm": 11.764166831970215,
      "learning_rate": 0.00019844363599579498,
      "loss": 6.9789,
      "step": 51630
    },
    {
      "epoch": 6.214199759326113,
      "grad_norm": 15.638115882873535,
      "learning_rate": 0.00019844296735670722,
      "loss": 7.0158,
      "step": 51640
    },
    {
      "epoch": 6.21540312876053,
      "grad_norm": 38.82330322265625,
      "learning_rate": 0.00019844229857514818,
      "loss": 6.9124,
      "step": 51650
    },
    {
      "epoch": 6.216606498194946,
      "grad_norm": 29.755327224731445,
      "learning_rate": 0.0001984416296511188,
      "loss": 6.969,
      "step": 51660
    },
    {
      "epoch": 6.217809867629362,
      "grad_norm": 20.518600463867188,
      "learning_rate": 0.00019844096058462007,
      "loss": 6.834,
      "step": 51670
    },
    {
      "epoch": 6.219013237063779,
      "grad_norm": 22.286468505859375,
      "learning_rate": 0.000198440291375653,
      "loss": 6.8331,
      "step": 51680
    },
    {
      "epoch": 6.2202166064981945,
      "grad_norm": 18.754711151123047,
      "learning_rate": 0.00019843962202421853,
      "loss": 6.8883,
      "step": 51690
    },
    {
      "epoch": 6.221419975932611,
      "grad_norm": 14.924210548400879,
      "learning_rate": 0.0001984389525303176,
      "loss": 6.9185,
      "step": 51700
    },
    {
      "epoch": 6.222623345367028,
      "grad_norm": 267.7951354980469,
      "learning_rate": 0.00019843828289395118,
      "loss": 7.0813,
      "step": 51710
    },
    {
      "epoch": 6.223826714801444,
      "grad_norm": 46.562496185302734,
      "learning_rate": 0.0001984376131151203,
      "loss": 6.8518,
      "step": 51720
    },
    {
      "epoch": 6.22503008423586,
      "grad_norm": 32.995338439941406,
      "learning_rate": 0.00019843694319382585,
      "loss": 6.8675,
      "step": 51730
    },
    {
      "epoch": 6.226233453670277,
      "grad_norm": 33.15148162841797,
      "learning_rate": 0.0001984362731300689,
      "loss": 6.7144,
      "step": 51740
    },
    {
      "epoch": 6.227436823104693,
      "grad_norm": 103.49129486083984,
      "learning_rate": 0.00019843560292385034,
      "loss": 6.6703,
      "step": 51750
    },
    {
      "epoch": 6.2286401925391095,
      "grad_norm": 23.396648406982422,
      "learning_rate": 0.00019843493257517113,
      "loss": 6.8905,
      "step": 51760
    },
    {
      "epoch": 6.229843561973526,
      "grad_norm": 66.21638488769531,
      "learning_rate": 0.00019843426208403228,
      "loss": 6.8329,
      "step": 51770
    },
    {
      "epoch": 6.231046931407942,
      "grad_norm": 25.776018142700195,
      "learning_rate": 0.00019843359145043477,
      "loss": 6.7987,
      "step": 51780
    },
    {
      "epoch": 6.232250300842359,
      "grad_norm": 74.73793029785156,
      "learning_rate": 0.00019843292067437952,
      "loss": 6.7562,
      "step": 51790
    },
    {
      "epoch": 6.233453670276775,
      "grad_norm": 20.916358947753906,
      "learning_rate": 0.00019843224975586756,
      "loss": 6.8506,
      "step": 51800
    },
    {
      "epoch": 6.234657039711191,
      "grad_norm": 35.76206970214844,
      "learning_rate": 0.0001984315786948998,
      "loss": 6.6572,
      "step": 51810
    },
    {
      "epoch": 6.235860409145608,
      "grad_norm": 20.380142211914062,
      "learning_rate": 0.0001984309074914773,
      "loss": 6.746,
      "step": 51820
    },
    {
      "epoch": 6.2370637785800245,
      "grad_norm": 60.54458236694336,
      "learning_rate": 0.0001984302361456009,
      "loss": 6.7675,
      "step": 51830
    },
    {
      "epoch": 6.23826714801444,
      "grad_norm": 103.00148010253906,
      "learning_rate": 0.00019842956465727173,
      "loss": 6.8724,
      "step": 51840
    },
    {
      "epoch": 6.239470517448857,
      "grad_norm": 105.30757141113281,
      "learning_rate": 0.00019842889302649062,
      "loss": 6.7432,
      "step": 51850
    },
    {
      "epoch": 6.240673886883273,
      "grad_norm": 44.128936767578125,
      "learning_rate": 0.0001984282212532586,
      "loss": 6.7083,
      "step": 51860
    },
    {
      "epoch": 6.241877256317689,
      "grad_norm": 561.1841430664062,
      "learning_rate": 0.0001984275493375767,
      "loss": 6.9873,
      "step": 51870
    },
    {
      "epoch": 6.243080625752106,
      "grad_norm": 41.606361389160156,
      "learning_rate": 0.00019842687727944576,
      "loss": 6.873,
      "step": 51880
    },
    {
      "epoch": 6.244283995186522,
      "grad_norm": 67.96810150146484,
      "learning_rate": 0.00019842620507886687,
      "loss": 6.9031,
      "step": 51890
    },
    {
      "epoch": 6.245487364620939,
      "grad_norm": 48.38755416870117,
      "learning_rate": 0.00019842553273584094,
      "loss": 6.9124,
      "step": 51900
    },
    {
      "epoch": 6.246690734055355,
      "grad_norm": 79.95443725585938,
      "learning_rate": 0.00019842486025036898,
      "loss": 6.8622,
      "step": 51910
    },
    {
      "epoch": 6.247894103489771,
      "grad_norm": 110.63379669189453,
      "learning_rate": 0.00019842418762245194,
      "loss": 6.773,
      "step": 51920
    },
    {
      "epoch": 6.249097472924188,
      "grad_norm": 295.70623779296875,
      "learning_rate": 0.00019842351485209082,
      "loss": 6.977,
      "step": 51930
    },
    {
      "epoch": 6.250300842358604,
      "grad_norm": 324.2843017578125,
      "learning_rate": 0.00019842284193928654,
      "loss": 7.1088,
      "step": 51940
    },
    {
      "epoch": 6.25150421179302,
      "grad_norm": 128.91664123535156,
      "learning_rate": 0.00019842216888404013,
      "loss": 6.9703,
      "step": 51950
    },
    {
      "epoch": 6.252707581227437,
      "grad_norm": 90.00405883789062,
      "learning_rate": 0.00019842149568635254,
      "loss": 7.0504,
      "step": 51960
    },
    {
      "epoch": 6.2539109506618535,
      "grad_norm": 28.90699577331543,
      "learning_rate": 0.0001984208223462247,
      "loss": 7.0963,
      "step": 51970
    },
    {
      "epoch": 6.255114320096269,
      "grad_norm": 170.06979370117188,
      "learning_rate": 0.00019842014886365772,
      "loss": 7.0379,
      "step": 51980
    },
    {
      "epoch": 6.256317689530686,
      "grad_norm": 73.8154525756836,
      "learning_rate": 0.00019841947523865244,
      "loss": 7.0657,
      "step": 51990
    },
    {
      "epoch": 6.257521058965103,
      "grad_norm": 47.21464538574219,
      "learning_rate": 0.00019841880147120988,
      "loss": 6.8675,
      "step": 52000
    },
    {
      "epoch": 6.2587244283995185,
      "grad_norm": 62.993038177490234,
      "learning_rate": 0.000198418127561331,
      "loss": 6.7525,
      "step": 52010
    },
    {
      "epoch": 6.259927797833935,
      "grad_norm": 70.13294982910156,
      "learning_rate": 0.00019841745350901682,
      "loss": 6.8223,
      "step": 52020
    },
    {
      "epoch": 6.261131167268351,
      "grad_norm": 97.21419525146484,
      "learning_rate": 0.00019841677931426826,
      "loss": 6.7882,
      "step": 52030
    },
    {
      "epoch": 6.262334536702768,
      "grad_norm": 1414.242431640625,
      "learning_rate": 0.00019841610497708633,
      "loss": 7.2025,
      "step": 52040
    },
    {
      "epoch": 6.263537906137184,
      "grad_norm": 261.9495544433594,
      "learning_rate": 0.000198415430497472,
      "loss": 7.3967,
      "step": 52050
    },
    {
      "epoch": 6.2647412755716,
      "grad_norm": 248.5710906982422,
      "learning_rate": 0.00019841475587542627,
      "loss": 7.1057,
      "step": 52060
    },
    {
      "epoch": 6.265944645006017,
      "grad_norm": 253.21038818359375,
      "learning_rate": 0.00019841408111095004,
      "loss": 7.2014,
      "step": 52070
    },
    {
      "epoch": 6.2671480144404335,
      "grad_norm": 277.64862060546875,
      "learning_rate": 0.00019841340620404437,
      "loss": 7.2097,
      "step": 52080
    },
    {
      "epoch": 6.268351383874849,
      "grad_norm": 501.5820007324219,
      "learning_rate": 0.0001984127311547102,
      "loss": 7.2405,
      "step": 52090
    },
    {
      "epoch": 6.269554753309266,
      "grad_norm": 36.811588287353516,
      "learning_rate": 0.0001984120559629485,
      "loss": 7.3904,
      "step": 52100
    },
    {
      "epoch": 6.270758122743683,
      "grad_norm": 1088.7659912109375,
      "learning_rate": 0.0001984113806287603,
      "loss": 7.2532,
      "step": 52110
    },
    {
      "epoch": 6.271961492178098,
      "grad_norm": 114.97406768798828,
      "learning_rate": 0.0001984107051521465,
      "loss": 7.7797,
      "step": 52120
    },
    {
      "epoch": 6.273164861612515,
      "grad_norm": 13.400031089782715,
      "learning_rate": 0.00019841002953310808,
      "loss": 7.5005,
      "step": 52130
    },
    {
      "epoch": 6.274368231046932,
      "grad_norm": 9.023686408996582,
      "learning_rate": 0.0001984093537716461,
      "loss": 7.6065,
      "step": 52140
    },
    {
      "epoch": 6.2755716004813475,
      "grad_norm": 35.21834182739258,
      "learning_rate": 0.00019840867786776145,
      "loss": 7.6507,
      "step": 52150
    },
    {
      "epoch": 6.276774969915764,
      "grad_norm": 10.762063026428223,
      "learning_rate": 0.0001984080018214552,
      "loss": 7.7764,
      "step": 52160
    },
    {
      "epoch": 6.277978339350181,
      "grad_norm": 9.541496276855469,
      "learning_rate": 0.00019840732563272822,
      "loss": 7.4684,
      "step": 52170
    },
    {
      "epoch": 6.279181708784597,
      "grad_norm": 26.444568634033203,
      "learning_rate": 0.00019840664930158156,
      "loss": 7.5107,
      "step": 52180
    },
    {
      "epoch": 6.280385078219013,
      "grad_norm": 18.916412353515625,
      "learning_rate": 0.00019840597282801622,
      "loss": 7.5227,
      "step": 52190
    },
    {
      "epoch": 6.28158844765343,
      "grad_norm": 12.912468910217285,
      "learning_rate": 0.0001984052962120331,
      "loss": 7.4949,
      "step": 52200
    },
    {
      "epoch": 6.282791817087846,
      "grad_norm": 9.408341407775879,
      "learning_rate": 0.00019840461945363322,
      "loss": 7.2832,
      "step": 52210
    },
    {
      "epoch": 6.2839951865222625,
      "grad_norm": 27.467823028564453,
      "learning_rate": 0.0001984039425528176,
      "loss": 7.2377,
      "step": 52220
    },
    {
      "epoch": 6.285198555956678,
      "grad_norm": 32.02880096435547,
      "learning_rate": 0.00019840326550958715,
      "loss": 7.3939,
      "step": 52230
    },
    {
      "epoch": 6.286401925391095,
      "grad_norm": 15.904004096984863,
      "learning_rate": 0.00019840258832394287,
      "loss": 7.4093,
      "step": 52240
    },
    {
      "epoch": 6.287605294825512,
      "grad_norm": 30.023584365844727,
      "learning_rate": 0.00019840191099588576,
      "loss": 7.3163,
      "step": 52250
    },
    {
      "epoch": 6.2888086642599275,
      "grad_norm": 28.092247009277344,
      "learning_rate": 0.0001984012335254168,
      "loss": 7.447,
      "step": 52260
    },
    {
      "epoch": 6.290012033694344,
      "grad_norm": 6.808090686798096,
      "learning_rate": 0.00019840055591253698,
      "loss": 7.4688,
      "step": 52270
    },
    {
      "epoch": 6.291215403128761,
      "grad_norm": 38.90576934814453,
      "learning_rate": 0.00019839987815724723,
      "loss": 7.4105,
      "step": 52280
    },
    {
      "epoch": 6.292418772563177,
      "grad_norm": 14.360461235046387,
      "learning_rate": 0.00019839920025954854,
      "loss": 7.4389,
      "step": 52290
    },
    {
      "epoch": 6.293622141997593,
      "grad_norm": 31.73518180847168,
      "learning_rate": 0.00019839852221944195,
      "loss": 7.3699,
      "step": 52300
    },
    {
      "epoch": 6.29482551143201,
      "grad_norm": 9.294116973876953,
      "learning_rate": 0.0001983978440369284,
      "loss": 7.3123,
      "step": 52310
    },
    {
      "epoch": 6.296028880866426,
      "grad_norm": 10.44865608215332,
      "learning_rate": 0.00019839716571200887,
      "loss": 7.1166,
      "step": 52320
    },
    {
      "epoch": 6.297232250300842,
      "grad_norm": 12.122678756713867,
      "learning_rate": 0.00019839648724468437,
      "loss": 7.1422,
      "step": 52330
    },
    {
      "epoch": 6.298435619735259,
      "grad_norm": 22.11392593383789,
      "learning_rate": 0.00019839580863495582,
      "loss": 6.9034,
      "step": 52340
    },
    {
      "epoch": 6.299638989169675,
      "grad_norm": 46.20428466796875,
      "learning_rate": 0.00019839512988282428,
      "loss": 7.1029,
      "step": 52350
    },
    {
      "epoch": 6.300842358604092,
      "grad_norm": 425.1414794921875,
      "learning_rate": 0.00019839445098829067,
      "loss": 7.1642,
      "step": 52360
    },
    {
      "epoch": 6.302045728038507,
      "grad_norm": 61.78722381591797,
      "learning_rate": 0.000198393771951356,
      "loss": 7.6449,
      "step": 52370
    },
    {
      "epoch": 6.303249097472924,
      "grad_norm": 54.612648010253906,
      "learning_rate": 0.00019839309277202126,
      "loss": 7.4344,
      "step": 52380
    },
    {
      "epoch": 6.304452466907341,
      "grad_norm": 52.32176208496094,
      "learning_rate": 0.0001983924134502874,
      "loss": 7.1869,
      "step": 52390
    },
    {
      "epoch": 6.3056558363417565,
      "grad_norm": 137.26119995117188,
      "learning_rate": 0.00019839173398615546,
      "loss": 6.9554,
      "step": 52400
    },
    {
      "epoch": 6.306859205776173,
      "grad_norm": 83.4488754272461,
      "learning_rate": 0.00019839105437962636,
      "loss": 6.9961,
      "step": 52410
    },
    {
      "epoch": 6.30806257521059,
      "grad_norm": 300.5815734863281,
      "learning_rate": 0.0001983903746307011,
      "loss": 7.0563,
      "step": 52420
    },
    {
      "epoch": 6.309265944645006,
      "grad_norm": 119.7309341430664,
      "learning_rate": 0.0001983896947393807,
      "loss": 7.0218,
      "step": 52430
    },
    {
      "epoch": 6.310469314079422,
      "grad_norm": 143.92747497558594,
      "learning_rate": 0.00019838901470566614,
      "loss": 7.1164,
      "step": 52440
    },
    {
      "epoch": 6.311672683513839,
      "grad_norm": 181.8562469482422,
      "learning_rate": 0.00019838833452955835,
      "loss": 7.1949,
      "step": 52450
    },
    {
      "epoch": 6.312876052948255,
      "grad_norm": 62.823307037353516,
      "learning_rate": 0.00019838765421105836,
      "loss": 7.2763,
      "step": 52460
    },
    {
      "epoch": 6.3140794223826715,
      "grad_norm": 345.1018371582031,
      "learning_rate": 0.00019838697375016714,
      "loss": 7.2706,
      "step": 52470
    },
    {
      "epoch": 6.315282791817088,
      "grad_norm": 41.3524055480957,
      "learning_rate": 0.00019838629314688567,
      "loss": 7.2011,
      "step": 52480
    },
    {
      "epoch": 6.316486161251504,
      "grad_norm": 87.23824310302734,
      "learning_rate": 0.00019838561240121495,
      "loss": 7.1037,
      "step": 52490
    },
    {
      "epoch": 6.317689530685921,
      "grad_norm": 92.90904235839844,
      "learning_rate": 0.000198384931513156,
      "loss": 6.9603,
      "step": 52500
    },
    {
      "epoch": 6.318892900120337,
      "grad_norm": 78.8731460571289,
      "learning_rate": 0.00019838425048270968,
      "loss": 6.9291,
      "step": 52510
    },
    {
      "epoch": 6.320096269554753,
      "grad_norm": 27.37542724609375,
      "learning_rate": 0.00019838356930987713,
      "loss": 6.9328,
      "step": 52520
    },
    {
      "epoch": 6.32129963898917,
      "grad_norm": 15.593992233276367,
      "learning_rate": 0.00019838288799465924,
      "loss": 7.1231,
      "step": 52530
    },
    {
      "epoch": 6.3225030084235865,
      "grad_norm": 78.6000747680664,
      "learning_rate": 0.000198382206537057,
      "loss": 7.2843,
      "step": 52540
    },
    {
      "epoch": 6.323706377858002,
      "grad_norm": 165.88641357421875,
      "learning_rate": 0.00019838152493707146,
      "loss": 7.2011,
      "step": 52550
    },
    {
      "epoch": 6.324909747292419,
      "grad_norm": 32.2805061340332,
      "learning_rate": 0.00019838084319470355,
      "loss": 7.118,
      "step": 52560
    },
    {
      "epoch": 6.326113116726835,
      "grad_norm": 29.490070343017578,
      "learning_rate": 0.00019838016130995425,
      "loss": 7.1597,
      "step": 52570
    },
    {
      "epoch": 6.327316486161251,
      "grad_norm": 17.017850875854492,
      "learning_rate": 0.00019837947928282458,
      "loss": 6.9764,
      "step": 52580
    },
    {
      "epoch": 6.328519855595668,
      "grad_norm": 35.79988098144531,
      "learning_rate": 0.00019837879711331552,
      "loss": 7.027,
      "step": 52590
    },
    {
      "epoch": 6.329723225030084,
      "grad_norm": 452.3404541015625,
      "learning_rate": 0.00019837811480142802,
      "loss": 6.9008,
      "step": 52600
    },
    {
      "epoch": 6.330926594464501,
      "grad_norm": 958.8521728515625,
      "learning_rate": 0.00019837743234716313,
      "loss": 7.2571,
      "step": 52610
    },
    {
      "epoch": 6.332129963898917,
      "grad_norm": 33.366722106933594,
      "learning_rate": 0.0001983767497505218,
      "loss": 7.408,
      "step": 52620
    },
    {
      "epoch": 6.333333333333333,
      "grad_norm": 43.752098083496094,
      "learning_rate": 0.00019837606701150504,
      "loss": 7.3278,
      "step": 52630
    },
    {
      "epoch": 6.33453670276775,
      "grad_norm": 51.28098678588867,
      "learning_rate": 0.0001983753841301138,
      "loss": 7.2685,
      "step": 52640
    },
    {
      "epoch": 6.335740072202166,
      "grad_norm": 76.79080200195312,
      "learning_rate": 0.0001983747011063491,
      "loss": 7.1186,
      "step": 52650
    },
    {
      "epoch": 6.336943441636582,
      "grad_norm": 329.1863098144531,
      "learning_rate": 0.00019837401794021192,
      "loss": 7.3061,
      "step": 52660
    },
    {
      "epoch": 6.338146811070999,
      "grad_norm": 201.25904846191406,
      "learning_rate": 0.00019837333463170324,
      "loss": 7.2656,
      "step": 52670
    },
    {
      "epoch": 6.3393501805054155,
      "grad_norm": 122.38607025146484,
      "learning_rate": 0.00019837265118082407,
      "loss": 7.1808,
      "step": 52680
    },
    {
      "epoch": 6.340553549939831,
      "grad_norm": 494.30841064453125,
      "learning_rate": 0.00019837196758757539,
      "loss": 7.2204,
      "step": 52690
    },
    {
      "epoch": 6.341756919374248,
      "grad_norm": 160.0564727783203,
      "learning_rate": 0.0001983712838519582,
      "loss": 7.0689,
      "step": 52700
    },
    {
      "epoch": 6.342960288808664,
      "grad_norm": 244.82846069335938,
      "learning_rate": 0.00019837059997397343,
      "loss": 7.0221,
      "step": 52710
    },
    {
      "epoch": 6.3441636582430805,
      "grad_norm": 69.26526641845703,
      "learning_rate": 0.00019836991595362214,
      "loss": 7.0907,
      "step": 52720
    },
    {
      "epoch": 6.345367027677497,
      "grad_norm": 32.76903533935547,
      "learning_rate": 0.0001983692317909053,
      "loss": 7.0811,
      "step": 52730
    },
    {
      "epoch": 6.346570397111913,
      "grad_norm": 91.3707275390625,
      "learning_rate": 0.00019836854748582388,
      "loss": 7.0809,
      "step": 52740
    },
    {
      "epoch": 6.34777376654633,
      "grad_norm": 92.4900131225586,
      "learning_rate": 0.00019836786303837887,
      "loss": 6.9811,
      "step": 52750
    },
    {
      "epoch": 6.348977135980746,
      "grad_norm": 985.42431640625,
      "learning_rate": 0.00019836717844857132,
      "loss": 7.1548,
      "step": 52760
    },
    {
      "epoch": 6.350180505415162,
      "grad_norm": 387.5338134765625,
      "learning_rate": 0.00019836649371640216,
      "loss": 7.2645,
      "step": 52770
    },
    {
      "epoch": 6.351383874849579,
      "grad_norm": 188.41053771972656,
      "learning_rate": 0.00019836580884187237,
      "loss": 6.9158,
      "step": 52780
    },
    {
      "epoch": 6.3525872442839955,
      "grad_norm": 460.3943176269531,
      "learning_rate": 0.000198365123824983,
      "loss": 7.4076,
      "step": 52790
    },
    {
      "epoch": 6.353790613718411,
      "grad_norm": 145.7770538330078,
      "learning_rate": 0.00019836443866573503,
      "loss": 7.3129,
      "step": 52800
    },
    {
      "epoch": 6.354993983152828,
      "grad_norm": 84.12643432617188,
      "learning_rate": 0.0001983637533641294,
      "loss": 7.2679,
      "step": 52810
    },
    {
      "epoch": 6.356197352587245,
      "grad_norm": 121.4907455444336,
      "learning_rate": 0.00019836306792016711,
      "loss": 7.2408,
      "step": 52820
    },
    {
      "epoch": 6.35740072202166,
      "grad_norm": 286.1339416503906,
      "learning_rate": 0.00019836238233384923,
      "loss": 7.182,
      "step": 52830
    },
    {
      "epoch": 6.358604091456077,
      "grad_norm": 987.18017578125,
      "learning_rate": 0.00019836169660517665,
      "loss": 7.2193,
      "step": 52840
    },
    {
      "epoch": 6.359807460890494,
      "grad_norm": 411.76727294921875,
      "learning_rate": 0.00019836101073415042,
      "loss": 7.2532,
      "step": 52850
    },
    {
      "epoch": 6.3610108303249095,
      "grad_norm": 118.34429931640625,
      "learning_rate": 0.00019836032472077156,
      "loss": 7.1956,
      "step": 52860
    },
    {
      "epoch": 6.362214199759326,
      "grad_norm": 623.2942504882812,
      "learning_rate": 0.000198359638565041,
      "loss": 7.1516,
      "step": 52870
    },
    {
      "epoch": 6.363417569193743,
      "grad_norm": 792.5098266601562,
      "learning_rate": 0.00019835895226695974,
      "loss": 7.0995,
      "step": 52880
    },
    {
      "epoch": 6.364620938628159,
      "grad_norm": 325.6565856933594,
      "learning_rate": 0.00019835826582652883,
      "loss": 7.1832,
      "step": 52890
    },
    {
      "epoch": 6.365824308062575,
      "grad_norm": 204.7774658203125,
      "learning_rate": 0.00019835757924374918,
      "loss": 7.1555,
      "step": 52900
    },
    {
      "epoch": 6.367027677496991,
      "grad_norm": 157.6756134033203,
      "learning_rate": 0.00019835689251862187,
      "loss": 7.1083,
      "step": 52910
    },
    {
      "epoch": 6.368231046931408,
      "grad_norm": 182.59307861328125,
      "learning_rate": 0.00019835620565114784,
      "loss": 7.1596,
      "step": 52920
    },
    {
      "epoch": 6.3694344163658245,
      "grad_norm": 414.76959228515625,
      "learning_rate": 0.0001983555186413281,
      "loss": 7.1812,
      "step": 52930
    },
    {
      "epoch": 6.37063778580024,
      "grad_norm": 375.0384216308594,
      "learning_rate": 0.00019835483148916363,
      "loss": 7.141,
      "step": 52940
    },
    {
      "epoch": 6.371841155234657,
      "grad_norm": 2651.772705078125,
      "learning_rate": 0.00019835414419465543,
      "loss": 7.3201,
      "step": 52950
    },
    {
      "epoch": 6.373044524669074,
      "grad_norm": 199.78663635253906,
      "learning_rate": 0.00019835345675780453,
      "loss": 7.2571,
      "step": 52960
    },
    {
      "epoch": 6.3742478941034895,
      "grad_norm": 822.8274536132812,
      "learning_rate": 0.00019835276917861186,
      "loss": 7.2299,
      "step": 52970
    },
    {
      "epoch": 6.375451263537906,
      "grad_norm": 485.8435974121094,
      "learning_rate": 0.0001983520814570785,
      "loss": 7.2557,
      "step": 52980
    },
    {
      "epoch": 6.376654632972323,
      "grad_norm": 396.6460876464844,
      "learning_rate": 0.00019835139359320535,
      "loss": 7.1102,
      "step": 52990
    },
    {
      "epoch": 6.377858002406739,
      "grad_norm": 855.5442504882812,
      "learning_rate": 0.00019835070558699348,
      "loss": 7.1931,
      "step": 53000
    },
    {
      "epoch": 6.379061371841155,
      "grad_norm": 376.45330810546875,
      "learning_rate": 0.00019835001743844383,
      "loss": 7.2304,
      "step": 53010
    },
    {
      "epoch": 6.380264741275572,
      "grad_norm": 98.51732635498047,
      "learning_rate": 0.00019834932914755744,
      "loss": 7.0782,
      "step": 53020
    },
    {
      "epoch": 6.381468110709988,
      "grad_norm": 69.67068481445312,
      "learning_rate": 0.0001983486407143353,
      "loss": 7.5735,
      "step": 53030
    },
    {
      "epoch": 6.382671480144404,
      "grad_norm": 201.29722595214844,
      "learning_rate": 0.00019834795213877836,
      "loss": 7.4532,
      "step": 53040
    },
    {
      "epoch": 6.38387484957882,
      "grad_norm": 345.9329833984375,
      "learning_rate": 0.00019834726342088767,
      "loss": 7.3236,
      "step": 53050
    },
    {
      "epoch": 6.385078219013237,
      "grad_norm": 118.62815856933594,
      "learning_rate": 0.0001983465745606642,
      "loss": 7.3724,
      "step": 53060
    },
    {
      "epoch": 6.386281588447654,
      "grad_norm": 44.16498565673828,
      "learning_rate": 0.000198345885558109,
      "loss": 7.4937,
      "step": 53070
    },
    {
      "epoch": 6.387484957882069,
      "grad_norm": 6.8613080978393555,
      "learning_rate": 0.00019834519641322298,
      "loss": 7.4436,
      "step": 53080
    },
    {
      "epoch": 6.388688327316486,
      "grad_norm": 62.453189849853516,
      "learning_rate": 0.0001983445071260072,
      "loss": 7.5031,
      "step": 53090
    },
    {
      "epoch": 6.389891696750903,
      "grad_norm": 29.752826690673828,
      "learning_rate": 0.00019834381769646262,
      "loss": 7.3467,
      "step": 53100
    },
    {
      "epoch": 6.3910950661853185,
      "grad_norm": 8.138988494873047,
      "learning_rate": 0.00019834312812459025,
      "loss": 7.4467,
      "step": 53110
    },
    {
      "epoch": 6.392298435619735,
      "grad_norm": 15.982645034790039,
      "learning_rate": 0.0001983424384103911,
      "loss": 7.5223,
      "step": 53120
    },
    {
      "epoch": 6.393501805054152,
      "grad_norm": 45.11640548706055,
      "learning_rate": 0.00019834174855386618,
      "loss": 7.2417,
      "step": 53130
    },
    {
      "epoch": 6.394705174488568,
      "grad_norm": 17.564626693725586,
      "learning_rate": 0.00019834105855501646,
      "loss": 7.2857,
      "step": 53140
    },
    {
      "epoch": 6.395908543922984,
      "grad_norm": 551.6398315429688,
      "learning_rate": 0.00019834036841384295,
      "loss": 7.6144,
      "step": 53150
    },
    {
      "epoch": 6.397111913357401,
      "grad_norm": 124.64905548095703,
      "learning_rate": 0.00019833967813034665,
      "loss": 7.5345,
      "step": 53160
    },
    {
      "epoch": 6.398315282791817,
      "grad_norm": 25.615991592407227,
      "learning_rate": 0.00019833898770452851,
      "loss": 7.3548,
      "step": 53170
    },
    {
      "epoch": 6.3995186522262335,
      "grad_norm": 24.818967819213867,
      "learning_rate": 0.00019833829713638963,
      "loss": 7.2835,
      "step": 53180
    },
    {
      "epoch": 6.40072202166065,
      "grad_norm": 160.3653564453125,
      "learning_rate": 0.00019833760642593096,
      "loss": 7.2268,
      "step": 53190
    },
    {
      "epoch": 6.401925391095066,
      "grad_norm": 2392.0791015625,
      "learning_rate": 0.00019833691557315345,
      "loss": 7.5182,
      "step": 53200
    },
    {
      "epoch": 6.403128760529483,
      "grad_norm": 674.7615356445312,
      "learning_rate": 0.00019833622457805818,
      "loss": 7.7121,
      "step": 53210
    },
    {
      "epoch": 6.404332129963899,
      "grad_norm": 326.4076843261719,
      "learning_rate": 0.00019833553344064612,
      "loss": 7.3973,
      "step": 53220
    },
    {
      "epoch": 6.405535499398315,
      "grad_norm": 266.168701171875,
      "learning_rate": 0.00019833484216091825,
      "loss": 7.2008,
      "step": 53230
    },
    {
      "epoch": 6.406738868832732,
      "grad_norm": 2855.228271484375,
      "learning_rate": 0.00019833415073887558,
      "loss": 7.2671,
      "step": 53240
    },
    {
      "epoch": 6.4079422382671485,
      "grad_norm": 908.1167602539062,
      "learning_rate": 0.00019833345917451913,
      "loss": 7.0994,
      "step": 53250
    },
    {
      "epoch": 6.409145607701564,
      "grad_norm": 250.6652374267578,
      "learning_rate": 0.0001983327674678499,
      "loss": 7.1249,
      "step": 53260
    },
    {
      "epoch": 6.410348977135981,
      "grad_norm": 185.2032928466797,
      "learning_rate": 0.00019833207561886882,
      "loss": 7.1267,
      "step": 53270
    },
    {
      "epoch": 6.411552346570397,
      "grad_norm": 173.52169799804688,
      "learning_rate": 0.000198331383627577,
      "loss": 8.2331,
      "step": 53280
    },
    {
      "epoch": 6.412755716004813,
      "grad_norm": 103.46623229980469,
      "learning_rate": 0.00019833069149397537,
      "loss": 8.351,
      "step": 53290
    },
    {
      "epoch": 6.41395908543923,
      "grad_norm": 54.158424377441406,
      "learning_rate": 0.00019832999921806497,
      "loss": 7.7284,
      "step": 53300
    },
    {
      "epoch": 6.415162454873646,
      "grad_norm": 28.00765037536621,
      "learning_rate": 0.00019832930679984675,
      "loss": 7.6515,
      "step": 53310
    },
    {
      "epoch": 6.416365824308063,
      "grad_norm": 15.205865859985352,
      "learning_rate": 0.0001983286142393218,
      "loss": 7.5569,
      "step": 53320
    },
    {
      "epoch": 6.417569193742479,
      "grad_norm": 17.263912200927734,
      "learning_rate": 0.000198327921536491,
      "loss": 7.6979,
      "step": 53330
    },
    {
      "epoch": 6.418772563176895,
      "grad_norm": 21.40951156616211,
      "learning_rate": 0.00019832722869135547,
      "loss": 7.7632,
      "step": 53340
    },
    {
      "epoch": 6.419975932611312,
      "grad_norm": 7.949574947357178,
      "learning_rate": 0.00019832653570391618,
      "loss": 7.572,
      "step": 53350
    },
    {
      "epoch": 6.421179302045728,
      "grad_norm": 39.62449645996094,
      "learning_rate": 0.00019832584257417407,
      "loss": 7.5378,
      "step": 53360
    },
    {
      "epoch": 6.422382671480144,
      "grad_norm": 29.13290023803711,
      "learning_rate": 0.0001983251493021302,
      "loss": 7.4672,
      "step": 53370
    },
    {
      "epoch": 6.423586040914561,
      "grad_norm": 29.931182861328125,
      "learning_rate": 0.00019832445588778558,
      "loss": 7.2516,
      "step": 53380
    },
    {
      "epoch": 6.4247894103489775,
      "grad_norm": 55.71449661254883,
      "learning_rate": 0.00019832376233114117,
      "loss": 7.3434,
      "step": 53390
    },
    {
      "epoch": 6.425992779783393,
      "grad_norm": 96.51713562011719,
      "learning_rate": 0.00019832306863219805,
      "loss": 7.3206,
      "step": 53400
    },
    {
      "epoch": 6.42719614921781,
      "grad_norm": 1127.621826171875,
      "learning_rate": 0.00019832237479095716,
      "loss": 7.3673,
      "step": 53410
    },
    {
      "epoch": 6.428399518652226,
      "grad_norm": 79.29362487792969,
      "learning_rate": 0.00019832168080741947,
      "loss": 7.3262,
      "step": 53420
    },
    {
      "epoch": 6.4296028880866425,
      "grad_norm": 78.43046569824219,
      "learning_rate": 0.00019832098668158608,
      "loss": 7.3669,
      "step": 53430
    },
    {
      "epoch": 6.430806257521059,
      "grad_norm": 79.61399841308594,
      "learning_rate": 0.00019832029241345793,
      "loss": 7.8748,
      "step": 53440
    },
    {
      "epoch": 6.432009626955475,
      "grad_norm": 55.962799072265625,
      "learning_rate": 0.00019831959800303603,
      "loss": 7.9114,
      "step": 53450
    },
    {
      "epoch": 6.433212996389892,
      "grad_norm": 25.874568939208984,
      "learning_rate": 0.00019831890345032145,
      "loss": 7.4856,
      "step": 53460
    },
    {
      "epoch": 6.434416365824308,
      "grad_norm": 30.580080032348633,
      "learning_rate": 0.00019831820875531507,
      "loss": 7.5684,
      "step": 53470
    },
    {
      "epoch": 6.435619735258724,
      "grad_norm": 26.631053924560547,
      "learning_rate": 0.000198317513918018,
      "loss": 7.5614,
      "step": 53480
    },
    {
      "epoch": 6.436823104693141,
      "grad_norm": 4.994171142578125,
      "learning_rate": 0.00019831681893843123,
      "loss": 7.4872,
      "step": 53490
    },
    {
      "epoch": 6.4380264741275575,
      "grad_norm": 11.913789749145508,
      "learning_rate": 0.00019831612381655575,
      "loss": 7.5216,
      "step": 53500
    },
    {
      "epoch": 6.439229843561973,
      "grad_norm": 256.5828857421875,
      "learning_rate": 0.00019831542855239254,
      "loss": 7.4161,
      "step": 53510
    },
    {
      "epoch": 6.44043321299639,
      "grad_norm": 9.101548194885254,
      "learning_rate": 0.00019831473314594264,
      "loss": 7.5105,
      "step": 53520
    },
    {
      "epoch": 6.441636582430807,
      "grad_norm": 3.497997760772705,
      "learning_rate": 0.00019831403759720706,
      "loss": 7.3792,
      "step": 53530
    },
    {
      "epoch": 6.442839951865222,
      "grad_norm": 15.275854110717773,
      "learning_rate": 0.0001983133419061868,
      "loss": 7.4054,
      "step": 53540
    },
    {
      "epoch": 6.444043321299639,
      "grad_norm": 4.371145248413086,
      "learning_rate": 0.00019831264607288285,
      "loss": 7.2603,
      "step": 53550
    },
    {
      "epoch": 6.445246690734056,
      "grad_norm": 21.592967987060547,
      "learning_rate": 0.00019831195009729623,
      "loss": 7.1969,
      "step": 53560
    },
    {
      "epoch": 6.4464500601684716,
      "grad_norm": 7.722259521484375,
      "learning_rate": 0.00019831125397942794,
      "loss": 7.1575,
      "step": 53570
    },
    {
      "epoch": 6.447653429602888,
      "grad_norm": 17.183794021606445,
      "learning_rate": 0.00019831055771927898,
      "loss": 7.1835,
      "step": 53580
    },
    {
      "epoch": 6.448856799037305,
      "grad_norm": 18.575815200805664,
      "learning_rate": 0.0001983098613168504,
      "loss": 7.1085,
      "step": 53590
    },
    {
      "epoch": 6.450060168471721,
      "grad_norm": 427.5148620605469,
      "learning_rate": 0.00019830916477214317,
      "loss": 7.3198,
      "step": 53600
    },
    {
      "epoch": 6.451263537906137,
      "grad_norm": 80.92161560058594,
      "learning_rate": 0.0001983084680851583,
      "loss": 7.7243,
      "step": 53610
    },
    {
      "epoch": 6.452466907340553,
      "grad_norm": 183.42733764648438,
      "learning_rate": 0.00019830777125589682,
      "loss": 7.6895,
      "step": 53620
    },
    {
      "epoch": 6.45367027677497,
      "grad_norm": 109.63179779052734,
      "learning_rate": 0.0001983070742843597,
      "loss": 7.3347,
      "step": 53630
    },
    {
      "epoch": 6.4548736462093865,
      "grad_norm": 58.21371841430664,
      "learning_rate": 0.00019830637717054798,
      "loss": 7.3218,
      "step": 53640
    },
    {
      "epoch": 6.456077015643802,
      "grad_norm": 30.193782806396484,
      "learning_rate": 0.00019830567991446264,
      "loss": 7.2972,
      "step": 53650
    },
    {
      "epoch": 6.457280385078219,
      "grad_norm": 33.620445251464844,
      "learning_rate": 0.00019830498251610475,
      "loss": 7.4496,
      "step": 53660
    },
    {
      "epoch": 6.458483754512636,
      "grad_norm": 27.148279190063477,
      "learning_rate": 0.00019830428497547527,
      "loss": 7.4207,
      "step": 53670
    },
    {
      "epoch": 6.4596871239470515,
      "grad_norm": 72.65086364746094,
      "learning_rate": 0.0001983035872925752,
      "loss": 7.2944,
      "step": 53680
    },
    {
      "epoch": 6.460890493381468,
      "grad_norm": 72.13494110107422,
      "learning_rate": 0.00019830288946740557,
      "loss": 7.3604,
      "step": 53690
    },
    {
      "epoch": 6.462093862815885,
      "grad_norm": 50.20564270019531,
      "learning_rate": 0.00019830219149996744,
      "loss": 7.4214,
      "step": 53700
    },
    {
      "epoch": 6.463297232250301,
      "grad_norm": 63.062278747558594,
      "learning_rate": 0.00019830149339026168,
      "loss": 7.2255,
      "step": 53710
    },
    {
      "epoch": 6.464500601684717,
      "grad_norm": 104.42791748046875,
      "learning_rate": 0.00019830079513828946,
      "loss": 7.2095,
      "step": 53720
    },
    {
      "epoch": 6.465703971119134,
      "grad_norm": 93.70234680175781,
      "learning_rate": 0.00019830009674405173,
      "loss": 7.1209,
      "step": 53730
    },
    {
      "epoch": 6.46690734055355,
      "grad_norm": 36.293392181396484,
      "learning_rate": 0.00019829939820754943,
      "loss": 7.1147,
      "step": 53740
    },
    {
      "epoch": 6.4681107099879664,
      "grad_norm": 78.2708740234375,
      "learning_rate": 0.00019829869952878366,
      "loss": 7.04,
      "step": 53750
    },
    {
      "epoch": 6.469314079422382,
      "grad_norm": 34.36333084106445,
      "learning_rate": 0.0001982980007077554,
      "loss": 7.0723,
      "step": 53760
    },
    {
      "epoch": 6.470517448856799,
      "grad_norm": 102.16293334960938,
      "learning_rate": 0.00019829730174446568,
      "loss": 7.1478,
      "step": 53770
    },
    {
      "epoch": 6.471720818291216,
      "grad_norm": 29.0257568359375,
      "learning_rate": 0.00019829660263891548,
      "loss": 7.3182,
      "step": 53780
    },
    {
      "epoch": 6.472924187725631,
      "grad_norm": 45.708526611328125,
      "learning_rate": 0.00019829590339110582,
      "loss": 7.2223,
      "step": 53790
    },
    {
      "epoch": 6.474127557160048,
      "grad_norm": 67.09626007080078,
      "learning_rate": 0.00019829520400103774,
      "loss": 7.2573,
      "step": 53800
    },
    {
      "epoch": 6.475330926594465,
      "grad_norm": 347.1438903808594,
      "learning_rate": 0.0001982945044687122,
      "loss": 7.094,
      "step": 53810
    },
    {
      "epoch": 6.4765342960288805,
      "grad_norm": 562.5654907226562,
      "learning_rate": 0.00019829380479413028,
      "loss": 7.1589,
      "step": 53820
    },
    {
      "epoch": 6.477737665463297,
      "grad_norm": 67.89314270019531,
      "learning_rate": 0.00019829310497729294,
      "loss": 7.1008,
      "step": 53830
    },
    {
      "epoch": 6.478941034897714,
      "grad_norm": 203.4894256591797,
      "learning_rate": 0.00019829240501820122,
      "loss": 7.276,
      "step": 53840
    },
    {
      "epoch": 6.48014440433213,
      "grad_norm": 27.177310943603516,
      "learning_rate": 0.0001982917049168561,
      "loss": 7.4379,
      "step": 53850
    },
    {
      "epoch": 6.481347773766546,
      "grad_norm": 18.682811737060547,
      "learning_rate": 0.00019829100467325862,
      "loss": 7.2928,
      "step": 53860
    },
    {
      "epoch": 6.482551143200963,
      "grad_norm": 110.0114974975586,
      "learning_rate": 0.00019829030428740983,
      "loss": 7.2596,
      "step": 53870
    },
    {
      "epoch": 6.483754512635379,
      "grad_norm": 217.84617614746094,
      "learning_rate": 0.00019828960375931068,
      "loss": 7.3008,
      "step": 53880
    },
    {
      "epoch": 6.4849578820697955,
      "grad_norm": 243.9082794189453,
      "learning_rate": 0.00019828890308896217,
      "loss": 7.78,
      "step": 53890
    },
    {
      "epoch": 6.486161251504212,
      "grad_norm": 9.155352592468262,
      "learning_rate": 0.00019828820227636539,
      "loss": 7.7512,
      "step": 53900
    },
    {
      "epoch": 6.487364620938628,
      "grad_norm": 10.2932767868042,
      "learning_rate": 0.0001982875013215213,
      "loss": 7.4618,
      "step": 53910
    },
    {
      "epoch": 6.488567990373045,
      "grad_norm": 17.100906372070312,
      "learning_rate": 0.0001982868002244309,
      "loss": 7.5332,
      "step": 53920
    },
    {
      "epoch": 6.489771359807461,
      "grad_norm": 12.623067855834961,
      "learning_rate": 0.00019828609898509529,
      "loss": 7.4229,
      "step": 53930
    },
    {
      "epoch": 6.490974729241877,
      "grad_norm": 50.046966552734375,
      "learning_rate": 0.0001982853976035154,
      "loss": 7.4933,
      "step": 53940
    },
    {
      "epoch": 6.492178098676294,
      "grad_norm": 21.30010414123535,
      "learning_rate": 0.0001982846960796923,
      "loss": 7.3344,
      "step": 53950
    },
    {
      "epoch": 6.49338146811071,
      "grad_norm": 183.37937927246094,
      "learning_rate": 0.00019828399441362696,
      "loss": 7.47,
      "step": 53960
    },
    {
      "epoch": 6.494584837545126,
      "grad_norm": 26.571096420288086,
      "learning_rate": 0.0001982832926053204,
      "loss": 7.4587,
      "step": 53970
    },
    {
      "epoch": 6.495788206979543,
      "grad_norm": 35.904136657714844,
      "learning_rate": 0.00019828259065477366,
      "loss": 7.6075,
      "step": 53980
    },
    {
      "epoch": 6.496991576413959,
      "grad_norm": 13.69642448425293,
      "learning_rate": 0.00019828188856198776,
      "loss": 7.5772,
      "step": 53990
    },
    {
      "epoch": 6.498194945848375,
      "grad_norm": 15.006000518798828,
      "learning_rate": 0.00019828118632696367,
      "loss": 7.5455,
      "step": 54000
    },
    {
      "epoch": 6.499398315282792,
      "grad_norm": 11.823795318603516,
      "learning_rate": 0.00019828048394970248,
      "loss": 7.4746,
      "step": 54010
    },
    {
      "epoch": 6.500601684717208,
      "grad_norm": 8.034140586853027,
      "learning_rate": 0.00019827978143020512,
      "loss": 7.3793,
      "step": 54020
    },
    {
      "epoch": 6.501805054151625,
      "grad_norm": 18.2496280670166,
      "learning_rate": 0.00019827907876847266,
      "loss": 7.4656,
      "step": 54030
    },
    {
      "epoch": 6.503008423586041,
      "grad_norm": 6.516838073730469,
      "learning_rate": 0.00019827837596450614,
      "loss": 7.4192,
      "step": 54040
    },
    {
      "epoch": 6.504211793020457,
      "grad_norm": 10.125895500183105,
      "learning_rate": 0.00019827767301830654,
      "loss": 7.314,
      "step": 54050
    },
    {
      "epoch": 6.505415162454874,
      "grad_norm": 5.568592071533203,
      "learning_rate": 0.00019827696992987486,
      "loss": 7.3927,
      "step": 54060
    },
    {
      "epoch": 6.5066185318892895,
      "grad_norm": 6.277139186859131,
      "learning_rate": 0.00019827626669921216,
      "loss": 7.2616,
      "step": 54070
    },
    {
      "epoch": 6.507821901323706,
      "grad_norm": 8.435803413391113,
      "learning_rate": 0.00019827556332631944,
      "loss": 7.3458,
      "step": 54080
    },
    {
      "epoch": 6.509025270758123,
      "grad_norm": 6.075100898742676,
      "learning_rate": 0.00019827485981119768,
      "loss": 7.2864,
      "step": 54090
    },
    {
      "epoch": 6.510228640192539,
      "grad_norm": 8.41025447845459,
      "learning_rate": 0.00019827415615384796,
      "loss": 7.3008,
      "step": 54100
    },
    {
      "epoch": 6.511432009626955,
      "grad_norm": 3.908461570739746,
      "learning_rate": 0.00019827345235427131,
      "loss": 7.2405,
      "step": 54110
    },
    {
      "epoch": 6.512635379061372,
      "grad_norm": 7.4334845542907715,
      "learning_rate": 0.00019827274841246865,
      "loss": 7.2647,
      "step": 54120
    },
    {
      "epoch": 6.513838748495788,
      "grad_norm": 7.621320724487305,
      "learning_rate": 0.0001982720443284411,
      "loss": 7.1673,
      "step": 54130
    },
    {
      "epoch": 6.5150421179302045,
      "grad_norm": 7.19697380065918,
      "learning_rate": 0.00019827134010218963,
      "loss": 7.1529,
      "step": 54140
    },
    {
      "epoch": 6.516245487364621,
      "grad_norm": 326.74365234375,
      "learning_rate": 0.00019827063573371527,
      "loss": 7.296,
      "step": 54150
    },
    {
      "epoch": 6.517448856799037,
      "grad_norm": 172.8400115966797,
      "learning_rate": 0.00019826993122301902,
      "loss": 7.8805,
      "step": 54160
    },
    {
      "epoch": 6.518652226233454,
      "grad_norm": 595.4959106445312,
      "learning_rate": 0.00019826922657010192,
      "loss": 7.9992,
      "step": 54170
    },
    {
      "epoch": 6.51985559566787,
      "grad_norm": 26.08918571472168,
      "learning_rate": 0.000198268521774965,
      "loss": 10.1198,
      "step": 54180
    },
    {
      "epoch": 6.521058965102286,
      "grad_norm": 11.50631046295166,
      "learning_rate": 0.00019826781683760926,
      "loss": 8.0815,
      "step": 54190
    },
    {
      "epoch": 6.522262334536703,
      "grad_norm": 10.769640922546387,
      "learning_rate": 0.00019826711175803573,
      "loss": 7.8827,
      "step": 54200
    },
    {
      "epoch": 6.5234657039711195,
      "grad_norm": 308.4264221191406,
      "learning_rate": 0.00019826640653624545,
      "loss": 7.5268,
      "step": 54210
    },
    {
      "epoch": 6.524669073405535,
      "grad_norm": 10.110445976257324,
      "learning_rate": 0.0001982657011722394,
      "loss": 7.5027,
      "step": 54220
    },
    {
      "epoch": 6.525872442839952,
      "grad_norm": 16.747638702392578,
      "learning_rate": 0.00019826499566601862,
      "loss": 7.4019,
      "step": 54230
    },
    {
      "epoch": 6.527075812274369,
      "grad_norm": 16.85960578918457,
      "learning_rate": 0.00019826429001758414,
      "loss": 7.4817,
      "step": 54240
    },
    {
      "epoch": 6.528279181708784,
      "grad_norm": 18.80135154724121,
      "learning_rate": 0.00019826358422693697,
      "loss": 7.4287,
      "step": 54250
    },
    {
      "epoch": 6.529482551143201,
      "grad_norm": 24.110595703125,
      "learning_rate": 0.0001982628782940781,
      "loss": 7.3289,
      "step": 54260
    },
    {
      "epoch": 6.530685920577618,
      "grad_norm": 39.86370849609375,
      "learning_rate": 0.0001982621722190086,
      "loss": 7.2644,
      "step": 54270
    },
    {
      "epoch": 6.5318892900120336,
      "grad_norm": 13.390047073364258,
      "learning_rate": 0.0001982614660017295,
      "loss": 7.4229,
      "step": 54280
    },
    {
      "epoch": 6.53309265944645,
      "grad_norm": 13.057512283325195,
      "learning_rate": 0.00019826075964224179,
      "loss": 7.381,
      "step": 54290
    },
    {
      "epoch": 6.534296028880867,
      "grad_norm": 13.116692543029785,
      "learning_rate": 0.00019826005314054648,
      "loss": 7.3792,
      "step": 54300
    },
    {
      "epoch": 6.535499398315283,
      "grad_norm": 17.642637252807617,
      "learning_rate": 0.00019825934649664464,
      "loss": 7.2654,
      "step": 54310
    },
    {
      "epoch": 6.536702767749699,
      "grad_norm": 21.632186889648438,
      "learning_rate": 0.00019825863971053724,
      "loss": 7.3399,
      "step": 54320
    },
    {
      "epoch": 6.537906137184115,
      "grad_norm": 34.379337310791016,
      "learning_rate": 0.00019825793278222534,
      "loss": 7.1887,
      "step": 54330
    },
    {
      "epoch": 6.539109506618532,
      "grad_norm": 20.180341720581055,
      "learning_rate": 0.00019825722571170998,
      "loss": 7.3595,
      "step": 54340
    },
    {
      "epoch": 6.5403128760529485,
      "grad_norm": 30.102216720581055,
      "learning_rate": 0.00019825651849899212,
      "loss": 7.1705,
      "step": 54350
    },
    {
      "epoch": 6.541516245487364,
      "grad_norm": 28.891704559326172,
      "learning_rate": 0.00019825581114407283,
      "loss": 7.4064,
      "step": 54360
    },
    {
      "epoch": 6.542719614921781,
      "grad_norm": 58.338565826416016,
      "learning_rate": 0.00019825510364695313,
      "loss": 7.1886,
      "step": 54370
    },
    {
      "epoch": 6.543922984356198,
      "grad_norm": 74.21753692626953,
      "learning_rate": 0.00019825439600763402,
      "loss": 7.2705,
      "step": 54380
    },
    {
      "epoch": 6.5451263537906135,
      "grad_norm": 92.1325454711914,
      "learning_rate": 0.00019825368822611657,
      "loss": 7.1965,
      "step": 54390
    },
    {
      "epoch": 6.54632972322503,
      "grad_norm": 27.837526321411133,
      "learning_rate": 0.00019825298030240173,
      "loss": 7.1966,
      "step": 54400
    },
    {
      "epoch": 6.547533092659447,
      "grad_norm": 25.98700523376465,
      "learning_rate": 0.00019825227223649062,
      "loss": 7.2939,
      "step": 54410
    },
    {
      "epoch": 6.548736462093863,
      "grad_norm": 45.24747848510742,
      "learning_rate": 0.00019825156402838417,
      "loss": 7.1676,
      "step": 54420
    },
    {
      "epoch": 6.549939831528279,
      "grad_norm": 14.43458080291748,
      "learning_rate": 0.00019825085567808345,
      "loss": 7.2205,
      "step": 54430
    },
    {
      "epoch": 6.551143200962695,
      "grad_norm": 14.4967041015625,
      "learning_rate": 0.00019825014718558953,
      "loss": 7.1258,
      "step": 54440
    },
    {
      "epoch": 6.552346570397112,
      "grad_norm": 9.712308883666992,
      "learning_rate": 0.00019824943855090336,
      "loss": 7.1183,
      "step": 54450
    },
    {
      "epoch": 6.5535499398315284,
      "grad_norm": 11.7882661819458,
      "learning_rate": 0.000198248729774026,
      "loss": 7.1117,
      "step": 54460
    },
    {
      "epoch": 6.554753309265944,
      "grad_norm": 24.683195114135742,
      "learning_rate": 0.00019824802085495846,
      "loss": 7.0663,
      "step": 54470
    },
    {
      "epoch": 6.555956678700361,
      "grad_norm": 26.272802352905273,
      "learning_rate": 0.0001982473117937018,
      "loss": 7.1405,
      "step": 54480
    },
    {
      "epoch": 6.557160048134778,
      "grad_norm": 15.428451538085938,
      "learning_rate": 0.000198246602590257,
      "loss": 7.117,
      "step": 54490
    },
    {
      "epoch": 6.558363417569193,
      "grad_norm": 20.457744598388672,
      "learning_rate": 0.00019824589324462513,
      "loss": 7.2366,
      "step": 54500
    },
    {
      "epoch": 6.55956678700361,
      "grad_norm": 6.069239616394043,
      "learning_rate": 0.00019824518375680716,
      "loss": 7.1005,
      "step": 54510
    },
    {
      "epoch": 6.560770156438027,
      "grad_norm": 8.31113052368164,
      "learning_rate": 0.0001982444741268042,
      "loss": 7.0395,
      "step": 54520
    },
    {
      "epoch": 6.5619735258724425,
      "grad_norm": 45.817283630371094,
      "learning_rate": 0.0001982437643546172,
      "loss": 7.11,
      "step": 54530
    },
    {
      "epoch": 6.563176895306859,
      "grad_norm": 46.114253997802734,
      "learning_rate": 0.00019824305444024725,
      "loss": 7.0689,
      "step": 54540
    },
    {
      "epoch": 6.564380264741276,
      "grad_norm": 190.5920867919922,
      "learning_rate": 0.0001982423443836953,
      "loss": 7.2594,
      "step": 54550
    },
    {
      "epoch": 6.565583634175692,
      "grad_norm": 215.85943603515625,
      "learning_rate": 0.00019824163418496245,
      "loss": 7.1505,
      "step": 54560
    },
    {
      "epoch": 6.566787003610108,
      "grad_norm": 284.28448486328125,
      "learning_rate": 0.00019824092384404968,
      "loss": 7.109,
      "step": 54570
    },
    {
      "epoch": 6.567990373044525,
      "grad_norm": 51.15251159667969,
      "learning_rate": 0.00019824021336095808,
      "loss": 7.1543,
      "step": 54580
    },
    {
      "epoch": 6.569193742478941,
      "grad_norm": 1200.6505126953125,
      "learning_rate": 0.00019823950273568862,
      "loss": 7.0203,
      "step": 54590
    },
    {
      "epoch": 6.5703971119133575,
      "grad_norm": 210.41134643554688,
      "learning_rate": 0.0001982387919682423,
      "loss": 7.1171,
      "step": 54600
    },
    {
      "epoch": 6.571600481347774,
      "grad_norm": 97.78995513916016,
      "learning_rate": 0.00019823808105862023,
      "loss": 7.2416,
      "step": 54610
    },
    {
      "epoch": 6.57280385078219,
      "grad_norm": 53.442588806152344,
      "learning_rate": 0.00019823737000682342,
      "loss": 7.1407,
      "step": 54620
    },
    {
      "epoch": 6.574007220216607,
      "grad_norm": 25.63939094543457,
      "learning_rate": 0.00019823665881285284,
      "loss": 7.4223,
      "step": 54630
    },
    {
      "epoch": 6.575210589651023,
      "grad_norm": 40.05807113647461,
      "learning_rate": 0.00019823594747670959,
      "loss": 7.5493,
      "step": 54640
    },
    {
      "epoch": 6.576413959085439,
      "grad_norm": 25.501604080200195,
      "learning_rate": 0.00019823523599839466,
      "loss": 7.4365,
      "step": 54650
    },
    {
      "epoch": 6.577617328519856,
      "grad_norm": 18.687335968017578,
      "learning_rate": 0.0001982345243779091,
      "loss": 7.6404,
      "step": 54660
    },
    {
      "epoch": 6.578820697954272,
      "grad_norm": 13.596409797668457,
      "learning_rate": 0.00019823381261525392,
      "loss": 7.5357,
      "step": 54670
    },
    {
      "epoch": 6.580024067388688,
      "grad_norm": 32.96208953857422,
      "learning_rate": 0.00019823310071043015,
      "loss": 7.6519,
      "step": 54680
    },
    {
      "epoch": 6.581227436823105,
      "grad_norm": 16.275880813598633,
      "learning_rate": 0.00019823238866343887,
      "loss": 7.5876,
      "step": 54690
    },
    {
      "epoch": 6.582430806257521,
      "grad_norm": 16.6007137298584,
      "learning_rate": 0.00019823167647428104,
      "loss": 7.5432,
      "step": 54700
    },
    {
      "epoch": 6.583634175691937,
      "grad_norm": 5.2712297439575195,
      "learning_rate": 0.00019823096414295772,
      "loss": 7.5932,
      "step": 54710
    },
    {
      "epoch": 6.584837545126354,
      "grad_norm": 6.463351249694824,
      "learning_rate": 0.00019823025166946994,
      "loss": 7.4931,
      "step": 54720
    },
    {
      "epoch": 6.58604091456077,
      "grad_norm": 34.08944320678711,
      "learning_rate": 0.00019822953905381877,
      "loss": 7.5064,
      "step": 54730
    },
    {
      "epoch": 6.587244283995187,
      "grad_norm": 171.5081787109375,
      "learning_rate": 0.00019822882629600517,
      "loss": 7.7336,
      "step": 54740
    },
    {
      "epoch": 6.588447653429603,
      "grad_norm": 23.387256622314453,
      "learning_rate": 0.0001982281133960302,
      "loss": 7.7548,
      "step": 54750
    },
    {
      "epoch": 6.589651022864019,
      "grad_norm": 7.655777931213379,
      "learning_rate": 0.0001982274003538949,
      "loss": 7.7193,
      "step": 54760
    },
    {
      "epoch": 6.590854392298436,
      "grad_norm": 27.844728469848633,
      "learning_rate": 0.00019822668716960032,
      "loss": 7.6539,
      "step": 54770
    },
    {
      "epoch": 6.5920577617328515,
      "grad_norm": 7.127274513244629,
      "learning_rate": 0.00019822597384314746,
      "loss": 7.4987,
      "step": 54780
    },
    {
      "epoch": 6.593261131167268,
      "grad_norm": 73.10989379882812,
      "learning_rate": 0.00019822526037453737,
      "loss": 7.3229,
      "step": 54790
    },
    {
      "epoch": 6.594464500601685,
      "grad_norm": 63.49520492553711,
      "learning_rate": 0.00019822454676377108,
      "loss": 7.2696,
      "step": 54800
    },
    {
      "epoch": 6.595667870036101,
      "grad_norm": 672.5573120117188,
      "learning_rate": 0.00019822383301084965,
      "loss": 7.5097,
      "step": 54810
    },
    {
      "epoch": 6.596871239470517,
      "grad_norm": 62.491119384765625,
      "learning_rate": 0.000198223119115774,
      "loss": 8.6739,
      "step": 54820
    },
    {
      "epoch": 6.598074608904934,
      "grad_norm": 61.918418884277344,
      "learning_rate": 0.00019822240507854532,
      "loss": 8.3629,
      "step": 54830
    },
    {
      "epoch": 6.59927797833935,
      "grad_norm": 15.558662414550781,
      "learning_rate": 0.00019822169089916454,
      "loss": 7.7627,
      "step": 54840
    },
    {
      "epoch": 6.6004813477737665,
      "grad_norm": 15.076722145080566,
      "learning_rate": 0.00019822097657763274,
      "loss": 7.6219,
      "step": 54850
    },
    {
      "epoch": 6.601684717208183,
      "grad_norm": 80.84474182128906,
      "learning_rate": 0.00019822026211395092,
      "loss": 7.5262,
      "step": 54860
    },
    {
      "epoch": 6.602888086642599,
      "grad_norm": 226.2410888671875,
      "learning_rate": 0.00019821954750812013,
      "loss": 7.389,
      "step": 54870
    },
    {
      "epoch": 6.604091456077016,
      "grad_norm": 663.0928955078125,
      "learning_rate": 0.0001982188327601414,
      "loss": 7.3833,
      "step": 54880
    },
    {
      "epoch": 6.605294825511432,
      "grad_norm": 57.77903747558594,
      "learning_rate": 0.00019821811787001578,
      "loss": 7.4233,
      "step": 54890
    },
    {
      "epoch": 6.606498194945848,
      "grad_norm": 1358.3236083984375,
      "learning_rate": 0.0001982174028377443,
      "loss": 7.1157,
      "step": 54900
    },
    {
      "epoch": 6.607701564380265,
      "grad_norm": 316.8460998535156,
      "learning_rate": 0.00019821668766332797,
      "loss": 7.3714,
      "step": 54910
    },
    {
      "epoch": 6.6089049338146815,
      "grad_norm": 197.40431213378906,
      "learning_rate": 0.00019821597234676787,
      "loss": 7.3758,
      "step": 54920
    },
    {
      "epoch": 6.610108303249097,
      "grad_norm": 28.91938591003418,
      "learning_rate": 0.000198215256888065,
      "loss": 7.4244,
      "step": 54930
    },
    {
      "epoch": 6.611311672683514,
      "grad_norm": 206.230224609375,
      "learning_rate": 0.00019821454128722042,
      "loss": 7.4257,
      "step": 54940
    },
    {
      "epoch": 6.612515042117931,
      "grad_norm": 34.72187042236328,
      "learning_rate": 0.0001982138255442351,
      "loss": 7.4846,
      "step": 54950
    },
    {
      "epoch": 6.613718411552346,
      "grad_norm": 26.24178123474121,
      "learning_rate": 0.00019821310965911017,
      "loss": 7.5512,
      "step": 54960
    },
    {
      "epoch": 6.614921780986763,
      "grad_norm": 81.10030364990234,
      "learning_rate": 0.0001982123936318466,
      "loss": 7.2925,
      "step": 54970
    },
    {
      "epoch": 6.61612515042118,
      "grad_norm": 23.236541748046875,
      "learning_rate": 0.00019821167746244547,
      "loss": 7.2629,
      "step": 54980
    },
    {
      "epoch": 6.617328519855596,
      "grad_norm": 228.65492248535156,
      "learning_rate": 0.0001982109611509078,
      "loss": 7.2236,
      "step": 54990
    },
    {
      "epoch": 6.618531889290012,
      "grad_norm": 12.98916244506836,
      "learning_rate": 0.00019821024469723458,
      "loss": 7.2756,
      "step": 55000
    },
    {
      "epoch": 6.619735258724428,
      "grad_norm": 94.37308502197266,
      "learning_rate": 0.00019820952810142693,
      "loss": 7.1736,
      "step": 55010
    },
    {
      "epoch": 6.620938628158845,
      "grad_norm": 80.3369140625,
      "learning_rate": 0.00019820881136348582,
      "loss": 7.2707,
      "step": 55020
    },
    {
      "epoch": 6.622141997593261,
      "grad_norm": 118.46678924560547,
      "learning_rate": 0.00019820809448341234,
      "loss": 7.1903,
      "step": 55030
    },
    {
      "epoch": 6.623345367027677,
      "grad_norm": 115.43425750732422,
      "learning_rate": 0.00019820737746120747,
      "loss": 7.1323,
      "step": 55040
    },
    {
      "epoch": 6.624548736462094,
      "grad_norm": 32.32550048828125,
      "learning_rate": 0.00019820666029687228,
      "loss": 7.3732,
      "step": 55050
    },
    {
      "epoch": 6.6257521058965105,
      "grad_norm": 361.280517578125,
      "learning_rate": 0.0001982059429904078,
      "loss": 7.4074,
      "step": 55060
    },
    {
      "epoch": 6.626955475330926,
      "grad_norm": 57.889488220214844,
      "learning_rate": 0.00019820522554181511,
      "loss": 7.3973,
      "step": 55070
    },
    {
      "epoch": 6.628158844765343,
      "grad_norm": 23.05805778503418,
      "learning_rate": 0.0001982045079510952,
      "loss": 7.301,
      "step": 55080
    },
    {
      "epoch": 6.62936221419976,
      "grad_norm": 41.13179397583008,
      "learning_rate": 0.0001982037902182491,
      "loss": 7.24,
      "step": 55090
    },
    {
      "epoch": 6.6305655836341755,
      "grad_norm": 12.71737003326416,
      "learning_rate": 0.00019820307234327792,
      "loss": 7.3102,
      "step": 55100
    },
    {
      "epoch": 6.631768953068592,
      "grad_norm": 25.804162979125977,
      "learning_rate": 0.0001982023543261826,
      "loss": 7.2828,
      "step": 55110
    },
    {
      "epoch": 6.632972322503008,
      "grad_norm": 18.82998275756836,
      "learning_rate": 0.00019820163616696422,
      "loss": 7.5072,
      "step": 55120
    },
    {
      "epoch": 6.634175691937425,
      "grad_norm": 58.981834411621094,
      "learning_rate": 0.00019820091786562386,
      "loss": 7.5733,
      "step": 55130
    },
    {
      "epoch": 6.635379061371841,
      "grad_norm": 102.47163391113281,
      "learning_rate": 0.0001982001994221625,
      "loss": 7.5286,
      "step": 55140
    },
    {
      "epoch": 6.636582430806257,
      "grad_norm": 173.20286560058594,
      "learning_rate": 0.00019819948083658125,
      "loss": 7.5001,
      "step": 55150
    },
    {
      "epoch": 6.637785800240674,
      "grad_norm": 59.042510986328125,
      "learning_rate": 0.00019819876210888106,
      "loss": 7.3809,
      "step": 55160
    },
    {
      "epoch": 6.6389891696750905,
      "grad_norm": 17.95256233215332,
      "learning_rate": 0.00019819804323906306,
      "loss": 7.3043,
      "step": 55170
    },
    {
      "epoch": 6.640192539109506,
      "grad_norm": 135.5763397216797,
      "learning_rate": 0.00019819732422712822,
      "loss": 7.3822,
      "step": 55180
    },
    {
      "epoch": 6.641395908543923,
      "grad_norm": 9.904345512390137,
      "learning_rate": 0.0001981966050730776,
      "loss": 7.3074,
      "step": 55190
    },
    {
      "epoch": 6.64259927797834,
      "grad_norm": 10.047538757324219,
      "learning_rate": 0.00019819588577691226,
      "loss": 7.1864,
      "step": 55200
    },
    {
      "epoch": 6.643802647412755,
      "grad_norm": 8.307875633239746,
      "learning_rate": 0.0001981951663386332,
      "loss": 7.1257,
      "step": 55210
    },
    {
      "epoch": 6.645006016847172,
      "grad_norm": 10.847360610961914,
      "learning_rate": 0.00019819444675824155,
      "loss": 7.1619,
      "step": 55220
    },
    {
      "epoch": 6.646209386281589,
      "grad_norm": 8.394771575927734,
      "learning_rate": 0.00019819372703573827,
      "loss": 7.3264,
      "step": 55230
    },
    {
      "epoch": 6.6474127557160045,
      "grad_norm": 26.60249900817871,
      "learning_rate": 0.0001981930071711244,
      "loss": 7.1023,
      "step": 55240
    },
    {
      "epoch": 6.648616125150421,
      "grad_norm": 103.9189453125,
      "learning_rate": 0.00019819228716440104,
      "loss": 7.2382,
      "step": 55250
    },
    {
      "epoch": 6.649819494584838,
      "grad_norm": 217.11172485351562,
      "learning_rate": 0.00019819156701556917,
      "loss": 7.2982,
      "step": 55260
    },
    {
      "epoch": 6.651022864019254,
      "grad_norm": 112.395751953125,
      "learning_rate": 0.00019819084672462986,
      "loss": 7.4447,
      "step": 55270
    },
    {
      "epoch": 6.65222623345367,
      "grad_norm": 36.52063751220703,
      "learning_rate": 0.00019819012629158415,
      "loss": 7.3282,
      "step": 55280
    },
    {
      "epoch": 6.653429602888087,
      "grad_norm": 1100.7315673828125,
      "learning_rate": 0.00019818940571643312,
      "loss": 7.2989,
      "step": 55290
    },
    {
      "epoch": 6.654632972322503,
      "grad_norm": 5.3416748046875,
      "learning_rate": 0.00019818868499917774,
      "loss": 7.3448,
      "step": 55300
    },
    {
      "epoch": 6.6558363417569195,
      "grad_norm": 3.8963637351989746,
      "learning_rate": 0.0001981879641398191,
      "loss": 7.393,
      "step": 55310
    },
    {
      "epoch": 6.657039711191336,
      "grad_norm": 20.22577667236328,
      "learning_rate": 0.00019818724313835825,
      "loss": 7.1391,
      "step": 55320
    },
    {
      "epoch": 6.658243080625752,
      "grad_norm": 23.58957862854004,
      "learning_rate": 0.0001981865219947962,
      "loss": 7.1837,
      "step": 55330
    },
    {
      "epoch": 6.659446450060169,
      "grad_norm": 6.533280372619629,
      "learning_rate": 0.000198185800709134,
      "loss": 7.359,
      "step": 55340
    },
    {
      "epoch": 6.6606498194945845,
      "grad_norm": 10.857223510742188,
      "learning_rate": 0.00019818507928137273,
      "loss": 7.2454,
      "step": 55350
    },
    {
      "epoch": 6.661853188929001,
      "grad_norm": 14.95755386352539,
      "learning_rate": 0.0001981843577115134,
      "loss": 7.1278,
      "step": 55360
    },
    {
      "epoch": 6.663056558363418,
      "grad_norm": 40.79299545288086,
      "learning_rate": 0.00019818363599955708,
      "loss": 7.4723,
      "step": 55370
    },
    {
      "epoch": 6.664259927797834,
      "grad_norm": 45.469818115234375,
      "learning_rate": 0.00019818291414550477,
      "loss": 7.2531,
      "step": 55380
    },
    {
      "epoch": 6.66546329723225,
      "grad_norm": 17.84055519104004,
      "learning_rate": 0.00019818219214935755,
      "loss": 7.3298,
      "step": 55390
    },
    {
      "epoch": 6.666666666666667,
      "grad_norm": 5.900567531585693,
      "learning_rate": 0.00019818147001111647,
      "loss": 7.31,
      "step": 55400
    },
    {
      "epoch": 6.667870036101083,
      "grad_norm": 5.828828811645508,
      "learning_rate": 0.00019818074773078257,
      "loss": 7.254,
      "step": 55410
    },
    {
      "epoch": 6.669073405535499,
      "grad_norm": 7.5560736656188965,
      "learning_rate": 0.00019818002530835687,
      "loss": 7.0643,
      "step": 55420
    },
    {
      "epoch": 6.670276774969916,
      "grad_norm": 7.557502269744873,
      "learning_rate": 0.00019817930274384043,
      "loss": 7.1738,
      "step": 55430
    },
    {
      "epoch": 6.671480144404332,
      "grad_norm": 81.7336196899414,
      "learning_rate": 0.0001981785800372343,
      "loss": 7.1299,
      "step": 55440
    },
    {
      "epoch": 6.672683513838749,
      "grad_norm": 5.938595294952393,
      "learning_rate": 0.00019817785718853953,
      "loss": 7.1194,
      "step": 55450
    },
    {
      "epoch": 6.673886883273164,
      "grad_norm": 12.312219619750977,
      "learning_rate": 0.00019817713419775717,
      "loss": 7.1877,
      "step": 55460
    },
    {
      "epoch": 6.675090252707581,
      "grad_norm": 47.27899932861328,
      "learning_rate": 0.00019817641106488826,
      "loss": 7.2036,
      "step": 55470
    },
    {
      "epoch": 6.676293622141998,
      "grad_norm": 77.00691986083984,
      "learning_rate": 0.00019817568778993382,
      "loss": 7.2,
      "step": 55480
    },
    {
      "epoch": 6.6774969915764135,
      "grad_norm": 35.118560791015625,
      "learning_rate": 0.00019817496437289493,
      "loss": 7.3714,
      "step": 55490
    },
    {
      "epoch": 6.67870036101083,
      "grad_norm": 23.781373977661133,
      "learning_rate": 0.00019817424081377264,
      "loss": 7.2999,
      "step": 55500
    },
    {
      "epoch": 6.679903730445247,
      "grad_norm": 605.828857421875,
      "learning_rate": 0.00019817351711256796,
      "loss": 7.4063,
      "step": 55510
    },
    {
      "epoch": 6.681107099879663,
      "grad_norm": 120.28981018066406,
      "learning_rate": 0.000198172793269282,
      "loss": 7.3056,
      "step": 55520
    },
    {
      "epoch": 6.682310469314079,
      "grad_norm": 14.749140739440918,
      "learning_rate": 0.00019817206928391575,
      "loss": 7.1653,
      "step": 55530
    },
    {
      "epoch": 6.683513838748496,
      "grad_norm": 19.87139129638672,
      "learning_rate": 0.00019817134515647027,
      "loss": 7.1294,
      "step": 55540
    },
    {
      "epoch": 6.684717208182912,
      "grad_norm": 51.131141662597656,
      "learning_rate": 0.00019817062088694663,
      "loss": 7.0697,
      "step": 55550
    },
    {
      "epoch": 6.6859205776173285,
      "grad_norm": 19.66100311279297,
      "learning_rate": 0.00019816989647534588,
      "loss": 7.2484,
      "step": 55560
    },
    {
      "epoch": 6.687123947051745,
      "grad_norm": 59.18501663208008,
      "learning_rate": 0.000198169171921669,
      "loss": 7.1588,
      "step": 55570
    },
    {
      "epoch": 6.688327316486161,
      "grad_norm": 12.373442649841309,
      "learning_rate": 0.00019816844722591714,
      "loss": 7.129,
      "step": 55580
    },
    {
      "epoch": 6.689530685920578,
      "grad_norm": 46.0589485168457,
      "learning_rate": 0.00019816772238809128,
      "loss": 7.0885,
      "step": 55590
    },
    {
      "epoch": 6.690734055354994,
      "grad_norm": 35.988494873046875,
      "learning_rate": 0.00019816699740819247,
      "loss": 7.2061,
      "step": 55600
    },
    {
      "epoch": 6.69193742478941,
      "grad_norm": 30.199434280395508,
      "learning_rate": 0.0001981662722862218,
      "loss": 7.152,
      "step": 55610
    },
    {
      "epoch": 6.693140794223827,
      "grad_norm": 26.043407440185547,
      "learning_rate": 0.0001981655470221803,
      "loss": 7.0291,
      "step": 55620
    },
    {
      "epoch": 6.6943441636582435,
      "grad_norm": 101.2144546508789,
      "learning_rate": 0.000198164821616069,
      "loss": 7.2461,
      "step": 55630
    },
    {
      "epoch": 6.695547533092659,
      "grad_norm": 17.652393341064453,
      "learning_rate": 0.000198164096067889,
      "loss": 7.1416,
      "step": 55640
    },
    {
      "epoch": 6.696750902527076,
      "grad_norm": 42.61686325073242,
      "learning_rate": 0.0001981633703776413,
      "loss": 7.1776,
      "step": 55650
    },
    {
      "epoch": 6.697954271961493,
      "grad_norm": 151.83102416992188,
      "learning_rate": 0.000198162644545327,
      "loss": 7.1747,
      "step": 55660
    },
    {
      "epoch": 6.699157641395908,
      "grad_norm": 26.311147689819336,
      "learning_rate": 0.00019816191857094704,
      "loss": 6.9852,
      "step": 55670
    },
    {
      "epoch": 6.700361010830325,
      "grad_norm": 34.56694793701172,
      "learning_rate": 0.00019816119245450259,
      "loss": 6.9377,
      "step": 55680
    },
    {
      "epoch": 6.701564380264742,
      "grad_norm": 38.924007415771484,
      "learning_rate": 0.00019816046619599466,
      "loss": 7.0062,
      "step": 55690
    },
    {
      "epoch": 6.702767749699158,
      "grad_norm": 89.71614074707031,
      "learning_rate": 0.00019815973979542433,
      "loss": 7.1952,
      "step": 55700
    },
    {
      "epoch": 6.703971119133574,
      "grad_norm": 95.2943115234375,
      "learning_rate": 0.00019815901325279258,
      "loss": 7.1886,
      "step": 55710
    },
    {
      "epoch": 6.70517448856799,
      "grad_norm": 88.1927719116211,
      "learning_rate": 0.0001981582865681005,
      "loss": 7.0989,
      "step": 55720
    },
    {
      "epoch": 6.706377858002407,
      "grad_norm": 95.01557922363281,
      "learning_rate": 0.00019815755974134918,
      "loss": 7.063,
      "step": 55730
    },
    {
      "epoch": 6.707581227436823,
      "grad_norm": 15.927534103393555,
      "learning_rate": 0.0001981568327725396,
      "loss": 7.0364,
      "step": 55740
    },
    {
      "epoch": 6.708784596871239,
      "grad_norm": 19.66518211364746,
      "learning_rate": 0.00019815610566167288,
      "loss": 7.1141,
      "step": 55750
    },
    {
      "epoch": 6.709987966305656,
      "grad_norm": 108.02747344970703,
      "learning_rate": 0.00019815537840875002,
      "loss": 7.1128,
      "step": 55760
    },
    {
      "epoch": 6.7111913357400725,
      "grad_norm": 20.291255950927734,
      "learning_rate": 0.00019815465101377208,
      "loss": 7.1718,
      "step": 55770
    },
    {
      "epoch": 6.712394705174488,
      "grad_norm": 67.08749389648438,
      "learning_rate": 0.00019815392347674019,
      "loss": 7.0628,
      "step": 55780
    },
    {
      "epoch": 6.713598074608905,
      "grad_norm": 53.44804000854492,
      "learning_rate": 0.00019815319579765528,
      "loss": 6.9924,
      "step": 55790
    },
    {
      "epoch": 6.714801444043322,
      "grad_norm": 42.22039794921875,
      "learning_rate": 0.00019815246797651848,
      "loss": 7.0726,
      "step": 55800
    },
    {
      "epoch": 6.7160048134777375,
      "grad_norm": 77.47686767578125,
      "learning_rate": 0.0001981517400133308,
      "loss": 7.0795,
      "step": 55810
    },
    {
      "epoch": 6.717208182912154,
      "grad_norm": 125.3265380859375,
      "learning_rate": 0.0001981510119080934,
      "loss": 7.2251,
      "step": 55820
    },
    {
      "epoch": 6.71841155234657,
      "grad_norm": 70.47822570800781,
      "learning_rate": 0.00019815028366080716,
      "loss": 7.2634,
      "step": 55830
    },
    {
      "epoch": 6.719614921780987,
      "grad_norm": 96.91410064697266,
      "learning_rate": 0.00019814955527147325,
      "loss": 6.9081,
      "step": 55840
    },
    {
      "epoch": 6.720818291215403,
      "grad_norm": 14.184112548828125,
      "learning_rate": 0.00019814882674009274,
      "loss": 6.9851,
      "step": 55850
    },
    {
      "epoch": 6.722021660649819,
      "grad_norm": 24.29714012145996,
      "learning_rate": 0.00019814809806666662,
      "loss": 6.9962,
      "step": 55860
    },
    {
      "epoch": 6.723225030084236,
      "grad_norm": 12.130228042602539,
      "learning_rate": 0.00019814736925119597,
      "loss": 6.9878,
      "step": 55870
    },
    {
      "epoch": 6.7244283995186525,
      "grad_norm": 127.275634765625,
      "learning_rate": 0.00019814664029368183,
      "loss": 6.9878,
      "step": 55880
    },
    {
      "epoch": 6.725631768953068,
      "grad_norm": 60.52707290649414,
      "learning_rate": 0.0001981459111941253,
      "loss": 7.0435,
      "step": 55890
    },
    {
      "epoch": 6.726835138387485,
      "grad_norm": 33.796531677246094,
      "learning_rate": 0.0001981451819525274,
      "loss": 7.0629,
      "step": 55900
    },
    {
      "epoch": 6.728038507821902,
      "grad_norm": 24.30449676513672,
      "learning_rate": 0.0001981444525688892,
      "loss": 7.0058,
      "step": 55910
    },
    {
      "epoch": 6.729241877256317,
      "grad_norm": 5.832573413848877,
      "learning_rate": 0.0001981437230432117,
      "loss": 7.1164,
      "step": 55920
    },
    {
      "epoch": 6.730445246690734,
      "grad_norm": 10.127008438110352,
      "learning_rate": 0.000198142993375496,
      "loss": 6.9855,
      "step": 55930
    },
    {
      "epoch": 6.731648616125151,
      "grad_norm": 14.363208770751953,
      "learning_rate": 0.0001981422635657432,
      "loss": 6.9767,
      "step": 55940
    },
    {
      "epoch": 6.7328519855595665,
      "grad_norm": 12.817060470581055,
      "learning_rate": 0.0001981415336139543,
      "loss": 6.9231,
      "step": 55950
    },
    {
      "epoch": 6.734055354993983,
      "grad_norm": 14.756559371948242,
      "learning_rate": 0.0001981408035201304,
      "loss": 7.0664,
      "step": 55960
    },
    {
      "epoch": 6.7352587244284,
      "grad_norm": 31.317472457885742,
      "learning_rate": 0.0001981400732842725,
      "loss": 7.2129,
      "step": 55970
    },
    {
      "epoch": 6.736462093862816,
      "grad_norm": 17.89832305908203,
      "learning_rate": 0.00019813934290638168,
      "loss": 7.0897,
      "step": 55980
    },
    {
      "epoch": 6.737665463297232,
      "grad_norm": 13.871992111206055,
      "learning_rate": 0.000198138612386459,
      "loss": 6.9262,
      "step": 55990
    },
    {
      "epoch": 6.738868832731649,
      "grad_norm": 40.8659553527832,
      "learning_rate": 0.0001981378817245055,
      "loss": 6.9915,
      "step": 56000
    },
    {
      "epoch": 6.740072202166065,
      "grad_norm": 18.634611129760742,
      "learning_rate": 0.00019813715092052229,
      "loss": 7.0372,
      "step": 56010
    },
    {
      "epoch": 6.7412755716004815,
      "grad_norm": 56.851539611816406,
      "learning_rate": 0.00019813641997451037,
      "loss": 6.9743,
      "step": 56020
    },
    {
      "epoch": 6.742478941034898,
      "grad_norm": 13.186022758483887,
      "learning_rate": 0.00019813568888647084,
      "loss": 6.9053,
      "step": 56030
    },
    {
      "epoch": 6.743682310469314,
      "grad_norm": 12.31736946105957,
      "learning_rate": 0.00019813495765640474,
      "loss": 6.7843,
      "step": 56040
    },
    {
      "epoch": 6.744885679903731,
      "grad_norm": 38.10343551635742,
      "learning_rate": 0.0001981342262843131,
      "loss": 6.8849,
      "step": 56050
    },
    {
      "epoch": 6.7460890493381465,
      "grad_norm": 12.185663223266602,
      "learning_rate": 0.00019813349477019703,
      "loss": 6.8924,
      "step": 56060
    },
    {
      "epoch": 6.747292418772563,
      "grad_norm": 202.5308380126953,
      "learning_rate": 0.00019813276311405757,
      "loss": 6.9407,
      "step": 56070
    },
    {
      "epoch": 6.74849578820698,
      "grad_norm": 31.466867446899414,
      "learning_rate": 0.00019813203131589577,
      "loss": 6.9976,
      "step": 56080
    },
    {
      "epoch": 6.749699157641396,
      "grad_norm": 36.996185302734375,
      "learning_rate": 0.0001981312993757127,
      "loss": 6.8978,
      "step": 56090
    },
    {
      "epoch": 6.750902527075812,
      "grad_norm": 14.816555976867676,
      "learning_rate": 0.00019813056729350937,
      "loss": 6.8933,
      "step": 56100
    },
    {
      "epoch": 6.752105896510229,
      "grad_norm": 8.242816925048828,
      "learning_rate": 0.00019812983506928693,
      "loss": 6.9547,
      "step": 56110
    },
    {
      "epoch": 6.753309265944645,
      "grad_norm": 11.20868968963623,
      "learning_rate": 0.00019812910270304633,
      "loss": 6.7677,
      "step": 56120
    },
    {
      "epoch": 6.754512635379061,
      "grad_norm": 13.01804256439209,
      "learning_rate": 0.00019812837019478875,
      "loss": 6.8561,
      "step": 56130
    },
    {
      "epoch": 6.755716004813478,
      "grad_norm": 12.331233978271484,
      "learning_rate": 0.00019812763754451518,
      "loss": 6.8899,
      "step": 56140
    },
    {
      "epoch": 6.756919374247894,
      "grad_norm": 10.18593978881836,
      "learning_rate": 0.0001981269047522267,
      "loss": 6.8891,
      "step": 56150
    },
    {
      "epoch": 6.758122743682311,
      "grad_norm": 15.321721076965332,
      "learning_rate": 0.00019812617181792433,
      "loss": 7.0075,
      "step": 56160
    },
    {
      "epoch": 6.759326113116726,
      "grad_norm": 20.115995407104492,
      "learning_rate": 0.00019812543874160918,
      "loss": 6.9075,
      "step": 56170
    },
    {
      "epoch": 6.760529482551143,
      "grad_norm": 25.264480590820312,
      "learning_rate": 0.0001981247055232823,
      "loss": 6.9353,
      "step": 56180
    },
    {
      "epoch": 6.76173285198556,
      "grad_norm": 9.208985328674316,
      "learning_rate": 0.00019812397216294474,
      "loss": 7.0104,
      "step": 56190
    },
    {
      "epoch": 6.7629362214199755,
      "grad_norm": 24.12775421142578,
      "learning_rate": 0.00019812323866059758,
      "loss": 6.9505,
      "step": 56200
    },
    {
      "epoch": 6.764139590854392,
      "grad_norm": 25.862085342407227,
      "learning_rate": 0.00019812250501624184,
      "loss": 6.9637,
      "step": 56210
    },
    {
      "epoch": 6.765342960288809,
      "grad_norm": 108.80167388916016,
      "learning_rate": 0.0001981217712298786,
      "loss": 7.3221,
      "step": 56220
    },
    {
      "epoch": 6.766546329723225,
      "grad_norm": 33.10988998413086,
      "learning_rate": 0.00019812103730150894,
      "loss": 7.0247,
      "step": 56230
    },
    {
      "epoch": 6.767749699157641,
      "grad_norm": 16.83145523071289,
      "learning_rate": 0.00019812030323113395,
      "loss": 6.9151,
      "step": 56240
    },
    {
      "epoch": 6.768953068592058,
      "grad_norm": 13.506775856018066,
      "learning_rate": 0.00019811956901875463,
      "loss": 6.9167,
      "step": 56250
    },
    {
      "epoch": 6.770156438026474,
      "grad_norm": 16.81162452697754,
      "learning_rate": 0.0001981188346643721,
      "loss": 6.9284,
      "step": 56260
    },
    {
      "epoch": 6.7713598074608905,
      "grad_norm": 10.348248481750488,
      "learning_rate": 0.00019811810016798735,
      "loss": 6.943,
      "step": 56270
    },
    {
      "epoch": 6.772563176895307,
      "grad_norm": 8.620027542114258,
      "learning_rate": 0.0001981173655296015,
      "loss": 6.8358,
      "step": 56280
    },
    {
      "epoch": 6.773766546329723,
      "grad_norm": 9.598176956176758,
      "learning_rate": 0.00019811663074921558,
      "loss": 6.8134,
      "step": 56290
    },
    {
      "epoch": 6.77496991576414,
      "grad_norm": 28.866487503051758,
      "learning_rate": 0.00019811589582683067,
      "loss": 6.8745,
      "step": 56300
    },
    {
      "epoch": 6.776173285198556,
      "grad_norm": 90.2942886352539,
      "learning_rate": 0.00019811516076244788,
      "loss": 6.8092,
      "step": 56310
    },
    {
      "epoch": 6.777376654632972,
      "grad_norm": 42.91451644897461,
      "learning_rate": 0.0001981144255560682,
      "loss": 7.0118,
      "step": 56320
    },
    {
      "epoch": 6.778580024067389,
      "grad_norm": 48.719337463378906,
      "learning_rate": 0.00019811369020769274,
      "loss": 7.3572,
      "step": 56330
    },
    {
      "epoch": 6.7797833935018055,
      "grad_norm": 9.14430046081543,
      "learning_rate": 0.0001981129547173225,
      "loss": 7.5251,
      "step": 56340
    },
    {
      "epoch": 6.780986762936221,
      "grad_norm": 4.427257061004639,
      "learning_rate": 0.0001981122190849586,
      "loss": 7.3696,
      "step": 56350
    },
    {
      "epoch": 6.782190132370638,
      "grad_norm": 3.3529253005981445,
      "learning_rate": 0.00019811148331060215,
      "loss": 7.3322,
      "step": 56360
    },
    {
      "epoch": 6.783393501805055,
      "grad_norm": 8.207467079162598,
      "learning_rate": 0.0001981107473942541,
      "loss": 7.1657,
      "step": 56370
    },
    {
      "epoch": 6.78459687123947,
      "grad_norm": 7.812586784362793,
      "learning_rate": 0.0001981100113359156,
      "loss": 7.1968,
      "step": 56380
    },
    {
      "epoch": 6.785800240673887,
      "grad_norm": 4.500979423522949,
      "learning_rate": 0.0001981092751355877,
      "loss": 7.0436,
      "step": 56390
    },
    {
      "epoch": 6.787003610108303,
      "grad_norm": 6.167128086090088,
      "learning_rate": 0.00019810853879327148,
      "loss": 7.0426,
      "step": 56400
    },
    {
      "epoch": 6.78820697954272,
      "grad_norm": 21.189769744873047,
      "learning_rate": 0.00019810780230896793,
      "loss": 7.011,
      "step": 56410
    },
    {
      "epoch": 6.789410348977136,
      "grad_norm": 8.848641395568848,
      "learning_rate": 0.00019810706568267817,
      "loss": 6.946,
      "step": 56420
    },
    {
      "epoch": 6.790613718411552,
      "grad_norm": 23.617721557617188,
      "learning_rate": 0.0001981063289144033,
      "loss": 7.0302,
      "step": 56430
    },
    {
      "epoch": 6.791817087845969,
      "grad_norm": 12.344385147094727,
      "learning_rate": 0.00019810559200414437,
      "loss": 6.9754,
      "step": 56440
    },
    {
      "epoch": 6.793020457280385,
      "grad_norm": 46.22349548339844,
      "learning_rate": 0.00019810485495190235,
      "loss": 6.967,
      "step": 56450
    },
    {
      "epoch": 6.794223826714801,
      "grad_norm": 36.92741394042969,
      "learning_rate": 0.00019810411775767842,
      "loss": 7.0277,
      "step": 56460
    },
    {
      "epoch": 6.795427196149218,
      "grad_norm": 40.07979965209961,
      "learning_rate": 0.00019810338042147363,
      "loss": 7.062,
      "step": 56470
    },
    {
      "epoch": 6.7966305655836345,
      "grad_norm": 10.478665351867676,
      "learning_rate": 0.00019810264294328902,
      "loss": 7.1715,
      "step": 56480
    },
    {
      "epoch": 6.79783393501805,
      "grad_norm": 13.519403457641602,
      "learning_rate": 0.00019810190532312566,
      "loss": 7.2005,
      "step": 56490
    },
    {
      "epoch": 6.799037304452467,
      "grad_norm": 61.327674865722656,
      "learning_rate": 0.0001981011675609846,
      "loss": 7.0186,
      "step": 56500
    },
    {
      "epoch": 6.800240673886883,
      "grad_norm": 24.989665985107422,
      "learning_rate": 0.00019810042965686696,
      "loss": 6.8714,
      "step": 56510
    },
    {
      "epoch": 6.8014440433212995,
      "grad_norm": 44.213470458984375,
      "learning_rate": 0.00019809969161077378,
      "loss": 6.8833,
      "step": 56520
    },
    {
      "epoch": 6.802647412755716,
      "grad_norm": 10.676111221313477,
      "learning_rate": 0.00019809895342270611,
      "loss": 6.9762,
      "step": 56530
    },
    {
      "epoch": 6.803850782190132,
      "grad_norm": 45.14767837524414,
      "learning_rate": 0.00019809821509266505,
      "loss": 6.9556,
      "step": 56540
    },
    {
      "epoch": 6.805054151624549,
      "grad_norm": 43.18658447265625,
      "learning_rate": 0.00019809747662065163,
      "loss": 6.8845,
      "step": 56550
    },
    {
      "epoch": 6.806257521058965,
      "grad_norm": 166.0917205810547,
      "learning_rate": 0.00019809673800666694,
      "loss": 7.1932,
      "step": 56560
    },
    {
      "epoch": 6.807460890493381,
      "grad_norm": 133.87234497070312,
      "learning_rate": 0.0001980959992507121,
      "loss": 7.0549,
      "step": 56570
    },
    {
      "epoch": 6.808664259927798,
      "grad_norm": 27.237041473388672,
      "learning_rate": 0.0001980952603527881,
      "loss": 7.2864,
      "step": 56580
    },
    {
      "epoch": 6.8098676293622145,
      "grad_norm": 9.687499046325684,
      "learning_rate": 0.000198094521312896,
      "loss": 7.1425,
      "step": 56590
    },
    {
      "epoch": 6.81107099879663,
      "grad_norm": 19.088077545166016,
      "learning_rate": 0.00019809378213103695,
      "loss": 7.0598,
      "step": 56600
    },
    {
      "epoch": 6.812274368231047,
      "grad_norm": 26.874658584594727,
      "learning_rate": 0.000198093042807212,
      "loss": 7.3514,
      "step": 56610
    },
    {
      "epoch": 6.813477737665464,
      "grad_norm": 29.382793426513672,
      "learning_rate": 0.00019809230334142216,
      "loss": 7.3406,
      "step": 56620
    },
    {
      "epoch": 6.814681107099879,
      "grad_norm": 20.1160831451416,
      "learning_rate": 0.00019809156373366854,
      "loss": 7.106,
      "step": 56630
    },
    {
      "epoch": 6.815884476534296,
      "grad_norm": 93.99827575683594,
      "learning_rate": 0.00019809082398395222,
      "loss": 7.0595,
      "step": 56640
    },
    {
      "epoch": 6.817087845968713,
      "grad_norm": 64.36432647705078,
      "learning_rate": 0.00019809008409227426,
      "loss": 6.985,
      "step": 56650
    },
    {
      "epoch": 6.8182912154031285,
      "grad_norm": 63.12652587890625,
      "learning_rate": 0.00019808934405863573,
      "loss": 7.1129,
      "step": 56660
    },
    {
      "epoch": 6.819494584837545,
      "grad_norm": 273.1567687988281,
      "learning_rate": 0.00019808860388303772,
      "loss": 6.8513,
      "step": 56670
    },
    {
      "epoch": 6.820697954271962,
      "grad_norm": 54.955684661865234,
      "learning_rate": 0.00019808786356548125,
      "loss": 7.033,
      "step": 56680
    },
    {
      "epoch": 6.821901323706378,
      "grad_norm": 179.6001739501953,
      "learning_rate": 0.00019808712310596743,
      "loss": 6.8798,
      "step": 56690
    },
    {
      "epoch": 6.823104693140794,
      "grad_norm": 24.781766891479492,
      "learning_rate": 0.00019808638250449733,
      "loss": 7.0504,
      "step": 56700
    },
    {
      "epoch": 6.824308062575211,
      "grad_norm": 96.25439453125,
      "learning_rate": 0.00019808564176107202,
      "loss": 7.2775,
      "step": 56710
    },
    {
      "epoch": 6.825511432009627,
      "grad_norm": 60.37154006958008,
      "learning_rate": 0.00019808490087569258,
      "loss": 7.2759,
      "step": 56720
    },
    {
      "epoch": 6.8267148014440435,
      "grad_norm": 52.00590515136719,
      "learning_rate": 0.00019808415984836006,
      "loss": 7.2158,
      "step": 56730
    },
    {
      "epoch": 6.827918170878459,
      "grad_norm": 25.168285369873047,
      "learning_rate": 0.00019808341867907556,
      "loss": 7.1152,
      "step": 56740
    },
    {
      "epoch": 6.829121540312876,
      "grad_norm": 47.336814880371094,
      "learning_rate": 0.00019808267736784013,
      "loss": 7.064,
      "step": 56750
    },
    {
      "epoch": 6.830324909747293,
      "grad_norm": 127.8853759765625,
      "learning_rate": 0.00019808193591465485,
      "loss": 7.0396,
      "step": 56760
    },
    {
      "epoch": 6.8315282791817085,
      "grad_norm": 48.18959426879883,
      "learning_rate": 0.00019808119431952078,
      "loss": 7.1016,
      "step": 56770
    },
    {
      "epoch": 6.832731648616125,
      "grad_norm": 117.4593276977539,
      "learning_rate": 0.000198080452582439,
      "loss": 7.0179,
      "step": 56780
    },
    {
      "epoch": 6.833935018050542,
      "grad_norm": 19.329599380493164,
      "learning_rate": 0.0001980797107034106,
      "loss": 7.0359,
      "step": 56790
    },
    {
      "epoch": 6.835138387484958,
      "grad_norm": 32.706485748291016,
      "learning_rate": 0.00019807896868243664,
      "loss": 6.9465,
      "step": 56800
    },
    {
      "epoch": 6.836341756919374,
      "grad_norm": 29.41881561279297,
      "learning_rate": 0.00019807822651951823,
      "loss": 7.0714,
      "step": 56810
    },
    {
      "epoch": 6.837545126353791,
      "grad_norm": 65.36539459228516,
      "learning_rate": 0.0001980774842146564,
      "loss": 7.1706,
      "step": 56820
    },
    {
      "epoch": 6.838748495788207,
      "grad_norm": 152.0336151123047,
      "learning_rate": 0.00019807674176785224,
      "loss": 7.2673,
      "step": 56830
    },
    {
      "epoch": 6.839951865222623,
      "grad_norm": 45.911563873291016,
      "learning_rate": 0.00019807599917910678,
      "loss": 7.3042,
      "step": 56840
    },
    {
      "epoch": 6.841155234657039,
      "grad_norm": 168.76766967773438,
      "learning_rate": 0.00019807525644842115,
      "loss": 7.5659,
      "step": 56850
    },
    {
      "epoch": 6.842358604091456,
      "grad_norm": 210.3383331298828,
      "learning_rate": 0.00019807451357579643,
      "loss": 7.4778,
      "step": 56860
    },
    {
      "epoch": 6.843561973525873,
      "grad_norm": 13.706093788146973,
      "learning_rate": 0.00019807377056123366,
      "loss": 7.4884,
      "step": 56870
    },
    {
      "epoch": 6.844765342960288,
      "grad_norm": 18.30304718017578,
      "learning_rate": 0.00019807302740473392,
      "loss": 7.5125,
      "step": 56880
    },
    {
      "epoch": 6.845968712394705,
      "grad_norm": 9.22196102142334,
      "learning_rate": 0.0001980722841062983,
      "loss": 7.4489,
      "step": 56890
    },
    {
      "epoch": 6.847172081829122,
      "grad_norm": 6.009471893310547,
      "learning_rate": 0.0001980715406659279,
      "loss": 7.3385,
      "step": 56900
    },
    {
      "epoch": 6.8483754512635375,
      "grad_norm": 17.32166290283203,
      "learning_rate": 0.00019807079708362375,
      "loss": 7.2548,
      "step": 56910
    },
    {
      "epoch": 6.849578820697954,
      "grad_norm": 33.649078369140625,
      "learning_rate": 0.00019807005335938693,
      "loss": 7.0738,
      "step": 56920
    },
    {
      "epoch": 6.850782190132371,
      "grad_norm": 45.04438781738281,
      "learning_rate": 0.00019806930949321856,
      "loss": 7.0602,
      "step": 56930
    },
    {
      "epoch": 6.851985559566787,
      "grad_norm": 36.69521713256836,
      "learning_rate": 0.00019806856548511967,
      "loss": 7.1064,
      "step": 56940
    },
    {
      "epoch": 6.853188929001203,
      "grad_norm": 29.256868362426758,
      "learning_rate": 0.00019806782133509134,
      "loss": 6.97,
      "step": 56950
    },
    {
      "epoch": 6.85439229843562,
      "grad_norm": 64.79109191894531,
      "learning_rate": 0.00019806707704313466,
      "loss": 6.985,
      "step": 56960
    },
    {
      "epoch": 6.855595667870036,
      "grad_norm": 525.453369140625,
      "learning_rate": 0.00019806633260925074,
      "loss": 6.9575,
      "step": 56970
    },
    {
      "epoch": 6.8567990373044525,
      "grad_norm": 277.41180419921875,
      "learning_rate": 0.0001980655880334406,
      "loss": 6.8746,
      "step": 56980
    },
    {
      "epoch": 6.858002406738869,
      "grad_norm": 606.5435791015625,
      "learning_rate": 0.00019806484331570535,
      "loss": 6.9853,
      "step": 56990
    },
    {
      "epoch": 6.859205776173285,
      "grad_norm": 713.16845703125,
      "learning_rate": 0.00019806409845604604,
      "loss": 7.0665,
      "step": 57000
    },
    {
      "epoch": 6.860409145607702,
      "grad_norm": 927.540771484375,
      "learning_rate": 0.0001980633534544638,
      "loss": 7.1799,
      "step": 57010
    },
    {
      "epoch": 6.861612515042118,
      "grad_norm": 360.078369140625,
      "learning_rate": 0.00019806260831095967,
      "loss": 7.2209,
      "step": 57020
    },
    {
      "epoch": 6.862815884476534,
      "grad_norm": 242.3676300048828,
      "learning_rate": 0.00019806186302553474,
      "loss": 7.3714,
      "step": 57030
    },
    {
      "epoch": 6.864019253910951,
      "grad_norm": 987.6715698242188,
      "learning_rate": 0.00019806111759819006,
      "loss": 7.2273,
      "step": 57040
    },
    {
      "epoch": 6.8652226233453675,
      "grad_norm": 413.6085205078125,
      "learning_rate": 0.00019806037202892675,
      "loss": 7.3345,
      "step": 57050
    },
    {
      "epoch": 6.866425992779783,
      "grad_norm": 55.95791244506836,
      "learning_rate": 0.00019805962631774587,
      "loss": 7.07,
      "step": 57060
    },
    {
      "epoch": 6.8676293622142,
      "grad_norm": 25.23629379272461,
      "learning_rate": 0.0001980588804646485,
      "loss": 6.9322,
      "step": 57070
    },
    {
      "epoch": 6.868832731648616,
      "grad_norm": 67.85858154296875,
      "learning_rate": 0.0001980581344696357,
      "loss": 7.1859,
      "step": 57080
    },
    {
      "epoch": 6.870036101083032,
      "grad_norm": 9.694706916809082,
      "learning_rate": 0.0001980573883327086,
      "loss": 6.9408,
      "step": 57090
    },
    {
      "epoch": 6.871239470517449,
      "grad_norm": 123.4223861694336,
      "learning_rate": 0.00019805664205386822,
      "loss": 7.0047,
      "step": 57100
    },
    {
      "epoch": 6.872442839951865,
      "grad_norm": 136.64669799804688,
      "learning_rate": 0.00019805589563311567,
      "loss": 6.9801,
      "step": 57110
    },
    {
      "epoch": 6.873646209386282,
      "grad_norm": 58.5440559387207,
      "learning_rate": 0.00019805514907045207,
      "loss": 6.8749,
      "step": 57120
    },
    {
      "epoch": 6.874849578820698,
      "grad_norm": 1241.7928466796875,
      "learning_rate": 0.00019805440236587843,
      "loss": 7.0498,
      "step": 57130
    },
    {
      "epoch": 6.876052948255114,
      "grad_norm": 118.74279022216797,
      "learning_rate": 0.00019805365551939588,
      "loss": 7.0736,
      "step": 57140
    },
    {
      "epoch": 6.877256317689531,
      "grad_norm": 10.805257797241211,
      "learning_rate": 0.00019805290853100544,
      "loss": 7.3568,
      "step": 57150
    },
    {
      "epoch": 6.878459687123947,
      "grad_norm": 43.8604736328125,
      "learning_rate": 0.00019805216140070825,
      "loss": 7.0794,
      "step": 57160
    },
    {
      "epoch": 6.879663056558363,
      "grad_norm": 54.924869537353516,
      "learning_rate": 0.00019805141412850542,
      "loss": 7.0036,
      "step": 57170
    },
    {
      "epoch": 6.88086642599278,
      "grad_norm": 17.287311553955078,
      "learning_rate": 0.00019805066671439793,
      "loss": 7.0393,
      "step": 57180
    },
    {
      "epoch": 6.882069795427196,
      "grad_norm": 139.8007049560547,
      "learning_rate": 0.00019804991915838694,
      "loss": 6.8933,
      "step": 57190
    },
    {
      "epoch": 6.883273164861612,
      "grad_norm": 20.29083824157715,
      "learning_rate": 0.0001980491714604735,
      "loss": 6.921,
      "step": 57200
    },
    {
      "epoch": 6.884476534296029,
      "grad_norm": 17.311073303222656,
      "learning_rate": 0.00019804842362065872,
      "loss": 6.9645,
      "step": 57210
    },
    {
      "epoch": 6.885679903730445,
      "grad_norm": 29.02623176574707,
      "learning_rate": 0.00019804767563894363,
      "loss": 6.8939,
      "step": 57220
    },
    {
      "epoch": 6.8868832731648615,
      "grad_norm": 33.343406677246094,
      "learning_rate": 0.0001980469275153294,
      "loss": 6.8359,
      "step": 57230
    },
    {
      "epoch": 6.888086642599278,
      "grad_norm": 141.8105926513672,
      "learning_rate": 0.000198046179249817,
      "loss": 7.1161,
      "step": 57240
    },
    {
      "epoch": 6.889290012033694,
      "grad_norm": 44.20021438598633,
      "learning_rate": 0.00019804543084240761,
      "loss": 6.9137,
      "step": 57250
    },
    {
      "epoch": 6.890493381468111,
      "grad_norm": 233.2378692626953,
      "learning_rate": 0.00019804468229310226,
      "loss": 6.9329,
      "step": 57260
    },
    {
      "epoch": 6.891696750902527,
      "grad_norm": 57.86533737182617,
      "learning_rate": 0.00019804393360190205,
      "loss": 6.8827,
      "step": 57270
    },
    {
      "epoch": 6.892900120336943,
      "grad_norm": 55.147926330566406,
      "learning_rate": 0.00019804318476880804,
      "loss": 7.0093,
      "step": 57280
    },
    {
      "epoch": 6.89410348977136,
      "grad_norm": 826.686767578125,
      "learning_rate": 0.0001980424357938214,
      "loss": 6.9077,
      "step": 57290
    },
    {
      "epoch": 6.8953068592057765,
      "grad_norm": 30.803756713867188,
      "learning_rate": 0.0001980416866769431,
      "loss": 6.9971,
      "step": 57300
    },
    {
      "epoch": 6.896510228640192,
      "grad_norm": 52.69157791137695,
      "learning_rate": 0.00019804093741817428,
      "loss": 6.9022,
      "step": 57310
    },
    {
      "epoch": 6.897713598074609,
      "grad_norm": 80.60667419433594,
      "learning_rate": 0.000198040188017516,
      "loss": 6.9859,
      "step": 57320
    },
    {
      "epoch": 6.898916967509026,
      "grad_norm": 31.60869598388672,
      "learning_rate": 0.0001980394384749694,
      "loss": 6.8363,
      "step": 57330
    },
    {
      "epoch": 6.900120336943441,
      "grad_norm": 84.88731384277344,
      "learning_rate": 0.0001980386887905355,
      "loss": 6.8107,
      "step": 57340
    },
    {
      "epoch": 6.901323706377858,
      "grad_norm": 52.038761138916016,
      "learning_rate": 0.00019803793896421544,
      "loss": 6.8244,
      "step": 57350
    },
    {
      "epoch": 6.902527075812275,
      "grad_norm": 518.0714721679688,
      "learning_rate": 0.00019803718899601025,
      "loss": 6.8211,
      "step": 57360
    },
    {
      "epoch": 6.9037304452466906,
      "grad_norm": 144.3697509765625,
      "learning_rate": 0.00019803643888592105,
      "loss": 6.9075,
      "step": 57370
    },
    {
      "epoch": 6.904933814681107,
      "grad_norm": 61.14954376220703,
      "learning_rate": 0.00019803568863394893,
      "loss": 6.8461,
      "step": 57380
    },
    {
      "epoch": 6.906137184115524,
      "grad_norm": 192.9075164794922,
      "learning_rate": 0.00019803493824009495,
      "loss": 6.9015,
      "step": 57390
    },
    {
      "epoch": 6.90734055354994,
      "grad_norm": 318.1616516113281,
      "learning_rate": 0.00019803418770436023,
      "loss": 6.8928,
      "step": 57400
    },
    {
      "epoch": 6.908543922984356,
      "grad_norm": 36.558746337890625,
      "learning_rate": 0.0001980334370267458,
      "loss": 6.8422,
      "step": 57410
    },
    {
      "epoch": 6.909747292418773,
      "grad_norm": 41.883949279785156,
      "learning_rate": 0.0001980326862072528,
      "loss": 6.8808,
      "step": 57420
    },
    {
      "epoch": 6.910950661853189,
      "grad_norm": 162.16358947753906,
      "learning_rate": 0.00019803193524588233,
      "loss": 6.7901,
      "step": 57430
    },
    {
      "epoch": 6.9121540312876055,
      "grad_norm": 31.169635772705078,
      "learning_rate": 0.00019803118414263543,
      "loss": 6.8673,
      "step": 57440
    },
    {
      "epoch": 6.913357400722021,
      "grad_norm": 1550.4622802734375,
      "learning_rate": 0.0001980304328975132,
      "loss": 6.9876,
      "step": 57450
    },
    {
      "epoch": 6.914560770156438,
      "grad_norm": 32.81758117675781,
      "learning_rate": 0.0001980296815105167,
      "loss": 7.0469,
      "step": 57460
    },
    {
      "epoch": 6.915764139590855,
      "grad_norm": 42.625938415527344,
      "learning_rate": 0.00019802892998164708,
      "loss": 6.9365,
      "step": 57470
    },
    {
      "epoch": 6.9169675090252705,
      "grad_norm": 66.55526733398438,
      "learning_rate": 0.0001980281783109054,
      "loss": 6.8728,
      "step": 57480
    },
    {
      "epoch": 6.918170878459687,
      "grad_norm": 26.685131072998047,
      "learning_rate": 0.00019802742649829274,
      "loss": 6.8764,
      "step": 57490
    },
    {
      "epoch": 6.919374247894104,
      "grad_norm": 24.899927139282227,
      "learning_rate": 0.00019802667454381018,
      "loss": 6.9962,
      "step": 57500
    },
    {
      "epoch": 6.92057761732852,
      "grad_norm": 458.8593444824219,
      "learning_rate": 0.0001980259224474588,
      "loss": 6.8451,
      "step": 57510
    },
    {
      "epoch": 6.921780986762936,
      "grad_norm": 248.4935302734375,
      "learning_rate": 0.00019802517020923977,
      "loss": 6.7186,
      "step": 57520
    },
    {
      "epoch": 6.922984356197353,
      "grad_norm": 62.006656646728516,
      "learning_rate": 0.00019802441782915407,
      "loss": 6.7663,
      "step": 57530
    },
    {
      "epoch": 6.924187725631769,
      "grad_norm": 87.1061782836914,
      "learning_rate": 0.00019802366530720283,
      "loss": 6.9221,
      "step": 57540
    },
    {
      "epoch": 6.925391095066185,
      "grad_norm": 238.37020874023438,
      "learning_rate": 0.00019802291264338717,
      "loss": 6.9728,
      "step": 57550
    },
    {
      "epoch": 6.926594464500601,
      "grad_norm": 59.73338317871094,
      "learning_rate": 0.00019802215983770816,
      "loss": 7.0036,
      "step": 57560
    },
    {
      "epoch": 6.927797833935018,
      "grad_norm": 571.4495239257812,
      "learning_rate": 0.00019802140689016687,
      "loss": 7.0787,
      "step": 57570
    },
    {
      "epoch": 6.929001203369435,
      "grad_norm": 52.87287521362305,
      "learning_rate": 0.00019802065380076437,
      "loss": 7.0749,
      "step": 57580
    },
    {
      "epoch": 6.93020457280385,
      "grad_norm": 46.382179260253906,
      "learning_rate": 0.00019801990056950183,
      "loss": 6.9606,
      "step": 57590
    },
    {
      "epoch": 6.931407942238267,
      "grad_norm": 1761.7576904296875,
      "learning_rate": 0.0001980191471963803,
      "loss": 7.0838,
      "step": 57600
    },
    {
      "epoch": 6.932611311672684,
      "grad_norm": 43.45762634277344,
      "learning_rate": 0.00019801839368140082,
      "loss": 7.0395,
      "step": 57610
    },
    {
      "epoch": 6.9338146811070995,
      "grad_norm": 53.85858917236328,
      "learning_rate": 0.00019801764002456455,
      "loss": 7.0695,
      "step": 57620
    },
    {
      "epoch": 6.935018050541516,
      "grad_norm": 200.94857788085938,
      "learning_rate": 0.00019801688622587255,
      "loss": 6.9848,
      "step": 57630
    },
    {
      "epoch": 6.936221419975933,
      "grad_norm": 733.8641967773438,
      "learning_rate": 0.00019801613228532592,
      "loss": 7.012,
      "step": 57640
    },
    {
      "epoch": 6.937424789410349,
      "grad_norm": 194.35470581054688,
      "learning_rate": 0.00019801537820292574,
      "loss": 7.2865,
      "step": 57650
    },
    {
      "epoch": 6.938628158844765,
      "grad_norm": 385.124755859375,
      "learning_rate": 0.00019801462397867314,
      "loss": 7.8188,
      "step": 57660
    },
    {
      "epoch": 6.939831528279182,
      "grad_norm": 57.436866760253906,
      "learning_rate": 0.00019801386961256916,
      "loss": 7.6012,
      "step": 57670
    },
    {
      "epoch": 6.941034897713598,
      "grad_norm": 58.678226470947266,
      "learning_rate": 0.0001980131151046149,
      "loss": 7.3455,
      "step": 57680
    },
    {
      "epoch": 6.9422382671480145,
      "grad_norm": 477.7326354980469,
      "learning_rate": 0.00019801236045481147,
      "loss": 7.2111,
      "step": 57690
    },
    {
      "epoch": 6.943441636582431,
      "grad_norm": 4454.03466796875,
      "learning_rate": 0.00019801160566315995,
      "loss": 7.019,
      "step": 57700
    },
    {
      "epoch": 6.944645006016847,
      "grad_norm": 380.172607421875,
      "learning_rate": 0.00019801085072966145,
      "loss": 7.4128,
      "step": 57710
    },
    {
      "epoch": 6.945848375451264,
      "grad_norm": 283.2594909667969,
      "learning_rate": 0.00019801009565431702,
      "loss": 7.0336,
      "step": 57720
    },
    {
      "epoch": 6.94705174488568,
      "grad_norm": 701.2532958984375,
      "learning_rate": 0.00019800934043712784,
      "loss": 7.1736,
      "step": 57730
    },
    {
      "epoch": 6.948255114320096,
      "grad_norm": 433.7130432128906,
      "learning_rate": 0.0001980085850780949,
      "loss": 7.0403,
      "step": 57740
    },
    {
      "epoch": 6.949458483754513,
      "grad_norm": 132.9949951171875,
      "learning_rate": 0.00019800782957721938,
      "loss": 7.1357,
      "step": 57750
    },
    {
      "epoch": 6.9506618531889295,
      "grad_norm": 323.5587158203125,
      "learning_rate": 0.00019800707393450228,
      "loss": 7.0936,
      "step": 57760
    },
    {
      "epoch": 6.951865222623345,
      "grad_norm": 207.72708129882812,
      "learning_rate": 0.00019800631814994479,
      "loss": 6.9885,
      "step": 57770
    },
    {
      "epoch": 6.953068592057762,
      "grad_norm": 24.468793869018555,
      "learning_rate": 0.00019800556222354795,
      "loss": 7.1302,
      "step": 57780
    },
    {
      "epoch": 6.954271961492178,
      "grad_norm": 51.776973724365234,
      "learning_rate": 0.00019800480615531286,
      "loss": 6.9301,
      "step": 57790
    },
    {
      "epoch": 6.955475330926594,
      "grad_norm": 77.13422393798828,
      "learning_rate": 0.0001980040499452406,
      "loss": 7.0983,
      "step": 57800
    },
    {
      "epoch": 6.956678700361011,
      "grad_norm": 280.56658935546875,
      "learning_rate": 0.00019800329359333233,
      "loss": 7.1142,
      "step": 57810
    },
    {
      "epoch": 6.957882069795427,
      "grad_norm": 233.68243408203125,
      "learning_rate": 0.00019800253709958905,
      "loss": 7.1303,
      "step": 57820
    },
    {
      "epoch": 6.959085439229844,
      "grad_norm": 2115.4267578125,
      "learning_rate": 0.00019800178046401192,
      "loss": 7.1151,
      "step": 57830
    },
    {
      "epoch": 6.96028880866426,
      "grad_norm": 178.87379455566406,
      "learning_rate": 0.00019800102368660203,
      "loss": 7.133,
      "step": 57840
    },
    {
      "epoch": 6.961492178098676,
      "grad_norm": 312.19281005859375,
      "learning_rate": 0.00019800026676736047,
      "loss": 7.3959,
      "step": 57850
    },
    {
      "epoch": 6.962695547533093,
      "grad_norm": 80.78385162353516,
      "learning_rate": 0.0001979995097062883,
      "loss": 7.7487,
      "step": 57860
    },
    {
      "epoch": 6.963898916967509,
      "grad_norm": 57.2595100402832,
      "learning_rate": 0.00019799875250338665,
      "loss": 7.9043,
      "step": 57870
    },
    {
      "epoch": 6.965102286401925,
      "grad_norm": 11.961369514465332,
      "learning_rate": 0.00019799799515865662,
      "loss": 7.5828,
      "step": 57880
    },
    {
      "epoch": 6.966305655836342,
      "grad_norm": 41.18761444091797,
      "learning_rate": 0.00019799723767209934,
      "loss": 7.5542,
      "step": 57890
    },
    {
      "epoch": 6.967509025270758,
      "grad_norm": 29.269014358520508,
      "learning_rate": 0.00019799648004371578,
      "loss": 7.4847,
      "step": 57900
    },
    {
      "epoch": 6.968712394705174,
      "grad_norm": 59.71797180175781,
      "learning_rate": 0.00019799572227350716,
      "loss": 7.5432,
      "step": 57910
    },
    {
      "epoch": 6.969915764139591,
      "grad_norm": 40.786354064941406,
      "learning_rate": 0.00019799496436147452,
      "loss": 7.5736,
      "step": 57920
    },
    {
      "epoch": 6.971119133574007,
      "grad_norm": 44.69639205932617,
      "learning_rate": 0.000197994206307619,
      "loss": 7.4789,
      "step": 57930
    },
    {
      "epoch": 6.9723225030084235,
      "grad_norm": 51.41767501831055,
      "learning_rate": 0.00019799344811194164,
      "loss": 7.4996,
      "step": 57940
    },
    {
      "epoch": 6.97352587244284,
      "grad_norm": 37.17129898071289,
      "learning_rate": 0.0001979926897744436,
      "loss": 7.5562,
      "step": 57950
    },
    {
      "epoch": 6.974729241877256,
      "grad_norm": 62.11970138549805,
      "learning_rate": 0.0001979919312951259,
      "loss": 7.5705,
      "step": 57960
    },
    {
      "epoch": 6.975932611311673,
      "grad_norm": 55.383914947509766,
      "learning_rate": 0.00019799117267398973,
      "loss": 7.5302,
      "step": 57970
    },
    {
      "epoch": 6.977135980746089,
      "grad_norm": 50.30924987792969,
      "learning_rate": 0.0001979904139110361,
      "loss": 7.6065,
      "step": 57980
    },
    {
      "epoch": 6.978339350180505,
      "grad_norm": 217.25390625,
      "learning_rate": 0.00019798965500626616,
      "loss": 7.4346,
      "step": 57990
    },
    {
      "epoch": 6.979542719614922,
      "grad_norm": 58.953006744384766,
      "learning_rate": 0.000197988895959681,
      "loss": 7.5839,
      "step": 58000
    },
    {
      "epoch": 6.9807460890493385,
      "grad_norm": 4133.89306640625,
      "learning_rate": 0.0001979881367712817,
      "loss": 7.6053,
      "step": 58010
    },
    {
      "epoch": 6.981949458483754,
      "grad_norm": 4566.65966796875,
      "learning_rate": 0.00019798737744106942,
      "loss": 7.7741,
      "step": 58020
    },
    {
      "epoch": 6.983152827918171,
      "grad_norm": 27203.322265625,
      "learning_rate": 0.0001979866179690452,
      "loss": 8.7773,
      "step": 58030
    },
    {
      "epoch": 6.984356197352588,
      "grad_norm": 381.32415771484375,
      "learning_rate": 0.00019798585835521012,
      "loss": 9.6103,
      "step": 58040
    },
    {
      "epoch": 6.985559566787003,
      "grad_norm": 80.57240295410156,
      "learning_rate": 0.0001979850985995653,
      "loss": 8.8075,
      "step": 58050
    },
    {
      "epoch": 6.98676293622142,
      "grad_norm": 55.19731140136719,
      "learning_rate": 0.00019798433870211188,
      "loss": 8.158,
      "step": 58060
    },
    {
      "epoch": 6.987966305655837,
      "grad_norm": 184.60914611816406,
      "learning_rate": 0.00019798357866285092,
      "loss": 7.5799,
      "step": 58070
    },
    {
      "epoch": 6.9891696750902526,
      "grad_norm": 350.00628662109375,
      "learning_rate": 0.00019798281848178352,
      "loss": 7.7077,
      "step": 58080
    },
    {
      "epoch": 6.990373044524669,
      "grad_norm": 3972.277099609375,
      "learning_rate": 0.00019798205815891082,
      "loss": 7.5186,
      "step": 58090
    },
    {
      "epoch": 6.991576413959086,
      "grad_norm": 3121.091552734375,
      "learning_rate": 0.0001979812976942339,
      "loss": 7.5143,
      "step": 58100
    },
    {
      "epoch": 6.992779783393502,
      "grad_norm": 646.8971557617188,
      "learning_rate": 0.00019798053708775383,
      "loss": 7.6622,
      "step": 58110
    },
    {
      "epoch": 6.993983152827918,
      "grad_norm": 2629.6357421875,
      "learning_rate": 0.00019797977633947172,
      "loss": 8.0694,
      "step": 58120
    },
    {
      "epoch": 6.995186522262334,
      "grad_norm": 683.5379638671875,
      "learning_rate": 0.00019797901544938869,
      "loss": 7.9134,
      "step": 58130
    },
    {
      "epoch": 6.996389891696751,
      "grad_norm": 96.93624877929688,
      "learning_rate": 0.00019797825441750583,
      "loss": 7.559,
      "step": 58140
    },
    {
      "epoch": 6.9975932611311675,
      "grad_norm": 23.396915435791016,
      "learning_rate": 0.00019797749324382426,
      "loss": 7.5456,
      "step": 58150
    },
    {
      "epoch": 6.998796630565583,
      "grad_norm": 19.570301055908203,
      "learning_rate": 0.00019797673192834507,
      "loss": 7.5895,
      "step": 58160
    },
    {
      "epoch": 7.0,
      "grad_norm": 20.815324783325195,
      "learning_rate": 0.00019797597047106934,
      "loss": 7.5813,
      "step": 58170
    },
    {
      "epoch": 7.0,
      "eval_loss": 7.590388774871826,
      "eval_runtime": 119.8697,
      "eval_samples_per_second": 61.625,
      "eval_steps_per_second": 7.708,
      "step": 58170
    },
    {
      "epoch": 7.001203369434417,
      "grad_norm": 6.156953811645508,
      "learning_rate": 0.0001979752088719982,
      "loss": 7.5375,
      "step": 58180
    },
    {
      "epoch": 7.0024067388688325,
      "grad_norm": 11.347652435302734,
      "learning_rate": 0.00019797444713113275,
      "loss": 7.6103,
      "step": 58190
    },
    {
      "epoch": 7.003610108303249,
      "grad_norm": 14.72617244720459,
      "learning_rate": 0.00019797368524847406,
      "loss": 7.7062,
      "step": 58200
    },
    {
      "epoch": 7.004813477737666,
      "grad_norm": 12.33837604522705,
      "learning_rate": 0.00019797292322402328,
      "loss": 7.5682,
      "step": 58210
    },
    {
      "epoch": 7.006016847172082,
      "grad_norm": 9.066107749938965,
      "learning_rate": 0.0001979721610577815,
      "loss": 7.5672,
      "step": 58220
    },
    {
      "epoch": 7.007220216606498,
      "grad_norm": 14.04942512512207,
      "learning_rate": 0.0001979713987497498,
      "loss": 7.512,
      "step": 58230
    },
    {
      "epoch": 7.008423586040915,
      "grad_norm": 3.176405906677246,
      "learning_rate": 0.0001979706362999293,
      "loss": 7.5756,
      "step": 58240
    },
    {
      "epoch": 7.009626955475331,
      "grad_norm": 12.903924942016602,
      "learning_rate": 0.00019796987370832112,
      "loss": 7.6313,
      "step": 58250
    },
    {
      "epoch": 7.0108303249097474,
      "grad_norm": 6.505326747894287,
      "learning_rate": 0.0001979691109749263,
      "loss": 7.6186,
      "step": 58260
    },
    {
      "epoch": 7.012033694344163,
      "grad_norm": 16.74981689453125,
      "learning_rate": 0.00019796834809974602,
      "loss": 7.5558,
      "step": 58270
    },
    {
      "epoch": 7.01323706377858,
      "grad_norm": 9.05154800415039,
      "learning_rate": 0.00019796758508278136,
      "loss": 7.5815,
      "step": 58280
    },
    {
      "epoch": 7.014440433212997,
      "grad_norm": 16.661067962646484,
      "learning_rate": 0.0001979668219240334,
      "loss": 7.6323,
      "step": 58290
    },
    {
      "epoch": 7.015643802647412,
      "grad_norm": 8.390691757202148,
      "learning_rate": 0.0001979660586235033,
      "loss": 7.6556,
      "step": 58300
    },
    {
      "epoch": 7.016847172081829,
      "grad_norm": 6.765058994293213,
      "learning_rate": 0.00019796529518119208,
      "loss": 7.6327,
      "step": 58310
    },
    {
      "epoch": 7.018050541516246,
      "grad_norm": 9.859917640686035,
      "learning_rate": 0.0001979645315971009,
      "loss": 7.5974,
      "step": 58320
    },
    {
      "epoch": 7.0192539109506615,
      "grad_norm": 5.7368998527526855,
      "learning_rate": 0.00019796376787123085,
      "loss": 7.5915,
      "step": 58330
    },
    {
      "epoch": 7.020457280385078,
      "grad_norm": 4.969225883483887,
      "learning_rate": 0.00019796300400358306,
      "loss": 7.6415,
      "step": 58340
    },
    {
      "epoch": 7.021660649819495,
      "grad_norm": 7.187626838684082,
      "learning_rate": 0.00019796223999415862,
      "loss": 7.5956,
      "step": 58350
    },
    {
      "epoch": 7.022864019253911,
      "grad_norm": 6.633703708648682,
      "learning_rate": 0.00019796147584295862,
      "loss": 7.5451,
      "step": 58360
    },
    {
      "epoch": 7.024067388688327,
      "grad_norm": 7.89232873916626,
      "learning_rate": 0.0001979607115499842,
      "loss": 7.6214,
      "step": 58370
    },
    {
      "epoch": 7.025270758122744,
      "grad_norm": 5.816625118255615,
      "learning_rate": 0.0001979599471152364,
      "loss": 7.5487,
      "step": 58380
    },
    {
      "epoch": 7.02647412755716,
      "grad_norm": 3.71209716796875,
      "learning_rate": 0.00019795918253871642,
      "loss": 7.5717,
      "step": 58390
    },
    {
      "epoch": 7.0276774969915765,
      "grad_norm": 9.613621711730957,
      "learning_rate": 0.0001979584178204253,
      "loss": 7.609,
      "step": 58400
    },
    {
      "epoch": 7.028880866425993,
      "grad_norm": 3.8423612117767334,
      "learning_rate": 0.00019795765296036413,
      "loss": 7.4974,
      "step": 58410
    },
    {
      "epoch": 7.030084235860409,
      "grad_norm": 4.460007667541504,
      "learning_rate": 0.0001979568879585341,
      "loss": 7.5948,
      "step": 58420
    },
    {
      "epoch": 7.031287605294826,
      "grad_norm": 4.862419128417969,
      "learning_rate": 0.00019795612281493625,
      "loss": 7.5215,
      "step": 58430
    },
    {
      "epoch": 7.0324909747292415,
      "grad_norm": 5.015869617462158,
      "learning_rate": 0.00019795535752957172,
      "loss": 7.6274,
      "step": 58440
    },
    {
      "epoch": 7.033694344163658,
      "grad_norm": 6.971224784851074,
      "learning_rate": 0.00019795459210244156,
      "loss": 7.7244,
      "step": 58450
    },
    {
      "epoch": 7.034897713598075,
      "grad_norm": 4.386664867401123,
      "learning_rate": 0.00019795382653354696,
      "loss": 7.6078,
      "step": 58460
    },
    {
      "epoch": 7.036101083032491,
      "grad_norm": 1728.6195068359375,
      "learning_rate": 0.00019795306082288897,
      "loss": 7.5955,
      "step": 58470
    },
    {
      "epoch": 7.037304452466907,
      "grad_norm": 14.76229190826416,
      "learning_rate": 0.00019795229497046872,
      "loss": 7.5277,
      "step": 58480
    },
    {
      "epoch": 7.038507821901324,
      "grad_norm": 13.176336288452148,
      "learning_rate": 0.0001979515289762873,
      "loss": 7.6122,
      "step": 58490
    },
    {
      "epoch": 7.03971119133574,
      "grad_norm": 7.656918048858643,
      "learning_rate": 0.00019795076284034586,
      "loss": 7.5423,
      "step": 58500
    },
    {
      "epoch": 7.040914560770156,
      "grad_norm": 5.574862957000732,
      "learning_rate": 0.0001979499965626455,
      "loss": 7.4972,
      "step": 58510
    },
    {
      "epoch": 7.042117930204573,
      "grad_norm": 8.460332870483398,
      "learning_rate": 0.0001979492301431873,
      "loss": 7.501,
      "step": 58520
    },
    {
      "epoch": 7.043321299638989,
      "grad_norm": 8.483569145202637,
      "learning_rate": 0.00019794846358197235,
      "loss": 7.555,
      "step": 58530
    },
    {
      "epoch": 7.044524669073406,
      "grad_norm": 5.071497917175293,
      "learning_rate": 0.00019794769687900182,
      "loss": 7.5449,
      "step": 58540
    },
    {
      "epoch": 7.045728038507822,
      "grad_norm": 5.318825721740723,
      "learning_rate": 0.00019794693003427676,
      "loss": 7.4687,
      "step": 58550
    },
    {
      "epoch": 7.046931407942238,
      "grad_norm": 7.241866111755371,
      "learning_rate": 0.00019794616304779833,
      "loss": 7.5028,
      "step": 58560
    },
    {
      "epoch": 7.048134777376655,
      "grad_norm": 7.500179767608643,
      "learning_rate": 0.00019794539591956763,
      "loss": 7.4961,
      "step": 58570
    },
    {
      "epoch": 7.049338146811071,
      "grad_norm": 15.585549354553223,
      "learning_rate": 0.00019794462864958576,
      "loss": 7.5545,
      "step": 58580
    },
    {
      "epoch": 7.050541516245487,
      "grad_norm": 20.122377395629883,
      "learning_rate": 0.00019794386123785383,
      "loss": 7.5008,
      "step": 58590
    },
    {
      "epoch": 7.051744885679904,
      "grad_norm": 13.105525016784668,
      "learning_rate": 0.0001979430936843729,
      "loss": 7.5519,
      "step": 58600
    },
    {
      "epoch": 7.05294825511432,
      "grad_norm": 6.437312602996826,
      "learning_rate": 0.0001979423259891442,
      "loss": 7.6007,
      "step": 58610
    },
    {
      "epoch": 7.054151624548736,
      "grad_norm": 141.77891540527344,
      "learning_rate": 0.00019794155815216872,
      "loss": 7.5104,
      "step": 58620
    },
    {
      "epoch": 7.055354993983153,
      "grad_norm": 5.608367919921875,
      "learning_rate": 0.00019794079017344766,
      "loss": 7.5207,
      "step": 58630
    },
    {
      "epoch": 7.056558363417569,
      "grad_norm": 6.196149826049805,
      "learning_rate": 0.0001979400220529821,
      "loss": 7.5716,
      "step": 58640
    },
    {
      "epoch": 7.0577617328519855,
      "grad_norm": 39.432212829589844,
      "learning_rate": 0.00019793925379077315,
      "loss": 7.5621,
      "step": 58650
    },
    {
      "epoch": 7.058965102286402,
      "grad_norm": 6.417609691619873,
      "learning_rate": 0.0001979384853868219,
      "loss": 7.507,
      "step": 58660
    },
    {
      "epoch": 7.060168471720818,
      "grad_norm": 7.268493175506592,
      "learning_rate": 0.00019793771684112948,
      "loss": 7.6224,
      "step": 58670
    },
    {
      "epoch": 7.061371841155235,
      "grad_norm": 6.067602157592773,
      "learning_rate": 0.000197936948153697,
      "loss": 7.5457,
      "step": 58680
    },
    {
      "epoch": 7.062575210589651,
      "grad_norm": 3.903944730758667,
      "learning_rate": 0.0001979361793245256,
      "loss": 7.5416,
      "step": 58690
    },
    {
      "epoch": 7.063778580024067,
      "grad_norm": 14.14567756652832,
      "learning_rate": 0.00019793541035361637,
      "loss": 7.5404,
      "step": 58700
    },
    {
      "epoch": 7.064981949458484,
      "grad_norm": 5.904326438903809,
      "learning_rate": 0.00019793464124097038,
      "loss": 7.5542,
      "step": 58710
    },
    {
      "epoch": 7.0661853188929005,
      "grad_norm": 13.580793380737305,
      "learning_rate": 0.00019793387198658883,
      "loss": 7.6201,
      "step": 58720
    },
    {
      "epoch": 7.067388688327316,
      "grad_norm": 9.535094261169434,
      "learning_rate": 0.00019793310259047276,
      "loss": 7.5076,
      "step": 58730
    },
    {
      "epoch": 7.068592057761733,
      "grad_norm": 13.778815269470215,
      "learning_rate": 0.0001979323330526233,
      "loss": 7.5298,
      "step": 58740
    },
    {
      "epoch": 7.06979542719615,
      "grad_norm": 8.662415504455566,
      "learning_rate": 0.00019793156337304162,
      "loss": 7.4654,
      "step": 58750
    },
    {
      "epoch": 7.070998796630565,
      "grad_norm": 4.333553791046143,
      "learning_rate": 0.00019793079355172877,
      "loss": 7.5594,
      "step": 58760
    },
    {
      "epoch": 7.072202166064982,
      "grad_norm": 61.4433708190918,
      "learning_rate": 0.00019793002358868585,
      "loss": 7.5706,
      "step": 58770
    },
    {
      "epoch": 7.073405535499398,
      "grad_norm": 27.061952590942383,
      "learning_rate": 0.00019792925348391407,
      "loss": 7.5555,
      "step": 58780
    },
    {
      "epoch": 7.074608904933815,
      "grad_norm": 4.7676849365234375,
      "learning_rate": 0.00019792848323741443,
      "loss": 7.5421,
      "step": 58790
    },
    {
      "epoch": 7.075812274368231,
      "grad_norm": 8.331336975097656,
      "learning_rate": 0.00019792771284918813,
      "loss": 7.5539,
      "step": 58800
    },
    {
      "epoch": 7.077015643802647,
      "grad_norm": 3.2121715545654297,
      "learning_rate": 0.00019792694231923622,
      "loss": 7.5369,
      "step": 58810
    },
    {
      "epoch": 7.078219013237064,
      "grad_norm": 18.14316749572754,
      "learning_rate": 0.00019792617164755987,
      "loss": 7.4792,
      "step": 58820
    },
    {
      "epoch": 7.07942238267148,
      "grad_norm": 5.8300395011901855,
      "learning_rate": 0.0001979254008341602,
      "loss": 7.4685,
      "step": 58830
    },
    {
      "epoch": 7.080625752105896,
      "grad_norm": 7.983304500579834,
      "learning_rate": 0.00019792462987903823,
      "loss": 7.523,
      "step": 58840
    },
    {
      "epoch": 7.081829121540313,
      "grad_norm": 8.311152458190918,
      "learning_rate": 0.00019792385878219517,
      "loss": 7.559,
      "step": 58850
    },
    {
      "epoch": 7.0830324909747295,
      "grad_norm": 174.98464965820312,
      "learning_rate": 0.00019792308754363214,
      "loss": 7.7203,
      "step": 58860
    },
    {
      "epoch": 7.084235860409145,
      "grad_norm": 38.772727966308594,
      "learning_rate": 0.0001979223161633502,
      "loss": 7.9951,
      "step": 58870
    },
    {
      "epoch": 7.085439229843562,
      "grad_norm": 46.782081604003906,
      "learning_rate": 0.00019792154464135048,
      "loss": 8.0025,
      "step": 58880
    },
    {
      "epoch": 7.086642599277979,
      "grad_norm": 11.872475624084473,
      "learning_rate": 0.00019792077297763413,
      "loss": 7.9058,
      "step": 58890
    },
    {
      "epoch": 7.0878459687123945,
      "grad_norm": 13.866183280944824,
      "learning_rate": 0.00019792000117220226,
      "loss": 7.6373,
      "step": 58900
    },
    {
      "epoch": 7.089049338146811,
      "grad_norm": 19.396339416503906,
      "learning_rate": 0.00019791922922505594,
      "loss": 7.6002,
      "step": 58910
    },
    {
      "epoch": 7.090252707581228,
      "grad_norm": 61.84298324584961,
      "learning_rate": 0.00019791845713619635,
      "loss": 7.6912,
      "step": 58920
    },
    {
      "epoch": 7.091456077015644,
      "grad_norm": 22.066791534423828,
      "learning_rate": 0.00019791768490562457,
      "loss": 7.5435,
      "step": 58930
    },
    {
      "epoch": 7.09265944645006,
      "grad_norm": 16.87318992614746,
      "learning_rate": 0.0001979169125333417,
      "loss": 7.6051,
      "step": 58940
    },
    {
      "epoch": 7.093862815884476,
      "grad_norm": 7.5266008377075195,
      "learning_rate": 0.0001979161400193489,
      "loss": 7.5309,
      "step": 58950
    },
    {
      "epoch": 7.095066185318893,
      "grad_norm": 9.935173034667969,
      "learning_rate": 0.00019791536736364727,
      "loss": 7.5462,
      "step": 58960
    },
    {
      "epoch": 7.0962695547533094,
      "grad_norm": 48.00737380981445,
      "learning_rate": 0.00019791459456623792,
      "loss": 7.6156,
      "step": 58970
    },
    {
      "epoch": 7.097472924187725,
      "grad_norm": 63.169559478759766,
      "learning_rate": 0.000197913821627122,
      "loss": 7.5662,
      "step": 58980
    },
    {
      "epoch": 7.098676293622142,
      "grad_norm": 38.13153076171875,
      "learning_rate": 0.0001979130485463006,
      "loss": 7.5425,
      "step": 58990
    },
    {
      "epoch": 7.099879663056559,
      "grad_norm": 10.797197341918945,
      "learning_rate": 0.0001979122753237748,
      "loss": 7.6917,
      "step": 59000
    },
    {
      "epoch": 7.101083032490974,
      "grad_norm": 7.944900035858154,
      "learning_rate": 0.0001979115019595458,
      "loss": 7.6342,
      "step": 59010
    },
    {
      "epoch": 7.102286401925391,
      "grad_norm": 5.772441864013672,
      "learning_rate": 0.0001979107284536147,
      "loss": 7.5624,
      "step": 59020
    },
    {
      "epoch": 7.103489771359808,
      "grad_norm": 3.6065547466278076,
      "learning_rate": 0.0001979099548059826,
      "loss": 7.5784,
      "step": 59030
    },
    {
      "epoch": 7.1046931407942235,
      "grad_norm": 20.077173233032227,
      "learning_rate": 0.00019790918101665058,
      "loss": 7.6786,
      "step": 59040
    },
    {
      "epoch": 7.10589651022864,
      "grad_norm": 6.177094459533691,
      "learning_rate": 0.00019790840708561986,
      "loss": 7.6383,
      "step": 59050
    },
    {
      "epoch": 7.107099879663057,
      "grad_norm": 9.216825485229492,
      "learning_rate": 0.00019790763301289147,
      "loss": 7.6239,
      "step": 59060
    },
    {
      "epoch": 7.108303249097473,
      "grad_norm": 23.94330596923828,
      "learning_rate": 0.00019790685879846656,
      "loss": 7.7322,
      "step": 59070
    },
    {
      "epoch": 7.109506618531889,
      "grad_norm": 5.786437511444092,
      "learning_rate": 0.00019790608444234623,
      "loss": 7.5141,
      "step": 59080
    },
    {
      "epoch": 7.110709987966306,
      "grad_norm": 3.0996527671813965,
      "learning_rate": 0.0001979053099445317,
      "loss": 7.6377,
      "step": 59090
    },
    {
      "epoch": 7.111913357400722,
      "grad_norm": 8.763830184936523,
      "learning_rate": 0.00019790453530502396,
      "loss": 7.5446,
      "step": 59100
    },
    {
      "epoch": 7.1131167268351385,
      "grad_norm": 2.4363465309143066,
      "learning_rate": 0.00019790376052382418,
      "loss": 7.6016,
      "step": 59110
    },
    {
      "epoch": 7.114320096269555,
      "grad_norm": 3.958065986633301,
      "learning_rate": 0.00019790298560093349,
      "loss": 7.6158,
      "step": 59120
    },
    {
      "epoch": 7.115523465703971,
      "grad_norm": 9.829353332519531,
      "learning_rate": 0.00019790221053635302,
      "loss": 7.557,
      "step": 59130
    },
    {
      "epoch": 7.116726835138388,
      "grad_norm": 7.361923694610596,
      "learning_rate": 0.00019790143533008389,
      "loss": 7.5716,
      "step": 59140
    },
    {
      "epoch": 7.1179302045728035,
      "grad_norm": 5.595799446105957,
      "learning_rate": 0.0001979006599821272,
      "loss": 7.5141,
      "step": 59150
    },
    {
      "epoch": 7.11913357400722,
      "grad_norm": 6.777280807495117,
      "learning_rate": 0.00019789988449248408,
      "loss": 7.6255,
      "step": 59160
    },
    {
      "epoch": 7.120336943441637,
      "grad_norm": 13.410460472106934,
      "learning_rate": 0.00019789910886115566,
      "loss": 7.5592,
      "step": 59170
    },
    {
      "epoch": 7.121540312876053,
      "grad_norm": 8.098434448242188,
      "learning_rate": 0.00019789833308814304,
      "loss": 7.5126,
      "step": 59180
    },
    {
      "epoch": 7.122743682310469,
      "grad_norm": 3.511575937271118,
      "learning_rate": 0.00019789755717344737,
      "loss": 7.5427,
      "step": 59190
    },
    {
      "epoch": 7.123947051744886,
      "grad_norm": 5.895703315734863,
      "learning_rate": 0.00019789678111706977,
      "loss": 7.4838,
      "step": 59200
    },
    {
      "epoch": 7.125150421179302,
      "grad_norm": 584.3899536132812,
      "learning_rate": 0.00019789600491901136,
      "loss": 7.6162,
      "step": 59210
    },
    {
      "epoch": 7.126353790613718,
      "grad_norm": 7.532637119293213,
      "learning_rate": 0.00019789522857927328,
      "loss": 7.6417,
      "step": 59220
    },
    {
      "epoch": 7.127557160048135,
      "grad_norm": 4.052803993225098,
      "learning_rate": 0.00019789445209785663,
      "loss": 7.5396,
      "step": 59230
    },
    {
      "epoch": 7.128760529482551,
      "grad_norm": 4.805096626281738,
      "learning_rate": 0.0001978936754747625,
      "loss": 7.499,
      "step": 59240
    },
    {
      "epoch": 7.129963898916968,
      "grad_norm": 6.047024726867676,
      "learning_rate": 0.0001978928987099921,
      "loss": 7.4485,
      "step": 59250
    },
    {
      "epoch": 7.131167268351384,
      "grad_norm": 427.5868835449219,
      "learning_rate": 0.0001978921218035465,
      "loss": 7.5272,
      "step": 59260
    },
    {
      "epoch": 7.1323706377858,
      "grad_norm": 65.29325103759766,
      "learning_rate": 0.0001978913447554268,
      "loss": 8.2954,
      "step": 59270
    },
    {
      "epoch": 7.133574007220217,
      "grad_norm": 5.001557350158691,
      "learning_rate": 0.00019789056756563418,
      "loss": 8.0184,
      "step": 59280
    },
    {
      "epoch": 7.1347773766546325,
      "grad_norm": 13.563468933105469,
      "learning_rate": 0.00019788979023416974,
      "loss": 7.7271,
      "step": 59290
    },
    {
      "epoch": 7.135980746089049,
      "grad_norm": 4.750330924987793,
      "learning_rate": 0.00019788901276103458,
      "loss": 7.6179,
      "step": 59300
    },
    {
      "epoch": 7.137184115523466,
      "grad_norm": 3.3980605602264404,
      "learning_rate": 0.00019788823514622987,
      "loss": 7.5529,
      "step": 59310
    },
    {
      "epoch": 7.138387484957882,
      "grad_norm": 5.319908618927002,
      "learning_rate": 0.0001978874573897567,
      "loss": 7.514,
      "step": 59320
    },
    {
      "epoch": 7.139590854392298,
      "grad_norm": 3.2248036861419678,
      "learning_rate": 0.00019788667949161622,
      "loss": 7.5623,
      "step": 59330
    },
    {
      "epoch": 7.140794223826715,
      "grad_norm": 5.58474588394165,
      "learning_rate": 0.00019788590145180957,
      "loss": 7.6767,
      "step": 59340
    },
    {
      "epoch": 7.141997593261131,
      "grad_norm": 2.9544930458068848,
      "learning_rate": 0.00019788512327033786,
      "loss": 7.727,
      "step": 59350
    },
    {
      "epoch": 7.1432009626955475,
      "grad_norm": 1.7140352725982666,
      "learning_rate": 0.00019788434494720218,
      "loss": 7.4993,
      "step": 59360
    },
    {
      "epoch": 7.144404332129964,
      "grad_norm": 3.823256731033325,
      "learning_rate": 0.0001978835664824037,
      "loss": 7.538,
      "step": 59370
    },
    {
      "epoch": 7.14560770156438,
      "grad_norm": 4.857741832733154,
      "learning_rate": 0.0001978827878759435,
      "loss": 7.4998,
      "step": 59380
    },
    {
      "epoch": 7.146811070998797,
      "grad_norm": 22.899444580078125,
      "learning_rate": 0.00019788200912782277,
      "loss": 7.6122,
      "step": 59390
    },
    {
      "epoch": 7.148014440433213,
      "grad_norm": 5.2488837242126465,
      "learning_rate": 0.00019788123023804262,
      "loss": 7.5971,
      "step": 59400
    },
    {
      "epoch": 7.149217809867629,
      "grad_norm": 2.129460334777832,
      "learning_rate": 0.0001978804512066041,
      "loss": 7.5568,
      "step": 59410
    },
    {
      "epoch": 7.150421179302046,
      "grad_norm": 4.204578876495361,
      "learning_rate": 0.00019787967203350847,
      "loss": 7.5412,
      "step": 59420
    },
    {
      "epoch": 7.1516245487364625,
      "grad_norm": 3.908076763153076,
      "learning_rate": 0.00019787889271875677,
      "loss": 7.4903,
      "step": 59430
    },
    {
      "epoch": 7.152827918170878,
      "grad_norm": 3.013484001159668,
      "learning_rate": 0.00019787811326235012,
      "loss": 7.5484,
      "step": 59440
    },
    {
      "epoch": 7.154031287605295,
      "grad_norm": 11.153820991516113,
      "learning_rate": 0.00019787733366428972,
      "loss": 7.4296,
      "step": 59450
    },
    {
      "epoch": 7.155234657039712,
      "grad_norm": 4.014166831970215,
      "learning_rate": 0.00019787655392457663,
      "loss": 7.4667,
      "step": 59460
    },
    {
      "epoch": 7.156438026474127,
      "grad_norm": 5.698665142059326,
      "learning_rate": 0.000197875774043212,
      "loss": 7.529,
      "step": 59470
    },
    {
      "epoch": 7.157641395908544,
      "grad_norm": 12.536340713500977,
      "learning_rate": 0.00019787499402019695,
      "loss": 7.512,
      "step": 59480
    },
    {
      "epoch": 7.15884476534296,
      "grad_norm": 4.035852909088135,
      "learning_rate": 0.00019787421385553263,
      "loss": 7.4643,
      "step": 59490
    },
    {
      "epoch": 7.160048134777377,
      "grad_norm": 4.346449375152588,
      "learning_rate": 0.00019787343354922017,
      "loss": 7.4415,
      "step": 59500
    },
    {
      "epoch": 7.161251504211793,
      "grad_norm": 4.9524736404418945,
      "learning_rate": 0.00019787265310126066,
      "loss": 7.4527,
      "step": 59510
    },
    {
      "epoch": 7.162454873646209,
      "grad_norm": 8.1602144241333,
      "learning_rate": 0.00019787187251165528,
      "loss": 7.5317,
      "step": 59520
    },
    {
      "epoch": 7.163658243080626,
      "grad_norm": 21.558897018432617,
      "learning_rate": 0.00019787109178040512,
      "loss": 7.6129,
      "step": 59530
    },
    {
      "epoch": 7.164861612515042,
      "grad_norm": 4.18744421005249,
      "learning_rate": 0.00019787031090751133,
      "loss": 7.6206,
      "step": 59540
    },
    {
      "epoch": 7.166064981949458,
      "grad_norm": 4.385098934173584,
      "learning_rate": 0.00019786952989297505,
      "loss": 7.4972,
      "step": 59550
    },
    {
      "epoch": 7.167268351383875,
      "grad_norm": 7.401921272277832,
      "learning_rate": 0.00019786874873679739,
      "loss": 7.4209,
      "step": 59560
    },
    {
      "epoch": 7.1684717208182915,
      "grad_norm": 761.7949829101562,
      "learning_rate": 0.0001978679674389795,
      "loss": 7.4231,
      "step": 59570
    },
    {
      "epoch": 7.169675090252707,
      "grad_norm": 569.0382690429688,
      "learning_rate": 0.0001978671859995225,
      "loss": 7.4035,
      "step": 59580
    },
    {
      "epoch": 7.170878459687124,
      "grad_norm": 289.3324890136719,
      "learning_rate": 0.00019786640441842748,
      "loss": 7.9095,
      "step": 59590
    },
    {
      "epoch": 7.172081829121541,
      "grad_norm": 3.3403782844543457,
      "learning_rate": 0.00019786562269569563,
      "loss": 7.854,
      "step": 59600
    },
    {
      "epoch": 7.1732851985559565,
      "grad_norm": 207.3368682861328,
      "learning_rate": 0.00019786484083132808,
      "loss": 7.621,
      "step": 59610
    },
    {
      "epoch": 7.174488567990373,
      "grad_norm": 6.522215843200684,
      "learning_rate": 0.00019786405882532592,
      "loss": 7.569,
      "step": 59620
    },
    {
      "epoch": 7.17569193742479,
      "grad_norm": 556.5330810546875,
      "learning_rate": 0.0001978632766776903,
      "loss": 7.5669,
      "step": 59630
    },
    {
      "epoch": 7.176895306859206,
      "grad_norm": 3.6186258792877197,
      "learning_rate": 0.00019786249438842238,
      "loss": 7.4964,
      "step": 59640
    },
    {
      "epoch": 7.178098676293622,
      "grad_norm": 5.747904300689697,
      "learning_rate": 0.00019786171195752329,
      "loss": 7.5989,
      "step": 59650
    },
    {
      "epoch": 7.179302045728038,
      "grad_norm": 17.137287139892578,
      "learning_rate": 0.0001978609293849941,
      "loss": 7.498,
      "step": 59660
    },
    {
      "epoch": 7.180505415162455,
      "grad_norm": 3.883235216140747,
      "learning_rate": 0.000197860146670836,
      "loss": 7.5472,
      "step": 59670
    },
    {
      "epoch": 7.1817087845968715,
      "grad_norm": 4.300616264343262,
      "learning_rate": 0.0001978593638150501,
      "loss": 7.5632,
      "step": 59680
    },
    {
      "epoch": 7.182912154031287,
      "grad_norm": 1.7276138067245483,
      "learning_rate": 0.00019785858081763756,
      "loss": 7.6019,
      "step": 59690
    },
    {
      "epoch": 7.184115523465704,
      "grad_norm": 5.939334392547607,
      "learning_rate": 0.00019785779767859946,
      "loss": 7.5971,
      "step": 59700
    },
    {
      "epoch": 7.185318892900121,
      "grad_norm": 6.778968334197998,
      "learning_rate": 0.00019785701439793698,
      "loss": 7.5578,
      "step": 59710
    },
    {
      "epoch": 7.186522262334536,
      "grad_norm": 7.693382263183594,
      "learning_rate": 0.00019785623097565125,
      "loss": 7.6227,
      "step": 59720
    },
    {
      "epoch": 7.187725631768953,
      "grad_norm": 5.100826263427734,
      "learning_rate": 0.00019785544741174338,
      "loss": 7.4925,
      "step": 59730
    },
    {
      "epoch": 7.18892900120337,
      "grad_norm": 3.846651554107666,
      "learning_rate": 0.00019785466370621452,
      "loss": 7.5773,
      "step": 59740
    },
    {
      "epoch": 7.1901323706377855,
      "grad_norm": 6.780317783355713,
      "learning_rate": 0.00019785387985906582,
      "loss": 7.5043,
      "step": 59750
    },
    {
      "epoch": 7.191335740072202,
      "grad_norm": 6.981686115264893,
      "learning_rate": 0.00019785309587029839,
      "loss": 7.4389,
      "step": 59760
    },
    {
      "epoch": 7.192539109506619,
      "grad_norm": 16.288442611694336,
      "learning_rate": 0.00019785231173991336,
      "loss": 7.4425,
      "step": 59770
    },
    {
      "epoch": 7.193742478941035,
      "grad_norm": 26.31230926513672,
      "learning_rate": 0.00019785152746791188,
      "loss": 7.4794,
      "step": 59780
    },
    {
      "epoch": 7.194945848375451,
      "grad_norm": 19.139209747314453,
      "learning_rate": 0.00019785074305429508,
      "loss": 7.5565,
      "step": 59790
    },
    {
      "epoch": 7.196149217809868,
      "grad_norm": 67.08979797363281,
      "learning_rate": 0.0001978499584990641,
      "loss": 7.5342,
      "step": 59800
    },
    {
      "epoch": 7.197352587244284,
      "grad_norm": 17.288860321044922,
      "learning_rate": 0.0001978491738022201,
      "loss": 7.5384,
      "step": 59810
    },
    {
      "epoch": 7.1985559566787005,
      "grad_norm": 74.42042541503906,
      "learning_rate": 0.00019784838896376412,
      "loss": 7.4855,
      "step": 59820
    },
    {
      "epoch": 7.199759326113116,
      "grad_norm": 15.520068168640137,
      "learning_rate": 0.00019784760398369738,
      "loss": 7.4961,
      "step": 59830
    },
    {
      "epoch": 7.200962695547533,
      "grad_norm": 12.522257804870605,
      "learning_rate": 0.00019784681886202105,
      "loss": 7.4992,
      "step": 59840
    },
    {
      "epoch": 7.20216606498195,
      "grad_norm": 29.99500274658203,
      "learning_rate": 0.00019784603359873618,
      "loss": 7.5273,
      "step": 59850
    },
    {
      "epoch": 7.2033694344163655,
      "grad_norm": 51.02986526489258,
      "learning_rate": 0.00019784524819384394,
      "loss": 7.4981,
      "step": 59860
    },
    {
      "epoch": 7.204572803850782,
      "grad_norm": 30.489192962646484,
      "learning_rate": 0.00019784446264734545,
      "loss": 7.4909,
      "step": 59870
    },
    {
      "epoch": 7.205776173285199,
      "grad_norm": 33.188053131103516,
      "learning_rate": 0.0001978436769592419,
      "loss": 7.4641,
      "step": 59880
    },
    {
      "epoch": 7.206979542719615,
      "grad_norm": 19.068748474121094,
      "learning_rate": 0.00019784289112953442,
      "loss": 7.5546,
      "step": 59890
    },
    {
      "epoch": 7.208182912154031,
      "grad_norm": 30.090360641479492,
      "learning_rate": 0.00019784210515822408,
      "loss": 7.4671,
      "step": 59900
    },
    {
      "epoch": 7.209386281588448,
      "grad_norm": 29.924680709838867,
      "learning_rate": 0.00019784131904531204,
      "loss": 7.402,
      "step": 59910
    },
    {
      "epoch": 7.210589651022864,
      "grad_norm": 37.59492111206055,
      "learning_rate": 0.00019784053279079948,
      "loss": 7.4358,
      "step": 59920
    },
    {
      "epoch": 7.21179302045728,
      "grad_norm": 55.84379196166992,
      "learning_rate": 0.00019783974639468749,
      "loss": 7.5398,
      "step": 59930
    },
    {
      "epoch": 7.212996389891697,
      "grad_norm": 82.36194610595703,
      "learning_rate": 0.00019783895985697728,
      "loss": 7.4889,
      "step": 59940
    },
    {
      "epoch": 7.214199759326113,
      "grad_norm": 32.53596115112305,
      "learning_rate": 0.00019783817317766992,
      "loss": 7.4392,
      "step": 59950
    },
    {
      "epoch": 7.21540312876053,
      "grad_norm": 46.800846099853516,
      "learning_rate": 0.00019783738635676654,
      "loss": 7.5065,
      "step": 59960
    },
    {
      "epoch": 7.216606498194946,
      "grad_norm": 114.05599212646484,
      "learning_rate": 0.0001978365993942683,
      "loss": 7.4845,
      "step": 59970
    },
    {
      "epoch": 7.217809867629362,
      "grad_norm": 48.833648681640625,
      "learning_rate": 0.0001978358122901764,
      "loss": 7.4846,
      "step": 59980
    },
    {
      "epoch": 7.219013237063779,
      "grad_norm": 27.91322135925293,
      "learning_rate": 0.0001978350250444919,
      "loss": 7.4814,
      "step": 59990
    },
    {
      "epoch": 7.2202166064981945,
      "grad_norm": 42.391719818115234,
      "learning_rate": 0.00019783423765721593,
      "loss": 7.4513,
      "step": 60000
    },
    {
      "epoch": 7.221419975932611,
      "grad_norm": 27.37371253967285,
      "learning_rate": 0.0001978334501283497,
      "loss": 7.5,
      "step": 60010
    },
    {
      "epoch": 7.222623345367028,
      "grad_norm": 64.112548828125,
      "learning_rate": 0.0001978326624578943,
      "loss": 7.4711,
      "step": 60020
    },
    {
      "epoch": 7.223826714801444,
      "grad_norm": 85.39812469482422,
      "learning_rate": 0.0001978318746458509,
      "loss": 7.5022,
      "step": 60030
    },
    {
      "epoch": 7.22503008423586,
      "grad_norm": 17.54473876953125,
      "learning_rate": 0.0001978310866922206,
      "loss": 7.4464,
      "step": 60040
    },
    {
      "epoch": 7.226233453670277,
      "grad_norm": 46.40385818481445,
      "learning_rate": 0.0001978302985970046,
      "loss": 7.4783,
      "step": 60050
    },
    {
      "epoch": 7.227436823104693,
      "grad_norm": 23.35736656188965,
      "learning_rate": 0.00019782951036020397,
      "loss": 7.4859,
      "step": 60060
    },
    {
      "epoch": 7.2286401925391095,
      "grad_norm": 36.57982635498047,
      "learning_rate": 0.0001978287219818199,
      "loss": 7.5214,
      "step": 60070
    },
    {
      "epoch": 7.229843561973526,
      "grad_norm": 75.2618179321289,
      "learning_rate": 0.0001978279334618535,
      "loss": 7.3987,
      "step": 60080
    },
    {
      "epoch": 7.231046931407942,
      "grad_norm": 44.916038513183594,
      "learning_rate": 0.00019782714480030594,
      "loss": 7.4643,
      "step": 60090
    },
    {
      "epoch": 7.232250300842359,
      "grad_norm": 57.88205337524414,
      "learning_rate": 0.00019782635599717836,
      "loss": 7.4747,
      "step": 60100
    },
    {
      "epoch": 7.233453670276775,
      "grad_norm": 77.02476501464844,
      "learning_rate": 0.00019782556705247187,
      "loss": 7.5516,
      "step": 60110
    },
    {
      "epoch": 7.234657039711191,
      "grad_norm": 278.16485595703125,
      "learning_rate": 0.00019782477796618765,
      "loss": 7.4994,
      "step": 60120
    },
    {
      "epoch": 7.235860409145608,
      "grad_norm": 102.18984985351562,
      "learning_rate": 0.00019782398873832683,
      "loss": 7.3611,
      "step": 60130
    },
    {
      "epoch": 7.2370637785800245,
      "grad_norm": 136.10565185546875,
      "learning_rate": 0.0001978231993688905,
      "loss": 7.5627,
      "step": 60140
    },
    {
      "epoch": 7.23826714801444,
      "grad_norm": 67.03374481201172,
      "learning_rate": 0.0001978224098578799,
      "loss": 7.4815,
      "step": 60150
    },
    {
      "epoch": 7.239470517448857,
      "grad_norm": 155.654541015625,
      "learning_rate": 0.0001978216202052961,
      "loss": 7.5172,
      "step": 60160
    },
    {
      "epoch": 7.240673886883273,
      "grad_norm": 115.03605651855469,
      "learning_rate": 0.00019782083041114027,
      "loss": 7.5528,
      "step": 60170
    },
    {
      "epoch": 7.241877256317689,
      "grad_norm": 104.7567138671875,
      "learning_rate": 0.00019782004047541355,
      "loss": 7.5031,
      "step": 60180
    },
    {
      "epoch": 7.243080625752106,
      "grad_norm": 42.97132873535156,
      "learning_rate": 0.00019781925039811707,
      "loss": 7.477,
      "step": 60190
    },
    {
      "epoch": 7.244283995186522,
      "grad_norm": 48.480133056640625,
      "learning_rate": 0.00019781846017925196,
      "loss": 7.5043,
      "step": 60200
    },
    {
      "epoch": 7.245487364620939,
      "grad_norm": 42.670719146728516,
      "learning_rate": 0.00019781766981881945,
      "loss": 7.49,
      "step": 60210
    },
    {
      "epoch": 7.246690734055355,
      "grad_norm": 72.44229125976562,
      "learning_rate": 0.00019781687931682055,
      "loss": 7.5018,
      "step": 60220
    },
    {
      "epoch": 7.247894103489771,
      "grad_norm": 96.29901885986328,
      "learning_rate": 0.00019781608867325653,
      "loss": 7.3994,
      "step": 60230
    },
    {
      "epoch": 7.249097472924188,
      "grad_norm": 29.10788917541504,
      "learning_rate": 0.00019781529788812845,
      "loss": 7.5485,
      "step": 60240
    },
    {
      "epoch": 7.250300842358604,
      "grad_norm": 52.739845275878906,
      "learning_rate": 0.0001978145069614375,
      "loss": 7.4365,
      "step": 60250
    },
    {
      "epoch": 7.25150421179302,
      "grad_norm": 64.79094696044922,
      "learning_rate": 0.0001978137158931848,
      "loss": 7.5318,
      "step": 60260
    },
    {
      "epoch": 7.252707581227437,
      "grad_norm": 53.2910270690918,
      "learning_rate": 0.0001978129246833715,
      "loss": 7.5018,
      "step": 60270
    },
    {
      "epoch": 7.2539109506618535,
      "grad_norm": 33.322811126708984,
      "learning_rate": 0.00019781213333199874,
      "loss": 7.4412,
      "step": 60280
    },
    {
      "epoch": 7.255114320096269,
      "grad_norm": 45.863426208496094,
      "learning_rate": 0.0001978113418390677,
      "loss": 7.4449,
      "step": 60290
    },
    {
      "epoch": 7.256317689530686,
      "grad_norm": 61.004669189453125,
      "learning_rate": 0.0001978105502045795,
      "loss": 7.438,
      "step": 60300
    },
    {
      "epoch": 7.257521058965103,
      "grad_norm": 38.47679901123047,
      "learning_rate": 0.00019780975842853527,
      "loss": 7.5001,
      "step": 60310
    },
    {
      "epoch": 7.2587244283995185,
      "grad_norm": 135.99485778808594,
      "learning_rate": 0.0001978089665109362,
      "loss": 7.518,
      "step": 60320
    },
    {
      "epoch": 7.259927797833935,
      "grad_norm": 31.63204574584961,
      "learning_rate": 0.00019780817445178337,
      "loss": 7.5673,
      "step": 60330
    },
    {
      "epoch": 7.261131167268351,
      "grad_norm": 28.619653701782227,
      "learning_rate": 0.00019780738225107797,
      "loss": 7.5921,
      "step": 60340
    },
    {
      "epoch": 7.262334536702768,
      "grad_norm": 56.345943450927734,
      "learning_rate": 0.00019780658990882114,
      "loss": 7.4882,
      "step": 60350
    },
    {
      "epoch": 7.263537906137184,
      "grad_norm": 49.259464263916016,
      "learning_rate": 0.00019780579742501405,
      "loss": 7.5011,
      "step": 60360
    },
    {
      "epoch": 7.2647412755716,
      "grad_norm": 42.734397888183594,
      "learning_rate": 0.0001978050047996578,
      "loss": 7.4518,
      "step": 60370
    },
    {
      "epoch": 7.265944645006017,
      "grad_norm": 55.917720794677734,
      "learning_rate": 0.00019780421203275357,
      "loss": 7.3579,
      "step": 60380
    },
    {
      "epoch": 7.2671480144404335,
      "grad_norm": 44.459991455078125,
      "learning_rate": 0.0001978034191243025,
      "loss": 7.5129,
      "step": 60390
    },
    {
      "epoch": 7.268351383874849,
      "grad_norm": 58.6427116394043,
      "learning_rate": 0.00019780262607430572,
      "loss": 7.4381,
      "step": 60400
    },
    {
      "epoch": 7.269554753309266,
      "grad_norm": 100.82861328125,
      "learning_rate": 0.0001978018328827644,
      "loss": 7.4171,
      "step": 60410
    },
    {
      "epoch": 7.270758122743683,
      "grad_norm": 54.6920051574707,
      "learning_rate": 0.0001978010395496797,
      "loss": 7.4001,
      "step": 60420
    },
    {
      "epoch": 7.271961492178098,
      "grad_norm": 71.27852630615234,
      "learning_rate": 0.00019780024607505273,
      "loss": 7.4267,
      "step": 60430
    },
    {
      "epoch": 7.273164861612515,
      "grad_norm": 61.42023849487305,
      "learning_rate": 0.00019779945245888467,
      "loss": 7.4185,
      "step": 60440
    },
    {
      "epoch": 7.274368231046932,
      "grad_norm": 60.58082962036133,
      "learning_rate": 0.00019779865870117664,
      "loss": 7.4937,
      "step": 60450
    },
    {
      "epoch": 7.2755716004813475,
      "grad_norm": 57.872398376464844,
      "learning_rate": 0.00019779786480192982,
      "loss": 7.4687,
      "step": 60460
    },
    {
      "epoch": 7.276774969915764,
      "grad_norm": 78.12239074707031,
      "learning_rate": 0.00019779707076114536,
      "loss": 7.4062,
      "step": 60470
    },
    {
      "epoch": 7.277978339350181,
      "grad_norm": 53.899505615234375,
      "learning_rate": 0.00019779627657882435,
      "loss": 7.4423,
      "step": 60480
    },
    {
      "epoch": 7.279181708784597,
      "grad_norm": 70.50180053710938,
      "learning_rate": 0.000197795482254968,
      "loss": 7.4284,
      "step": 60490
    },
    {
      "epoch": 7.280385078219013,
      "grad_norm": 74.60441589355469,
      "learning_rate": 0.00019779468778957746,
      "loss": 7.4769,
      "step": 60500
    },
    {
      "epoch": 7.28158844765343,
      "grad_norm": 85.81940460205078,
      "learning_rate": 0.00019779389318265385,
      "loss": 7.476,
      "step": 60510
    },
    {
      "epoch": 7.282791817087846,
      "grad_norm": 88.97525024414062,
      "learning_rate": 0.00019779309843419833,
      "loss": 7.4764,
      "step": 60520
    },
    {
      "epoch": 7.2839951865222625,
      "grad_norm": 108.99325561523438,
      "learning_rate": 0.00019779230354421206,
      "loss": 7.4186,
      "step": 60530
    },
    {
      "epoch": 7.285198555956678,
      "grad_norm": 56.356689453125,
      "learning_rate": 0.00019779150851269618,
      "loss": 7.4352,
      "step": 60540
    },
    {
      "epoch": 7.286401925391095,
      "grad_norm": 128.97547912597656,
      "learning_rate": 0.00019779071333965185,
      "loss": 7.3784,
      "step": 60550
    },
    {
      "epoch": 7.287605294825512,
      "grad_norm": 103.07829284667969,
      "learning_rate": 0.0001977899180250802,
      "loss": 7.4154,
      "step": 60560
    },
    {
      "epoch": 7.2888086642599275,
      "grad_norm": 198.7111053466797,
      "learning_rate": 0.0001977891225689824,
      "loss": 7.4376,
      "step": 60570
    },
    {
      "epoch": 7.290012033694344,
      "grad_norm": 384.5240173339844,
      "learning_rate": 0.00019778832697135959,
      "loss": 7.4949,
      "step": 60580
    },
    {
      "epoch": 7.291215403128761,
      "grad_norm": 413.0855407714844,
      "learning_rate": 0.00019778753123221292,
      "loss": 7.4138,
      "step": 60590
    },
    {
      "epoch": 7.292418772563177,
      "grad_norm": 143.67127990722656,
      "learning_rate": 0.00019778673535154355,
      "loss": 7.4467,
      "step": 60600
    },
    {
      "epoch": 7.293622141997593,
      "grad_norm": 300.8535461425781,
      "learning_rate": 0.00019778593932935265,
      "loss": 7.4996,
      "step": 60610
    },
    {
      "epoch": 7.29482551143201,
      "grad_norm": 280.2632141113281,
      "learning_rate": 0.00019778514316564136,
      "loss": 7.3682,
      "step": 60620
    },
    {
      "epoch": 7.296028880866426,
      "grad_norm": 215.53408813476562,
      "learning_rate": 0.0001977843468604108,
      "loss": 7.6457,
      "step": 60630
    },
    {
      "epoch": 7.297232250300842,
      "grad_norm": 88.64797973632812,
      "learning_rate": 0.00019778355041366217,
      "loss": 9.2716,
      "step": 60640
    },
    {
      "epoch": 7.298435619735259,
      "grad_norm": 400.6197204589844,
      "learning_rate": 0.0001977827538253966,
      "loss": 8.1227,
      "step": 60650
    },
    {
      "epoch": 7.299638989169675,
      "grad_norm": 204.63186645507812,
      "learning_rate": 0.00019778195709561523,
      "loss": 7.9466,
      "step": 60660
    },
    {
      "epoch": 7.300842358604092,
      "grad_norm": 142.33651733398438,
      "learning_rate": 0.00019778116022431923,
      "loss": 7.8356,
      "step": 60670
    },
    {
      "epoch": 7.302045728038507,
      "grad_norm": 355.62103271484375,
      "learning_rate": 0.00019778036321150974,
      "loss": 8.2918,
      "step": 60680
    },
    {
      "epoch": 7.303249097472924,
      "grad_norm": 39.75912857055664,
      "learning_rate": 0.00019777956605718794,
      "loss": 8.4857,
      "step": 60690
    },
    {
      "epoch": 7.304452466907341,
      "grad_norm": 118.32836151123047,
      "learning_rate": 0.00019777876876135494,
      "loss": 7.6227,
      "step": 60700
    },
    {
      "epoch": 7.3056558363417565,
      "grad_norm": 36.87958908081055,
      "learning_rate": 0.00019777797132401195,
      "loss": 7.594,
      "step": 60710
    },
    {
      "epoch": 7.306859205776173,
      "grad_norm": 28.497652053833008,
      "learning_rate": 0.0001977771737451601,
      "loss": 7.4891,
      "step": 60720
    },
    {
      "epoch": 7.30806257521059,
      "grad_norm": 28.90912628173828,
      "learning_rate": 0.0001977763760248005,
      "loss": 7.5382,
      "step": 60730
    },
    {
      "epoch": 7.309265944645006,
      "grad_norm": 24.26543617248535,
      "learning_rate": 0.00019777557816293438,
      "loss": 7.4958,
      "step": 60740
    },
    {
      "epoch": 7.310469314079422,
      "grad_norm": 38.23131561279297,
      "learning_rate": 0.00019777478015956285,
      "loss": 7.4338,
      "step": 60750
    },
    {
      "epoch": 7.311672683513839,
      "grad_norm": 25.858304977416992,
      "learning_rate": 0.00019777398201468706,
      "loss": 7.4713,
      "step": 60760
    },
    {
      "epoch": 7.312876052948255,
      "grad_norm": 44.17701721191406,
      "learning_rate": 0.0001977731837283082,
      "loss": 7.5925,
      "step": 60770
    },
    {
      "epoch": 7.3140794223826715,
      "grad_norm": 63.03474426269531,
      "learning_rate": 0.00019777238530042742,
      "loss": 7.4646,
      "step": 60780
    },
    {
      "epoch": 7.315282791817088,
      "grad_norm": 72.42682647705078,
      "learning_rate": 0.00019777158673104582,
      "loss": 7.5022,
      "step": 60790
    },
    {
      "epoch": 7.316486161251504,
      "grad_norm": 64.22356414794922,
      "learning_rate": 0.0001977707880201646,
      "loss": 7.5057,
      "step": 60800
    },
    {
      "epoch": 7.317689530685921,
      "grad_norm": 38.25808334350586,
      "learning_rate": 0.00019776998916778494,
      "loss": 7.4618,
      "step": 60810
    },
    {
      "epoch": 7.318892900120337,
      "grad_norm": 143.3399200439453,
      "learning_rate": 0.0001977691901739079,
      "loss": 7.4346,
      "step": 60820
    },
    {
      "epoch": 7.320096269554753,
      "grad_norm": 87.22521209716797,
      "learning_rate": 0.00019776839103853477,
      "loss": 7.4397,
      "step": 60830
    },
    {
      "epoch": 7.32129963898917,
      "grad_norm": 66.5234603881836,
      "learning_rate": 0.0001977675917616666,
      "loss": 7.4557,
      "step": 60840
    },
    {
      "epoch": 7.3225030084235865,
      "grad_norm": 62.35301971435547,
      "learning_rate": 0.00019776679234330462,
      "loss": 7.4729,
      "step": 60850
    },
    {
      "epoch": 7.323706377858002,
      "grad_norm": 69.43376159667969,
      "learning_rate": 0.00019776599278344993,
      "loss": 7.4977,
      "step": 60860
    },
    {
      "epoch": 7.324909747292419,
      "grad_norm": 86.56552124023438,
      "learning_rate": 0.00019776519308210372,
      "loss": 7.4436,
      "step": 60870
    },
    {
      "epoch": 7.326113116726835,
      "grad_norm": 94.27783203125,
      "learning_rate": 0.00019776439323926717,
      "loss": 7.4941,
      "step": 60880
    },
    {
      "epoch": 7.327316486161251,
      "grad_norm": 61.91367721557617,
      "learning_rate": 0.00019776359325494137,
      "loss": 7.4784,
      "step": 60890
    },
    {
      "epoch": 7.328519855595668,
      "grad_norm": 97.73733520507812,
      "learning_rate": 0.0001977627931291275,
      "loss": 7.4395,
      "step": 60900
    },
    {
      "epoch": 7.329723225030084,
      "grad_norm": 66.14411163330078,
      "learning_rate": 0.00019776199286182676,
      "loss": 7.4584,
      "step": 60910
    },
    {
      "epoch": 7.330926594464501,
      "grad_norm": 65.3894271850586,
      "learning_rate": 0.00019776119245304026,
      "loss": 7.5061,
      "step": 60920
    },
    {
      "epoch": 7.332129963898917,
      "grad_norm": 69.93453979492188,
      "learning_rate": 0.0001977603919027692,
      "loss": 7.3627,
      "step": 60930
    },
    {
      "epoch": 7.333333333333333,
      "grad_norm": 63.80093765258789,
      "learning_rate": 0.00019775959121101468,
      "loss": 7.5226,
      "step": 60940
    },
    {
      "epoch": 7.33453670276775,
      "grad_norm": 106.47486114501953,
      "learning_rate": 0.0001977587903777779,
      "loss": 7.4226,
      "step": 60950
    },
    {
      "epoch": 7.335740072202166,
      "grad_norm": 314.5417785644531,
      "learning_rate": 0.00019775798940306006,
      "loss": 7.4833,
      "step": 60960
    },
    {
      "epoch": 7.336943441636582,
      "grad_norm": 137.12054443359375,
      "learning_rate": 0.00019775718828686224,
      "loss": 7.4836,
      "step": 60970
    },
    {
      "epoch": 7.338146811070999,
      "grad_norm": 115.8062973022461,
      "learning_rate": 0.00019775638702918563,
      "loss": 7.3838,
      "step": 60980
    },
    {
      "epoch": 7.3393501805054155,
      "grad_norm": 128.8003387451172,
      "learning_rate": 0.0001977555856300314,
      "loss": 7.5599,
      "step": 60990
    },
    {
      "epoch": 7.340553549939831,
      "grad_norm": 103.74386596679688,
      "learning_rate": 0.0001977547840894007,
      "loss": 7.4646,
      "step": 61000
    },
    {
      "epoch": 7.341756919374248,
      "grad_norm": 101.12113952636719,
      "learning_rate": 0.0001977539824072947,
      "loss": 7.4665,
      "step": 61010
    },
    {
      "epoch": 7.342960288808664,
      "grad_norm": 116.65435028076172,
      "learning_rate": 0.00019775318058371457,
      "loss": 7.476,
      "step": 61020
    },
    {
      "epoch": 7.3441636582430805,
      "grad_norm": 67.12080383300781,
      "learning_rate": 0.00019775237861866142,
      "loss": 7.4221,
      "step": 61030
    },
    {
      "epoch": 7.345367027677497,
      "grad_norm": 93.72984313964844,
      "learning_rate": 0.00019775157651213645,
      "loss": 7.418,
      "step": 61040
    },
    {
      "epoch": 7.346570397111913,
      "grad_norm": 130.2222442626953,
      "learning_rate": 0.00019775077426414085,
      "loss": 7.412,
      "step": 61050
    },
    {
      "epoch": 7.34777376654633,
      "grad_norm": 134.0852813720703,
      "learning_rate": 0.00019774997187467573,
      "loss": 7.37,
      "step": 61060
    },
    {
      "epoch": 7.348977135980746,
      "grad_norm": 255.88497924804688,
      "learning_rate": 0.00019774916934374223,
      "loss": 7.463,
      "step": 61070
    },
    {
      "epoch": 7.350180505415162,
      "grad_norm": 97.97000122070312,
      "learning_rate": 0.00019774836667134157,
      "loss": 7.436,
      "step": 61080
    },
    {
      "epoch": 7.351383874849579,
      "grad_norm": 79.88434600830078,
      "learning_rate": 0.0001977475638574749,
      "loss": 7.4323,
      "step": 61090
    },
    {
      "epoch": 7.3525872442839955,
      "grad_norm": 204.21705627441406,
      "learning_rate": 0.00019774676090214338,
      "loss": 7.2863,
      "step": 61100
    },
    {
      "epoch": 7.353790613718411,
      "grad_norm": 111.7914047241211,
      "learning_rate": 0.00019774595780534812,
      "loss": 7.4165,
      "step": 61110
    },
    {
      "epoch": 7.354993983152828,
      "grad_norm": 159.9964599609375,
      "learning_rate": 0.00019774515456709034,
      "loss": 7.3932,
      "step": 61120
    },
    {
      "epoch": 7.356197352587245,
      "grad_norm": 174.8877716064453,
      "learning_rate": 0.00019774435118737125,
      "loss": 7.488,
      "step": 61130
    },
    {
      "epoch": 7.35740072202166,
      "grad_norm": 81.28499603271484,
      "learning_rate": 0.00019774354766619186,
      "loss": 7.4149,
      "step": 61140
    },
    {
      "epoch": 7.358604091456077,
      "grad_norm": 166.77029418945312,
      "learning_rate": 0.00019774274400355347,
      "loss": 7.3922,
      "step": 61150
    },
    {
      "epoch": 7.359807460890494,
      "grad_norm": 98.10020446777344,
      "learning_rate": 0.00019774194019945722,
      "loss": 7.4174,
      "step": 61160
    },
    {
      "epoch": 7.3610108303249095,
      "grad_norm": 199.62445068359375,
      "learning_rate": 0.00019774113625390424,
      "loss": 7.4153,
      "step": 61170
    },
    {
      "epoch": 7.362214199759326,
      "grad_norm": 248.86231994628906,
      "learning_rate": 0.0001977403321668957,
      "loss": 7.3757,
      "step": 61180
    },
    {
      "epoch": 7.363417569193743,
      "grad_norm": 169.08200073242188,
      "learning_rate": 0.00019773952793843274,
      "loss": 7.4045,
      "step": 61190
    },
    {
      "epoch": 7.364620938628159,
      "grad_norm": 237.693359375,
      "learning_rate": 0.00019773872356851658,
      "loss": 7.3645,
      "step": 61200
    },
    {
      "epoch": 7.365824308062575,
      "grad_norm": 181.377685546875,
      "learning_rate": 0.00019773791905714836,
      "loss": 7.4149,
      "step": 61210
    },
    {
      "epoch": 7.367027677496991,
      "grad_norm": 570.8060302734375,
      "learning_rate": 0.00019773711440432924,
      "loss": 7.482,
      "step": 61220
    },
    {
      "epoch": 7.368231046931408,
      "grad_norm": 692.4110107421875,
      "learning_rate": 0.00019773630961006036,
      "loss": 7.43,
      "step": 61230
    },
    {
      "epoch": 7.3694344163658245,
      "grad_norm": 145.51316833496094,
      "learning_rate": 0.00019773550467434291,
      "loss": 7.4216,
      "step": 61240
    },
    {
      "epoch": 7.37063778580024,
      "grad_norm": 515.2535400390625,
      "learning_rate": 0.00019773469959717807,
      "loss": 7.3935,
      "step": 61250
    },
    {
      "epoch": 7.371841155234657,
      "grad_norm": 300.3969421386719,
      "learning_rate": 0.000197733894378567,
      "loss": 7.4292,
      "step": 61260
    },
    {
      "epoch": 7.373044524669074,
      "grad_norm": 316.3150634765625,
      "learning_rate": 0.00019773308901851084,
      "loss": 7.4575,
      "step": 61270
    },
    {
      "epoch": 7.3742478941034895,
      "grad_norm": 779.3385009765625,
      "learning_rate": 0.00019773228351701078,
      "loss": 7.5211,
      "step": 61280
    },
    {
      "epoch": 7.375451263537906,
      "grad_norm": 161.46609497070312,
      "learning_rate": 0.00019773147787406796,
      "loss": 7.4393,
      "step": 61290
    },
    {
      "epoch": 7.376654632972323,
      "grad_norm": 468.7433166503906,
      "learning_rate": 0.00019773067208968357,
      "loss": 7.3958,
      "step": 61300
    },
    {
      "epoch": 7.377858002406739,
      "grad_norm": 550.7386474609375,
      "learning_rate": 0.00019772986616385877,
      "loss": 7.4006,
      "step": 61310
    },
    {
      "epoch": 7.379061371841155,
      "grad_norm": 420.22088623046875,
      "learning_rate": 0.00019772906009659472,
      "loss": 7.4031,
      "step": 61320
    },
    {
      "epoch": 7.380264741275572,
      "grad_norm": 334.4451599121094,
      "learning_rate": 0.00019772825388789263,
      "loss": 7.5428,
      "step": 61330
    },
    {
      "epoch": 7.381468110709988,
      "grad_norm": 465.0753173828125,
      "learning_rate": 0.00019772744753775357,
      "loss": 7.583,
      "step": 61340
    },
    {
      "epoch": 7.382671480144404,
      "grad_norm": 882.8710327148438,
      "learning_rate": 0.0001977266410461788,
      "loss": 7.404,
      "step": 61350
    },
    {
      "epoch": 7.38387484957882,
      "grad_norm": 407.3564453125,
      "learning_rate": 0.00019772583441316944,
      "loss": 7.6056,
      "step": 61360
    },
    {
      "epoch": 7.385078219013237,
      "grad_norm": 195.94752502441406,
      "learning_rate": 0.00019772502763872667,
      "loss": 7.565,
      "step": 61370
    },
    {
      "epoch": 7.386281588447654,
      "grad_norm": 758.3787231445312,
      "learning_rate": 0.00019772422072285165,
      "loss": 7.4727,
      "step": 61380
    },
    {
      "epoch": 7.387484957882069,
      "grad_norm": 115.20805358886719,
      "learning_rate": 0.00019772341366554557,
      "loss": 7.6022,
      "step": 61390
    },
    {
      "epoch": 7.388688327316486,
      "grad_norm": 900.2510986328125,
      "learning_rate": 0.0001977226064668096,
      "loss": 7.4808,
      "step": 61400
    },
    {
      "epoch": 7.389891696750903,
      "grad_norm": 91.54452514648438,
      "learning_rate": 0.00019772179912664486,
      "loss": 7.5298,
      "step": 61410
    },
    {
      "epoch": 7.3910950661853185,
      "grad_norm": 98.03681182861328,
      "learning_rate": 0.00019772099164505258,
      "loss": 7.7436,
      "step": 61420
    },
    {
      "epoch": 7.392298435619735,
      "grad_norm": 71757.5078125,
      "learning_rate": 0.00019772018402203388,
      "loss": 7.7631,
      "step": 61430
    },
    {
      "epoch": 7.393501805054152,
      "grad_norm": 6845.66796875,
      "learning_rate": 0.00019771937625758993,
      "loss": 8.3932,
      "step": 61440
    },
    {
      "epoch": 7.394705174488568,
      "grad_norm": 3221.813232421875,
      "learning_rate": 0.00019771856835172193,
      "loss": 8.2226,
      "step": 61450
    },
    {
      "epoch": 7.395908543922984,
      "grad_norm": 7.67422342300415,
      "learning_rate": 0.00019771776030443104,
      "loss": 8.0501,
      "step": 61460
    },
    {
      "epoch": 7.397111913357401,
      "grad_norm": 7.596240520477295,
      "learning_rate": 0.00019771695211571845,
      "loss": 7.731,
      "step": 61470
    },
    {
      "epoch": 7.398315282791817,
      "grad_norm": 2.8515572547912598,
      "learning_rate": 0.00019771614378558526,
      "loss": 7.6125,
      "step": 61480
    },
    {
      "epoch": 7.3995186522262335,
      "grad_norm": 4.044766426086426,
      "learning_rate": 0.00019771533531403273,
      "loss": 7.5527,
      "step": 61490
    },
    {
      "epoch": 7.40072202166065,
      "grad_norm": 2.84025239944458,
      "learning_rate": 0.00019771452670106195,
      "loss": 7.6032,
      "step": 61500
    },
    {
      "epoch": 7.401925391095066,
      "grad_norm": 6.143985271453857,
      "learning_rate": 0.00019771371794667414,
      "loss": 7.5401,
      "step": 61510
    },
    {
      "epoch": 7.403128760529483,
      "grad_norm": 3.8663153648376465,
      "learning_rate": 0.00019771290905087042,
      "loss": 7.5409,
      "step": 61520
    },
    {
      "epoch": 7.404332129963899,
      "grad_norm": 3.899678945541382,
      "learning_rate": 0.00019771210001365204,
      "loss": 7.6064,
      "step": 61530
    },
    {
      "epoch": 7.405535499398315,
      "grad_norm": 9.156954765319824,
      "learning_rate": 0.00019771129083502012,
      "loss": 7.6168,
      "step": 61540
    },
    {
      "epoch": 7.406738868832732,
      "grad_norm": 7.387956619262695,
      "learning_rate": 0.00019771048151497582,
      "loss": 7.5788,
      "step": 61550
    },
    {
      "epoch": 7.4079422382671485,
      "grad_norm": 5.763236999511719,
      "learning_rate": 0.00019770967205352035,
      "loss": 7.5211,
      "step": 61560
    },
    {
      "epoch": 7.409145607701564,
      "grad_norm": 2.982830047607422,
      "learning_rate": 0.00019770886245065486,
      "loss": 7.6556,
      "step": 61570
    },
    {
      "epoch": 7.410348977135981,
      "grad_norm": 5.544549942016602,
      "learning_rate": 0.00019770805270638052,
      "loss": 7.5588,
      "step": 61580
    },
    {
      "epoch": 7.411552346570397,
      "grad_norm": 8.750646591186523,
      "learning_rate": 0.00019770724282069848,
      "loss": 7.652,
      "step": 61590
    },
    {
      "epoch": 7.412755716004813,
      "grad_norm": 4.386682033538818,
      "learning_rate": 0.00019770643279360997,
      "loss": 7.591,
      "step": 61600
    },
    {
      "epoch": 7.41395908543923,
      "grad_norm": 6.326420783996582,
      "learning_rate": 0.0001977056226251161,
      "loss": 7.6151,
      "step": 61610
    },
    {
      "epoch": 7.415162454873646,
      "grad_norm": 4.076996326446533,
      "learning_rate": 0.0001977048123152181,
      "loss": 7.5841,
      "step": 61620
    },
    {
      "epoch": 7.416365824308063,
      "grad_norm": 2.429680824279785,
      "learning_rate": 0.00019770400186391708,
      "loss": 7.5782,
      "step": 61630
    },
    {
      "epoch": 7.417569193742479,
      "grad_norm": 3.590900421142578,
      "learning_rate": 0.00019770319127121425,
      "loss": 7.5815,
      "step": 61640
    },
    {
      "epoch": 7.418772563176895,
      "grad_norm": 5.025684833526611,
      "learning_rate": 0.0001977023805371108,
      "loss": 7.5659,
      "step": 61650
    },
    {
      "epoch": 7.419975932611312,
      "grad_norm": 1.833849549293518,
      "learning_rate": 0.0001977015696616079,
      "loss": 7.5136,
      "step": 61660
    },
    {
      "epoch": 7.421179302045728,
      "grad_norm": 2.585576057434082,
      "learning_rate": 0.00019770075864470666,
      "loss": 7.6369,
      "step": 61670
    },
    {
      "epoch": 7.422382671480144,
      "grad_norm": 1.9070616960525513,
      "learning_rate": 0.00019769994748640833,
      "loss": 7.6164,
      "step": 61680
    },
    {
      "epoch": 7.423586040914561,
      "grad_norm": 8.042707443237305,
      "learning_rate": 0.00019769913618671405,
      "loss": 7.6009,
      "step": 61690
    },
    {
      "epoch": 7.4247894103489775,
      "grad_norm": 1.3596677780151367,
      "learning_rate": 0.000197698324745625,
      "loss": 7.5459,
      "step": 61700
    },
    {
      "epoch": 7.425992779783393,
      "grad_norm": 3.9862680435180664,
      "learning_rate": 0.00019769751316314237,
      "loss": 7.6482,
      "step": 61710
    },
    {
      "epoch": 7.42719614921781,
      "grad_norm": 9.697206497192383,
      "learning_rate": 0.0001976967014392673,
      "loss": 7.7043,
      "step": 61720
    },
    {
      "epoch": 7.428399518652226,
      "grad_norm": 6.146006107330322,
      "learning_rate": 0.00019769588957400095,
      "loss": 7.6642,
      "step": 61730
    },
    {
      "epoch": 7.4296028880866425,
      "grad_norm": 7.936337471008301,
      "learning_rate": 0.00019769507756734457,
      "loss": 7.6674,
      "step": 61740
    },
    {
      "epoch": 7.430806257521059,
      "grad_norm": 3.8579227924346924,
      "learning_rate": 0.0001976942654192993,
      "loss": 7.5317,
      "step": 61750
    },
    {
      "epoch": 7.432009626955475,
      "grad_norm": 1.0830788612365723,
      "learning_rate": 0.0001976934531298663,
      "loss": 7.5707,
      "step": 61760
    },
    {
      "epoch": 7.433212996389892,
      "grad_norm": 4.701448917388916,
      "learning_rate": 0.00019769264069904673,
      "loss": 7.5479,
      "step": 61770
    },
    {
      "epoch": 7.434416365824308,
      "grad_norm": 5.61253547668457,
      "learning_rate": 0.0001976918281268418,
      "loss": 7.5542,
      "step": 61780
    },
    {
      "epoch": 7.435619735258724,
      "grad_norm": 1.6608448028564453,
      "learning_rate": 0.00019769101541325268,
      "loss": 7.5943,
      "step": 61790
    },
    {
      "epoch": 7.436823104693141,
      "grad_norm": 2.3284783363342285,
      "learning_rate": 0.00019769020255828055,
      "loss": 7.6296,
      "step": 61800
    },
    {
      "epoch": 7.4380264741275575,
      "grad_norm": 1.209886908531189,
      "learning_rate": 0.00019768938956192655,
      "loss": 7.5939,
      "step": 61810
    },
    {
      "epoch": 7.439229843561973,
      "grad_norm": 6.6313958168029785,
      "learning_rate": 0.00019768857642419192,
      "loss": 7.589,
      "step": 61820
    },
    {
      "epoch": 7.44043321299639,
      "grad_norm": 2.033318281173706,
      "learning_rate": 0.00019768776314507778,
      "loss": 7.5264,
      "step": 61830
    },
    {
      "epoch": 7.441636582430807,
      "grad_norm": 3.16302490234375,
      "learning_rate": 0.00019768694972458533,
      "loss": 7.6112,
      "step": 61840
    },
    {
      "epoch": 7.442839951865222,
      "grad_norm": 2.6318798065185547,
      "learning_rate": 0.00019768613616271576,
      "loss": 7.5498,
      "step": 61850
    },
    {
      "epoch": 7.444043321299639,
      "grad_norm": 3.1773931980133057,
      "learning_rate": 0.00019768532245947022,
      "loss": 7.5234,
      "step": 61860
    },
    {
      "epoch": 7.445246690734056,
      "grad_norm": 4.611644268035889,
      "learning_rate": 0.00019768450861484992,
      "loss": 7.5847,
      "step": 61870
    },
    {
      "epoch": 7.4464500601684716,
      "grad_norm": 3.9233875274658203,
      "learning_rate": 0.00019768369462885597,
      "loss": 7.508,
      "step": 61880
    },
    {
      "epoch": 7.447653429602888,
      "grad_norm": 3.1735899448394775,
      "learning_rate": 0.00019768288050148964,
      "loss": 7.5514,
      "step": 61890
    },
    {
      "epoch": 7.448856799037305,
      "grad_norm": 1.643195390701294,
      "learning_rate": 0.00019768206623275205,
      "loss": 7.6471,
      "step": 61900
    },
    {
      "epoch": 7.450060168471721,
      "grad_norm": 1.9416474103927612,
      "learning_rate": 0.0001976812518226444,
      "loss": 7.6117,
      "step": 61910
    },
    {
      "epoch": 7.451263537906137,
      "grad_norm": 5.642084121704102,
      "learning_rate": 0.00019768043727116786,
      "loss": 7.6559,
      "step": 61920
    },
    {
      "epoch": 7.452466907340553,
      "grad_norm": 2.461118221282959,
      "learning_rate": 0.0001976796225783236,
      "loss": 7.6176,
      "step": 61930
    },
    {
      "epoch": 7.45367027677497,
      "grad_norm": 1.362603783607483,
      "learning_rate": 0.00019767880774411282,
      "loss": 7.5749,
      "step": 61940
    },
    {
      "epoch": 7.4548736462093865,
      "grad_norm": 1.9335479736328125,
      "learning_rate": 0.0001976779927685367,
      "loss": 7.5407,
      "step": 61950
    },
    {
      "epoch": 7.456077015643802,
      "grad_norm": 27.64781379699707,
      "learning_rate": 0.0001976771776515964,
      "loss": 8.4723,
      "step": 61960
    },
    {
      "epoch": 7.457280385078219,
      "grad_norm": 3.2767229080200195,
      "learning_rate": 0.0001976763623932931,
      "loss": 7.6751,
      "step": 61970
    },
    {
      "epoch": 7.458483754512636,
      "grad_norm": 4.256045818328857,
      "learning_rate": 0.00019767554699362798,
      "loss": 7.6952,
      "step": 61980
    },
    {
      "epoch": 7.4596871239470515,
      "grad_norm": 4.228376865386963,
      "learning_rate": 0.00019767473145260221,
      "loss": 7.5736,
      "step": 61990
    },
    {
      "epoch": 7.460890493381468,
      "grad_norm": 3.365741491317749,
      "learning_rate": 0.00019767391577021703,
      "loss": 7.5411,
      "step": 62000
    },
    {
      "epoch": 7.462093862815885,
      "grad_norm": 3.4032626152038574,
      "learning_rate": 0.00019767309994647359,
      "loss": 7.5595,
      "step": 62010
    },
    {
      "epoch": 7.463297232250301,
      "grad_norm": 4.338688373565674,
      "learning_rate": 0.00019767228398137303,
      "loss": 7.6592,
      "step": 62020
    },
    {
      "epoch": 7.464500601684717,
      "grad_norm": 2.9457218647003174,
      "learning_rate": 0.00019767146787491658,
      "loss": 7.5986,
      "step": 62030
    },
    {
      "epoch": 7.465703971119134,
      "grad_norm": 3.9267494678497314,
      "learning_rate": 0.00019767065162710535,
      "loss": 7.5062,
      "step": 62040
    },
    {
      "epoch": 7.46690734055355,
      "grad_norm": 3.50807785987854,
      "learning_rate": 0.00019766983523794064,
      "loss": 7.5115,
      "step": 62050
    },
    {
      "epoch": 7.4681107099879664,
      "grad_norm": 2.1032209396362305,
      "learning_rate": 0.00019766901870742353,
      "loss": 7.6017,
      "step": 62060
    },
    {
      "epoch": 7.469314079422382,
      "grad_norm": 2.7084898948669434,
      "learning_rate": 0.00019766820203555524,
      "loss": 7.5257,
      "step": 62070
    },
    {
      "epoch": 7.470517448856799,
      "grad_norm": 2.000941514968872,
      "learning_rate": 0.00019766738522233697,
      "loss": 7.5845,
      "step": 62080
    },
    {
      "epoch": 7.471720818291216,
      "grad_norm": 1.2094730138778687,
      "learning_rate": 0.00019766656826776985,
      "loss": 7.5537,
      "step": 62090
    },
    {
      "epoch": 7.472924187725631,
      "grad_norm": 5.074915409088135,
      "learning_rate": 0.0001976657511718551,
      "loss": 7.6043,
      "step": 62100
    },
    {
      "epoch": 7.474127557160048,
      "grad_norm": 4.014949798583984,
      "learning_rate": 0.00019766493393459392,
      "loss": 7.6621,
      "step": 62110
    },
    {
      "epoch": 7.475330926594465,
      "grad_norm": 3.9575960636138916,
      "learning_rate": 0.00019766411655598742,
      "loss": 7.6231,
      "step": 62120
    },
    {
      "epoch": 7.4765342960288805,
      "grad_norm": 6.740754127502441,
      "learning_rate": 0.0001976632990360369,
      "loss": 7.5467,
      "step": 62130
    },
    {
      "epoch": 7.477737665463297,
      "grad_norm": 2.403111219406128,
      "learning_rate": 0.0001976624813747434,
      "loss": 7.6174,
      "step": 62140
    },
    {
      "epoch": 7.478941034897714,
      "grad_norm": 4.091782093048096,
      "learning_rate": 0.0001976616635721082,
      "loss": 7.494,
      "step": 62150
    },
    {
      "epoch": 7.48014440433213,
      "grad_norm": 2.9781603813171387,
      "learning_rate": 0.0001976608456281325,
      "loss": 7.5424,
      "step": 62160
    },
    {
      "epoch": 7.481347773766546,
      "grad_norm": 1.9330450296401978,
      "learning_rate": 0.0001976600275428174,
      "loss": 7.5204,
      "step": 62170
    },
    {
      "epoch": 7.482551143200963,
      "grad_norm": 1.7493013143539429,
      "learning_rate": 0.00019765920931616413,
      "loss": 7.6254,
      "step": 62180
    },
    {
      "epoch": 7.483754512635379,
      "grad_norm": 4.262869358062744,
      "learning_rate": 0.00019765839094817392,
      "loss": 7.5192,
      "step": 62190
    },
    {
      "epoch": 7.4849578820697955,
      "grad_norm": 1.4905524253845215,
      "learning_rate": 0.00019765757243884785,
      "loss": 7.5038,
      "step": 62200
    },
    {
      "epoch": 7.486161251504212,
      "grad_norm": 2.74172306060791,
      "learning_rate": 0.0001976567537881872,
      "loss": 7.5698,
      "step": 62210
    },
    {
      "epoch": 7.487364620938628,
      "grad_norm": 4.381186485290527,
      "learning_rate": 0.0001976559349961931,
      "loss": 7.6151,
      "step": 62220
    },
    {
      "epoch": 7.488567990373045,
      "grad_norm": 3.1179680824279785,
      "learning_rate": 0.00019765511606286676,
      "loss": 7.5753,
      "step": 62230
    },
    {
      "epoch": 7.489771359807461,
      "grad_norm": 2.6024482250213623,
      "learning_rate": 0.00019765429698820935,
      "loss": 7.4814,
      "step": 62240
    },
    {
      "epoch": 7.490974729241877,
      "grad_norm": 2.894402503967285,
      "learning_rate": 0.00019765347777222204,
      "loss": 7.5792,
      "step": 62250
    },
    {
      "epoch": 7.492178098676294,
      "grad_norm": 1.1348998546600342,
      "learning_rate": 0.00019765265841490608,
      "loss": 7.5516,
      "step": 62260
    },
    {
      "epoch": 7.49338146811071,
      "grad_norm": 3.59531569480896,
      "learning_rate": 0.00019765183891626257,
      "loss": 7.5775,
      "step": 62270
    },
    {
      "epoch": 7.494584837545126,
      "grad_norm": 5.755243301391602,
      "learning_rate": 0.00019765101927629277,
      "loss": 7.5668,
      "step": 62280
    },
    {
      "epoch": 7.495788206979543,
      "grad_norm": 1.2266857624053955,
      "learning_rate": 0.00019765019949499784,
      "loss": 7.5431,
      "step": 62290
    },
    {
      "epoch": 7.496991576413959,
      "grad_norm": 8.78464412689209,
      "learning_rate": 0.00019764937957237897,
      "loss": 7.5258,
      "step": 62300
    },
    {
      "epoch": 7.498194945848375,
      "grad_norm": 4.158756732940674,
      "learning_rate": 0.0001976485595084373,
      "loss": 7.5491,
      "step": 62310
    },
    {
      "epoch": 7.499398315282792,
      "grad_norm": 4.620486736297607,
      "learning_rate": 0.00019764773930317409,
      "loss": 7.6269,
      "step": 62320
    },
    {
      "epoch": 7.500601684717208,
      "grad_norm": 2.0696299076080322,
      "learning_rate": 0.00019764691895659045,
      "loss": 7.5648,
      "step": 62330
    },
    {
      "epoch": 7.501805054151625,
      "grad_norm": 5.2726826667785645,
      "learning_rate": 0.00019764609846868766,
      "loss": 7.6527,
      "step": 62340
    },
    {
      "epoch": 7.503008423586041,
      "grad_norm": 1.4752849340438843,
      "learning_rate": 0.00019764527783946686,
      "loss": 7.5715,
      "step": 62350
    },
    {
      "epoch": 7.504211793020457,
      "grad_norm": 3.716099262237549,
      "learning_rate": 0.00019764445706892917,
      "loss": 7.4312,
      "step": 62360
    },
    {
      "epoch": 7.505415162454874,
      "grad_norm": 2.872321605682373,
      "learning_rate": 0.00019764363615707592,
      "loss": 7.5898,
      "step": 62370
    },
    {
      "epoch": 7.5066185318892895,
      "grad_norm": 2.2229344844818115,
      "learning_rate": 0.00019764281510390816,
      "loss": 7.6523,
      "step": 62380
    },
    {
      "epoch": 7.507821901323706,
      "grad_norm": 3.699789047241211,
      "learning_rate": 0.00019764199390942717,
      "loss": 7.4637,
      "step": 62390
    },
    {
      "epoch": 7.509025270758123,
      "grad_norm": 4.786972999572754,
      "learning_rate": 0.0001976411725736341,
      "loss": 7.6154,
      "step": 62400
    },
    {
      "epoch": 7.510228640192539,
      "grad_norm": 2.555964946746826,
      "learning_rate": 0.00019764035109653017,
      "loss": 7.5778,
      "step": 62410
    },
    {
      "epoch": 7.511432009626955,
      "grad_norm": 1.9561940431594849,
      "learning_rate": 0.0001976395294781165,
      "loss": 7.5444,
      "step": 62420
    },
    {
      "epoch": 7.512635379061372,
      "grad_norm": 5.055975914001465,
      "learning_rate": 0.00019763870771839437,
      "loss": 7.5536,
      "step": 62430
    },
    {
      "epoch": 7.513838748495788,
      "grad_norm": 1.9030122756958008,
      "learning_rate": 0.0001976378858173649,
      "loss": 7.5497,
      "step": 62440
    },
    {
      "epoch": 7.5150421179302045,
      "grad_norm": 1.4870508909225464,
      "learning_rate": 0.0001976370637750293,
      "loss": 7.6542,
      "step": 62450
    },
    {
      "epoch": 7.516245487364621,
      "grad_norm": 1.8990765810012817,
      "learning_rate": 0.00019763624159138877,
      "loss": 7.6043,
      "step": 62460
    },
    {
      "epoch": 7.517448856799037,
      "grad_norm": 1.8807646036148071,
      "learning_rate": 0.00019763541926644452,
      "loss": 7.6038,
      "step": 62470
    },
    {
      "epoch": 7.518652226233454,
      "grad_norm": 1.7417924404144287,
      "learning_rate": 0.00019763459680019767,
      "loss": 7.5579,
      "step": 62480
    },
    {
      "epoch": 7.51985559566787,
      "grad_norm": 3.7538490295410156,
      "learning_rate": 0.00019763377419264945,
      "loss": 7.5665,
      "step": 62490
    },
    {
      "epoch": 7.521058965102286,
      "grad_norm": 3.280259132385254,
      "learning_rate": 0.00019763295144380107,
      "loss": 7.6211,
      "step": 62500
    },
    {
      "epoch": 7.522262334536703,
      "grad_norm": 2.8466787338256836,
      "learning_rate": 0.00019763212855365372,
      "loss": 7.5859,
      "step": 62510
    },
    {
      "epoch": 7.5234657039711195,
      "grad_norm": 2.8797900676727295,
      "learning_rate": 0.00019763130552220856,
      "loss": 7.6329,
      "step": 62520
    },
    {
      "epoch": 7.524669073405535,
      "grad_norm": 1.699373483657837,
      "learning_rate": 0.00019763048234946678,
      "loss": 7.5808,
      "step": 62530
    },
    {
      "epoch": 7.525872442839952,
      "grad_norm": 1.7151769399642944,
      "learning_rate": 0.00019762965903542964,
      "loss": 7.5738,
      "step": 62540
    },
    {
      "epoch": 7.527075812274369,
      "grad_norm": 2.4995622634887695,
      "learning_rate": 0.00019762883558009824,
      "loss": 7.539,
      "step": 62550
    },
    {
      "epoch": 7.528279181708784,
      "grad_norm": 1.6077799797058105,
      "learning_rate": 0.0001976280119834738,
      "loss": 7.5272,
      "step": 62560
    },
    {
      "epoch": 7.529482551143201,
      "grad_norm": 3.734455108642578,
      "learning_rate": 0.00019762718824555754,
      "loss": 7.5363,
      "step": 62570
    },
    {
      "epoch": 7.530685920577618,
      "grad_norm": 2.447986125946045,
      "learning_rate": 0.00019762636436635066,
      "loss": 7.5188,
      "step": 62580
    },
    {
      "epoch": 7.5318892900120336,
      "grad_norm": 5.249312877655029,
      "learning_rate": 0.0001976255403458543,
      "loss": 7.6377,
      "step": 62590
    },
    {
      "epoch": 7.53309265944645,
      "grad_norm": 2.261216640472412,
      "learning_rate": 0.00019762471618406968,
      "loss": 7.5725,
      "step": 62600
    },
    {
      "epoch": 7.534296028880867,
      "grad_norm": 3.281552314758301,
      "learning_rate": 0.00019762389188099797,
      "loss": 7.5659,
      "step": 62610
    },
    {
      "epoch": 7.535499398315283,
      "grad_norm": 2.4530956745147705,
      "learning_rate": 0.00019762306743664043,
      "loss": 7.5436,
      "step": 62620
    },
    {
      "epoch": 7.536702767749699,
      "grad_norm": 2.429037094116211,
      "learning_rate": 0.00019762224285099818,
      "loss": 7.5066,
      "step": 62630
    },
    {
      "epoch": 7.537906137184115,
      "grad_norm": 1.5216857194900513,
      "learning_rate": 0.00019762141812407246,
      "loss": 7.5222,
      "step": 62640
    },
    {
      "epoch": 7.539109506618532,
      "grad_norm": 4.295212268829346,
      "learning_rate": 0.00019762059325586443,
      "loss": 7.5354,
      "step": 62650
    },
    {
      "epoch": 7.5403128760529485,
      "grad_norm": 1.930651307106018,
      "learning_rate": 0.00019761976824637532,
      "loss": 7.5968,
      "step": 62660
    },
    {
      "epoch": 7.541516245487364,
      "grad_norm": 3.3710813522338867,
      "learning_rate": 0.0001976189430956063,
      "loss": 7.5271,
      "step": 62670
    },
    {
      "epoch": 7.542719614921781,
      "grad_norm": 6.041106700897217,
      "learning_rate": 0.00019761811780355854,
      "loss": 7.502,
      "step": 62680
    },
    {
      "epoch": 7.543922984356198,
      "grad_norm": 1.3379738330841064,
      "learning_rate": 0.0001976172923702333,
      "loss": 7.54,
      "step": 62690
    },
    {
      "epoch": 7.5451263537906135,
      "grad_norm": 6.153784275054932,
      "learning_rate": 0.00019761646679563173,
      "loss": 7.542,
      "step": 62700
    },
    {
      "epoch": 7.54632972322503,
      "grad_norm": 3.5414633750915527,
      "learning_rate": 0.000197615641079755,
      "loss": 7.5418,
      "step": 62710
    },
    {
      "epoch": 7.547533092659447,
      "grad_norm": 3.5140304565429688,
      "learning_rate": 0.00019761481522260436,
      "loss": 7.5398,
      "step": 62720
    },
    {
      "epoch": 7.548736462093863,
      "grad_norm": 3.721635341644287,
      "learning_rate": 0.00019761398922418097,
      "loss": 7.5333,
      "step": 62730
    },
    {
      "epoch": 7.549939831528279,
      "grad_norm": 2.585458755493164,
      "learning_rate": 0.00019761316308448604,
      "loss": 7.5586,
      "step": 62740
    },
    {
      "epoch": 7.551143200962695,
      "grad_norm": 2.3863401412963867,
      "learning_rate": 0.00019761233680352081,
      "loss": 7.5191,
      "step": 62750
    },
    {
      "epoch": 7.552346570397112,
      "grad_norm": 1.8514198064804077,
      "learning_rate": 0.00019761151038128638,
      "loss": 7.5751,
      "step": 62760
    },
    {
      "epoch": 7.5535499398315284,
      "grad_norm": 2.957806348800659,
      "learning_rate": 0.000197610683817784,
      "loss": 7.5367,
      "step": 62770
    },
    {
      "epoch": 7.554753309265944,
      "grad_norm": 2.435565233230591,
      "learning_rate": 0.00019760985711301488,
      "loss": 7.6302,
      "step": 62780
    },
    {
      "epoch": 7.555956678700361,
      "grad_norm": 2.6678693294525146,
      "learning_rate": 0.00019760903026698018,
      "loss": 7.5704,
      "step": 62790
    },
    {
      "epoch": 7.557160048134778,
      "grad_norm": 2.53086256980896,
      "learning_rate": 0.00019760820327968112,
      "loss": 7.4994,
      "step": 62800
    },
    {
      "epoch": 7.558363417569193,
      "grad_norm": 4.167178630828857,
      "learning_rate": 0.00019760737615111889,
      "loss": 7.5181,
      "step": 62810
    },
    {
      "epoch": 7.55956678700361,
      "grad_norm": 1.563988447189331,
      "learning_rate": 0.00019760654888129468,
      "loss": 7.6593,
      "step": 62820
    },
    {
      "epoch": 7.560770156438027,
      "grad_norm": 2.448904514312744,
      "learning_rate": 0.0001976057214702097,
      "loss": 7.6433,
      "step": 62830
    },
    {
      "epoch": 7.5619735258724425,
      "grad_norm": 5.7234206199646,
      "learning_rate": 0.00019760489391786515,
      "loss": 7.6568,
      "step": 62840
    },
    {
      "epoch": 7.563176895306859,
      "grad_norm": 1.2108515501022339,
      "learning_rate": 0.00019760406622426223,
      "loss": 7.5729,
      "step": 62850
    },
    {
      "epoch": 7.564380264741276,
      "grad_norm": 4.4809041023254395,
      "learning_rate": 0.0001976032383894021,
      "loss": 7.5406,
      "step": 62860
    },
    {
      "epoch": 7.565583634175692,
      "grad_norm": 4.5882086753845215,
      "learning_rate": 0.000197602410413286,
      "loss": 7.7044,
      "step": 62870
    },
    {
      "epoch": 7.566787003610108,
      "grad_norm": 1.8069469928741455,
      "learning_rate": 0.00019760158229591512,
      "loss": 7.5452,
      "step": 62880
    },
    {
      "epoch": 7.567990373044525,
      "grad_norm": 1.049312949180603,
      "learning_rate": 0.00019760075403729065,
      "loss": 7.5368,
      "step": 62890
    },
    {
      "epoch": 7.569193742478941,
      "grad_norm": 1.602185845375061,
      "learning_rate": 0.0001975999256374138,
      "loss": 7.5801,
      "step": 62900
    },
    {
      "epoch": 7.5703971119133575,
      "grad_norm": 1.6173040866851807,
      "learning_rate": 0.00019759909709628578,
      "loss": 7.6503,
      "step": 62910
    },
    {
      "epoch": 7.571600481347774,
      "grad_norm": 4.716909885406494,
      "learning_rate": 0.00019759826841390772,
      "loss": 7.53,
      "step": 62920
    },
    {
      "epoch": 7.57280385078219,
      "grad_norm": 2.9156434535980225,
      "learning_rate": 0.00019759743959028088,
      "loss": 7.5201,
      "step": 62930
    },
    {
      "epoch": 7.574007220216607,
      "grad_norm": 1.4834566116333008,
      "learning_rate": 0.0001975966106254065,
      "loss": 7.4602,
      "step": 62940
    },
    {
      "epoch": 7.575210589651023,
      "grad_norm": 1.8014588356018066,
      "learning_rate": 0.00019759578151928566,
      "loss": 7.581,
      "step": 62950
    },
    {
      "epoch": 7.576413959085439,
      "grad_norm": 4.3229451179504395,
      "learning_rate": 0.00019759495227191968,
      "loss": 7.6452,
      "step": 62960
    },
    {
      "epoch": 7.577617328519856,
      "grad_norm": 2.332765817642212,
      "learning_rate": 0.00019759412288330967,
      "loss": 7.6667,
      "step": 62970
    },
    {
      "epoch": 7.578820697954272,
      "grad_norm": 2.9040307998657227,
      "learning_rate": 0.0001975932933534569,
      "loss": 7.5922,
      "step": 62980
    },
    {
      "epoch": 7.580024067388688,
      "grad_norm": 3.8108394145965576,
      "learning_rate": 0.0001975924636823625,
      "loss": 7.5505,
      "step": 62990
    },
    {
      "epoch": 7.581227436823105,
      "grad_norm": 2.6067097187042236,
      "learning_rate": 0.00019759163387002775,
      "loss": 7.5408,
      "step": 63000
    },
    {
      "epoch": 7.582430806257521,
      "grad_norm": 2.1119534969329834,
      "learning_rate": 0.0001975908039164538,
      "loss": 7.5641,
      "step": 63010
    },
    {
      "epoch": 7.583634175691937,
      "grad_norm": 2.8722622394561768,
      "learning_rate": 0.00019758997382164186,
      "loss": 7.5431,
      "step": 63020
    },
    {
      "epoch": 7.584837545126354,
      "grad_norm": 2.3188719749450684,
      "learning_rate": 0.00019758914358559312,
      "loss": 7.5942,
      "step": 63030
    },
    {
      "epoch": 7.58604091456077,
      "grad_norm": 8.449291229248047,
      "learning_rate": 0.00019758831320830882,
      "loss": 7.546,
      "step": 63040
    },
    {
      "epoch": 7.587244283995187,
      "grad_norm": 2.0346593856811523,
      "learning_rate": 0.00019758748268979012,
      "loss": 7.5198,
      "step": 63050
    },
    {
      "epoch": 7.588447653429603,
      "grad_norm": 4.518615245819092,
      "learning_rate": 0.00019758665203003823,
      "loss": 7.6062,
      "step": 63060
    },
    {
      "epoch": 7.589651022864019,
      "grad_norm": 2.3619871139526367,
      "learning_rate": 0.0001975858212290544,
      "loss": 7.5481,
      "step": 63070
    },
    {
      "epoch": 7.590854392298436,
      "grad_norm": 1.6649744510650635,
      "learning_rate": 0.00019758499028683978,
      "loss": 7.5576,
      "step": 63080
    },
    {
      "epoch": 7.5920577617328515,
      "grad_norm": 7.858250141143799,
      "learning_rate": 0.00019758415920339558,
      "loss": 7.566,
      "step": 63090
    },
    {
      "epoch": 7.593261131167268,
      "grad_norm": 1.6704298257827759,
      "learning_rate": 0.000197583327978723,
      "loss": 7.5065,
      "step": 63100
    },
    {
      "epoch": 7.594464500601685,
      "grad_norm": 1.2919920682907104,
      "learning_rate": 0.00019758249661282323,
      "loss": 7.5531,
      "step": 63110
    },
    {
      "epoch": 7.595667870036101,
      "grad_norm": 4.33117151260376,
      "learning_rate": 0.00019758166510569755,
      "loss": 7.5211,
      "step": 63120
    },
    {
      "epoch": 7.596871239470517,
      "grad_norm": 1.717166781425476,
      "learning_rate": 0.00019758083345734707,
      "loss": 7.5163,
      "step": 63130
    },
    {
      "epoch": 7.598074608904934,
      "grad_norm": 6.342723369598389,
      "learning_rate": 0.00019758000166777303,
      "loss": 7.5172,
      "step": 63140
    },
    {
      "epoch": 7.59927797833935,
      "grad_norm": 5.0491461753845215,
      "learning_rate": 0.00019757916973697664,
      "loss": 7.4563,
      "step": 63150
    },
    {
      "epoch": 7.6004813477737665,
      "grad_norm": 1.950960636138916,
      "learning_rate": 0.00019757833766495913,
      "loss": 7.6231,
      "step": 63160
    },
    {
      "epoch": 7.601684717208183,
      "grad_norm": 3.9927687644958496,
      "learning_rate": 0.00019757750545172165,
      "loss": 7.5158,
      "step": 63170
    },
    {
      "epoch": 7.602888086642599,
      "grad_norm": 1.4569436311721802,
      "learning_rate": 0.00019757667309726543,
      "loss": 7.4586,
      "step": 63180
    },
    {
      "epoch": 7.604091456077016,
      "grad_norm": 7.387356758117676,
      "learning_rate": 0.00019757584060159166,
      "loss": 7.5484,
      "step": 63190
    },
    {
      "epoch": 7.605294825511432,
      "grad_norm": 2.513106346130371,
      "learning_rate": 0.0001975750079647016,
      "loss": 7.5849,
      "step": 63200
    },
    {
      "epoch": 7.606498194945848,
      "grad_norm": 2.2404375076293945,
      "learning_rate": 0.00019757417518659635,
      "loss": 7.619,
      "step": 63210
    },
    {
      "epoch": 7.607701564380265,
      "grad_norm": 3.4064767360687256,
      "learning_rate": 0.00019757334226727725,
      "loss": 7.5106,
      "step": 63220
    },
    {
      "epoch": 7.6089049338146815,
      "grad_norm": 1.926140308380127,
      "learning_rate": 0.00019757250920674537,
      "loss": 7.5097,
      "step": 63230
    },
    {
      "epoch": 7.610108303249097,
      "grad_norm": 1.4428722858428955,
      "learning_rate": 0.000197571676005002,
      "loss": 7.6128,
      "step": 63240
    },
    {
      "epoch": 7.611311672683514,
      "grad_norm": 3.848318099975586,
      "learning_rate": 0.00019757084266204834,
      "loss": 7.5696,
      "step": 63250
    },
    {
      "epoch": 7.612515042117931,
      "grad_norm": 2.6820499897003174,
      "learning_rate": 0.00019757000917788558,
      "loss": 7.561,
      "step": 63260
    },
    {
      "epoch": 7.613718411552346,
      "grad_norm": 2.538931131362915,
      "learning_rate": 0.0001975691755525149,
      "loss": 7.6254,
      "step": 63270
    },
    {
      "epoch": 7.614921780986763,
      "grad_norm": 1.9670612812042236,
      "learning_rate": 0.00019756834178593757,
      "loss": 7.5691,
      "step": 63280
    },
    {
      "epoch": 7.61612515042118,
      "grad_norm": 1.9742215871810913,
      "learning_rate": 0.00019756750787815473,
      "loss": 7.6541,
      "step": 63290
    },
    {
      "epoch": 7.617328519855596,
      "grad_norm": 2.971890926361084,
      "learning_rate": 0.00019756667382916763,
      "loss": 7.5045,
      "step": 63300
    },
    {
      "epoch": 7.618531889290012,
      "grad_norm": 1.8174625635147095,
      "learning_rate": 0.00019756583963897748,
      "loss": 7.5149,
      "step": 63310
    },
    {
      "epoch": 7.619735258724428,
      "grad_norm": 4.023271083831787,
      "learning_rate": 0.00019756500530758545,
      "loss": 7.573,
      "step": 63320
    },
    {
      "epoch": 7.620938628158845,
      "grad_norm": 1.5273938179016113,
      "learning_rate": 0.00019756417083499278,
      "loss": 7.5163,
      "step": 63330
    },
    {
      "epoch": 7.622141997593261,
      "grad_norm": 2.6752381324768066,
      "learning_rate": 0.00019756333622120067,
      "loss": 7.5582,
      "step": 63340
    },
    {
      "epoch": 7.623345367027677,
      "grad_norm": 1.7386105060577393,
      "learning_rate": 0.0001975625014662103,
      "loss": 7.554,
      "step": 63350
    },
    {
      "epoch": 7.624548736462094,
      "grad_norm": 2.0145838260650635,
      "learning_rate": 0.00019756166657002293,
      "loss": 7.4902,
      "step": 63360
    },
    {
      "epoch": 7.6257521058965105,
      "grad_norm": 2.0998079776763916,
      "learning_rate": 0.00019756083153263973,
      "loss": 7.5331,
      "step": 63370
    },
    {
      "epoch": 7.626955475330926,
      "grad_norm": 2.288391590118408,
      "learning_rate": 0.00019755999635406193,
      "loss": 7.543,
      "step": 63380
    },
    {
      "epoch": 7.628158844765343,
      "grad_norm": 1.6026840209960938,
      "learning_rate": 0.0001975591610342907,
      "loss": 7.517,
      "step": 63390
    },
    {
      "epoch": 7.62936221419976,
      "grad_norm": 2.1817140579223633,
      "learning_rate": 0.00019755832557332733,
      "loss": 7.4646,
      "step": 63400
    },
    {
      "epoch": 7.6305655836341755,
      "grad_norm": 2.484788179397583,
      "learning_rate": 0.00019755748997117293,
      "loss": 7.5459,
      "step": 63410
    },
    {
      "epoch": 7.631768953068592,
      "grad_norm": 2.66873836517334,
      "learning_rate": 0.00019755665422782878,
      "loss": 7.5803,
      "step": 63420
    },
    {
      "epoch": 7.632972322503008,
      "grad_norm": 1.8474595546722412,
      "learning_rate": 0.00019755581834329604,
      "loss": 7.582,
      "step": 63430
    },
    {
      "epoch": 7.634175691937425,
      "grad_norm": 6.507787227630615,
      "learning_rate": 0.00019755498231757595,
      "loss": 7.5455,
      "step": 63440
    },
    {
      "epoch": 7.635379061371841,
      "grad_norm": 2.517765760421753,
      "learning_rate": 0.00019755414615066976,
      "loss": 7.4921,
      "step": 63450
    },
    {
      "epoch": 7.636582430806257,
      "grad_norm": 8.62722110748291,
      "learning_rate": 0.0001975533098425786,
      "loss": 7.6468,
      "step": 63460
    },
    {
      "epoch": 7.637785800240674,
      "grad_norm": 3.2575252056121826,
      "learning_rate": 0.0001975524733933037,
      "loss": 7.5935,
      "step": 63470
    },
    {
      "epoch": 7.6389891696750905,
      "grad_norm": 5.830880165100098,
      "learning_rate": 0.00019755163680284631,
      "loss": 7.6109,
      "step": 63480
    },
    {
      "epoch": 7.640192539109506,
      "grad_norm": 5.177492141723633,
      "learning_rate": 0.00019755080007120764,
      "loss": 7.4877,
      "step": 63490
    },
    {
      "epoch": 7.641395908543923,
      "grad_norm": 1.9223870038986206,
      "learning_rate": 0.00019754996319838882,
      "loss": 7.5029,
      "step": 63500
    },
    {
      "epoch": 7.64259927797834,
      "grad_norm": 2.336559295654297,
      "learning_rate": 0.00019754912618439113,
      "loss": 7.5298,
      "step": 63510
    },
    {
      "epoch": 7.643802647412755,
      "grad_norm": 1.6006437540054321,
      "learning_rate": 0.00019754828902921582,
      "loss": 7.5539,
      "step": 63520
    },
    {
      "epoch": 7.645006016847172,
      "grad_norm": 2.190847396850586,
      "learning_rate": 0.00019754745173286402,
      "loss": 7.4398,
      "step": 63530
    },
    {
      "epoch": 7.646209386281589,
      "grad_norm": 3.2800686359405518,
      "learning_rate": 0.00019754661429533695,
      "loss": 7.5672,
      "step": 63540
    },
    {
      "epoch": 7.6474127557160045,
      "grad_norm": 2.1784653663635254,
      "learning_rate": 0.00019754577671663587,
      "loss": 7.4409,
      "step": 63550
    },
    {
      "epoch": 7.648616125150421,
      "grad_norm": 1.5865652561187744,
      "learning_rate": 0.00019754493899676198,
      "loss": 7.5773,
      "step": 63560
    },
    {
      "epoch": 7.649819494584838,
      "grad_norm": 3.0933616161346436,
      "learning_rate": 0.00019754410113571646,
      "loss": 7.5312,
      "step": 63570
    },
    {
      "epoch": 7.651022864019254,
      "grad_norm": 2.3897340297698975,
      "learning_rate": 0.00019754326313350056,
      "loss": 7.4591,
      "step": 63580
    },
    {
      "epoch": 7.65222623345367,
      "grad_norm": 3.7813637256622314,
      "learning_rate": 0.00019754242499011546,
      "loss": 7.5851,
      "step": 63590
    },
    {
      "epoch": 7.653429602888087,
      "grad_norm": 6.147426605224609,
      "learning_rate": 0.00019754158670556239,
      "loss": 7.4976,
      "step": 63600
    },
    {
      "epoch": 7.654632972322503,
      "grad_norm": 3.5563721656799316,
      "learning_rate": 0.00019754074827984256,
      "loss": 7.4387,
      "step": 63610
    },
    {
      "epoch": 7.6558363417569195,
      "grad_norm": 9.913110733032227,
      "learning_rate": 0.0001975399097129572,
      "loss": 7.4511,
      "step": 63620
    },
    {
      "epoch": 7.657039711191336,
      "grad_norm": 3.6053640842437744,
      "learning_rate": 0.0001975390710049075,
      "loss": 7.5784,
      "step": 63630
    },
    {
      "epoch": 7.658243080625752,
      "grad_norm": 1.8616575002670288,
      "learning_rate": 0.00019753823215569466,
      "loss": 7.636,
      "step": 63640
    },
    {
      "epoch": 7.659446450060169,
      "grad_norm": 2.616544485092163,
      "learning_rate": 0.00019753739316531995,
      "loss": 7.5859,
      "step": 63650
    },
    {
      "epoch": 7.6606498194945845,
      "grad_norm": 2.7370352745056152,
      "learning_rate": 0.00019753655403378454,
      "loss": 7.4884,
      "step": 63660
    },
    {
      "epoch": 7.661853188929001,
      "grad_norm": 6.263876438140869,
      "learning_rate": 0.00019753571476108968,
      "loss": 7.6124,
      "step": 63670
    },
    {
      "epoch": 7.663056558363418,
      "grad_norm": 5.364309787750244,
      "learning_rate": 0.0001975348753472365,
      "loss": 7.545,
      "step": 63680
    },
    {
      "epoch": 7.664259927797834,
      "grad_norm": 3.396055221557617,
      "learning_rate": 0.0001975340357922263,
      "loss": 7.5016,
      "step": 63690
    },
    {
      "epoch": 7.66546329723225,
      "grad_norm": 2.0077784061431885,
      "learning_rate": 0.0001975331960960603,
      "loss": 7.4803,
      "step": 63700
    },
    {
      "epoch": 7.666666666666667,
      "grad_norm": 3.4992411136627197,
      "learning_rate": 0.00019753235625873968,
      "loss": 7.5829,
      "step": 63710
    },
    {
      "epoch": 7.667870036101083,
      "grad_norm": 5.172300338745117,
      "learning_rate": 0.00019753151628026564,
      "loss": 7.4763,
      "step": 63720
    },
    {
      "epoch": 7.669073405535499,
      "grad_norm": 4.38728666305542,
      "learning_rate": 0.00019753067616063943,
      "loss": 7.609,
      "step": 63730
    },
    {
      "epoch": 7.670276774969916,
      "grad_norm": 6.026226997375488,
      "learning_rate": 0.0001975298358998622,
      "loss": 7.5913,
      "step": 63740
    },
    {
      "epoch": 7.671480144404332,
      "grad_norm": 37.50835418701172,
      "learning_rate": 0.0001975289954979353,
      "loss": 7.6434,
      "step": 63750
    },
    {
      "epoch": 7.672683513838749,
      "grad_norm": 4.4342498779296875,
      "learning_rate": 0.0001975281549548598,
      "loss": 7.5253,
      "step": 63760
    },
    {
      "epoch": 7.673886883273164,
      "grad_norm": 2.3639578819274902,
      "learning_rate": 0.00019752731427063702,
      "loss": 7.613,
      "step": 63770
    },
    {
      "epoch": 7.675090252707581,
      "grad_norm": 3.7852845191955566,
      "learning_rate": 0.0001975264734452681,
      "loss": 7.5704,
      "step": 63780
    },
    {
      "epoch": 7.676293622141998,
      "grad_norm": 2.4387924671173096,
      "learning_rate": 0.00019752563247875436,
      "loss": 7.4976,
      "step": 63790
    },
    {
      "epoch": 7.6774969915764135,
      "grad_norm": 6.428036212921143,
      "learning_rate": 0.00019752479137109688,
      "loss": 7.4921,
      "step": 63800
    },
    {
      "epoch": 7.67870036101083,
      "grad_norm": 116.8272476196289,
      "learning_rate": 0.00019752395012229701,
      "loss": 7.6181,
      "step": 63810
    },
    {
      "epoch": 7.679903730445247,
      "grad_norm": 418.4561462402344,
      "learning_rate": 0.00019752310873235587,
      "loss": 7.8285,
      "step": 63820
    },
    {
      "epoch": 7.681107099879663,
      "grad_norm": 5.228960990905762,
      "learning_rate": 0.00019752226720127472,
      "loss": 7.58,
      "step": 63830
    },
    {
      "epoch": 7.682310469314079,
      "grad_norm": 3.4253652095794678,
      "learning_rate": 0.00019752142552905476,
      "loss": 7.5709,
      "step": 63840
    },
    {
      "epoch": 7.683513838748496,
      "grad_norm": 2.6173439025878906,
      "learning_rate": 0.00019752058371569724,
      "loss": 7.5608,
      "step": 63850
    },
    {
      "epoch": 7.684717208182912,
      "grad_norm": 6.814177513122559,
      "learning_rate": 0.00019751974176120335,
      "loss": 7.6076,
      "step": 63860
    },
    {
      "epoch": 7.6859205776173285,
      "grad_norm": 1.6845054626464844,
      "learning_rate": 0.0001975188996655743,
      "loss": 7.5737,
      "step": 63870
    },
    {
      "epoch": 7.687123947051745,
      "grad_norm": 1.6492327451705933,
      "learning_rate": 0.00019751805742881134,
      "loss": 7.5565,
      "step": 63880
    },
    {
      "epoch": 7.688327316486161,
      "grad_norm": 4.752699375152588,
      "learning_rate": 0.0001975172150509157,
      "loss": 7.6199,
      "step": 63890
    },
    {
      "epoch": 7.689530685920578,
      "grad_norm": 1.4198944568634033,
      "learning_rate": 0.00019751637253188854,
      "loss": 7.5747,
      "step": 63900
    },
    {
      "epoch": 7.690734055354994,
      "grad_norm": 6.435874938964844,
      "learning_rate": 0.00019751552987173112,
      "loss": 7.5798,
      "step": 63910
    },
    {
      "epoch": 7.69193742478941,
      "grad_norm": 2.5221967697143555,
      "learning_rate": 0.00019751468707044465,
      "loss": 7.542,
      "step": 63920
    },
    {
      "epoch": 7.693140794223827,
      "grad_norm": 2.549466848373413,
      "learning_rate": 0.00019751384412803038,
      "loss": 7.4978,
      "step": 63930
    },
    {
      "epoch": 7.6943441636582435,
      "grad_norm": 0.9417816400527954,
      "learning_rate": 0.00019751300104448946,
      "loss": 7.6134,
      "step": 63940
    },
    {
      "epoch": 7.695547533092659,
      "grad_norm": 1.1502593755722046,
      "learning_rate": 0.00019751215781982318,
      "loss": 7.5314,
      "step": 63950
    },
    {
      "epoch": 7.696750902527076,
      "grad_norm": 1.5023001432418823,
      "learning_rate": 0.00019751131445403273,
      "loss": 7.5336,
      "step": 63960
    },
    {
      "epoch": 7.697954271961493,
      "grad_norm": 2.0197274684906006,
      "learning_rate": 0.00019751047094711933,
      "loss": 7.5932,
      "step": 63970
    },
    {
      "epoch": 7.699157641395908,
      "grad_norm": 2.939283847808838,
      "learning_rate": 0.0001975096272990842,
      "loss": 7.4884,
      "step": 63980
    },
    {
      "epoch": 7.700361010830325,
      "grad_norm": 4.106808185577393,
      "learning_rate": 0.00019750878350992856,
      "loss": 7.4562,
      "step": 63990
    },
    {
      "epoch": 7.701564380264742,
      "grad_norm": 2.5515804290771484,
      "learning_rate": 0.00019750793957965364,
      "loss": 7.589,
      "step": 64000
    },
    {
      "epoch": 7.702767749699158,
      "grad_norm": 2.34568190574646,
      "learning_rate": 0.00019750709550826065,
      "loss": 7.5463,
      "step": 64010
    },
    {
      "epoch": 7.703971119133574,
      "grad_norm": 7.649059295654297,
      "learning_rate": 0.00019750625129575085,
      "loss": 7.6698,
      "step": 64020
    },
    {
      "epoch": 7.70517448856799,
      "grad_norm": 2.6196110248565674,
      "learning_rate": 0.0001975054069421254,
      "loss": 7.5876,
      "step": 64030
    },
    {
      "epoch": 7.706377858002407,
      "grad_norm": 3.7890024185180664,
      "learning_rate": 0.00019750456244738555,
      "loss": 7.5667,
      "step": 64040
    },
    {
      "epoch": 7.707581227436823,
      "grad_norm": 1.9010287523269653,
      "learning_rate": 0.00019750371781153257,
      "loss": 7.5407,
      "step": 64050
    },
    {
      "epoch": 7.708784596871239,
      "grad_norm": 2.4469101428985596,
      "learning_rate": 0.00019750287303456758,
      "loss": 7.5519,
      "step": 64060
    },
    {
      "epoch": 7.709987966305656,
      "grad_norm": 3.000920295715332,
      "learning_rate": 0.0001975020281164919,
      "loss": 7.5291,
      "step": 64070
    },
    {
      "epoch": 7.7111913357400725,
      "grad_norm": 3.5249977111816406,
      "learning_rate": 0.0001975011830573067,
      "loss": 7.5339,
      "step": 64080
    },
    {
      "epoch": 7.712394705174488,
      "grad_norm": 6.092122554779053,
      "learning_rate": 0.0001975003378570132,
      "loss": 7.5936,
      "step": 64090
    },
    {
      "epoch": 7.713598074608905,
      "grad_norm": 4.037398338317871,
      "learning_rate": 0.00019749949251561267,
      "loss": 7.6532,
      "step": 64100
    },
    {
      "epoch": 7.714801444043322,
      "grad_norm": 1.274417757987976,
      "learning_rate": 0.00019749864703310626,
      "loss": 7.5724,
      "step": 64110
    },
    {
      "epoch": 7.7160048134777375,
      "grad_norm": 2.7037127017974854,
      "learning_rate": 0.00019749780140949528,
      "loss": 7.5893,
      "step": 64120
    },
    {
      "epoch": 7.717208182912154,
      "grad_norm": 5.475039958953857,
      "learning_rate": 0.00019749695564478091,
      "loss": 7.5529,
      "step": 64130
    },
    {
      "epoch": 7.71841155234657,
      "grad_norm": 7.58928108215332,
      "learning_rate": 0.00019749610973896433,
      "loss": 7.6414,
      "step": 64140
    },
    {
      "epoch": 7.719614921780987,
      "grad_norm": 1.3528416156768799,
      "learning_rate": 0.00019749526369204685,
      "loss": 7.6468,
      "step": 64150
    },
    {
      "epoch": 7.720818291215403,
      "grad_norm": 2.443830728530884,
      "learning_rate": 0.00019749441750402963,
      "loss": 7.5163,
      "step": 64160
    },
    {
      "epoch": 7.722021660649819,
      "grad_norm": 5.909456729888916,
      "learning_rate": 0.00019749357117491393,
      "loss": 7.5757,
      "step": 64170
    },
    {
      "epoch": 7.723225030084236,
      "grad_norm": 1.958285927772522,
      "learning_rate": 0.00019749272470470094,
      "loss": 7.4558,
      "step": 64180
    },
    {
      "epoch": 7.7244283995186525,
      "grad_norm": 5.574573516845703,
      "learning_rate": 0.0001974918780933919,
      "loss": 7.4697,
      "step": 64190
    },
    {
      "epoch": 7.725631768953068,
      "grad_norm": 2.4745922088623047,
      "learning_rate": 0.0001974910313409881,
      "loss": 7.6013,
      "step": 64200
    },
    {
      "epoch": 7.726835138387485,
      "grad_norm": 1.5560861825942993,
      "learning_rate": 0.00019749018444749066,
      "loss": 7.6324,
      "step": 64210
    },
    {
      "epoch": 7.728038507821902,
      "grad_norm": 1.7659121751785278,
      "learning_rate": 0.0001974893374129009,
      "loss": 7.5592,
      "step": 64220
    },
    {
      "epoch": 7.729241877256317,
      "grad_norm": 1.9414066076278687,
      "learning_rate": 0.00019748849023721996,
      "loss": 7.5265,
      "step": 64230
    },
    {
      "epoch": 7.730445246690734,
      "grad_norm": 4.28660249710083,
      "learning_rate": 0.00019748764292044907,
      "loss": 7.5062,
      "step": 64240
    },
    {
      "epoch": 7.731648616125151,
      "grad_norm": 2.162717342376709,
      "learning_rate": 0.00019748679546258954,
      "loss": 7.6001,
      "step": 64250
    },
    {
      "epoch": 7.7328519855595665,
      "grad_norm": 1.2348604202270508,
      "learning_rate": 0.00019748594786364257,
      "loss": 7.5055,
      "step": 64260
    },
    {
      "epoch": 7.734055354993983,
      "grad_norm": 4.4819722175598145,
      "learning_rate": 0.0001974851001236093,
      "loss": 7.5513,
      "step": 64270
    },
    {
      "epoch": 7.7352587244284,
      "grad_norm": 2.6607470512390137,
      "learning_rate": 0.00019748425224249108,
      "loss": 7.5627,
      "step": 64280
    },
    {
      "epoch": 7.736462093862816,
      "grad_norm": 1.8192871809005737,
      "learning_rate": 0.00019748340422028906,
      "loss": 7.7036,
      "step": 64290
    },
    {
      "epoch": 7.737665463297232,
      "grad_norm": 1.4009596109390259,
      "learning_rate": 0.00019748255605700447,
      "loss": 7.4962,
      "step": 64300
    },
    {
      "epoch": 7.738868832731649,
      "grad_norm": 1.0280046463012695,
      "learning_rate": 0.0001974817077526386,
      "loss": 7.5562,
      "step": 64310
    },
    {
      "epoch": 7.740072202166065,
      "grad_norm": 1.7691929340362549,
      "learning_rate": 0.00019748085930719258,
      "loss": 7.5742,
      "step": 64320
    },
    {
      "epoch": 7.7412755716004815,
      "grad_norm": 3.197049856185913,
      "learning_rate": 0.00019748001072066772,
      "loss": 7.5702,
      "step": 64330
    },
    {
      "epoch": 7.742478941034898,
      "grad_norm": 1.4524508714675903,
      "learning_rate": 0.0001974791619930652,
      "loss": 7.6947,
      "step": 64340
    },
    {
      "epoch": 7.743682310469314,
      "grad_norm": 1.4267940521240234,
      "learning_rate": 0.00019747831312438626,
      "loss": 7.5172,
      "step": 64350
    },
    {
      "epoch": 7.744885679903731,
      "grad_norm": 2.4507248401641846,
      "learning_rate": 0.00019747746411463214,
      "loss": 7.5614,
      "step": 64360
    },
    {
      "epoch": 7.7460890493381465,
      "grad_norm": 5.619117736816406,
      "learning_rate": 0.0001974766149638041,
      "loss": 7.6002,
      "step": 64370
    },
    {
      "epoch": 7.747292418772563,
      "grad_norm": 1.2406738996505737,
      "learning_rate": 0.00019747576567190328,
      "loss": 7.4875,
      "step": 64380
    },
    {
      "epoch": 7.74849578820698,
      "grad_norm": 1.8460060358047485,
      "learning_rate": 0.000197474916238931,
      "loss": 7.593,
      "step": 64390
    },
    {
      "epoch": 7.749699157641396,
      "grad_norm": 5.018037796020508,
      "learning_rate": 0.00019747406666488843,
      "loss": 7.4923,
      "step": 64400
    },
    {
      "epoch": 7.750902527075812,
      "grad_norm": 1.853564977645874,
      "learning_rate": 0.00019747321694977684,
      "loss": 7.5799,
      "step": 64410
    },
    {
      "epoch": 7.752105896510229,
      "grad_norm": 3.093491554260254,
      "learning_rate": 0.0001974723670935974,
      "loss": 7.5675,
      "step": 64420
    },
    {
      "epoch": 7.753309265944645,
      "grad_norm": 2.326819658279419,
      "learning_rate": 0.0001974715170963514,
      "loss": 7.5225,
      "step": 64430
    },
    {
      "epoch": 7.754512635379061,
      "grad_norm": 4.111430644989014,
      "learning_rate": 0.00019747066695804003,
      "loss": 7.6652,
      "step": 64440
    },
    {
      "epoch": 7.755716004813478,
      "grad_norm": 1.2467533349990845,
      "learning_rate": 0.0001974698166786646,
      "loss": 7.5544,
      "step": 64450
    },
    {
      "epoch": 7.756919374247894,
      "grad_norm": 4.279912948608398,
      "learning_rate": 0.00019746896625822622,
      "loss": 7.5619,
      "step": 64460
    },
    {
      "epoch": 7.758122743682311,
      "grad_norm": 4.139098644256592,
      "learning_rate": 0.00019746811569672622,
      "loss": 7.5576,
      "step": 64470
    },
    {
      "epoch": 7.759326113116726,
      "grad_norm": 1.6513266563415527,
      "learning_rate": 0.00019746726499416577,
      "loss": 7.6236,
      "step": 64480
    },
    {
      "epoch": 7.760529482551143,
      "grad_norm": 2.026365280151367,
      "learning_rate": 0.00019746641415054614,
      "loss": 7.5971,
      "step": 64490
    },
    {
      "epoch": 7.76173285198556,
      "grad_norm": 3.4769864082336426,
      "learning_rate": 0.00019746556316586853,
      "loss": 7.5674,
      "step": 64500
    },
    {
      "epoch": 7.7629362214199755,
      "grad_norm": 1.7914206981658936,
      "learning_rate": 0.0001974647120401342,
      "loss": 7.6313,
      "step": 64510
    },
    {
      "epoch": 7.764139590854392,
      "grad_norm": 7.555795192718506,
      "learning_rate": 0.00019746386077334433,
      "loss": 7.5473,
      "step": 64520
    },
    {
      "epoch": 7.765342960288809,
      "grad_norm": 2.8244681358337402,
      "learning_rate": 0.00019746300936550023,
      "loss": 7.5729,
      "step": 64530
    },
    {
      "epoch": 7.766546329723225,
      "grad_norm": 4.239053726196289,
      "learning_rate": 0.00019746215781660308,
      "loss": 7.5083,
      "step": 64540
    },
    {
      "epoch": 7.767749699157641,
      "grad_norm": 2.4355297088623047,
      "learning_rate": 0.00019746130612665415,
      "loss": 7.583,
      "step": 64550
    },
    {
      "epoch": 7.768953068592058,
      "grad_norm": 1.4362925291061401,
      "learning_rate": 0.00019746045429565464,
      "loss": 7.5604,
      "step": 64560
    },
    {
      "epoch": 7.770156438026474,
      "grad_norm": 7.067419528961182,
      "learning_rate": 0.00019745960232360577,
      "loss": 7.5351,
      "step": 64570
    },
    {
      "epoch": 7.7713598074608905,
      "grad_norm": 1.3868975639343262,
      "learning_rate": 0.00019745875021050878,
      "loss": 7.476,
      "step": 64580
    },
    {
      "epoch": 7.772563176895307,
      "grad_norm": 5.19750452041626,
      "learning_rate": 0.00019745789795636492,
      "loss": 7.5075,
      "step": 64590
    },
    {
      "epoch": 7.773766546329723,
      "grad_norm": 1.595211148262024,
      "learning_rate": 0.00019745704556117547,
      "loss": 7.5509,
      "step": 64600
    },
    {
      "epoch": 7.77496991576414,
      "grad_norm": 1.366173505783081,
      "learning_rate": 0.00019745619302494159,
      "loss": 7.5437,
      "step": 64610
    },
    {
      "epoch": 7.776173285198556,
      "grad_norm": 4.056373596191406,
      "learning_rate": 0.00019745534034766452,
      "loss": 7.6292,
      "step": 64620
    },
    {
      "epoch": 7.777376654632972,
      "grad_norm": 3.093219757080078,
      "learning_rate": 0.0001974544875293455,
      "loss": 7.5332,
      "step": 64630
    },
    {
      "epoch": 7.778580024067389,
      "grad_norm": 2.3698809146881104,
      "learning_rate": 0.0001974536345699858,
      "loss": 7.4866,
      "step": 64640
    },
    {
      "epoch": 7.7797833935018055,
      "grad_norm": 1.3150427341461182,
      "learning_rate": 0.0001974527814695866,
      "loss": 7.52,
      "step": 64650
    },
    {
      "epoch": 7.780986762936221,
      "grad_norm": 2.1142914295196533,
      "learning_rate": 0.00019745192822814923,
      "loss": 7.6069,
      "step": 64660
    },
    {
      "epoch": 7.782190132370638,
      "grad_norm": 3.0181469917297363,
      "learning_rate": 0.00019745107484567481,
      "loss": 7.6176,
      "step": 64670
    },
    {
      "epoch": 7.783393501805055,
      "grad_norm": 1.8113892078399658,
      "learning_rate": 0.00019745022132216464,
      "loss": 7.5021,
      "step": 64680
    },
    {
      "epoch": 7.78459687123947,
      "grad_norm": 0.9960190057754517,
      "learning_rate": 0.0001974493676576199,
      "loss": 7.6028,
      "step": 64690
    },
    {
      "epoch": 7.785800240673887,
      "grad_norm": 1.6795947551727295,
      "learning_rate": 0.00019744851385204191,
      "loss": 7.547,
      "step": 64700
    },
    {
      "epoch": 7.787003610108303,
      "grad_norm": 2.323227882385254,
      "learning_rate": 0.00019744765990543188,
      "loss": 7.5919,
      "step": 64710
    },
    {
      "epoch": 7.78820697954272,
      "grad_norm": 1.5074779987335205,
      "learning_rate": 0.00019744680581779101,
      "loss": 7.5274,
      "step": 64720
    },
    {
      "epoch": 7.789410348977136,
      "grad_norm": 2.540370464324951,
      "learning_rate": 0.00019744595158912052,
      "loss": 7.568,
      "step": 64730
    },
    {
      "epoch": 7.790613718411552,
      "grad_norm": 1.8090728521347046,
      "learning_rate": 0.0001974450972194217,
      "loss": 7.5752,
      "step": 64740
    },
    {
      "epoch": 7.791817087845969,
      "grad_norm": 1.3719005584716797,
      "learning_rate": 0.00019744424270869577,
      "loss": 7.5466,
      "step": 64750
    },
    {
      "epoch": 7.793020457280385,
      "grad_norm": 1.5757919549942017,
      "learning_rate": 0.00019744338805694395,
      "loss": 7.5842,
      "step": 64760
    },
    {
      "epoch": 7.794223826714801,
      "grad_norm": 4.796267986297607,
      "learning_rate": 0.0001974425332641675,
      "loss": 7.535,
      "step": 64770
    },
    {
      "epoch": 7.795427196149218,
      "grad_norm": 3.333724021911621,
      "learning_rate": 0.00019744167833036765,
      "loss": 7.5415,
      "step": 64780
    },
    {
      "epoch": 7.7966305655836345,
      "grad_norm": 7.976353168487549,
      "learning_rate": 0.00019744082325554564,
      "loss": 7.582,
      "step": 64790
    },
    {
      "epoch": 7.79783393501805,
      "grad_norm": 3.6790027618408203,
      "learning_rate": 0.00019743996803970267,
      "loss": 7.6207,
      "step": 64800
    },
    {
      "epoch": 7.799037304452467,
      "grad_norm": 4.515024185180664,
      "learning_rate": 0.00019743911268284005,
      "loss": 7.5974,
      "step": 64810
    },
    {
      "epoch": 7.800240673886883,
      "grad_norm": 2.5830047130584717,
      "learning_rate": 0.00019743825718495893,
      "loss": 7.6111,
      "step": 64820
    },
    {
      "epoch": 7.8014440433212995,
      "grad_norm": 2.3890299797058105,
      "learning_rate": 0.00019743740154606063,
      "loss": 7.5557,
      "step": 64830
    },
    {
      "epoch": 7.802647412755716,
      "grad_norm": 2.4947264194488525,
      "learning_rate": 0.00019743654576614636,
      "loss": 7.5119,
      "step": 64840
    },
    {
      "epoch": 7.803850782190132,
      "grad_norm": 1.304914116859436,
      "learning_rate": 0.00019743568984521734,
      "loss": 7.5988,
      "step": 64850
    },
    {
      "epoch": 7.805054151624549,
      "grad_norm": 1.2074496746063232,
      "learning_rate": 0.0001974348337832748,
      "loss": 7.5569,
      "step": 64860
    },
    {
      "epoch": 7.806257521058965,
      "grad_norm": 1.843406319618225,
      "learning_rate": 0.00019743397758032006,
      "loss": 7.5541,
      "step": 64870
    },
    {
      "epoch": 7.807460890493381,
      "grad_norm": 1.682096004486084,
      "learning_rate": 0.0001974331212363542,
      "loss": 7.5764,
      "step": 64880
    },
    {
      "epoch": 7.808664259927798,
      "grad_norm": 5.8746018409729,
      "learning_rate": 0.00019743226475137866,
      "loss": 7.6384,
      "step": 64890
    },
    {
      "epoch": 7.8098676293622145,
      "grad_norm": 2.056720495223999,
      "learning_rate": 0.0001974314081253945,
      "loss": 7.4991,
      "step": 64900
    },
    {
      "epoch": 7.81107099879663,
      "grad_norm": 2.306234836578369,
      "learning_rate": 0.0001974305513584031,
      "loss": 7.4743,
      "step": 64910
    },
    {
      "epoch": 7.812274368231047,
      "grad_norm": 3.863231658935547,
      "learning_rate": 0.0001974296944504056,
      "loss": 7.4782,
      "step": 64920
    },
    {
      "epoch": 7.813477737665464,
      "grad_norm": 0.9819413423538208,
      "learning_rate": 0.00019742883740140327,
      "loss": 7.5229,
      "step": 64930
    },
    {
      "epoch": 7.814681107099879,
      "grad_norm": 3.1213436126708984,
      "learning_rate": 0.0001974279802113974,
      "loss": 7.5257,
      "step": 64940
    },
    {
      "epoch": 7.815884476534296,
      "grad_norm": 4.330084800720215,
      "learning_rate": 0.00019742712288038912,
      "loss": 7.5378,
      "step": 64950
    },
    {
      "epoch": 7.817087845968713,
      "grad_norm": 1.4419766664505005,
      "learning_rate": 0.00019742626540837982,
      "loss": 7.5202,
      "step": 64960
    },
    {
      "epoch": 7.8182912154031285,
      "grad_norm": 4.496762275695801,
      "learning_rate": 0.0001974254077953706,
      "loss": 7.5922,
      "step": 64970
    },
    {
      "epoch": 7.819494584837545,
      "grad_norm": 1.9259644746780396,
      "learning_rate": 0.0001974245500413628,
      "loss": 7.5476,
      "step": 64980
    },
    {
      "epoch": 7.820697954271962,
      "grad_norm": 5.210505962371826,
      "learning_rate": 0.0001974236921463576,
      "loss": 7.5164,
      "step": 64990
    },
    {
      "epoch": 7.821901323706378,
      "grad_norm": 1.6667044162750244,
      "learning_rate": 0.00019742283411035627,
      "loss": 7.5339,
      "step": 65000
    },
    {
      "epoch": 7.823104693140794,
      "grad_norm": 2.4324257373809814,
      "learning_rate": 0.00019742197593336004,
      "loss": 7.6013,
      "step": 65010
    },
    {
      "epoch": 7.824308062575211,
      "grad_norm": 4.153875827789307,
      "learning_rate": 0.00019742111761537017,
      "loss": 7.6247,
      "step": 65020
    },
    {
      "epoch": 7.825511432009627,
      "grad_norm": 1.1132521629333496,
      "learning_rate": 0.00019742025915638788,
      "loss": 7.4552,
      "step": 65030
    },
    {
      "epoch": 7.8267148014440435,
      "grad_norm": 2.402911901473999,
      "learning_rate": 0.0001974194005564144,
      "loss": 7.5211,
      "step": 65040
    },
    {
      "epoch": 7.827918170878459,
      "grad_norm": 2.978116035461426,
      "learning_rate": 0.00019741854181545105,
      "loss": 7.5419,
      "step": 65050
    },
    {
      "epoch": 7.829121540312876,
      "grad_norm": 3.3875086307525635,
      "learning_rate": 0.000197417682933499,
      "loss": 7.5099,
      "step": 65060
    },
    {
      "epoch": 7.830324909747293,
      "grad_norm": 1.9529696702957153,
      "learning_rate": 0.0001974168239105595,
      "loss": 7.556,
      "step": 65070
    },
    {
      "epoch": 7.8315282791817085,
      "grad_norm": 2.051300287246704,
      "learning_rate": 0.0001974159647466338,
      "loss": 7.5171,
      "step": 65080
    },
    {
      "epoch": 7.832731648616125,
      "grad_norm": 3.8050243854522705,
      "learning_rate": 0.00019741510544172314,
      "loss": 7.5617,
      "step": 65090
    },
    {
      "epoch": 7.833935018050542,
      "grad_norm": 1.7248845100402832,
      "learning_rate": 0.0001974142459958288,
      "loss": 7.4871,
      "step": 65100
    },
    {
      "epoch": 7.835138387484958,
      "grad_norm": 1.7705193758010864,
      "learning_rate": 0.00019741338640895197,
      "loss": 7.5426,
      "step": 65110
    },
    {
      "epoch": 7.836341756919374,
      "grad_norm": 2.031907796859741,
      "learning_rate": 0.00019741252668109395,
      "loss": 7.4904,
      "step": 65120
    },
    {
      "epoch": 7.837545126353791,
      "grad_norm": 2.721560001373291,
      "learning_rate": 0.00019741166681225592,
      "loss": 7.5527,
      "step": 65130
    },
    {
      "epoch": 7.838748495788207,
      "grad_norm": 5.362534999847412,
      "learning_rate": 0.00019741080680243917,
      "loss": 7.4884,
      "step": 65140
    },
    {
      "epoch": 7.839951865222623,
      "grad_norm": 3.95763897895813,
      "learning_rate": 0.00019740994665164494,
      "loss": 7.5728,
      "step": 65150
    },
    {
      "epoch": 7.841155234657039,
      "grad_norm": 3.410754919052124,
      "learning_rate": 0.00019740908635987445,
      "loss": 7.6117,
      "step": 65160
    },
    {
      "epoch": 7.842358604091456,
      "grad_norm": 2.575103998184204,
      "learning_rate": 0.00019740822592712897,
      "loss": 7.5994,
      "step": 65170
    },
    {
      "epoch": 7.843561973525873,
      "grad_norm": 1.800849199295044,
      "learning_rate": 0.00019740736535340975,
      "loss": 7.5598,
      "step": 65180
    },
    {
      "epoch": 7.844765342960288,
      "grad_norm": 1.6818286180496216,
      "learning_rate": 0.000197406504638718,
      "loss": 7.5488,
      "step": 65190
    },
    {
      "epoch": 7.845968712394705,
      "grad_norm": 1.826019525527954,
      "learning_rate": 0.00019740564378305502,
      "loss": 7.5313,
      "step": 65200
    },
    {
      "epoch": 7.847172081829122,
      "grad_norm": 0.9545811414718628,
      "learning_rate": 0.000197404782786422,
      "loss": 7.6071,
      "step": 65210
    },
    {
      "epoch": 7.8483754512635375,
      "grad_norm": 2.7919225692749023,
      "learning_rate": 0.00019740392164882022,
      "loss": 7.5058,
      "step": 65220
    },
    {
      "epoch": 7.849578820697954,
      "grad_norm": 2.8619964122772217,
      "learning_rate": 0.00019740306037025092,
      "loss": 7.4722,
      "step": 65230
    },
    {
      "epoch": 7.850782190132371,
      "grad_norm": 3.166888952255249,
      "learning_rate": 0.0001974021989507153,
      "loss": 7.5576,
      "step": 65240
    },
    {
      "epoch": 7.851985559566787,
      "grad_norm": 3.817974805831909,
      "learning_rate": 0.00019740133739021474,
      "loss": 7.5461,
      "step": 65250
    },
    {
      "epoch": 7.853188929001203,
      "grad_norm": 2.8046810626983643,
      "learning_rate": 0.00019740047568875033,
      "loss": 7.5449,
      "step": 65260
    },
    {
      "epoch": 7.85439229843562,
      "grad_norm": 1.7972290515899658,
      "learning_rate": 0.0001973996138463234,
      "loss": 7.5372,
      "step": 65270
    },
    {
      "epoch": 7.855595667870036,
      "grad_norm": 3.015254497528076,
      "learning_rate": 0.0001973987518629352,
      "loss": 7.5555,
      "step": 65280
    },
    {
      "epoch": 7.8567990373044525,
      "grad_norm": 4.372307300567627,
      "learning_rate": 0.00019739788973858693,
      "loss": 7.5718,
      "step": 65290
    },
    {
      "epoch": 7.858002406738869,
      "grad_norm": 2.7161402702331543,
      "learning_rate": 0.00019739702747327987,
      "loss": 7.5227,
      "step": 65300
    },
    {
      "epoch": 7.859205776173285,
      "grad_norm": 1.8533118963241577,
      "learning_rate": 0.00019739616506701528,
      "loss": 7.4359,
      "step": 65310
    },
    {
      "epoch": 7.860409145607702,
      "grad_norm": 1.7904753684997559,
      "learning_rate": 0.0001973953025197944,
      "loss": 7.5533,
      "step": 65320
    },
    {
      "epoch": 7.861612515042118,
      "grad_norm": 2.156414270401001,
      "learning_rate": 0.00019739443983161846,
      "loss": 7.5456,
      "step": 65330
    },
    {
      "epoch": 7.862815884476534,
      "grad_norm": 3.3818130493164062,
      "learning_rate": 0.0001973935770024887,
      "loss": 7.5957,
      "step": 65340
    },
    {
      "epoch": 7.864019253910951,
      "grad_norm": 4.580965042114258,
      "learning_rate": 0.00019739271403240643,
      "loss": 7.5773,
      "step": 65350
    },
    {
      "epoch": 7.8652226233453675,
      "grad_norm": 2.1996190547943115,
      "learning_rate": 0.0001973918509213728,
      "loss": 7.4879,
      "step": 65360
    },
    {
      "epoch": 7.866425992779783,
      "grad_norm": 5.844728946685791,
      "learning_rate": 0.00019739098766938918,
      "loss": 7.5364,
      "step": 65370
    },
    {
      "epoch": 7.8676293622142,
      "grad_norm": 1.9474704265594482,
      "learning_rate": 0.0001973901242764567,
      "loss": 7.5071,
      "step": 65380
    },
    {
      "epoch": 7.868832731648616,
      "grad_norm": 2.776552438735962,
      "learning_rate": 0.0001973892607425767,
      "loss": 7.4707,
      "step": 65390
    },
    {
      "epoch": 7.870036101083032,
      "grad_norm": 2.329772472381592,
      "learning_rate": 0.0001973883970677504,
      "loss": 7.5491,
      "step": 65400
    },
    {
      "epoch": 7.871239470517449,
      "grad_norm": 1.808611273765564,
      "learning_rate": 0.00019738753325197902,
      "loss": 7.4872,
      "step": 65410
    },
    {
      "epoch": 7.872442839951865,
      "grad_norm": 2.5463521480560303,
      "learning_rate": 0.00019738666929526385,
      "loss": 7.6258,
      "step": 65420
    },
    {
      "epoch": 7.873646209386282,
      "grad_norm": 1.359895944595337,
      "learning_rate": 0.0001973858051976061,
      "loss": 7.5124,
      "step": 65430
    },
    {
      "epoch": 7.874849578820698,
      "grad_norm": 1.0344352722167969,
      "learning_rate": 0.00019738494095900707,
      "loss": 7.6403,
      "step": 65440
    },
    {
      "epoch": 7.876052948255114,
      "grad_norm": 4.382324695587158,
      "learning_rate": 0.000197384076579468,
      "loss": 7.5588,
      "step": 65450
    },
    {
      "epoch": 7.877256317689531,
      "grad_norm": 1.6695517301559448,
      "learning_rate": 0.0001973832120589901,
      "loss": 7.6111,
      "step": 65460
    },
    {
      "epoch": 7.878459687123947,
      "grad_norm": 1.7769498825073242,
      "learning_rate": 0.00019738234739757467,
      "loss": 7.5345,
      "step": 65470
    },
    {
      "epoch": 7.879663056558363,
      "grad_norm": 2.275113821029663,
      "learning_rate": 0.00019738148259522293,
      "loss": 7.5103,
      "step": 65480
    },
    {
      "epoch": 7.88086642599278,
      "grad_norm": 2.661782741546631,
      "learning_rate": 0.00019738061765193615,
      "loss": 7.564,
      "step": 65490
    },
    {
      "epoch": 7.882069795427196,
      "grad_norm": 1.5649412870407104,
      "learning_rate": 0.00019737975256771555,
      "loss": 7.5496,
      "step": 65500
    },
    {
      "epoch": 7.883273164861612,
      "grad_norm": 1.8353400230407715,
      "learning_rate": 0.00019737888734256242,
      "loss": 7.6234,
      "step": 65510
    },
    {
      "epoch": 7.884476534296029,
      "grad_norm": 3.566084623336792,
      "learning_rate": 0.00019737802197647802,
      "loss": 7.5368,
      "step": 65520
    },
    {
      "epoch": 7.885679903730445,
      "grad_norm": 2.392169713973999,
      "learning_rate": 0.00019737715646946354,
      "loss": 7.4623,
      "step": 65530
    },
    {
      "epoch": 7.8868832731648615,
      "grad_norm": 1.8639453649520874,
      "learning_rate": 0.0001973762908215203,
      "loss": 7.5273,
      "step": 65540
    },
    {
      "epoch": 7.888086642599278,
      "grad_norm": 3.9149012565612793,
      "learning_rate": 0.00019737542503264952,
      "loss": 7.5995,
      "step": 65550
    },
    {
      "epoch": 7.889290012033694,
      "grad_norm": 1.9269750118255615,
      "learning_rate": 0.00019737455910285245,
      "loss": 7.5675,
      "step": 65560
    },
    {
      "epoch": 7.890493381468111,
      "grad_norm": 7.425387859344482,
      "learning_rate": 0.0001973736930321304,
      "loss": 7.5878,
      "step": 65570
    },
    {
      "epoch": 7.891696750902527,
      "grad_norm": 1.50200355052948,
      "learning_rate": 0.0001973728268204845,
      "loss": 7.5821,
      "step": 65580
    },
    {
      "epoch": 7.892900120336943,
      "grad_norm": 1.8554785251617432,
      "learning_rate": 0.00019737196046791612,
      "loss": 7.5473,
      "step": 65590
    },
    {
      "epoch": 7.89410348977136,
      "grad_norm": 1.7223453521728516,
      "learning_rate": 0.00019737109397442647,
      "loss": 7.4737,
      "step": 65600
    },
    {
      "epoch": 7.8953068592057765,
      "grad_norm": 2.594144105911255,
      "learning_rate": 0.0001973702273400168,
      "loss": 7.5732,
      "step": 65610
    },
    {
      "epoch": 7.896510228640192,
      "grad_norm": 3.629218101501465,
      "learning_rate": 0.00019736936056468838,
      "loss": 7.5807,
      "step": 65620
    },
    {
      "epoch": 7.897713598074609,
      "grad_norm": 1.520920753479004,
      "learning_rate": 0.00019736849364844245,
      "loss": 7.5864,
      "step": 65630
    },
    {
      "epoch": 7.898916967509026,
      "grad_norm": 3.9262914657592773,
      "learning_rate": 0.00019736762659128027,
      "loss": 7.4949,
      "step": 65640
    },
    {
      "epoch": 7.900120336943441,
      "grad_norm": 2.1921727657318115,
      "learning_rate": 0.0001973667593932031,
      "loss": 7.5909,
      "step": 65650
    },
    {
      "epoch": 7.901323706377858,
      "grad_norm": 2.235055446624756,
      "learning_rate": 0.00019736589205421217,
      "loss": 7.5696,
      "step": 65660
    },
    {
      "epoch": 7.902527075812275,
      "grad_norm": 4.1180315017700195,
      "learning_rate": 0.00019736502457430878,
      "loss": 7.5985,
      "step": 65670
    },
    {
      "epoch": 7.9037304452466906,
      "grad_norm": 3.2192816734313965,
      "learning_rate": 0.00019736415695349418,
      "loss": 7.4839,
      "step": 65680
    },
    {
      "epoch": 7.904933814681107,
      "grad_norm": 2.169773578643799,
      "learning_rate": 0.00019736328919176956,
      "loss": 7.5479,
      "step": 65690
    },
    {
      "epoch": 7.906137184115524,
      "grad_norm": 5.835244655609131,
      "learning_rate": 0.00019736242128913625,
      "loss": 7.5532,
      "step": 65700
    },
    {
      "epoch": 7.90734055354994,
      "grad_norm": 3.51247239112854,
      "learning_rate": 0.00019736155324559548,
      "loss": 7.5576,
      "step": 65710
    },
    {
      "epoch": 7.908543922984356,
      "grad_norm": 2.271472215652466,
      "learning_rate": 0.00019736068506114848,
      "loss": 7.5475,
      "step": 65720
    },
    {
      "epoch": 7.909747292418773,
      "grad_norm": 5.35723876953125,
      "learning_rate": 0.00019735981673579654,
      "loss": 7.57,
      "step": 65730
    },
    {
      "epoch": 7.910950661853189,
      "grad_norm": 4.704807758331299,
      "learning_rate": 0.00019735894826954092,
      "loss": 7.6706,
      "step": 65740
    },
    {
      "epoch": 7.9121540312876055,
      "grad_norm": 1.3137952089309692,
      "learning_rate": 0.00019735807966238285,
      "loss": 7.5887,
      "step": 65750
    },
    {
      "epoch": 7.913357400722021,
      "grad_norm": 2.429102897644043,
      "learning_rate": 0.00019735721091432363,
      "loss": 7.5469,
      "step": 65760
    },
    {
      "epoch": 7.914560770156438,
      "grad_norm": 2.188586950302124,
      "learning_rate": 0.00019735634202536445,
      "loss": 7.5864,
      "step": 65770
    },
    {
      "epoch": 7.915764139590855,
      "grad_norm": 2.7870771884918213,
      "learning_rate": 0.00019735547299550664,
      "loss": 7.6143,
      "step": 65780
    },
    {
      "epoch": 7.9169675090252705,
      "grad_norm": 2.501793384552002,
      "learning_rate": 0.00019735460382475138,
      "loss": 7.5295,
      "step": 65790
    },
    {
      "epoch": 7.918170878459687,
      "grad_norm": 2.051754951477051,
      "learning_rate": 0.00019735373451310002,
      "loss": 7.6133,
      "step": 65800
    },
    {
      "epoch": 7.919374247894104,
      "grad_norm": 0.946588397026062,
      "learning_rate": 0.00019735286506055374,
      "loss": 7.5772,
      "step": 65810
    },
    {
      "epoch": 7.92057761732852,
      "grad_norm": 4.108164310455322,
      "learning_rate": 0.00019735199546711387,
      "loss": 7.5376,
      "step": 65820
    },
    {
      "epoch": 7.921780986762936,
      "grad_norm": 1.4005546569824219,
      "learning_rate": 0.0001973511257327816,
      "loss": 7.5352,
      "step": 65830
    },
    {
      "epoch": 7.922984356197353,
      "grad_norm": 2.1767354011535645,
      "learning_rate": 0.00019735025585755823,
      "loss": 7.606,
      "step": 65840
    },
    {
      "epoch": 7.924187725631769,
      "grad_norm": 2.0439133644104004,
      "learning_rate": 0.000197349385841445,
      "loss": 7.4631,
      "step": 65850
    },
    {
      "epoch": 7.925391095066185,
      "grad_norm": 0.9920651316642761,
      "learning_rate": 0.0001973485156844431,
      "loss": 7.5778,
      "step": 65860
    },
    {
      "epoch": 7.926594464500601,
      "grad_norm": 1.7963697910308838,
      "learning_rate": 0.00019734764538655394,
      "loss": 7.47,
      "step": 65870
    },
    {
      "epoch": 7.927797833935018,
      "grad_norm": 1.4925529956817627,
      "learning_rate": 0.00019734677494777868,
      "loss": 7.5808,
      "step": 65880
    },
    {
      "epoch": 7.929001203369435,
      "grad_norm": 1.299932599067688,
      "learning_rate": 0.00019734590436811864,
      "loss": 7.5486,
      "step": 65890
    },
    {
      "epoch": 7.93020457280385,
      "grad_norm": 2.4741785526275635,
      "learning_rate": 0.00019734503364757502,
      "loss": 7.5412,
      "step": 65900
    },
    {
      "epoch": 7.931407942238267,
      "grad_norm": 2.414104700088501,
      "learning_rate": 0.00019734416278614908,
      "loss": 7.6205,
      "step": 65910
    },
    {
      "epoch": 7.932611311672684,
      "grad_norm": 4.809906005859375,
      "learning_rate": 0.00019734329178384212,
      "loss": 7.6088,
      "step": 65920
    },
    {
      "epoch": 7.9338146811070995,
      "grad_norm": 2.123011350631714,
      "learning_rate": 0.00019734242064065537,
      "loss": 7.4841,
      "step": 65930
    },
    {
      "epoch": 7.935018050541516,
      "grad_norm": 1.3958371877670288,
      "learning_rate": 0.00019734154935659013,
      "loss": 7.6147,
      "step": 65940
    },
    {
      "epoch": 7.936221419975933,
      "grad_norm": 3.714866876602173,
      "learning_rate": 0.00019734067793164762,
      "loss": 7.5309,
      "step": 65950
    },
    {
      "epoch": 7.937424789410349,
      "grad_norm": 2.742020845413208,
      "learning_rate": 0.0001973398063658291,
      "loss": 7.4426,
      "step": 65960
    },
    {
      "epoch": 7.938628158844765,
      "grad_norm": 7.133040428161621,
      "learning_rate": 0.0001973389346591359,
      "loss": 7.4584,
      "step": 65970
    },
    {
      "epoch": 7.939831528279182,
      "grad_norm": 2.097890853881836,
      "learning_rate": 0.00019733806281156918,
      "loss": 7.5057,
      "step": 65980
    },
    {
      "epoch": 7.941034897713598,
      "grad_norm": 45.25676345825195,
      "learning_rate": 0.00019733719082313026,
      "loss": 7.3778,
      "step": 65990
    },
    {
      "epoch": 7.9422382671480145,
      "grad_norm": 1.7738686800003052,
      "learning_rate": 0.0001973363186938204,
      "loss": 7.5804,
      "step": 66000
    },
    {
      "epoch": 7.943441636582431,
      "grad_norm": 6.599555015563965,
      "learning_rate": 0.00019733544642364085,
      "loss": 7.551,
      "step": 66010
    },
    {
      "epoch": 7.944645006016847,
      "grad_norm": 19.756610870361328,
      "learning_rate": 0.0001973345740125929,
      "loss": 7.572,
      "step": 66020
    },
    {
      "epoch": 7.945848375451264,
      "grad_norm": 5.177730083465576,
      "learning_rate": 0.00019733370146067777,
      "loss": 7.5903,
      "step": 66030
    },
    {
      "epoch": 7.94705174488568,
      "grad_norm": 2.7433736324310303,
      "learning_rate": 0.00019733282876789674,
      "loss": 7.5059,
      "step": 66040
    },
    {
      "epoch": 7.948255114320096,
      "grad_norm": 11.886504173278809,
      "learning_rate": 0.00019733195593425108,
      "loss": 7.4819,
      "step": 66050
    },
    {
      "epoch": 7.949458483754513,
      "grad_norm": 2.1322600841522217,
      "learning_rate": 0.00019733108295974207,
      "loss": 7.6008,
      "step": 66060
    },
    {
      "epoch": 7.9506618531889295,
      "grad_norm": 18.446819305419922,
      "learning_rate": 0.00019733020984437092,
      "loss": 7.6633,
      "step": 66070
    },
    {
      "epoch": 7.951865222623345,
      "grad_norm": 2.9190242290496826,
      "learning_rate": 0.00019732933658813895,
      "loss": 7.526,
      "step": 66080
    },
    {
      "epoch": 7.953068592057762,
      "grad_norm": 6.180509090423584,
      "learning_rate": 0.0001973284631910474,
      "loss": 7.5624,
      "step": 66090
    },
    {
      "epoch": 7.954271961492178,
      "grad_norm": 3.771601676940918,
      "learning_rate": 0.00019732758965309752,
      "loss": 7.5549,
      "step": 66100
    },
    {
      "epoch": 7.955475330926594,
      "grad_norm": 5.095956325531006,
      "learning_rate": 0.0001973267159742906,
      "loss": 7.6696,
      "step": 66110
    },
    {
      "epoch": 7.956678700361011,
      "grad_norm": 3.6988613605499268,
      "learning_rate": 0.0001973258421546279,
      "loss": 7.5666,
      "step": 66120
    },
    {
      "epoch": 7.957882069795427,
      "grad_norm": 3.2755191326141357,
      "learning_rate": 0.00019732496819411063,
      "loss": 7.5828,
      "step": 66130
    },
    {
      "epoch": 7.959085439229844,
      "grad_norm": 4.143715858459473,
      "learning_rate": 0.00019732409409274016,
      "loss": 7.5146,
      "step": 66140
    },
    {
      "epoch": 7.96028880866426,
      "grad_norm": 1.505613923072815,
      "learning_rate": 0.00019732321985051768,
      "loss": 7.6425,
      "step": 66150
    },
    {
      "epoch": 7.961492178098676,
      "grad_norm": 2.0005664825439453,
      "learning_rate": 0.00019732234546744444,
      "loss": 7.5653,
      "step": 66160
    },
    {
      "epoch": 7.962695547533093,
      "grad_norm": 7.554701805114746,
      "learning_rate": 0.00019732147094352179,
      "loss": 7.4561,
      "step": 66170
    },
    {
      "epoch": 7.963898916967509,
      "grad_norm": 2.0083789825439453,
      "learning_rate": 0.0001973205962787509,
      "loss": 7.6,
      "step": 66180
    },
    {
      "epoch": 7.965102286401925,
      "grad_norm": 3.126695156097412,
      "learning_rate": 0.00019731972147313308,
      "loss": 7.581,
      "step": 66190
    },
    {
      "epoch": 7.966305655836342,
      "grad_norm": 2.4581592082977295,
      "learning_rate": 0.00019731884652666963,
      "loss": 7.6485,
      "step": 66200
    },
    {
      "epoch": 7.967509025270758,
      "grad_norm": 3.34667706489563,
      "learning_rate": 0.00019731797143936178,
      "loss": 7.4706,
      "step": 66210
    },
    {
      "epoch": 7.968712394705174,
      "grad_norm": 9.819497108459473,
      "learning_rate": 0.00019731709621121076,
      "loss": 7.4327,
      "step": 66220
    },
    {
      "epoch": 7.969915764139591,
      "grad_norm": 256.25848388671875,
      "learning_rate": 0.0001973162208422179,
      "loss": 7.5515,
      "step": 66230
    },
    {
      "epoch": 7.971119133574007,
      "grad_norm": 4.266899585723877,
      "learning_rate": 0.00019731534533238442,
      "loss": 7.4488,
      "step": 66240
    },
    {
      "epoch": 7.9723225030084235,
      "grad_norm": 3.634145975112915,
      "learning_rate": 0.00019731446968171166,
      "loss": 7.6172,
      "step": 66250
    },
    {
      "epoch": 7.97352587244284,
      "grad_norm": 1.9419814348220825,
      "learning_rate": 0.00019731359389020078,
      "loss": 7.5243,
      "step": 66260
    },
    {
      "epoch": 7.974729241877256,
      "grad_norm": 2.2661983966827393,
      "learning_rate": 0.00019731271795785316,
      "loss": 7.5668,
      "step": 66270
    },
    {
      "epoch": 7.975932611311673,
      "grad_norm": 5.704434394836426,
      "learning_rate": 0.00019731184188466995,
      "loss": 7.5363,
      "step": 66280
    },
    {
      "epoch": 7.977135980746089,
      "grad_norm": 3.7606465816497803,
      "learning_rate": 0.00019731096567065254,
      "loss": 7.5769,
      "step": 66290
    },
    {
      "epoch": 7.978339350180505,
      "grad_norm": 3.810349464416504,
      "learning_rate": 0.0001973100893158021,
      "loss": 7.5451,
      "step": 66300
    },
    {
      "epoch": 7.979542719614922,
      "grad_norm": 3.087070941925049,
      "learning_rate": 0.00019730921282011993,
      "loss": 7.6326,
      "step": 66310
    },
    {
      "epoch": 7.9807460890493385,
      "grad_norm": 2.7472946643829346,
      "learning_rate": 0.0001973083361836073,
      "loss": 7.6323,
      "step": 66320
    },
    {
      "epoch": 7.981949458483754,
      "grad_norm": 4.732410907745361,
      "learning_rate": 0.00019730745940626553,
      "loss": 7.573,
      "step": 66330
    },
    {
      "epoch": 7.983152827918171,
      "grad_norm": 3.2656383514404297,
      "learning_rate": 0.0001973065824880958,
      "loss": 7.5678,
      "step": 66340
    },
    {
      "epoch": 7.984356197352588,
      "grad_norm": 2.0426857471466064,
      "learning_rate": 0.00019730570542909943,
      "loss": 7.5306,
      "step": 66350
    },
    {
      "epoch": 7.985559566787003,
      "grad_norm": 5.330846786499023,
      "learning_rate": 0.00019730482822927768,
      "loss": 7.6089,
      "step": 66360
    },
    {
      "epoch": 7.98676293622142,
      "grad_norm": 2.1498169898986816,
      "learning_rate": 0.0001973039508886318,
      "loss": 7.5299,
      "step": 66370
    },
    {
      "epoch": 7.987966305655837,
      "grad_norm": 2.945660352706909,
      "learning_rate": 0.00019730307340716312,
      "loss": 7.5255,
      "step": 66380
    },
    {
      "epoch": 7.9891696750902526,
      "grad_norm": 2.038069725036621,
      "learning_rate": 0.00019730219578487285,
      "loss": 7.6004,
      "step": 66390
    },
    {
      "epoch": 7.990373044524669,
      "grad_norm": 2.870570659637451,
      "learning_rate": 0.00019730131802176229,
      "loss": 7.5769,
      "step": 66400
    },
    {
      "epoch": 7.991576413959086,
      "grad_norm": 4.0290985107421875,
      "learning_rate": 0.0001973004401178327,
      "loss": 7.5337,
      "step": 66410
    },
    {
      "epoch": 7.992779783393502,
      "grad_norm": 5.626413822174072,
      "learning_rate": 0.0001972995620730853,
      "loss": 7.591,
      "step": 66420
    },
    {
      "epoch": 7.993983152827918,
      "grad_norm": 2.0646297931671143,
      "learning_rate": 0.00019729868388752148,
      "loss": 7.5542,
      "step": 66430
    },
    {
      "epoch": 7.995186522262334,
      "grad_norm": 2.8770172595977783,
      "learning_rate": 0.0001972978055611424,
      "loss": 7.441,
      "step": 66440
    },
    {
      "epoch": 7.996389891696751,
      "grad_norm": 2.1362249851226807,
      "learning_rate": 0.0001972969270939494,
      "loss": 7.5164,
      "step": 66450
    },
    {
      "epoch": 7.9975932611311675,
      "grad_norm": 1.335581660270691,
      "learning_rate": 0.0001972960484859437,
      "loss": 7.6045,
      "step": 66460
    },
    {
      "epoch": 7.998796630565583,
      "grad_norm": 3.3891594409942627,
      "learning_rate": 0.00019729516973712662,
      "loss": 7.5193,
      "step": 66470
    },
    {
      "epoch": 8.0,
      "grad_norm": 3.9859588146209717,
      "learning_rate": 0.0001972942908474994,
      "loss": 7.4777,
      "step": 66480
    },
    {
      "epoch": 8.0,
      "eval_loss": 7.555968284606934,
      "eval_runtime": 119.8679,
      "eval_samples_per_second": 61.626,
      "eval_steps_per_second": 7.708,
      "step": 66480
    },
    {
      "epoch": 8.001203369434416,
      "grad_norm": 1.7889467477798462,
      "learning_rate": 0.0001972934118170633,
      "loss": 7.4666,
      "step": 66490
    },
    {
      "epoch": 8.002406738868833,
      "grad_norm": 2.1631128787994385,
      "learning_rate": 0.00019729253264581963,
      "loss": 7.6277,
      "step": 66500
    },
    {
      "epoch": 8.00361010830325,
      "grad_norm": 2.403017282485962,
      "learning_rate": 0.00019729165333376964,
      "loss": 7.5817,
      "step": 66510
    },
    {
      "epoch": 8.004813477737665,
      "grad_norm": 4.499189376831055,
      "learning_rate": 0.00019729077388091462,
      "loss": 7.4501,
      "step": 66520
    },
    {
      "epoch": 8.006016847172083,
      "grad_norm": 2.48382306098938,
      "learning_rate": 0.0001972898942872558,
      "loss": 7.5931,
      "step": 66530
    },
    {
      "epoch": 8.007220216606498,
      "grad_norm": 1.6500462293624878,
      "learning_rate": 0.00019728901455279453,
      "loss": 7.587,
      "step": 66540
    },
    {
      "epoch": 8.008423586040914,
      "grad_norm": 3.5017309188842773,
      "learning_rate": 0.000197288134677532,
      "loss": 7.5509,
      "step": 66550
    },
    {
      "epoch": 8.009626955475332,
      "grad_norm": 2.3547470569610596,
      "learning_rate": 0.00019728725466146952,
      "loss": 7.4834,
      "step": 66560
    },
    {
      "epoch": 8.010830324909747,
      "grad_norm": 1.6275417804718018,
      "learning_rate": 0.0001972863745046084,
      "loss": 7.4582,
      "step": 66570
    },
    {
      "epoch": 8.012033694344163,
      "grad_norm": 1.9797002077102661,
      "learning_rate": 0.00019728549420694984,
      "loss": 7.5016,
      "step": 66580
    },
    {
      "epoch": 8.01323706377858,
      "grad_norm": 3.043384075164795,
      "learning_rate": 0.00019728461376849517,
      "loss": 7.5484,
      "step": 66590
    },
    {
      "epoch": 8.014440433212997,
      "grad_norm": 2.2602577209472656,
      "learning_rate": 0.0001972837331892456,
      "loss": 7.592,
      "step": 66600
    },
    {
      "epoch": 8.015643802647412,
      "grad_norm": 4.684236526489258,
      "learning_rate": 0.00019728285246920247,
      "loss": 7.5489,
      "step": 66610
    },
    {
      "epoch": 8.01684717208183,
      "grad_norm": 1.8694789409637451,
      "learning_rate": 0.00019728197160836708,
      "loss": 7.6464,
      "step": 66620
    },
    {
      "epoch": 8.018050541516246,
      "grad_norm": 3.5975682735443115,
      "learning_rate": 0.0001972810906067406,
      "loss": 7.6482,
      "step": 66630
    },
    {
      "epoch": 8.019253910950662,
      "grad_norm": 1.554791808128357,
      "learning_rate": 0.00019728020946432442,
      "loss": 7.5689,
      "step": 66640
    },
    {
      "epoch": 8.020457280385079,
      "grad_norm": 3.4765594005584717,
      "learning_rate": 0.00019727932818111975,
      "loss": 7.5757,
      "step": 66650
    },
    {
      "epoch": 8.021660649819495,
      "grad_norm": 1.4756439924240112,
      "learning_rate": 0.00019727844675712782,
      "loss": 7.4566,
      "step": 66660
    },
    {
      "epoch": 8.02286401925391,
      "grad_norm": 2.2486424446105957,
      "learning_rate": 0.00019727756519235002,
      "loss": 7.6053,
      "step": 66670
    },
    {
      "epoch": 8.024067388688326,
      "grad_norm": 3.7347216606140137,
      "learning_rate": 0.0001972766834867875,
      "loss": 7.6275,
      "step": 66680
    },
    {
      "epoch": 8.025270758122744,
      "grad_norm": 1.8567224740982056,
      "learning_rate": 0.00019727580164044165,
      "loss": 7.4503,
      "step": 66690
    },
    {
      "epoch": 8.02647412755716,
      "grad_norm": 1.1227103471755981,
      "learning_rate": 0.0001972749196533137,
      "loss": 7.5952,
      "step": 66700
    },
    {
      "epoch": 8.027677496991576,
      "grad_norm": 3.2187764644622803,
      "learning_rate": 0.00019727403752540493,
      "loss": 7.5426,
      "step": 66710
    },
    {
      "epoch": 8.028880866425993,
      "grad_norm": 4.35701322555542,
      "learning_rate": 0.0001972731552567166,
      "loss": 7.5884,
      "step": 66720
    },
    {
      "epoch": 8.030084235860409,
      "grad_norm": 3.640974283218384,
      "learning_rate": 0.00019727227284725,
      "loss": 7.5374,
      "step": 66730
    },
    {
      "epoch": 8.031287605294825,
      "grad_norm": 1.4204896688461304,
      "learning_rate": 0.00019727139029700639,
      "loss": 7.6199,
      "step": 66740
    },
    {
      "epoch": 8.032490974729242,
      "grad_norm": 1.9910739660263062,
      "learning_rate": 0.0001972705076059871,
      "loss": 7.3852,
      "step": 66750
    },
    {
      "epoch": 8.033694344163658,
      "grad_norm": 1.7718479633331299,
      "learning_rate": 0.00019726962477419332,
      "loss": 7.5771,
      "step": 66760
    },
    {
      "epoch": 8.034897713598074,
      "grad_norm": 1.0193895101547241,
      "learning_rate": 0.0001972687418016264,
      "loss": 7.5511,
      "step": 66770
    },
    {
      "epoch": 8.036101083032491,
      "grad_norm": 3.0317351818084717,
      "learning_rate": 0.0001972678586882876,
      "loss": 7.5749,
      "step": 66780
    },
    {
      "epoch": 8.037304452466907,
      "grad_norm": 2.6694209575653076,
      "learning_rate": 0.00019726697543417818,
      "loss": 7.5776,
      "step": 66790
    },
    {
      "epoch": 8.038507821901323,
      "grad_norm": 1.310793161392212,
      "learning_rate": 0.00019726609203929947,
      "loss": 7.6521,
      "step": 66800
    },
    {
      "epoch": 8.03971119133574,
      "grad_norm": 2.222972869873047,
      "learning_rate": 0.00019726520850365267,
      "loss": 7.5005,
      "step": 66810
    },
    {
      "epoch": 8.040914560770156,
      "grad_norm": 1.9121289253234863,
      "learning_rate": 0.0001972643248272391,
      "loss": 7.5079,
      "step": 66820
    },
    {
      "epoch": 8.042117930204572,
      "grad_norm": 4.8868584632873535,
      "learning_rate": 0.00019726344101006005,
      "loss": 7.5309,
      "step": 66830
    },
    {
      "epoch": 8.04332129963899,
      "grad_norm": 1.1957162618637085,
      "learning_rate": 0.0001972625570521168,
      "loss": 7.5178,
      "step": 66840
    },
    {
      "epoch": 8.044524669073406,
      "grad_norm": 2.1886768341064453,
      "learning_rate": 0.00019726167295341061,
      "loss": 7.5092,
      "step": 66850
    },
    {
      "epoch": 8.045728038507821,
      "grad_norm": 2.4477555751800537,
      "learning_rate": 0.00019726078871394275,
      "loss": 7.4921,
      "step": 66860
    },
    {
      "epoch": 8.046931407942239,
      "grad_norm": 1.644256591796875,
      "learning_rate": 0.00019725990433371452,
      "loss": 7.499,
      "step": 66870
    },
    {
      "epoch": 8.048134777376655,
      "grad_norm": 2.111665964126587,
      "learning_rate": 0.00019725901981272723,
      "loss": 7.5616,
      "step": 66880
    },
    {
      "epoch": 8.04933814681107,
      "grad_norm": 1.5597729682922363,
      "learning_rate": 0.00019725813515098208,
      "loss": 7.5732,
      "step": 66890
    },
    {
      "epoch": 8.050541516245488,
      "grad_norm": 5.364047050476074,
      "learning_rate": 0.00019725725034848042,
      "loss": 7.5655,
      "step": 66900
    },
    {
      "epoch": 8.051744885679904,
      "grad_norm": 2.172454833984375,
      "learning_rate": 0.00019725636540522353,
      "loss": 7.5581,
      "step": 66910
    },
    {
      "epoch": 8.05294825511432,
      "grad_norm": 1.8547550439834595,
      "learning_rate": 0.00019725548032121262,
      "loss": 7.5491,
      "step": 66920
    },
    {
      "epoch": 8.054151624548737,
      "grad_norm": 1.9544776678085327,
      "learning_rate": 0.00019725459509644904,
      "loss": 7.5166,
      "step": 66930
    },
    {
      "epoch": 8.055354993983153,
      "grad_norm": 3.6576926708221436,
      "learning_rate": 0.00019725370973093403,
      "loss": 7.5342,
      "step": 66940
    },
    {
      "epoch": 8.056558363417569,
      "grad_norm": 4.2791428565979,
      "learning_rate": 0.0001972528242246689,
      "loss": 7.5077,
      "step": 66950
    },
    {
      "epoch": 8.057761732851986,
      "grad_norm": 3.432894229888916,
      "learning_rate": 0.00019725193857765495,
      "loss": 7.609,
      "step": 66960
    },
    {
      "epoch": 8.058965102286402,
      "grad_norm": 3.0376970767974854,
      "learning_rate": 0.00019725105278989343,
      "loss": 7.5796,
      "step": 66970
    },
    {
      "epoch": 8.060168471720818,
      "grad_norm": 1.7038400173187256,
      "learning_rate": 0.0001972501668613856,
      "loss": 7.5954,
      "step": 66980
    },
    {
      "epoch": 8.061371841155236,
      "grad_norm": 2.2276458740234375,
      "learning_rate": 0.00019724928079213277,
      "loss": 7.4984,
      "step": 66990
    },
    {
      "epoch": 8.062575210589651,
      "grad_norm": 1.1704379320144653,
      "learning_rate": 0.00019724839458213624,
      "loss": 7.5318,
      "step": 67000
    },
    {
      "epoch": 8.063778580024067,
      "grad_norm": 3.316469430923462,
      "learning_rate": 0.00019724750823139727,
      "loss": 7.5661,
      "step": 67010
    },
    {
      "epoch": 8.064981949458483,
      "grad_norm": 2.5439579486846924,
      "learning_rate": 0.00019724662173991716,
      "loss": 7.5363,
      "step": 67020
    },
    {
      "epoch": 8.0661853188929,
      "grad_norm": 3.5956685543060303,
      "learning_rate": 0.00019724573510769713,
      "loss": 7.5097,
      "step": 67030
    },
    {
      "epoch": 8.067388688327316,
      "grad_norm": 3.3855502605438232,
      "learning_rate": 0.00019724484833473854,
      "loss": 7.5059,
      "step": 67040
    },
    {
      "epoch": 8.068592057761732,
      "grad_norm": 2.211360454559326,
      "learning_rate": 0.00019724396142104265,
      "loss": 7.5294,
      "step": 67050
    },
    {
      "epoch": 8.06979542719615,
      "grad_norm": 2.3276491165161133,
      "learning_rate": 0.0001972430743666107,
      "loss": 7.6116,
      "step": 67060
    },
    {
      "epoch": 8.070998796630565,
      "grad_norm": 1.7261228561401367,
      "learning_rate": 0.0001972421871714441,
      "loss": 7.5458,
      "step": 67070
    },
    {
      "epoch": 8.072202166064981,
      "grad_norm": 4.647141933441162,
      "learning_rate": 0.00019724129983554397,
      "loss": 7.6452,
      "step": 67080
    },
    {
      "epoch": 8.073405535499399,
      "grad_norm": 3.790757179260254,
      "learning_rate": 0.0001972404123589117,
      "loss": 7.5392,
      "step": 67090
    },
    {
      "epoch": 8.074608904933815,
      "grad_norm": 3.0562901496887207,
      "learning_rate": 0.00019723952474154854,
      "loss": 7.5011,
      "step": 67100
    },
    {
      "epoch": 8.07581227436823,
      "grad_norm": 3.311819314956665,
      "learning_rate": 0.0001972386369834558,
      "loss": 7.6471,
      "step": 67110
    },
    {
      "epoch": 8.077015643802648,
      "grad_norm": 2.8634488582611084,
      "learning_rate": 0.00019723774908463472,
      "loss": 7.5742,
      "step": 67120
    },
    {
      "epoch": 8.078219013237064,
      "grad_norm": 5.867244243621826,
      "learning_rate": 0.0001972368610450866,
      "loss": 7.6673,
      "step": 67130
    },
    {
      "epoch": 8.07942238267148,
      "grad_norm": 1.945067048072815,
      "learning_rate": 0.00019723597286481278,
      "loss": 7.5856,
      "step": 67140
    },
    {
      "epoch": 8.080625752105897,
      "grad_norm": 3.4768807888031006,
      "learning_rate": 0.00019723508454381445,
      "loss": 7.4726,
      "step": 67150
    },
    {
      "epoch": 8.081829121540313,
      "grad_norm": 4.385050296783447,
      "learning_rate": 0.00019723419608209296,
      "loss": 7.4804,
      "step": 67160
    },
    {
      "epoch": 8.083032490974729,
      "grad_norm": 2.0109615325927734,
      "learning_rate": 0.0001972333074796496,
      "loss": 7.6148,
      "step": 67170
    },
    {
      "epoch": 8.084235860409146,
      "grad_norm": 4.329806804656982,
      "learning_rate": 0.0001972324187364856,
      "loss": 7.5095,
      "step": 67180
    },
    {
      "epoch": 8.085439229843562,
      "grad_norm": 2.5663046836853027,
      "learning_rate": 0.00019723152985260233,
      "loss": 7.5293,
      "step": 67190
    },
    {
      "epoch": 8.086642599277978,
      "grad_norm": 3.8938119411468506,
      "learning_rate": 0.000197230640828001,
      "loss": 7.6064,
      "step": 67200
    },
    {
      "epoch": 8.087845968712395,
      "grad_norm": 3.559451103210449,
      "learning_rate": 0.00019722975166268296,
      "loss": 7.5777,
      "step": 67210
    },
    {
      "epoch": 8.089049338146811,
      "grad_norm": 4.946438789367676,
      "learning_rate": 0.00019722886235664944,
      "loss": 7.5046,
      "step": 67220
    },
    {
      "epoch": 8.090252707581227,
      "grad_norm": 4.496367454528809,
      "learning_rate": 0.00019722797290990177,
      "loss": 7.5286,
      "step": 67230
    },
    {
      "epoch": 8.091456077015645,
      "grad_norm": 2.403841257095337,
      "learning_rate": 0.00019722708332244116,
      "loss": 7.5267,
      "step": 67240
    },
    {
      "epoch": 8.09265944645006,
      "grad_norm": 3.0910139083862305,
      "learning_rate": 0.000197226193594269,
      "loss": 7.5893,
      "step": 67250
    },
    {
      "epoch": 8.093862815884476,
      "grad_norm": 2.407782793045044,
      "learning_rate": 0.00019722530372538657,
      "loss": 7.5826,
      "step": 67260
    },
    {
      "epoch": 8.095066185318894,
      "grad_norm": 4.66002082824707,
      "learning_rate": 0.00019722441371579507,
      "loss": 7.5305,
      "step": 67270
    },
    {
      "epoch": 8.09626955475331,
      "grad_norm": 8.19322395324707,
      "learning_rate": 0.00019722352356549584,
      "loss": 7.6439,
      "step": 67280
    },
    {
      "epoch": 8.097472924187725,
      "grad_norm": 2.7666609287261963,
      "learning_rate": 0.0001972226332744902,
      "loss": 7.5423,
      "step": 67290
    },
    {
      "epoch": 8.098676293622143,
      "grad_norm": 2.804218053817749,
      "learning_rate": 0.0001972217428427794,
      "loss": 7.569,
      "step": 67300
    },
    {
      "epoch": 8.099879663056559,
      "grad_norm": 3.691422462463379,
      "learning_rate": 0.0001972208522703647,
      "loss": 7.6215,
      "step": 67310
    },
    {
      "epoch": 8.101083032490974,
      "grad_norm": 2.050039768218994,
      "learning_rate": 0.00019721996155724745,
      "loss": 7.522,
      "step": 67320
    },
    {
      "epoch": 8.102286401925392,
      "grad_norm": 3.7799856662750244,
      "learning_rate": 0.0001972190707034289,
      "loss": 7.5668,
      "step": 67330
    },
    {
      "epoch": 8.103489771359808,
      "grad_norm": 2.6276156902313232,
      "learning_rate": 0.00019721817970891036,
      "loss": 7.5166,
      "step": 67340
    },
    {
      "epoch": 8.104693140794224,
      "grad_norm": 1.7708622217178345,
      "learning_rate": 0.0001972172885736931,
      "loss": 7.5256,
      "step": 67350
    },
    {
      "epoch": 8.10589651022864,
      "grad_norm": 3.777482271194458,
      "learning_rate": 0.00019721639729777842,
      "loss": 7.568,
      "step": 67360
    },
    {
      "epoch": 8.107099879663057,
      "grad_norm": 3.338593006134033,
      "learning_rate": 0.00019721550588116765,
      "loss": 7.5303,
      "step": 67370
    },
    {
      "epoch": 8.108303249097473,
      "grad_norm": 2.4056456089019775,
      "learning_rate": 0.00019721461432386202,
      "loss": 7.5602,
      "step": 67380
    },
    {
      "epoch": 8.109506618531888,
      "grad_norm": 2.8567841053009033,
      "learning_rate": 0.00019721372262586283,
      "loss": 7.5319,
      "step": 67390
    },
    {
      "epoch": 8.110709987966306,
      "grad_norm": 3.788039445877075,
      "learning_rate": 0.00019721283078717142,
      "loss": 7.5456,
      "step": 67400
    },
    {
      "epoch": 8.111913357400722,
      "grad_norm": 3.1496784687042236,
      "learning_rate": 0.00019721193880778897,
      "loss": 7.5476,
      "step": 67410
    },
    {
      "epoch": 8.113116726835138,
      "grad_norm": 5.800292491912842,
      "learning_rate": 0.0001972110466877169,
      "loss": 7.4844,
      "step": 67420
    },
    {
      "epoch": 8.114320096269555,
      "grad_norm": 5.42637825012207,
      "learning_rate": 0.00019721015442695643,
      "loss": 7.5015,
      "step": 67430
    },
    {
      "epoch": 8.115523465703971,
      "grad_norm": 5.717906951904297,
      "learning_rate": 0.00019720926202550888,
      "loss": 7.6371,
      "step": 67440
    },
    {
      "epoch": 8.116726835138387,
      "grad_norm": 6.923117637634277,
      "learning_rate": 0.00019720836948337552,
      "loss": 7.5295,
      "step": 67450
    },
    {
      "epoch": 8.117930204572804,
      "grad_norm": 4.9793548583984375,
      "learning_rate": 0.0001972074768005576,
      "loss": 7.5399,
      "step": 67460
    },
    {
      "epoch": 8.11913357400722,
      "grad_norm": 2.6926350593566895,
      "learning_rate": 0.00019720658397705653,
      "loss": 7.4808,
      "step": 67470
    },
    {
      "epoch": 8.120336943441636,
      "grad_norm": 4.779403209686279,
      "learning_rate": 0.0001972056910128735,
      "loss": 7.5939,
      "step": 67480
    },
    {
      "epoch": 8.121540312876053,
      "grad_norm": 5.075592517852783,
      "learning_rate": 0.00019720479790800985,
      "loss": 7.5442,
      "step": 67490
    },
    {
      "epoch": 8.12274368231047,
      "grad_norm": 3.9302806854248047,
      "learning_rate": 0.00019720390466246686,
      "loss": 7.5373,
      "step": 67500
    },
    {
      "epoch": 8.123947051744885,
      "grad_norm": 4.288082122802734,
      "learning_rate": 0.0001972030112762458,
      "loss": 7.4558,
      "step": 67510
    },
    {
      "epoch": 8.125150421179303,
      "grad_norm": 3.6908655166625977,
      "learning_rate": 0.000197202117749348,
      "loss": 7.5385,
      "step": 67520
    },
    {
      "epoch": 8.126353790613718,
      "grad_norm": 5.126953125,
      "learning_rate": 0.00019720122408177474,
      "loss": 7.5608,
      "step": 67530
    },
    {
      "epoch": 8.127557160048134,
      "grad_norm": 74.11157989501953,
      "learning_rate": 0.0001972003302735273,
      "loss": 7.5073,
      "step": 67540
    },
    {
      "epoch": 8.128760529482552,
      "grad_norm": 3.187445640563965,
      "learning_rate": 0.000197199436324607,
      "loss": 7.6169,
      "step": 67550
    },
    {
      "epoch": 8.129963898916968,
      "grad_norm": 10.13817310333252,
      "learning_rate": 0.0001971985422350151,
      "loss": 7.5623,
      "step": 67560
    },
    {
      "epoch": 8.131167268351383,
      "grad_norm": 3.3027138710021973,
      "learning_rate": 0.00019719764800475291,
      "loss": 7.4782,
      "step": 67570
    },
    {
      "epoch": 8.132370637785801,
      "grad_norm": 7.6401448249816895,
      "learning_rate": 0.00019719675363382176,
      "loss": 7.5925,
      "step": 67580
    },
    {
      "epoch": 8.133574007220217,
      "grad_norm": 6.250296115875244,
      "learning_rate": 0.00019719585912222288,
      "loss": 7.537,
      "step": 67590
    },
    {
      "epoch": 8.134777376654633,
      "grad_norm": 4.511796474456787,
      "learning_rate": 0.0001971949644699576,
      "loss": 7.5018,
      "step": 67600
    },
    {
      "epoch": 8.13598074608905,
      "grad_norm": 5.038487911224365,
      "learning_rate": 0.0001971940696770272,
      "loss": 7.5311,
      "step": 67610
    },
    {
      "epoch": 8.137184115523466,
      "grad_norm": 2.808227062225342,
      "learning_rate": 0.00019719317474343301,
      "loss": 7.6221,
      "step": 67620
    },
    {
      "epoch": 8.138387484957882,
      "grad_norm": 6.342968940734863,
      "learning_rate": 0.00019719227966917625,
      "loss": 7.5024,
      "step": 67630
    },
    {
      "epoch": 8.1395908543923,
      "grad_norm": 4.256033420562744,
      "learning_rate": 0.00019719138445425832,
      "loss": 7.5954,
      "step": 67640
    },
    {
      "epoch": 8.140794223826715,
      "grad_norm": 2.9034934043884277,
      "learning_rate": 0.00019719048909868045,
      "loss": 7.6146,
      "step": 67650
    },
    {
      "epoch": 8.14199759326113,
      "grad_norm": 4.3751702308654785,
      "learning_rate": 0.00019718959360244393,
      "loss": 7.4616,
      "step": 67660
    },
    {
      "epoch": 8.143200962695548,
      "grad_norm": 4.708079814910889,
      "learning_rate": 0.00019718869796555007,
      "loss": 7.5555,
      "step": 67670
    },
    {
      "epoch": 8.144404332129964,
      "grad_norm": 10.061758995056152,
      "learning_rate": 0.0001971878021880002,
      "loss": 7.5387,
      "step": 67680
    },
    {
      "epoch": 8.14560770156438,
      "grad_norm": 3.6461355686187744,
      "learning_rate": 0.00019718690626979556,
      "loss": 7.5192,
      "step": 67690
    },
    {
      "epoch": 8.146811070998796,
      "grad_norm": 2.372333526611328,
      "learning_rate": 0.00019718601021093748,
      "loss": 7.5778,
      "step": 67700
    },
    {
      "epoch": 8.148014440433213,
      "grad_norm": 3.2085626125335693,
      "learning_rate": 0.00019718511401142723,
      "loss": 7.6006,
      "step": 67710
    },
    {
      "epoch": 8.14921780986763,
      "grad_norm": 3.4295706748962402,
      "learning_rate": 0.00019718421767126614,
      "loss": 7.5677,
      "step": 67720
    },
    {
      "epoch": 8.150421179302045,
      "grad_norm": 4.437617778778076,
      "learning_rate": 0.0001971833211904555,
      "loss": 7.5051,
      "step": 67730
    },
    {
      "epoch": 8.151624548736462,
      "grad_norm": 3.899390935897827,
      "learning_rate": 0.00019718242456899662,
      "loss": 7.4893,
      "step": 67740
    },
    {
      "epoch": 8.152827918170878,
      "grad_norm": 4.297066688537598,
      "learning_rate": 0.00019718152780689074,
      "loss": 7.5436,
      "step": 67750
    },
    {
      "epoch": 8.154031287605294,
      "grad_norm": 4.00190544128418,
      "learning_rate": 0.0001971806309041392,
      "loss": 7.5472,
      "step": 67760
    },
    {
      "epoch": 8.155234657039712,
      "grad_norm": 3.7725093364715576,
      "learning_rate": 0.00019717973386074332,
      "loss": 7.4599,
      "step": 67770
    },
    {
      "epoch": 8.156438026474127,
      "grad_norm": 4.101502895355225,
      "learning_rate": 0.00019717883667670435,
      "loss": 7.4917,
      "step": 67780
    },
    {
      "epoch": 8.157641395908543,
      "grad_norm": 4.189589023590088,
      "learning_rate": 0.00019717793935202361,
      "loss": 7.5229,
      "step": 67790
    },
    {
      "epoch": 8.15884476534296,
      "grad_norm": 4.835085868835449,
      "learning_rate": 0.0001971770418867024,
      "loss": 7.5586,
      "step": 67800
    },
    {
      "epoch": 8.160048134777377,
      "grad_norm": 3.7349352836608887,
      "learning_rate": 0.00019717614428074203,
      "loss": 7.5528,
      "step": 67810
    },
    {
      "epoch": 8.161251504211792,
      "grad_norm": 12.529157638549805,
      "learning_rate": 0.00019717524653414378,
      "loss": 7.5559,
      "step": 67820
    },
    {
      "epoch": 8.16245487364621,
      "grad_norm": 6.1256279945373535,
      "learning_rate": 0.00019717434864690897,
      "loss": 7.5865,
      "step": 67830
    },
    {
      "epoch": 8.163658243080626,
      "grad_norm": 2.5910558700561523,
      "learning_rate": 0.00019717345061903886,
      "loss": 7.5505,
      "step": 67840
    },
    {
      "epoch": 8.164861612515042,
      "grad_norm": 9.59126091003418,
      "learning_rate": 0.0001971725524505348,
      "loss": 7.6519,
      "step": 67850
    },
    {
      "epoch": 8.166064981949459,
      "grad_norm": 2.7174177169799805,
      "learning_rate": 0.00019717165414139803,
      "loss": 7.6099,
      "step": 67860
    },
    {
      "epoch": 8.167268351383875,
      "grad_norm": 2.9353551864624023,
      "learning_rate": 0.00019717075569162992,
      "loss": 7.5707,
      "step": 67870
    },
    {
      "epoch": 8.16847172081829,
      "grad_norm": 2.3061249256134033,
      "learning_rate": 0.00019716985710123173,
      "loss": 7.4742,
      "step": 67880
    },
    {
      "epoch": 8.169675090252708,
      "grad_norm": 2.0400211811065674,
      "learning_rate": 0.00019716895837020476,
      "loss": 7.4908,
      "step": 67890
    },
    {
      "epoch": 8.170878459687124,
      "grad_norm": 5.309356689453125,
      "learning_rate": 0.00019716805949855033,
      "loss": 7.5348,
      "step": 67900
    },
    {
      "epoch": 8.17208182912154,
      "grad_norm": 2.3623602390289307,
      "learning_rate": 0.00019716716048626972,
      "loss": 7.5545,
      "step": 67910
    },
    {
      "epoch": 8.173285198555957,
      "grad_norm": 2.786099910736084,
      "learning_rate": 0.0001971662613333642,
      "loss": 7.5822,
      "step": 67920
    },
    {
      "epoch": 8.174488567990373,
      "grad_norm": 2.34267258644104,
      "learning_rate": 0.0001971653620398352,
      "loss": 7.5849,
      "step": 67930
    },
    {
      "epoch": 8.175691937424789,
      "grad_norm": 1.97996187210083,
      "learning_rate": 0.00019716446260568385,
      "loss": 7.6152,
      "step": 67940
    },
    {
      "epoch": 8.176895306859207,
      "grad_norm": 2.22361421585083,
      "learning_rate": 0.00019716356303091156,
      "loss": 7.65,
      "step": 67950
    },
    {
      "epoch": 8.178098676293622,
      "grad_norm": 3.01961088180542,
      "learning_rate": 0.0001971626633155196,
      "loss": 7.617,
      "step": 67960
    },
    {
      "epoch": 8.179302045728038,
      "grad_norm": 2.3614389896392822,
      "learning_rate": 0.00019716176345950927,
      "loss": 7.5211,
      "step": 67970
    },
    {
      "epoch": 8.180505415162456,
      "grad_norm": 2.727503538131714,
      "learning_rate": 0.0001971608634628819,
      "loss": 7.5328,
      "step": 67980
    },
    {
      "epoch": 8.181708784596871,
      "grad_norm": 3.683574914932251,
      "learning_rate": 0.00019715996332563877,
      "loss": 7.6351,
      "step": 67990
    },
    {
      "epoch": 8.182912154031287,
      "grad_norm": 0.9345389604568481,
      "learning_rate": 0.0001971590630477812,
      "loss": 7.6196,
      "step": 68000
    },
    {
      "epoch": 8.184115523465705,
      "grad_norm": 1.5343817472457886,
      "learning_rate": 0.00019715816262931045,
      "loss": 7.5788,
      "step": 68010
    },
    {
      "epoch": 8.18531889290012,
      "grad_norm": 1.9073002338409424,
      "learning_rate": 0.00019715726207022787,
      "loss": 7.5732,
      "step": 68020
    },
    {
      "epoch": 8.186522262334536,
      "grad_norm": 3.421137809753418,
      "learning_rate": 0.00019715636137053471,
      "loss": 7.5483,
      "step": 68030
    },
    {
      "epoch": 8.187725631768952,
      "grad_norm": 2.6363823413848877,
      "learning_rate": 0.00019715546053023232,
      "loss": 7.5479,
      "step": 68040
    },
    {
      "epoch": 8.18892900120337,
      "grad_norm": 2.0040910243988037,
      "learning_rate": 0.00019715455954932203,
      "loss": 7.5816,
      "step": 68050
    },
    {
      "epoch": 8.190132370637786,
      "grad_norm": 1.6898119449615479,
      "learning_rate": 0.00019715365842780508,
      "loss": 7.5438,
      "step": 68060
    },
    {
      "epoch": 8.191335740072201,
      "grad_norm": 3.1103663444519043,
      "learning_rate": 0.0001971527571656828,
      "loss": 7.52,
      "step": 68070
    },
    {
      "epoch": 8.192539109506619,
      "grad_norm": 3.5032927989959717,
      "learning_rate": 0.0001971518557629565,
      "loss": 7.5365,
      "step": 68080
    },
    {
      "epoch": 8.193742478941035,
      "grad_norm": 2.230195999145508,
      "learning_rate": 0.00019715095421962746,
      "loss": 7.4769,
      "step": 68090
    },
    {
      "epoch": 8.19494584837545,
      "grad_norm": 5.8746161460876465,
      "learning_rate": 0.00019715005253569703,
      "loss": 7.6013,
      "step": 68100
    },
    {
      "epoch": 8.196149217809868,
      "grad_norm": 1.3629354238510132,
      "learning_rate": 0.00019714915071116645,
      "loss": 7.5615,
      "step": 68110
    },
    {
      "epoch": 8.197352587244284,
      "grad_norm": 7.240510940551758,
      "learning_rate": 0.0001971482487460371,
      "loss": 7.5066,
      "step": 68120
    },
    {
      "epoch": 8.1985559566787,
      "grad_norm": 1.6762794256210327,
      "learning_rate": 0.00019714734664031026,
      "loss": 7.5589,
      "step": 68130
    },
    {
      "epoch": 8.199759326113117,
      "grad_norm": 1.8678590059280396,
      "learning_rate": 0.00019714644439398717,
      "loss": 7.4935,
      "step": 68140
    },
    {
      "epoch": 8.200962695547533,
      "grad_norm": 3.108381509780884,
      "learning_rate": 0.00019714554200706921,
      "loss": 7.5725,
      "step": 68150
    },
    {
      "epoch": 8.202166064981949,
      "grad_norm": 3.1506729125976562,
      "learning_rate": 0.0001971446394795577,
      "loss": 7.5115,
      "step": 68160
    },
    {
      "epoch": 8.203369434416366,
      "grad_norm": 3.2303717136383057,
      "learning_rate": 0.00019714373681145388,
      "loss": 7.5672,
      "step": 68170
    },
    {
      "epoch": 8.204572803850782,
      "grad_norm": 3.26997971534729,
      "learning_rate": 0.0001971428340027591,
      "loss": 7.4824,
      "step": 68180
    },
    {
      "epoch": 8.205776173285198,
      "grad_norm": 1.6657923460006714,
      "learning_rate": 0.00019714193105347468,
      "loss": 7.5582,
      "step": 68190
    },
    {
      "epoch": 8.206979542719615,
      "grad_norm": 1.4390580654144287,
      "learning_rate": 0.00019714102796360185,
      "loss": 7.5975,
      "step": 68200
    },
    {
      "epoch": 8.208182912154031,
      "grad_norm": 2.1655938625335693,
      "learning_rate": 0.000197140124733142,
      "loss": 7.5292,
      "step": 68210
    },
    {
      "epoch": 8.209386281588447,
      "grad_norm": 1.6513818502426147,
      "learning_rate": 0.00019713922136209642,
      "loss": 7.5303,
      "step": 68220
    },
    {
      "epoch": 8.210589651022865,
      "grad_norm": 1.6579651832580566,
      "learning_rate": 0.0001971383178504664,
      "loss": 7.5422,
      "step": 68230
    },
    {
      "epoch": 8.21179302045728,
      "grad_norm": 1.0994912385940552,
      "learning_rate": 0.0001971374141982532,
      "loss": 7.5492,
      "step": 68240
    },
    {
      "epoch": 8.212996389891696,
      "grad_norm": 4.455509662628174,
      "learning_rate": 0.00019713651040545823,
      "loss": 7.5579,
      "step": 68250
    },
    {
      "epoch": 8.214199759326114,
      "grad_norm": 2.238157272338867,
      "learning_rate": 0.00019713560647208273,
      "loss": 7.5954,
      "step": 68260
    },
    {
      "epoch": 8.21540312876053,
      "grad_norm": 1.090720534324646,
      "learning_rate": 0.00019713470239812804,
      "loss": 7.4525,
      "step": 68270
    },
    {
      "epoch": 8.216606498194945,
      "grad_norm": 3.328411340713501,
      "learning_rate": 0.00019713379818359544,
      "loss": 7.5188,
      "step": 68280
    },
    {
      "epoch": 8.217809867629363,
      "grad_norm": 1.9475572109222412,
      "learning_rate": 0.00019713289382848625,
      "loss": 7.6525,
      "step": 68290
    },
    {
      "epoch": 8.219013237063779,
      "grad_norm": 1.8784102201461792,
      "learning_rate": 0.0001971319893328018,
      "loss": 7.5081,
      "step": 68300
    },
    {
      "epoch": 8.220216606498195,
      "grad_norm": 2.0379552841186523,
      "learning_rate": 0.00019713108469654336,
      "loss": 7.5691,
      "step": 68310
    },
    {
      "epoch": 8.221419975932612,
      "grad_norm": 4.128599166870117,
      "learning_rate": 0.00019713017991971227,
      "loss": 7.5606,
      "step": 68320
    },
    {
      "epoch": 8.222623345367028,
      "grad_norm": 3.831167697906494,
      "learning_rate": 0.0001971292750023098,
      "loss": 7.5611,
      "step": 68330
    },
    {
      "epoch": 8.223826714801444,
      "grad_norm": 2.0155131816864014,
      "learning_rate": 0.00019712836994433732,
      "loss": 7.6011,
      "step": 68340
    },
    {
      "epoch": 8.225030084235861,
      "grad_norm": 0.9491598606109619,
      "learning_rate": 0.00019712746474579614,
      "loss": 7.6215,
      "step": 68350
    },
    {
      "epoch": 8.226233453670277,
      "grad_norm": 2.5010132789611816,
      "learning_rate": 0.00019712655940668747,
      "loss": 7.5869,
      "step": 68360
    },
    {
      "epoch": 8.227436823104693,
      "grad_norm": 4.692080974578857,
      "learning_rate": 0.00019712565392701273,
      "loss": 7.5292,
      "step": 68370
    },
    {
      "epoch": 8.22864019253911,
      "grad_norm": 7.14740514755249,
      "learning_rate": 0.00019712474830677317,
      "loss": 7.6161,
      "step": 68380
    },
    {
      "epoch": 8.229843561973526,
      "grad_norm": 2.1906774044036865,
      "learning_rate": 0.00019712384254597014,
      "loss": 7.5704,
      "step": 68390
    },
    {
      "epoch": 8.231046931407942,
      "grad_norm": 2.0275540351867676,
      "learning_rate": 0.00019712293664460494,
      "loss": 7.4773,
      "step": 68400
    },
    {
      "epoch": 8.232250300842358,
      "grad_norm": 1.672620177268982,
      "learning_rate": 0.00019712203060267883,
      "loss": 7.4938,
      "step": 68410
    },
    {
      "epoch": 8.233453670276775,
      "grad_norm": 1.1433289051055908,
      "learning_rate": 0.0001971211244201932,
      "loss": 7.6208,
      "step": 68420
    },
    {
      "epoch": 8.234657039711191,
      "grad_norm": 3.5545032024383545,
      "learning_rate": 0.0001971202180971493,
      "loss": 7.6019,
      "step": 68430
    },
    {
      "epoch": 8.235860409145607,
      "grad_norm": 3.189627170562744,
      "learning_rate": 0.0001971193116335485,
      "loss": 7.5796,
      "step": 68440
    },
    {
      "epoch": 8.237063778580024,
      "grad_norm": 1.6809436082839966,
      "learning_rate": 0.00019711840502939203,
      "loss": 7.5374,
      "step": 68450
    },
    {
      "epoch": 8.23826714801444,
      "grad_norm": 3.4670281410217285,
      "learning_rate": 0.00019711749828468127,
      "loss": 7.6031,
      "step": 68460
    },
    {
      "epoch": 8.239470517448856,
      "grad_norm": 2.8628058433532715,
      "learning_rate": 0.0001971165913994175,
      "loss": 7.5635,
      "step": 68470
    },
    {
      "epoch": 8.240673886883274,
      "grad_norm": 4.718408584594727,
      "learning_rate": 0.00019711568437360207,
      "loss": 7.5076,
      "step": 68480
    },
    {
      "epoch": 8.24187725631769,
      "grad_norm": 1.6630380153656006,
      "learning_rate": 0.00019711477720723628,
      "loss": 7.5079,
      "step": 68490
    },
    {
      "epoch": 8.243080625752105,
      "grad_norm": 5.847504138946533,
      "learning_rate": 0.00019711386990032138,
      "loss": 7.5828,
      "step": 68500
    },
    {
      "epoch": 8.244283995186523,
      "grad_norm": 2.3594069480895996,
      "learning_rate": 0.0001971129624528588,
      "loss": 7.5139,
      "step": 68510
    },
    {
      "epoch": 8.245487364620939,
      "grad_norm": 1.472467064857483,
      "learning_rate": 0.00019711205486484974,
      "loss": 7.4547,
      "step": 68520
    },
    {
      "epoch": 8.246690734055354,
      "grad_norm": 1.9103935956954956,
      "learning_rate": 0.00019711114713629556,
      "loss": 7.663,
      "step": 68530
    },
    {
      "epoch": 8.247894103489772,
      "grad_norm": 1.3938982486724854,
      "learning_rate": 0.0001971102392671976,
      "loss": 7.523,
      "step": 68540
    },
    {
      "epoch": 8.249097472924188,
      "grad_norm": 1.1023763418197632,
      "learning_rate": 0.0001971093312575571,
      "loss": 7.5299,
      "step": 68550
    },
    {
      "epoch": 8.250300842358604,
      "grad_norm": 2.892200231552124,
      "learning_rate": 0.00019710842310737548,
      "loss": 7.5397,
      "step": 68560
    },
    {
      "epoch": 8.251504211793021,
      "grad_norm": 1.197816252708435,
      "learning_rate": 0.00019710751481665398,
      "loss": 7.5108,
      "step": 68570
    },
    {
      "epoch": 8.252707581227437,
      "grad_norm": 3.8547701835632324,
      "learning_rate": 0.00019710660638539392,
      "loss": 7.5239,
      "step": 68580
    },
    {
      "epoch": 8.253910950661853,
      "grad_norm": 2.4872729778289795,
      "learning_rate": 0.0001971056978135966,
      "loss": 7.6455,
      "step": 68590
    },
    {
      "epoch": 8.25511432009627,
      "grad_norm": 4.06846809387207,
      "learning_rate": 0.0001971047891012634,
      "loss": 7.5681,
      "step": 68600
    },
    {
      "epoch": 8.256317689530686,
      "grad_norm": 2.2631430625915527,
      "learning_rate": 0.0001971038802483956,
      "loss": 7.5589,
      "step": 68610
    },
    {
      "epoch": 8.257521058965102,
      "grad_norm": 3.6923489570617676,
      "learning_rate": 0.00019710297125499448,
      "loss": 7.5747,
      "step": 68620
    },
    {
      "epoch": 8.25872442839952,
      "grad_norm": 5.725185394287109,
      "learning_rate": 0.00019710206212106139,
      "loss": 7.5944,
      "step": 68630
    },
    {
      "epoch": 8.259927797833935,
      "grad_norm": 3.2183244228363037,
      "learning_rate": 0.00019710115284659767,
      "loss": 7.5221,
      "step": 68640
    },
    {
      "epoch": 8.261131167268351,
      "grad_norm": 1.4189926385879517,
      "learning_rate": 0.0001971002434316046,
      "loss": 7.543,
      "step": 68650
    },
    {
      "epoch": 8.262334536702769,
      "grad_norm": 1.2534085512161255,
      "learning_rate": 0.0001970993338760835,
      "loss": 7.4875,
      "step": 68660
    },
    {
      "epoch": 8.263537906137184,
      "grad_norm": 2.4593253135681152,
      "learning_rate": 0.00019709842418003568,
      "loss": 7.4666,
      "step": 68670
    },
    {
      "epoch": 8.2647412755716,
      "grad_norm": 2.0745296478271484,
      "learning_rate": 0.0001970975143434625,
      "loss": 7.6271,
      "step": 68680
    },
    {
      "epoch": 8.265944645006018,
      "grad_norm": 3.8010599613189697,
      "learning_rate": 0.0001970966043663652,
      "loss": 7.5012,
      "step": 68690
    },
    {
      "epoch": 8.267148014440433,
      "grad_norm": 1.0840163230895996,
      "learning_rate": 0.00019709569424874513,
      "loss": 7.5432,
      "step": 68700
    },
    {
      "epoch": 8.26835138387485,
      "grad_norm": 1.1043710708618164,
      "learning_rate": 0.00019709478399060367,
      "loss": 7.6078,
      "step": 68710
    },
    {
      "epoch": 8.269554753309265,
      "grad_norm": 2.4659457206726074,
      "learning_rate": 0.00019709387359194208,
      "loss": 7.5378,
      "step": 68720
    },
    {
      "epoch": 8.270758122743683,
      "grad_norm": 5.003413200378418,
      "learning_rate": 0.00019709296305276165,
      "loss": 7.6184,
      "step": 68730
    },
    {
      "epoch": 8.271961492178098,
      "grad_norm": 1.0706757307052612,
      "learning_rate": 0.00019709205237306376,
      "loss": 7.4712,
      "step": 68740
    },
    {
      "epoch": 8.273164861612514,
      "grad_norm": 3.0960686206817627,
      "learning_rate": 0.0001970911415528497,
      "loss": 7.5741,
      "step": 68750
    },
    {
      "epoch": 8.274368231046932,
      "grad_norm": 1.5803383588790894,
      "learning_rate": 0.00019709023059212078,
      "loss": 7.6581,
      "step": 68760
    },
    {
      "epoch": 8.275571600481348,
      "grad_norm": 4.492832183837891,
      "learning_rate": 0.0001970893194908783,
      "loss": 7.5397,
      "step": 68770
    },
    {
      "epoch": 8.276774969915763,
      "grad_norm": 3.1308069229125977,
      "learning_rate": 0.00019708840824912363,
      "loss": 7.5709,
      "step": 68780
    },
    {
      "epoch": 8.277978339350181,
      "grad_norm": 1.3884894847869873,
      "learning_rate": 0.00019708749686685805,
      "loss": 7.513,
      "step": 68790
    },
    {
      "epoch": 8.279181708784597,
      "grad_norm": 4.107443809509277,
      "learning_rate": 0.0001970865853440829,
      "loss": 7.5936,
      "step": 68800
    },
    {
      "epoch": 8.280385078219012,
      "grad_norm": 1.3624972105026245,
      "learning_rate": 0.00019708567368079951,
      "loss": 7.577,
      "step": 68810
    },
    {
      "epoch": 8.28158844765343,
      "grad_norm": 4.004277229309082,
      "learning_rate": 0.00019708476187700914,
      "loss": 7.5543,
      "step": 68820
    },
    {
      "epoch": 8.282791817087846,
      "grad_norm": 5.335235595703125,
      "learning_rate": 0.00019708384993271318,
      "loss": 7.5473,
      "step": 68830
    },
    {
      "epoch": 8.283995186522262,
      "grad_norm": 4.383147239685059,
      "learning_rate": 0.00019708293784791292,
      "loss": 7.5184,
      "step": 68840
    },
    {
      "epoch": 8.28519855595668,
      "grad_norm": 2.059014081954956,
      "learning_rate": 0.00019708202562260967,
      "loss": 7.571,
      "step": 68850
    },
    {
      "epoch": 8.286401925391095,
      "grad_norm": 4.604721546173096,
      "learning_rate": 0.00019708111325680475,
      "loss": 7.5548,
      "step": 68860
    },
    {
      "epoch": 8.28760529482551,
      "grad_norm": 0.9158333539962769,
      "learning_rate": 0.00019708020075049948,
      "loss": 7.6148,
      "step": 68870
    },
    {
      "epoch": 8.288808664259928,
      "grad_norm": 3.3548567295074463,
      "learning_rate": 0.0001970792881036952,
      "loss": 7.59,
      "step": 68880
    },
    {
      "epoch": 8.290012033694344,
      "grad_norm": 4.503845691680908,
      "learning_rate": 0.00019707837531639325,
      "loss": 7.5678,
      "step": 68890
    },
    {
      "epoch": 8.29121540312876,
      "grad_norm": 1.3565207719802856,
      "learning_rate": 0.00019707746238859493,
      "loss": 7.5724,
      "step": 68900
    },
    {
      "epoch": 8.292418772563177,
      "grad_norm": 1.3087491989135742,
      "learning_rate": 0.00019707654932030153,
      "loss": 7.5304,
      "step": 68910
    },
    {
      "epoch": 8.293622141997593,
      "grad_norm": 4.389442443847656,
      "learning_rate": 0.00019707563611151438,
      "loss": 7.5141,
      "step": 68920
    },
    {
      "epoch": 8.294825511432009,
      "grad_norm": 3.10077166557312,
      "learning_rate": 0.00019707472276223484,
      "loss": 7.6327,
      "step": 68930
    },
    {
      "epoch": 8.296028880866427,
      "grad_norm": 1.7084767818450928,
      "learning_rate": 0.00019707380927246423,
      "loss": 7.4655,
      "step": 68940
    },
    {
      "epoch": 8.297232250300842,
      "grad_norm": 5.375579357147217,
      "learning_rate": 0.0001970728956422038,
      "loss": 7.5407,
      "step": 68950
    },
    {
      "epoch": 8.298435619735258,
      "grad_norm": 1.3260475397109985,
      "learning_rate": 0.00019707198187145496,
      "loss": 7.6192,
      "step": 68960
    },
    {
      "epoch": 8.299638989169676,
      "grad_norm": 4.735003471374512,
      "learning_rate": 0.000197071067960219,
      "loss": 7.5451,
      "step": 68970
    },
    {
      "epoch": 8.300842358604092,
      "grad_norm": 2.3451125621795654,
      "learning_rate": 0.0001970701539084972,
      "loss": 7.533,
      "step": 68980
    },
    {
      "epoch": 8.302045728038507,
      "grad_norm": 4.834249496459961,
      "learning_rate": 0.00019706923971629097,
      "loss": 7.4932,
      "step": 68990
    },
    {
      "epoch": 8.303249097472925,
      "grad_norm": 1.004472017288208,
      "learning_rate": 0.00019706832538360155,
      "loss": 7.5187,
      "step": 69000
    },
    {
      "epoch": 8.30445246690734,
      "grad_norm": 3.006849527359009,
      "learning_rate": 0.00019706741091043032,
      "loss": 7.4423,
      "step": 69010
    },
    {
      "epoch": 8.305655836341757,
      "grad_norm": 1.71808922290802,
      "learning_rate": 0.00019706649629677855,
      "loss": 7.5339,
      "step": 69020
    },
    {
      "epoch": 8.306859205776174,
      "grad_norm": 1.9772635698318481,
      "learning_rate": 0.0001970655815426476,
      "loss": 7.6016,
      "step": 69030
    },
    {
      "epoch": 8.30806257521059,
      "grad_norm": 1.686840295791626,
      "learning_rate": 0.00019706466664803882,
      "loss": 7.544,
      "step": 69040
    },
    {
      "epoch": 8.309265944645006,
      "grad_norm": 2.325416088104248,
      "learning_rate": 0.00019706375161295347,
      "loss": 7.5321,
      "step": 69050
    },
    {
      "epoch": 8.310469314079423,
      "grad_norm": 1.6634761095046997,
      "learning_rate": 0.00019706283643739295,
      "loss": 7.4735,
      "step": 69060
    },
    {
      "epoch": 8.311672683513839,
      "grad_norm": 2.6000471115112305,
      "learning_rate": 0.0001970619211213585,
      "loss": 7.5945,
      "step": 69070
    },
    {
      "epoch": 8.312876052948255,
      "grad_norm": 3.5839436054229736,
      "learning_rate": 0.0001970610056648515,
      "loss": 7.5622,
      "step": 69080
    },
    {
      "epoch": 8.314079422382672,
      "grad_norm": 1.9345314502716064,
      "learning_rate": 0.00019706009006787326,
      "loss": 7.5756,
      "step": 69090
    },
    {
      "epoch": 8.315282791817088,
      "grad_norm": 2.8428375720977783,
      "learning_rate": 0.00019705917433042512,
      "loss": 7.4851,
      "step": 69100
    },
    {
      "epoch": 8.316486161251504,
      "grad_norm": 1.6608713865280151,
      "learning_rate": 0.00019705825845250837,
      "loss": 7.5587,
      "step": 69110
    },
    {
      "epoch": 8.31768953068592,
      "grad_norm": 1.403356909751892,
      "learning_rate": 0.00019705734243412437,
      "loss": 7.5413,
      "step": 69120
    },
    {
      "epoch": 8.318892900120337,
      "grad_norm": 4.303885459899902,
      "learning_rate": 0.00019705642627527444,
      "loss": 7.5151,
      "step": 69130
    },
    {
      "epoch": 8.320096269554753,
      "grad_norm": 2.9453518390655518,
      "learning_rate": 0.0001970555099759599,
      "loss": 7.7009,
      "step": 69140
    },
    {
      "epoch": 8.321299638989169,
      "grad_norm": 1.3282310962677002,
      "learning_rate": 0.00019705459353618206,
      "loss": 7.5429,
      "step": 69150
    },
    {
      "epoch": 8.322503008423586,
      "grad_norm": 1.5983847379684448,
      "learning_rate": 0.00019705367695594227,
      "loss": 7.5995,
      "step": 69160
    },
    {
      "epoch": 8.323706377858002,
      "grad_norm": 4.7585320472717285,
      "learning_rate": 0.0001970527602352418,
      "loss": 7.5006,
      "step": 69170
    },
    {
      "epoch": 8.324909747292418,
      "grad_norm": 2.053856372833252,
      "learning_rate": 0.00019705184337408209,
      "loss": 7.5291,
      "step": 69180
    },
    {
      "epoch": 8.326113116726836,
      "grad_norm": 1.6657274961471558,
      "learning_rate": 0.00019705092637246438,
      "loss": 7.5205,
      "step": 69190
    },
    {
      "epoch": 8.327316486161251,
      "grad_norm": 1.5674819946289062,
      "learning_rate": 0.00019705000923039002,
      "loss": 7.56,
      "step": 69200
    },
    {
      "epoch": 8.328519855595667,
      "grad_norm": 3.1776132583618164,
      "learning_rate": 0.00019704909194786033,
      "loss": 7.5724,
      "step": 69210
    },
    {
      "epoch": 8.329723225030085,
      "grad_norm": 0.933430016040802,
      "learning_rate": 0.00019704817452487663,
      "loss": 7.5107,
      "step": 69220
    },
    {
      "epoch": 8.3309265944645,
      "grad_norm": 2.722656011581421,
      "learning_rate": 0.0001970472569614403,
      "loss": 7.5993,
      "step": 69230
    },
    {
      "epoch": 8.332129963898916,
      "grad_norm": 2.6975135803222656,
      "learning_rate": 0.00019704633925755262,
      "loss": 7.5472,
      "step": 69240
    },
    {
      "epoch": 8.333333333333334,
      "grad_norm": 2.1026878356933594,
      "learning_rate": 0.00019704542141321487,
      "loss": 7.5919,
      "step": 69250
    },
    {
      "epoch": 8.33453670276775,
      "grad_norm": 1.7851208448410034,
      "learning_rate": 0.00019704450342842848,
      "loss": 7.5893,
      "step": 69260
    },
    {
      "epoch": 8.335740072202166,
      "grad_norm": 1.280189037322998,
      "learning_rate": 0.00019704358530319476,
      "loss": 7.5784,
      "step": 69270
    },
    {
      "epoch": 8.336943441636583,
      "grad_norm": 1.7406482696533203,
      "learning_rate": 0.00019704266703751496,
      "loss": 7.5389,
      "step": 69280
    },
    {
      "epoch": 8.338146811070999,
      "grad_norm": 2.391625165939331,
      "learning_rate": 0.00019704174863139048,
      "loss": 7.562,
      "step": 69290
    },
    {
      "epoch": 8.339350180505415,
      "grad_norm": 4.894526958465576,
      "learning_rate": 0.00019704083008482263,
      "loss": 7.5398,
      "step": 69300
    },
    {
      "epoch": 8.340553549939832,
      "grad_norm": 3.314066171646118,
      "learning_rate": 0.00019703991139781278,
      "loss": 7.5561,
      "step": 69310
    },
    {
      "epoch": 8.341756919374248,
      "grad_norm": 4.265951633453369,
      "learning_rate": 0.00019703899257036218,
      "loss": 7.5035,
      "step": 69320
    },
    {
      "epoch": 8.342960288808664,
      "grad_norm": 2.081082582473755,
      "learning_rate": 0.00019703807360247223,
      "loss": 7.5066,
      "step": 69330
    },
    {
      "epoch": 8.344163658243081,
      "grad_norm": 1.7546112537384033,
      "learning_rate": 0.0001970371544941442,
      "loss": 7.5573,
      "step": 69340
    },
    {
      "epoch": 8.345367027677497,
      "grad_norm": 2.2228806018829346,
      "learning_rate": 0.00019703623524537945,
      "loss": 7.5062,
      "step": 69350
    },
    {
      "epoch": 8.346570397111913,
      "grad_norm": 1.3018989562988281,
      "learning_rate": 0.0001970353158561793,
      "loss": 7.4785,
      "step": 69360
    },
    {
      "epoch": 8.34777376654633,
      "grad_norm": 1.700270652770996,
      "learning_rate": 0.00019703439632654513,
      "loss": 7.5577,
      "step": 69370
    },
    {
      "epoch": 8.348977135980746,
      "grad_norm": 5.086368083953857,
      "learning_rate": 0.00019703347665647817,
      "loss": 7.577,
      "step": 69380
    },
    {
      "epoch": 8.350180505415162,
      "grad_norm": 2.2030587196350098,
      "learning_rate": 0.00019703255684597987,
      "loss": 7.5343,
      "step": 69390
    },
    {
      "epoch": 8.35138387484958,
      "grad_norm": 2.14054012298584,
      "learning_rate": 0.00019703163689505147,
      "loss": 7.5883,
      "step": 69400
    },
    {
      "epoch": 8.352587244283995,
      "grad_norm": 1.624119758605957,
      "learning_rate": 0.00019703071680369437,
      "loss": 7.6223,
      "step": 69410
    },
    {
      "epoch": 8.353790613718411,
      "grad_norm": 2.0320706367492676,
      "learning_rate": 0.00019702979657190985,
      "loss": 7.5987,
      "step": 69420
    },
    {
      "epoch": 8.354993983152827,
      "grad_norm": 3.275172710418701,
      "learning_rate": 0.00019702887619969922,
      "loss": 7.5904,
      "step": 69430
    },
    {
      "epoch": 8.356197352587245,
      "grad_norm": 2.4781830310821533,
      "learning_rate": 0.00019702795568706392,
      "loss": 7.51,
      "step": 69440
    },
    {
      "epoch": 8.35740072202166,
      "grad_norm": 2.811317205429077,
      "learning_rate": 0.00019702703503400517,
      "loss": 7.5035,
      "step": 69450
    },
    {
      "epoch": 8.358604091456076,
      "grad_norm": 2.913123846054077,
      "learning_rate": 0.00019702611424052434,
      "loss": 7.5816,
      "step": 69460
    },
    {
      "epoch": 8.359807460890494,
      "grad_norm": 3.629225492477417,
      "learning_rate": 0.0001970251933066228,
      "loss": 7.5186,
      "step": 69470
    },
    {
      "epoch": 8.36101083032491,
      "grad_norm": 2.3426499366760254,
      "learning_rate": 0.00019702427223230182,
      "loss": 7.611,
      "step": 69480
    },
    {
      "epoch": 8.362214199759325,
      "grad_norm": 2.5738203525543213,
      "learning_rate": 0.00019702335101756278,
      "loss": 7.5928,
      "step": 69490
    },
    {
      "epoch": 8.363417569193743,
      "grad_norm": 4.000921249389648,
      "learning_rate": 0.00019702242966240698,
      "loss": 7.6396,
      "step": 69500
    },
    {
      "epoch": 8.364620938628159,
      "grad_norm": 2.1955952644348145,
      "learning_rate": 0.0001970215081668358,
      "loss": 7.5869,
      "step": 69510
    },
    {
      "epoch": 8.365824308062574,
      "grad_norm": 2.526052236557007,
      "learning_rate": 0.0001970205865308505,
      "loss": 7.5259,
      "step": 69520
    },
    {
      "epoch": 8.367027677496992,
      "grad_norm": 2.636810541152954,
      "learning_rate": 0.0001970196647544525,
      "loss": 7.4302,
      "step": 69530
    },
    {
      "epoch": 8.368231046931408,
      "grad_norm": 1.7989444732666016,
      "learning_rate": 0.00019701874283764307,
      "loss": 7.5755,
      "step": 69540
    },
    {
      "epoch": 8.369434416365824,
      "grad_norm": 3.394885301589966,
      "learning_rate": 0.0001970178207804236,
      "loss": 7.5249,
      "step": 69550
    },
    {
      "epoch": 8.370637785800241,
      "grad_norm": 3.081509828567505,
      "learning_rate": 0.00019701689858279533,
      "loss": 7.5408,
      "step": 69560
    },
    {
      "epoch": 8.371841155234657,
      "grad_norm": 1.3034275770187378,
      "learning_rate": 0.00019701597624475968,
      "loss": 7.586,
      "step": 69570
    },
    {
      "epoch": 8.373044524669073,
      "grad_norm": 3.7937939167022705,
      "learning_rate": 0.00019701505376631796,
      "loss": 7.6025,
      "step": 69580
    },
    {
      "epoch": 8.37424789410349,
      "grad_norm": 0.9847928285598755,
      "learning_rate": 0.00019701413114747154,
      "loss": 7.6243,
      "step": 69590
    },
    {
      "epoch": 8.375451263537906,
      "grad_norm": 2.6686904430389404,
      "learning_rate": 0.00019701320838822167,
      "loss": 7.5863,
      "step": 69600
    },
    {
      "epoch": 8.376654632972322,
      "grad_norm": 3.7679669857025146,
      "learning_rate": 0.00019701228548856979,
      "loss": 7.4367,
      "step": 69610
    },
    {
      "epoch": 8.37785800240674,
      "grad_norm": 1.0155153274536133,
      "learning_rate": 0.0001970113624485171,
      "loss": 7.5095,
      "step": 69620
    },
    {
      "epoch": 8.379061371841155,
      "grad_norm": 6.318257808685303,
      "learning_rate": 0.0001970104392680651,
      "loss": 7.6557,
      "step": 69630
    },
    {
      "epoch": 8.380264741275571,
      "grad_norm": 1.2143404483795166,
      "learning_rate": 0.000197009515947215,
      "loss": 7.4981,
      "step": 69640
    },
    {
      "epoch": 8.381468110709989,
      "grad_norm": 2.227907180786133,
      "learning_rate": 0.0001970085924859682,
      "loss": 7.5467,
      "step": 69650
    },
    {
      "epoch": 8.382671480144404,
      "grad_norm": 3.4254238605499268,
      "learning_rate": 0.000197007668884326,
      "loss": 7.6048,
      "step": 69660
    },
    {
      "epoch": 8.38387484957882,
      "grad_norm": 1.2634449005126953,
      "learning_rate": 0.00019700674514228975,
      "loss": 7.5963,
      "step": 69670
    },
    {
      "epoch": 8.385078219013238,
      "grad_norm": 3.4352493286132812,
      "learning_rate": 0.0001970058212598608,
      "loss": 7.5584,
      "step": 69680
    },
    {
      "epoch": 8.386281588447654,
      "grad_norm": 2.7097935676574707,
      "learning_rate": 0.0001970048972370405,
      "loss": 7.5435,
      "step": 69690
    },
    {
      "epoch": 8.38748495788207,
      "grad_norm": 1.6629337072372437,
      "learning_rate": 0.00019700397307383013,
      "loss": 7.6898,
      "step": 69700
    },
    {
      "epoch": 8.388688327316487,
      "grad_norm": 2.548401117324829,
      "learning_rate": 0.00019700304877023106,
      "loss": 7.5624,
      "step": 69710
    },
    {
      "epoch": 8.389891696750903,
      "grad_norm": 1.9017082452774048,
      "learning_rate": 0.00019700212432624463,
      "loss": 7.4993,
      "step": 69720
    },
    {
      "epoch": 8.391095066185319,
      "grad_norm": 1.875321626663208,
      "learning_rate": 0.0001970011997418722,
      "loss": 7.4741,
      "step": 69730
    },
    {
      "epoch": 8.392298435619736,
      "grad_norm": 2.2850897312164307,
      "learning_rate": 0.00019700027501711506,
      "loss": 7.5867,
      "step": 69740
    },
    {
      "epoch": 8.393501805054152,
      "grad_norm": 1.5570414066314697,
      "learning_rate": 0.00019699935015197458,
      "loss": 7.6479,
      "step": 69750
    },
    {
      "epoch": 8.394705174488568,
      "grad_norm": 1.8163044452667236,
      "learning_rate": 0.0001969984251464521,
      "loss": 7.5948,
      "step": 69760
    },
    {
      "epoch": 8.395908543922985,
      "grad_norm": 1.5942484140396118,
      "learning_rate": 0.00019699750000054896,
      "loss": 7.4422,
      "step": 69770
    },
    {
      "epoch": 8.397111913357401,
      "grad_norm": 4.415695667266846,
      "learning_rate": 0.00019699657471426647,
      "loss": 7.6148,
      "step": 69780
    },
    {
      "epoch": 8.398315282791817,
      "grad_norm": 1.3548928499221802,
      "learning_rate": 0.000196995649287606,
      "loss": 7.56,
      "step": 69790
    },
    {
      "epoch": 8.399518652226233,
      "grad_norm": 2.413607358932495,
      "learning_rate": 0.00019699472372056886,
      "loss": 7.5521,
      "step": 69800
    },
    {
      "epoch": 8.40072202166065,
      "grad_norm": 2.9080374240875244,
      "learning_rate": 0.00019699379801315644,
      "loss": 7.4956,
      "step": 69810
    },
    {
      "epoch": 8.401925391095066,
      "grad_norm": 4.320467472076416,
      "learning_rate": 0.00019699287216537,
      "loss": 7.5482,
      "step": 69820
    },
    {
      "epoch": 8.403128760529482,
      "grad_norm": 6.3931450843811035,
      "learning_rate": 0.00019699194617721097,
      "loss": 7.614,
      "step": 69830
    },
    {
      "epoch": 8.4043321299639,
      "grad_norm": 1.192388653755188,
      "learning_rate": 0.00019699102004868064,
      "loss": 7.6063,
      "step": 69840
    },
    {
      "epoch": 8.405535499398315,
      "grad_norm": 2.541799783706665,
      "learning_rate": 0.00019699009377978035,
      "loss": 7.6549,
      "step": 69850
    },
    {
      "epoch": 8.406738868832731,
      "grad_norm": 1.4081885814666748,
      "learning_rate": 0.00019698916737051145,
      "loss": 7.5446,
      "step": 69860
    },
    {
      "epoch": 8.407942238267148,
      "grad_norm": 2.5232038497924805,
      "learning_rate": 0.00019698824082087528,
      "loss": 7.5346,
      "step": 69870
    },
    {
      "epoch": 8.409145607701564,
      "grad_norm": 2.2329695224761963,
      "learning_rate": 0.00019698731413087317,
      "loss": 7.4967,
      "step": 69880
    },
    {
      "epoch": 8.41034897713598,
      "grad_norm": 1.0750991106033325,
      "learning_rate": 0.00019698638730050649,
      "loss": 7.5455,
      "step": 69890
    },
    {
      "epoch": 8.411552346570398,
      "grad_norm": 3.1521692276000977,
      "learning_rate": 0.00019698546032977655,
      "loss": 7.5921,
      "step": 69900
    },
    {
      "epoch": 8.412755716004813,
      "grad_norm": 1.1493871212005615,
      "learning_rate": 0.00019698453321868468,
      "loss": 7.4912,
      "step": 69910
    },
    {
      "epoch": 8.41395908543923,
      "grad_norm": 1.6853001117706299,
      "learning_rate": 0.00019698360596723228,
      "loss": 7.5831,
      "step": 69920
    },
    {
      "epoch": 8.415162454873647,
      "grad_norm": 2.5247533321380615,
      "learning_rate": 0.00019698267857542066,
      "loss": 7.5224,
      "step": 69930
    },
    {
      "epoch": 8.416365824308063,
      "grad_norm": 1.6904728412628174,
      "learning_rate": 0.00019698175104325113,
      "loss": 7.5352,
      "step": 69940
    },
    {
      "epoch": 8.417569193742478,
      "grad_norm": 1.9604597091674805,
      "learning_rate": 0.0001969808233707251,
      "loss": 7.5345,
      "step": 69950
    },
    {
      "epoch": 8.418772563176896,
      "grad_norm": 3.829887628555298,
      "learning_rate": 0.00019697989555784385,
      "loss": 7.6028,
      "step": 69960
    },
    {
      "epoch": 8.419975932611312,
      "grad_norm": 1.9152189493179321,
      "learning_rate": 0.00019697896760460874,
      "loss": 7.4951,
      "step": 69970
    },
    {
      "epoch": 8.421179302045728,
      "grad_norm": 4.484202861785889,
      "learning_rate": 0.00019697803951102113,
      "loss": 7.5123,
      "step": 69980
    },
    {
      "epoch": 8.422382671480145,
      "grad_norm": 2.0393359661102295,
      "learning_rate": 0.00019697711127708236,
      "loss": 7.5755,
      "step": 69990
    },
    {
      "epoch": 8.42358604091456,
      "grad_norm": 2.936713457107544,
      "learning_rate": 0.00019697618290279376,
      "loss": 7.5164,
      "step": 70000
    },
    {
      "epoch": 8.424789410348977,
      "grad_norm": 1.60394287109375,
      "learning_rate": 0.0001969752543881567,
      "loss": 7.5921,
      "step": 70010
    },
    {
      "epoch": 8.425992779783394,
      "grad_norm": 1.1328998804092407,
      "learning_rate": 0.00019697432573317247,
      "loss": 7.5357,
      "step": 70020
    },
    {
      "epoch": 8.42719614921781,
      "grad_norm": 3.7305259704589844,
      "learning_rate": 0.00019697339693784248,
      "loss": 7.5175,
      "step": 70030
    },
    {
      "epoch": 8.428399518652226,
      "grad_norm": 1.4460114240646362,
      "learning_rate": 0.000196972468002168,
      "loss": 7.4581,
      "step": 70040
    },
    {
      "epoch": 8.429602888086643,
      "grad_norm": 2.8230886459350586,
      "learning_rate": 0.00019697153892615047,
      "loss": 7.5622,
      "step": 70050
    },
    {
      "epoch": 8.43080625752106,
      "grad_norm": 1.8971261978149414,
      "learning_rate": 0.00019697060970979115,
      "loss": 7.5077,
      "step": 70060
    },
    {
      "epoch": 8.432009626955475,
      "grad_norm": 1.3763706684112549,
      "learning_rate": 0.00019696968035309146,
      "loss": 7.5594,
      "step": 70070
    },
    {
      "epoch": 8.433212996389893,
      "grad_norm": 4.566793918609619,
      "learning_rate": 0.00019696875085605266,
      "loss": 7.5197,
      "step": 70080
    },
    {
      "epoch": 8.434416365824308,
      "grad_norm": 3.3624672889709473,
      "learning_rate": 0.00019696782121867614,
      "loss": 7.5901,
      "step": 70090
    },
    {
      "epoch": 8.435619735258724,
      "grad_norm": 3.586397647857666,
      "learning_rate": 0.00019696689144096325,
      "loss": 7.5154,
      "step": 70100
    },
    {
      "epoch": 8.43682310469314,
      "grad_norm": 1.238344430923462,
      "learning_rate": 0.0001969659615229153,
      "loss": 7.5655,
      "step": 70110
    },
    {
      "epoch": 8.438026474127557,
      "grad_norm": 1.6360137462615967,
      "learning_rate": 0.0001969650314645337,
      "loss": 7.6414,
      "step": 70120
    },
    {
      "epoch": 8.439229843561973,
      "grad_norm": 6.115720272064209,
      "learning_rate": 0.00019696410126581975,
      "loss": 7.5798,
      "step": 70130
    },
    {
      "epoch": 8.440433212996389,
      "grad_norm": 3.8817498683929443,
      "learning_rate": 0.0001969631709267748,
      "loss": 7.5112,
      "step": 70140
    },
    {
      "epoch": 8.441636582430807,
      "grad_norm": 2.3776049613952637,
      "learning_rate": 0.0001969622404474002,
      "loss": 7.5668,
      "step": 70150
    },
    {
      "epoch": 8.442839951865222,
      "grad_norm": 1.765113115310669,
      "learning_rate": 0.00019696130982769733,
      "loss": 7.5458,
      "step": 70160
    },
    {
      "epoch": 8.444043321299638,
      "grad_norm": 1.173095703125,
      "learning_rate": 0.00019696037906766746,
      "loss": 7.5786,
      "step": 70170
    },
    {
      "epoch": 8.445246690734056,
      "grad_norm": 3.3456172943115234,
      "learning_rate": 0.000196959448167312,
      "loss": 7.6441,
      "step": 70180
    },
    {
      "epoch": 8.446450060168472,
      "grad_norm": 2.850996971130371,
      "learning_rate": 0.0001969585171266323,
      "loss": 7.5242,
      "step": 70190
    },
    {
      "epoch": 8.447653429602887,
      "grad_norm": 1.7878836393356323,
      "learning_rate": 0.00019695758594562965,
      "loss": 7.5496,
      "step": 70200
    },
    {
      "epoch": 8.448856799037305,
      "grad_norm": 2.1548261642456055,
      "learning_rate": 0.00019695665462430545,
      "loss": 7.5587,
      "step": 70210
    },
    {
      "epoch": 8.45006016847172,
      "grad_norm": 1.7706656455993652,
      "learning_rate": 0.00019695572316266103,
      "loss": 7.5371,
      "step": 70220
    },
    {
      "epoch": 8.451263537906136,
      "grad_norm": 3.8058576583862305,
      "learning_rate": 0.00019695479156069776,
      "loss": 7.563,
      "step": 70230
    },
    {
      "epoch": 8.452466907340554,
      "grad_norm": 2.564669370651245,
      "learning_rate": 0.00019695385981841697,
      "loss": 7.6053,
      "step": 70240
    },
    {
      "epoch": 8.45367027677497,
      "grad_norm": 1.0102994441986084,
      "learning_rate": 0.00019695292793581997,
      "loss": 7.6064,
      "step": 70250
    },
    {
      "epoch": 8.454873646209386,
      "grad_norm": 1.3062176704406738,
      "learning_rate": 0.00019695199591290822,
      "loss": 7.5409,
      "step": 70260
    },
    {
      "epoch": 8.456077015643803,
      "grad_norm": 0.8895031809806824,
      "learning_rate": 0.00019695106374968292,
      "loss": 7.5031,
      "step": 70270
    },
    {
      "epoch": 8.457280385078219,
      "grad_norm": 1.7016993761062622,
      "learning_rate": 0.00019695013144614552,
      "loss": 7.5381,
      "step": 70280
    },
    {
      "epoch": 8.458483754512635,
      "grad_norm": 1.497953176498413,
      "learning_rate": 0.00019694919900229736,
      "loss": 7.559,
      "step": 70290
    },
    {
      "epoch": 8.459687123947052,
      "grad_norm": 2.9686248302459717,
      "learning_rate": 0.00019694826641813973,
      "loss": 7.5928,
      "step": 70300
    },
    {
      "epoch": 8.460890493381468,
      "grad_norm": 1.3379900455474854,
      "learning_rate": 0.00019694733369367406,
      "loss": 7.5338,
      "step": 70310
    },
    {
      "epoch": 8.462093862815884,
      "grad_norm": 5.383237838745117,
      "learning_rate": 0.00019694640082890165,
      "loss": 7.481,
      "step": 70320
    },
    {
      "epoch": 8.463297232250302,
      "grad_norm": 2.661841630935669,
      "learning_rate": 0.0001969454678238239,
      "loss": 7.5019,
      "step": 70330
    },
    {
      "epoch": 8.464500601684717,
      "grad_norm": 3.2884738445281982,
      "learning_rate": 0.00019694453467844207,
      "loss": 7.5538,
      "step": 70340
    },
    {
      "epoch": 8.465703971119133,
      "grad_norm": 2.750004529953003,
      "learning_rate": 0.00019694360139275755,
      "loss": 7.5315,
      "step": 70350
    },
    {
      "epoch": 8.46690734055355,
      "grad_norm": 2.629000663757324,
      "learning_rate": 0.00019694266796677176,
      "loss": 7.4495,
      "step": 70360
    },
    {
      "epoch": 8.468110709987966,
      "grad_norm": 4.275704383850098,
      "learning_rate": 0.00019694173440048596,
      "loss": 7.473,
      "step": 70370
    },
    {
      "epoch": 8.469314079422382,
      "grad_norm": 1.3949379920959473,
      "learning_rate": 0.00019694080069390158,
      "loss": 7.5222,
      "step": 70380
    },
    {
      "epoch": 8.4705174488568,
      "grad_norm": 2.105985641479492,
      "learning_rate": 0.00019693986684701987,
      "loss": 7.6076,
      "step": 70390
    },
    {
      "epoch": 8.471720818291216,
      "grad_norm": 4.698059558868408,
      "learning_rate": 0.0001969389328598423,
      "loss": 7.5532,
      "step": 70400
    },
    {
      "epoch": 8.472924187725631,
      "grad_norm": 1.465378999710083,
      "learning_rate": 0.0001969379987323701,
      "loss": 7.5151,
      "step": 70410
    },
    {
      "epoch": 8.474127557160049,
      "grad_norm": 1.6236846446990967,
      "learning_rate": 0.0001969370644646047,
      "loss": 7.4925,
      "step": 70420
    },
    {
      "epoch": 8.475330926594465,
      "grad_norm": 5.72561502456665,
      "learning_rate": 0.00019693613005654744,
      "loss": 7.5585,
      "step": 70430
    },
    {
      "epoch": 8.47653429602888,
      "grad_norm": 2.293527364730835,
      "learning_rate": 0.0001969351955081997,
      "loss": 7.5193,
      "step": 70440
    },
    {
      "epoch": 8.477737665463298,
      "grad_norm": 5.550730228424072,
      "learning_rate": 0.00019693426081956273,
      "loss": 7.5706,
      "step": 70450
    },
    {
      "epoch": 8.478941034897714,
      "grad_norm": 1.1612584590911865,
      "learning_rate": 0.00019693332599063801,
      "loss": 7.6691,
      "step": 70460
    },
    {
      "epoch": 8.48014440433213,
      "grad_norm": 1.364365577697754,
      "learning_rate": 0.00019693239102142683,
      "loss": 7.5741,
      "step": 70470
    },
    {
      "epoch": 8.481347773766545,
      "grad_norm": 3.126809597015381,
      "learning_rate": 0.0001969314559119305,
      "loss": 7.6143,
      "step": 70480
    },
    {
      "epoch": 8.482551143200963,
      "grad_norm": 2.3165807723999023,
      "learning_rate": 0.00019693052066215047,
      "loss": 7.5971,
      "step": 70490
    },
    {
      "epoch": 8.483754512635379,
      "grad_norm": 1.6918388605117798,
      "learning_rate": 0.00019692958527208805,
      "loss": 7.5596,
      "step": 70500
    },
    {
      "epoch": 8.484957882069795,
      "grad_norm": 2.422433614730835,
      "learning_rate": 0.00019692864974174456,
      "loss": 7.4805,
      "step": 70510
    },
    {
      "epoch": 8.486161251504212,
      "grad_norm": 1.60307776927948,
      "learning_rate": 0.00019692771407112139,
      "loss": 7.5824,
      "step": 70520
    },
    {
      "epoch": 8.487364620938628,
      "grad_norm": 4.616024494171143,
      "learning_rate": 0.0001969267782602199,
      "loss": 7.4813,
      "step": 70530
    },
    {
      "epoch": 8.488567990373044,
      "grad_norm": 1.1546640396118164,
      "learning_rate": 0.00019692584230904142,
      "loss": 7.4846,
      "step": 70540
    },
    {
      "epoch": 8.489771359807461,
      "grad_norm": 2.1627120971679688,
      "learning_rate": 0.0001969249062175873,
      "loss": 7.5749,
      "step": 70550
    },
    {
      "epoch": 8.490974729241877,
      "grad_norm": 1.7128040790557861,
      "learning_rate": 0.00019692396998585894,
      "loss": 7.499,
      "step": 70560
    },
    {
      "epoch": 8.492178098676293,
      "grad_norm": 1.1143193244934082,
      "learning_rate": 0.00019692303361385765,
      "loss": 7.598,
      "step": 70570
    },
    {
      "epoch": 8.49338146811071,
      "grad_norm": 6.3489155769348145,
      "learning_rate": 0.0001969220971015848,
      "loss": 7.6215,
      "step": 70580
    },
    {
      "epoch": 8.494584837545126,
      "grad_norm": 2.653006076812744,
      "learning_rate": 0.00019692116044904177,
      "loss": 7.493,
      "step": 70590
    },
    {
      "epoch": 8.495788206979542,
      "grad_norm": 1.5156558752059937,
      "learning_rate": 0.00019692022365622988,
      "loss": 7.5372,
      "step": 70600
    },
    {
      "epoch": 8.49699157641396,
      "grad_norm": 2.8469014167785645,
      "learning_rate": 0.00019691928672315047,
      "loss": 7.5831,
      "step": 70610
    },
    {
      "epoch": 8.498194945848375,
      "grad_norm": 1.1309077739715576,
      "learning_rate": 0.00019691834964980496,
      "loss": 7.5292,
      "step": 70620
    },
    {
      "epoch": 8.499398315282791,
      "grad_norm": 1.6052037477493286,
      "learning_rate": 0.00019691741243619464,
      "loss": 7.6114,
      "step": 70630
    },
    {
      "epoch": 8.500601684717209,
      "grad_norm": 1.7786142826080322,
      "learning_rate": 0.00019691647508232092,
      "loss": 7.5592,
      "step": 70640
    },
    {
      "epoch": 8.501805054151625,
      "grad_norm": 2.218583583831787,
      "learning_rate": 0.00019691553758818513,
      "loss": 7.5153,
      "step": 70650
    },
    {
      "epoch": 8.50300842358604,
      "grad_norm": 2.2946512699127197,
      "learning_rate": 0.0001969145999537886,
      "loss": 7.5829,
      "step": 70660
    },
    {
      "epoch": 8.504211793020458,
      "grad_norm": 1.7440555095672607,
      "learning_rate": 0.00019691366217913276,
      "loss": 7.5762,
      "step": 70670
    },
    {
      "epoch": 8.505415162454874,
      "grad_norm": 2.6018712520599365,
      "learning_rate": 0.0001969127242642189,
      "loss": 7.5387,
      "step": 70680
    },
    {
      "epoch": 8.50661853188929,
      "grad_norm": 1.3647104501724243,
      "learning_rate": 0.0001969117862090484,
      "loss": 7.508,
      "step": 70690
    },
    {
      "epoch": 8.507821901323707,
      "grad_norm": 1.5754398107528687,
      "learning_rate": 0.00019691084801362263,
      "loss": 7.5032,
      "step": 70700
    },
    {
      "epoch": 8.509025270758123,
      "grad_norm": 2.110182762145996,
      "learning_rate": 0.00019690990967794294,
      "loss": 7.4783,
      "step": 70710
    },
    {
      "epoch": 8.510228640192539,
      "grad_norm": 10.546307563781738,
      "learning_rate": 0.00019690897120201064,
      "loss": 7.5894,
      "step": 70720
    },
    {
      "epoch": 8.511432009626956,
      "grad_norm": 874.0382690429688,
      "learning_rate": 0.0001969080325858272,
      "loss": 7.5401,
      "step": 70730
    },
    {
      "epoch": 8.512635379061372,
      "grad_norm": 1.38412344455719,
      "learning_rate": 0.00019690709382939384,
      "loss": 7.7281,
      "step": 70740
    },
    {
      "epoch": 8.513838748495788,
      "grad_norm": 9.912545204162598,
      "learning_rate": 0.00019690615493271203,
      "loss": 7.6684,
      "step": 70750
    },
    {
      "epoch": 8.515042117930205,
      "grad_norm": 4.450211048126221,
      "learning_rate": 0.0001969052158957831,
      "loss": 7.4833,
      "step": 70760
    },
    {
      "epoch": 8.516245487364621,
      "grad_norm": 5.637786388397217,
      "learning_rate": 0.0001969042767186084,
      "loss": 7.5644,
      "step": 70770
    },
    {
      "epoch": 8.517448856799037,
      "grad_norm": 3.0561916828155518,
      "learning_rate": 0.00019690333740118924,
      "loss": 7.5679,
      "step": 70780
    },
    {
      "epoch": 8.518652226233453,
      "grad_norm": 2.1943979263305664,
      "learning_rate": 0.00019690239794352708,
      "loss": 7.5543,
      "step": 70790
    },
    {
      "epoch": 8.51985559566787,
      "grad_norm": 3.0288281440734863,
      "learning_rate": 0.00019690145834562318,
      "loss": 7.548,
      "step": 70800
    },
    {
      "epoch": 8.521058965102286,
      "grad_norm": 1.6482231616973877,
      "learning_rate": 0.00019690051860747894,
      "loss": 7.5413,
      "step": 70810
    },
    {
      "epoch": 8.522262334536702,
      "grad_norm": 3.06941556930542,
      "learning_rate": 0.0001968995787290958,
      "loss": 7.5749,
      "step": 70820
    },
    {
      "epoch": 8.52346570397112,
      "grad_norm": 3.1419012546539307,
      "learning_rate": 0.00019689863871047498,
      "loss": 7.5293,
      "step": 70830
    },
    {
      "epoch": 8.524669073405535,
      "grad_norm": 1.204333782196045,
      "learning_rate": 0.0001968976985516179,
      "loss": 7.4705,
      "step": 70840
    },
    {
      "epoch": 8.525872442839951,
      "grad_norm": 2.838078498840332,
      "learning_rate": 0.000196896758252526,
      "loss": 7.5857,
      "step": 70850
    },
    {
      "epoch": 8.527075812274369,
      "grad_norm": 4.2524800300598145,
      "learning_rate": 0.0001968958178132005,
      "loss": 7.572,
      "step": 70860
    },
    {
      "epoch": 8.528279181708784,
      "grad_norm": 2.989532232284546,
      "learning_rate": 0.00019689487723364283,
      "loss": 7.4776,
      "step": 70870
    },
    {
      "epoch": 8.5294825511432,
      "grad_norm": 3.1052374839782715,
      "learning_rate": 0.00019689393651385438,
      "loss": 7.5736,
      "step": 70880
    },
    {
      "epoch": 8.530685920577618,
      "grad_norm": 2.4602415561676025,
      "learning_rate": 0.00019689299565383648,
      "loss": 7.5589,
      "step": 70890
    },
    {
      "epoch": 8.531889290012034,
      "grad_norm": 1.5469058752059937,
      "learning_rate": 0.0001968920546535905,
      "loss": 7.5763,
      "step": 70900
    },
    {
      "epoch": 8.53309265944645,
      "grad_norm": 1.3221018314361572,
      "learning_rate": 0.00019689111351311775,
      "loss": 7.5067,
      "step": 70910
    },
    {
      "epoch": 8.534296028880867,
      "grad_norm": 1.084427833557129,
      "learning_rate": 0.00019689017223241967,
      "loss": 7.523,
      "step": 70920
    },
    {
      "epoch": 8.535499398315283,
      "grad_norm": 2.507516622543335,
      "learning_rate": 0.00019688923081149758,
      "loss": 7.5407,
      "step": 70930
    },
    {
      "epoch": 8.536702767749698,
      "grad_norm": 1.2151966094970703,
      "learning_rate": 0.00019688828925035285,
      "loss": 7.5362,
      "step": 70940
    },
    {
      "epoch": 8.537906137184116,
      "grad_norm": 3.324131965637207,
      "learning_rate": 0.00019688734754898685,
      "loss": 7.542,
      "step": 70950
    },
    {
      "epoch": 8.539109506618532,
      "grad_norm": 2.180873394012451,
      "learning_rate": 0.00019688640570740095,
      "loss": 7.614,
      "step": 70960
    },
    {
      "epoch": 8.540312876052948,
      "grad_norm": 1.2994470596313477,
      "learning_rate": 0.0001968854637255965,
      "loss": 7.5134,
      "step": 70970
    },
    {
      "epoch": 8.541516245487365,
      "grad_norm": 2.6376659870147705,
      "learning_rate": 0.00019688452160357485,
      "loss": 7.5046,
      "step": 70980
    },
    {
      "epoch": 8.542719614921781,
      "grad_norm": 1.9583570957183838,
      "learning_rate": 0.00019688357934133737,
      "loss": 7.5359,
      "step": 70990
    },
    {
      "epoch": 8.543922984356197,
      "grad_norm": 1.505137324333191,
      "learning_rate": 0.00019688263693888543,
      "loss": 7.5449,
      "step": 71000
    },
    {
      "epoch": 8.545126353790614,
      "grad_norm": 3.4148805141448975,
      "learning_rate": 0.0001968816943962204,
      "loss": 7.5003,
      "step": 71010
    },
    {
      "epoch": 8.54632972322503,
      "grad_norm": 1.5637755393981934,
      "learning_rate": 0.00019688075171334367,
      "loss": 7.4096,
      "step": 71020
    },
    {
      "epoch": 8.547533092659446,
      "grad_norm": 4.581235885620117,
      "learning_rate": 0.00019687980889025652,
      "loss": 7.3817,
      "step": 71030
    },
    {
      "epoch": 8.548736462093864,
      "grad_norm": 1.3608286380767822,
      "learning_rate": 0.0001968788659269604,
      "loss": 7.4101,
      "step": 71040
    },
    {
      "epoch": 8.54993983152828,
      "grad_norm": 2.594306707382202,
      "learning_rate": 0.00019687792282345665,
      "loss": 7.4077,
      "step": 71050
    },
    {
      "epoch": 8.551143200962695,
      "grad_norm": 4.229619026184082,
      "learning_rate": 0.0001968769795797466,
      "loss": 7.3315,
      "step": 71060
    },
    {
      "epoch": 8.552346570397113,
      "grad_norm": 5.648379802703857,
      "learning_rate": 0.00019687603619583167,
      "loss": 7.4154,
      "step": 71070
    },
    {
      "epoch": 8.553549939831528,
      "grad_norm": 3.063415288925171,
      "learning_rate": 0.00019687509267171316,
      "loss": 7.3526,
      "step": 71080
    },
    {
      "epoch": 8.554753309265944,
      "grad_norm": 2.3927276134490967,
      "learning_rate": 0.0001968741490073925,
      "loss": 7.3678,
      "step": 71090
    },
    {
      "epoch": 8.555956678700362,
      "grad_norm": 4.117802143096924,
      "learning_rate": 0.00019687320520287105,
      "loss": 7.3824,
      "step": 71100
    },
    {
      "epoch": 8.557160048134778,
      "grad_norm": 3.0834388732910156,
      "learning_rate": 0.00019687226125815013,
      "loss": 7.3224,
      "step": 71110
    },
    {
      "epoch": 8.558363417569193,
      "grad_norm": 3.2998011112213135,
      "learning_rate": 0.00019687131717323114,
      "loss": 7.4092,
      "step": 71120
    },
    {
      "epoch": 8.559566787003611,
      "grad_norm": 3.658404588699341,
      "learning_rate": 0.0001968703729481154,
      "loss": 7.3639,
      "step": 71130
    },
    {
      "epoch": 8.560770156438027,
      "grad_norm": 2.8112010955810547,
      "learning_rate": 0.00019686942858280435,
      "loss": 7.2514,
      "step": 71140
    },
    {
      "epoch": 8.561973525872443,
      "grad_norm": 3.114164352416992,
      "learning_rate": 0.00019686848407729932,
      "loss": 7.4219,
      "step": 71150
    },
    {
      "epoch": 8.56317689530686,
      "grad_norm": 7.612882614135742,
      "learning_rate": 0.00019686753943160167,
      "loss": 7.309,
      "step": 71160
    },
    {
      "epoch": 8.564380264741276,
      "grad_norm": 3.221165418624878,
      "learning_rate": 0.00019686659464571277,
      "loss": 7.418,
      "step": 71170
    },
    {
      "epoch": 8.565583634175692,
      "grad_norm": 3.081568717956543,
      "learning_rate": 0.000196865649719634,
      "loss": 7.3208,
      "step": 71180
    },
    {
      "epoch": 8.566787003610107,
      "grad_norm": 2.8525428771972656,
      "learning_rate": 0.00019686470465336673,
      "loss": 7.2933,
      "step": 71190
    },
    {
      "epoch": 8.567990373044525,
      "grad_norm": 3.2281641960144043,
      "learning_rate": 0.00019686375944691231,
      "loss": 7.4252,
      "step": 71200
    },
    {
      "epoch": 8.56919374247894,
      "grad_norm": 2.482732057571411,
      "learning_rate": 0.0001968628141002721,
      "loss": 7.2929,
      "step": 71210
    },
    {
      "epoch": 8.570397111913357,
      "grad_norm": 2.630949020385742,
      "learning_rate": 0.0001968618686134475,
      "loss": 7.3732,
      "step": 71220
    },
    {
      "epoch": 8.571600481347774,
      "grad_norm": 3.7805721759796143,
      "learning_rate": 0.00019686092298643986,
      "loss": 7.3923,
      "step": 71230
    },
    {
      "epoch": 8.57280385078219,
      "grad_norm": 7.153574466705322,
      "learning_rate": 0.00019685997721925053,
      "loss": 7.2577,
      "step": 71240
    },
    {
      "epoch": 8.574007220216606,
      "grad_norm": 4.336424827575684,
      "learning_rate": 0.0001968590313118809,
      "loss": 7.4687,
      "step": 71250
    },
    {
      "epoch": 8.575210589651023,
      "grad_norm": 3.935218572616577,
      "learning_rate": 0.00019685808526433238,
      "loss": 7.221,
      "step": 71260
    },
    {
      "epoch": 8.57641395908544,
      "grad_norm": 7.177575588226318,
      "learning_rate": 0.00019685713907660626,
      "loss": 7.3643,
      "step": 71270
    },
    {
      "epoch": 8.577617328519855,
      "grad_norm": 4.28988790512085,
      "learning_rate": 0.00019685619274870396,
      "loss": 7.3117,
      "step": 71280
    },
    {
      "epoch": 8.578820697954272,
      "grad_norm": 5.960178375244141,
      "learning_rate": 0.00019685524628062684,
      "loss": 7.248,
      "step": 71290
    },
    {
      "epoch": 8.580024067388688,
      "grad_norm": 6.8065643310546875,
      "learning_rate": 0.00019685429967237627,
      "loss": 7.2714,
      "step": 71300
    },
    {
      "epoch": 8.581227436823104,
      "grad_norm": 7.6462788581848145,
      "learning_rate": 0.00019685335292395358,
      "loss": 7.2764,
      "step": 71310
    },
    {
      "epoch": 8.582430806257522,
      "grad_norm": 10.469433784484863,
      "learning_rate": 0.0001968524060353602,
      "loss": 7.2509,
      "step": 71320
    },
    {
      "epoch": 8.583634175691937,
      "grad_norm": 6.646592140197754,
      "learning_rate": 0.00019685145900659746,
      "loss": 7.3452,
      "step": 71330
    },
    {
      "epoch": 8.584837545126353,
      "grad_norm": 7.079794406890869,
      "learning_rate": 0.00019685051183766677,
      "loss": 7.3314,
      "step": 71340
    },
    {
      "epoch": 8.58604091456077,
      "grad_norm": 3.4195752143859863,
      "learning_rate": 0.00019684956452856945,
      "loss": 7.2803,
      "step": 71350
    },
    {
      "epoch": 8.587244283995187,
      "grad_norm": 9.062658309936523,
      "learning_rate": 0.00019684861707930696,
      "loss": 7.2947,
      "step": 71360
    },
    {
      "epoch": 8.588447653429602,
      "grad_norm": 6.66270112991333,
      "learning_rate": 0.00019684766948988055,
      "loss": 7.18,
      "step": 71370
    },
    {
      "epoch": 8.58965102286402,
      "grad_norm": 8.045583724975586,
      "learning_rate": 0.00019684672176029168,
      "loss": 7.3799,
      "step": 71380
    },
    {
      "epoch": 8.590854392298436,
      "grad_norm": 9.857762336730957,
      "learning_rate": 0.00019684577389054167,
      "loss": 7.2175,
      "step": 71390
    },
    {
      "epoch": 8.592057761732852,
      "grad_norm": 13.152425765991211,
      "learning_rate": 0.00019684482588063192,
      "loss": 7.2631,
      "step": 71400
    },
    {
      "epoch": 8.593261131167269,
      "grad_norm": 23.076932907104492,
      "learning_rate": 0.0001968438777305638,
      "loss": 7.1899,
      "step": 71410
    },
    {
      "epoch": 8.594464500601685,
      "grad_norm": 24.89419937133789,
      "learning_rate": 0.0001968429294403387,
      "loss": 7.1728,
      "step": 71420
    },
    {
      "epoch": 8.5956678700361,
      "grad_norm": 33.4057502746582,
      "learning_rate": 0.00019684198100995795,
      "loss": 7.3295,
      "step": 71430
    },
    {
      "epoch": 8.596871239470518,
      "grad_norm": 14.6361083984375,
      "learning_rate": 0.00019684103243942296,
      "loss": 7.2593,
      "step": 71440
    },
    {
      "epoch": 8.598074608904934,
      "grad_norm": 22.988183975219727,
      "learning_rate": 0.0001968400837287351,
      "loss": 7.2897,
      "step": 71450
    },
    {
      "epoch": 8.59927797833935,
      "grad_norm": 23.829151153564453,
      "learning_rate": 0.0001968391348778957,
      "loss": 7.3138,
      "step": 71460
    },
    {
      "epoch": 8.600481347773766,
      "grad_norm": 46.30461883544922,
      "learning_rate": 0.00019683818588690617,
      "loss": 7.2216,
      "step": 71470
    },
    {
      "epoch": 8.601684717208183,
      "grad_norm": 51.192527770996094,
      "learning_rate": 0.00019683723675576785,
      "loss": 7.2637,
      "step": 71480
    },
    {
      "epoch": 8.602888086642599,
      "grad_norm": 20.90011215209961,
      "learning_rate": 0.00019683628748448217,
      "loss": 7.1765,
      "step": 71490
    },
    {
      "epoch": 8.604091456077015,
      "grad_norm": 20.127248764038086,
      "learning_rate": 0.00019683533807305045,
      "loss": 7.2282,
      "step": 71500
    },
    {
      "epoch": 8.605294825511432,
      "grad_norm": 21.1735782623291,
      "learning_rate": 0.00019683438852147414,
      "loss": 7.2744,
      "step": 71510
    },
    {
      "epoch": 8.606498194945848,
      "grad_norm": 14.524378776550293,
      "learning_rate": 0.00019683343882975453,
      "loss": 7.2047,
      "step": 71520
    },
    {
      "epoch": 8.607701564380264,
      "grad_norm": 27.374757766723633,
      "learning_rate": 0.00019683248899789302,
      "loss": 7.2963,
      "step": 71530
    },
    {
      "epoch": 8.608904933814681,
      "grad_norm": 45.3835563659668,
      "learning_rate": 0.000196831539025891,
      "loss": 7.1218,
      "step": 71540
    },
    {
      "epoch": 8.610108303249097,
      "grad_norm": 30.102523803710938,
      "learning_rate": 0.00019683058891374985,
      "loss": 7.2178,
      "step": 71550
    },
    {
      "epoch": 8.611311672683513,
      "grad_norm": 398.0894470214844,
      "learning_rate": 0.00019682963866147089,
      "loss": 7.1868,
      "step": 71560
    },
    {
      "epoch": 8.61251504211793,
      "grad_norm": 55.32802200317383,
      "learning_rate": 0.00019682868826905557,
      "loss": 7.2882,
      "step": 71570
    },
    {
      "epoch": 8.613718411552346,
      "grad_norm": 56.809810638427734,
      "learning_rate": 0.00019682773773650522,
      "loss": 7.2094,
      "step": 71580
    },
    {
      "epoch": 8.614921780986762,
      "grad_norm": 52.83161926269531,
      "learning_rate": 0.00019682678706382122,
      "loss": 7.1541,
      "step": 71590
    },
    {
      "epoch": 8.61612515042118,
      "grad_norm": 46.8318977355957,
      "learning_rate": 0.00019682583625100497,
      "loss": 7.3496,
      "step": 71600
    },
    {
      "epoch": 8.617328519855596,
      "grad_norm": 92.50406646728516,
      "learning_rate": 0.00019682488529805783,
      "loss": 7.268,
      "step": 71610
    },
    {
      "epoch": 8.618531889290011,
      "grad_norm": 54.915550231933594,
      "learning_rate": 0.00019682393420498116,
      "loss": 7.2467,
      "step": 71620
    },
    {
      "epoch": 8.619735258724429,
      "grad_norm": 101.87727355957031,
      "learning_rate": 0.00019682298297177636,
      "loss": 7.2832,
      "step": 71630
    },
    {
      "epoch": 8.620938628158845,
      "grad_norm": 1934.8565673828125,
      "learning_rate": 0.00019682203159844482,
      "loss": 7.1457,
      "step": 71640
    },
    {
      "epoch": 8.62214199759326,
      "grad_norm": 86.66309356689453,
      "learning_rate": 0.00019682108008498784,
      "loss": 7.2731,
      "step": 71650
    },
    {
      "epoch": 8.623345367027678,
      "grad_norm": 125.84171295166016,
      "learning_rate": 0.0001968201284314069,
      "loss": 7.2701,
      "step": 71660
    },
    {
      "epoch": 8.624548736462094,
      "grad_norm": 69.844482421875,
      "learning_rate": 0.00019681917663770334,
      "loss": 7.3399,
      "step": 71670
    },
    {
      "epoch": 8.62575210589651,
      "grad_norm": 258.81988525390625,
      "learning_rate": 0.0001968182247038785,
      "loss": 7.3469,
      "step": 71680
    },
    {
      "epoch": 8.626955475330927,
      "grad_norm": 140.39862060546875,
      "learning_rate": 0.00019681727262993377,
      "loss": 7.2338,
      "step": 71690
    },
    {
      "epoch": 8.628158844765343,
      "grad_norm": 65.96165466308594,
      "learning_rate": 0.00019681632041587056,
      "loss": 7.2687,
      "step": 71700
    },
    {
      "epoch": 8.629362214199759,
      "grad_norm": 89.50534057617188,
      "learning_rate": 0.00019681536806169026,
      "loss": 7.2222,
      "step": 71710
    },
    {
      "epoch": 8.630565583634176,
      "grad_norm": 48.90932083129883,
      "learning_rate": 0.0001968144155673942,
      "loss": 7.2026,
      "step": 71720
    },
    {
      "epoch": 8.631768953068592,
      "grad_norm": 53.440757751464844,
      "learning_rate": 0.00019681346293298378,
      "loss": 7.3175,
      "step": 71730
    },
    {
      "epoch": 8.632972322503008,
      "grad_norm": 70.87751770019531,
      "learning_rate": 0.00019681251015846036,
      "loss": 7.293,
      "step": 71740
    },
    {
      "epoch": 8.634175691937426,
      "grad_norm": 62.420082092285156,
      "learning_rate": 0.00019681155724382535,
      "loss": 7.345,
      "step": 71750
    },
    {
      "epoch": 8.635379061371841,
      "grad_norm": 110.84393310546875,
      "learning_rate": 0.0001968106041890801,
      "loss": 7.2736,
      "step": 71760
    },
    {
      "epoch": 8.636582430806257,
      "grad_norm": 66.7536849975586,
      "learning_rate": 0.00019680965099422602,
      "loss": 7.2736,
      "step": 71770
    },
    {
      "epoch": 8.637785800240675,
      "grad_norm": 106.4581069946289,
      "learning_rate": 0.00019680869765926446,
      "loss": 7.2642,
      "step": 71780
    },
    {
      "epoch": 8.63898916967509,
      "grad_norm": 164.11788940429688,
      "learning_rate": 0.0001968077441841968,
      "loss": 7.2527,
      "step": 71790
    },
    {
      "epoch": 8.640192539109506,
      "grad_norm": 130.2183837890625,
      "learning_rate": 0.00019680679056902447,
      "loss": 7.3717,
      "step": 71800
    },
    {
      "epoch": 8.641395908543924,
      "grad_norm": 3453.31494140625,
      "learning_rate": 0.00019680583681374878,
      "loss": 7.2511,
      "step": 71810
    },
    {
      "epoch": 8.64259927797834,
      "grad_norm": 577.374755859375,
      "learning_rate": 0.00019680488291837117,
      "loss": 7.3618,
      "step": 71820
    },
    {
      "epoch": 8.643802647412755,
      "grad_norm": 172.49139404296875,
      "learning_rate": 0.00019680392888289297,
      "loss": 7.3405,
      "step": 71830
    },
    {
      "epoch": 8.645006016847173,
      "grad_norm": 82.99141693115234,
      "learning_rate": 0.0001968029747073156,
      "loss": 7.2138,
      "step": 71840
    },
    {
      "epoch": 8.646209386281589,
      "grad_norm": 97.69009399414062,
      "learning_rate": 0.00019680202039164043,
      "loss": 7.3318,
      "step": 71850
    },
    {
      "epoch": 8.647412755716005,
      "grad_norm": 138.60333251953125,
      "learning_rate": 0.00019680106593586885,
      "loss": 7.1722,
      "step": 71860
    },
    {
      "epoch": 8.648616125150422,
      "grad_norm": 101.49639892578125,
      "learning_rate": 0.0001968001113400022,
      "loss": 7.2632,
      "step": 71870
    },
    {
      "epoch": 8.649819494584838,
      "grad_norm": 165.77996826171875,
      "learning_rate": 0.0001967991566040419,
      "loss": 7.2798,
      "step": 71880
    },
    {
      "epoch": 8.651022864019254,
      "grad_norm": 126.64981079101562,
      "learning_rate": 0.00019679820172798933,
      "loss": 7.2899,
      "step": 71890
    },
    {
      "epoch": 8.65222623345367,
      "grad_norm": 103.47054290771484,
      "learning_rate": 0.00019679724671184585,
      "loss": 7.2584,
      "step": 71900
    },
    {
      "epoch": 8.653429602888087,
      "grad_norm": 120.01701354980469,
      "learning_rate": 0.00019679629155561285,
      "loss": 7.2726,
      "step": 71910
    },
    {
      "epoch": 8.654632972322503,
      "grad_norm": 50.30412673950195,
      "learning_rate": 0.00019679533625929175,
      "loss": 7.4641,
      "step": 71920
    },
    {
      "epoch": 8.655836341756919,
      "grad_norm": 157.47386169433594,
      "learning_rate": 0.00019679438082288386,
      "loss": 7.2505,
      "step": 71930
    },
    {
      "epoch": 8.657039711191336,
      "grad_norm": 155.23707580566406,
      "learning_rate": 0.00019679342524639063,
      "loss": 7.2249,
      "step": 71940
    },
    {
      "epoch": 8.658243080625752,
      "grad_norm": 224.6519317626953,
      "learning_rate": 0.0001967924695298134,
      "loss": 7.3257,
      "step": 71950
    },
    {
      "epoch": 8.659446450060168,
      "grad_norm": 211.5713653564453,
      "learning_rate": 0.00019679151367315358,
      "loss": 7.3759,
      "step": 71960
    },
    {
      "epoch": 8.660649819494585,
      "grad_norm": 126.58041381835938,
      "learning_rate": 0.00019679055767641255,
      "loss": 7.2925,
      "step": 71970
    },
    {
      "epoch": 8.661853188929001,
      "grad_norm": 227.46240234375,
      "learning_rate": 0.00019678960153959166,
      "loss": 7.2896,
      "step": 71980
    },
    {
      "epoch": 8.663056558363417,
      "grad_norm": 116.68350219726562,
      "learning_rate": 0.00019678864526269233,
      "loss": 7.3468,
      "step": 71990
    },
    {
      "epoch": 8.664259927797834,
      "grad_norm": 219.05282592773438,
      "learning_rate": 0.00019678768884571595,
      "loss": 7.2951,
      "step": 72000
    },
    {
      "epoch": 8.66546329723225,
      "grad_norm": 250.28330993652344,
      "learning_rate": 0.00019678673228866388,
      "loss": 7.3772,
      "step": 72010
    },
    {
      "epoch": 8.666666666666666,
      "grad_norm": 133.87220764160156,
      "learning_rate": 0.00019678577559153752,
      "loss": 7.2715,
      "step": 72020
    },
    {
      "epoch": 8.667870036101084,
      "grad_norm": 175.6766815185547,
      "learning_rate": 0.00019678481875433823,
      "loss": 7.3487,
      "step": 72030
    },
    {
      "epoch": 8.6690734055355,
      "grad_norm": 366.0092468261719,
      "learning_rate": 0.00019678386177706742,
      "loss": 7.4083,
      "step": 72040
    },
    {
      "epoch": 8.670276774969915,
      "grad_norm": 76.88314056396484,
      "learning_rate": 0.00019678290465972647,
      "loss": 7.352,
      "step": 72050
    },
    {
      "epoch": 8.671480144404333,
      "grad_norm": 79.90327453613281,
      "learning_rate": 0.00019678194740231674,
      "loss": 7.3391,
      "step": 72060
    },
    {
      "epoch": 8.672683513838749,
      "grad_norm": 78.71815490722656,
      "learning_rate": 0.00019678099000483966,
      "loss": 7.3185,
      "step": 72070
    },
    {
      "epoch": 8.673886883273164,
      "grad_norm": 309.57965087890625,
      "learning_rate": 0.0001967800324672966,
      "loss": 7.2959,
      "step": 72080
    },
    {
      "epoch": 8.675090252707582,
      "grad_norm": 109.93051147460938,
      "learning_rate": 0.00019677907478968894,
      "loss": 7.2478,
      "step": 72090
    },
    {
      "epoch": 8.676293622141998,
      "grad_norm": 211.07933044433594,
      "learning_rate": 0.00019677811697201803,
      "loss": 7.3607,
      "step": 72100
    },
    {
      "epoch": 8.677496991576414,
      "grad_norm": 356.20892333984375,
      "learning_rate": 0.00019677715901428533,
      "loss": 7.3882,
      "step": 72110
    },
    {
      "epoch": 8.678700361010831,
      "grad_norm": 163.8538818359375,
      "learning_rate": 0.00019677620091649217,
      "loss": 7.278,
      "step": 72120
    },
    {
      "epoch": 8.679903730445247,
      "grad_norm": 541.84228515625,
      "learning_rate": 0.00019677524267863996,
      "loss": 7.36,
      "step": 72130
    },
    {
      "epoch": 8.681107099879663,
      "grad_norm": 260.95477294921875,
      "learning_rate": 0.00019677428430073005,
      "loss": 7.2536,
      "step": 72140
    },
    {
      "epoch": 8.68231046931408,
      "grad_norm": 83.66963195800781,
      "learning_rate": 0.00019677332578276388,
      "loss": 7.4829,
      "step": 72150
    },
    {
      "epoch": 8.683513838748496,
      "grad_norm": 202.00425720214844,
      "learning_rate": 0.00019677236712474283,
      "loss": 7.3373,
      "step": 72160
    },
    {
      "epoch": 8.684717208182912,
      "grad_norm": 401.5888366699219,
      "learning_rate": 0.00019677140832666826,
      "loss": 7.3372,
      "step": 72170
    },
    {
      "epoch": 8.685920577617328,
      "grad_norm": 379.677490234375,
      "learning_rate": 0.00019677044938854157,
      "loss": 7.3094,
      "step": 72180
    },
    {
      "epoch": 8.687123947051745,
      "grad_norm": 384.5784912109375,
      "learning_rate": 0.00019676949031036415,
      "loss": 7.369,
      "step": 72190
    },
    {
      "epoch": 8.688327316486161,
      "grad_norm": 180.51034545898438,
      "learning_rate": 0.0001967685310921374,
      "loss": 7.3727,
      "step": 72200
    },
    {
      "epoch": 8.689530685920577,
      "grad_norm": 93.56525421142578,
      "learning_rate": 0.00019676757173386265,
      "loss": 7.3642,
      "step": 72210
    },
    {
      "epoch": 8.690734055354994,
      "grad_norm": 165.45242309570312,
      "learning_rate": 0.0001967666122355414,
      "loss": 7.4359,
      "step": 72220
    },
    {
      "epoch": 8.69193742478941,
      "grad_norm": 252.12612915039062,
      "learning_rate": 0.00019676565259717493,
      "loss": 7.3754,
      "step": 72230
    },
    {
      "epoch": 8.693140794223826,
      "grad_norm": 267.654541015625,
      "learning_rate": 0.00019676469281876467,
      "loss": 7.3834,
      "step": 72240
    },
    {
      "epoch": 8.694344163658243,
      "grad_norm": 359.25933837890625,
      "learning_rate": 0.00019676373290031197,
      "loss": 7.3656,
      "step": 72250
    },
    {
      "epoch": 8.69554753309266,
      "grad_norm": 960.4390869140625,
      "learning_rate": 0.00019676277284181833,
      "loss": 7.5518,
      "step": 72260
    },
    {
      "epoch": 8.696750902527075,
      "grad_norm": 373.3681640625,
      "learning_rate": 0.00019676181264328504,
      "loss": 7.3583,
      "step": 72270
    },
    {
      "epoch": 8.697954271961493,
      "grad_norm": 796.1593627929688,
      "learning_rate": 0.0001967608523047135,
      "loss": 7.467,
      "step": 72280
    },
    {
      "epoch": 8.699157641395908,
      "grad_norm": 479.3305358886719,
      "learning_rate": 0.00019675989182610514,
      "loss": 7.5587,
      "step": 72290
    },
    {
      "epoch": 8.700361010830324,
      "grad_norm": 585.9391479492188,
      "learning_rate": 0.00019675893120746132,
      "loss": 7.4181,
      "step": 72300
    },
    {
      "epoch": 8.701564380264742,
      "grad_norm": 971.416748046875,
      "learning_rate": 0.00019675797044878344,
      "loss": 7.4307,
      "step": 72310
    },
    {
      "epoch": 8.702767749699158,
      "grad_norm": 298.24786376953125,
      "learning_rate": 0.00019675700955007288,
      "loss": 7.416,
      "step": 72320
    },
    {
      "epoch": 8.703971119133573,
      "grad_norm": 321.9739990234375,
      "learning_rate": 0.00019675604851133105,
      "loss": 7.3523,
      "step": 72330
    },
    {
      "epoch": 8.705174488567991,
      "grad_norm": 280.7763366699219,
      "learning_rate": 0.00019675508733255933,
      "loss": 7.4245,
      "step": 72340
    },
    {
      "epoch": 8.706377858002407,
      "grad_norm": 551.8917236328125,
      "learning_rate": 0.0001967541260137591,
      "loss": 7.4216,
      "step": 72350
    },
    {
      "epoch": 8.707581227436823,
      "grad_norm": 135.53799438476562,
      "learning_rate": 0.00019675316455493175,
      "loss": 7.4067,
      "step": 72360
    },
    {
      "epoch": 8.70878459687124,
      "grad_norm": 256.7081604003906,
      "learning_rate": 0.0001967522029560787,
      "loss": 7.4115,
      "step": 72370
    },
    {
      "epoch": 8.709987966305656,
      "grad_norm": 167.232177734375,
      "learning_rate": 0.00019675124121720132,
      "loss": 7.412,
      "step": 72380
    },
    {
      "epoch": 8.711191335740072,
      "grad_norm": 239.28404235839844,
      "learning_rate": 0.000196750279338301,
      "loss": 7.3593,
      "step": 72390
    },
    {
      "epoch": 8.71239470517449,
      "grad_norm": 119.85040283203125,
      "learning_rate": 0.00019674931731937916,
      "loss": 7.4017,
      "step": 72400
    },
    {
      "epoch": 8.713598074608905,
      "grad_norm": 196.09478759765625,
      "learning_rate": 0.00019674835516043717,
      "loss": 7.3264,
      "step": 72410
    },
    {
      "epoch": 8.71480144404332,
      "grad_norm": 523.9920043945312,
      "learning_rate": 0.00019674739286147644,
      "loss": 7.3477,
      "step": 72420
    },
    {
      "epoch": 8.716004813477738,
      "grad_norm": 475.8111877441406,
      "learning_rate": 0.0001967464304224983,
      "loss": 7.4119,
      "step": 72430
    },
    {
      "epoch": 8.717208182912154,
      "grad_norm": 76.03440856933594,
      "learning_rate": 0.0001967454678435042,
      "loss": 7.2692,
      "step": 72440
    },
    {
      "epoch": 8.71841155234657,
      "grad_norm": 122.36022186279297,
      "learning_rate": 0.00019674450512449553,
      "loss": 7.3012,
      "step": 72450
    },
    {
      "epoch": 8.719614921780988,
      "grad_norm": 89.64482879638672,
      "learning_rate": 0.0001967435422654737,
      "loss": 7.2409,
      "step": 72460
    },
    {
      "epoch": 8.720818291215403,
      "grad_norm": 41.002525329589844,
      "learning_rate": 0.00019674257926644004,
      "loss": 7.239,
      "step": 72470
    },
    {
      "epoch": 8.722021660649819,
      "grad_norm": 114.916259765625,
      "learning_rate": 0.00019674161612739602,
      "loss": 7.3411,
      "step": 72480
    },
    {
      "epoch": 8.723225030084237,
      "grad_norm": 131.14639282226562,
      "learning_rate": 0.00019674065284834295,
      "loss": 7.3203,
      "step": 72490
    },
    {
      "epoch": 8.724428399518652,
      "grad_norm": 40.085140228271484,
      "learning_rate": 0.00019673968942928233,
      "loss": 7.2774,
      "step": 72500
    },
    {
      "epoch": 8.725631768953068,
      "grad_norm": 69.3632583618164,
      "learning_rate": 0.00019673872587021546,
      "loss": 7.2698,
      "step": 72510
    },
    {
      "epoch": 8.726835138387486,
      "grad_norm": 52.196964263916016,
      "learning_rate": 0.00019673776217114374,
      "loss": 7.2324,
      "step": 72520
    },
    {
      "epoch": 8.728038507821902,
      "grad_norm": 99.98628234863281,
      "learning_rate": 0.00019673679833206865,
      "loss": 7.3338,
      "step": 72530
    },
    {
      "epoch": 8.729241877256317,
      "grad_norm": 90.21719360351562,
      "learning_rate": 0.0001967358343529915,
      "loss": 7.2763,
      "step": 72540
    },
    {
      "epoch": 8.730445246690735,
      "grad_norm": 56.11993408203125,
      "learning_rate": 0.00019673487023391372,
      "loss": 7.191,
      "step": 72550
    },
    {
      "epoch": 8.73164861612515,
      "grad_norm": 102.80853271484375,
      "learning_rate": 0.0001967339059748367,
      "loss": 7.2479,
      "step": 72560
    },
    {
      "epoch": 8.732851985559567,
      "grad_norm": 64.7136001586914,
      "learning_rate": 0.00019673294157576183,
      "loss": 7.1636,
      "step": 72570
    },
    {
      "epoch": 8.734055354993982,
      "grad_norm": 85.17577362060547,
      "learning_rate": 0.00019673197703669052,
      "loss": 7.2703,
      "step": 72580
    },
    {
      "epoch": 8.7352587244284,
      "grad_norm": 34.33113098144531,
      "learning_rate": 0.00019673101235762416,
      "loss": 7.2237,
      "step": 72590
    },
    {
      "epoch": 8.736462093862816,
      "grad_norm": 11.690998077392578,
      "learning_rate": 0.00019673004753856414,
      "loss": 7.2095,
      "step": 72600
    },
    {
      "epoch": 8.737665463297231,
      "grad_norm": 35.17302703857422,
      "learning_rate": 0.00019672908257951184,
      "loss": 7.1793,
      "step": 72610
    },
    {
      "epoch": 8.738868832731649,
      "grad_norm": 25.318931579589844,
      "learning_rate": 0.0001967281174804687,
      "loss": 7.2641,
      "step": 72620
    },
    {
      "epoch": 8.740072202166065,
      "grad_norm": 84.61393737792969,
      "learning_rate": 0.0001967271522414361,
      "loss": 7.3209,
      "step": 72630
    },
    {
      "epoch": 8.74127557160048,
      "grad_norm": 69.09017181396484,
      "learning_rate": 0.0001967261868624154,
      "loss": 7.2556,
      "step": 72640
    },
    {
      "epoch": 8.742478941034898,
      "grad_norm": 48.37390899658203,
      "learning_rate": 0.00019672522134340806,
      "loss": 7.2947,
      "step": 72650
    },
    {
      "epoch": 8.743682310469314,
      "grad_norm": 45.822628021240234,
      "learning_rate": 0.0001967242556844154,
      "loss": 7.2985,
      "step": 72660
    },
    {
      "epoch": 8.74488567990373,
      "grad_norm": 247.39044189453125,
      "learning_rate": 0.0001967232898854389,
      "loss": 7.2706,
      "step": 72670
    },
    {
      "epoch": 8.746089049338147,
      "grad_norm": 81.0790023803711,
      "learning_rate": 0.0001967223239464799,
      "loss": 7.3618,
      "step": 72680
    },
    {
      "epoch": 8.747292418772563,
      "grad_norm": 58.9289665222168,
      "learning_rate": 0.0001967213578675398,
      "loss": 7.2608,
      "step": 72690
    },
    {
      "epoch": 8.748495788206979,
      "grad_norm": 72.7082290649414,
      "learning_rate": 0.00019672039164862003,
      "loss": 7.1639,
      "step": 72700
    },
    {
      "epoch": 8.749699157641396,
      "grad_norm": 51.148956298828125,
      "learning_rate": 0.00019671942528972202,
      "loss": 7.1739,
      "step": 72710
    },
    {
      "epoch": 8.750902527075812,
      "grad_norm": 51.62891387939453,
      "learning_rate": 0.00019671845879084708,
      "loss": 7.2465,
      "step": 72720
    },
    {
      "epoch": 8.752105896510228,
      "grad_norm": 109.6996841430664,
      "learning_rate": 0.00019671749215199665,
      "loss": 7.2561,
      "step": 72730
    },
    {
      "epoch": 8.753309265944646,
      "grad_norm": 97.04100036621094,
      "learning_rate": 0.00019671652537317212,
      "loss": 7.2702,
      "step": 72740
    },
    {
      "epoch": 8.754512635379061,
      "grad_norm": 63.299251556396484,
      "learning_rate": 0.00019671555845437492,
      "loss": 7.255,
      "step": 72750
    },
    {
      "epoch": 8.755716004813477,
      "grad_norm": 119.66692352294922,
      "learning_rate": 0.0001967145913956064,
      "loss": 7.2322,
      "step": 72760
    },
    {
      "epoch": 8.756919374247895,
      "grad_norm": 52.720359802246094,
      "learning_rate": 0.00019671362419686802,
      "loss": 7.2204,
      "step": 72770
    },
    {
      "epoch": 8.75812274368231,
      "grad_norm": 166.690185546875,
      "learning_rate": 0.00019671265685816116,
      "loss": 7.3432,
      "step": 72780
    },
    {
      "epoch": 8.759326113116726,
      "grad_norm": 54.056705474853516,
      "learning_rate": 0.0001967116893794872,
      "loss": 7.1537,
      "step": 72790
    },
    {
      "epoch": 8.760529482551144,
      "grad_norm": 35.2452278137207,
      "learning_rate": 0.00019671072176084752,
      "loss": 7.2601,
      "step": 72800
    },
    {
      "epoch": 8.76173285198556,
      "grad_norm": 86.97594451904297,
      "learning_rate": 0.00019670975400224355,
      "loss": 7.1536,
      "step": 72810
    },
    {
      "epoch": 8.762936221419976,
      "grad_norm": 84.7102279663086,
      "learning_rate": 0.0001967087861036767,
      "loss": 7.1861,
      "step": 72820
    },
    {
      "epoch": 8.764139590854393,
      "grad_norm": 84.19380950927734,
      "learning_rate": 0.00019670781806514838,
      "loss": 7.3394,
      "step": 72830
    },
    {
      "epoch": 8.765342960288809,
      "grad_norm": 125.45092010498047,
      "learning_rate": 0.00019670684988665996,
      "loss": 7.1695,
      "step": 72840
    },
    {
      "epoch": 8.766546329723225,
      "grad_norm": 88.8373794555664,
      "learning_rate": 0.00019670588156821283,
      "loss": 7.2827,
      "step": 72850
    },
    {
      "epoch": 8.76774969915764,
      "grad_norm": 135.4306182861328,
      "learning_rate": 0.00019670491310980843,
      "loss": 7.2219,
      "step": 72860
    },
    {
      "epoch": 8.768953068592058,
      "grad_norm": 83.65979766845703,
      "learning_rate": 0.00019670394451144817,
      "loss": 7.191,
      "step": 72870
    },
    {
      "epoch": 8.770156438026474,
      "grad_norm": 116.87435913085938,
      "learning_rate": 0.0001967029757731334,
      "loss": 7.273,
      "step": 72880
    },
    {
      "epoch": 8.77135980746089,
      "grad_norm": 132.07652282714844,
      "learning_rate": 0.00019670200689486553,
      "loss": 7.2315,
      "step": 72890
    },
    {
      "epoch": 8.772563176895307,
      "grad_norm": 102.33387756347656,
      "learning_rate": 0.00019670103787664602,
      "loss": 7.3263,
      "step": 72900
    },
    {
      "epoch": 8.773766546329723,
      "grad_norm": 267.4897155761719,
      "learning_rate": 0.0001967000687184762,
      "loss": 7.269,
      "step": 72910
    },
    {
      "epoch": 8.774969915764139,
      "grad_norm": 113.48148345947266,
      "learning_rate": 0.00019669909942035752,
      "loss": 7.2553,
      "step": 72920
    },
    {
      "epoch": 8.776173285198556,
      "grad_norm": 244.1143798828125,
      "learning_rate": 0.00019669812998229138,
      "loss": 7.3231,
      "step": 72930
    },
    {
      "epoch": 8.777376654632972,
      "grad_norm": 165.69198608398438,
      "learning_rate": 0.00019669716040427915,
      "loss": 7.3271,
      "step": 72940
    },
    {
      "epoch": 8.778580024067388,
      "grad_norm": 140.86488342285156,
      "learning_rate": 0.00019669619068632228,
      "loss": 7.1785,
      "step": 72950
    },
    {
      "epoch": 8.779783393501805,
      "grad_norm": 232.5905303955078,
      "learning_rate": 0.0001966952208284221,
      "loss": 7.2789,
      "step": 72960
    },
    {
      "epoch": 8.780986762936221,
      "grad_norm": 151.40029907226562,
      "learning_rate": 0.0001966942508305801,
      "loss": 7.3207,
      "step": 72970
    },
    {
      "epoch": 8.782190132370637,
      "grad_norm": 425.86578369140625,
      "learning_rate": 0.00019669328069279765,
      "loss": 7.3771,
      "step": 72980
    },
    {
      "epoch": 8.783393501805055,
      "grad_norm": 977.79296875,
      "learning_rate": 0.00019669231041507615,
      "loss": 7.3079,
      "step": 72990
    },
    {
      "epoch": 8.78459687123947,
      "grad_norm": 410.035888671875,
      "learning_rate": 0.000196691339997417,
      "loss": 7.3176,
      "step": 73000
    },
    {
      "epoch": 8.785800240673886,
      "grad_norm": 10397.0966796875,
      "learning_rate": 0.00019669036943982163,
      "loss": 7.4211,
      "step": 73010
    },
    {
      "epoch": 8.787003610108304,
      "grad_norm": 72771.4921875,
      "learning_rate": 0.0001966893987422914,
      "loss": 7.6,
      "step": 73020
    },
    {
      "epoch": 8.78820697954272,
      "grad_norm": 440206.6875,
      "learning_rate": 0.00019668842790482774,
      "loss": 7.4552,
      "step": 73030
    },
    {
      "epoch": 8.789410348977135,
      "grad_norm": 180011.390625,
      "learning_rate": 0.00019668745692743206,
      "loss": 7.6077,
      "step": 73040
    },
    {
      "epoch": 8.790613718411553,
      "grad_norm": 134450.203125,
      "learning_rate": 0.00019668648581010572,
      "loss": 7.5985,
      "step": 73050
    },
    {
      "epoch": 8.791817087845969,
      "grad_norm": 611143.25,
      "learning_rate": 0.00019668551455285022,
      "loss": 7.807,
      "step": 73060
    },
    {
      "epoch": 8.793020457280385,
      "grad_norm": 46172.26953125,
      "learning_rate": 0.00019668454315566685,
      "loss": 7.7018,
      "step": 73070
    },
    {
      "epoch": 8.794223826714802,
      "grad_norm": 94033.984375,
      "learning_rate": 0.0001966835716185571,
      "loss": 7.8388,
      "step": 73080
    },
    {
      "epoch": 8.795427196149218,
      "grad_norm": 73787.5625,
      "learning_rate": 0.00019668259994152235,
      "loss": 7.9338,
      "step": 73090
    },
    {
      "epoch": 8.796630565583634,
      "grad_norm": 77311.5546875,
      "learning_rate": 0.00019668162812456404,
      "loss": 7.9655,
      "step": 73100
    },
    {
      "epoch": 8.797833935018051,
      "grad_norm": 66708.921875,
      "learning_rate": 0.00019668065616768355,
      "loss": 8.0024,
      "step": 73110
    },
    {
      "epoch": 8.799037304452467,
      "grad_norm": 77981.0390625,
      "learning_rate": 0.00019667968407088222,
      "loss": 7.9471,
      "step": 73120
    },
    {
      "epoch": 8.800240673886883,
      "grad_norm": 42229.23828125,
      "learning_rate": 0.00019667871183416156,
      "loss": 7.7234,
      "step": 73130
    },
    {
      "epoch": 8.8014440433213,
      "grad_norm": 74030.171875,
      "learning_rate": 0.00019667773945752294,
      "loss": 7.6446,
      "step": 73140
    },
    {
      "epoch": 8.802647412755716,
      "grad_norm": 42493.125,
      "learning_rate": 0.00019667676694096775,
      "loss": 7.5837,
      "step": 73150
    },
    {
      "epoch": 8.803850782190132,
      "grad_norm": 113942.7265625,
      "learning_rate": 0.0001966757942844974,
      "loss": 7.6676,
      "step": 73160
    },
    {
      "epoch": 8.80505415162455,
      "grad_norm": 301.4753723144531,
      "learning_rate": 0.0001966748214881133,
      "loss": 7.4604,
      "step": 73170
    },
    {
      "epoch": 8.806257521058965,
      "grad_norm": 209.85147094726562,
      "learning_rate": 0.00019667384855181688,
      "loss": 7.5237,
      "step": 73180
    },
    {
      "epoch": 8.807460890493381,
      "grad_norm": 152.94668579101562,
      "learning_rate": 0.00019667287547560952,
      "loss": 7.2736,
      "step": 73190
    },
    {
      "epoch": 8.808664259927799,
      "grad_norm": 568.5696411132812,
      "learning_rate": 0.0001966719022594927,
      "loss": 7.504,
      "step": 73200
    },
    {
      "epoch": 8.809867629362214,
      "grad_norm": 154.757568359375,
      "learning_rate": 0.0001966709289034677,
      "loss": 7.3922,
      "step": 73210
    },
    {
      "epoch": 8.81107099879663,
      "grad_norm": 198.43405151367188,
      "learning_rate": 0.000196669955407536,
      "loss": 7.3032,
      "step": 73220
    },
    {
      "epoch": 8.812274368231048,
      "grad_norm": 1810.9339599609375,
      "learning_rate": 0.00019666898177169903,
      "loss": 7.3786,
      "step": 73230
    },
    {
      "epoch": 8.813477737665464,
      "grad_norm": 4194.4169921875,
      "learning_rate": 0.0001966680079959582,
      "loss": 7.5287,
      "step": 73240
    },
    {
      "epoch": 8.81468110709988,
      "grad_norm": 310.0009460449219,
      "learning_rate": 0.00019666703408031486,
      "loss": 7.4477,
      "step": 73250
    },
    {
      "epoch": 8.815884476534297,
      "grad_norm": 145.40956115722656,
      "learning_rate": 0.00019666606002477046,
      "loss": 7.3341,
      "step": 73260
    },
    {
      "epoch": 8.817087845968713,
      "grad_norm": 63.58095169067383,
      "learning_rate": 0.00019666508582932642,
      "loss": 7.411,
      "step": 73270
    },
    {
      "epoch": 8.818291215403129,
      "grad_norm": 148.66078186035156,
      "learning_rate": 0.00019666411149398414,
      "loss": 7.351,
      "step": 73280
    },
    {
      "epoch": 8.819494584837544,
      "grad_norm": 200.17071533203125,
      "learning_rate": 0.000196663137018745,
      "loss": 7.2373,
      "step": 73290
    },
    {
      "epoch": 8.820697954271962,
      "grad_norm": 148.67649841308594,
      "learning_rate": 0.00019666216240361045,
      "loss": 7.2577,
      "step": 73300
    },
    {
      "epoch": 8.821901323706378,
      "grad_norm": 254.94749450683594,
      "learning_rate": 0.0001966611876485819,
      "loss": 7.4477,
      "step": 73310
    },
    {
      "epoch": 8.823104693140793,
      "grad_norm": 355.9664306640625,
      "learning_rate": 0.00019666021275366072,
      "loss": 7.3074,
      "step": 73320
    },
    {
      "epoch": 8.824308062575211,
      "grad_norm": 1004.6361694335938,
      "learning_rate": 0.00019665923771884837,
      "loss": 7.3727,
      "step": 73330
    },
    {
      "epoch": 8.825511432009627,
      "grad_norm": 292.1346435546875,
      "learning_rate": 0.0001966582625441462,
      "loss": 7.2991,
      "step": 73340
    },
    {
      "epoch": 8.826714801444043,
      "grad_norm": 318.120361328125,
      "learning_rate": 0.0001966572872295557,
      "loss": 7.4043,
      "step": 73350
    },
    {
      "epoch": 8.82791817087846,
      "grad_norm": 274.3759460449219,
      "learning_rate": 0.00019665631177507824,
      "loss": 7.2219,
      "step": 73360
    },
    {
      "epoch": 8.829121540312876,
      "grad_norm": 264.17401123046875,
      "learning_rate": 0.00019665533618071522,
      "loss": 7.433,
      "step": 73370
    },
    {
      "epoch": 8.830324909747292,
      "grad_norm": 420.0724182128906,
      "learning_rate": 0.00019665436044646805,
      "loss": 7.291,
      "step": 73380
    },
    {
      "epoch": 8.83152827918171,
      "grad_norm": 235.51654052734375,
      "learning_rate": 0.00019665338457233817,
      "loss": 7.3185,
      "step": 73390
    },
    {
      "epoch": 8.832731648616125,
      "grad_norm": 459.8696594238281,
      "learning_rate": 0.000196652408558327,
      "loss": 7.3206,
      "step": 73400
    },
    {
      "epoch": 8.833935018050541,
      "grad_norm": 1089.8634033203125,
      "learning_rate": 0.0001966514324044359,
      "loss": 7.3873,
      "step": 73410
    },
    {
      "epoch": 8.835138387484958,
      "grad_norm": 503.56817626953125,
      "learning_rate": 0.00019665045611066634,
      "loss": 7.462,
      "step": 73420
    },
    {
      "epoch": 8.836341756919374,
      "grad_norm": 1028.46923828125,
      "learning_rate": 0.0001966494796770197,
      "loss": 7.5007,
      "step": 73430
    },
    {
      "epoch": 8.83754512635379,
      "grad_norm": 1281.8826904296875,
      "learning_rate": 0.00019664850310349737,
      "loss": 7.5903,
      "step": 73440
    },
    {
      "epoch": 8.838748495788208,
      "grad_norm": 577.5121459960938,
      "learning_rate": 0.00019664752639010085,
      "loss": 7.617,
      "step": 73450
    },
    {
      "epoch": 8.839951865222623,
      "grad_norm": 695.2171020507812,
      "learning_rate": 0.00019664654953683146,
      "loss": 7.4535,
      "step": 73460
    },
    {
      "epoch": 8.84115523465704,
      "grad_norm": 687.0921020507812,
      "learning_rate": 0.00019664557254369067,
      "loss": 7.524,
      "step": 73470
    },
    {
      "epoch": 8.842358604091457,
      "grad_norm": 1461.2998046875,
      "learning_rate": 0.00019664459541067988,
      "loss": 7.5276,
      "step": 73480
    },
    {
      "epoch": 8.843561973525873,
      "grad_norm": 1477.6116943359375,
      "learning_rate": 0.00019664361813780045,
      "loss": 7.4721,
      "step": 73490
    },
    {
      "epoch": 8.844765342960288,
      "grad_norm": 384.4921875,
      "learning_rate": 0.0001966426407250539,
      "loss": 7.7023,
      "step": 73500
    },
    {
      "epoch": 8.845968712394706,
      "grad_norm": 439.4667663574219,
      "learning_rate": 0.00019664166317244158,
      "loss": 7.5249,
      "step": 73510
    },
    {
      "epoch": 8.847172081829122,
      "grad_norm": 439.1540832519531,
      "learning_rate": 0.0001966406854799649,
      "loss": 7.5037,
      "step": 73520
    },
    {
      "epoch": 8.848375451263538,
      "grad_norm": 309.0791015625,
      "learning_rate": 0.00019663970764762528,
      "loss": 7.5758,
      "step": 73530
    },
    {
      "epoch": 8.849578820697955,
      "grad_norm": 356.8844909667969,
      "learning_rate": 0.00019663872967542415,
      "loss": 7.4662,
      "step": 73540
    },
    {
      "epoch": 8.85078219013237,
      "grad_norm": 275.45648193359375,
      "learning_rate": 0.00019663775156336292,
      "loss": 7.5084,
      "step": 73550
    },
    {
      "epoch": 8.851985559566787,
      "grad_norm": 379.85504150390625,
      "learning_rate": 0.00019663677331144302,
      "loss": 7.5011,
      "step": 73560
    },
    {
      "epoch": 8.853188929001202,
      "grad_norm": 300.57647705078125,
      "learning_rate": 0.00019663579491966584,
      "loss": 7.45,
      "step": 73570
    },
    {
      "epoch": 8.85439229843562,
      "grad_norm": 642.899169921875,
      "learning_rate": 0.0001966348163880328,
      "loss": 7.5217,
      "step": 73580
    },
    {
      "epoch": 8.855595667870036,
      "grad_norm": 531.8223876953125,
      "learning_rate": 0.00019663383771654534,
      "loss": 7.3338,
      "step": 73590
    },
    {
      "epoch": 8.856799037304452,
      "grad_norm": 331.4888916015625,
      "learning_rate": 0.00019663285890520485,
      "loss": 7.4976,
      "step": 73600
    },
    {
      "epoch": 8.85800240673887,
      "grad_norm": 1119.4039306640625,
      "learning_rate": 0.00019663187995401274,
      "loss": 7.4766,
      "step": 73610
    },
    {
      "epoch": 8.859205776173285,
      "grad_norm": 741.74267578125,
      "learning_rate": 0.00019663090086297045,
      "loss": 7.3044,
      "step": 73620
    },
    {
      "epoch": 8.8604091456077,
      "grad_norm": 772.233642578125,
      "learning_rate": 0.00019662992163207941,
      "loss": 7.3415,
      "step": 73630
    },
    {
      "epoch": 8.861612515042118,
      "grad_norm": 451.1818542480469,
      "learning_rate": 0.000196628942261341,
      "loss": 7.3441,
      "step": 73640
    },
    {
      "epoch": 8.862815884476534,
      "grad_norm": 143.37416076660156,
      "learning_rate": 0.00019662796275075667,
      "loss": 7.2599,
      "step": 73650
    },
    {
      "epoch": 8.86401925391095,
      "grad_norm": 178.45677185058594,
      "learning_rate": 0.0001966269831003278,
      "loss": 7.3787,
      "step": 73660
    },
    {
      "epoch": 8.865222623345367,
      "grad_norm": 147.97024536132812,
      "learning_rate": 0.0001966260033100558,
      "loss": 7.438,
      "step": 73670
    },
    {
      "epoch": 8.866425992779783,
      "grad_norm": 420.7137756347656,
      "learning_rate": 0.00019662502337994216,
      "loss": 7.4596,
      "step": 73680
    },
    {
      "epoch": 8.867629362214199,
      "grad_norm": 453.4184265136719,
      "learning_rate": 0.00019662404330998823,
      "loss": 7.3101,
      "step": 73690
    },
    {
      "epoch": 8.868832731648617,
      "grad_norm": 302.7304382324219,
      "learning_rate": 0.00019662306310019547,
      "loss": 7.2601,
      "step": 73700
    },
    {
      "epoch": 8.870036101083032,
      "grad_norm": 398.3207092285156,
      "learning_rate": 0.00019662208275056527,
      "loss": 7.3412,
      "step": 73710
    },
    {
      "epoch": 8.871239470517448,
      "grad_norm": 319.54083251953125,
      "learning_rate": 0.0001966211022610991,
      "loss": 7.3469,
      "step": 73720
    },
    {
      "epoch": 8.872442839951866,
      "grad_norm": 407.7167663574219,
      "learning_rate": 0.0001966201216317983,
      "loss": 7.3415,
      "step": 73730
    },
    {
      "epoch": 8.873646209386282,
      "grad_norm": 118.96154022216797,
      "learning_rate": 0.00019661914086266434,
      "loss": 7.316,
      "step": 73740
    },
    {
      "epoch": 8.874849578820697,
      "grad_norm": 111.42121887207031,
      "learning_rate": 0.00019661815995369864,
      "loss": 7.2347,
      "step": 73750
    },
    {
      "epoch": 8.876052948255115,
      "grad_norm": 199.5009307861328,
      "learning_rate": 0.0001966171789049026,
      "loss": 7.2554,
      "step": 73760
    },
    {
      "epoch": 8.87725631768953,
      "grad_norm": 349.4769287109375,
      "learning_rate": 0.00019661619771627764,
      "loss": 7.4234,
      "step": 73770
    },
    {
      "epoch": 8.878459687123947,
      "grad_norm": 535.7664184570312,
      "learning_rate": 0.0001966152163878252,
      "loss": 7.3354,
      "step": 73780
    },
    {
      "epoch": 8.879663056558364,
      "grad_norm": 404.7473449707031,
      "learning_rate": 0.00019661423491954666,
      "loss": 7.2791,
      "step": 73790
    },
    {
      "epoch": 8.88086642599278,
      "grad_norm": 498.8824768066406,
      "learning_rate": 0.00019661325331144349,
      "loss": 7.2278,
      "step": 73800
    },
    {
      "epoch": 8.882069795427196,
      "grad_norm": 1814.0084228515625,
      "learning_rate": 0.00019661227156351712,
      "loss": 7.3612,
      "step": 73810
    },
    {
      "epoch": 8.883273164861613,
      "grad_norm": 280.3747253417969,
      "learning_rate": 0.0001966112896757689,
      "loss": 7.2753,
      "step": 73820
    },
    {
      "epoch": 8.884476534296029,
      "grad_norm": 771.80810546875,
      "learning_rate": 0.0001966103076482003,
      "loss": 7.3609,
      "step": 73830
    },
    {
      "epoch": 8.885679903730445,
      "grad_norm": 454.1147766113281,
      "learning_rate": 0.00019660932548081275,
      "loss": 7.3298,
      "step": 73840
    },
    {
      "epoch": 8.886883273164862,
      "grad_norm": 322.5664367675781,
      "learning_rate": 0.0001966083431736076,
      "loss": 7.2804,
      "step": 73850
    },
    {
      "epoch": 8.888086642599278,
      "grad_norm": 793.1326293945312,
      "learning_rate": 0.0001966073607265864,
      "loss": 7.2871,
      "step": 73860
    },
    {
      "epoch": 8.889290012033694,
      "grad_norm": 383.89630126953125,
      "learning_rate": 0.00019660637813975044,
      "loss": 7.2863,
      "step": 73870
    },
    {
      "epoch": 8.890493381468112,
      "grad_norm": 704.010498046875,
      "learning_rate": 0.00019660539541310122,
      "loss": 7.3659,
      "step": 73880
    },
    {
      "epoch": 8.891696750902527,
      "grad_norm": 416.768798828125,
      "learning_rate": 0.00019660441254664015,
      "loss": 7.3856,
      "step": 73890
    },
    {
      "epoch": 8.892900120336943,
      "grad_norm": 987.7268676757812,
      "learning_rate": 0.00019660342954036863,
      "loss": 7.2553,
      "step": 73900
    },
    {
      "epoch": 8.89410348977136,
      "grad_norm": 310.909423828125,
      "learning_rate": 0.0001966024463942881,
      "loss": 7.3154,
      "step": 73910
    },
    {
      "epoch": 8.895306859205776,
      "grad_norm": 2970.412841796875,
      "learning_rate": 0.0001966014631084,
      "loss": 7.3766,
      "step": 73920
    },
    {
      "epoch": 8.896510228640192,
      "grad_norm": 28545.12890625,
      "learning_rate": 0.00019660047968270573,
      "loss": 7.8155,
      "step": 73930
    },
    {
      "epoch": 8.89771359807461,
      "grad_norm": 26139.6875,
      "learning_rate": 0.00019659949611720666,
      "loss": 8.5855,
      "step": 73940
    },
    {
      "epoch": 8.898916967509026,
      "grad_norm": 15616.185546875,
      "learning_rate": 0.00019659851241190433,
      "loss": 8.8785,
      "step": 73950
    },
    {
      "epoch": 8.900120336943441,
      "grad_norm": 13234.80859375,
      "learning_rate": 0.0001965975285668001,
      "loss": 8.0822,
      "step": 73960
    },
    {
      "epoch": 8.901323706377857,
      "grad_norm": 3385.303955078125,
      "learning_rate": 0.00019659654458189537,
      "loss": 7.8559,
      "step": 73970
    },
    {
      "epoch": 8.902527075812275,
      "grad_norm": 4327.6064453125,
      "learning_rate": 0.0001965955604571916,
      "loss": 7.7036,
      "step": 73980
    },
    {
      "epoch": 8.90373044524669,
      "grad_norm": 4876.15087890625,
      "learning_rate": 0.0001965945761926902,
      "loss": 7.644,
      "step": 73990
    },
    {
      "epoch": 8.904933814681106,
      "grad_norm": 12927.3759765625,
      "learning_rate": 0.00019659359178839263,
      "loss": 7.9037,
      "step": 74000
    },
    {
      "epoch": 8.906137184115524,
      "grad_norm": 18909.4921875,
      "learning_rate": 0.00019659260724430024,
      "loss": 7.592,
      "step": 74010
    },
    {
      "epoch": 8.90734055354994,
      "grad_norm": 3630.268798828125,
      "learning_rate": 0.00019659162256041452,
      "loss": 7.5257,
      "step": 74020
    },
    {
      "epoch": 8.908543922984355,
      "grad_norm": 5463.34375,
      "learning_rate": 0.0001965906377367369,
      "loss": 7.4377,
      "step": 74030
    },
    {
      "epoch": 8.909747292418773,
      "grad_norm": 2743.06494140625,
      "learning_rate": 0.00019658965277326876,
      "loss": 7.3535,
      "step": 74040
    },
    {
      "epoch": 8.910950661853189,
      "grad_norm": 7481.822265625,
      "learning_rate": 0.00019658866767001155,
      "loss": 7.4033,
      "step": 74050
    },
    {
      "epoch": 8.912154031287605,
      "grad_norm": 1642.2529296875,
      "learning_rate": 0.00019658768242696665,
      "loss": 7.3684,
      "step": 74060
    },
    {
      "epoch": 8.913357400722022,
      "grad_norm": 755.1593017578125,
      "learning_rate": 0.00019658669704413558,
      "loss": 7.2633,
      "step": 74070
    },
    {
      "epoch": 8.914560770156438,
      "grad_norm": 4966.22705078125,
      "learning_rate": 0.00019658571152151967,
      "loss": 7.3571,
      "step": 74080
    },
    {
      "epoch": 8.915764139590854,
      "grad_norm": 2784.0341796875,
      "learning_rate": 0.00019658472585912042,
      "loss": 7.4542,
      "step": 74090
    },
    {
      "epoch": 8.916967509025271,
      "grad_norm": 4800.8154296875,
      "learning_rate": 0.00019658374005693922,
      "loss": 7.501,
      "step": 74100
    },
    {
      "epoch": 8.918170878459687,
      "grad_norm": 5794.126953125,
      "learning_rate": 0.00019658275411497752,
      "loss": 7.5696,
      "step": 74110
    },
    {
      "epoch": 8.919374247894103,
      "grad_norm": 7495.03125,
      "learning_rate": 0.00019658176803323668,
      "loss": 7.5415,
      "step": 74120
    },
    {
      "epoch": 8.92057761732852,
      "grad_norm": 5297.14990234375,
      "learning_rate": 0.00019658078181171822,
      "loss": 7.5821,
      "step": 74130
    },
    {
      "epoch": 8.921780986762936,
      "grad_norm": 1676.0108642578125,
      "learning_rate": 0.00019657979545042352,
      "loss": 7.5723,
      "step": 74140
    },
    {
      "epoch": 8.922984356197352,
      "grad_norm": 3404.173828125,
      "learning_rate": 0.000196578808949354,
      "loss": 7.4392,
      "step": 74150
    },
    {
      "epoch": 8.92418772563177,
      "grad_norm": 1335.2464599609375,
      "learning_rate": 0.00019657782230851108,
      "loss": 7.6545,
      "step": 74160
    },
    {
      "epoch": 8.925391095066185,
      "grad_norm": 1651.59033203125,
      "learning_rate": 0.00019657683552789623,
      "loss": 7.5228,
      "step": 74170
    },
    {
      "epoch": 8.926594464500601,
      "grad_norm": 3013.30322265625,
      "learning_rate": 0.00019657584860751083,
      "loss": 7.6587,
      "step": 74180
    },
    {
      "epoch": 8.927797833935019,
      "grad_norm": 798.3281860351562,
      "learning_rate": 0.00019657486154735638,
      "loss": 7.6309,
      "step": 74190
    },
    {
      "epoch": 8.929001203369435,
      "grad_norm": 755.2177734375,
      "learning_rate": 0.0001965738743474342,
      "loss": 7.6763,
      "step": 74200
    },
    {
      "epoch": 8.93020457280385,
      "grad_norm": 1988.45458984375,
      "learning_rate": 0.00019657288700774585,
      "loss": 7.568,
      "step": 74210
    },
    {
      "epoch": 8.931407942238268,
      "grad_norm": 1843.828369140625,
      "learning_rate": 0.00019657189952829266,
      "loss": 7.5603,
      "step": 74220
    },
    {
      "epoch": 8.932611311672684,
      "grad_norm": 2340.12548828125,
      "learning_rate": 0.00019657091190907608,
      "loss": 7.6733,
      "step": 74230
    },
    {
      "epoch": 8.9338146811071,
      "grad_norm": 5211.78955078125,
      "learning_rate": 0.00019656992415009757,
      "loss": 7.6235,
      "step": 74240
    },
    {
      "epoch": 8.935018050541515,
      "grad_norm": 660.459228515625,
      "learning_rate": 0.0001965689362513585,
      "loss": 7.546,
      "step": 74250
    },
    {
      "epoch": 8.936221419975933,
      "grad_norm": 1261.160888671875,
      "learning_rate": 0.00019656794821286037,
      "loss": 7.5232,
      "step": 74260
    },
    {
      "epoch": 8.937424789410349,
      "grad_norm": 425.0375671386719,
      "learning_rate": 0.00019656696003460456,
      "loss": 7.4593,
      "step": 74270
    },
    {
      "epoch": 8.938628158844764,
      "grad_norm": 2253.50537109375,
      "learning_rate": 0.0001965659717165925,
      "loss": 7.4774,
      "step": 74280
    },
    {
      "epoch": 8.939831528279182,
      "grad_norm": 6685.58447265625,
      "learning_rate": 0.00019656498325882566,
      "loss": 7.5481,
      "step": 74290
    },
    {
      "epoch": 8.941034897713598,
      "grad_norm": 18517.52734375,
      "learning_rate": 0.00019656399466130546,
      "loss": 7.6617,
      "step": 74300
    },
    {
      "epoch": 8.942238267148014,
      "grad_norm": 17794.576171875,
      "learning_rate": 0.0001965630059240333,
      "loss": 7.5496,
      "step": 74310
    },
    {
      "epoch": 8.943441636582431,
      "grad_norm": 12111.9296875,
      "learning_rate": 0.00019656201704701065,
      "loss": 7.5412,
      "step": 74320
    },
    {
      "epoch": 8.944645006016847,
      "grad_norm": 63324.16796875,
      "learning_rate": 0.00019656102803023887,
      "loss": 7.4649,
      "step": 74330
    },
    {
      "epoch": 8.945848375451263,
      "grad_norm": 34897.78125,
      "learning_rate": 0.0001965600388737195,
      "loss": 7.343,
      "step": 74340
    },
    {
      "epoch": 8.94705174488568,
      "grad_norm": 58111.921875,
      "learning_rate": 0.0001965590495774539,
      "loss": 7.4078,
      "step": 74350
    },
    {
      "epoch": 8.948255114320096,
      "grad_norm": 124278.6640625,
      "learning_rate": 0.0001965580601414435,
      "loss": 7.5357,
      "step": 74360
    },
    {
      "epoch": 8.949458483754512,
      "grad_norm": 155516.234375,
      "learning_rate": 0.00019655707056568975,
      "loss": 7.6066,
      "step": 74370
    },
    {
      "epoch": 8.95066185318893,
      "grad_norm": 53974.58203125,
      "learning_rate": 0.0001965560808501941,
      "loss": 7.4978,
      "step": 74380
    },
    {
      "epoch": 8.951865222623345,
      "grad_norm": 28184.689453125,
      "learning_rate": 0.0001965550909949579,
      "loss": 7.5567,
      "step": 74390
    },
    {
      "epoch": 8.953068592057761,
      "grad_norm": 91322.3828125,
      "learning_rate": 0.00019655410099998273,
      "loss": 7.8956,
      "step": 74400
    },
    {
      "epoch": 8.954271961492179,
      "grad_norm": 71777.9375,
      "learning_rate": 0.00019655311086526987,
      "loss": 7.735,
      "step": 74410
    },
    {
      "epoch": 8.955475330926594,
      "grad_norm": 234126.375,
      "learning_rate": 0.00019655212059082087,
      "loss": 7.511,
      "step": 74420
    },
    {
      "epoch": 8.95667870036101,
      "grad_norm": 533.6937866210938,
      "learning_rate": 0.0001965511301766371,
      "loss": 7.4187,
      "step": 74430
    },
    {
      "epoch": 8.957882069795428,
      "grad_norm": 752.577880859375,
      "learning_rate": 0.00019655013962272002,
      "loss": 7.3956,
      "step": 74440
    },
    {
      "epoch": 8.959085439229844,
      "grad_norm": 1881.1563720703125,
      "learning_rate": 0.00019654914892907103,
      "loss": 7.4691,
      "step": 74450
    },
    {
      "epoch": 8.96028880866426,
      "grad_norm": 6551.96875,
      "learning_rate": 0.00019654815809569157,
      "loss": 7.4811,
      "step": 74460
    },
    {
      "epoch": 8.961492178098677,
      "grad_norm": 4642.265625,
      "learning_rate": 0.0001965471671225831,
      "loss": 7.3208,
      "step": 74470
    },
    {
      "epoch": 8.962695547533093,
      "grad_norm": 7261.7666015625,
      "learning_rate": 0.00019654617600974706,
      "loss": 7.4772,
      "step": 74480
    },
    {
      "epoch": 8.963898916967509,
      "grad_norm": 62814.953125,
      "learning_rate": 0.00019654518475718485,
      "loss": 8.1708,
      "step": 74490
    },
    {
      "epoch": 8.965102286401926,
      "grad_norm": 5553.8818359375,
      "learning_rate": 0.0001965441933648979,
      "loss": 9.1301,
      "step": 74500
    },
    {
      "epoch": 8.966305655836342,
      "grad_norm": 441.4981384277344,
      "learning_rate": 0.0001965432018328877,
      "loss": 7.7538,
      "step": 74510
    },
    {
      "epoch": 8.967509025270758,
      "grad_norm": 291.3560485839844,
      "learning_rate": 0.00019654221016115564,
      "loss": 7.7199,
      "step": 74520
    },
    {
      "epoch": 8.968712394705175,
      "grad_norm": 1062.9539794921875,
      "learning_rate": 0.00019654121834970315,
      "loss": 7.7126,
      "step": 74530
    },
    {
      "epoch": 8.969915764139591,
      "grad_norm": 1585.7899169921875,
      "learning_rate": 0.00019654022639853174,
      "loss": 7.4453,
      "step": 74540
    },
    {
      "epoch": 8.971119133574007,
      "grad_norm": 267.0061340332031,
      "learning_rate": 0.00019653923430764273,
      "loss": 7.4073,
      "step": 74550
    },
    {
      "epoch": 8.972322503008424,
      "grad_norm": 15941.5517578125,
      "learning_rate": 0.00019653824207703765,
      "loss": 7.4764,
      "step": 74560
    },
    {
      "epoch": 8.97352587244284,
      "grad_norm": 35514.24609375,
      "learning_rate": 0.00019653724970671786,
      "loss": 7.6851,
      "step": 74570
    },
    {
      "epoch": 8.974729241877256,
      "grad_norm": 55953.18359375,
      "learning_rate": 0.00019653625719668485,
      "loss": 7.544,
      "step": 74580
    },
    {
      "epoch": 8.975932611311674,
      "grad_norm": 61923.56640625,
      "learning_rate": 0.00019653526454694003,
      "loss": 7.8099,
      "step": 74590
    },
    {
      "epoch": 8.97713598074609,
      "grad_norm": 490680.71875,
      "learning_rate": 0.0001965342717574849,
      "loss": 7.7647,
      "step": 74600
    },
    {
      "epoch": 8.978339350180505,
      "grad_norm": 1914434.125,
      "learning_rate": 0.0001965332788283208,
      "loss": 8.1098,
      "step": 74610
    },
    {
      "epoch": 8.979542719614923,
      "grad_norm": 1534326.875,
      "learning_rate": 0.00019653228575944922,
      "loss": 8.1381,
      "step": 74620
    },
    {
      "epoch": 8.980746089049338,
      "grad_norm": 744071.5,
      "learning_rate": 0.00019653129255087158,
      "loss": 7.9991,
      "step": 74630
    },
    {
      "epoch": 8.981949458483754,
      "grad_norm": 1192951.5,
      "learning_rate": 0.00019653029920258934,
      "loss": 7.9607,
      "step": 74640
    },
    {
      "epoch": 8.98315282791817,
      "grad_norm": 206944.8125,
      "learning_rate": 0.00019652930571460394,
      "loss": 7.5681,
      "step": 74650
    },
    {
      "epoch": 8.984356197352588,
      "grad_norm": 389405.21875,
      "learning_rate": 0.0001965283120869168,
      "loss": 7.5574,
      "step": 74660
    },
    {
      "epoch": 8.985559566787003,
      "grad_norm": 28191.26953125,
      "learning_rate": 0.00019652731831952934,
      "loss": 7.4444,
      "step": 74670
    },
    {
      "epoch": 8.98676293622142,
      "grad_norm": 530025.0625,
      "learning_rate": 0.00019652632441244302,
      "loss": 7.6509,
      "step": 74680
    },
    {
      "epoch": 8.987966305655837,
      "grad_norm": 1986.7574462890625,
      "learning_rate": 0.00019652533036565927,
      "loss": 7.7447,
      "step": 74690
    },
    {
      "epoch": 8.989169675090253,
      "grad_norm": 773.9241943359375,
      "learning_rate": 0.00019652433617917952,
      "loss": 7.6749,
      "step": 74700
    },
    {
      "epoch": 8.990373044524668,
      "grad_norm": 1008.8065185546875,
      "learning_rate": 0.00019652334185300528,
      "loss": 7.5409,
      "step": 74710
    },
    {
      "epoch": 8.991576413959086,
      "grad_norm": 652.8068237304688,
      "learning_rate": 0.0001965223473871379,
      "loss": 7.4327,
      "step": 74720
    },
    {
      "epoch": 8.992779783393502,
      "grad_norm": 426.9729309082031,
      "learning_rate": 0.00019652135278157886,
      "loss": 7.4154,
      "step": 74730
    },
    {
      "epoch": 8.993983152827917,
      "grad_norm": 1204.4501953125,
      "learning_rate": 0.0001965203580363296,
      "loss": 7.2929,
      "step": 74740
    },
    {
      "epoch": 8.995186522262335,
      "grad_norm": 493.05084228515625,
      "learning_rate": 0.00019651936315139153,
      "loss": 7.361,
      "step": 74750
    },
    {
      "epoch": 8.99638989169675,
      "grad_norm": 527.6356201171875,
      "learning_rate": 0.00019651836812676612,
      "loss": 7.3862,
      "step": 74760
    },
    {
      "epoch": 8.997593261131167,
      "grad_norm": 1537.05517578125,
      "learning_rate": 0.0001965173729624548,
      "loss": 7.3668,
      "step": 74770
    },
    {
      "epoch": 8.998796630565584,
      "grad_norm": 255.41732788085938,
      "learning_rate": 0.00019651637765845902,
      "loss": 7.5919,
      "step": 74780
    },
    {
      "epoch": 9.0,
      "grad_norm": 1466.275390625,
      "learning_rate": 0.00019651538221478022,
      "loss": 7.5058,
      "step": 74790
    },
    {
      "epoch": 9.0,
      "eval_loss": 7.465070724487305,
      "eval_runtime": 119.5011,
      "eval_samples_per_second": 61.815,
      "eval_steps_per_second": 7.732,
      "step": 74790
    },
    {
      "epoch": 9.001203369434416,
      "grad_norm": 597.3857421875,
      "learning_rate": 0.00019651438663141981,
      "loss": 7.3772,
      "step": 74800
    },
    {
      "epoch": 9.002406738868833,
      "grad_norm": 1271.1337890625,
      "learning_rate": 0.00019651339090837928,
      "loss": 7.4259,
      "step": 74810
    },
    {
      "epoch": 9.00361010830325,
      "grad_norm": 827.50537109375,
      "learning_rate": 0.00019651239504566003,
      "loss": 7.4705,
      "step": 74820
    },
    {
      "epoch": 9.004813477737665,
      "grad_norm": 242.7447967529297,
      "learning_rate": 0.0001965113990432635,
      "loss": 7.4658,
      "step": 74830
    },
    {
      "epoch": 9.006016847172083,
      "grad_norm": 425.5442199707031,
      "learning_rate": 0.00019651040290119117,
      "loss": 7.5238,
      "step": 74840
    },
    {
      "epoch": 9.007220216606498,
      "grad_norm": 3862.75830078125,
      "learning_rate": 0.00019650940661944448,
      "loss": 7.2356,
      "step": 74850
    },
    {
      "epoch": 9.008423586040914,
      "grad_norm": 6282.27685546875,
      "learning_rate": 0.00019650841019802482,
      "loss": 7.3926,
      "step": 74860
    },
    {
      "epoch": 9.009626955475332,
      "grad_norm": 2767.276611328125,
      "learning_rate": 0.00019650741363693369,
      "loss": 7.5318,
      "step": 74870
    },
    {
      "epoch": 9.010830324909747,
      "grad_norm": 1277.268310546875,
      "learning_rate": 0.00019650641693617246,
      "loss": 7.477,
      "step": 74880
    },
    {
      "epoch": 9.012033694344163,
      "grad_norm": 10262.96875,
      "learning_rate": 0.00019650542009574265,
      "loss": 7.6104,
      "step": 74890
    },
    {
      "epoch": 9.01323706377858,
      "grad_norm": 5794.29736328125,
      "learning_rate": 0.00019650442311564564,
      "loss": 7.5953,
      "step": 74900
    },
    {
      "epoch": 9.014440433212997,
      "grad_norm": 3976.21826171875,
      "learning_rate": 0.00019650342599588294,
      "loss": 7.5964,
      "step": 74910
    },
    {
      "epoch": 9.015643802647412,
      "grad_norm": 4755.27099609375,
      "learning_rate": 0.00019650242873645593,
      "loss": 7.57,
      "step": 74920
    },
    {
      "epoch": 9.01684717208183,
      "grad_norm": 5756.71533203125,
      "learning_rate": 0.00019650143133736613,
      "loss": 7.8332,
      "step": 74930
    },
    {
      "epoch": 9.018050541516246,
      "grad_norm": 3540.826416015625,
      "learning_rate": 0.0001965004337986149,
      "loss": 7.6195,
      "step": 74940
    },
    {
      "epoch": 9.019253910950662,
      "grad_norm": 1619.6551513671875,
      "learning_rate": 0.0001964994361202037,
      "loss": 7.4998,
      "step": 74950
    },
    {
      "epoch": 9.020457280385079,
      "grad_norm": 1862.5577392578125,
      "learning_rate": 0.00019649843830213398,
      "loss": 7.393,
      "step": 74960
    },
    {
      "epoch": 9.021660649819495,
      "grad_norm": 2073.433349609375,
      "learning_rate": 0.00019649744034440722,
      "loss": 7.3996,
      "step": 74970
    },
    {
      "epoch": 9.02286401925391,
      "grad_norm": 3002.72265625,
      "learning_rate": 0.00019649644224702486,
      "loss": 7.4384,
      "step": 74980
    },
    {
      "epoch": 9.024067388688326,
      "grad_norm": 2637.7998046875,
      "learning_rate": 0.0001964954440099883,
      "loss": 7.4376,
      "step": 74990
    },
    {
      "epoch": 9.025270758122744,
      "grad_norm": 2617.635498046875,
      "learning_rate": 0.00019649444563329903,
      "loss": 7.366,
      "step": 75000
    },
    {
      "epoch": 9.02647412755716,
      "grad_norm": 3600.6943359375,
      "learning_rate": 0.00019649344711695842,
      "loss": 7.3583,
      "step": 75010
    },
    {
      "epoch": 9.027677496991576,
      "grad_norm": 3394.476806640625,
      "learning_rate": 0.00019649244846096803,
      "loss": 7.3795,
      "step": 75020
    },
    {
      "epoch": 9.028880866425993,
      "grad_norm": 3939.307373046875,
      "learning_rate": 0.00019649144966532922,
      "loss": 7.477,
      "step": 75030
    },
    {
      "epoch": 9.030084235860409,
      "grad_norm": 6318.85693359375,
      "learning_rate": 0.00019649045073004344,
      "loss": 7.5165,
      "step": 75040
    },
    {
      "epoch": 9.031287605294825,
      "grad_norm": 3790.567626953125,
      "learning_rate": 0.00019648945165511218,
      "loss": 7.6241,
      "step": 75050
    },
    {
      "epoch": 9.032490974729242,
      "grad_norm": 2998.84228515625,
      "learning_rate": 0.00019648845244053687,
      "loss": 7.6216,
      "step": 75060
    },
    {
      "epoch": 9.033694344163658,
      "grad_norm": 4382.93505859375,
      "learning_rate": 0.00019648745308631892,
      "loss": 7.6728,
      "step": 75070
    },
    {
      "epoch": 9.034897713598074,
      "grad_norm": 3263.46142578125,
      "learning_rate": 0.00019648645359245982,
      "loss": 7.7167,
      "step": 75080
    },
    {
      "epoch": 9.036101083032491,
      "grad_norm": 4775.517578125,
      "learning_rate": 0.00019648545395896098,
      "loss": 7.7483,
      "step": 75090
    },
    {
      "epoch": 9.037304452466907,
      "grad_norm": 2728.435302734375,
      "learning_rate": 0.00019648445418582388,
      "loss": 7.6134,
      "step": 75100
    },
    {
      "epoch": 9.038507821901323,
      "grad_norm": 8050.763671875,
      "learning_rate": 0.00019648345427304996,
      "loss": 7.5364,
      "step": 75110
    },
    {
      "epoch": 9.03971119133574,
      "grad_norm": 5224.39111328125,
      "learning_rate": 0.00019648245422064067,
      "loss": 7.7044,
      "step": 75120
    },
    {
      "epoch": 9.040914560770156,
      "grad_norm": 5393.15966796875,
      "learning_rate": 0.0001964814540285974,
      "loss": 7.6266,
      "step": 75130
    },
    {
      "epoch": 9.042117930204572,
      "grad_norm": 4150.26416015625,
      "learning_rate": 0.0001964804536969217,
      "loss": 7.6594,
      "step": 75140
    },
    {
      "epoch": 9.04332129963899,
      "grad_norm": 3149.526611328125,
      "learning_rate": 0.00019647945322561494,
      "loss": 7.6063,
      "step": 75150
    },
    {
      "epoch": 9.044524669073406,
      "grad_norm": 2388.351318359375,
      "learning_rate": 0.00019647845261467858,
      "loss": 7.5508,
      "step": 75160
    },
    {
      "epoch": 9.045728038507821,
      "grad_norm": 4836.52001953125,
      "learning_rate": 0.0001964774518641141,
      "loss": 7.5867,
      "step": 75170
    },
    {
      "epoch": 9.046931407942239,
      "grad_norm": 1788.6094970703125,
      "learning_rate": 0.00019647645097392293,
      "loss": 7.5901,
      "step": 75180
    },
    {
      "epoch": 9.048134777376655,
      "grad_norm": 7338.75634765625,
      "learning_rate": 0.0001964754499441065,
      "loss": 7.5614,
      "step": 75190
    },
    {
      "epoch": 9.04933814681107,
      "grad_norm": 3098.3955078125,
      "learning_rate": 0.00019647444877466628,
      "loss": 7.495,
      "step": 75200
    },
    {
      "epoch": 9.050541516245488,
      "grad_norm": 2653.363037109375,
      "learning_rate": 0.00019647344746560373,
      "loss": 7.6479,
      "step": 75210
    },
    {
      "epoch": 9.051744885679904,
      "grad_norm": 6382.98974609375,
      "learning_rate": 0.00019647244601692025,
      "loss": 7.5038,
      "step": 75220
    },
    {
      "epoch": 9.05294825511432,
      "grad_norm": 2274.214599609375,
      "learning_rate": 0.00019647144442861734,
      "loss": 7.5092,
      "step": 75230
    },
    {
      "epoch": 9.054151624548737,
      "grad_norm": 5597.10009765625,
      "learning_rate": 0.00019647044270069644,
      "loss": 7.5237,
      "step": 75240
    },
    {
      "epoch": 9.055354993983153,
      "grad_norm": 8310.5517578125,
      "learning_rate": 0.00019646944083315896,
      "loss": 7.572,
      "step": 75250
    },
    {
      "epoch": 9.056558363417569,
      "grad_norm": 2285.239501953125,
      "learning_rate": 0.00019646843882600642,
      "loss": 7.5238,
      "step": 75260
    },
    {
      "epoch": 9.057761732851986,
      "grad_norm": 9253.587890625,
      "learning_rate": 0.00019646743667924021,
      "loss": 7.5775,
      "step": 75270
    },
    {
      "epoch": 9.058965102286402,
      "grad_norm": 2495.662109375,
      "learning_rate": 0.0001964664343928618,
      "loss": 7.4708,
      "step": 75280
    },
    {
      "epoch": 9.060168471720818,
      "grad_norm": 3574.8486328125,
      "learning_rate": 0.00019646543196687267,
      "loss": 7.4904,
      "step": 75290
    },
    {
      "epoch": 9.061371841155236,
      "grad_norm": 3205.84912109375,
      "learning_rate": 0.0001964644294012742,
      "loss": 7.4509,
      "step": 75300
    },
    {
      "epoch": 9.062575210589651,
      "grad_norm": 4271.21533203125,
      "learning_rate": 0.00019646342669606793,
      "loss": 7.4857,
      "step": 75310
    },
    {
      "epoch": 9.063778580024067,
      "grad_norm": 2383.53515625,
      "learning_rate": 0.00019646242385125523,
      "loss": 7.2864,
      "step": 75320
    },
    {
      "epoch": 9.064981949458483,
      "grad_norm": 1065.3895263671875,
      "learning_rate": 0.0001964614208668376,
      "loss": 7.3884,
      "step": 75330
    },
    {
      "epoch": 9.0661853188929,
      "grad_norm": 2439.354736328125,
      "learning_rate": 0.0001964604177428165,
      "loss": 7.5093,
      "step": 75340
    },
    {
      "epoch": 9.067388688327316,
      "grad_norm": 1850.3184814453125,
      "learning_rate": 0.00019645941447919333,
      "loss": 7.3219,
      "step": 75350
    },
    {
      "epoch": 9.068592057761732,
      "grad_norm": 1488.564697265625,
      "learning_rate": 0.0001964584110759696,
      "loss": 7.3398,
      "step": 75360
    },
    {
      "epoch": 9.06979542719615,
      "grad_norm": 3440.6279296875,
      "learning_rate": 0.00019645740753314672,
      "loss": 7.4433,
      "step": 75370
    },
    {
      "epoch": 9.070998796630565,
      "grad_norm": 2456.555419921875,
      "learning_rate": 0.00019645640385072613,
      "loss": 7.473,
      "step": 75380
    },
    {
      "epoch": 9.072202166064981,
      "grad_norm": 11485.5634765625,
      "learning_rate": 0.00019645540002870935,
      "loss": 7.4575,
      "step": 75390
    },
    {
      "epoch": 9.073405535499399,
      "grad_norm": 4019.61669921875,
      "learning_rate": 0.00019645439606709778,
      "loss": 7.3776,
      "step": 75400
    },
    {
      "epoch": 9.074608904933815,
      "grad_norm": 2951.8466796875,
      "learning_rate": 0.00019645339196589285,
      "loss": 7.4738,
      "step": 75410
    },
    {
      "epoch": 9.07581227436823,
      "grad_norm": 2534.1728515625,
      "learning_rate": 0.0001964523877250961,
      "loss": 7.5318,
      "step": 75420
    },
    {
      "epoch": 9.077015643802648,
      "grad_norm": 5971.91552734375,
      "learning_rate": 0.0001964513833447089,
      "loss": 7.6485,
      "step": 75430
    },
    {
      "epoch": 9.078219013237064,
      "grad_norm": 7832.55810546875,
      "learning_rate": 0.00019645037882473276,
      "loss": 7.6937,
      "step": 75440
    },
    {
      "epoch": 9.07942238267148,
      "grad_norm": 2537.312744140625,
      "learning_rate": 0.00019644937416516908,
      "loss": 7.4627,
      "step": 75450
    },
    {
      "epoch": 9.080625752105897,
      "grad_norm": 2070.791015625,
      "learning_rate": 0.00019644836936601935,
      "loss": 7.5414,
      "step": 75460
    },
    {
      "epoch": 9.081829121540313,
      "grad_norm": 5187.1982421875,
      "learning_rate": 0.00019644736442728505,
      "loss": 7.4998,
      "step": 75470
    },
    {
      "epoch": 9.083032490974729,
      "grad_norm": 2618.777587890625,
      "learning_rate": 0.00019644635934896757,
      "loss": 7.5091,
      "step": 75480
    },
    {
      "epoch": 9.084235860409146,
      "grad_norm": 3503.40966796875,
      "learning_rate": 0.0001964453541310684,
      "loss": 7.5392,
      "step": 75490
    },
    {
      "epoch": 9.085439229843562,
      "grad_norm": 2291.1005859375,
      "learning_rate": 0.000196444348773589,
      "loss": 7.4334,
      "step": 75500
    },
    {
      "epoch": 9.086642599277978,
      "grad_norm": 4875.828125,
      "learning_rate": 0.0001964433432765308,
      "loss": 7.3775,
      "step": 75510
    },
    {
      "epoch": 9.087845968712395,
      "grad_norm": 4910.685546875,
      "learning_rate": 0.00019644233763989528,
      "loss": 7.3955,
      "step": 75520
    },
    {
      "epoch": 9.089049338146811,
      "grad_norm": 1899.69091796875,
      "learning_rate": 0.00019644133186368388,
      "loss": 7.2959,
      "step": 75530
    },
    {
      "epoch": 9.090252707581227,
      "grad_norm": 5022.66796875,
      "learning_rate": 0.00019644032594789808,
      "loss": 7.3931,
      "step": 75540
    },
    {
      "epoch": 9.091456077015645,
      "grad_norm": 2716.967529296875,
      "learning_rate": 0.0001964393198925393,
      "loss": 7.4728,
      "step": 75550
    },
    {
      "epoch": 9.09265944645006,
      "grad_norm": 4830.9619140625,
      "learning_rate": 0.00019643831369760905,
      "loss": 7.5441,
      "step": 75560
    },
    {
      "epoch": 9.093862815884476,
      "grad_norm": 3634.146240234375,
      "learning_rate": 0.00019643730736310867,
      "loss": 7.4721,
      "step": 75570
    },
    {
      "epoch": 9.095066185318894,
      "grad_norm": 4341.45166015625,
      "learning_rate": 0.00019643630088903976,
      "loss": 7.5323,
      "step": 75580
    },
    {
      "epoch": 9.09626955475331,
      "grad_norm": 2593.2265625,
      "learning_rate": 0.0001964352942754037,
      "loss": 7.4874,
      "step": 75590
    },
    {
      "epoch": 9.097472924187725,
      "grad_norm": 2320.23876953125,
      "learning_rate": 0.00019643428752220195,
      "loss": 7.3997,
      "step": 75600
    },
    {
      "epoch": 9.098676293622143,
      "grad_norm": 2152.728271484375,
      "learning_rate": 0.000196433280629436,
      "loss": 7.3698,
      "step": 75610
    },
    {
      "epoch": 9.099879663056559,
      "grad_norm": 2708.65625,
      "learning_rate": 0.00019643227359710728,
      "loss": 7.4336,
      "step": 75620
    },
    {
      "epoch": 9.101083032490974,
      "grad_norm": 2264.180908203125,
      "learning_rate": 0.00019643126642521726,
      "loss": 7.4608,
      "step": 75630
    },
    {
      "epoch": 9.102286401925392,
      "grad_norm": 2932.721435546875,
      "learning_rate": 0.00019643025911376735,
      "loss": 7.4104,
      "step": 75640
    },
    {
      "epoch": 9.103489771359808,
      "grad_norm": 3209.4423828125,
      "learning_rate": 0.0001964292516627591,
      "loss": 7.354,
      "step": 75650
    },
    {
      "epoch": 9.104693140794224,
      "grad_norm": 1590.328857421875,
      "learning_rate": 0.00019642824407219386,
      "loss": 7.3862,
      "step": 75660
    },
    {
      "epoch": 9.10589651022864,
      "grad_norm": 4043.4326171875,
      "learning_rate": 0.00019642723634207319,
      "loss": 7.3696,
      "step": 75670
    },
    {
      "epoch": 9.107099879663057,
      "grad_norm": 1901.3751220703125,
      "learning_rate": 0.00019642622847239848,
      "loss": 7.3036,
      "step": 75680
    },
    {
      "epoch": 9.108303249097473,
      "grad_norm": 2234.895751953125,
      "learning_rate": 0.0001964252204631712,
      "loss": 7.3256,
      "step": 75690
    },
    {
      "epoch": 9.109506618531888,
      "grad_norm": 1258.59130859375,
      "learning_rate": 0.00019642421231439285,
      "loss": 7.2707,
      "step": 75700
    },
    {
      "epoch": 9.110709987966306,
      "grad_norm": 1214.1578369140625,
      "learning_rate": 0.00019642320402606483,
      "loss": 7.291,
      "step": 75710
    },
    {
      "epoch": 9.111913357400722,
      "grad_norm": 1541.5499267578125,
      "learning_rate": 0.00019642219559818863,
      "loss": 7.2909,
      "step": 75720
    },
    {
      "epoch": 9.113116726835138,
      "grad_norm": 5279.283203125,
      "learning_rate": 0.00019642118703076572,
      "loss": 7.3405,
      "step": 75730
    },
    {
      "epoch": 9.114320096269555,
      "grad_norm": 2948.894287109375,
      "learning_rate": 0.00019642017832379756,
      "loss": 7.3548,
      "step": 75740
    },
    {
      "epoch": 9.115523465703971,
      "grad_norm": 1446.7877197265625,
      "learning_rate": 0.00019641916947728556,
      "loss": 7.2449,
      "step": 75750
    },
    {
      "epoch": 9.116726835138387,
      "grad_norm": 3335.70849609375,
      "learning_rate": 0.00019641816049123124,
      "loss": 7.2542,
      "step": 75760
    },
    {
      "epoch": 9.117930204572804,
      "grad_norm": 2856.826904296875,
      "learning_rate": 0.000196417151365636,
      "loss": 7.3918,
      "step": 75770
    },
    {
      "epoch": 9.11913357400722,
      "grad_norm": 830.1557006835938,
      "learning_rate": 0.00019641614210050138,
      "loss": 7.2572,
      "step": 75780
    },
    {
      "epoch": 9.120336943441636,
      "grad_norm": 3191.406982421875,
      "learning_rate": 0.0001964151326958288,
      "loss": 7.4104,
      "step": 75790
    },
    {
      "epoch": 9.121540312876053,
      "grad_norm": 2908.66015625,
      "learning_rate": 0.00019641412315161971,
      "loss": 7.3513,
      "step": 75800
    },
    {
      "epoch": 9.12274368231047,
      "grad_norm": 727.5166015625,
      "learning_rate": 0.00019641311346787558,
      "loss": 7.3175,
      "step": 75810
    },
    {
      "epoch": 9.123947051744885,
      "grad_norm": 2046.9010009765625,
      "learning_rate": 0.00019641210364459784,
      "loss": 7.2328,
      "step": 75820
    },
    {
      "epoch": 9.125150421179303,
      "grad_norm": 1243.1632080078125,
      "learning_rate": 0.00019641109368178804,
      "loss": 7.33,
      "step": 75830
    },
    {
      "epoch": 9.126353790613718,
      "grad_norm": 1148.9654541015625,
      "learning_rate": 0.00019641008357944753,
      "loss": 7.3078,
      "step": 75840
    },
    {
      "epoch": 9.127557160048134,
      "grad_norm": 969.3536987304688,
      "learning_rate": 0.00019640907333757785,
      "loss": 7.331,
      "step": 75850
    },
    {
      "epoch": 9.128760529482552,
      "grad_norm": 1206.808837890625,
      "learning_rate": 0.00019640806295618046,
      "loss": 7.3609,
      "step": 75860
    },
    {
      "epoch": 9.129963898916968,
      "grad_norm": 1089.9910888671875,
      "learning_rate": 0.00019640705243525677,
      "loss": 7.2742,
      "step": 75870
    },
    {
      "epoch": 9.131167268351383,
      "grad_norm": 1392.9652099609375,
      "learning_rate": 0.0001964060417748083,
      "loss": 7.3062,
      "step": 75880
    },
    {
      "epoch": 9.132370637785801,
      "grad_norm": 2660.32568359375,
      "learning_rate": 0.00019640503097483644,
      "loss": 7.4094,
      "step": 75890
    },
    {
      "epoch": 9.133574007220217,
      "grad_norm": 2828.388916015625,
      "learning_rate": 0.00019640402003534275,
      "loss": 7.2343,
      "step": 75900
    },
    {
      "epoch": 9.134777376654633,
      "grad_norm": 798.2307739257812,
      "learning_rate": 0.0001964030089563286,
      "loss": 7.333,
      "step": 75910
    },
    {
      "epoch": 9.13598074608905,
      "grad_norm": 2811.5615234375,
      "learning_rate": 0.0001964019977377955,
      "loss": 7.3057,
      "step": 75920
    },
    {
      "epoch": 9.137184115523466,
      "grad_norm": 1283.5013427734375,
      "learning_rate": 0.00019640098637974495,
      "loss": 7.2729,
      "step": 75930
    },
    {
      "epoch": 9.138387484957882,
      "grad_norm": 602.6644287109375,
      "learning_rate": 0.0001963999748821783,
      "loss": 7.2362,
      "step": 75940
    },
    {
      "epoch": 9.1395908543923,
      "grad_norm": 846.1353149414062,
      "learning_rate": 0.00019639896324509716,
      "loss": 7.2874,
      "step": 75950
    },
    {
      "epoch": 9.140794223826715,
      "grad_norm": 719.5494995117188,
      "learning_rate": 0.0001963979514685029,
      "loss": 7.3526,
      "step": 75960
    },
    {
      "epoch": 9.14199759326113,
      "grad_norm": 882.4271240234375,
      "learning_rate": 0.00019639693955239698,
      "loss": 7.2699,
      "step": 75970
    },
    {
      "epoch": 9.143200962695548,
      "grad_norm": 416.4239196777344,
      "learning_rate": 0.0001963959274967809,
      "loss": 7.3739,
      "step": 75980
    },
    {
      "epoch": 9.144404332129964,
      "grad_norm": 3346.033935546875,
      "learning_rate": 0.00019639491530165613,
      "loss": 7.2879,
      "step": 75990
    },
    {
      "epoch": 9.14560770156438,
      "grad_norm": 1748.701171875,
      "learning_rate": 0.0001963939029670241,
      "loss": 7.2639,
      "step": 76000
    },
    {
      "epoch": 9.146811070998796,
      "grad_norm": 967.0650634765625,
      "learning_rate": 0.00019639289049288628,
      "loss": 7.3475,
      "step": 76010
    },
    {
      "epoch": 9.148014440433213,
      "grad_norm": 925.6656494140625,
      "learning_rate": 0.00019639187787924418,
      "loss": 7.3854,
      "step": 76020
    },
    {
      "epoch": 9.14921780986763,
      "grad_norm": 1009.5439453125,
      "learning_rate": 0.00019639086512609924,
      "loss": 7.214,
      "step": 76030
    },
    {
      "epoch": 9.150421179302045,
      "grad_norm": 719.2240600585938,
      "learning_rate": 0.0001963898522334529,
      "loss": 7.4178,
      "step": 76040
    },
    {
      "epoch": 9.151624548736462,
      "grad_norm": 1386.6724853515625,
      "learning_rate": 0.00019638883920130665,
      "loss": 7.2835,
      "step": 76050
    },
    {
      "epoch": 9.152827918170878,
      "grad_norm": 523.4008178710938,
      "learning_rate": 0.00019638782602966192,
      "loss": 7.2417,
      "step": 76060
    },
    {
      "epoch": 9.154031287605294,
      "grad_norm": 1144.1290283203125,
      "learning_rate": 0.00019638681271852025,
      "loss": 7.1999,
      "step": 76070
    },
    {
      "epoch": 9.155234657039712,
      "grad_norm": 1325.0841064453125,
      "learning_rate": 0.00019638579926788306,
      "loss": 7.2872,
      "step": 76080
    },
    {
      "epoch": 9.156438026474127,
      "grad_norm": 2343.41845703125,
      "learning_rate": 0.0001963847856777518,
      "loss": 7.2956,
      "step": 76090
    },
    {
      "epoch": 9.157641395908543,
      "grad_norm": 606.81494140625,
      "learning_rate": 0.00019638377194812799,
      "loss": 7.438,
      "step": 76100
    },
    {
      "epoch": 9.15884476534296,
      "grad_norm": 415.5561828613281,
      "learning_rate": 0.00019638275807901302,
      "loss": 7.2043,
      "step": 76110
    },
    {
      "epoch": 9.160048134777377,
      "grad_norm": 832.2674560546875,
      "learning_rate": 0.00019638174407040845,
      "loss": 7.2921,
      "step": 76120
    },
    {
      "epoch": 9.161251504211792,
      "grad_norm": 376.2272644042969,
      "learning_rate": 0.00019638072992231566,
      "loss": 7.2569,
      "step": 76130
    },
    {
      "epoch": 9.16245487364621,
      "grad_norm": 1342.66357421875,
      "learning_rate": 0.00019637971563473617,
      "loss": 7.2735,
      "step": 76140
    },
    {
      "epoch": 9.163658243080626,
      "grad_norm": 2213.8486328125,
      "learning_rate": 0.00019637870120767147,
      "loss": 7.2528,
      "step": 76150
    },
    {
      "epoch": 9.164861612515042,
      "grad_norm": 456.4389343261719,
      "learning_rate": 0.00019637768664112297,
      "loss": 7.2569,
      "step": 76160
    },
    {
      "epoch": 9.166064981949459,
      "grad_norm": 1014.8824462890625,
      "learning_rate": 0.00019637667193509218,
      "loss": 7.2242,
      "step": 76170
    },
    {
      "epoch": 9.167268351383875,
      "grad_norm": 1248.652587890625,
      "learning_rate": 0.0001963756570895805,
      "loss": 7.3643,
      "step": 76180
    },
    {
      "epoch": 9.16847172081829,
      "grad_norm": 844.365478515625,
      "learning_rate": 0.00019637464210458952,
      "loss": 7.2705,
      "step": 76190
    },
    {
      "epoch": 9.169675090252708,
      "grad_norm": 3494.603515625,
      "learning_rate": 0.00019637362698012058,
      "loss": 7.1682,
      "step": 76200
    },
    {
      "epoch": 9.170878459687124,
      "grad_norm": 851.4253540039062,
      "learning_rate": 0.00019637261171617521,
      "loss": 7.2478,
      "step": 76210
    },
    {
      "epoch": 9.17208182912154,
      "grad_norm": 1071.74169921875,
      "learning_rate": 0.0001963715963127549,
      "loss": 7.3127,
      "step": 76220
    },
    {
      "epoch": 9.173285198555957,
      "grad_norm": 780.7906494140625,
      "learning_rate": 0.00019637058076986112,
      "loss": 7.1742,
      "step": 76230
    },
    {
      "epoch": 9.174488567990373,
      "grad_norm": 2235.20556640625,
      "learning_rate": 0.00019636956508749527,
      "loss": 7.2388,
      "step": 76240
    },
    {
      "epoch": 9.175691937424789,
      "grad_norm": 1000.28662109375,
      "learning_rate": 0.00019636854926565888,
      "loss": 7.2958,
      "step": 76250
    },
    {
      "epoch": 9.176895306859207,
      "grad_norm": 591.0308837890625,
      "learning_rate": 0.00019636753330435345,
      "loss": 7.223,
      "step": 76260
    },
    {
      "epoch": 9.178098676293622,
      "grad_norm": 830.2313232421875,
      "learning_rate": 0.00019636651720358035,
      "loss": 7.3221,
      "step": 76270
    },
    {
      "epoch": 9.179302045728038,
      "grad_norm": 445.8819580078125,
      "learning_rate": 0.00019636550096334112,
      "loss": 7.2662,
      "step": 76280
    },
    {
      "epoch": 9.180505415162456,
      "grad_norm": 994.1788330078125,
      "learning_rate": 0.00019636448458363722,
      "loss": 7.2521,
      "step": 76290
    },
    {
      "epoch": 9.181708784596871,
      "grad_norm": 809.820068359375,
      "learning_rate": 0.00019636346806447015,
      "loss": 7.3052,
      "step": 76300
    },
    {
      "epoch": 9.182912154031287,
      "grad_norm": 587.2872314453125,
      "learning_rate": 0.0001963624514058413,
      "loss": 7.2471,
      "step": 76310
    },
    {
      "epoch": 9.184115523465705,
      "grad_norm": 1231.2158203125,
      "learning_rate": 0.00019636143460775222,
      "loss": 7.2079,
      "step": 76320
    },
    {
      "epoch": 9.18531889290012,
      "grad_norm": 1350.02978515625,
      "learning_rate": 0.00019636041767020435,
      "loss": 7.1862,
      "step": 76330
    },
    {
      "epoch": 9.186522262334536,
      "grad_norm": 2103.52294921875,
      "learning_rate": 0.00019635940059319916,
      "loss": 7.319,
      "step": 76340
    },
    {
      "epoch": 9.187725631768952,
      "grad_norm": 2919.232666015625,
      "learning_rate": 0.00019635838337673814,
      "loss": 7.0929,
      "step": 76350
    },
    {
      "epoch": 9.18892900120337,
      "grad_norm": 1127.0545654296875,
      "learning_rate": 0.00019635736602082276,
      "loss": 7.2528,
      "step": 76360
    },
    {
      "epoch": 9.190132370637786,
      "grad_norm": 3849.189208984375,
      "learning_rate": 0.00019635634852545447,
      "loss": 7.2087,
      "step": 76370
    },
    {
      "epoch": 9.191335740072201,
      "grad_norm": 2272.21728515625,
      "learning_rate": 0.00019635533089063474,
      "loss": 7.2832,
      "step": 76380
    },
    {
      "epoch": 9.192539109506619,
      "grad_norm": 4344.23828125,
      "learning_rate": 0.00019635431311636504,
      "loss": 7.3083,
      "step": 76390
    },
    {
      "epoch": 9.193742478941035,
      "grad_norm": 2692.489013671875,
      "learning_rate": 0.0001963532952026469,
      "loss": 7.2261,
      "step": 76400
    },
    {
      "epoch": 9.19494584837545,
      "grad_norm": 2531.884033203125,
      "learning_rate": 0.00019635227714948173,
      "loss": 7.3983,
      "step": 76410
    },
    {
      "epoch": 9.196149217809868,
      "grad_norm": 3592.314453125,
      "learning_rate": 0.00019635125895687105,
      "loss": 7.2888,
      "step": 76420
    },
    {
      "epoch": 9.197352587244284,
      "grad_norm": 2629.387451171875,
      "learning_rate": 0.00019635024062481627,
      "loss": 7.2016,
      "step": 76430
    },
    {
      "epoch": 9.1985559566787,
      "grad_norm": 1859.731689453125,
      "learning_rate": 0.00019634922215331892,
      "loss": 7.3499,
      "step": 76440
    },
    {
      "epoch": 9.199759326113117,
      "grad_norm": 2324.42138671875,
      "learning_rate": 0.00019634820354238048,
      "loss": 7.2116,
      "step": 76450
    },
    {
      "epoch": 9.200962695547533,
      "grad_norm": 2354.31103515625,
      "learning_rate": 0.00019634718479200237,
      "loss": 7.2484,
      "step": 76460
    },
    {
      "epoch": 9.202166064981949,
      "grad_norm": 2572.722900390625,
      "learning_rate": 0.00019634616590218613,
      "loss": 7.2555,
      "step": 76470
    },
    {
      "epoch": 9.203369434416366,
      "grad_norm": 1928.0738525390625,
      "learning_rate": 0.0001963451468729332,
      "loss": 7.2725,
      "step": 76480
    },
    {
      "epoch": 9.204572803850782,
      "grad_norm": 2449.217529296875,
      "learning_rate": 0.00019634412770424504,
      "loss": 7.2963,
      "step": 76490
    },
    {
      "epoch": 9.205776173285198,
      "grad_norm": 2320.357177734375,
      "learning_rate": 0.00019634310839612313,
      "loss": 7.272,
      "step": 76500
    },
    {
      "epoch": 9.206979542719615,
      "grad_norm": 1060.3175048828125,
      "learning_rate": 0.00019634208894856895,
      "loss": 7.3014,
      "step": 76510
    },
    {
      "epoch": 9.208182912154031,
      "grad_norm": 2839.287353515625,
      "learning_rate": 0.00019634106936158402,
      "loss": 7.3121,
      "step": 76520
    },
    {
      "epoch": 9.209386281588447,
      "grad_norm": 3156.68115234375,
      "learning_rate": 0.00019634004963516973,
      "loss": 7.2479,
      "step": 76530
    },
    {
      "epoch": 9.210589651022865,
      "grad_norm": 1798.8192138671875,
      "learning_rate": 0.00019633902976932762,
      "loss": 7.2339,
      "step": 76540
    },
    {
      "epoch": 9.21179302045728,
      "grad_norm": 1643.83740234375,
      "learning_rate": 0.00019633800976405917,
      "loss": 7.1835,
      "step": 76550
    },
    {
      "epoch": 9.212996389891696,
      "grad_norm": 1385.792724609375,
      "learning_rate": 0.0001963369896193658,
      "loss": 7.3373,
      "step": 76560
    },
    {
      "epoch": 9.214199759326114,
      "grad_norm": 3337.798583984375,
      "learning_rate": 0.00019633596933524907,
      "loss": 7.2332,
      "step": 76570
    },
    {
      "epoch": 9.21540312876053,
      "grad_norm": 1712.10400390625,
      "learning_rate": 0.00019633494891171038,
      "loss": 7.3596,
      "step": 76580
    },
    {
      "epoch": 9.216606498194945,
      "grad_norm": 3498.635498046875,
      "learning_rate": 0.0001963339283487512,
      "loss": 7.268,
      "step": 76590
    },
    {
      "epoch": 9.217809867629363,
      "grad_norm": 3339.609375,
      "learning_rate": 0.0001963329076463731,
      "loss": 7.1177,
      "step": 76600
    },
    {
      "epoch": 9.219013237063779,
      "grad_norm": 1963.313232421875,
      "learning_rate": 0.00019633188680457744,
      "loss": 7.2571,
      "step": 76610
    },
    {
      "epoch": 9.220216606498195,
      "grad_norm": 1857.890625,
      "learning_rate": 0.00019633086582336582,
      "loss": 7.2689,
      "step": 76620
    },
    {
      "epoch": 9.221419975932612,
      "grad_norm": 1188.5067138671875,
      "learning_rate": 0.0001963298447027396,
      "loss": 7.2145,
      "step": 76630
    },
    {
      "epoch": 9.222623345367028,
      "grad_norm": 786.3170776367188,
      "learning_rate": 0.00019632882344270037,
      "loss": 7.2828,
      "step": 76640
    },
    {
      "epoch": 9.223826714801444,
      "grad_norm": 2681.61865234375,
      "learning_rate": 0.0001963278020432495,
      "loss": 7.3441,
      "step": 76650
    },
    {
      "epoch": 9.225030084235861,
      "grad_norm": 787.3405151367188,
      "learning_rate": 0.00019632678050438853,
      "loss": 7.2636,
      "step": 76660
    },
    {
      "epoch": 9.226233453670277,
      "grad_norm": 832.1189575195312,
      "learning_rate": 0.00019632575882611893,
      "loss": 7.2567,
      "step": 76670
    },
    {
      "epoch": 9.227436823104693,
      "grad_norm": 919.2532348632812,
      "learning_rate": 0.00019632473700844218,
      "loss": 7.1968,
      "step": 76680
    },
    {
      "epoch": 9.22864019253911,
      "grad_norm": 2160.03271484375,
      "learning_rate": 0.00019632371505135977,
      "loss": 7.2358,
      "step": 76690
    },
    {
      "epoch": 9.229843561973526,
      "grad_norm": 1070.5589599609375,
      "learning_rate": 0.0001963226929548731,
      "loss": 7.2749,
      "step": 76700
    },
    {
      "epoch": 9.231046931407942,
      "grad_norm": 1854.709716796875,
      "learning_rate": 0.00019632167071898375,
      "loss": 7.2184,
      "step": 76710
    },
    {
      "epoch": 9.232250300842358,
      "grad_norm": 2756.621337890625,
      "learning_rate": 0.0001963206483436932,
      "loss": 7.234,
      "step": 76720
    },
    {
      "epoch": 9.233453670276775,
      "grad_norm": 2554.576416015625,
      "learning_rate": 0.00019631962582900285,
      "loss": 7.2721,
      "step": 76730
    },
    {
      "epoch": 9.234657039711191,
      "grad_norm": 1862.0181884765625,
      "learning_rate": 0.00019631860317491422,
      "loss": 7.2227,
      "step": 76740
    },
    {
      "epoch": 9.235860409145607,
      "grad_norm": 1320.4754638671875,
      "learning_rate": 0.00019631758038142879,
      "loss": 7.2316,
      "step": 76750
    },
    {
      "epoch": 9.237063778580024,
      "grad_norm": 1313.3798828125,
      "learning_rate": 0.00019631655744854805,
      "loss": 7.3046,
      "step": 76760
    },
    {
      "epoch": 9.23826714801444,
      "grad_norm": 1622.3133544921875,
      "learning_rate": 0.00019631553437627348,
      "loss": 7.2199,
      "step": 76770
    },
    {
      "epoch": 9.239470517448856,
      "grad_norm": 2560.140380859375,
      "learning_rate": 0.00019631451116460652,
      "loss": 7.3148,
      "step": 76780
    },
    {
      "epoch": 9.240673886883274,
      "grad_norm": 8938.111328125,
      "learning_rate": 0.0001963134878135487,
      "loss": 7.2599,
      "step": 76790
    },
    {
      "epoch": 9.24187725631769,
      "grad_norm": 3016.765380859375,
      "learning_rate": 0.00019631246432310148,
      "loss": 7.2187,
      "step": 76800
    },
    {
      "epoch": 9.243080625752105,
      "grad_norm": 1779.9124755859375,
      "learning_rate": 0.00019631144069326637,
      "loss": 7.2957,
      "step": 76810
    },
    {
      "epoch": 9.244283995186523,
      "grad_norm": 1240.015625,
      "learning_rate": 0.00019631041692404482,
      "loss": 7.3554,
      "step": 76820
    },
    {
      "epoch": 9.245487364620939,
      "grad_norm": 2843.078857421875,
      "learning_rate": 0.00019630939301543832,
      "loss": 7.2836,
      "step": 76830
    },
    {
      "epoch": 9.246690734055354,
      "grad_norm": 2494.840576171875,
      "learning_rate": 0.00019630836896744833,
      "loss": 7.2299,
      "step": 76840
    },
    {
      "epoch": 9.247894103489772,
      "grad_norm": 6024.77099609375,
      "learning_rate": 0.00019630734478007637,
      "loss": 7.2111,
      "step": 76850
    },
    {
      "epoch": 9.249097472924188,
      "grad_norm": 4501.42041015625,
      "learning_rate": 0.00019630632045332388,
      "loss": 7.177,
      "step": 76860
    },
    {
      "epoch": 9.250300842358604,
      "grad_norm": 14304.2392578125,
      "learning_rate": 0.00019630529598719238,
      "loss": 7.2481,
      "step": 76870
    },
    {
      "epoch": 9.251504211793021,
      "grad_norm": 9051.515625,
      "learning_rate": 0.00019630427138168336,
      "loss": 7.3745,
      "step": 76880
    },
    {
      "epoch": 9.252707581227437,
      "grad_norm": 4632.99365234375,
      "learning_rate": 0.00019630324663679827,
      "loss": 7.3549,
      "step": 76890
    },
    {
      "epoch": 9.253910950661853,
      "grad_norm": 8207.3876953125,
      "learning_rate": 0.0001963022217525386,
      "loss": 7.3544,
      "step": 76900
    },
    {
      "epoch": 9.25511432009627,
      "grad_norm": 4218.57666015625,
      "learning_rate": 0.0001963011967289058,
      "loss": 7.2332,
      "step": 76910
    },
    {
      "epoch": 9.256317689530686,
      "grad_norm": 3511.77783203125,
      "learning_rate": 0.00019630017156590146,
      "loss": 7.2147,
      "step": 76920
    },
    {
      "epoch": 9.257521058965102,
      "grad_norm": 1797.2041015625,
      "learning_rate": 0.00019629914626352696,
      "loss": 7.234,
      "step": 76930
    },
    {
      "epoch": 9.25872442839952,
      "grad_norm": 4163.26806640625,
      "learning_rate": 0.00019629812082178384,
      "loss": 7.1906,
      "step": 76940
    },
    {
      "epoch": 9.259927797833935,
      "grad_norm": 6068.142578125,
      "learning_rate": 0.00019629709524067356,
      "loss": 7.2345,
      "step": 76950
    },
    {
      "epoch": 9.261131167268351,
      "grad_norm": 4087.80517578125,
      "learning_rate": 0.0001962960695201976,
      "loss": 7.2311,
      "step": 76960
    },
    {
      "epoch": 9.262334536702769,
      "grad_norm": 4219.29150390625,
      "learning_rate": 0.00019629504366035745,
      "loss": 7.1954,
      "step": 76970
    },
    {
      "epoch": 9.263537906137184,
      "grad_norm": 3657.016357421875,
      "learning_rate": 0.00019629401766115462,
      "loss": 7.3137,
      "step": 76980
    },
    {
      "epoch": 9.2647412755716,
      "grad_norm": 4069.469482421875,
      "learning_rate": 0.00019629299152259052,
      "loss": 7.302,
      "step": 76990
    },
    {
      "epoch": 9.265944645006018,
      "grad_norm": 2477.508544921875,
      "learning_rate": 0.00019629196524466674,
      "loss": 7.2541,
      "step": 77000
    },
    {
      "epoch": 9.267148014440433,
      "grad_norm": 3701.619140625,
      "learning_rate": 0.00019629093882738468,
      "loss": 7.3844,
      "step": 77010
    },
    {
      "epoch": 9.26835138387485,
      "grad_norm": 4865.80029296875,
      "learning_rate": 0.00019628991227074586,
      "loss": 7.2111,
      "step": 77020
    },
    {
      "epoch": 9.269554753309265,
      "grad_norm": 2069.574951171875,
      "learning_rate": 0.00019628888557475177,
      "loss": 7.2608,
      "step": 77030
    },
    {
      "epoch": 9.270758122743683,
      "grad_norm": 6883.31787109375,
      "learning_rate": 0.00019628785873940393,
      "loss": 7.3075,
      "step": 77040
    },
    {
      "epoch": 9.271961492178098,
      "grad_norm": 2148.96533203125,
      "learning_rate": 0.00019628683176470371,
      "loss": 7.1896,
      "step": 77050
    },
    {
      "epoch": 9.273164861612514,
      "grad_norm": 1338.982421875,
      "learning_rate": 0.00019628580465065273,
      "loss": 7.2983,
      "step": 77060
    },
    {
      "epoch": 9.274368231046932,
      "grad_norm": 6684.205078125,
      "learning_rate": 0.00019628477739725239,
      "loss": 7.112,
      "step": 77070
    },
    {
      "epoch": 9.275571600481348,
      "grad_norm": 1727.3746337890625,
      "learning_rate": 0.00019628375000450423,
      "loss": 7.0922,
      "step": 77080
    },
    {
      "epoch": 9.276774969915763,
      "grad_norm": 3326.0224609375,
      "learning_rate": 0.0001962827224724097,
      "loss": 7.1227,
      "step": 77090
    },
    {
      "epoch": 9.277978339350181,
      "grad_norm": 3629.23828125,
      "learning_rate": 0.00019628169480097028,
      "loss": 7.2084,
      "step": 77100
    },
    {
      "epoch": 9.279181708784597,
      "grad_norm": 2339.129150390625,
      "learning_rate": 0.0001962806669901875,
      "loss": 7.2363,
      "step": 77110
    },
    {
      "epoch": 9.280385078219012,
      "grad_norm": 3463.510986328125,
      "learning_rate": 0.00019627963904006283,
      "loss": 7.2249,
      "step": 77120
    },
    {
      "epoch": 9.28158844765343,
      "grad_norm": 3700.374755859375,
      "learning_rate": 0.0001962786109505977,
      "loss": 7.2272,
      "step": 77130
    },
    {
      "epoch": 9.282791817087846,
      "grad_norm": 9186.841796875,
      "learning_rate": 0.00019627758272179372,
      "loss": 7.1917,
      "step": 77140
    },
    {
      "epoch": 9.283995186522262,
      "grad_norm": 4616.56201171875,
      "learning_rate": 0.00019627655435365226,
      "loss": 7.3142,
      "step": 77150
    },
    {
      "epoch": 9.28519855595668,
      "grad_norm": 2341.943115234375,
      "learning_rate": 0.0001962755258461749,
      "loss": 7.289,
      "step": 77160
    },
    {
      "epoch": 9.286401925391095,
      "grad_norm": 4500.80615234375,
      "learning_rate": 0.00019627449719936303,
      "loss": 7.2633,
      "step": 77170
    },
    {
      "epoch": 9.28760529482551,
      "grad_norm": 2312.70654296875,
      "learning_rate": 0.00019627346841321824,
      "loss": 7.2438,
      "step": 77180
    },
    {
      "epoch": 9.288808664259928,
      "grad_norm": 4723.912109375,
      "learning_rate": 0.00019627243948774196,
      "loss": 7.2204,
      "step": 77190
    },
    {
      "epoch": 9.290012033694344,
      "grad_norm": 6762.6513671875,
      "learning_rate": 0.00019627141042293566,
      "loss": 7.2854,
      "step": 77200
    },
    {
      "epoch": 9.29121540312876,
      "grad_norm": 2534.995849609375,
      "learning_rate": 0.00019627038121880087,
      "loss": 7.2793,
      "step": 77210
    },
    {
      "epoch": 9.292418772563177,
      "grad_norm": 10283.8681640625,
      "learning_rate": 0.00019626935187533912,
      "loss": 7.1858,
      "step": 77220
    },
    {
      "epoch": 9.293622141997593,
      "grad_norm": 4278.6865234375,
      "learning_rate": 0.0001962683223925518,
      "loss": 7.2183,
      "step": 77230
    },
    {
      "epoch": 9.294825511432009,
      "grad_norm": 12018.0615234375,
      "learning_rate": 0.00019626729277044046,
      "loss": 7.2766,
      "step": 77240
    },
    {
      "epoch": 9.296028880866427,
      "grad_norm": 5349.31640625,
      "learning_rate": 0.0001962662630090066,
      "loss": 7.2706,
      "step": 77250
    },
    {
      "epoch": 9.297232250300842,
      "grad_norm": 3850.95068359375,
      "learning_rate": 0.00019626523310825163,
      "loss": 7.2422,
      "step": 77260
    },
    {
      "epoch": 9.298435619735258,
      "grad_norm": 4175.66015625,
      "learning_rate": 0.00019626420306817716,
      "loss": 7.2801,
      "step": 77270
    },
    {
      "epoch": 9.299638989169676,
      "grad_norm": 2477.304931640625,
      "learning_rate": 0.00019626317288878463,
      "loss": 7.2748,
      "step": 77280
    },
    {
      "epoch": 9.300842358604092,
      "grad_norm": 8570.326171875,
      "learning_rate": 0.00019626214257007548,
      "loss": 7.3768,
      "step": 77290
    },
    {
      "epoch": 9.302045728038507,
      "grad_norm": 5402.13232421875,
      "learning_rate": 0.00019626111211205126,
      "loss": 7.3386,
      "step": 77300
    },
    {
      "epoch": 9.303249097472925,
      "grad_norm": 3276.2119140625,
      "learning_rate": 0.00019626008151471346,
      "loss": 7.4104,
      "step": 77310
    },
    {
      "epoch": 9.30445246690734,
      "grad_norm": 4396.63525390625,
      "learning_rate": 0.00019625905077806354,
      "loss": 7.2342,
      "step": 77320
    },
    {
      "epoch": 9.305655836341757,
      "grad_norm": 3854.626220703125,
      "learning_rate": 0.00019625801990210299,
      "loss": 7.2903,
      "step": 77330
    },
    {
      "epoch": 9.306859205776174,
      "grad_norm": 8129.89404296875,
      "learning_rate": 0.00019625698888683335,
      "loss": 7.3334,
      "step": 77340
    },
    {
      "epoch": 9.30806257521059,
      "grad_norm": 6008.79248046875,
      "learning_rate": 0.00019625595773225607,
      "loss": 7.2288,
      "step": 77350
    },
    {
      "epoch": 9.309265944645006,
      "grad_norm": 6686.689453125,
      "learning_rate": 0.00019625492643837264,
      "loss": 7.2692,
      "step": 77360
    },
    {
      "epoch": 9.310469314079423,
      "grad_norm": 4833.21923828125,
      "learning_rate": 0.0001962538950051846,
      "loss": 7.2991,
      "step": 77370
    },
    {
      "epoch": 9.311672683513839,
      "grad_norm": 3234.591064453125,
      "learning_rate": 0.00019625286343269338,
      "loss": 7.172,
      "step": 77380
    },
    {
      "epoch": 9.312876052948255,
      "grad_norm": 5472.28125,
      "learning_rate": 0.00019625183172090053,
      "loss": 7.2136,
      "step": 77390
    },
    {
      "epoch": 9.314079422382672,
      "grad_norm": 8887.9833984375,
      "learning_rate": 0.0001962507998698075,
      "loss": 7.2837,
      "step": 77400
    },
    {
      "epoch": 9.315282791817088,
      "grad_norm": 3831.244384765625,
      "learning_rate": 0.00019624976787941577,
      "loss": 7.2685,
      "step": 77410
    },
    {
      "epoch": 9.316486161251504,
      "grad_norm": 2643.47705078125,
      "learning_rate": 0.0001962487357497269,
      "loss": 7.3147,
      "step": 77420
    },
    {
      "epoch": 9.31768953068592,
      "grad_norm": 4234.41015625,
      "learning_rate": 0.00019624770348074236,
      "loss": 7.405,
      "step": 77430
    },
    {
      "epoch": 9.318892900120337,
      "grad_norm": 9308.5439453125,
      "learning_rate": 0.00019624667107246357,
      "loss": 7.1495,
      "step": 77440
    },
    {
      "epoch": 9.320096269554753,
      "grad_norm": 8027.15185546875,
      "learning_rate": 0.00019624563852489212,
      "loss": 7.3376,
      "step": 77450
    },
    {
      "epoch": 9.321299638989169,
      "grad_norm": 6350.17236328125,
      "learning_rate": 0.00019624460583802947,
      "loss": 7.4338,
      "step": 77460
    },
    {
      "epoch": 9.322503008423586,
      "grad_norm": 11119.685546875,
      "learning_rate": 0.00019624357301187715,
      "loss": 7.3742,
      "step": 77470
    },
    {
      "epoch": 9.323706377858002,
      "grad_norm": 4849.087890625,
      "learning_rate": 0.00019624254004643656,
      "loss": 7.4504,
      "step": 77480
    },
    {
      "epoch": 9.324909747292418,
      "grad_norm": 7170.11572265625,
      "learning_rate": 0.00019624150694170925,
      "loss": 7.6526,
      "step": 77490
    },
    {
      "epoch": 9.326113116726836,
      "grad_norm": 4548.6494140625,
      "learning_rate": 0.00019624047369769677,
      "loss": 7.4069,
      "step": 77500
    },
    {
      "epoch": 9.327316486161251,
      "grad_norm": 3030.703125,
      "learning_rate": 0.00019623944031440054,
      "loss": 7.3581,
      "step": 77510
    },
    {
      "epoch": 9.328519855595667,
      "grad_norm": 3149.457763671875,
      "learning_rate": 0.00019623840679182207,
      "loss": 7.3331,
      "step": 77520
    },
    {
      "epoch": 9.329723225030085,
      "grad_norm": 8242.6728515625,
      "learning_rate": 0.0001962373731299629,
      "loss": 7.3794,
      "step": 77530
    },
    {
      "epoch": 9.3309265944645,
      "grad_norm": 4847.974609375,
      "learning_rate": 0.00019623633932882445,
      "loss": 7.442,
      "step": 77540
    },
    {
      "epoch": 9.332129963898916,
      "grad_norm": 2945.122802734375,
      "learning_rate": 0.00019623530538840825,
      "loss": 7.2981,
      "step": 77550
    },
    {
      "epoch": 9.333333333333334,
      "grad_norm": 2150.29541015625,
      "learning_rate": 0.00019623427130871583,
      "loss": 7.3586,
      "step": 77560
    },
    {
      "epoch": 9.33453670276775,
      "grad_norm": 3948.70947265625,
      "learning_rate": 0.00019623323708974866,
      "loss": 7.1776,
      "step": 77570
    },
    {
      "epoch": 9.335740072202166,
      "grad_norm": 8347.814453125,
      "learning_rate": 0.00019623220273150822,
      "loss": 7.3039,
      "step": 77580
    },
    {
      "epoch": 9.336943441636583,
      "grad_norm": 5592.7275390625,
      "learning_rate": 0.00019623116823399604,
      "loss": 7.3945,
      "step": 77590
    },
    {
      "epoch": 9.338146811070999,
      "grad_norm": 4071.8798828125,
      "learning_rate": 0.00019623013359721357,
      "loss": 7.2104,
      "step": 77600
    },
    {
      "epoch": 9.339350180505415,
      "grad_norm": 3863.02099609375,
      "learning_rate": 0.00019622909882116237,
      "loss": 7.4239,
      "step": 77610
    },
    {
      "epoch": 9.340553549939832,
      "grad_norm": 8054.66162109375,
      "learning_rate": 0.0001962280639058439,
      "loss": 7.3631,
      "step": 77620
    },
    {
      "epoch": 9.341756919374248,
      "grad_norm": 4694.1416015625,
      "learning_rate": 0.00019622702885125967,
      "loss": 7.3631,
      "step": 77630
    },
    {
      "epoch": 9.342960288808664,
      "grad_norm": 6013.3544921875,
      "learning_rate": 0.00019622599365741117,
      "loss": 7.3621,
      "step": 77640
    },
    {
      "epoch": 9.344163658243081,
      "grad_norm": 4333.583984375,
      "learning_rate": 0.00019622495832429988,
      "loss": 7.2826,
      "step": 77650
    },
    {
      "epoch": 9.345367027677497,
      "grad_norm": 2671.57763671875,
      "learning_rate": 0.00019622392285192732,
      "loss": 7.2949,
      "step": 77660
    },
    {
      "epoch": 9.346570397111913,
      "grad_norm": 4898.59716796875,
      "learning_rate": 0.000196222887240295,
      "loss": 7.3193,
      "step": 77670
    },
    {
      "epoch": 9.34777376654633,
      "grad_norm": 5069.6416015625,
      "learning_rate": 0.0001962218514894044,
      "loss": 7.2262,
      "step": 77680
    },
    {
      "epoch": 9.348977135980746,
      "grad_norm": 6325.212890625,
      "learning_rate": 0.00019622081559925704,
      "loss": 7.2731,
      "step": 77690
    },
    {
      "epoch": 9.350180505415162,
      "grad_norm": 4033.055419921875,
      "learning_rate": 0.0001962197795698544,
      "loss": 7.2964,
      "step": 77700
    },
    {
      "epoch": 9.35138387484958,
      "grad_norm": 3727.21875,
      "learning_rate": 0.00019621874340119798,
      "loss": 7.2466,
      "step": 77710
    },
    {
      "epoch": 9.352587244283995,
      "grad_norm": 3326.578857421875,
      "learning_rate": 0.00019621770709328927,
      "loss": 7.2284,
      "step": 77720
    },
    {
      "epoch": 9.353790613718411,
      "grad_norm": 3845.493408203125,
      "learning_rate": 0.00019621667064612978,
      "loss": 7.2972,
      "step": 77730
    },
    {
      "epoch": 9.354993983152827,
      "grad_norm": 6615.982421875,
      "learning_rate": 0.00019621563405972103,
      "loss": 7.2749,
      "step": 77740
    },
    {
      "epoch": 9.356197352587245,
      "grad_norm": 5356.36328125,
      "learning_rate": 0.00019621459733406448,
      "loss": 7.3435,
      "step": 77750
    },
    {
      "epoch": 9.35740072202166,
      "grad_norm": 3604.121826171875,
      "learning_rate": 0.00019621356046916168,
      "loss": 7.2066,
      "step": 77760
    },
    {
      "epoch": 9.358604091456076,
      "grad_norm": 5260.88037109375,
      "learning_rate": 0.0001962125234650141,
      "loss": 7.23,
      "step": 77770
    },
    {
      "epoch": 9.359807460890494,
      "grad_norm": 1868.055908203125,
      "learning_rate": 0.0001962114863216232,
      "loss": 7.2848,
      "step": 77780
    },
    {
      "epoch": 9.36101083032491,
      "grad_norm": 2795.3427734375,
      "learning_rate": 0.00019621044903899058,
      "loss": 7.2138,
      "step": 77790
    },
    {
      "epoch": 9.362214199759325,
      "grad_norm": 8162.1064453125,
      "learning_rate": 0.00019620941161711767,
      "loss": 7.2871,
      "step": 77800
    },
    {
      "epoch": 9.363417569193743,
      "grad_norm": 1776.7362060546875,
      "learning_rate": 0.000196208374056006,
      "loss": 7.2871,
      "step": 77810
    },
    {
      "epoch": 9.364620938628159,
      "grad_norm": 5059.310546875,
      "learning_rate": 0.00019620733635565703,
      "loss": 7.2553,
      "step": 77820
    },
    {
      "epoch": 9.365824308062574,
      "grad_norm": 2266.853759765625,
      "learning_rate": 0.0001962062985160723,
      "loss": 7.3381,
      "step": 77830
    },
    {
      "epoch": 9.367027677496992,
      "grad_norm": 2640.204833984375,
      "learning_rate": 0.0001962052605372533,
      "loss": 7.2424,
      "step": 77840
    },
    {
      "epoch": 9.368231046931408,
      "grad_norm": 1352.280517578125,
      "learning_rate": 0.00019620422241920155,
      "loss": 7.1813,
      "step": 77850
    },
    {
      "epoch": 9.369434416365824,
      "grad_norm": 2594.178955078125,
      "learning_rate": 0.00019620318416191854,
      "loss": 7.233,
      "step": 77860
    },
    {
      "epoch": 9.370637785800241,
      "grad_norm": 5957.42529296875,
      "learning_rate": 0.00019620214576540577,
      "loss": 7.2733,
      "step": 77870
    },
    {
      "epoch": 9.371841155234657,
      "grad_norm": 3715.390625,
      "learning_rate": 0.00019620110722966475,
      "loss": 7.3269,
      "step": 77880
    },
    {
      "epoch": 9.373044524669073,
      "grad_norm": 2783.47607421875,
      "learning_rate": 0.00019620006855469694,
      "loss": 7.258,
      "step": 77890
    },
    {
      "epoch": 9.37424789410349,
      "grad_norm": 4849.11376953125,
      "learning_rate": 0.0001961990297405039,
      "loss": 7.1817,
      "step": 77900
    },
    {
      "epoch": 9.375451263537906,
      "grad_norm": 3400.09326171875,
      "learning_rate": 0.00019619799078708715,
      "loss": 7.2494,
      "step": 77910
    },
    {
      "epoch": 9.376654632972322,
      "grad_norm": 5464.45556640625,
      "learning_rate": 0.0001961969516944481,
      "loss": 7.1859,
      "step": 77920
    },
    {
      "epoch": 9.37785800240674,
      "grad_norm": 2846.5673828125,
      "learning_rate": 0.00019619591246258834,
      "loss": 7.3879,
      "step": 77930
    },
    {
      "epoch": 9.379061371841155,
      "grad_norm": 3701.296630859375,
      "learning_rate": 0.00019619487309150934,
      "loss": 7.2437,
      "step": 77940
    },
    {
      "epoch": 9.380264741275571,
      "grad_norm": 2765.167724609375,
      "learning_rate": 0.00019619383358121262,
      "loss": 7.2886,
      "step": 77950
    },
    {
      "epoch": 9.381468110709989,
      "grad_norm": 2480.395751953125,
      "learning_rate": 0.00019619279393169966,
      "loss": 7.2509,
      "step": 77960
    },
    {
      "epoch": 9.382671480144404,
      "grad_norm": 1947.8099365234375,
      "learning_rate": 0.00019619175414297197,
      "loss": 7.3437,
      "step": 77970
    },
    {
      "epoch": 9.38387484957882,
      "grad_norm": 2265.4072265625,
      "learning_rate": 0.0001961907142150311,
      "loss": 7.1724,
      "step": 77980
    },
    {
      "epoch": 9.385078219013238,
      "grad_norm": 6581.12255859375,
      "learning_rate": 0.00019618967414787846,
      "loss": 7.3022,
      "step": 77990
    },
    {
      "epoch": 9.386281588447654,
      "grad_norm": 3810.63330078125,
      "learning_rate": 0.00019618863394151566,
      "loss": 7.259,
      "step": 78000
    },
    {
      "epoch": 9.38748495788207,
      "grad_norm": 4691.70166015625,
      "learning_rate": 0.0001961875935959441,
      "loss": 7.2887,
      "step": 78010
    },
    {
      "epoch": 9.388688327316487,
      "grad_norm": 2855.01025390625,
      "learning_rate": 0.00019618655311116543,
      "loss": 7.245,
      "step": 78020
    },
    {
      "epoch": 9.389891696750903,
      "grad_norm": 3010.580078125,
      "learning_rate": 0.00019618551248718104,
      "loss": 7.2509,
      "step": 78030
    },
    {
      "epoch": 9.391095066185319,
      "grad_norm": 4552.91162109375,
      "learning_rate": 0.0001961844717239924,
      "loss": 7.3897,
      "step": 78040
    },
    {
      "epoch": 9.392298435619736,
      "grad_norm": 10382.8818359375,
      "learning_rate": 0.00019618343082160116,
      "loss": 7.2379,
      "step": 78050
    },
    {
      "epoch": 9.393501805054152,
      "grad_norm": 10426.2216796875,
      "learning_rate": 0.0001961823897800087,
      "loss": 7.245,
      "step": 78060
    },
    {
      "epoch": 9.394705174488568,
      "grad_norm": 6853.20068359375,
      "learning_rate": 0.00019618134859921663,
      "loss": 7.3681,
      "step": 78070
    },
    {
      "epoch": 9.395908543922985,
      "grad_norm": 3953.904541015625,
      "learning_rate": 0.00019618030727922635,
      "loss": 7.4511,
      "step": 78080
    },
    {
      "epoch": 9.397111913357401,
      "grad_norm": 4179.099609375,
      "learning_rate": 0.00019617926582003945,
      "loss": 7.24,
      "step": 78090
    },
    {
      "epoch": 9.398315282791817,
      "grad_norm": 4626.548828125,
      "learning_rate": 0.00019617822422165739,
      "loss": 7.199,
      "step": 78100
    },
    {
      "epoch": 9.399518652226233,
      "grad_norm": 3491.727294921875,
      "learning_rate": 0.00019617718248408174,
      "loss": 7.2903,
      "step": 78110
    },
    {
      "epoch": 9.40072202166065,
      "grad_norm": 1451.7998046875,
      "learning_rate": 0.00019617614060731388,
      "loss": 7.2262,
      "step": 78120
    },
    {
      "epoch": 9.401925391095066,
      "grad_norm": 1606.47802734375,
      "learning_rate": 0.00019617509859135544,
      "loss": 7.317,
      "step": 78130
    },
    {
      "epoch": 9.403128760529482,
      "grad_norm": 1304.86572265625,
      "learning_rate": 0.0001961740564362079,
      "loss": 7.1825,
      "step": 78140
    },
    {
      "epoch": 9.4043321299639,
      "grad_norm": 1305.392578125,
      "learning_rate": 0.00019617301414187274,
      "loss": 7.2438,
      "step": 78150
    },
    {
      "epoch": 9.405535499398315,
      "grad_norm": 461.7162780761719,
      "learning_rate": 0.00019617197170835148,
      "loss": 7.3259,
      "step": 78160
    },
    {
      "epoch": 9.406738868832731,
      "grad_norm": 835.4100952148438,
      "learning_rate": 0.00019617092913564566,
      "loss": 7.2617,
      "step": 78170
    },
    {
      "epoch": 9.407942238267148,
      "grad_norm": 2162.604248046875,
      "learning_rate": 0.00019616988642375674,
      "loss": 7.3059,
      "step": 78180
    },
    {
      "epoch": 9.409145607701564,
      "grad_norm": 5187.7822265625,
      "learning_rate": 0.00019616884357268627,
      "loss": 7.2772,
      "step": 78190
    },
    {
      "epoch": 9.41034897713598,
      "grad_norm": 2610.12841796875,
      "learning_rate": 0.0001961678005824357,
      "loss": 7.3588,
      "step": 78200
    },
    {
      "epoch": 9.411552346570398,
      "grad_norm": 2318.993408203125,
      "learning_rate": 0.0001961667574530066,
      "loss": 7.4081,
      "step": 78210
    },
    {
      "epoch": 9.412755716004813,
      "grad_norm": 5456.6708984375,
      "learning_rate": 0.0001961657141844005,
      "loss": 7.2809,
      "step": 78220
    },
    {
      "epoch": 9.41395908543923,
      "grad_norm": 1491.068115234375,
      "learning_rate": 0.0001961646707766188,
      "loss": 7.2779,
      "step": 78230
    },
    {
      "epoch": 9.415162454873647,
      "grad_norm": 2164.580322265625,
      "learning_rate": 0.0001961636272296631,
      "loss": 7.1812,
      "step": 78240
    },
    {
      "epoch": 9.416365824308063,
      "grad_norm": 3099.32275390625,
      "learning_rate": 0.00019616258354353491,
      "loss": 7.3556,
      "step": 78250
    },
    {
      "epoch": 9.417569193742478,
      "grad_norm": 1073.731201171875,
      "learning_rate": 0.0001961615397182357,
      "loss": 7.3093,
      "step": 78260
    },
    {
      "epoch": 9.418772563176896,
      "grad_norm": 1488.4786376953125,
      "learning_rate": 0.00019616049575376702,
      "loss": 7.1745,
      "step": 78270
    },
    {
      "epoch": 9.419975932611312,
      "grad_norm": 835.6521606445312,
      "learning_rate": 0.00019615945165013033,
      "loss": 7.0926,
      "step": 78280
    },
    {
      "epoch": 9.421179302045728,
      "grad_norm": 1333.96142578125,
      "learning_rate": 0.00019615840740732724,
      "loss": 7.1712,
      "step": 78290
    },
    {
      "epoch": 9.422382671480145,
      "grad_norm": 2048.01416015625,
      "learning_rate": 0.00019615736302535912,
      "loss": 7.2467,
      "step": 78300
    },
    {
      "epoch": 9.42358604091456,
      "grad_norm": 759.7511596679688,
      "learning_rate": 0.0001961563185042276,
      "loss": 7.2253,
      "step": 78310
    },
    {
      "epoch": 9.424789410348977,
      "grad_norm": 1097.372314453125,
      "learning_rate": 0.0001961552738439341,
      "loss": 7.2261,
      "step": 78320
    },
    {
      "epoch": 9.425992779783394,
      "grad_norm": 602.5449829101562,
      "learning_rate": 0.00019615422904448022,
      "loss": 7.269,
      "step": 78330
    },
    {
      "epoch": 9.42719614921781,
      "grad_norm": 2424.92041015625,
      "learning_rate": 0.00019615318410586742,
      "loss": 7.2529,
      "step": 78340
    },
    {
      "epoch": 9.428399518652226,
      "grad_norm": 1713.84716796875,
      "learning_rate": 0.00019615213902809725,
      "loss": 7.1447,
      "step": 78350
    },
    {
      "epoch": 9.429602888086643,
      "grad_norm": 1154.8934326171875,
      "learning_rate": 0.00019615109381117117,
      "loss": 7.2958,
      "step": 78360
    },
    {
      "epoch": 9.43080625752106,
      "grad_norm": 3433.16455078125,
      "learning_rate": 0.0001961500484550907,
      "loss": 7.2972,
      "step": 78370
    },
    {
      "epoch": 9.432009626955475,
      "grad_norm": 2607.750732421875,
      "learning_rate": 0.00019614900295985738,
      "loss": 7.2306,
      "step": 78380
    },
    {
      "epoch": 9.433212996389893,
      "grad_norm": 2142.672119140625,
      "learning_rate": 0.00019614795732547272,
      "loss": 7.3421,
      "step": 78390
    },
    {
      "epoch": 9.434416365824308,
      "grad_norm": 2936.884033203125,
      "learning_rate": 0.00019614691155193822,
      "loss": 7.3912,
      "step": 78400
    },
    {
      "epoch": 9.435619735258724,
      "grad_norm": 326.2511901855469,
      "learning_rate": 0.00019614586563925543,
      "loss": 7.6074,
      "step": 78410
    },
    {
      "epoch": 9.43682310469314,
      "grad_norm": 962.3978271484375,
      "learning_rate": 0.0001961448195874258,
      "loss": 7.2849,
      "step": 78420
    },
    {
      "epoch": 9.438026474127557,
      "grad_norm": 540.7434692382812,
      "learning_rate": 0.0001961437733964509,
      "loss": 7.433,
      "step": 78430
    },
    {
      "epoch": 9.439229843561973,
      "grad_norm": 472.0730285644531,
      "learning_rate": 0.00019614272706633222,
      "loss": 7.4856,
      "step": 78440
    },
    {
      "epoch": 9.440433212996389,
      "grad_norm": 399.0505676269531,
      "learning_rate": 0.00019614168059707129,
      "loss": 7.5461,
      "step": 78450
    },
    {
      "epoch": 9.441636582430807,
      "grad_norm": 245.52577209472656,
      "learning_rate": 0.0001961406339886696,
      "loss": 7.5329,
      "step": 78460
    },
    {
      "epoch": 9.442839951865222,
      "grad_norm": 398.97039794921875,
      "learning_rate": 0.00019613958724112868,
      "loss": 7.4528,
      "step": 78470
    },
    {
      "epoch": 9.444043321299638,
      "grad_norm": 395.4848937988281,
      "learning_rate": 0.00019613854035445,
      "loss": 7.4018,
      "step": 78480
    },
    {
      "epoch": 9.445246690734056,
      "grad_norm": 275.9483642578125,
      "learning_rate": 0.00019613749332863517,
      "loss": 7.417,
      "step": 78490
    },
    {
      "epoch": 9.446450060168472,
      "grad_norm": 313.7587890625,
      "learning_rate": 0.00019613644616368563,
      "loss": 7.4118,
      "step": 78500
    },
    {
      "epoch": 9.447653429602887,
      "grad_norm": 440.3321838378906,
      "learning_rate": 0.0001961353988596029,
      "loss": 7.4619,
      "step": 78510
    },
    {
      "epoch": 9.448856799037305,
      "grad_norm": 371.33453369140625,
      "learning_rate": 0.00019613435141638855,
      "loss": 7.3197,
      "step": 78520
    },
    {
      "epoch": 9.45006016847172,
      "grad_norm": 146.487060546875,
      "learning_rate": 0.00019613330383404405,
      "loss": 7.3658,
      "step": 78530
    },
    {
      "epoch": 9.451263537906136,
      "grad_norm": 385.2295837402344,
      "learning_rate": 0.00019613225611257091,
      "loss": 7.3773,
      "step": 78540
    },
    {
      "epoch": 9.452466907340554,
      "grad_norm": 462.59490966796875,
      "learning_rate": 0.00019613120825197068,
      "loss": 7.3651,
      "step": 78550
    },
    {
      "epoch": 9.45367027677497,
      "grad_norm": 204.2476043701172,
      "learning_rate": 0.00019613016025224483,
      "loss": 7.3405,
      "step": 78560
    },
    {
      "epoch": 9.454873646209386,
      "grad_norm": 172.2794189453125,
      "learning_rate": 0.00019612911211339492,
      "loss": 7.2601,
      "step": 78570
    },
    {
      "epoch": 9.456077015643803,
      "grad_norm": 129.8483428955078,
      "learning_rate": 0.0001961280638354224,
      "loss": 7.2698,
      "step": 78580
    },
    {
      "epoch": 9.457280385078219,
      "grad_norm": 154.7813720703125,
      "learning_rate": 0.00019612701541832893,
      "loss": 7.3033,
      "step": 78590
    },
    {
      "epoch": 9.458483754512635,
      "grad_norm": 82.32933044433594,
      "learning_rate": 0.00019612596686211588,
      "loss": 7.3932,
      "step": 78600
    },
    {
      "epoch": 9.459687123947052,
      "grad_norm": 388.26068115234375,
      "learning_rate": 0.00019612491816678484,
      "loss": 7.3569,
      "step": 78610
    },
    {
      "epoch": 9.460890493381468,
      "grad_norm": 2829.243896484375,
      "learning_rate": 0.0001961238693323373,
      "loss": 7.684,
      "step": 78620
    },
    {
      "epoch": 9.462093862815884,
      "grad_norm": 3014.725341796875,
      "learning_rate": 0.00019612282035877477,
      "loss": 7.4296,
      "step": 78630
    },
    {
      "epoch": 9.463297232250302,
      "grad_norm": 22669.455078125,
      "learning_rate": 0.00019612177124609886,
      "loss": 8.0225,
      "step": 78640
    },
    {
      "epoch": 9.464500601684717,
      "grad_norm": 16556.046875,
      "learning_rate": 0.00019612072199431095,
      "loss": 8.7562,
      "step": 78650
    },
    {
      "epoch": 9.465703971119133,
      "grad_norm": 2059.367431640625,
      "learning_rate": 0.00019611967260341263,
      "loss": 7.8364,
      "step": 78660
    },
    {
      "epoch": 9.46690734055355,
      "grad_norm": 7453.05517578125,
      "learning_rate": 0.00019611862307340543,
      "loss": 8.4579,
      "step": 78670
    },
    {
      "epoch": 9.468110709987966,
      "grad_norm": 98882.0390625,
      "learning_rate": 0.00019611757340429087,
      "loss": 10.7681,
      "step": 78680
    },
    {
      "epoch": 9.469314079422382,
      "grad_norm": 5717.16796875,
      "learning_rate": 0.00019611652359607042,
      "loss": 9.2477,
      "step": 78690
    },
    {
      "epoch": 9.4705174488568,
      "grad_norm": 423.6444091796875,
      "learning_rate": 0.00019611547364874562,
      "loss": 9.2902,
      "step": 78700
    },
    {
      "epoch": 9.471720818291216,
      "grad_norm": 2988.169921875,
      "learning_rate": 0.000196114423562318,
      "loss": 8.2041,
      "step": 78710
    },
    {
      "epoch": 9.472924187725631,
      "grad_norm": 14273.8759765625,
      "learning_rate": 0.00019611337333678912,
      "loss": 8.4192,
      "step": 78720
    },
    {
      "epoch": 9.474127557160049,
      "grad_norm": 10774.318359375,
      "learning_rate": 0.00019611232297216043,
      "loss": 8.6146,
      "step": 78730
    },
    {
      "epoch": 9.475330926594465,
      "grad_norm": 18324.392578125,
      "learning_rate": 0.00019611127246843348,
      "loss": 8.4233,
      "step": 78740
    },
    {
      "epoch": 9.47653429602888,
      "grad_norm": 4517.78857421875,
      "learning_rate": 0.00019611022182560982,
      "loss": 7.8533,
      "step": 78750
    },
    {
      "epoch": 9.477737665463298,
      "grad_norm": 1231.1884765625,
      "learning_rate": 0.0001961091710436909,
      "loss": 7.6132,
      "step": 78760
    },
    {
      "epoch": 9.478941034897714,
      "grad_norm": 442.2496032714844,
      "learning_rate": 0.00019610812012267828,
      "loss": 7.6836,
      "step": 78770
    },
    {
      "epoch": 9.48014440433213,
      "grad_norm": 4462.33056640625,
      "learning_rate": 0.00019610706906257353,
      "loss": 7.512,
      "step": 78780
    },
    {
      "epoch": 9.481347773766545,
      "grad_norm": 87.68080139160156,
      "learning_rate": 0.00019610601786337805,
      "loss": 7.657,
      "step": 78790
    },
    {
      "epoch": 9.482551143200963,
      "grad_norm": 68.20985412597656,
      "learning_rate": 0.0001961049665250935,
      "loss": 7.55,
      "step": 78800
    },
    {
      "epoch": 9.483754512635379,
      "grad_norm": 16.40642738342285,
      "learning_rate": 0.00019610391504772129,
      "loss": 7.5414,
      "step": 78810
    },
    {
      "epoch": 9.484957882069795,
      "grad_norm": 29.987028121948242,
      "learning_rate": 0.00019610286343126304,
      "loss": 7.6145,
      "step": 78820
    },
    {
      "epoch": 9.486161251504212,
      "grad_norm": 14.566014289855957,
      "learning_rate": 0.00019610181167572014,
      "loss": 7.6461,
      "step": 78830
    },
    {
      "epoch": 9.487364620938628,
      "grad_norm": 12.040142059326172,
      "learning_rate": 0.00019610075978109426,
      "loss": 7.6969,
      "step": 78840
    },
    {
      "epoch": 9.488567990373044,
      "grad_norm": 6.859003067016602,
      "learning_rate": 0.00019609970774738683,
      "loss": 7.5266,
      "step": 78850
    },
    {
      "epoch": 9.489771359807461,
      "grad_norm": 6.0998358726501465,
      "learning_rate": 0.0001960986555745994,
      "loss": 7.5581,
      "step": 78860
    },
    {
      "epoch": 9.490974729241877,
      "grad_norm": 6.660614967346191,
      "learning_rate": 0.0001960976032627335,
      "loss": 7.5943,
      "step": 78870
    },
    {
      "epoch": 9.492178098676293,
      "grad_norm": 9.384659767150879,
      "learning_rate": 0.00019609655081179064,
      "loss": 7.5543,
      "step": 78880
    },
    {
      "epoch": 9.49338146811071,
      "grad_norm": 19.02987289428711,
      "learning_rate": 0.00019609549822177232,
      "loss": 7.6117,
      "step": 78890
    },
    {
      "epoch": 9.494584837545126,
      "grad_norm": 13.510137557983398,
      "learning_rate": 0.00019609444549268014,
      "loss": 7.664,
      "step": 78900
    },
    {
      "epoch": 9.495788206979542,
      "grad_norm": 105.62084197998047,
      "learning_rate": 0.00019609339262451552,
      "loss": 7.4671,
      "step": 78910
    },
    {
      "epoch": 9.49699157641396,
      "grad_norm": 29.44380760192871,
      "learning_rate": 0.00019609233961728005,
      "loss": 7.6215,
      "step": 78920
    },
    {
      "epoch": 9.498194945848375,
      "grad_norm": 43.94749069213867,
      "learning_rate": 0.00019609128647097524,
      "loss": 7.464,
      "step": 78930
    },
    {
      "epoch": 9.499398315282791,
      "grad_norm": 12.527983665466309,
      "learning_rate": 0.0001960902331856026,
      "loss": 7.5784,
      "step": 78940
    },
    {
      "epoch": 9.500601684717209,
      "grad_norm": 28.495134353637695,
      "learning_rate": 0.0001960891797611637,
      "loss": 7.5494,
      "step": 78950
    },
    {
      "epoch": 9.501805054151625,
      "grad_norm": 10.201322555541992,
      "learning_rate": 0.00019608812619766,
      "loss": 7.5191,
      "step": 78960
    },
    {
      "epoch": 9.50300842358604,
      "grad_norm": 27.35040283203125,
      "learning_rate": 0.00019608707249509308,
      "loss": 7.5623,
      "step": 78970
    },
    {
      "epoch": 9.504211793020458,
      "grad_norm": 8.544074058532715,
      "learning_rate": 0.00019608601865346445,
      "loss": 7.5798,
      "step": 78980
    },
    {
      "epoch": 9.505415162454874,
      "grad_norm": 34.04827117919922,
      "learning_rate": 0.00019608496467277565,
      "loss": 7.459,
      "step": 78990
    },
    {
      "epoch": 9.50661853188929,
      "grad_norm": 12.461686134338379,
      "learning_rate": 0.00019608391055302813,
      "loss": 7.6613,
      "step": 79000
    },
    {
      "epoch": 9.507821901323707,
      "grad_norm": 11.033659934997559,
      "learning_rate": 0.0001960828562942235,
      "loss": 7.5834,
      "step": 79010
    },
    {
      "epoch": 9.509025270758123,
      "grad_norm": 15.903694152832031,
      "learning_rate": 0.00019608180189636322,
      "loss": 7.5608,
      "step": 79020
    },
    {
      "epoch": 9.510228640192539,
      "grad_norm": 10.244565963745117,
      "learning_rate": 0.0001960807473594489,
      "loss": 7.6276,
      "step": 79030
    },
    {
      "epoch": 9.511432009626956,
      "grad_norm": 26.45344352722168,
      "learning_rate": 0.000196079692683482,
      "loss": 7.6082,
      "step": 79040
    },
    {
      "epoch": 9.512635379061372,
      "grad_norm": 14.464168548583984,
      "learning_rate": 0.00019607863786846405,
      "loss": 7.5861,
      "step": 79050
    },
    {
      "epoch": 9.513838748495788,
      "grad_norm": 12.145282745361328,
      "learning_rate": 0.0001960775829143966,
      "loss": 7.5265,
      "step": 79060
    },
    {
      "epoch": 9.515042117930205,
      "grad_norm": 23.147916793823242,
      "learning_rate": 0.00019607652782128117,
      "loss": 7.6055,
      "step": 79070
    },
    {
      "epoch": 9.516245487364621,
      "grad_norm": 19.37364959716797,
      "learning_rate": 0.00019607547258911927,
      "loss": 7.5509,
      "step": 79080
    },
    {
      "epoch": 9.517448856799037,
      "grad_norm": 45.35286331176758,
      "learning_rate": 0.00019607441721791245,
      "loss": 7.6603,
      "step": 79090
    },
    {
      "epoch": 9.518652226233453,
      "grad_norm": 37.83170700073242,
      "learning_rate": 0.00019607336170766223,
      "loss": 7.5165,
      "step": 79100
    },
    {
      "epoch": 9.51985559566787,
      "grad_norm": 15.612932205200195,
      "learning_rate": 0.00019607230605837015,
      "loss": 7.5318,
      "step": 79110
    },
    {
      "epoch": 9.521058965102286,
      "grad_norm": 30.468122482299805,
      "learning_rate": 0.00019607125027003769,
      "loss": 7.5491,
      "step": 79120
    },
    {
      "epoch": 9.522262334536702,
      "grad_norm": 7.987763404846191,
      "learning_rate": 0.00019607019434266643,
      "loss": 7.546,
      "step": 79130
    },
    {
      "epoch": 9.52346570397112,
      "grad_norm": 13.636329650878906,
      "learning_rate": 0.00019606913827625787,
      "loss": 7.575,
      "step": 79140
    },
    {
      "epoch": 9.524669073405535,
      "grad_norm": 14.846640586853027,
      "learning_rate": 0.00019606808207081359,
      "loss": 7.5855,
      "step": 79150
    },
    {
      "epoch": 9.525872442839951,
      "grad_norm": 19.705636978149414,
      "learning_rate": 0.000196067025726335,
      "loss": 7.4977,
      "step": 79160
    },
    {
      "epoch": 9.527075812274369,
      "grad_norm": 26.944072723388672,
      "learning_rate": 0.00019606596924282376,
      "loss": 7.6127,
      "step": 79170
    },
    {
      "epoch": 9.528279181708784,
      "grad_norm": 39.36925506591797,
      "learning_rate": 0.00019606491262028132,
      "loss": 7.4681,
      "step": 79180
    },
    {
      "epoch": 9.5294825511432,
      "grad_norm": 75.97850036621094,
      "learning_rate": 0.00019606385585870924,
      "loss": 7.6021,
      "step": 79190
    },
    {
      "epoch": 9.530685920577618,
      "grad_norm": 34.29001998901367,
      "learning_rate": 0.00019606279895810908,
      "loss": 7.4164,
      "step": 79200
    },
    {
      "epoch": 9.531889290012034,
      "grad_norm": 36.16572952270508,
      "learning_rate": 0.0001960617419184823,
      "loss": 7.3311,
      "step": 79210
    },
    {
      "epoch": 9.53309265944645,
      "grad_norm": 77.1339340209961,
      "learning_rate": 0.00019606068473983047,
      "loss": 7.4225,
      "step": 79220
    },
    {
      "epoch": 9.534296028880867,
      "grad_norm": 92.11058044433594,
      "learning_rate": 0.0001960596274221551,
      "loss": 7.4033,
      "step": 79230
    },
    {
      "epoch": 9.535499398315283,
      "grad_norm": 374.5948791503906,
      "learning_rate": 0.00019605856996545774,
      "loss": 7.4027,
      "step": 79240
    },
    {
      "epoch": 9.536702767749698,
      "grad_norm": 194.46343994140625,
      "learning_rate": 0.00019605751236973992,
      "loss": 7.351,
      "step": 79250
    },
    {
      "epoch": 9.537906137184116,
      "grad_norm": 121.67532348632812,
      "learning_rate": 0.00019605645463500318,
      "loss": 7.2614,
      "step": 79260
    },
    {
      "epoch": 9.539109506618532,
      "grad_norm": 104.19454956054688,
      "learning_rate": 0.000196055396761249,
      "loss": 7.3343,
      "step": 79270
    },
    {
      "epoch": 9.540312876052948,
      "grad_norm": 159.23812866210938,
      "learning_rate": 0.00019605433874847895,
      "loss": 7.4227,
      "step": 79280
    },
    {
      "epoch": 9.541516245487365,
      "grad_norm": 169.9329833984375,
      "learning_rate": 0.00019605328059669458,
      "loss": 7.2454,
      "step": 79290
    },
    {
      "epoch": 9.542719614921781,
      "grad_norm": 72.37603759765625,
      "learning_rate": 0.00019605222230589738,
      "loss": 7.2949,
      "step": 79300
    },
    {
      "epoch": 9.543922984356197,
      "grad_norm": 53.81119155883789,
      "learning_rate": 0.0001960511638760889,
      "loss": 7.4175,
      "step": 79310
    },
    {
      "epoch": 9.545126353790614,
      "grad_norm": 242.05604553222656,
      "learning_rate": 0.00019605010530727066,
      "loss": 7.3465,
      "step": 79320
    },
    {
      "epoch": 9.54632972322503,
      "grad_norm": 216.44239807128906,
      "learning_rate": 0.00019604904659944423,
      "loss": 7.2111,
      "step": 79330
    },
    {
      "epoch": 9.547533092659446,
      "grad_norm": 130.51792907714844,
      "learning_rate": 0.0001960479877526111,
      "loss": 7.2793,
      "step": 79340
    },
    {
      "epoch": 9.548736462093864,
      "grad_norm": 121.87419128417969,
      "learning_rate": 0.0001960469287667728,
      "loss": 7.3841,
      "step": 79350
    },
    {
      "epoch": 9.54993983152828,
      "grad_norm": 158.7510528564453,
      "learning_rate": 0.00019604586964193095,
      "loss": 7.3366,
      "step": 79360
    },
    {
      "epoch": 9.551143200962695,
      "grad_norm": 159.97264099121094,
      "learning_rate": 0.00019604481037808696,
      "loss": 7.1752,
      "step": 79370
    },
    {
      "epoch": 9.552346570397113,
      "grad_norm": 153.2975616455078,
      "learning_rate": 0.0001960437509752424,
      "loss": 7.2265,
      "step": 79380
    },
    {
      "epoch": 9.553549939831528,
      "grad_norm": 160.29055786132812,
      "learning_rate": 0.00019604269143339885,
      "loss": 7.2581,
      "step": 79390
    },
    {
      "epoch": 9.554753309265944,
      "grad_norm": 271.9485168457031,
      "learning_rate": 0.00019604163175255783,
      "loss": 7.2888,
      "step": 79400
    },
    {
      "epoch": 9.555956678700362,
      "grad_norm": 233.46905517578125,
      "learning_rate": 0.00019604057193272083,
      "loss": 7.3562,
      "step": 79410
    },
    {
      "epoch": 9.557160048134778,
      "grad_norm": 452.61572265625,
      "learning_rate": 0.0001960395119738894,
      "loss": 7.3663,
      "step": 79420
    },
    {
      "epoch": 9.558363417569193,
      "grad_norm": 157.4434356689453,
      "learning_rate": 0.0001960384518760651,
      "loss": 7.27,
      "step": 79430
    },
    {
      "epoch": 9.559566787003611,
      "grad_norm": 405.47259521484375,
      "learning_rate": 0.00019603739163924947,
      "loss": 7.2544,
      "step": 79440
    },
    {
      "epoch": 9.560770156438027,
      "grad_norm": 102.44945526123047,
      "learning_rate": 0.000196036331263444,
      "loss": 7.1992,
      "step": 79450
    },
    {
      "epoch": 9.561973525872443,
      "grad_norm": 143.56369018554688,
      "learning_rate": 0.00019603527074865023,
      "loss": 7.2965,
      "step": 79460
    },
    {
      "epoch": 9.56317689530686,
      "grad_norm": 190.5968475341797,
      "learning_rate": 0.00019603421009486974,
      "loss": 7.2631,
      "step": 79470
    },
    {
      "epoch": 9.564380264741276,
      "grad_norm": 133.09567260742188,
      "learning_rate": 0.00019603314930210402,
      "loss": 7.2565,
      "step": 79480
    },
    {
      "epoch": 9.565583634175692,
      "grad_norm": 73.62224578857422,
      "learning_rate": 0.00019603208837035464,
      "loss": 7.2491,
      "step": 79490
    },
    {
      "epoch": 9.566787003610107,
      "grad_norm": 157.73590087890625,
      "learning_rate": 0.00019603102729962311,
      "loss": 7.2924,
      "step": 79500
    },
    {
      "epoch": 9.567990373044525,
      "grad_norm": 1052.1524658203125,
      "learning_rate": 0.00019602996608991097,
      "loss": 7.3758,
      "step": 79510
    },
    {
      "epoch": 9.56919374247894,
      "grad_norm": 10060.75390625,
      "learning_rate": 0.00019602890474121977,
      "loss": 7.3981,
      "step": 79520
    },
    {
      "epoch": 9.570397111913357,
      "grad_norm": 30700.2890625,
      "learning_rate": 0.00019602784325355104,
      "loss": 7.7364,
      "step": 79530
    },
    {
      "epoch": 9.571600481347774,
      "grad_norm": 282.262451171875,
      "learning_rate": 0.00019602678162690632,
      "loss": 7.9551,
      "step": 79540
    },
    {
      "epoch": 9.57280385078219,
      "grad_norm": 11075.5908203125,
      "learning_rate": 0.0001960257198612871,
      "loss": 7.7967,
      "step": 79550
    },
    {
      "epoch": 9.574007220216606,
      "grad_norm": 39592.89453125,
      "learning_rate": 0.00019602465795669498,
      "loss": 7.5915,
      "step": 79560
    },
    {
      "epoch": 9.575210589651023,
      "grad_norm": 951.6878662109375,
      "learning_rate": 0.00019602359591313148,
      "loss": 7.6229,
      "step": 79570
    },
    {
      "epoch": 9.57641395908544,
      "grad_norm": 2799.107177734375,
      "learning_rate": 0.00019602253373059814,
      "loss": 7.7397,
      "step": 79580
    },
    {
      "epoch": 9.577617328519855,
      "grad_norm": 1263.59521484375,
      "learning_rate": 0.00019602147140909644,
      "loss": 7.7926,
      "step": 79590
    },
    {
      "epoch": 9.578820697954272,
      "grad_norm": 13267.4990234375,
      "learning_rate": 0.00019602040894862798,
      "loss": 7.8047,
      "step": 79600
    },
    {
      "epoch": 9.580024067388688,
      "grad_norm": 5765.8056640625,
      "learning_rate": 0.0001960193463491943,
      "loss": 7.9076,
      "step": 79610
    },
    {
      "epoch": 9.581227436823104,
      "grad_norm": 46395.68359375,
      "learning_rate": 0.00019601828361079694,
      "loss": 7.9171,
      "step": 79620
    },
    {
      "epoch": 9.582430806257522,
      "grad_norm": 42442.5625,
      "learning_rate": 0.00019601722073343738,
      "loss": 7.8046,
      "step": 79630
    },
    {
      "epoch": 9.583634175691937,
      "grad_norm": 14398.544921875,
      "learning_rate": 0.0001960161577171172,
      "loss": 7.7582,
      "step": 79640
    },
    {
      "epoch": 9.584837545126353,
      "grad_norm": 319.1088562011719,
      "learning_rate": 0.00019601509456183792,
      "loss": 7.7495,
      "step": 79650
    },
    {
      "epoch": 9.58604091456077,
      "grad_norm": 2.814500570297241,
      "learning_rate": 0.00019601403126760114,
      "loss": 7.5919,
      "step": 79660
    },
    {
      "epoch": 9.587244283995187,
      "grad_norm": 2.316964626312256,
      "learning_rate": 0.0001960129678344083,
      "loss": 7.6064,
      "step": 79670
    },
    {
      "epoch": 9.588447653429602,
      "grad_norm": 2.642634868621826,
      "learning_rate": 0.00019601190426226105,
      "loss": 7.6044,
      "step": 79680
    },
    {
      "epoch": 9.58965102286402,
      "grad_norm": 6.033572196960449,
      "learning_rate": 0.00019601084055116084,
      "loss": 7.5677,
      "step": 79690
    },
    {
      "epoch": 9.590854392298436,
      "grad_norm": 4.895756244659424,
      "learning_rate": 0.00019600977670110922,
      "loss": 7.5958,
      "step": 79700
    },
    {
      "epoch": 9.592057761732852,
      "grad_norm": 5.3907470703125,
      "learning_rate": 0.00019600871271210774,
      "loss": 7.5904,
      "step": 79710
    },
    {
      "epoch": 9.593261131167269,
      "grad_norm": 3.6726460456848145,
      "learning_rate": 0.00019600764858415797,
      "loss": 7.5695,
      "step": 79720
    },
    {
      "epoch": 9.594464500601685,
      "grad_norm": 3.873385190963745,
      "learning_rate": 0.00019600658431726144,
      "loss": 7.5725,
      "step": 79730
    },
    {
      "epoch": 9.5956678700361,
      "grad_norm": 4.151281833648682,
      "learning_rate": 0.00019600551991141968,
      "loss": 7.6107,
      "step": 79740
    },
    {
      "epoch": 9.596871239470518,
      "grad_norm": 9.565984725952148,
      "learning_rate": 0.00019600445536663421,
      "loss": 7.6013,
      "step": 79750
    },
    {
      "epoch": 9.598074608904934,
      "grad_norm": 16.456396102905273,
      "learning_rate": 0.0001960033906829066,
      "loss": 7.5136,
      "step": 79760
    },
    {
      "epoch": 9.59927797833935,
      "grad_norm": 3.6445560455322266,
      "learning_rate": 0.0001960023258602384,
      "loss": 7.5437,
      "step": 79770
    },
    {
      "epoch": 9.600481347773766,
      "grad_norm": 6.523273468017578,
      "learning_rate": 0.0001960012608986311,
      "loss": 7.6049,
      "step": 79780
    },
    {
      "epoch": 9.601684717208183,
      "grad_norm": 947.410888671875,
      "learning_rate": 0.0001960001957980863,
      "loss": 7.4322,
      "step": 79790
    },
    {
      "epoch": 9.602888086642599,
      "grad_norm": 4.064530849456787,
      "learning_rate": 0.0001959991305586055,
      "loss": 7.4855,
      "step": 79800
    },
    {
      "epoch": 9.604091456077015,
      "grad_norm": 11.220860481262207,
      "learning_rate": 0.00019599806518019025,
      "loss": 7.5319,
      "step": 79810
    },
    {
      "epoch": 9.605294825511432,
      "grad_norm": 18.553869247436523,
      "learning_rate": 0.00019599699966284213,
      "loss": 7.5367,
      "step": 79820
    },
    {
      "epoch": 9.606498194945848,
      "grad_norm": 10.639506340026855,
      "learning_rate": 0.0001959959340065626,
      "loss": 7.554,
      "step": 79830
    },
    {
      "epoch": 9.607701564380264,
      "grad_norm": 28.654579162597656,
      "learning_rate": 0.0001959948682113533,
      "loss": 7.5017,
      "step": 79840
    },
    {
      "epoch": 9.608904933814681,
      "grad_norm": 21.98432731628418,
      "learning_rate": 0.00019599380227721568,
      "loss": 7.3128,
      "step": 79850
    },
    {
      "epoch": 9.610108303249097,
      "grad_norm": 22.034852981567383,
      "learning_rate": 0.00019599273620415138,
      "loss": 7.307,
      "step": 79860
    },
    {
      "epoch": 9.611311672683513,
      "grad_norm": 21.325101852416992,
      "learning_rate": 0.00019599166999216186,
      "loss": 7.4465,
      "step": 79870
    },
    {
      "epoch": 9.61251504211793,
      "grad_norm": 42.87892150878906,
      "learning_rate": 0.0001959906036412487,
      "loss": 7.3836,
      "step": 79880
    },
    {
      "epoch": 9.613718411552346,
      "grad_norm": 31.864187240600586,
      "learning_rate": 0.00019598953715141345,
      "loss": 7.4534,
      "step": 79890
    },
    {
      "epoch": 9.614921780986762,
      "grad_norm": 26.607751846313477,
      "learning_rate": 0.00019598847052265761,
      "loss": 7.3211,
      "step": 79900
    },
    {
      "epoch": 9.61612515042118,
      "grad_norm": 59.20132064819336,
      "learning_rate": 0.00019598740375498277,
      "loss": 7.3567,
      "step": 79910
    },
    {
      "epoch": 9.617328519855596,
      "grad_norm": 41.33769607543945,
      "learning_rate": 0.00019598633684839048,
      "loss": 7.441,
      "step": 79920
    },
    {
      "epoch": 9.618531889290011,
      "grad_norm": 20.586872100830078,
      "learning_rate": 0.0001959852698028822,
      "loss": 7.3002,
      "step": 79930
    },
    {
      "epoch": 9.619735258724429,
      "grad_norm": 11.548286437988281,
      "learning_rate": 0.0001959842026184596,
      "loss": 7.3633,
      "step": 79940
    },
    {
      "epoch": 9.620938628158845,
      "grad_norm": 84.25369262695312,
      "learning_rate": 0.00019598313529512414,
      "loss": 7.3129,
      "step": 79950
    },
    {
      "epoch": 9.62214199759326,
      "grad_norm": 17.92951774597168,
      "learning_rate": 0.00019598206783287737,
      "loss": 7.2589,
      "step": 79960
    },
    {
      "epoch": 9.623345367027678,
      "grad_norm": 14.395515441894531,
      "learning_rate": 0.00019598100023172089,
      "loss": 7.2708,
      "step": 79970
    },
    {
      "epoch": 9.624548736462094,
      "grad_norm": 24.21454429626465,
      "learning_rate": 0.00019597993249165617,
      "loss": 7.2915,
      "step": 79980
    },
    {
      "epoch": 9.62575210589651,
      "grad_norm": 72.24156951904297,
      "learning_rate": 0.00019597886461268483,
      "loss": 7.306,
      "step": 79990
    },
    {
      "epoch": 9.626955475330927,
      "grad_norm": 30.216567993164062,
      "learning_rate": 0.00019597779659480833,
      "loss": 7.3008,
      "step": 80000
    },
    {
      "epoch": 9.628158844765343,
      "grad_norm": 38.20553970336914,
      "learning_rate": 0.0001959767284380283,
      "loss": 7.2473,
      "step": 80010
    },
    {
      "epoch": 9.629362214199759,
      "grad_norm": 54.55497741699219,
      "learning_rate": 0.0001959756601423462,
      "loss": 7.3417,
      "step": 80020
    },
    {
      "epoch": 9.630565583634176,
      "grad_norm": 20.382944107055664,
      "learning_rate": 0.00019597459170776367,
      "loss": 7.147,
      "step": 80030
    },
    {
      "epoch": 9.631768953068592,
      "grad_norm": 46.73661422729492,
      "learning_rate": 0.00019597352313428218,
      "loss": 7.3668,
      "step": 80040
    },
    {
      "epoch": 9.632972322503008,
      "grad_norm": 18.51910972595215,
      "learning_rate": 0.00019597245442190333,
      "loss": 7.2688,
      "step": 80050
    },
    {
      "epoch": 9.634175691937426,
      "grad_norm": 36.11069869995117,
      "learning_rate": 0.00019597138557062862,
      "loss": 7.2074,
      "step": 80060
    },
    {
      "epoch": 9.635379061371841,
      "grad_norm": 30.255470275878906,
      "learning_rate": 0.00019597031658045966,
      "loss": 7.2938,
      "step": 80070
    },
    {
      "epoch": 9.636582430806257,
      "grad_norm": 10.197362899780273,
      "learning_rate": 0.00019596924745139793,
      "loss": 7.4057,
      "step": 80080
    },
    {
      "epoch": 9.637785800240675,
      "grad_norm": 23.368932723999023,
      "learning_rate": 0.000195968178183445,
      "loss": 7.2778,
      "step": 80090
    },
    {
      "epoch": 9.63898916967509,
      "grad_norm": 19.56130599975586,
      "learning_rate": 0.00019596710877660243,
      "loss": 7.3187,
      "step": 80100
    },
    {
      "epoch": 9.640192539109506,
      "grad_norm": 24.682722091674805,
      "learning_rate": 0.00019596603923087175,
      "loss": 7.3107,
      "step": 80110
    },
    {
      "epoch": 9.641395908543924,
      "grad_norm": 43.84751510620117,
      "learning_rate": 0.00019596496954625453,
      "loss": 7.3186,
      "step": 80120
    },
    {
      "epoch": 9.64259927797834,
      "grad_norm": 15.692676544189453,
      "learning_rate": 0.00019596389972275234,
      "loss": 7.2783,
      "step": 80130
    },
    {
      "epoch": 9.643802647412755,
      "grad_norm": 6.348845958709717,
      "learning_rate": 0.0001959628297603666,
      "loss": 7.3805,
      "step": 80140
    },
    {
      "epoch": 9.645006016847173,
      "grad_norm": 2.7350292205810547,
      "learning_rate": 0.00019596175965909903,
      "loss": 7.2617,
      "step": 80150
    },
    {
      "epoch": 9.646209386281589,
      "grad_norm": 7.285604476928711,
      "learning_rate": 0.0001959606894189511,
      "loss": 7.4061,
      "step": 80160
    },
    {
      "epoch": 9.647412755716005,
      "grad_norm": 10.559433937072754,
      "learning_rate": 0.00019595961903992435,
      "loss": 7.3202,
      "step": 80170
    },
    {
      "epoch": 9.648616125150422,
      "grad_norm": 4.576714515686035,
      "learning_rate": 0.00019595854852202032,
      "loss": 7.3242,
      "step": 80180
    },
    {
      "epoch": 9.649819494584838,
      "grad_norm": 3.536768913269043,
      "learning_rate": 0.0001959574778652406,
      "loss": 7.2417,
      "step": 80190
    },
    {
      "epoch": 9.651022864019254,
      "grad_norm": 8.18758487701416,
      "learning_rate": 0.0001959564070695867,
      "loss": 7.1858,
      "step": 80200
    },
    {
      "epoch": 9.65222623345367,
      "grad_norm": 4.7222700119018555,
      "learning_rate": 0.0001959553361350602,
      "loss": 7.3187,
      "step": 80210
    },
    {
      "epoch": 9.653429602888087,
      "grad_norm": 42.41435241699219,
      "learning_rate": 0.00019595426506166264,
      "loss": 7.3391,
      "step": 80220
    },
    {
      "epoch": 9.654632972322503,
      "grad_norm": 15.976512908935547,
      "learning_rate": 0.00019595319384939556,
      "loss": 7.2505,
      "step": 80230
    },
    {
      "epoch": 9.655836341756919,
      "grad_norm": 6.87771463394165,
      "learning_rate": 0.00019595212249826053,
      "loss": 7.271,
      "step": 80240
    },
    {
      "epoch": 9.657039711191336,
      "grad_norm": 15.128507614135742,
      "learning_rate": 0.00019595105100825907,
      "loss": 7.2773,
      "step": 80250
    },
    {
      "epoch": 9.658243080625752,
      "grad_norm": 3.9212698936462402,
      "learning_rate": 0.00019594997937939277,
      "loss": 7.3445,
      "step": 80260
    },
    {
      "epoch": 9.659446450060168,
      "grad_norm": 6.612792015075684,
      "learning_rate": 0.00019594890761166317,
      "loss": 7.3511,
      "step": 80270
    },
    {
      "epoch": 9.660649819494585,
      "grad_norm": 9.729209899902344,
      "learning_rate": 0.00019594783570507177,
      "loss": 7.2522,
      "step": 80280
    },
    {
      "epoch": 9.661853188929001,
      "grad_norm": 25.043237686157227,
      "learning_rate": 0.0001959467636596202,
      "loss": 7.3197,
      "step": 80290
    },
    {
      "epoch": 9.663056558363417,
      "grad_norm": 189.68516540527344,
      "learning_rate": 0.00019594569147530995,
      "loss": 7.2938,
      "step": 80300
    },
    {
      "epoch": 9.664259927797834,
      "grad_norm": 301.1114807128906,
      "learning_rate": 0.00019594461915214262,
      "loss": 7.2735,
      "step": 80310
    },
    {
      "epoch": 9.66546329723225,
      "grad_norm": 61.184085845947266,
      "learning_rate": 0.00019594354669011973,
      "loss": 7.2298,
      "step": 80320
    },
    {
      "epoch": 9.666666666666666,
      "grad_norm": 14.759530067443848,
      "learning_rate": 0.00019594247408924283,
      "loss": 7.2912,
      "step": 80330
    },
    {
      "epoch": 9.667870036101084,
      "grad_norm": 32.6790657043457,
      "learning_rate": 0.00019594140134951348,
      "loss": 7.2896,
      "step": 80340
    },
    {
      "epoch": 9.6690734055355,
      "grad_norm": 10.808216094970703,
      "learning_rate": 0.00019594032847093326,
      "loss": 7.3244,
      "step": 80350
    },
    {
      "epoch": 9.670276774969915,
      "grad_norm": 6.880530834197998,
      "learning_rate": 0.00019593925545350366,
      "loss": 7.3362,
      "step": 80360
    },
    {
      "epoch": 9.671480144404333,
      "grad_norm": 10.589981079101562,
      "learning_rate": 0.0001959381822972263,
      "loss": 7.3277,
      "step": 80370
    },
    {
      "epoch": 9.672683513838749,
      "grad_norm": 17.26545524597168,
      "learning_rate": 0.0001959371090021027,
      "loss": 7.2339,
      "step": 80380
    },
    {
      "epoch": 9.673886883273164,
      "grad_norm": 30.79890251159668,
      "learning_rate": 0.00019593603556813442,
      "loss": 7.3052,
      "step": 80390
    },
    {
      "epoch": 9.675090252707582,
      "grad_norm": 9.122573852539062,
      "learning_rate": 0.00019593496199532299,
      "loss": 7.3575,
      "step": 80400
    },
    {
      "epoch": 9.676293622141998,
      "grad_norm": 10.84290885925293,
      "learning_rate": 0.00019593388828367,
      "loss": 7.3456,
      "step": 80410
    },
    {
      "epoch": 9.677496991576414,
      "grad_norm": 7.429494380950928,
      "learning_rate": 0.000195932814433177,
      "loss": 7.3102,
      "step": 80420
    },
    {
      "epoch": 9.678700361010831,
      "grad_norm": 7.6119303703308105,
      "learning_rate": 0.00019593174044384552,
      "loss": 7.2874,
      "step": 80430
    },
    {
      "epoch": 9.679903730445247,
      "grad_norm": 11.422502517700195,
      "learning_rate": 0.00019593066631567712,
      "loss": 7.2702,
      "step": 80440
    },
    {
      "epoch": 9.681107099879663,
      "grad_norm": 11.697199821472168,
      "learning_rate": 0.00019592959204867337,
      "loss": 7.2839,
      "step": 80450
    },
    {
      "epoch": 9.68231046931408,
      "grad_norm": 14.388079643249512,
      "learning_rate": 0.00019592851764283582,
      "loss": 7.3083,
      "step": 80460
    },
    {
      "epoch": 9.683513838748496,
      "grad_norm": 28.95720672607422,
      "learning_rate": 0.000195927443098166,
      "loss": 7.2277,
      "step": 80470
    },
    {
      "epoch": 9.684717208182912,
      "grad_norm": 8.118637084960938,
      "learning_rate": 0.0001959263684146655,
      "loss": 7.2757,
      "step": 80480
    },
    {
      "epoch": 9.685920577617328,
      "grad_norm": 20.11807632446289,
      "learning_rate": 0.00019592529359233586,
      "loss": 7.3231,
      "step": 80490
    },
    {
      "epoch": 9.687123947051745,
      "grad_norm": 7.092824459075928,
      "learning_rate": 0.00019592421863117863,
      "loss": 7.2876,
      "step": 80500
    },
    {
      "epoch": 9.688327316486161,
      "grad_norm": 5.425356864929199,
      "learning_rate": 0.00019592314353119542,
      "loss": 7.3649,
      "step": 80510
    },
    {
      "epoch": 9.689530685920577,
      "grad_norm": 5.126730918884277,
      "learning_rate": 0.00019592206829238768,
      "loss": 7.46,
      "step": 80520
    },
    {
      "epoch": 9.690734055354994,
      "grad_norm": 15.088762283325195,
      "learning_rate": 0.00019592099291475704,
      "loss": 7.3103,
      "step": 80530
    },
    {
      "epoch": 9.69193742478941,
      "grad_norm": 4.680660724639893,
      "learning_rate": 0.00019591991739830505,
      "loss": 7.2495,
      "step": 80540
    },
    {
      "epoch": 9.693140794223826,
      "grad_norm": 7.10062837600708,
      "learning_rate": 0.00019591884174303324,
      "loss": 7.2549,
      "step": 80550
    },
    {
      "epoch": 9.694344163658243,
      "grad_norm": 5.300331115722656,
      "learning_rate": 0.0001959177659489432,
      "loss": 7.1931,
      "step": 80560
    },
    {
      "epoch": 9.69554753309266,
      "grad_norm": 5.892553329467773,
      "learning_rate": 0.00019591669001603647,
      "loss": 7.3507,
      "step": 80570
    },
    {
      "epoch": 9.696750902527075,
      "grad_norm": 8.73146915435791,
      "learning_rate": 0.00019591561394431461,
      "loss": 7.266,
      "step": 80580
    },
    {
      "epoch": 9.697954271961493,
      "grad_norm": 11.49372386932373,
      "learning_rate": 0.00019591453773377916,
      "loss": 7.2896,
      "step": 80590
    },
    {
      "epoch": 9.699157641395908,
      "grad_norm": 4.181190490722656,
      "learning_rate": 0.00019591346138443175,
      "loss": 7.2576,
      "step": 80600
    },
    {
      "epoch": 9.700361010830324,
      "grad_norm": 1.6814228296279907,
      "learning_rate": 0.0001959123848962738,
      "loss": 7.2934,
      "step": 80610
    },
    {
      "epoch": 9.701564380264742,
      "grad_norm": 6.716777324676514,
      "learning_rate": 0.000195911308269307,
      "loss": 7.2646,
      "step": 80620
    },
    {
      "epoch": 9.702767749699158,
      "grad_norm": 1.973939061164856,
      "learning_rate": 0.00019591023150353286,
      "loss": 7.3238,
      "step": 80630
    },
    {
      "epoch": 9.703971119133573,
      "grad_norm": 1.9682577848434448,
      "learning_rate": 0.0001959091545989529,
      "loss": 7.2739,
      "step": 80640
    },
    {
      "epoch": 9.705174488567991,
      "grad_norm": 3.713451385498047,
      "learning_rate": 0.00019590807755556875,
      "loss": 7.3343,
      "step": 80650
    },
    {
      "epoch": 9.706377858002407,
      "grad_norm": 4.571086883544922,
      "learning_rate": 0.00019590700037338193,
      "loss": 7.2604,
      "step": 80660
    },
    {
      "epoch": 9.707581227436823,
      "grad_norm": 3.9480361938476562,
      "learning_rate": 0.00019590592305239398,
      "loss": 7.3839,
      "step": 80670
    },
    {
      "epoch": 9.70878459687124,
      "grad_norm": 5.151997089385986,
      "learning_rate": 0.00019590484559260646,
      "loss": 7.3549,
      "step": 80680
    },
    {
      "epoch": 9.709987966305656,
      "grad_norm": 5.794747352600098,
      "learning_rate": 0.00019590376799402098,
      "loss": 7.2307,
      "step": 80690
    },
    {
      "epoch": 9.711191335740072,
      "grad_norm": 10.18275260925293,
      "learning_rate": 0.0001959026902566391,
      "loss": 7.1675,
      "step": 80700
    },
    {
      "epoch": 9.71239470517449,
      "grad_norm": 3.547149896621704,
      "learning_rate": 0.00019590161238046232,
      "loss": 7.2141,
      "step": 80710
    },
    {
      "epoch": 9.713598074608905,
      "grad_norm": 3.9798226356506348,
      "learning_rate": 0.00019590053436549222,
      "loss": 7.2735,
      "step": 80720
    },
    {
      "epoch": 9.71480144404332,
      "grad_norm": 4.546346187591553,
      "learning_rate": 0.00019589945621173037,
      "loss": 7.3026,
      "step": 80730
    },
    {
      "epoch": 9.716004813477738,
      "grad_norm": 6.8329386711120605,
      "learning_rate": 0.00019589837791917836,
      "loss": 7.2736,
      "step": 80740
    },
    {
      "epoch": 9.717208182912154,
      "grad_norm": 6.138765811920166,
      "learning_rate": 0.00019589729948783769,
      "loss": 7.2282,
      "step": 80750
    },
    {
      "epoch": 9.71841155234657,
      "grad_norm": 4.7079691886901855,
      "learning_rate": 0.00019589622091770994,
      "loss": 7.1251,
      "step": 80760
    },
    {
      "epoch": 9.719614921780988,
      "grad_norm": 12.007296562194824,
      "learning_rate": 0.00019589514220879673,
      "loss": 7.2584,
      "step": 80770
    },
    {
      "epoch": 9.720818291215403,
      "grad_norm": 5.398209095001221,
      "learning_rate": 0.00019589406336109955,
      "loss": 7.1358,
      "step": 80780
    },
    {
      "epoch": 9.722021660649819,
      "grad_norm": 4.695461750030518,
      "learning_rate": 0.00019589298437461996,
      "loss": 7.2449,
      "step": 80790
    },
    {
      "epoch": 9.723225030084237,
      "grad_norm": 4.134794235229492,
      "learning_rate": 0.00019589190524935956,
      "loss": 7.1649,
      "step": 80800
    },
    {
      "epoch": 9.724428399518652,
      "grad_norm": 3.4911704063415527,
      "learning_rate": 0.0001958908259853199,
      "loss": 7.2282,
      "step": 80810
    },
    {
      "epoch": 9.725631768953068,
      "grad_norm": 7.523705005645752,
      "learning_rate": 0.00019588974658250256,
      "loss": 7.3258,
      "step": 80820
    },
    {
      "epoch": 9.726835138387486,
      "grad_norm": 5.453453540802002,
      "learning_rate": 0.00019588866704090907,
      "loss": 7.2819,
      "step": 80830
    },
    {
      "epoch": 9.728038507821902,
      "grad_norm": 13.083098411560059,
      "learning_rate": 0.000195887587360541,
      "loss": 7.2137,
      "step": 80840
    },
    {
      "epoch": 9.729241877256317,
      "grad_norm": 5.979798316955566,
      "learning_rate": 0.00019588650754139995,
      "loss": 7.156,
      "step": 80850
    },
    {
      "epoch": 9.730445246690735,
      "grad_norm": 3.339381694793701,
      "learning_rate": 0.00019588542758348741,
      "loss": 7.2267,
      "step": 80860
    },
    {
      "epoch": 9.73164861612515,
      "grad_norm": 3.9839985370635986,
      "learning_rate": 0.00019588434748680497,
      "loss": 7.1336,
      "step": 80870
    },
    {
      "epoch": 9.732851985559567,
      "grad_norm": 3.4965097904205322,
      "learning_rate": 0.00019588326725135423,
      "loss": 7.2444,
      "step": 80880
    },
    {
      "epoch": 9.734055354993982,
      "grad_norm": 4.205904006958008,
      "learning_rate": 0.00019588218687713674,
      "loss": 7.2232,
      "step": 80890
    },
    {
      "epoch": 9.7352587244284,
      "grad_norm": 5.698105335235596,
      "learning_rate": 0.00019588110636415405,
      "loss": 7.2114,
      "step": 80900
    },
    {
      "epoch": 9.736462093862816,
      "grad_norm": 12.120505332946777,
      "learning_rate": 0.0001958800257124077,
      "loss": 7.2099,
      "step": 80910
    },
    {
      "epoch": 9.737665463297231,
      "grad_norm": 5.2960662841796875,
      "learning_rate": 0.00019587894492189932,
      "loss": 7.2902,
      "step": 80920
    },
    {
      "epoch": 9.738868832731649,
      "grad_norm": 3.8109209537506104,
      "learning_rate": 0.0001958778639926304,
      "loss": 7.1597,
      "step": 80930
    },
    {
      "epoch": 9.740072202166065,
      "grad_norm": 5.949132919311523,
      "learning_rate": 0.00019587678292460256,
      "loss": 7.1693,
      "step": 80940
    },
    {
      "epoch": 9.74127557160048,
      "grad_norm": 6.312081336975098,
      "learning_rate": 0.00019587570171781735,
      "loss": 7.2273,
      "step": 80950
    },
    {
      "epoch": 9.742478941034898,
      "grad_norm": 13.132644653320312,
      "learning_rate": 0.0001958746203722763,
      "loss": 7.1698,
      "step": 80960
    },
    {
      "epoch": 9.743682310469314,
      "grad_norm": 10.809191703796387,
      "learning_rate": 0.000195873538887981,
      "loss": 7.1982,
      "step": 80970
    },
    {
      "epoch": 9.74488567990373,
      "grad_norm": 19.389211654663086,
      "learning_rate": 0.00019587245726493307,
      "loss": 7.1589,
      "step": 80980
    },
    {
      "epoch": 9.746089049338147,
      "grad_norm": 15.189022064208984,
      "learning_rate": 0.00019587137550313398,
      "loss": 7.1406,
      "step": 80990
    },
    {
      "epoch": 9.747292418772563,
      "grad_norm": 14.81157398223877,
      "learning_rate": 0.00019587029360258537,
      "loss": 7.1925,
      "step": 81000
    },
    {
      "epoch": 9.748495788206979,
      "grad_norm": 8.749092102050781,
      "learning_rate": 0.00019586921156328874,
      "loss": 7.0593,
      "step": 81010
    },
    {
      "epoch": 9.749699157641396,
      "grad_norm": 24.159252166748047,
      "learning_rate": 0.0001958681293852457,
      "loss": 7.2597,
      "step": 81020
    },
    {
      "epoch": 9.750902527075812,
      "grad_norm": 21.47060775756836,
      "learning_rate": 0.00019586704706845781,
      "loss": 7.2261,
      "step": 81030
    },
    {
      "epoch": 9.752105896510228,
      "grad_norm": 23.625019073486328,
      "learning_rate": 0.00019586596461292665,
      "loss": 7.2175,
      "step": 81040
    },
    {
      "epoch": 9.753309265944646,
      "grad_norm": 19.579246520996094,
      "learning_rate": 0.00019586488201865377,
      "loss": 7.1089,
      "step": 81050
    },
    {
      "epoch": 9.754512635379061,
      "grad_norm": 18.162458419799805,
      "learning_rate": 0.00019586379928564074,
      "loss": 7.1443,
      "step": 81060
    },
    {
      "epoch": 9.755716004813477,
      "grad_norm": 36.512001037597656,
      "learning_rate": 0.00019586271641388908,
      "loss": 7.0819,
      "step": 81070
    },
    {
      "epoch": 9.756919374247895,
      "grad_norm": 29.0833740234375,
      "learning_rate": 0.00019586163340340048,
      "loss": 7.2203,
      "step": 81080
    },
    {
      "epoch": 9.75812274368231,
      "grad_norm": 53.947410583496094,
      "learning_rate": 0.00019586055025417638,
      "loss": 7.2633,
      "step": 81090
    },
    {
      "epoch": 9.759326113116726,
      "grad_norm": 28.762466430664062,
      "learning_rate": 0.00019585946696621837,
      "loss": 7.1178,
      "step": 81100
    },
    {
      "epoch": 9.760529482551144,
      "grad_norm": 21.785465240478516,
      "learning_rate": 0.0001958583835395281,
      "loss": 7.1446,
      "step": 81110
    },
    {
      "epoch": 9.76173285198556,
      "grad_norm": 28.514663696289062,
      "learning_rate": 0.00019585729997410707,
      "loss": 7.1186,
      "step": 81120
    },
    {
      "epoch": 9.762936221419976,
      "grad_norm": 44.98613357543945,
      "learning_rate": 0.00019585621626995683,
      "loss": 7.1605,
      "step": 81130
    },
    {
      "epoch": 9.764139590854393,
      "grad_norm": 24.573091506958008,
      "learning_rate": 0.000195855132427079,
      "loss": 7.2074,
      "step": 81140
    },
    {
      "epoch": 9.765342960288809,
      "grad_norm": 47.555606842041016,
      "learning_rate": 0.00019585404844547512,
      "loss": 7.2037,
      "step": 81150
    },
    {
      "epoch": 9.766546329723225,
      "grad_norm": 22.620088577270508,
      "learning_rate": 0.00019585296432514677,
      "loss": 7.1592,
      "step": 81160
    },
    {
      "epoch": 9.76774969915764,
      "grad_norm": 18.58357048034668,
      "learning_rate": 0.00019585188006609553,
      "loss": 7.1873,
      "step": 81170
    },
    {
      "epoch": 9.768953068592058,
      "grad_norm": 26.78301239013672,
      "learning_rate": 0.0001958507956683229,
      "loss": 7.1344,
      "step": 81180
    },
    {
      "epoch": 9.770156438026474,
      "grad_norm": 55.00008773803711,
      "learning_rate": 0.00019584971113183056,
      "loss": 7.1803,
      "step": 81190
    },
    {
      "epoch": 9.77135980746089,
      "grad_norm": 63.540374755859375,
      "learning_rate": 0.00019584862645662004,
      "loss": 7.1767,
      "step": 81200
    },
    {
      "epoch": 9.772563176895307,
      "grad_norm": 27.377532958984375,
      "learning_rate": 0.00019584754164269285,
      "loss": 7.1682,
      "step": 81210
    },
    {
      "epoch": 9.773766546329723,
      "grad_norm": 42.744606018066406,
      "learning_rate": 0.0001958464566900506,
      "loss": 7.0821,
      "step": 81220
    },
    {
      "epoch": 9.774969915764139,
      "grad_norm": 65.01910400390625,
      "learning_rate": 0.0001958453715986949,
      "loss": 7.1563,
      "step": 81230
    },
    {
      "epoch": 9.776173285198556,
      "grad_norm": 95.84552001953125,
      "learning_rate": 0.00019584428636862727,
      "loss": 7.1289,
      "step": 81240
    },
    {
      "epoch": 9.777376654632972,
      "grad_norm": 39.06284713745117,
      "learning_rate": 0.00019584320099984924,
      "loss": 7.2058,
      "step": 81250
    },
    {
      "epoch": 9.778580024067388,
      "grad_norm": 28.717742919921875,
      "learning_rate": 0.0001958421154923625,
      "loss": 7.2044,
      "step": 81260
    },
    {
      "epoch": 9.779783393501805,
      "grad_norm": 37.450199127197266,
      "learning_rate": 0.00019584102984616854,
      "loss": 7.2618,
      "step": 81270
    },
    {
      "epoch": 9.780986762936221,
      "grad_norm": 24.722015380859375,
      "learning_rate": 0.00019583994406126894,
      "loss": 7.1449,
      "step": 81280
    },
    {
      "epoch": 9.782190132370637,
      "grad_norm": 34.97111892700195,
      "learning_rate": 0.0001958388581376653,
      "loss": 7.2478,
      "step": 81290
    },
    {
      "epoch": 9.783393501805055,
      "grad_norm": 15.848621368408203,
      "learning_rate": 0.00019583777207535916,
      "loss": 7.0592,
      "step": 81300
    },
    {
      "epoch": 9.78459687123947,
      "grad_norm": 14.618273735046387,
      "learning_rate": 0.0001958366858743521,
      "loss": 7.1619,
      "step": 81310
    },
    {
      "epoch": 9.785800240673886,
      "grad_norm": 46.58594512939453,
      "learning_rate": 0.0001958355995346457,
      "loss": 7.0839,
      "step": 81320
    },
    {
      "epoch": 9.787003610108304,
      "grad_norm": 14.210058212280273,
      "learning_rate": 0.0001958345130562415,
      "loss": 7.1581,
      "step": 81330
    },
    {
      "epoch": 9.78820697954272,
      "grad_norm": 55.863956451416016,
      "learning_rate": 0.0001958334264391411,
      "loss": 7.1962,
      "step": 81340
    },
    {
      "epoch": 9.789410348977135,
      "grad_norm": 16.850446701049805,
      "learning_rate": 0.0001958323396833461,
      "loss": 7.1214,
      "step": 81350
    },
    {
      "epoch": 9.790613718411553,
      "grad_norm": 31.345497131347656,
      "learning_rate": 0.00019583125278885803,
      "loss": 7.1457,
      "step": 81360
    },
    {
      "epoch": 9.791817087845969,
      "grad_norm": 54.03544235229492,
      "learning_rate": 0.00019583016575567847,
      "loss": 7.1623,
      "step": 81370
    },
    {
      "epoch": 9.793020457280385,
      "grad_norm": 40.371238708496094,
      "learning_rate": 0.00019582907858380901,
      "loss": 7.0905,
      "step": 81380
    },
    {
      "epoch": 9.794223826714802,
      "grad_norm": 46.57606887817383,
      "learning_rate": 0.00019582799127325123,
      "loss": 7.0522,
      "step": 81390
    },
    {
      "epoch": 9.795427196149218,
      "grad_norm": 24.272066116333008,
      "learning_rate": 0.00019582690382400668,
      "loss": 7.1489,
      "step": 81400
    },
    {
      "epoch": 9.796630565583634,
      "grad_norm": 33.30632781982422,
      "learning_rate": 0.00019582581623607691,
      "loss": 7.1172,
      "step": 81410
    },
    {
      "epoch": 9.797833935018051,
      "grad_norm": 49.117000579833984,
      "learning_rate": 0.00019582472850946357,
      "loss": 7.14,
      "step": 81420
    },
    {
      "epoch": 9.799037304452467,
      "grad_norm": 30.63140869140625,
      "learning_rate": 0.00019582364064416817,
      "loss": 7.1226,
      "step": 81430
    },
    {
      "epoch": 9.800240673886883,
      "grad_norm": 83.77889251708984,
      "learning_rate": 0.0001958225526401923,
      "loss": 7.1363,
      "step": 81440
    },
    {
      "epoch": 9.8014440433213,
      "grad_norm": 21.450422286987305,
      "learning_rate": 0.00019582146449753754,
      "loss": 7.2253,
      "step": 81450
    },
    {
      "epoch": 9.802647412755716,
      "grad_norm": 16.15947151184082,
      "learning_rate": 0.00019582037621620547,
      "loss": 7.1707,
      "step": 81460
    },
    {
      "epoch": 9.803850782190132,
      "grad_norm": 65.86964416503906,
      "learning_rate": 0.00019581928779619762,
      "loss": 7.1195,
      "step": 81470
    },
    {
      "epoch": 9.80505415162455,
      "grad_norm": 37.578853607177734,
      "learning_rate": 0.00019581819923751565,
      "loss": 7.2109,
      "step": 81480
    },
    {
      "epoch": 9.806257521058965,
      "grad_norm": 203.12808227539062,
      "learning_rate": 0.00019581711054016108,
      "loss": 7.1567,
      "step": 81490
    },
    {
      "epoch": 9.807460890493381,
      "grad_norm": 141.58651733398438,
      "learning_rate": 0.00019581602170413547,
      "loss": 7.1059,
      "step": 81500
    },
    {
      "epoch": 9.808664259927799,
      "grad_norm": 177.87869262695312,
      "learning_rate": 0.00019581493272944044,
      "loss": 7.1007,
      "step": 81510
    },
    {
      "epoch": 9.809867629362214,
      "grad_norm": 148.8583526611328,
      "learning_rate": 0.00019581384361607755,
      "loss": 7.1554,
      "step": 81520
    },
    {
      "epoch": 9.81107099879663,
      "grad_norm": 221.38961791992188,
      "learning_rate": 0.00019581275436404835,
      "loss": 7.1346,
      "step": 81530
    },
    {
      "epoch": 9.812274368231048,
      "grad_norm": 363.3433837890625,
      "learning_rate": 0.00019581166497335448,
      "loss": 7.2178,
      "step": 81540
    },
    {
      "epoch": 9.813477737665464,
      "grad_norm": 288.44677734375,
      "learning_rate": 0.00019581057544399742,
      "loss": 7.2879,
      "step": 81550
    },
    {
      "epoch": 9.81468110709988,
      "grad_norm": 140.5692596435547,
      "learning_rate": 0.00019580948577597887,
      "loss": 7.1986,
      "step": 81560
    },
    {
      "epoch": 9.815884476534297,
      "grad_norm": 222.69528198242188,
      "learning_rate": 0.0001958083959693003,
      "loss": 7.0938,
      "step": 81570
    },
    {
      "epoch": 9.817087845968713,
      "grad_norm": 306.8842468261719,
      "learning_rate": 0.00019580730602396332,
      "loss": 7.1569,
      "step": 81580
    },
    {
      "epoch": 9.818291215403129,
      "grad_norm": 356.0410461425781,
      "learning_rate": 0.00019580621593996952,
      "loss": 7.2226,
      "step": 81590
    },
    {
      "epoch": 9.819494584837544,
      "grad_norm": 398.6214904785156,
      "learning_rate": 0.0001958051257173205,
      "loss": 7.3989,
      "step": 81600
    },
    {
      "epoch": 9.820697954271962,
      "grad_norm": 401.3757629394531,
      "learning_rate": 0.00019580403535601778,
      "loss": 7.2843,
      "step": 81610
    },
    {
      "epoch": 9.821901323706378,
      "grad_norm": 134.86692810058594,
      "learning_rate": 0.00019580294485606298,
      "loss": 7.233,
      "step": 81620
    },
    {
      "epoch": 9.823104693140793,
      "grad_norm": 211.73191833496094,
      "learning_rate": 0.00019580185421745765,
      "loss": 7.2583,
      "step": 81630
    },
    {
      "epoch": 9.824308062575211,
      "grad_norm": 269.29052734375,
      "learning_rate": 0.0001958007634402034,
      "loss": 7.1652,
      "step": 81640
    },
    {
      "epoch": 9.825511432009627,
      "grad_norm": 118.49571990966797,
      "learning_rate": 0.0001957996725243018,
      "loss": 7.1154,
      "step": 81650
    },
    {
      "epoch": 9.826714801444043,
      "grad_norm": 44.02663040161133,
      "learning_rate": 0.0001957985814697544,
      "loss": 7.2774,
      "step": 81660
    },
    {
      "epoch": 9.82791817087846,
      "grad_norm": 79.97306823730469,
      "learning_rate": 0.0001957974902765628,
      "loss": 7.246,
      "step": 81670
    },
    {
      "epoch": 9.829121540312876,
      "grad_norm": 77.45560455322266,
      "learning_rate": 0.00019579639894472858,
      "loss": 7.193,
      "step": 81680
    },
    {
      "epoch": 9.830324909747292,
      "grad_norm": 126.63589477539062,
      "learning_rate": 0.00019579530747425335,
      "loss": 7.238,
      "step": 81690
    },
    {
      "epoch": 9.83152827918171,
      "grad_norm": 89.69837951660156,
      "learning_rate": 0.00019579421586513862,
      "loss": 7.1709,
      "step": 81700
    },
    {
      "epoch": 9.832731648616125,
      "grad_norm": 33.425048828125,
      "learning_rate": 0.00019579312411738603,
      "loss": 7.2799,
      "step": 81710
    },
    {
      "epoch": 9.833935018050541,
      "grad_norm": 16.653369903564453,
      "learning_rate": 0.00019579203223099714,
      "loss": 7.2248,
      "step": 81720
    },
    {
      "epoch": 9.835138387484958,
      "grad_norm": 117.22438049316406,
      "learning_rate": 0.00019579094020597354,
      "loss": 7.2314,
      "step": 81730
    },
    {
      "epoch": 9.836341756919374,
      "grad_norm": 163.76300048828125,
      "learning_rate": 0.00019578984804231677,
      "loss": 7.4453,
      "step": 81740
    },
    {
      "epoch": 9.83754512635379,
      "grad_norm": 53.07315444946289,
      "learning_rate": 0.00019578875574002846,
      "loss": 7.2892,
      "step": 81750
    },
    {
      "epoch": 9.838748495788208,
      "grad_norm": 31.876468658447266,
      "learning_rate": 0.00019578766329911016,
      "loss": 7.2551,
      "step": 81760
    },
    {
      "epoch": 9.839951865222623,
      "grad_norm": 204.01930236816406,
      "learning_rate": 0.00019578657071956344,
      "loss": 7.1353,
      "step": 81770
    },
    {
      "epoch": 9.84115523465704,
      "grad_norm": 329.9084777832031,
      "learning_rate": 0.00019578547800138996,
      "loss": 7.3476,
      "step": 81780
    },
    {
      "epoch": 9.842358604091457,
      "grad_norm": 284.59259033203125,
      "learning_rate": 0.00019578438514459122,
      "loss": 7.3247,
      "step": 81790
    },
    {
      "epoch": 9.843561973525873,
      "grad_norm": 308.4414367675781,
      "learning_rate": 0.0001957832921491688,
      "loss": 7.5106,
      "step": 81800
    },
    {
      "epoch": 9.844765342960288,
      "grad_norm": 245.64149475097656,
      "learning_rate": 0.00019578219901512434,
      "loss": 7.5745,
      "step": 81810
    },
    {
      "epoch": 9.845968712394706,
      "grad_norm": 287.5294494628906,
      "learning_rate": 0.00019578110574245935,
      "loss": 7.4791,
      "step": 81820
    },
    {
      "epoch": 9.847172081829122,
      "grad_norm": 127.05554962158203,
      "learning_rate": 0.00019578001233117552,
      "loss": 7.2435,
      "step": 81830
    },
    {
      "epoch": 9.848375451263538,
      "grad_norm": 230.76974487304688,
      "learning_rate": 0.00019577891878127434,
      "loss": 7.2437,
      "step": 81840
    },
    {
      "epoch": 9.849578820697955,
      "grad_norm": 102.9111099243164,
      "learning_rate": 0.00019577782509275736,
      "loss": 7.2524,
      "step": 81850
    },
    {
      "epoch": 9.85078219013237,
      "grad_norm": 74.16764068603516,
      "learning_rate": 0.00019577673126562628,
      "loss": 7.2241,
      "step": 81860
    },
    {
      "epoch": 9.851985559566787,
      "grad_norm": 60.01537322998047,
      "learning_rate": 0.0001957756372998826,
      "loss": 7.1939,
      "step": 81870
    },
    {
      "epoch": 9.853188929001202,
      "grad_norm": 49.157100677490234,
      "learning_rate": 0.00019577454319552796,
      "loss": 7.2806,
      "step": 81880
    },
    {
      "epoch": 9.85439229843562,
      "grad_norm": 11.818543434143066,
      "learning_rate": 0.00019577344895256387,
      "loss": 7.377,
      "step": 81890
    },
    {
      "epoch": 9.855595667870036,
      "grad_norm": 23.056699752807617,
      "learning_rate": 0.00019577235457099196,
      "loss": 7.307,
      "step": 81900
    },
    {
      "epoch": 9.856799037304452,
      "grad_norm": 29.7425594329834,
      "learning_rate": 0.00019577126005081383,
      "loss": 7.3352,
      "step": 81910
    },
    {
      "epoch": 9.85800240673887,
      "grad_norm": 95.12638092041016,
      "learning_rate": 0.00019577016539203102,
      "loss": 7.3838,
      "step": 81920
    },
    {
      "epoch": 9.859205776173285,
      "grad_norm": 11.68164348602295,
      "learning_rate": 0.00019576907059464516,
      "loss": 7.3004,
      "step": 81930
    },
    {
      "epoch": 9.8604091456077,
      "grad_norm": 24.45294761657715,
      "learning_rate": 0.00019576797565865777,
      "loss": 7.3887,
      "step": 81940
    },
    {
      "epoch": 9.861612515042118,
      "grad_norm": 30.26659393310547,
      "learning_rate": 0.0001957668805840705,
      "loss": 7.3831,
      "step": 81950
    },
    {
      "epoch": 9.862815884476534,
      "grad_norm": 14.567951202392578,
      "learning_rate": 0.00019576578537088492,
      "loss": 7.3683,
      "step": 81960
    },
    {
      "epoch": 9.86401925391095,
      "grad_norm": 34.78707504272461,
      "learning_rate": 0.00019576469001910258,
      "loss": 7.3319,
      "step": 81970
    },
    {
      "epoch": 9.865222623345367,
      "grad_norm": 13.080894470214844,
      "learning_rate": 0.0001957635945287251,
      "loss": 7.3395,
      "step": 81980
    },
    {
      "epoch": 9.866425992779783,
      "grad_norm": 21.612865447998047,
      "learning_rate": 0.00019576249889975407,
      "loss": 7.3189,
      "step": 81990
    },
    {
      "epoch": 9.867629362214199,
      "grad_norm": 20.334867477416992,
      "learning_rate": 0.00019576140313219104,
      "loss": 7.2705,
      "step": 82000
    },
    {
      "epoch": 9.868832731648617,
      "grad_norm": 12.033313751220703,
      "learning_rate": 0.00019576030722603762,
      "loss": 7.3791,
      "step": 82010
    },
    {
      "epoch": 9.870036101083032,
      "grad_norm": 40.773624420166016,
      "learning_rate": 0.0001957592111812954,
      "loss": 7.2292,
      "step": 82020
    },
    {
      "epoch": 9.871239470517448,
      "grad_norm": 196.13064575195312,
      "learning_rate": 0.00019575811499796596,
      "loss": 7.398,
      "step": 82030
    },
    {
      "epoch": 9.872442839951866,
      "grad_norm": 344.34698486328125,
      "learning_rate": 0.0001957570186760509,
      "loss": 7.4888,
      "step": 82040
    },
    {
      "epoch": 9.873646209386282,
      "grad_norm": 19.01952362060547,
      "learning_rate": 0.00019575592221555177,
      "loss": 7.2922,
      "step": 82050
    },
    {
      "epoch": 9.874849578820697,
      "grad_norm": 85.31414031982422,
      "learning_rate": 0.00019575482561647018,
      "loss": 7.433,
      "step": 82060
    },
    {
      "epoch": 9.876052948255115,
      "grad_norm": 98.12935638427734,
      "learning_rate": 0.00019575372887880774,
      "loss": 7.5917,
      "step": 82070
    },
    {
      "epoch": 9.87725631768953,
      "grad_norm": 9.710317611694336,
      "learning_rate": 0.000195752632002566,
      "loss": 7.6765,
      "step": 82080
    },
    {
      "epoch": 9.878459687123947,
      "grad_norm": 8.849117279052734,
      "learning_rate": 0.00019575153498774652,
      "loss": 7.5691,
      "step": 82090
    },
    {
      "epoch": 9.879663056558364,
      "grad_norm": 9.453598022460938,
      "learning_rate": 0.00019575043783435098,
      "loss": 7.5222,
      "step": 82100
    },
    {
      "epoch": 9.88086642599278,
      "grad_norm": 9.823592185974121,
      "learning_rate": 0.0001957493405423809,
      "loss": 7.5208,
      "step": 82110
    },
    {
      "epoch": 9.882069795427196,
      "grad_norm": 9.096379280090332,
      "learning_rate": 0.0001957482431118379,
      "loss": 7.4863,
      "step": 82120
    },
    {
      "epoch": 9.883273164861613,
      "grad_norm": 17.901180267333984,
      "learning_rate": 0.0001957471455427235,
      "loss": 7.6243,
      "step": 82130
    },
    {
      "epoch": 9.884476534296029,
      "grad_norm": 101.95468139648438,
      "learning_rate": 0.0001957460478350394,
      "loss": 7.566,
      "step": 82140
    },
    {
      "epoch": 9.885679903730445,
      "grad_norm": 39.395477294921875,
      "learning_rate": 0.00019574494998878713,
      "loss": 7.6857,
      "step": 82150
    },
    {
      "epoch": 9.886883273164862,
      "grad_norm": 11.01801872253418,
      "learning_rate": 0.00019574385200396823,
      "loss": 7.6153,
      "step": 82160
    },
    {
      "epoch": 9.888086642599278,
      "grad_norm": 8.296485900878906,
      "learning_rate": 0.0001957427538805844,
      "loss": 7.6091,
      "step": 82170
    },
    {
      "epoch": 9.889290012033694,
      "grad_norm": 11.651066780090332,
      "learning_rate": 0.0001957416556186371,
      "loss": 7.5882,
      "step": 82180
    },
    {
      "epoch": 9.890493381468112,
      "grad_norm": 15.031363487243652,
      "learning_rate": 0.00019574055721812802,
      "loss": 7.4771,
      "step": 82190
    },
    {
      "epoch": 9.891696750902527,
      "grad_norm": 22.049480438232422,
      "learning_rate": 0.00019573945867905871,
      "loss": 7.5367,
      "step": 82200
    },
    {
      "epoch": 9.892900120336943,
      "grad_norm": 6.620582103729248,
      "learning_rate": 0.00019573836000143078,
      "loss": 7.4201,
      "step": 82210
    },
    {
      "epoch": 9.89410348977136,
      "grad_norm": 44.3409423828125,
      "learning_rate": 0.0001957372611852458,
      "loss": 7.4448,
      "step": 82220
    },
    {
      "epoch": 9.895306859205776,
      "grad_norm": 11.555047988891602,
      "learning_rate": 0.00019573616223050538,
      "loss": 7.4677,
      "step": 82230
    },
    {
      "epoch": 9.896510228640192,
      "grad_norm": 42.511558532714844,
      "learning_rate": 0.00019573506313721106,
      "loss": 7.382,
      "step": 82240
    },
    {
      "epoch": 9.89771359807461,
      "grad_norm": 33.44620132446289,
      "learning_rate": 0.00019573396390536448,
      "loss": 7.4196,
      "step": 82250
    },
    {
      "epoch": 9.898916967509026,
      "grad_norm": 35.21254348754883,
      "learning_rate": 0.00019573286453496725,
      "loss": 7.3664,
      "step": 82260
    },
    {
      "epoch": 9.900120336943441,
      "grad_norm": 167.31251525878906,
      "learning_rate": 0.0001957317650260209,
      "loss": 7.4636,
      "step": 82270
    },
    {
      "epoch": 9.901323706377857,
      "grad_norm": 279.4736633300781,
      "learning_rate": 0.00019573066537852704,
      "loss": 7.5109,
      "step": 82280
    },
    {
      "epoch": 9.902527075812275,
      "grad_norm": 152.21102905273438,
      "learning_rate": 0.0001957295655924873,
      "loss": 7.4334,
      "step": 82290
    },
    {
      "epoch": 9.90373044524669,
      "grad_norm": 26.969270706176758,
      "learning_rate": 0.00019572846566790324,
      "loss": 7.3984,
      "step": 82300
    },
    {
      "epoch": 9.904933814681106,
      "grad_norm": 9.886170387268066,
      "learning_rate": 0.00019572736560477644,
      "loss": 7.4776,
      "step": 82310
    },
    {
      "epoch": 9.906137184115524,
      "grad_norm": 100.31951904296875,
      "learning_rate": 0.00019572626540310852,
      "loss": 7.3959,
      "step": 82320
    },
    {
      "epoch": 9.90734055354994,
      "grad_norm": 36.880149841308594,
      "learning_rate": 0.00019572516506290106,
      "loss": 7.4326,
      "step": 82330
    },
    {
      "epoch": 9.908543922984355,
      "grad_norm": 96.23030090332031,
      "learning_rate": 0.00019572406458415564,
      "loss": 7.3163,
      "step": 82340
    },
    {
      "epoch": 9.909747292418773,
      "grad_norm": 64.73321533203125,
      "learning_rate": 0.0001957229639668739,
      "loss": 7.3193,
      "step": 82350
    },
    {
      "epoch": 9.910950661853189,
      "grad_norm": 221.0384063720703,
      "learning_rate": 0.00019572186321105734,
      "loss": 7.3878,
      "step": 82360
    },
    {
      "epoch": 9.912154031287605,
      "grad_norm": 77.85171508789062,
      "learning_rate": 0.00019572076231670767,
      "loss": 7.4973,
      "step": 82370
    },
    {
      "epoch": 9.913357400722022,
      "grad_norm": 47.297149658203125,
      "learning_rate": 0.00019571966128382638,
      "loss": 7.4137,
      "step": 82380
    },
    {
      "epoch": 9.914560770156438,
      "grad_norm": 204.392822265625,
      "learning_rate": 0.0001957185601124151,
      "loss": 7.4217,
      "step": 82390
    },
    {
      "epoch": 9.915764139590854,
      "grad_norm": 27.453489303588867,
      "learning_rate": 0.00019571745880247546,
      "loss": 7.511,
      "step": 82400
    },
    {
      "epoch": 9.916967509025271,
      "grad_norm": 23.24348258972168,
      "learning_rate": 0.000195716357354009,
      "loss": 7.5252,
      "step": 82410
    },
    {
      "epoch": 9.918170878459687,
      "grad_norm": 9.649572372436523,
      "learning_rate": 0.00019571525576701738,
      "loss": 7.3736,
      "step": 82420
    },
    {
      "epoch": 9.919374247894103,
      "grad_norm": 10.647658348083496,
      "learning_rate": 0.00019571415404150212,
      "loss": 7.5302,
      "step": 82430
    },
    {
      "epoch": 9.92057761732852,
      "grad_norm": 28.23257064819336,
      "learning_rate": 0.00019571305217746488,
      "loss": 7.4453,
      "step": 82440
    },
    {
      "epoch": 9.921780986762936,
      "grad_norm": 16.840059280395508,
      "learning_rate": 0.00019571195017490723,
      "loss": 7.3965,
      "step": 82450
    },
    {
      "epoch": 9.922984356197352,
      "grad_norm": 9.777364730834961,
      "learning_rate": 0.0001957108480338307,
      "loss": 7.4309,
      "step": 82460
    },
    {
      "epoch": 9.92418772563177,
      "grad_norm": 21.885236740112305,
      "learning_rate": 0.000195709745754237,
      "loss": 7.4862,
      "step": 82470
    },
    {
      "epoch": 9.925391095066185,
      "grad_norm": 17.263158798217773,
      "learning_rate": 0.00019570864333612762,
      "loss": 7.4125,
      "step": 82480
    },
    {
      "epoch": 9.926594464500601,
      "grad_norm": 16.948944091796875,
      "learning_rate": 0.00019570754077950423,
      "loss": 7.4226,
      "step": 82490
    },
    {
      "epoch": 9.927797833935019,
      "grad_norm": 16.359371185302734,
      "learning_rate": 0.0001957064380843684,
      "loss": 7.4369,
      "step": 82500
    },
    {
      "epoch": 9.929001203369435,
      "grad_norm": 12.681251525878906,
      "learning_rate": 0.0001957053352507217,
      "loss": 7.3728,
      "step": 82510
    },
    {
      "epoch": 9.93020457280385,
      "grad_norm": 17.827796936035156,
      "learning_rate": 0.0001957042322785658,
      "loss": 7.3548,
      "step": 82520
    },
    {
      "epoch": 9.931407942238268,
      "grad_norm": 41.84023666381836,
      "learning_rate": 0.0001957031291679022,
      "loss": 7.3884,
      "step": 82530
    },
    {
      "epoch": 9.932611311672684,
      "grad_norm": 34.2298583984375,
      "learning_rate": 0.00019570202591873254,
      "loss": 7.3613,
      "step": 82540
    },
    {
      "epoch": 9.9338146811071,
      "grad_norm": 26.370372772216797,
      "learning_rate": 0.00019570092253105848,
      "loss": 7.3187,
      "step": 82550
    },
    {
      "epoch": 9.935018050541515,
      "grad_norm": 158.3146209716797,
      "learning_rate": 0.0001956998190048815,
      "loss": 7.8905,
      "step": 82560
    },
    {
      "epoch": 9.936221419975933,
      "grad_norm": 83.25333404541016,
      "learning_rate": 0.0001956987153402033,
      "loss": 8.5049,
      "step": 82570
    },
    {
      "epoch": 9.937424789410349,
      "grad_norm": 67.73261260986328,
      "learning_rate": 0.00019569761153702539,
      "loss": 7.752,
      "step": 82580
    },
    {
      "epoch": 9.938628158844764,
      "grad_norm": 41.0427360534668,
      "learning_rate": 0.0001956965075953494,
      "loss": 9.0304,
      "step": 82590
    },
    {
      "epoch": 9.939831528279182,
      "grad_norm": 28.815858840942383,
      "learning_rate": 0.00019569540351517696,
      "loss": 8.2543,
      "step": 82600
    },
    {
      "epoch": 9.941034897713598,
      "grad_norm": 28.119356155395508,
      "learning_rate": 0.00019569429929650967,
      "loss": 7.6497,
      "step": 82610
    },
    {
      "epoch": 9.942238267148014,
      "grad_norm": 30.308202743530273,
      "learning_rate": 0.00019569319493934904,
      "loss": 7.5888,
      "step": 82620
    },
    {
      "epoch": 9.943441636582431,
      "grad_norm": 15.763078689575195,
      "learning_rate": 0.00019569209044369677,
      "loss": 7.3972,
      "step": 82630
    },
    {
      "epoch": 9.944645006016847,
      "grad_norm": 35.42869567871094,
      "learning_rate": 0.00019569098580955445,
      "loss": 7.5268,
      "step": 82640
    },
    {
      "epoch": 9.945848375451263,
      "grad_norm": 30.2524356842041,
      "learning_rate": 0.0001956898810369236,
      "loss": 7.501,
      "step": 82650
    },
    {
      "epoch": 9.94705174488568,
      "grad_norm": 16.35939598083496,
      "learning_rate": 0.0001956887761258059,
      "loss": 7.4544,
      "step": 82660
    },
    {
      "epoch": 9.948255114320096,
      "grad_norm": 13.832306861877441,
      "learning_rate": 0.0001956876710762029,
      "loss": 7.4556,
      "step": 82670
    },
    {
      "epoch": 9.949458483754512,
      "grad_norm": 26.814851760864258,
      "learning_rate": 0.00019568656588811622,
      "loss": 7.5283,
      "step": 82680
    },
    {
      "epoch": 9.95066185318893,
      "grad_norm": 33.63003921508789,
      "learning_rate": 0.00019568546056154744,
      "loss": 7.517,
      "step": 82690
    },
    {
      "epoch": 9.951865222623345,
      "grad_norm": 15.94406795501709,
      "learning_rate": 0.0001956843550964982,
      "loss": 7.5016,
      "step": 82700
    },
    {
      "epoch": 9.953068592057761,
      "grad_norm": 9.614724159240723,
      "learning_rate": 0.00019568324949297003,
      "loss": 7.5599,
      "step": 82710
    },
    {
      "epoch": 9.954271961492179,
      "grad_norm": 11.35432243347168,
      "learning_rate": 0.0001956821437509646,
      "loss": 7.4455,
      "step": 82720
    },
    {
      "epoch": 9.955475330926594,
      "grad_norm": 28.24508285522461,
      "learning_rate": 0.0001956810378704835,
      "loss": 7.5293,
      "step": 82730
    },
    {
      "epoch": 9.95667870036101,
      "grad_norm": 16.813451766967773,
      "learning_rate": 0.00019567993185152832,
      "loss": 7.4443,
      "step": 82740
    },
    {
      "epoch": 9.957882069795428,
      "grad_norm": 14.316716194152832,
      "learning_rate": 0.00019567882569410062,
      "loss": 7.4105,
      "step": 82750
    },
    {
      "epoch": 9.959085439229844,
      "grad_norm": 11.617361068725586,
      "learning_rate": 0.00019567771939820206,
      "loss": 7.4074,
      "step": 82760
    },
    {
      "epoch": 9.96028880866426,
      "grad_norm": 16.331735610961914,
      "learning_rate": 0.00019567661296383425,
      "loss": 7.4777,
      "step": 82770
    },
    {
      "epoch": 9.961492178098677,
      "grad_norm": 14.681106567382812,
      "learning_rate": 0.00019567550639099869,
      "loss": 7.4397,
      "step": 82780
    },
    {
      "epoch": 9.962695547533093,
      "grad_norm": 29.191822052001953,
      "learning_rate": 0.0001956743996796971,
      "loss": 7.408,
      "step": 82790
    },
    {
      "epoch": 9.963898916967509,
      "grad_norm": 137.0829620361328,
      "learning_rate": 0.000195673292829931,
      "loss": 7.4947,
      "step": 82800
    },
    {
      "epoch": 9.965102286401926,
      "grad_norm": 11.389457702636719,
      "learning_rate": 0.00019567218584170206,
      "loss": 7.3979,
      "step": 82810
    },
    {
      "epoch": 9.966305655836342,
      "grad_norm": 54.594200134277344,
      "learning_rate": 0.00019567107871501182,
      "loss": 7.5069,
      "step": 82820
    },
    {
      "epoch": 9.967509025270758,
      "grad_norm": 13.628167152404785,
      "learning_rate": 0.00019566997144986192,
      "loss": 7.3462,
      "step": 82830
    },
    {
      "epoch": 9.968712394705175,
      "grad_norm": 50.4848518371582,
      "learning_rate": 0.00019566886404625395,
      "loss": 7.4765,
      "step": 82840
    },
    {
      "epoch": 9.969915764139591,
      "grad_norm": 163.96688842773438,
      "learning_rate": 0.00019566775650418951,
      "loss": 7.693,
      "step": 82850
    },
    {
      "epoch": 9.971119133574007,
      "grad_norm": 20.04106330871582,
      "learning_rate": 0.0001956666488236702,
      "loss": 7.5473,
      "step": 82860
    },
    {
      "epoch": 9.972322503008424,
      "grad_norm": 53.01713562011719,
      "learning_rate": 0.00019566554100469764,
      "loss": 7.5402,
      "step": 82870
    },
    {
      "epoch": 9.97352587244284,
      "grad_norm": 7.600004196166992,
      "learning_rate": 0.00019566443304727343,
      "loss": 7.4574,
      "step": 82880
    },
    {
      "epoch": 9.974729241877256,
      "grad_norm": 14.555046081542969,
      "learning_rate": 0.00019566332495139915,
      "loss": 7.3451,
      "step": 82890
    },
    {
      "epoch": 9.975932611311674,
      "grad_norm": 235.01226806640625,
      "learning_rate": 0.00019566221671707642,
      "loss": 7.3825,
      "step": 82900
    },
    {
      "epoch": 9.97713598074609,
      "grad_norm": 12.984586715698242,
      "learning_rate": 0.00019566110834430686,
      "loss": 7.3919,
      "step": 82910
    },
    {
      "epoch": 9.978339350180505,
      "grad_norm": 187.3184051513672,
      "learning_rate": 0.00019565999983309207,
      "loss": 7.3823,
      "step": 82920
    },
    {
      "epoch": 9.979542719614923,
      "grad_norm": 12.903731346130371,
      "learning_rate": 0.0001956588911834336,
      "loss": 7.3833,
      "step": 82930
    },
    {
      "epoch": 9.980746089049338,
      "grad_norm": 10.085701942443848,
      "learning_rate": 0.00019565778239533313,
      "loss": 7.3357,
      "step": 82940
    },
    {
      "epoch": 9.981949458483754,
      "grad_norm": 6.894649982452393,
      "learning_rate": 0.0001956566734687922,
      "loss": 7.2906,
      "step": 82950
    },
    {
      "epoch": 9.98315282791817,
      "grad_norm": 5.988856315612793,
      "learning_rate": 0.0001956555644038125,
      "loss": 7.3891,
      "step": 82960
    },
    {
      "epoch": 9.984356197352588,
      "grad_norm": 6.759174823760986,
      "learning_rate": 0.00019565445520039555,
      "loss": 7.3043,
      "step": 82970
    },
    {
      "epoch": 9.985559566787003,
      "grad_norm": 11.749421119689941,
      "learning_rate": 0.000195653345858543,
      "loss": 7.2955,
      "step": 82980
    },
    {
      "epoch": 9.98676293622142,
      "grad_norm": 9.498119354248047,
      "learning_rate": 0.00019565223637825642,
      "loss": 7.4558,
      "step": 82990
    },
    {
      "epoch": 9.987966305655837,
      "grad_norm": 6.589890480041504,
      "learning_rate": 0.00019565112675953744,
      "loss": 7.388,
      "step": 83000
    },
    {
      "epoch": 9.989169675090253,
      "grad_norm": 103.37834167480469,
      "learning_rate": 0.00019565001700238768,
      "loss": 7.4097,
      "step": 83010
    },
    {
      "epoch": 9.990373044524668,
      "grad_norm": 2.0575625896453857,
      "learning_rate": 0.00019564890710680873,
      "loss": 7.2956,
      "step": 83020
    },
    {
      "epoch": 9.991576413959086,
      "grad_norm": 3.804295778274536,
      "learning_rate": 0.00019564779707280218,
      "loss": 7.3008,
      "step": 83030
    },
    {
      "epoch": 9.992779783393502,
      "grad_norm": 56.05696105957031,
      "learning_rate": 0.00019564668690036965,
      "loss": 7.3577,
      "step": 83040
    },
    {
      "epoch": 9.993983152827917,
      "grad_norm": 8.73436450958252,
      "learning_rate": 0.00019564557658951278,
      "loss": 7.2376,
      "step": 83050
    },
    {
      "epoch": 9.995186522262335,
      "grad_norm": 7.325547695159912,
      "learning_rate": 0.0001956444661402331,
      "loss": 7.3478,
      "step": 83060
    },
    {
      "epoch": 9.99638989169675,
      "grad_norm": 30.338605880737305,
      "learning_rate": 0.00019564335555253233,
      "loss": 7.3733,
      "step": 83070
    },
    {
      "epoch": 9.997593261131167,
      "grad_norm": 113.83546447753906,
      "learning_rate": 0.00019564224482641196,
      "loss": 7.3178,
      "step": 83080
    },
    {
      "epoch": 9.998796630565584,
      "grad_norm": 54.39371871948242,
      "learning_rate": 0.00019564113396187364,
      "loss": 7.1751,
      "step": 83090
    },
    {
      "epoch": 10.0,
      "grad_norm": 7.800882339477539,
      "learning_rate": 0.00019564002295891904,
      "loss": 7.4801,
      "step": 83100
    },
    {
      "epoch": 10.0,
      "eval_loss": 7.518892765045166,
      "eval_runtime": 119.6337,
      "eval_samples_per_second": 61.747,
      "eval_steps_per_second": 7.724,
      "step": 83100
    },
    {
      "epoch": 10.001203369434416,
      "grad_norm": 544.3108520507812,
      "learning_rate": 0.00019563891181754967,
      "loss": 7.4537,
      "step": 83110
    },
    {
      "epoch": 10.002406738868833,
      "grad_norm": 45.86470031738281,
      "learning_rate": 0.0001956378005377672,
      "loss": 7.4266,
      "step": 83120
    },
    {
      "epoch": 10.00361010830325,
      "grad_norm": 23.418373107910156,
      "learning_rate": 0.0001956366891195732,
      "loss": 7.293,
      "step": 83130
    },
    {
      "epoch": 10.004813477737665,
      "grad_norm": 9.664619445800781,
      "learning_rate": 0.00019563557756296934,
      "loss": 7.3395,
      "step": 83140
    },
    {
      "epoch": 10.006016847172083,
      "grad_norm": 28.96971893310547,
      "learning_rate": 0.00019563446586795714,
      "loss": 7.3513,
      "step": 83150
    },
    {
      "epoch": 10.007220216606498,
      "grad_norm": 100.30049133300781,
      "learning_rate": 0.00019563335403453827,
      "loss": 7.368,
      "step": 83160
    },
    {
      "epoch": 10.008423586040914,
      "grad_norm": 61.96688461303711,
      "learning_rate": 0.00019563224206271436,
      "loss": 7.3882,
      "step": 83170
    },
    {
      "epoch": 10.009626955475332,
      "grad_norm": 56.330135345458984,
      "learning_rate": 0.00019563112995248693,
      "loss": 7.5772,
      "step": 83180
    },
    {
      "epoch": 10.010830324909747,
      "grad_norm": 564.260986328125,
      "learning_rate": 0.0001956300177038577,
      "loss": 7.4997,
      "step": 83190
    },
    {
      "epoch": 10.012033694344163,
      "grad_norm": 201.80166625976562,
      "learning_rate": 0.0001956289053168282,
      "loss": 7.4236,
      "step": 83200
    },
    {
      "epoch": 10.01323706377858,
      "grad_norm": 410.536376953125,
      "learning_rate": 0.00019562779279140005,
      "loss": 7.5006,
      "step": 83210
    },
    {
      "epoch": 10.014440433212997,
      "grad_norm": 3044.130126953125,
      "learning_rate": 0.00019562668012757488,
      "loss": 8.1232,
      "step": 83220
    },
    {
      "epoch": 10.015643802647412,
      "grad_norm": 36.56355285644531,
      "learning_rate": 0.00019562556732535428,
      "loss": 7.7171,
      "step": 83230
    },
    {
      "epoch": 10.01684717208183,
      "grad_norm": 79.8092269897461,
      "learning_rate": 0.0001956244543847399,
      "loss": 7.6449,
      "step": 83240
    },
    {
      "epoch": 10.018050541516246,
      "grad_norm": 17.27240753173828,
      "learning_rate": 0.00019562334130573332,
      "loss": 7.5587,
      "step": 83250
    },
    {
      "epoch": 10.019253910950662,
      "grad_norm": 18.181196212768555,
      "learning_rate": 0.00019562222808833615,
      "loss": 7.4222,
      "step": 83260
    },
    {
      "epoch": 10.020457280385079,
      "grad_norm": 103.67266845703125,
      "learning_rate": 0.00019562111473255,
      "loss": 7.3783,
      "step": 83270
    },
    {
      "epoch": 10.021660649819495,
      "grad_norm": 207.486572265625,
      "learning_rate": 0.0001956200012383765,
      "loss": 7.3435,
      "step": 83280
    },
    {
      "epoch": 10.02286401925391,
      "grad_norm": 7375.67578125,
      "learning_rate": 0.0001956188876058173,
      "loss": 7.4211,
      "step": 83290
    },
    {
      "epoch": 10.024067388688326,
      "grad_norm": 365.6496276855469,
      "learning_rate": 0.0001956177738348739,
      "loss": 7.5252,
      "step": 83300
    },
    {
      "epoch": 10.025270758122744,
      "grad_norm": 54.60540771484375,
      "learning_rate": 0.00019561665992554796,
      "loss": 7.4325,
      "step": 83310
    },
    {
      "epoch": 10.02647412755716,
      "grad_norm": 28.61625099182129,
      "learning_rate": 0.00019561554587784114,
      "loss": 7.4601,
      "step": 83320
    },
    {
      "epoch": 10.027677496991576,
      "grad_norm": 50.83510208129883,
      "learning_rate": 0.00019561443169175503,
      "loss": 7.4433,
      "step": 83330
    },
    {
      "epoch": 10.028880866425993,
      "grad_norm": 57.21589279174805,
      "learning_rate": 0.0001956133173672912,
      "loss": 7.4556,
      "step": 83340
    },
    {
      "epoch": 10.030084235860409,
      "grad_norm": 981.710205078125,
      "learning_rate": 0.00019561220290445133,
      "loss": 7.36,
      "step": 83350
    },
    {
      "epoch": 10.031287605294825,
      "grad_norm": 75.56356048583984,
      "learning_rate": 0.00019561108830323695,
      "loss": 7.3356,
      "step": 83360
    },
    {
      "epoch": 10.032490974729242,
      "grad_norm": 70.26111602783203,
      "learning_rate": 0.00019560997356364977,
      "loss": 7.4164,
      "step": 83370
    },
    {
      "epoch": 10.033694344163658,
      "grad_norm": 535.666748046875,
      "learning_rate": 0.00019560885868569132,
      "loss": 7.402,
      "step": 83380
    },
    {
      "epoch": 10.034897713598074,
      "grad_norm": 127.51241302490234,
      "learning_rate": 0.00019560774366936324,
      "loss": 7.4143,
      "step": 83390
    },
    {
      "epoch": 10.036101083032491,
      "grad_norm": 307.8184814453125,
      "learning_rate": 0.00019560662851466715,
      "loss": 7.4348,
      "step": 83400
    },
    {
      "epoch": 10.037304452466907,
      "grad_norm": 9.060751914978027,
      "learning_rate": 0.00019560551322160468,
      "loss": 7.4283,
      "step": 83410
    },
    {
      "epoch": 10.038507821901323,
      "grad_norm": 9.682147026062012,
      "learning_rate": 0.00019560439779017742,
      "loss": 7.3959,
      "step": 83420
    },
    {
      "epoch": 10.03971119133574,
      "grad_norm": 8.636005401611328,
      "learning_rate": 0.00019560328222038698,
      "loss": 7.4284,
      "step": 83430
    },
    {
      "epoch": 10.040914560770156,
      "grad_norm": 9.469674110412598,
      "learning_rate": 0.00019560216651223503,
      "loss": 7.2697,
      "step": 83440
    },
    {
      "epoch": 10.042117930204572,
      "grad_norm": 19.751909255981445,
      "learning_rate": 0.0001956010506657231,
      "loss": 7.418,
      "step": 83450
    },
    {
      "epoch": 10.04332129963899,
      "grad_norm": 166.0884552001953,
      "learning_rate": 0.0001955999346808529,
      "loss": 7.3375,
      "step": 83460
    },
    {
      "epoch": 10.044524669073406,
      "grad_norm": 5969.85107421875,
      "learning_rate": 0.00019559881855762593,
      "loss": 7.4188,
      "step": 83470
    },
    {
      "epoch": 10.045728038507821,
      "grad_norm": 115.87480163574219,
      "learning_rate": 0.0001955977022960439,
      "loss": 7.3576,
      "step": 83480
    },
    {
      "epoch": 10.046931407942239,
      "grad_norm": 5.944371700286865,
      "learning_rate": 0.00019559658589610838,
      "loss": 7.413,
      "step": 83490
    },
    {
      "epoch": 10.048134777376655,
      "grad_norm": 481.37255859375,
      "learning_rate": 0.00019559546935782101,
      "loss": 7.3182,
      "step": 83500
    },
    {
      "epoch": 10.04933814681107,
      "grad_norm": 40.134395599365234,
      "learning_rate": 0.00019559435268118338,
      "loss": 7.2567,
      "step": 83510
    },
    {
      "epoch": 10.050541516245488,
      "grad_norm": 183.8351593017578,
      "learning_rate": 0.00019559323586619713,
      "loss": 7.3474,
      "step": 83520
    },
    {
      "epoch": 10.051744885679904,
      "grad_norm": 118.63419342041016,
      "learning_rate": 0.00019559211891286387,
      "loss": 7.2396,
      "step": 83530
    },
    {
      "epoch": 10.05294825511432,
      "grad_norm": 27.949094772338867,
      "learning_rate": 0.0001955910018211852,
      "loss": 7.3118,
      "step": 83540
    },
    {
      "epoch": 10.054151624548737,
      "grad_norm": 102.37348175048828,
      "learning_rate": 0.00019558988459116274,
      "loss": 7.3392,
      "step": 83550
    },
    {
      "epoch": 10.055354993983153,
      "grad_norm": 71.63182067871094,
      "learning_rate": 0.00019558876722279816,
      "loss": 7.3955,
      "step": 83560
    },
    {
      "epoch": 10.056558363417569,
      "grad_norm": 1578.4404296875,
      "learning_rate": 0.000195587649716093,
      "loss": 7.4133,
      "step": 83570
    },
    {
      "epoch": 10.057761732851986,
      "grad_norm": 124.20844268798828,
      "learning_rate": 0.0001955865320710489,
      "loss": 7.3468,
      "step": 83580
    },
    {
      "epoch": 10.058965102286402,
      "grad_norm": 40.077510833740234,
      "learning_rate": 0.0001955854142876675,
      "loss": 7.2826,
      "step": 83590
    },
    {
      "epoch": 10.060168471720818,
      "grad_norm": 33.752506256103516,
      "learning_rate": 0.00019558429636595045,
      "loss": 7.3967,
      "step": 83600
    },
    {
      "epoch": 10.061371841155236,
      "grad_norm": 52.681339263916016,
      "learning_rate": 0.00019558317830589928,
      "loss": 7.3475,
      "step": 83610
    },
    {
      "epoch": 10.062575210589651,
      "grad_norm": 312.16448974609375,
      "learning_rate": 0.00019558206010751564,
      "loss": 7.2011,
      "step": 83620
    },
    {
      "epoch": 10.063778580024067,
      "grad_norm": 10.673030853271484,
      "learning_rate": 0.00019558094177080118,
      "loss": 7.265,
      "step": 83630
    },
    {
      "epoch": 10.064981949458483,
      "grad_norm": 60.73483657836914,
      "learning_rate": 0.00019557982329575747,
      "loss": 7.2372,
      "step": 83640
    },
    {
      "epoch": 10.0661853188929,
      "grad_norm": 30.778305053710938,
      "learning_rate": 0.00019557870468238622,
      "loss": 7.3471,
      "step": 83650
    },
    {
      "epoch": 10.067388688327316,
      "grad_norm": 69.0111312866211,
      "learning_rate": 0.00019557758593068892,
      "loss": 7.3048,
      "step": 83660
    },
    {
      "epoch": 10.068592057761732,
      "grad_norm": 30.080812454223633,
      "learning_rate": 0.00019557646704066733,
      "loss": 7.4109,
      "step": 83670
    },
    {
      "epoch": 10.06979542719615,
      "grad_norm": 32.3433723449707,
      "learning_rate": 0.00019557534801232292,
      "loss": 7.365,
      "step": 83680
    },
    {
      "epoch": 10.070998796630565,
      "grad_norm": 71.39273834228516,
      "learning_rate": 0.0001955742288456574,
      "loss": 7.408,
      "step": 83690
    },
    {
      "epoch": 10.072202166064981,
      "grad_norm": 133.87762451171875,
      "learning_rate": 0.00019557310954067242,
      "loss": 7.389,
      "step": 83700
    },
    {
      "epoch": 10.073405535499399,
      "grad_norm": 63.1323356628418,
      "learning_rate": 0.00019557199009736953,
      "loss": 7.4022,
      "step": 83710
    },
    {
      "epoch": 10.074608904933815,
      "grad_norm": 81.40836334228516,
      "learning_rate": 0.00019557087051575036,
      "loss": 7.2864,
      "step": 83720
    },
    {
      "epoch": 10.07581227436823,
      "grad_norm": 37.00311279296875,
      "learning_rate": 0.00019556975079581656,
      "loss": 7.2,
      "step": 83730
    },
    {
      "epoch": 10.077015643802648,
      "grad_norm": 32.62196350097656,
      "learning_rate": 0.0001955686309375697,
      "loss": 7.3818,
      "step": 83740
    },
    {
      "epoch": 10.078219013237064,
      "grad_norm": 931.1118774414062,
      "learning_rate": 0.0001955675109410115,
      "loss": 7.3582,
      "step": 83750
    },
    {
      "epoch": 10.07942238267148,
      "grad_norm": 74.43765258789062,
      "learning_rate": 0.0001955663908061435,
      "loss": 7.3079,
      "step": 83760
    },
    {
      "epoch": 10.080625752105897,
      "grad_norm": 387.19573974609375,
      "learning_rate": 0.00019556527053296731,
      "loss": 7.3001,
      "step": 83770
    },
    {
      "epoch": 10.081829121540313,
      "grad_norm": 535.4677734375,
      "learning_rate": 0.0001955641501214846,
      "loss": 7.3723,
      "step": 83780
    },
    {
      "epoch": 10.083032490974729,
      "grad_norm": 440.4124450683594,
      "learning_rate": 0.00019556302957169695,
      "loss": 7.3969,
      "step": 83790
    },
    {
      "epoch": 10.084235860409146,
      "grad_norm": 26.299827575683594,
      "learning_rate": 0.000195561908883606,
      "loss": 7.3512,
      "step": 83800
    },
    {
      "epoch": 10.085439229843562,
      "grad_norm": 9.611572265625,
      "learning_rate": 0.0001955607880572134,
      "loss": 7.3437,
      "step": 83810
    },
    {
      "epoch": 10.086642599277978,
      "grad_norm": 10106.060546875,
      "learning_rate": 0.00019555966709252072,
      "loss": 7.4588,
      "step": 83820
    },
    {
      "epoch": 10.087845968712395,
      "grad_norm": 51.20191955566406,
      "learning_rate": 0.00019555854598952963,
      "loss": 7.4455,
      "step": 83830
    },
    {
      "epoch": 10.089049338146811,
      "grad_norm": 49.935585021972656,
      "learning_rate": 0.00019555742474824172,
      "loss": 7.3975,
      "step": 83840
    },
    {
      "epoch": 10.090252707581227,
      "grad_norm": 672.3780517578125,
      "learning_rate": 0.00019555630336865864,
      "loss": 7.4609,
      "step": 83850
    },
    {
      "epoch": 10.091456077015645,
      "grad_norm": 791.9589233398438,
      "learning_rate": 0.000195555181850782,
      "loss": 7.4114,
      "step": 83860
    },
    {
      "epoch": 10.09265944645006,
      "grad_norm": 96690.9765625,
      "learning_rate": 0.0001955540601946134,
      "loss": 7.4866,
      "step": 83870
    },
    {
      "epoch": 10.093862815884476,
      "grad_norm": 169.25511169433594,
      "learning_rate": 0.00019555293840015447,
      "loss": 7.5957,
      "step": 83880
    },
    {
      "epoch": 10.095066185318894,
      "grad_norm": 217.37278747558594,
      "learning_rate": 0.00019555181646740687,
      "loss": 7.5802,
      "step": 83890
    },
    {
      "epoch": 10.09626955475331,
      "grad_norm": 143.0635986328125,
      "learning_rate": 0.00019555069439637223,
      "loss": 7.6048,
      "step": 83900
    },
    {
      "epoch": 10.097472924187725,
      "grad_norm": 792.082763671875,
      "learning_rate": 0.00019554957218705207,
      "loss": 7.4206,
      "step": 83910
    },
    {
      "epoch": 10.098676293622143,
      "grad_norm": 181.4624786376953,
      "learning_rate": 0.00019554844983944817,
      "loss": 7.4301,
      "step": 83920
    },
    {
      "epoch": 10.099879663056559,
      "grad_norm": 1086.484375,
      "learning_rate": 0.000195547327353562,
      "loss": 7.5052,
      "step": 83930
    },
    {
      "epoch": 10.101083032490974,
      "grad_norm": 216.3699951171875,
      "learning_rate": 0.0001955462047293953,
      "loss": 7.4086,
      "step": 83940
    },
    {
      "epoch": 10.102286401925392,
      "grad_norm": 159.45480346679688,
      "learning_rate": 0.00019554508196694962,
      "loss": 7.3906,
      "step": 83950
    },
    {
      "epoch": 10.103489771359808,
      "grad_norm": 36455.9140625,
      "learning_rate": 0.00019554395906622664,
      "loss": 7.4926,
      "step": 83960
    },
    {
      "epoch": 10.104693140794224,
      "grad_norm": 6384.41796875,
      "learning_rate": 0.000195542836027228,
      "loss": 7.4972,
      "step": 83970
    },
    {
      "epoch": 10.10589651022864,
      "grad_norm": 6696.1494140625,
      "learning_rate": 0.00019554171284995523,
      "loss": 7.3718,
      "step": 83980
    },
    {
      "epoch": 10.107099879663057,
      "grad_norm": 799.7194213867188,
      "learning_rate": 0.00019554058953441005,
      "loss": 7.4873,
      "step": 83990
    },
    {
      "epoch": 10.108303249097473,
      "grad_norm": 1213.58837890625,
      "learning_rate": 0.000195539466080594,
      "loss": 7.5494,
      "step": 84000
    },
    {
      "epoch": 10.109506618531888,
      "grad_norm": 3183014.75,
      "learning_rate": 0.0001955383424885088,
      "loss": 7.5762,
      "step": 84010
    },
    {
      "epoch": 10.110709987966306,
      "grad_norm": 44857.57421875,
      "learning_rate": 0.00019553721875815603,
      "loss": 7.493,
      "step": 84020
    },
    {
      "epoch": 10.111913357400722,
      "grad_norm": 1882.9136962890625,
      "learning_rate": 0.0001955360948895373,
      "loss": 7.4567,
      "step": 84030
    },
    {
      "epoch": 10.113116726835138,
      "grad_norm": 500.9793395996094,
      "learning_rate": 0.00019553497088265425,
      "loss": 7.4522,
      "step": 84040
    },
    {
      "epoch": 10.114320096269555,
      "grad_norm": 417.8712158203125,
      "learning_rate": 0.00019553384673750852,
      "loss": 7.4704,
      "step": 84050
    },
    {
      "epoch": 10.115523465703971,
      "grad_norm": 1876.11083984375,
      "learning_rate": 0.0001955327224541017,
      "loss": 7.4824,
      "step": 84060
    },
    {
      "epoch": 10.116726835138387,
      "grad_norm": 299.41705322265625,
      "learning_rate": 0.00019553159803243548,
      "loss": 7.4902,
      "step": 84070
    },
    {
      "epoch": 10.117930204572804,
      "grad_norm": 82.94864654541016,
      "learning_rate": 0.00019553047347251146,
      "loss": 7.3832,
      "step": 84080
    },
    {
      "epoch": 10.11913357400722,
      "grad_norm": 36297.7109375,
      "learning_rate": 0.00019552934877433122,
      "loss": 7.3803,
      "step": 84090
    },
    {
      "epoch": 10.120336943441636,
      "grad_norm": 783.7518310546875,
      "learning_rate": 0.00019552822393789645,
      "loss": 7.3939,
      "step": 84100
    },
    {
      "epoch": 10.121540312876053,
      "grad_norm": 3914.262939453125,
      "learning_rate": 0.00019552709896320875,
      "loss": 7.3988,
      "step": 84110
    },
    {
      "epoch": 10.12274368231047,
      "grad_norm": 494.4291076660156,
      "learning_rate": 0.00019552597385026976,
      "loss": 7.4736,
      "step": 84120
    },
    {
      "epoch": 10.123947051744885,
      "grad_norm": 3709.756591796875,
      "learning_rate": 0.0001955248485990811,
      "loss": 7.481,
      "step": 84130
    },
    {
      "epoch": 10.125150421179303,
      "grad_norm": 670.2003173828125,
      "learning_rate": 0.00019552372320964437,
      "loss": 7.4812,
      "step": 84140
    },
    {
      "epoch": 10.126353790613718,
      "grad_norm": 566.203125,
      "learning_rate": 0.00019552259768196127,
      "loss": 7.6785,
      "step": 84150
    },
    {
      "epoch": 10.127557160048134,
      "grad_norm": 552.3314208984375,
      "learning_rate": 0.00019552147201603337,
      "loss": 7.461,
      "step": 84160
    },
    {
      "epoch": 10.128760529482552,
      "grad_norm": 42738.7734375,
      "learning_rate": 0.0001955203462118623,
      "loss": 7.5014,
      "step": 84170
    },
    {
      "epoch": 10.129963898916968,
      "grad_norm": 1573.9173583984375,
      "learning_rate": 0.00019551922026944974,
      "loss": 7.5085,
      "step": 84180
    },
    {
      "epoch": 10.131167268351383,
      "grad_norm": 1960.232421875,
      "learning_rate": 0.00019551809418879727,
      "loss": 7.4702,
      "step": 84190
    },
    {
      "epoch": 10.132370637785801,
      "grad_norm": 2213.644287109375,
      "learning_rate": 0.00019551696796990653,
      "loss": 7.5749,
      "step": 84200
    },
    {
      "epoch": 10.133574007220217,
      "grad_norm": 541.3018188476562,
      "learning_rate": 0.00019551584161277917,
      "loss": 7.5032,
      "step": 84210
    },
    {
      "epoch": 10.134777376654633,
      "grad_norm": 43037.62109375,
      "learning_rate": 0.0001955147151174168,
      "loss": 7.664,
      "step": 84220
    },
    {
      "epoch": 10.13598074608905,
      "grad_norm": 5107.90185546875,
      "learning_rate": 0.00019551358848382105,
      "loss": 7.6627,
      "step": 84230
    },
    {
      "epoch": 10.137184115523466,
      "grad_norm": 8360.072265625,
      "learning_rate": 0.00019551246171199354,
      "loss": 7.6046,
      "step": 84240
    },
    {
      "epoch": 10.138387484957882,
      "grad_norm": 2013.773681640625,
      "learning_rate": 0.00019551133480193592,
      "loss": 7.6087,
      "step": 84250
    },
    {
      "epoch": 10.1395908543923,
      "grad_norm": 10944.0859375,
      "learning_rate": 0.00019551020775364984,
      "loss": 7.7149,
      "step": 84260
    },
    {
      "epoch": 10.140794223826715,
      "grad_norm": 1577.2322998046875,
      "learning_rate": 0.0001955090805671369,
      "loss": 7.6496,
      "step": 84270
    },
    {
      "epoch": 10.14199759326113,
      "grad_norm": 38615.421875,
      "learning_rate": 0.00019550795324239874,
      "loss": 7.8053,
      "step": 84280
    },
    {
      "epoch": 10.143200962695548,
      "grad_norm": 17308.25,
      "learning_rate": 0.00019550682577943699,
      "loss": 7.7949,
      "step": 84290
    },
    {
      "epoch": 10.144404332129964,
      "grad_norm": 21218.119140625,
      "learning_rate": 0.00019550569817825328,
      "loss": 7.818,
      "step": 84300
    },
    {
      "epoch": 10.14560770156438,
      "grad_norm": 7600.408203125,
      "learning_rate": 0.00019550457043884928,
      "loss": 7.8349,
      "step": 84310
    },
    {
      "epoch": 10.146811070998796,
      "grad_norm": 4755.52734375,
      "learning_rate": 0.00019550344256122652,
      "loss": 7.8572,
      "step": 84320
    },
    {
      "epoch": 10.148014440433213,
      "grad_norm": 11302.9296875,
      "learning_rate": 0.00019550231454538674,
      "loss": 7.8566,
      "step": 84330
    },
    {
      "epoch": 10.14921780986763,
      "grad_norm": 9766.044921875,
      "learning_rate": 0.00019550118639133154,
      "loss": 7.8228,
      "step": 84340
    },
    {
      "epoch": 10.150421179302045,
      "grad_norm": 2017.5308837890625,
      "learning_rate": 0.00019550005809906252,
      "loss": 7.8841,
      "step": 84350
    },
    {
      "epoch": 10.151624548736462,
      "grad_norm": 2552.633056640625,
      "learning_rate": 0.00019549892966858135,
      "loss": 7.7728,
      "step": 84360
    },
    {
      "epoch": 10.152827918170878,
      "grad_norm": 21814.564453125,
      "learning_rate": 0.00019549780109988965,
      "loss": 7.8626,
      "step": 84370
    },
    {
      "epoch": 10.154031287605294,
      "grad_norm": 34714.97265625,
      "learning_rate": 0.00019549667239298905,
      "loss": 7.8497,
      "step": 84380
    },
    {
      "epoch": 10.155234657039712,
      "grad_norm": 31947.96484375,
      "learning_rate": 0.0001954955435478812,
      "loss": 7.911,
      "step": 84390
    },
    {
      "epoch": 10.156438026474127,
      "grad_norm": 494912.46875,
      "learning_rate": 0.00019549441456456773,
      "loss": 7.8158,
      "step": 84400
    },
    {
      "epoch": 10.157641395908543,
      "grad_norm": 47175.609375,
      "learning_rate": 0.00019549328544305023,
      "loss": 7.8153,
      "step": 84410
    },
    {
      "epoch": 10.15884476534296,
      "grad_norm": 51441.97265625,
      "learning_rate": 0.0001954921561833304,
      "loss": 7.8902,
      "step": 84420
    },
    {
      "epoch": 10.160048134777377,
      "grad_norm": 2050.4072265625,
      "learning_rate": 0.0001954910267854098,
      "loss": 7.7684,
      "step": 84430
    },
    {
      "epoch": 10.161251504211792,
      "grad_norm": 4532.08056640625,
      "learning_rate": 0.00019548989724929014,
      "loss": 7.8678,
      "step": 84440
    },
    {
      "epoch": 10.16245487364621,
      "grad_norm": 3253.48046875,
      "learning_rate": 0.00019548876757497302,
      "loss": 7.8032,
      "step": 84450
    },
    {
      "epoch": 10.163658243080626,
      "grad_norm": 1844.1534423828125,
      "learning_rate": 0.0001954876377624601,
      "loss": 7.6659,
      "step": 84460
    },
    {
      "epoch": 10.164861612515042,
      "grad_norm": 6227.05126953125,
      "learning_rate": 0.00019548650781175297,
      "loss": 7.7648,
      "step": 84470
    },
    {
      "epoch": 10.166064981949459,
      "grad_norm": 3526.93798828125,
      "learning_rate": 0.0001954853777228533,
      "loss": 7.6573,
      "step": 84480
    },
    {
      "epoch": 10.167268351383875,
      "grad_norm": 437.5130615234375,
      "learning_rate": 0.0001954842474957627,
      "loss": 7.6103,
      "step": 84490
    },
    {
      "epoch": 10.16847172081829,
      "grad_norm": 290.9523620605469,
      "learning_rate": 0.00019548311713048282,
      "loss": 7.5856,
      "step": 84500
    },
    {
      "epoch": 10.169675090252708,
      "grad_norm": 184.8275146484375,
      "learning_rate": 0.0001954819866270153,
      "loss": 7.693,
      "step": 84510
    },
    {
      "epoch": 10.170878459687124,
      "grad_norm": 15172.384765625,
      "learning_rate": 0.0001954808559853618,
      "loss": 7.7003,
      "step": 84520
    },
    {
      "epoch": 10.17208182912154,
      "grad_norm": 2435.298583984375,
      "learning_rate": 0.0001954797252055239,
      "loss": 7.7421,
      "step": 84530
    },
    {
      "epoch": 10.173285198555957,
      "grad_norm": 1336.593505859375,
      "learning_rate": 0.00019547859428750326,
      "loss": 7.6244,
      "step": 84540
    },
    {
      "epoch": 10.174488567990373,
      "grad_norm": 1790.1878662109375,
      "learning_rate": 0.00019547746323130152,
      "loss": 7.6178,
      "step": 84550
    },
    {
      "epoch": 10.175691937424789,
      "grad_norm": 996.5396118164062,
      "learning_rate": 0.0001954763320369203,
      "loss": 7.5489,
      "step": 84560
    },
    {
      "epoch": 10.176895306859207,
      "grad_norm": 304.2719421386719,
      "learning_rate": 0.0001954752007043613,
      "loss": 7.5244,
      "step": 84570
    },
    {
      "epoch": 10.178098676293622,
      "grad_norm": 1636.525634765625,
      "learning_rate": 0.0001954740692336261,
      "loss": 7.6303,
      "step": 84580
    },
    {
      "epoch": 10.179302045728038,
      "grad_norm": 1122.403076171875,
      "learning_rate": 0.00019547293762471631,
      "loss": 7.646,
      "step": 84590
    },
    {
      "epoch": 10.180505415162456,
      "grad_norm": 3601.208251953125,
      "learning_rate": 0.00019547180587763365,
      "loss": 7.6218,
      "step": 84600
    },
    {
      "epoch": 10.181708784596871,
      "grad_norm": 224.7150115966797,
      "learning_rate": 0.00019547067399237973,
      "loss": 7.576,
      "step": 84610
    },
    {
      "epoch": 10.182912154031287,
      "grad_norm": 5668.404296875,
      "learning_rate": 0.00019546954196895616,
      "loss": 7.592,
      "step": 84620
    },
    {
      "epoch": 10.184115523465705,
      "grad_norm": 2218.216796875,
      "learning_rate": 0.00019546840980736455,
      "loss": 7.6003,
      "step": 84630
    },
    {
      "epoch": 10.18531889290012,
      "grad_norm": 3921.322998046875,
      "learning_rate": 0.00019546727750760663,
      "loss": 7.517,
      "step": 84640
    },
    {
      "epoch": 10.186522262334536,
      "grad_norm": 186.01300048828125,
      "learning_rate": 0.00019546614506968398,
      "loss": 7.6106,
      "step": 84650
    },
    {
      "epoch": 10.187725631768952,
      "grad_norm": 771.4862060546875,
      "learning_rate": 0.00019546501249359822,
      "loss": 7.5792,
      "step": 84660
    },
    {
      "epoch": 10.18892900120337,
      "grad_norm": 283.5538635253906,
      "learning_rate": 0.00019546387977935106,
      "loss": 7.5603,
      "step": 84670
    },
    {
      "epoch": 10.190132370637786,
      "grad_norm": 736.301513671875,
      "learning_rate": 0.00019546274692694408,
      "loss": 7.5591,
      "step": 84680
    },
    {
      "epoch": 10.191335740072201,
      "grad_norm": 1668.3681640625,
      "learning_rate": 0.0001954616139363789,
      "loss": 7.6162,
      "step": 84690
    },
    {
      "epoch": 10.192539109506619,
      "grad_norm": 12183.900390625,
      "learning_rate": 0.00019546048080765723,
      "loss": 7.4798,
      "step": 84700
    },
    {
      "epoch": 10.193742478941035,
      "grad_norm": 1348.5616455078125,
      "learning_rate": 0.00019545934754078067,
      "loss": 7.5387,
      "step": 84710
    },
    {
      "epoch": 10.19494584837545,
      "grad_norm": 3114.279052734375,
      "learning_rate": 0.00019545821413575088,
      "loss": 7.4717,
      "step": 84720
    },
    {
      "epoch": 10.196149217809868,
      "grad_norm": 4764.4140625,
      "learning_rate": 0.00019545708059256947,
      "loss": 7.4346,
      "step": 84730
    },
    {
      "epoch": 10.197352587244284,
      "grad_norm": 8097.38720703125,
      "learning_rate": 0.0001954559469112381,
      "loss": 7.3994,
      "step": 84740
    },
    {
      "epoch": 10.1985559566787,
      "grad_norm": 4141.76318359375,
      "learning_rate": 0.00019545481309175838,
      "loss": 7.4519,
      "step": 84750
    },
    {
      "epoch": 10.199759326113117,
      "grad_norm": 464.6991882324219,
      "learning_rate": 0.00019545367913413204,
      "loss": 7.404,
      "step": 84760
    },
    {
      "epoch": 10.200962695547533,
      "grad_norm": 194.63821411132812,
      "learning_rate": 0.00019545254503836065,
      "loss": 7.4688,
      "step": 84770
    },
    {
      "epoch": 10.202166064981949,
      "grad_norm": 3374.51318359375,
      "learning_rate": 0.0001954514108044458,
      "loss": 7.463,
      "step": 84780
    },
    {
      "epoch": 10.203369434416366,
      "grad_norm": 3136.606689453125,
      "learning_rate": 0.00019545027643238925,
      "loss": 7.4433,
      "step": 84790
    },
    {
      "epoch": 10.204572803850782,
      "grad_norm": 2281.577880859375,
      "learning_rate": 0.00019544914192219255,
      "loss": 7.5009,
      "step": 84800
    },
    {
      "epoch": 10.205776173285198,
      "grad_norm": 558.43212890625,
      "learning_rate": 0.00019544800727385738,
      "loss": 7.5724,
      "step": 84810
    },
    {
      "epoch": 10.206979542719615,
      "grad_norm": 4649.240234375,
      "learning_rate": 0.0001954468724873854,
      "loss": 7.5039,
      "step": 84820
    },
    {
      "epoch": 10.208182912154031,
      "grad_norm": 12884.7939453125,
      "learning_rate": 0.00019544573756277822,
      "loss": 7.5177,
      "step": 84830
    },
    {
      "epoch": 10.209386281588447,
      "grad_norm": 264.2885437011719,
      "learning_rate": 0.00019544460250003749,
      "loss": 7.439,
      "step": 84840
    },
    {
      "epoch": 10.210589651022865,
      "grad_norm": 546.5271606445312,
      "learning_rate": 0.00019544346729916486,
      "loss": 7.563,
      "step": 84850
    },
    {
      "epoch": 10.21179302045728,
      "grad_norm": 61.41135025024414,
      "learning_rate": 0.00019544233196016199,
      "loss": 7.4349,
      "step": 84860
    },
    {
      "epoch": 10.212996389891696,
      "grad_norm": 11019.5673828125,
      "learning_rate": 0.00019544119648303045,
      "loss": 7.4064,
      "step": 84870
    },
    {
      "epoch": 10.214199759326114,
      "grad_norm": 91.89659118652344,
      "learning_rate": 0.00019544006086777196,
      "loss": 7.4024,
      "step": 84880
    },
    {
      "epoch": 10.21540312876053,
      "grad_norm": 1944.5679931640625,
      "learning_rate": 0.00019543892511438813,
      "loss": 7.4678,
      "step": 84890
    },
    {
      "epoch": 10.216606498194945,
      "grad_norm": 2484.067626953125,
      "learning_rate": 0.00019543778922288064,
      "loss": 7.5887,
      "step": 84900
    },
    {
      "epoch": 10.217809867629363,
      "grad_norm": 2496.830810546875,
      "learning_rate": 0.0001954366531932511,
      "loss": 7.4322,
      "step": 84910
    },
    {
      "epoch": 10.219013237063779,
      "grad_norm": 35778.47265625,
      "learning_rate": 0.00019543551702550115,
      "loss": 7.5733,
      "step": 84920
    },
    {
      "epoch": 10.220216606498195,
      "grad_norm": 329.30694580078125,
      "learning_rate": 0.00019543438071963241,
      "loss": 7.5197,
      "step": 84930
    },
    {
      "epoch": 10.221419975932612,
      "grad_norm": 501.95611572265625,
      "learning_rate": 0.00019543324427564662,
      "loss": 7.4337,
      "step": 84940
    },
    {
      "epoch": 10.222623345367028,
      "grad_norm": 1292.3909912109375,
      "learning_rate": 0.00019543210769354535,
      "loss": 7.499,
      "step": 84950
    },
    {
      "epoch": 10.223826714801444,
      "grad_norm": 24684.328125,
      "learning_rate": 0.00019543097097333024,
      "loss": 7.5541,
      "step": 84960
    },
    {
      "epoch": 10.225030084235861,
      "grad_norm": 1482.8800048828125,
      "learning_rate": 0.00019542983411500293,
      "loss": 7.4271,
      "step": 84970
    },
    {
      "epoch": 10.226233453670277,
      "grad_norm": 2366.8427734375,
      "learning_rate": 0.00019542869711856512,
      "loss": 7.5985,
      "step": 84980
    },
    {
      "epoch": 10.227436823104693,
      "grad_norm": 2164.325439453125,
      "learning_rate": 0.00019542755998401844,
      "loss": 7.5098,
      "step": 84990
    },
    {
      "epoch": 10.22864019253911,
      "grad_norm": 1001.6786499023438,
      "learning_rate": 0.0001954264227113645,
      "loss": 7.4615,
      "step": 85000
    },
    {
      "epoch": 10.229843561973526,
      "grad_norm": 262.1122131347656,
      "learning_rate": 0.00019542528530060497,
      "loss": 7.5176,
      "step": 85010
    },
    {
      "epoch": 10.231046931407942,
      "grad_norm": 130.0224151611328,
      "learning_rate": 0.00019542414775174148,
      "loss": 7.3711,
      "step": 85020
    },
    {
      "epoch": 10.232250300842358,
      "grad_norm": 402.8792419433594,
      "learning_rate": 0.00019542301006477572,
      "loss": 7.4132,
      "step": 85030
    },
    {
      "epoch": 10.233453670276775,
      "grad_norm": 49.24036407470703,
      "learning_rate": 0.0001954218722397093,
      "loss": 7.4445,
      "step": 85040
    },
    {
      "epoch": 10.234657039711191,
      "grad_norm": 931.08984375,
      "learning_rate": 0.00019542073427654383,
      "loss": 7.3981,
      "step": 85050
    },
    {
      "epoch": 10.235860409145607,
      "grad_norm": 81.37020874023438,
      "learning_rate": 0.00019541959617528102,
      "loss": 7.3585,
      "step": 85060
    },
    {
      "epoch": 10.237063778580024,
      "grad_norm": 288.3209228515625,
      "learning_rate": 0.00019541845793592252,
      "loss": 7.3375,
      "step": 85070
    },
    {
      "epoch": 10.23826714801444,
      "grad_norm": 81.19265747070312,
      "learning_rate": 0.00019541731955846994,
      "loss": 7.3465,
      "step": 85080
    },
    {
      "epoch": 10.239470517448856,
      "grad_norm": 68.54808807373047,
      "learning_rate": 0.00019541618104292494,
      "loss": 7.4104,
      "step": 85090
    },
    {
      "epoch": 10.240673886883274,
      "grad_norm": 103.49225616455078,
      "learning_rate": 0.00019541504238928915,
      "loss": 7.4004,
      "step": 85100
    },
    {
      "epoch": 10.24187725631769,
      "grad_norm": 142.51296997070312,
      "learning_rate": 0.00019541390359756425,
      "loss": 7.3708,
      "step": 85110
    },
    {
      "epoch": 10.243080625752105,
      "grad_norm": 6.464770793914795,
      "learning_rate": 0.0001954127646677519,
      "loss": 7.497,
      "step": 85120
    },
    {
      "epoch": 10.244283995186523,
      "grad_norm": 33.73357009887695,
      "learning_rate": 0.00019541162559985368,
      "loss": 7.4298,
      "step": 85130
    },
    {
      "epoch": 10.245487364620939,
      "grad_norm": 22.592464447021484,
      "learning_rate": 0.0001954104863938713,
      "loss": 7.3442,
      "step": 85140
    },
    {
      "epoch": 10.246690734055354,
      "grad_norm": 17.11105728149414,
      "learning_rate": 0.0001954093470498064,
      "loss": 7.522,
      "step": 85150
    },
    {
      "epoch": 10.247894103489772,
      "grad_norm": 20.17689323425293,
      "learning_rate": 0.0001954082075676606,
      "loss": 7.3856,
      "step": 85160
    },
    {
      "epoch": 10.249097472924188,
      "grad_norm": 30.028621673583984,
      "learning_rate": 0.0001954070679474356,
      "loss": 7.3863,
      "step": 85170
    },
    {
      "epoch": 10.250300842358604,
      "grad_norm": 21.267133712768555,
      "learning_rate": 0.00019540592818913298,
      "loss": 7.4415,
      "step": 85180
    },
    {
      "epoch": 10.251504211793021,
      "grad_norm": 19.69413185119629,
      "learning_rate": 0.00019540478829275447,
      "loss": 7.5127,
      "step": 85190
    },
    {
      "epoch": 10.252707581227437,
      "grad_norm": 3.984941244125366,
      "learning_rate": 0.0001954036482583016,
      "loss": 7.5265,
      "step": 85200
    },
    {
      "epoch": 10.253910950661853,
      "grad_norm": 2.579620599746704,
      "learning_rate": 0.00019540250808577618,
      "loss": 7.4825,
      "step": 85210
    },
    {
      "epoch": 10.25511432009627,
      "grad_norm": 6.304272174835205,
      "learning_rate": 0.00019540136777517974,
      "loss": 7.5466,
      "step": 85220
    },
    {
      "epoch": 10.256317689530686,
      "grad_norm": 4.438540458679199,
      "learning_rate": 0.00019540022732651398,
      "loss": 7.5504,
      "step": 85230
    },
    {
      "epoch": 10.257521058965102,
      "grad_norm": 2.3285071849823,
      "learning_rate": 0.00019539908673978052,
      "loss": 7.5308,
      "step": 85240
    },
    {
      "epoch": 10.25872442839952,
      "grad_norm": 7.048343658447266,
      "learning_rate": 0.00019539794601498102,
      "loss": 7.4571,
      "step": 85250
    },
    {
      "epoch": 10.259927797833935,
      "grad_norm": 3.191307306289673,
      "learning_rate": 0.00019539680515211715,
      "loss": 7.479,
      "step": 85260
    },
    {
      "epoch": 10.261131167268351,
      "grad_norm": 5.28131628036499,
      "learning_rate": 0.00019539566415119057,
      "loss": 7.4891,
      "step": 85270
    },
    {
      "epoch": 10.262334536702769,
      "grad_norm": 4.179459095001221,
      "learning_rate": 0.0001953945230122029,
      "loss": 7.4649,
      "step": 85280
    },
    {
      "epoch": 10.263537906137184,
      "grad_norm": 3.379351854324341,
      "learning_rate": 0.0001953933817351558,
      "loss": 7.4909,
      "step": 85290
    },
    {
      "epoch": 10.2647412755716,
      "grad_norm": 9.762781143188477,
      "learning_rate": 0.00019539224032005093,
      "loss": 7.4819,
      "step": 85300
    },
    {
      "epoch": 10.265944645006018,
      "grad_norm": 8.950075149536133,
      "learning_rate": 0.00019539109876688993,
      "loss": 7.5743,
      "step": 85310
    },
    {
      "epoch": 10.267148014440433,
      "grad_norm": 9.266764640808105,
      "learning_rate": 0.00019538995707567446,
      "loss": 7.493,
      "step": 85320
    },
    {
      "epoch": 10.26835138387485,
      "grad_norm": 2.029279947280884,
      "learning_rate": 0.00019538881524640614,
      "loss": 7.4323,
      "step": 85330
    },
    {
      "epoch": 10.269554753309265,
      "grad_norm": 1.764323115348816,
      "learning_rate": 0.00019538767327908672,
      "loss": 7.3986,
      "step": 85340
    },
    {
      "epoch": 10.270758122743683,
      "grad_norm": 3.6983046531677246,
      "learning_rate": 0.00019538653117371775,
      "loss": 7.4004,
      "step": 85350
    },
    {
      "epoch": 10.271961492178098,
      "grad_norm": 1.9362916946411133,
      "learning_rate": 0.00019538538893030093,
      "loss": 7.4334,
      "step": 85360
    },
    {
      "epoch": 10.273164861612514,
      "grad_norm": 3.71360445022583,
      "learning_rate": 0.0001953842465488379,
      "loss": 7.5023,
      "step": 85370
    },
    {
      "epoch": 10.274368231046932,
      "grad_norm": 3.131895065307617,
      "learning_rate": 0.0001953831040293303,
      "loss": 7.5417,
      "step": 85380
    },
    {
      "epoch": 10.275571600481348,
      "grad_norm": 1.9646403789520264,
      "learning_rate": 0.0001953819613717798,
      "loss": 7.4674,
      "step": 85390
    },
    {
      "epoch": 10.276774969915763,
      "grad_norm": 2.816831111907959,
      "learning_rate": 0.00019538081857618805,
      "loss": 7.4984,
      "step": 85400
    },
    {
      "epoch": 10.277978339350181,
      "grad_norm": 1.4591765403747559,
      "learning_rate": 0.00019537967564255671,
      "loss": 7.4286,
      "step": 85410
    },
    {
      "epoch": 10.279181708784597,
      "grad_norm": 7.126994609832764,
      "learning_rate": 0.00019537853257088744,
      "loss": 7.4799,
      "step": 85420
    },
    {
      "epoch": 10.280385078219012,
      "grad_norm": 1.5111916065216064,
      "learning_rate": 0.00019537738936118188,
      "loss": 7.4561,
      "step": 85430
    },
    {
      "epoch": 10.28158844765343,
      "grad_norm": 3.422874927520752,
      "learning_rate": 0.00019537624601344172,
      "loss": 7.5421,
      "step": 85440
    },
    {
      "epoch": 10.282791817087846,
      "grad_norm": 8.737654685974121,
      "learning_rate": 0.00019537510252766853,
      "loss": 7.4657,
      "step": 85450
    },
    {
      "epoch": 10.283995186522262,
      "grad_norm": 2.050358533859253,
      "learning_rate": 0.00019537395890386404,
      "loss": 7.5599,
      "step": 85460
    },
    {
      "epoch": 10.28519855595668,
      "grad_norm": 5.419197082519531,
      "learning_rate": 0.0001953728151420299,
      "loss": 7.3527,
      "step": 85470
    },
    {
      "epoch": 10.286401925391095,
      "grad_norm": 4.262731552124023,
      "learning_rate": 0.00019537167124216774,
      "loss": 7.5019,
      "step": 85480
    },
    {
      "epoch": 10.28760529482551,
      "grad_norm": 1.5936861038208008,
      "learning_rate": 0.00019537052720427922,
      "loss": 7.5104,
      "step": 85490
    },
    {
      "epoch": 10.288808664259928,
      "grad_norm": 6.046593189239502,
      "learning_rate": 0.000195369383028366,
      "loss": 7.4177,
      "step": 85500
    },
    {
      "epoch": 10.290012033694344,
      "grad_norm": 4.11210298538208,
      "learning_rate": 0.00019536823871442978,
      "loss": 7.4079,
      "step": 85510
    },
    {
      "epoch": 10.29121540312876,
      "grad_norm": 4.220069885253906,
      "learning_rate": 0.00019536709426247212,
      "loss": 7.4925,
      "step": 85520
    },
    {
      "epoch": 10.292418772563177,
      "grad_norm": 2.521742105484009,
      "learning_rate": 0.00019536594967249475,
      "loss": 7.4716,
      "step": 85530
    },
    {
      "epoch": 10.293622141997593,
      "grad_norm": 2.1066694259643555,
      "learning_rate": 0.0001953648049444993,
      "loss": 7.495,
      "step": 85540
    },
    {
      "epoch": 10.294825511432009,
      "grad_norm": 4.6667656898498535,
      "learning_rate": 0.00019536366007848744,
      "loss": 7.4612,
      "step": 85550
    },
    {
      "epoch": 10.296028880866427,
      "grad_norm": 1.8581546545028687,
      "learning_rate": 0.00019536251507446082,
      "loss": 7.479,
      "step": 85560
    },
    {
      "epoch": 10.297232250300842,
      "grad_norm": 2.111947536468506,
      "learning_rate": 0.00019536136993242106,
      "loss": 7.4875,
      "step": 85570
    },
    {
      "epoch": 10.298435619735258,
      "grad_norm": 3.2716856002807617,
      "learning_rate": 0.0001953602246523699,
      "loss": 7.5409,
      "step": 85580
    },
    {
      "epoch": 10.299638989169676,
      "grad_norm": 3.554607629776001,
      "learning_rate": 0.0001953590792343089,
      "loss": 7.5228,
      "step": 85590
    },
    {
      "epoch": 10.300842358604092,
      "grad_norm": 3.1566975116729736,
      "learning_rate": 0.00019535793367823985,
      "loss": 7.4526,
      "step": 85600
    },
    {
      "epoch": 10.302045728038507,
      "grad_norm": 2.891200542449951,
      "learning_rate": 0.0001953567879841643,
      "loss": 7.4107,
      "step": 85610
    },
    {
      "epoch": 10.303249097472925,
      "grad_norm": 3.0826222896575928,
      "learning_rate": 0.00019535564215208387,
      "loss": 7.4472,
      "step": 85620
    },
    {
      "epoch": 10.30445246690734,
      "grad_norm": 4.875820159912109,
      "learning_rate": 0.0001953544961820003,
      "loss": 7.5037,
      "step": 85630
    },
    {
      "epoch": 10.305655836341757,
      "grad_norm": 1.9762930870056152,
      "learning_rate": 0.0001953533500739153,
      "loss": 7.388,
      "step": 85640
    },
    {
      "epoch": 10.306859205776174,
      "grad_norm": 2.4597346782684326,
      "learning_rate": 0.0001953522038278304,
      "loss": 7.5856,
      "step": 85650
    },
    {
      "epoch": 10.30806257521059,
      "grad_norm": 1.838729739189148,
      "learning_rate": 0.00019535105744374734,
      "loss": 7.545,
      "step": 85660
    },
    {
      "epoch": 10.309265944645006,
      "grad_norm": 1.7955926656723022,
      "learning_rate": 0.00019534991092166778,
      "loss": 7.5446,
      "step": 85670
    },
    {
      "epoch": 10.310469314079423,
      "grad_norm": 2.9095957279205322,
      "learning_rate": 0.00019534876426159333,
      "loss": 7.414,
      "step": 85680
    },
    {
      "epoch": 10.311672683513839,
      "grad_norm": 1.6297860145568848,
      "learning_rate": 0.0001953476174635257,
      "loss": 7.4953,
      "step": 85690
    },
    {
      "epoch": 10.312876052948255,
      "grad_norm": 3.330533504486084,
      "learning_rate": 0.0001953464705274665,
      "loss": 7.4244,
      "step": 85700
    },
    {
      "epoch": 10.314079422382672,
      "grad_norm": 2.861637830734253,
      "learning_rate": 0.00019534532345341743,
      "loss": 7.4212,
      "step": 85710
    },
    {
      "epoch": 10.315282791817088,
      "grad_norm": 2507.746826171875,
      "learning_rate": 0.00019534417624138012,
      "loss": 7.4614,
      "step": 85720
    },
    {
      "epoch": 10.316486161251504,
      "grad_norm": 7.7929301261901855,
      "learning_rate": 0.00019534302889135627,
      "loss": 7.409,
      "step": 85730
    },
    {
      "epoch": 10.31768953068592,
      "grad_norm": 6.382420063018799,
      "learning_rate": 0.0001953418814033475,
      "loss": 7.4432,
      "step": 85740
    },
    {
      "epoch": 10.318892900120337,
      "grad_norm": 19.105998992919922,
      "learning_rate": 0.0001953407337773555,
      "loss": 7.4761,
      "step": 85750
    },
    {
      "epoch": 10.320096269554753,
      "grad_norm": 5.199395179748535,
      "learning_rate": 0.00019533958601338194,
      "loss": 7.3907,
      "step": 85760
    },
    {
      "epoch": 10.321299638989169,
      "grad_norm": 5.394845485687256,
      "learning_rate": 0.00019533843811142843,
      "loss": 7.4173,
      "step": 85770
    },
    {
      "epoch": 10.322503008423586,
      "grad_norm": 6.040703773498535,
      "learning_rate": 0.00019533729007149665,
      "loss": 7.4228,
      "step": 85780
    },
    {
      "epoch": 10.323706377858002,
      "grad_norm": 3.328591823577881,
      "learning_rate": 0.0001953361418935883,
      "loss": 7.4293,
      "step": 85790
    },
    {
      "epoch": 10.324909747292418,
      "grad_norm": 2.576733112335205,
      "learning_rate": 0.00019533499357770503,
      "loss": 7.3564,
      "step": 85800
    },
    {
      "epoch": 10.326113116726836,
      "grad_norm": 85.13755798339844,
      "learning_rate": 0.00019533384512384843,
      "loss": 7.4474,
      "step": 85810
    },
    {
      "epoch": 10.327316486161251,
      "grad_norm": 8.398889541625977,
      "learning_rate": 0.00019533269653202026,
      "loss": 7.4373,
      "step": 85820
    },
    {
      "epoch": 10.328519855595667,
      "grad_norm": 4.141637802124023,
      "learning_rate": 0.00019533154780222213,
      "loss": 7.4746,
      "step": 85830
    },
    {
      "epoch": 10.329723225030085,
      "grad_norm": 5.044012546539307,
      "learning_rate": 0.00019533039893445572,
      "loss": 7.4292,
      "step": 85840
    },
    {
      "epoch": 10.3309265944645,
      "grad_norm": 5.562819480895996,
      "learning_rate": 0.0001953292499287227,
      "loss": 7.4262,
      "step": 85850
    },
    {
      "epoch": 10.332129963898916,
      "grad_norm": 3.3015358448028564,
      "learning_rate": 0.0001953281007850247,
      "loss": 7.3704,
      "step": 85860
    },
    {
      "epoch": 10.333333333333334,
      "grad_norm": 1.2157748937606812,
      "learning_rate": 0.0001953269515033634,
      "loss": 7.4286,
      "step": 85870
    },
    {
      "epoch": 10.33453670276775,
      "grad_norm": 31.767892837524414,
      "learning_rate": 0.00019532580208374048,
      "loss": 7.3897,
      "step": 85880
    },
    {
      "epoch": 10.335740072202166,
      "grad_norm": 8.570412635803223,
      "learning_rate": 0.00019532465252615757,
      "loss": 7.4112,
      "step": 85890
    },
    {
      "epoch": 10.336943441636583,
      "grad_norm": 6.177369594573975,
      "learning_rate": 0.00019532350283061638,
      "loss": 7.4998,
      "step": 85900
    },
    {
      "epoch": 10.338146811070999,
      "grad_norm": 1302.7275390625,
      "learning_rate": 0.0001953223529971185,
      "loss": 7.2976,
      "step": 85910
    },
    {
      "epoch": 10.339350180505415,
      "grad_norm": 469.8651123046875,
      "learning_rate": 0.0001953212030256657,
      "loss": 7.4372,
      "step": 85920
    },
    {
      "epoch": 10.340553549939832,
      "grad_norm": 4.476493835449219,
      "learning_rate": 0.00019532005291625956,
      "loss": 7.448,
      "step": 85930
    },
    {
      "epoch": 10.341756919374248,
      "grad_norm": 2.6381514072418213,
      "learning_rate": 0.00019531890266890177,
      "loss": 7.533,
      "step": 85940
    },
    {
      "epoch": 10.342960288808664,
      "grad_norm": 8.27548885345459,
      "learning_rate": 0.00019531775228359398,
      "loss": 7.381,
      "step": 85950
    },
    {
      "epoch": 10.344163658243081,
      "grad_norm": 6.770925998687744,
      "learning_rate": 0.0001953166017603379,
      "loss": 7.4246,
      "step": 85960
    },
    {
      "epoch": 10.345367027677497,
      "grad_norm": 6.874700546264648,
      "learning_rate": 0.00019531545109913512,
      "loss": 7.4873,
      "step": 85970
    },
    {
      "epoch": 10.346570397111913,
      "grad_norm": 2.744999885559082,
      "learning_rate": 0.00019531430029998739,
      "loss": 7.3974,
      "step": 85980
    },
    {
      "epoch": 10.34777376654633,
      "grad_norm": 29.037128448486328,
      "learning_rate": 0.0001953131493628963,
      "loss": 7.3437,
      "step": 85990
    },
    {
      "epoch": 10.348977135980746,
      "grad_norm": 22.944522857666016,
      "learning_rate": 0.0001953119982878636,
      "loss": 7.3648,
      "step": 86000
    },
    {
      "epoch": 10.350180505415162,
      "grad_norm": 4.095037937164307,
      "learning_rate": 0.00019531084707489086,
      "loss": 7.3007,
      "step": 86010
    },
    {
      "epoch": 10.35138387484958,
      "grad_norm": 5.845776081085205,
      "learning_rate": 0.00019530969572397982,
      "loss": 7.3914,
      "step": 86020
    },
    {
      "epoch": 10.352587244283995,
      "grad_norm": 206.785888671875,
      "learning_rate": 0.0001953085442351321,
      "loss": 7.3753,
      "step": 86030
    },
    {
      "epoch": 10.353790613718411,
      "grad_norm": 2.7486515045166016,
      "learning_rate": 0.00019530739260834937,
      "loss": 7.4032,
      "step": 86040
    },
    {
      "epoch": 10.354993983152827,
      "grad_norm": 5.12001895904541,
      "learning_rate": 0.00019530624084363336,
      "loss": 7.4674,
      "step": 86050
    },
    {
      "epoch": 10.356197352587245,
      "grad_norm": 4.1538262367248535,
      "learning_rate": 0.00019530508894098568,
      "loss": 7.3124,
      "step": 86060
    },
    {
      "epoch": 10.35740072202166,
      "grad_norm": 26.622573852539062,
      "learning_rate": 0.00019530393690040797,
      "loss": 7.3205,
      "step": 86070
    },
    {
      "epoch": 10.358604091456076,
      "grad_norm": 24.624265670776367,
      "learning_rate": 0.00019530278472190194,
      "loss": 7.356,
      "step": 86080
    },
    {
      "epoch": 10.359807460890494,
      "grad_norm": 1126.248046875,
      "learning_rate": 0.0001953016324054693,
      "loss": 7.3292,
      "step": 86090
    },
    {
      "epoch": 10.36101083032491,
      "grad_norm": 12.730032920837402,
      "learning_rate": 0.00019530047995111163,
      "loss": 7.3581,
      "step": 86100
    },
    {
      "epoch": 10.362214199759325,
      "grad_norm": 2211.71435546875,
      "learning_rate": 0.00019529932735883065,
      "loss": 7.4205,
      "step": 86110
    },
    {
      "epoch": 10.363417569193743,
      "grad_norm": 1.5781265497207642,
      "learning_rate": 0.000195298174628628,
      "loss": 7.4111,
      "step": 86120
    },
    {
      "epoch": 10.364620938628159,
      "grad_norm": 3.420382261276245,
      "learning_rate": 0.00019529702176050538,
      "loss": 7.3086,
      "step": 86130
    },
    {
      "epoch": 10.365824308062574,
      "grad_norm": 4.60121488571167,
      "learning_rate": 0.00019529586875446442,
      "loss": 7.298,
      "step": 86140
    },
    {
      "epoch": 10.367027677496992,
      "grad_norm": 6.83242130279541,
      "learning_rate": 0.0001952947156105068,
      "loss": 7.3454,
      "step": 86150
    },
    {
      "epoch": 10.368231046931408,
      "grad_norm": 50.682098388671875,
      "learning_rate": 0.00019529356232863423,
      "loss": 7.39,
      "step": 86160
    },
    {
      "epoch": 10.369434416365824,
      "grad_norm": 2.5132851600646973,
      "learning_rate": 0.00019529240890884834,
      "loss": 7.4204,
      "step": 86170
    },
    {
      "epoch": 10.370637785800241,
      "grad_norm": 6.81795072555542,
      "learning_rate": 0.00019529125535115084,
      "loss": 7.3806,
      "step": 86180
    },
    {
      "epoch": 10.371841155234657,
      "grad_norm": 2.5079987049102783,
      "learning_rate": 0.00019529010165554329,
      "loss": 7.2356,
      "step": 86190
    },
    {
      "epoch": 10.373044524669073,
      "grad_norm": 75.7482681274414,
      "learning_rate": 0.0001952889478220275,
      "loss": 7.3059,
      "step": 86200
    },
    {
      "epoch": 10.37424789410349,
      "grad_norm": 6.17049503326416,
      "learning_rate": 0.00019528779385060506,
      "loss": 7.3547,
      "step": 86210
    },
    {
      "epoch": 10.375451263537906,
      "grad_norm": 110.1769027709961,
      "learning_rate": 0.00019528663974127764,
      "loss": 7.204,
      "step": 86220
    },
    {
      "epoch": 10.376654632972322,
      "grad_norm": 3.8423500061035156,
      "learning_rate": 0.00019528548549404694,
      "loss": 7.3803,
      "step": 86230
    },
    {
      "epoch": 10.37785800240674,
      "grad_norm": 680.0330810546875,
      "learning_rate": 0.00019528433110891461,
      "loss": 7.363,
      "step": 86240
    },
    {
      "epoch": 10.379061371841155,
      "grad_norm": 2.896757125854492,
      "learning_rate": 0.00019528317658588234,
      "loss": 7.2612,
      "step": 86250
    },
    {
      "epoch": 10.380264741275571,
      "grad_norm": 2.5943007469177246,
      "learning_rate": 0.00019528202192495178,
      "loss": 7.3528,
      "step": 86260
    },
    {
      "epoch": 10.381468110709989,
      "grad_norm": 10.391607284545898,
      "learning_rate": 0.0001952808671261246,
      "loss": 7.3012,
      "step": 86270
    },
    {
      "epoch": 10.382671480144404,
      "grad_norm": 4.298302173614502,
      "learning_rate": 0.0001952797121894025,
      "loss": 7.3386,
      "step": 86280
    },
    {
      "epoch": 10.38387484957882,
      "grad_norm": 22.879108428955078,
      "learning_rate": 0.00019527855711478712,
      "loss": 7.3371,
      "step": 86290
    },
    {
      "epoch": 10.385078219013238,
      "grad_norm": 4.031415939331055,
      "learning_rate": 0.00019527740190228017,
      "loss": 7.3839,
      "step": 86300
    },
    {
      "epoch": 10.386281588447654,
      "grad_norm": 3.8353116512298584,
      "learning_rate": 0.00019527624655188327,
      "loss": 7.3102,
      "step": 86310
    },
    {
      "epoch": 10.38748495788207,
      "grad_norm": 28.881017684936523,
      "learning_rate": 0.00019527509106359814,
      "loss": 7.2641,
      "step": 86320
    },
    {
      "epoch": 10.388688327316487,
      "grad_norm": 3.1833651065826416,
      "learning_rate": 0.0001952739354374264,
      "loss": 7.3326,
      "step": 86330
    },
    {
      "epoch": 10.389891696750903,
      "grad_norm": 4.188170909881592,
      "learning_rate": 0.00019527277967336975,
      "loss": 7.334,
      "step": 86340
    },
    {
      "epoch": 10.391095066185319,
      "grad_norm": 101.34941864013672,
      "learning_rate": 0.00019527162377142992,
      "loss": 7.2211,
      "step": 86350
    },
    {
      "epoch": 10.392298435619736,
      "grad_norm": 89.14458465576172,
      "learning_rate": 0.00019527046773160848,
      "loss": 7.3642,
      "step": 86360
    },
    {
      "epoch": 10.393501805054152,
      "grad_norm": 11.451932907104492,
      "learning_rate": 0.0001952693115539072,
      "loss": 7.3821,
      "step": 86370
    },
    {
      "epoch": 10.394705174488568,
      "grad_norm": 102.76563262939453,
      "learning_rate": 0.00019526815523832763,
      "loss": 7.391,
      "step": 86380
    },
    {
      "epoch": 10.395908543922985,
      "grad_norm": 9.27910327911377,
      "learning_rate": 0.00019526699878487155,
      "loss": 7.3022,
      "step": 86390
    },
    {
      "epoch": 10.397111913357401,
      "grad_norm": 12.482718467712402,
      "learning_rate": 0.00019526584219354062,
      "loss": 7.2903,
      "step": 86400
    },
    {
      "epoch": 10.398315282791817,
      "grad_norm": 5.972568511962891,
      "learning_rate": 0.0001952646854643365,
      "loss": 7.3264,
      "step": 86410
    },
    {
      "epoch": 10.399518652226233,
      "grad_norm": 6.8508076667785645,
      "learning_rate": 0.00019526352859726082,
      "loss": 7.3016,
      "step": 86420
    },
    {
      "epoch": 10.40072202166065,
      "grad_norm": 2.6398284435272217,
      "learning_rate": 0.00019526237159231532,
      "loss": 7.3302,
      "step": 86430
    },
    {
      "epoch": 10.401925391095066,
      "grad_norm": 24.66536521911621,
      "learning_rate": 0.00019526121444950168,
      "loss": 7.31,
      "step": 86440
    },
    {
      "epoch": 10.403128760529482,
      "grad_norm": 9.506525993347168,
      "learning_rate": 0.00019526005716882149,
      "loss": 7.2653,
      "step": 86450
    },
    {
      "epoch": 10.4043321299639,
      "grad_norm": 91.26265716552734,
      "learning_rate": 0.0001952588997502765,
      "loss": 7.3002,
      "step": 86460
    },
    {
      "epoch": 10.405535499398315,
      "grad_norm": 11.353433609008789,
      "learning_rate": 0.00019525774219386838,
      "loss": 7.4245,
      "step": 86470
    },
    {
      "epoch": 10.406738868832731,
      "grad_norm": 12.2548828125,
      "learning_rate": 0.00019525658449959878,
      "loss": 7.2518,
      "step": 86480
    },
    {
      "epoch": 10.407942238267148,
      "grad_norm": 13.156693458557129,
      "learning_rate": 0.00019525542666746937,
      "loss": 7.2892,
      "step": 86490
    },
    {
      "epoch": 10.409145607701564,
      "grad_norm": 127.03163146972656,
      "learning_rate": 0.00019525426869748184,
      "loss": 7.1979,
      "step": 86500
    },
    {
      "epoch": 10.41034897713598,
      "grad_norm": 387.25146484375,
      "learning_rate": 0.0001952531105896379,
      "loss": 7.3386,
      "step": 86510
    },
    {
      "epoch": 10.411552346570398,
      "grad_norm": 134.3849639892578,
      "learning_rate": 0.00019525195234393916,
      "loss": 7.4107,
      "step": 86520
    },
    {
      "epoch": 10.412755716004813,
      "grad_norm": 57.4593505859375,
      "learning_rate": 0.00019525079396038732,
      "loss": 7.3787,
      "step": 86530
    },
    {
      "epoch": 10.41395908543923,
      "grad_norm": 528.2086791992188,
      "learning_rate": 0.00019524963543898408,
      "loss": 7.3134,
      "step": 86540
    },
    {
      "epoch": 10.415162454873647,
      "grad_norm": 8.248697280883789,
      "learning_rate": 0.0001952484767797311,
      "loss": 7.4736,
      "step": 86550
    },
    {
      "epoch": 10.416365824308063,
      "grad_norm": 26.072460174560547,
      "learning_rate": 0.00019524731798263006,
      "loss": 7.3903,
      "step": 86560
    },
    {
      "epoch": 10.417569193742478,
      "grad_norm": 339.4084777832031,
      "learning_rate": 0.00019524615904768262,
      "loss": 7.3795,
      "step": 86570
    },
    {
      "epoch": 10.418772563176896,
      "grad_norm": 2.1741859912872314,
      "learning_rate": 0.00019524499997489048,
      "loss": 7.3697,
      "step": 86580
    },
    {
      "epoch": 10.419975932611312,
      "grad_norm": 14.492084503173828,
      "learning_rate": 0.00019524384076425534,
      "loss": 7.267,
      "step": 86590
    },
    {
      "epoch": 10.421179302045728,
      "grad_norm": 9.385497093200684,
      "learning_rate": 0.0001952426814157788,
      "loss": 7.3738,
      "step": 86600
    },
    {
      "epoch": 10.422382671480145,
      "grad_norm": 138.29335021972656,
      "learning_rate": 0.00019524152192946262,
      "loss": 7.3135,
      "step": 86610
    },
    {
      "epoch": 10.42358604091456,
      "grad_norm": 11.764824867248535,
      "learning_rate": 0.00019524036230530843,
      "loss": 7.4418,
      "step": 86620
    },
    {
      "epoch": 10.424789410348977,
      "grad_norm": 203.98117065429688,
      "learning_rate": 0.0001952392025433179,
      "loss": 7.388,
      "step": 86630
    },
    {
      "epoch": 10.425992779783394,
      "grad_norm": 40.192779541015625,
      "learning_rate": 0.00019523804264349277,
      "loss": 7.3609,
      "step": 86640
    },
    {
      "epoch": 10.42719614921781,
      "grad_norm": 426.7164306640625,
      "learning_rate": 0.00019523688260583466,
      "loss": 7.4305,
      "step": 86650
    },
    {
      "epoch": 10.428399518652226,
      "grad_norm": 1.9205007553100586,
      "learning_rate": 0.00019523572243034524,
      "loss": 7.3843,
      "step": 86660
    },
    {
      "epoch": 10.429602888086643,
      "grad_norm": 39.38263702392578,
      "learning_rate": 0.00019523456211702626,
      "loss": 7.3838,
      "step": 86670
    },
    {
      "epoch": 10.43080625752106,
      "grad_norm": 16.812620162963867,
      "learning_rate": 0.00019523340166587936,
      "loss": 7.3231,
      "step": 86680
    },
    {
      "epoch": 10.432009626955475,
      "grad_norm": 6.904186248779297,
      "learning_rate": 0.00019523224107690618,
      "loss": 7.3075,
      "step": 86690
    },
    {
      "epoch": 10.433212996389893,
      "grad_norm": 41.70980453491211,
      "learning_rate": 0.00019523108035010846,
      "loss": 7.4046,
      "step": 86700
    },
    {
      "epoch": 10.434416365824308,
      "grad_norm": 685.0161743164062,
      "learning_rate": 0.00019522991948548785,
      "loss": 7.2371,
      "step": 86710
    },
    {
      "epoch": 10.435619735258724,
      "grad_norm": 13.445799827575684,
      "learning_rate": 0.00019522875848304602,
      "loss": 7.2452,
      "step": 86720
    },
    {
      "epoch": 10.43682310469314,
      "grad_norm": 168.27622985839844,
      "learning_rate": 0.00019522759734278466,
      "loss": 7.2095,
      "step": 86730
    },
    {
      "epoch": 10.438026474127557,
      "grad_norm": 24.310766220092773,
      "learning_rate": 0.0001952264360647055,
      "loss": 7.2637,
      "step": 86740
    },
    {
      "epoch": 10.439229843561973,
      "grad_norm": 31.537172317504883,
      "learning_rate": 0.00019522527464881014,
      "loss": 7.3222,
      "step": 86750
    },
    {
      "epoch": 10.440433212996389,
      "grad_norm": 9.848044395446777,
      "learning_rate": 0.0001952241130951003,
      "loss": 7.3228,
      "step": 86760
    },
    {
      "epoch": 10.441636582430807,
      "grad_norm": 2950.752197265625,
      "learning_rate": 0.00019522295140357766,
      "loss": 7.2449,
      "step": 86770
    },
    {
      "epoch": 10.442839951865222,
      "grad_norm": 8.810911178588867,
      "learning_rate": 0.0001952217895742439,
      "loss": 7.3124,
      "step": 86780
    },
    {
      "epoch": 10.444043321299638,
      "grad_norm": 169.3764190673828,
      "learning_rate": 0.00019522062760710072,
      "loss": 7.2767,
      "step": 86790
    },
    {
      "epoch": 10.445246690734056,
      "grad_norm": 8.374594688415527,
      "learning_rate": 0.00019521946550214977,
      "loss": 7.3505,
      "step": 86800
    },
    {
      "epoch": 10.446450060168472,
      "grad_norm": 29.65797233581543,
      "learning_rate": 0.00019521830325939274,
      "loss": 7.3024,
      "step": 86810
    },
    {
      "epoch": 10.447653429602887,
      "grad_norm": 2913.634765625,
      "learning_rate": 0.0001952171408788313,
      "loss": 7.311,
      "step": 86820
    },
    {
      "epoch": 10.448856799037305,
      "grad_norm": 94.82404327392578,
      "learning_rate": 0.00019521597836046718,
      "loss": 7.2979,
      "step": 86830
    },
    {
      "epoch": 10.45006016847172,
      "grad_norm": 16.484973907470703,
      "learning_rate": 0.00019521481570430201,
      "loss": 7.3126,
      "step": 86840
    },
    {
      "epoch": 10.451263537906136,
      "grad_norm": 9.324202537536621,
      "learning_rate": 0.00019521365291033752,
      "loss": 7.361,
      "step": 86850
    },
    {
      "epoch": 10.452466907340554,
      "grad_norm": 236.1169891357422,
      "learning_rate": 0.00019521248997857535,
      "loss": 7.2888,
      "step": 86860
    },
    {
      "epoch": 10.45367027677497,
      "grad_norm": 42.557838439941406,
      "learning_rate": 0.00019521132690901718,
      "loss": 7.2597,
      "step": 86870
    },
    {
      "epoch": 10.454873646209386,
      "grad_norm": 63.07026672363281,
      "learning_rate": 0.00019521016370166474,
      "loss": 7.385,
      "step": 86880
    },
    {
      "epoch": 10.456077015643803,
      "grad_norm": 9.713822364807129,
      "learning_rate": 0.00019520900035651967,
      "loss": 7.399,
      "step": 86890
    },
    {
      "epoch": 10.457280385078219,
      "grad_norm": 24.991436004638672,
      "learning_rate": 0.00019520783687358367,
      "loss": 7.458,
      "step": 86900
    },
    {
      "epoch": 10.458483754512635,
      "grad_norm": 71.0694580078125,
      "learning_rate": 0.00019520667325285845,
      "loss": 7.4802,
      "step": 86910
    },
    {
      "epoch": 10.459687123947052,
      "grad_norm": 257.9768371582031,
      "learning_rate": 0.00019520550949434564,
      "loss": 7.5513,
      "step": 86920
    },
    {
      "epoch": 10.460890493381468,
      "grad_norm": 10389.453125,
      "learning_rate": 0.00019520434559804696,
      "loss": 7.5223,
      "step": 86930
    },
    {
      "epoch": 10.462093862815884,
      "grad_norm": 16.677762985229492,
      "learning_rate": 0.0001952031815639641,
      "loss": 7.5917,
      "step": 86940
    },
    {
      "epoch": 10.463297232250302,
      "grad_norm": 26.780977249145508,
      "learning_rate": 0.00019520201739209873,
      "loss": 7.5718,
      "step": 86950
    },
    {
      "epoch": 10.464500601684717,
      "grad_norm": 261.8631896972656,
      "learning_rate": 0.0001952008530824525,
      "loss": 7.4584,
      "step": 86960
    },
    {
      "epoch": 10.465703971119133,
      "grad_norm": 106.22891235351562,
      "learning_rate": 0.0001951996886350272,
      "loss": 7.5674,
      "step": 86970
    },
    {
      "epoch": 10.46690734055355,
      "grad_norm": 335.2868347167969,
      "learning_rate": 0.00019519852404982437,
      "loss": 7.4891,
      "step": 86980
    },
    {
      "epoch": 10.468110709987966,
      "grad_norm": 185.00413513183594,
      "learning_rate": 0.00019519735932684582,
      "loss": 7.4306,
      "step": 86990
    },
    {
      "epoch": 10.469314079422382,
      "grad_norm": 385.6830749511719,
      "learning_rate": 0.00019519619446609315,
      "loss": 7.5965,
      "step": 87000
    },
    {
      "epoch": 10.4705174488568,
      "grad_norm": 3.527390480041504,
      "learning_rate": 0.00019519502946756815,
      "loss": 7.5177,
      "step": 87010
    },
    {
      "epoch": 10.471720818291216,
      "grad_norm": 157.47364807128906,
      "learning_rate": 0.00019519386433127238,
      "loss": 7.5277,
      "step": 87020
    },
    {
      "epoch": 10.472924187725631,
      "grad_norm": 158.25482177734375,
      "learning_rate": 0.0001951926990572076,
      "loss": 7.4535,
      "step": 87030
    },
    {
      "epoch": 10.474127557160049,
      "grad_norm": 120.78103637695312,
      "learning_rate": 0.00019519153364537547,
      "loss": 7.5546,
      "step": 87040
    },
    {
      "epoch": 10.475330926594465,
      "grad_norm": 13.872295379638672,
      "learning_rate": 0.00019519036809577771,
      "loss": 7.5583,
      "step": 87050
    },
    {
      "epoch": 10.47653429602888,
      "grad_norm": 187.38462829589844,
      "learning_rate": 0.000195189202408416,
      "loss": 7.4062,
      "step": 87060
    },
    {
      "epoch": 10.477737665463298,
      "grad_norm": 152.8062286376953,
      "learning_rate": 0.00019518803658329198,
      "loss": 7.405,
      "step": 87070
    },
    {
      "epoch": 10.478941034897714,
      "grad_norm": 9.428537368774414,
      "learning_rate": 0.0001951868706204074,
      "loss": 7.4634,
      "step": 87080
    },
    {
      "epoch": 10.48014440433213,
      "grad_norm": 55.365570068359375,
      "learning_rate": 0.0001951857045197639,
      "loss": 7.4388,
      "step": 87090
    },
    {
      "epoch": 10.481347773766545,
      "grad_norm": 57.88100814819336,
      "learning_rate": 0.00019518453828136316,
      "loss": 7.5701,
      "step": 87100
    },
    {
      "epoch": 10.482551143200963,
      "grad_norm": 34.78044891357422,
      "learning_rate": 0.0001951833719052069,
      "loss": 7.4406,
      "step": 87110
    },
    {
      "epoch": 10.483754512635379,
      "grad_norm": 40.717681884765625,
      "learning_rate": 0.00019518220539129684,
      "loss": 7.4336,
      "step": 87120
    },
    {
      "epoch": 10.484957882069795,
      "grad_norm": 694.4647827148438,
      "learning_rate": 0.00019518103873963463,
      "loss": 7.4763,
      "step": 87130
    },
    {
      "epoch": 10.486161251504212,
      "grad_norm": 79.98706817626953,
      "learning_rate": 0.00019517987195022193,
      "loss": 7.4129,
      "step": 87140
    },
    {
      "epoch": 10.487364620938628,
      "grad_norm": 8.193269729614258,
      "learning_rate": 0.00019517870502306045,
      "loss": 7.4447,
      "step": 87150
    },
    {
      "epoch": 10.488567990373044,
      "grad_norm": 51494.640625,
      "learning_rate": 0.0001951775379581519,
      "loss": 7.5182,
      "step": 87160
    },
    {
      "epoch": 10.489771359807461,
      "grad_norm": 180.87831115722656,
      "learning_rate": 0.00019517637075549795,
      "loss": 7.4129,
      "step": 87170
    },
    {
      "epoch": 10.490974729241877,
      "grad_norm": 63.636260986328125,
      "learning_rate": 0.0001951752034151003,
      "loss": 7.5,
      "step": 87180
    },
    {
      "epoch": 10.492178098676293,
      "grad_norm": 14.191164016723633,
      "learning_rate": 0.00019517403593696063,
      "loss": 7.5157,
      "step": 87190
    },
    {
      "epoch": 10.49338146811071,
      "grad_norm": 28.7744140625,
      "learning_rate": 0.00019517286832108061,
      "loss": 7.445,
      "step": 87200
    },
    {
      "epoch": 10.494584837545126,
      "grad_norm": 37.393409729003906,
      "learning_rate": 0.00019517170056746196,
      "loss": 7.5317,
      "step": 87210
    },
    {
      "epoch": 10.495788206979542,
      "grad_norm": 6.052454471588135,
      "learning_rate": 0.0001951705326761064,
      "loss": 7.497,
      "step": 87220
    },
    {
      "epoch": 10.49699157641396,
      "grad_norm": 41.5262565612793,
      "learning_rate": 0.00019516936464701555,
      "loss": 7.4961,
      "step": 87230
    },
    {
      "epoch": 10.498194945848375,
      "grad_norm": 15.388359069824219,
      "learning_rate": 0.00019516819648019112,
      "loss": 7.4264,
      "step": 87240
    },
    {
      "epoch": 10.499398315282791,
      "grad_norm": 191.34458923339844,
      "learning_rate": 0.00019516702817563484,
      "loss": 7.4435,
      "step": 87250
    },
    {
      "epoch": 10.500601684717209,
      "grad_norm": 18.088212966918945,
      "learning_rate": 0.0001951658597333484,
      "loss": 7.5107,
      "step": 87260
    },
    {
      "epoch": 10.501805054151625,
      "grad_norm": 787.8485717773438,
      "learning_rate": 0.0001951646911533334,
      "loss": 7.5042,
      "step": 87270
    },
    {
      "epoch": 10.50300842358604,
      "grad_norm": 35.14695358276367,
      "learning_rate": 0.00019516352243559163,
      "loss": 7.5672,
      "step": 87280
    },
    {
      "epoch": 10.504211793020458,
      "grad_norm": 1208.171875,
      "learning_rate": 0.00019516235358012473,
      "loss": 7.4827,
      "step": 87290
    },
    {
      "epoch": 10.505415162454874,
      "grad_norm": 93.63135528564453,
      "learning_rate": 0.00019516118458693443,
      "loss": 7.5129,
      "step": 87300
    },
    {
      "epoch": 10.50661853188929,
      "grad_norm": 111.2813491821289,
      "learning_rate": 0.0001951600154560224,
      "loss": 7.5537,
      "step": 87310
    },
    {
      "epoch": 10.507821901323707,
      "grad_norm": 30.587112426757812,
      "learning_rate": 0.00019515884618739033,
      "loss": 7.5448,
      "step": 87320
    },
    {
      "epoch": 10.509025270758123,
      "grad_norm": 18.9459285736084,
      "learning_rate": 0.0001951576767810399,
      "loss": 7.4557,
      "step": 87330
    },
    {
      "epoch": 10.510228640192539,
      "grad_norm": 148.022216796875,
      "learning_rate": 0.00019515650723697284,
      "loss": 7.5728,
      "step": 87340
    },
    {
      "epoch": 10.511432009626956,
      "grad_norm": 8.671709060668945,
      "learning_rate": 0.00019515533755519083,
      "loss": 7.5929,
      "step": 87350
    },
    {
      "epoch": 10.512635379061372,
      "grad_norm": 99.92613220214844,
      "learning_rate": 0.0001951541677356955,
      "loss": 7.458,
      "step": 87360
    },
    {
      "epoch": 10.513838748495788,
      "grad_norm": 51.72817611694336,
      "learning_rate": 0.00019515299777848866,
      "loss": 7.5404,
      "step": 87370
    },
    {
      "epoch": 10.515042117930205,
      "grad_norm": 20.989768981933594,
      "learning_rate": 0.0001951518276835719,
      "loss": 7.4775,
      "step": 87380
    },
    {
      "epoch": 10.516245487364621,
      "grad_norm": 384.83258056640625,
      "learning_rate": 0.00019515065745094696,
      "loss": 7.4001,
      "step": 87390
    },
    {
      "epoch": 10.517448856799037,
      "grad_norm": 25.202699661254883,
      "learning_rate": 0.00019514948708061554,
      "loss": 7.4639,
      "step": 87400
    },
    {
      "epoch": 10.518652226233453,
      "grad_norm": 14.15613842010498,
      "learning_rate": 0.00019514831657257932,
      "loss": 7.4721,
      "step": 87410
    },
    {
      "epoch": 10.51985559566787,
      "grad_norm": 8.060870170593262,
      "learning_rate": 0.00019514714592684,
      "loss": 7.5325,
      "step": 87420
    },
    {
      "epoch": 10.521058965102286,
      "grad_norm": 41.303680419921875,
      "learning_rate": 0.00019514597514339925,
      "loss": 7.5139,
      "step": 87430
    },
    {
      "epoch": 10.522262334536702,
      "grad_norm": 2.9415969848632812,
      "learning_rate": 0.0001951448042222588,
      "loss": 7.5952,
      "step": 87440
    },
    {
      "epoch": 10.52346570397112,
      "grad_norm": 6.626253604888916,
      "learning_rate": 0.0001951436331634203,
      "loss": 7.5419,
      "step": 87450
    },
    {
      "epoch": 10.524669073405535,
      "grad_norm": 7.343200206756592,
      "learning_rate": 0.0001951424619668855,
      "loss": 7.4667,
      "step": 87460
    },
    {
      "epoch": 10.525872442839951,
      "grad_norm": 11.2632417678833,
      "learning_rate": 0.00019514129063265604,
      "loss": 7.4708,
      "step": 87470
    },
    {
      "epoch": 10.527075812274369,
      "grad_norm": 5.937346935272217,
      "learning_rate": 0.00019514011916073367,
      "loss": 7.5314,
      "step": 87480
    },
    {
      "epoch": 10.528279181708784,
      "grad_norm": 9.836857795715332,
      "learning_rate": 0.00019513894755112006,
      "loss": 7.5127,
      "step": 87490
    },
    {
      "epoch": 10.5294825511432,
      "grad_norm": 4.930632591247559,
      "learning_rate": 0.00019513777580381687,
      "loss": 7.5156,
      "step": 87500
    },
    {
      "epoch": 10.530685920577618,
      "grad_norm": 27.229860305786133,
      "learning_rate": 0.00019513660391882586,
      "loss": 7.6346,
      "step": 87510
    },
    {
      "epoch": 10.531889290012034,
      "grad_norm": 7.525158882141113,
      "learning_rate": 0.00019513543189614868,
      "loss": 7.5316,
      "step": 87520
    },
    {
      "epoch": 10.53309265944645,
      "grad_norm": 6.16154146194458,
      "learning_rate": 0.00019513425973578707,
      "loss": 7.4763,
      "step": 87530
    },
    {
      "epoch": 10.534296028880867,
      "grad_norm": 3.742701768875122,
      "learning_rate": 0.00019513308743774265,
      "loss": 7.5409,
      "step": 87540
    },
    {
      "epoch": 10.535499398315283,
      "grad_norm": 7.226905822753906,
      "learning_rate": 0.00019513191500201721,
      "loss": 7.5202,
      "step": 87550
    },
    {
      "epoch": 10.536702767749698,
      "grad_norm": 9.411243438720703,
      "learning_rate": 0.00019513074242861241,
      "loss": 7.4987,
      "step": 87560
    },
    {
      "epoch": 10.537906137184116,
      "grad_norm": 6.339433670043945,
      "learning_rate": 0.0001951295697175299,
      "loss": 7.5569,
      "step": 87570
    },
    {
      "epoch": 10.539109506618532,
      "grad_norm": 3.9573373794555664,
      "learning_rate": 0.0001951283968687714,
      "loss": 7.4227,
      "step": 87580
    },
    {
      "epoch": 10.540312876052948,
      "grad_norm": 6.60520601272583,
      "learning_rate": 0.00019512722388233868,
      "loss": 7.2681,
      "step": 87590
    },
    {
      "epoch": 10.541516245487365,
      "grad_norm": 9.820063591003418,
      "learning_rate": 0.00019512605075823337,
      "loss": 7.411,
      "step": 87600
    },
    {
      "epoch": 10.542719614921781,
      "grad_norm": 18.547088623046875,
      "learning_rate": 0.00019512487749645714,
      "loss": 7.3299,
      "step": 87610
    },
    {
      "epoch": 10.543922984356197,
      "grad_norm": 3.7740328311920166,
      "learning_rate": 0.00019512370409701174,
      "loss": 7.2751,
      "step": 87620
    },
    {
      "epoch": 10.545126353790614,
      "grad_norm": 39.1048698425293,
      "learning_rate": 0.00019512253055989888,
      "loss": 7.367,
      "step": 87630
    },
    {
      "epoch": 10.54632972322503,
      "grad_norm": 12.924285888671875,
      "learning_rate": 0.0001951213568851202,
      "loss": 7.3617,
      "step": 87640
    },
    {
      "epoch": 10.547533092659446,
      "grad_norm": 11.680988311767578,
      "learning_rate": 0.00019512018307267748,
      "loss": 7.3535,
      "step": 87650
    },
    {
      "epoch": 10.548736462093864,
      "grad_norm": 5.430835247039795,
      "learning_rate": 0.00019511900912257233,
      "loss": 7.3529,
      "step": 87660
    },
    {
      "epoch": 10.54993983152828,
      "grad_norm": 5.599913120269775,
      "learning_rate": 0.0001951178350348065,
      "loss": 7.387,
      "step": 87670
    },
    {
      "epoch": 10.551143200962695,
      "grad_norm": 9.078004837036133,
      "learning_rate": 0.00019511666080938167,
      "loss": 7.2856,
      "step": 87680
    },
    {
      "epoch": 10.552346570397113,
      "grad_norm": 7.466926097869873,
      "learning_rate": 0.00019511548644629955,
      "loss": 7.355,
      "step": 87690
    },
    {
      "epoch": 10.553549939831528,
      "grad_norm": 29.83484649658203,
      "learning_rate": 0.00019511431194556183,
      "loss": 7.4152,
      "step": 87700
    },
    {
      "epoch": 10.554753309265944,
      "grad_norm": 7.700797080993652,
      "learning_rate": 0.00019511313730717026,
      "loss": 7.308,
      "step": 87710
    },
    {
      "epoch": 10.555956678700362,
      "grad_norm": 10.354599952697754,
      "learning_rate": 0.00019511196253112645,
      "loss": 7.3433,
      "step": 87720
    },
    {
      "epoch": 10.557160048134778,
      "grad_norm": 13.304309844970703,
      "learning_rate": 0.00019511078761743218,
      "loss": 7.3175,
      "step": 87730
    },
    {
      "epoch": 10.558363417569193,
      "grad_norm": 11.474148750305176,
      "learning_rate": 0.0001951096125660891,
      "loss": 7.4401,
      "step": 87740
    },
    {
      "epoch": 10.559566787003611,
      "grad_norm": 33.384464263916016,
      "learning_rate": 0.00019510843737709894,
      "loss": 7.385,
      "step": 87750
    },
    {
      "epoch": 10.560770156438027,
      "grad_norm": 22.092769622802734,
      "learning_rate": 0.0001951072620504634,
      "loss": 7.3322,
      "step": 87760
    },
    {
      "epoch": 10.561973525872443,
      "grad_norm": 8.763447761535645,
      "learning_rate": 0.00019510608658618414,
      "loss": 7.3348,
      "step": 87770
    },
    {
      "epoch": 10.56317689530686,
      "grad_norm": 22.685483932495117,
      "learning_rate": 0.0001951049109842629,
      "loss": 7.3653,
      "step": 87780
    },
    {
      "epoch": 10.564380264741276,
      "grad_norm": 45.91816329956055,
      "learning_rate": 0.00019510373524470138,
      "loss": 7.3695,
      "step": 87790
    },
    {
      "epoch": 10.565583634175692,
      "grad_norm": 30.154470443725586,
      "learning_rate": 0.00019510255936750127,
      "loss": 7.3982,
      "step": 87800
    },
    {
      "epoch": 10.566787003610107,
      "grad_norm": 16.189109802246094,
      "learning_rate": 0.0001951013833526643,
      "loss": 7.3565,
      "step": 87810
    },
    {
      "epoch": 10.567990373044525,
      "grad_norm": 8.944977760314941,
      "learning_rate": 0.00019510020720019214,
      "loss": 7.4081,
      "step": 87820
    },
    {
      "epoch": 10.56919374247894,
      "grad_norm": 8.774030685424805,
      "learning_rate": 0.00019509903091008646,
      "loss": 7.2951,
      "step": 87830
    },
    {
      "epoch": 10.570397111913357,
      "grad_norm": 6.231395244598389,
      "learning_rate": 0.00019509785448234906,
      "loss": 7.3018,
      "step": 87840
    },
    {
      "epoch": 10.571600481347774,
      "grad_norm": 10.9324951171875,
      "learning_rate": 0.00019509667791698155,
      "loss": 7.3708,
      "step": 87850
    },
    {
      "epoch": 10.57280385078219,
      "grad_norm": 5.3968329429626465,
      "learning_rate": 0.00019509550121398568,
      "loss": 7.2815,
      "step": 87860
    },
    {
      "epoch": 10.574007220216606,
      "grad_norm": 5.910791873931885,
      "learning_rate": 0.00019509432437336312,
      "loss": 7.2696,
      "step": 87870
    },
    {
      "epoch": 10.575210589651023,
      "grad_norm": 12.487360954284668,
      "learning_rate": 0.00019509314739511564,
      "loss": 7.3683,
      "step": 87880
    },
    {
      "epoch": 10.57641395908544,
      "grad_norm": 10.230481147766113,
      "learning_rate": 0.00019509197027924485,
      "loss": 7.3155,
      "step": 87890
    },
    {
      "epoch": 10.577617328519855,
      "grad_norm": 20.70260238647461,
      "learning_rate": 0.00019509079302575252,
      "loss": 7.2053,
      "step": 87900
    },
    {
      "epoch": 10.578820697954272,
      "grad_norm": 5.648168563842773,
      "learning_rate": 0.00019508961563464034,
      "loss": 7.3536,
      "step": 87910
    },
    {
      "epoch": 10.580024067388688,
      "grad_norm": 11.615484237670898,
      "learning_rate": 0.00019508843810591003,
      "loss": 7.3168,
      "step": 87920
    },
    {
      "epoch": 10.581227436823104,
      "grad_norm": 44.52412033081055,
      "learning_rate": 0.00019508726043956324,
      "loss": 7.3478,
      "step": 87930
    },
    {
      "epoch": 10.582430806257522,
      "grad_norm": 15.564360618591309,
      "learning_rate": 0.0001950860826356017,
      "loss": 7.344,
      "step": 87940
    },
    {
      "epoch": 10.583634175691937,
      "grad_norm": 14.57740592956543,
      "learning_rate": 0.00019508490469402715,
      "loss": 7.4948,
      "step": 87950
    },
    {
      "epoch": 10.584837545126353,
      "grad_norm": 37.10041046142578,
      "learning_rate": 0.00019508372661484127,
      "loss": 7.3218,
      "step": 87960
    },
    {
      "epoch": 10.58604091456077,
      "grad_norm": 20.117599487304688,
      "learning_rate": 0.00019508254839804577,
      "loss": 7.4361,
      "step": 87970
    },
    {
      "epoch": 10.587244283995187,
      "grad_norm": 15.298394203186035,
      "learning_rate": 0.0001950813700436423,
      "loss": 7.335,
      "step": 87980
    },
    {
      "epoch": 10.588447653429602,
      "grad_norm": 10.638603210449219,
      "learning_rate": 0.00019508019155163264,
      "loss": 7.3783,
      "step": 87990
    },
    {
      "epoch": 10.58965102286402,
      "grad_norm": 31.82084083557129,
      "learning_rate": 0.00019507901292201848,
      "loss": 7.4785,
      "step": 88000
    },
    {
      "epoch": 10.590854392298436,
      "grad_norm": 27.33327293395996,
      "learning_rate": 0.0001950778341548015,
      "loss": 7.2916,
      "step": 88010
    },
    {
      "epoch": 10.592057761732852,
      "grad_norm": 9.71547794342041,
      "learning_rate": 0.00019507665524998342,
      "loss": 7.4436,
      "step": 88020
    },
    {
      "epoch": 10.593261131167269,
      "grad_norm": 18.68053436279297,
      "learning_rate": 0.00019507547620756593,
      "loss": 7.364,
      "step": 88030
    },
    {
      "epoch": 10.594464500601685,
      "grad_norm": 15.65990924835205,
      "learning_rate": 0.0001950742970275508,
      "loss": 7.3381,
      "step": 88040
    },
    {
      "epoch": 10.5956678700361,
      "grad_norm": 14.456318855285645,
      "learning_rate": 0.00019507311770993965,
      "loss": 7.3626,
      "step": 88050
    },
    {
      "epoch": 10.596871239470518,
      "grad_norm": 15.827197074890137,
      "learning_rate": 0.00019507193825473424,
      "loss": 7.308,
      "step": 88060
    },
    {
      "epoch": 10.598074608904934,
      "grad_norm": 13.547784805297852,
      "learning_rate": 0.00019507075866193624,
      "loss": 7.3595,
      "step": 88070
    },
    {
      "epoch": 10.59927797833935,
      "grad_norm": 25.571096420288086,
      "learning_rate": 0.0001950695789315474,
      "loss": 7.279,
      "step": 88080
    },
    {
      "epoch": 10.600481347773766,
      "grad_norm": 23.713600158691406,
      "learning_rate": 0.00019506839906356942,
      "loss": 7.35,
      "step": 88090
    },
    {
      "epoch": 10.601684717208183,
      "grad_norm": 40.97940444946289,
      "learning_rate": 0.000195067219058004,
      "loss": 7.4039,
      "step": 88100
    },
    {
      "epoch": 10.602888086642599,
      "grad_norm": 14.471735000610352,
      "learning_rate": 0.00019506603891485282,
      "loss": 7.4411,
      "step": 88110
    },
    {
      "epoch": 10.604091456077015,
      "grad_norm": 174.0249481201172,
      "learning_rate": 0.0001950648586341176,
      "loss": 7.4085,
      "step": 88120
    },
    {
      "epoch": 10.605294825511432,
      "grad_norm": 22.778398513793945,
      "learning_rate": 0.0001950636782158001,
      "loss": 7.4301,
      "step": 88130
    },
    {
      "epoch": 10.606498194945848,
      "grad_norm": 21.839618682861328,
      "learning_rate": 0.00019506249765990195,
      "loss": 7.3172,
      "step": 88140
    },
    {
      "epoch": 10.607701564380264,
      "grad_norm": 25.78067398071289,
      "learning_rate": 0.0001950613169664249,
      "loss": 7.3865,
      "step": 88150
    },
    {
      "epoch": 10.608904933814681,
      "grad_norm": 37.15629959106445,
      "learning_rate": 0.00019506013613537066,
      "loss": 7.2948,
      "step": 88160
    },
    {
      "epoch": 10.610108303249097,
      "grad_norm": 36.7580680847168,
      "learning_rate": 0.00019505895516674096,
      "loss": 7.3327,
      "step": 88170
    },
    {
      "epoch": 10.611311672683513,
      "grad_norm": 75.67523193359375,
      "learning_rate": 0.00019505777406053745,
      "loss": 7.2686,
      "step": 88180
    },
    {
      "epoch": 10.61251504211793,
      "grad_norm": 56.04183578491211,
      "learning_rate": 0.00019505659281676186,
      "loss": 7.3916,
      "step": 88190
    },
    {
      "epoch": 10.613718411552346,
      "grad_norm": 55.43891906738281,
      "learning_rate": 0.00019505541143541595,
      "loss": 7.2491,
      "step": 88200
    },
    {
      "epoch": 10.614921780986762,
      "grad_norm": 50.380245208740234,
      "learning_rate": 0.00019505422991650137,
      "loss": 7.4029,
      "step": 88210
    },
    {
      "epoch": 10.61612515042118,
      "grad_norm": 35.543277740478516,
      "learning_rate": 0.00019505304826001983,
      "loss": 7.3614,
      "step": 88220
    },
    {
      "epoch": 10.617328519855596,
      "grad_norm": 61.23712158203125,
      "learning_rate": 0.00019505186646597312,
      "loss": 7.4214,
      "step": 88230
    },
    {
      "epoch": 10.618531889290011,
      "grad_norm": 26.799434661865234,
      "learning_rate": 0.00019505068453436284,
      "loss": 7.324,
      "step": 88240
    },
    {
      "epoch": 10.619735258724429,
      "grad_norm": 43.014469146728516,
      "learning_rate": 0.00019504950246519075,
      "loss": 7.4568,
      "step": 88250
    },
    {
      "epoch": 10.620938628158845,
      "grad_norm": 268.6820983886719,
      "learning_rate": 0.00019504832025845858,
      "loss": 7.3789,
      "step": 88260
    },
    {
      "epoch": 10.62214199759326,
      "grad_norm": 48.661808013916016,
      "learning_rate": 0.00019504713791416805,
      "loss": 7.4168,
      "step": 88270
    },
    {
      "epoch": 10.623345367027678,
      "grad_norm": 72.69803619384766,
      "learning_rate": 0.00019504595543232082,
      "loss": 7.3598,
      "step": 88280
    },
    {
      "epoch": 10.624548736462094,
      "grad_norm": 101.9991455078125,
      "learning_rate": 0.00019504477281291862,
      "loss": 7.3906,
      "step": 88290
    },
    {
      "epoch": 10.62575210589651,
      "grad_norm": 204.7308349609375,
      "learning_rate": 0.00019504359005596315,
      "loss": 7.3497,
      "step": 88300
    },
    {
      "epoch": 10.626955475330927,
      "grad_norm": 112.8714599609375,
      "learning_rate": 0.0001950424071614562,
      "loss": 7.2988,
      "step": 88310
    },
    {
      "epoch": 10.628158844765343,
      "grad_norm": 229.57504272460938,
      "learning_rate": 0.00019504122412939938,
      "loss": 7.2327,
      "step": 88320
    },
    {
      "epoch": 10.629362214199759,
      "grad_norm": 247.52845764160156,
      "learning_rate": 0.00019504004095979445,
      "loss": 7.3221,
      "step": 88330
    },
    {
      "epoch": 10.630565583634176,
      "grad_norm": 152.36831665039062,
      "learning_rate": 0.0001950388576526431,
      "loss": 7.2489,
      "step": 88340
    },
    {
      "epoch": 10.631768953068592,
      "grad_norm": 147.5357208251953,
      "learning_rate": 0.00019503767420794708,
      "loss": 7.3202,
      "step": 88350
    },
    {
      "epoch": 10.632972322503008,
      "grad_norm": 181.00552368164062,
      "learning_rate": 0.00019503649062570808,
      "loss": 7.3037,
      "step": 88360
    },
    {
      "epoch": 10.634175691937426,
      "grad_norm": 67.28950500488281,
      "learning_rate": 0.00019503530690592781,
      "loss": 7.3349,
      "step": 88370
    },
    {
      "epoch": 10.635379061371841,
      "grad_norm": 208.2025909423828,
      "learning_rate": 0.000195034123048608,
      "loss": 7.3797,
      "step": 88380
    },
    {
      "epoch": 10.636582430806257,
      "grad_norm": 123.33839416503906,
      "learning_rate": 0.00019503293905375033,
      "loss": 7.3294,
      "step": 88390
    },
    {
      "epoch": 10.637785800240675,
      "grad_norm": 141.5004119873047,
      "learning_rate": 0.00019503175492135654,
      "loss": 7.2263,
      "step": 88400
    },
    {
      "epoch": 10.63898916967509,
      "grad_norm": 173.01364135742188,
      "learning_rate": 0.00019503057065142833,
      "loss": 7.2706,
      "step": 88410
    },
    {
      "epoch": 10.640192539109506,
      "grad_norm": 66.8580093383789,
      "learning_rate": 0.00019502938624396744,
      "loss": 7.438,
      "step": 88420
    },
    {
      "epoch": 10.641395908543924,
      "grad_norm": 209.61973571777344,
      "learning_rate": 0.00019502820169897556,
      "loss": 7.3632,
      "step": 88430
    },
    {
      "epoch": 10.64259927797834,
      "grad_norm": 132.6012420654297,
      "learning_rate": 0.0001950270170164544,
      "loss": 7.3851,
      "step": 88440
    },
    {
      "epoch": 10.643802647412755,
      "grad_norm": 149.63917541503906,
      "learning_rate": 0.0001950258321964057,
      "loss": 7.4109,
      "step": 88450
    },
    {
      "epoch": 10.645006016847173,
      "grad_norm": 141.02175903320312,
      "learning_rate": 0.00019502464723883116,
      "loss": 7.3332,
      "step": 88460
    },
    {
      "epoch": 10.646209386281589,
      "grad_norm": 161.88961791992188,
      "learning_rate": 0.0001950234621437325,
      "loss": 7.3588,
      "step": 88470
    },
    {
      "epoch": 10.647412755716005,
      "grad_norm": 250.7991180419922,
      "learning_rate": 0.0001950222769111114,
      "loss": 7.2971,
      "step": 88480
    },
    {
      "epoch": 10.648616125150422,
      "grad_norm": 95.60113525390625,
      "learning_rate": 0.00019502109154096963,
      "loss": 7.3376,
      "step": 88490
    },
    {
      "epoch": 10.649819494584838,
      "grad_norm": 187.06863403320312,
      "learning_rate": 0.0001950199060333089,
      "loss": 7.2651,
      "step": 88500
    },
    {
      "epoch": 10.651022864019254,
      "grad_norm": 115.3002700805664,
      "learning_rate": 0.00019501872038813086,
      "loss": 7.286,
      "step": 88510
    },
    {
      "epoch": 10.65222623345367,
      "grad_norm": 66.42501831054688,
      "learning_rate": 0.0001950175346054373,
      "loss": 7.3652,
      "step": 88520
    },
    {
      "epoch": 10.653429602888087,
      "grad_norm": 140.0968780517578,
      "learning_rate": 0.0001950163486852299,
      "loss": 7.2641,
      "step": 88530
    },
    {
      "epoch": 10.654632972322503,
      "grad_norm": 129.69400024414062,
      "learning_rate": 0.00019501516262751038,
      "loss": 7.3465,
      "step": 88540
    },
    {
      "epoch": 10.655836341756919,
      "grad_norm": 166.33387756347656,
      "learning_rate": 0.00019501397643228047,
      "loss": 7.3526,
      "step": 88550
    },
    {
      "epoch": 10.657039711191336,
      "grad_norm": 69.1812515258789,
      "learning_rate": 0.00019501279009954186,
      "loss": 7.286,
      "step": 88560
    },
    {
      "epoch": 10.658243080625752,
      "grad_norm": 142.8148651123047,
      "learning_rate": 0.00019501160362929632,
      "loss": 7.2852,
      "step": 88570
    },
    {
      "epoch": 10.659446450060168,
      "grad_norm": 141.66908264160156,
      "learning_rate": 0.0001950104170215455,
      "loss": 7.2608,
      "step": 88580
    },
    {
      "epoch": 10.660649819494585,
      "grad_norm": 51.66764450073242,
      "learning_rate": 0.00019500923027629118,
      "loss": 7.2969,
      "step": 88590
    },
    {
      "epoch": 10.661853188929001,
      "grad_norm": 109.10205078125,
      "learning_rate": 0.00019500804339353504,
      "loss": 7.3724,
      "step": 88600
    },
    {
      "epoch": 10.663056558363417,
      "grad_norm": 93.02632141113281,
      "learning_rate": 0.00019500685637327875,
      "loss": 7.2457,
      "step": 88610
    },
    {
      "epoch": 10.664259927797834,
      "grad_norm": 73.33341979980469,
      "learning_rate": 0.00019500566921552414,
      "loss": 7.3102,
      "step": 88620
    },
    {
      "epoch": 10.66546329723225,
      "grad_norm": 102.7935791015625,
      "learning_rate": 0.00019500448192027284,
      "loss": 7.3332,
      "step": 88630
    },
    {
      "epoch": 10.666666666666666,
      "grad_norm": 108.53228759765625,
      "learning_rate": 0.00019500329448752664,
      "loss": 7.3152,
      "step": 88640
    },
    {
      "epoch": 10.667870036101084,
      "grad_norm": 103.6209716796875,
      "learning_rate": 0.00019500210691728717,
      "loss": 7.2561,
      "step": 88650
    },
    {
      "epoch": 10.6690734055355,
      "grad_norm": 99.7047119140625,
      "learning_rate": 0.00019500091920955623,
      "loss": 7.2516,
      "step": 88660
    },
    {
      "epoch": 10.670276774969915,
      "grad_norm": 62.162837982177734,
      "learning_rate": 0.0001949997313643355,
      "loss": 7.2708,
      "step": 88670
    },
    {
      "epoch": 10.671480144404333,
      "grad_norm": 84.71681213378906,
      "learning_rate": 0.0001949985433816267,
      "loss": 7.2769,
      "step": 88680
    },
    {
      "epoch": 10.672683513838749,
      "grad_norm": 69.47581481933594,
      "learning_rate": 0.00019499735526143152,
      "loss": 7.2773,
      "step": 88690
    },
    {
      "epoch": 10.673886883273164,
      "grad_norm": 210.3521270751953,
      "learning_rate": 0.00019499616700375177,
      "loss": 7.3398,
      "step": 88700
    },
    {
      "epoch": 10.675090252707582,
      "grad_norm": 79.78470611572266,
      "learning_rate": 0.00019499497860858909,
      "loss": 7.1656,
      "step": 88710
    },
    {
      "epoch": 10.676293622141998,
      "grad_norm": 62.30799865722656,
      "learning_rate": 0.00019499379007594517,
      "loss": 7.2361,
      "step": 88720
    },
    {
      "epoch": 10.677496991576414,
      "grad_norm": 91.21034240722656,
      "learning_rate": 0.00019499260140582188,
      "loss": 7.3132,
      "step": 88730
    },
    {
      "epoch": 10.678700361010831,
      "grad_norm": 43.911380767822266,
      "learning_rate": 0.00019499141259822077,
      "loss": 7.3459,
      "step": 88740
    },
    {
      "epoch": 10.679903730445247,
      "grad_norm": 105.96215057373047,
      "learning_rate": 0.00019499022365314367,
      "loss": 7.2649,
      "step": 88750
    },
    {
      "epoch": 10.681107099879663,
      "grad_norm": 248.10601806640625,
      "learning_rate": 0.00019498903457059222,
      "loss": 7.2417,
      "step": 88760
    },
    {
      "epoch": 10.68231046931408,
      "grad_norm": 38.80183410644531,
      "learning_rate": 0.00019498784535056825,
      "loss": 7.2187,
      "step": 88770
    },
    {
      "epoch": 10.683513838748496,
      "grad_norm": 59.740116119384766,
      "learning_rate": 0.00019498665599307336,
      "loss": 7.2648,
      "step": 88780
    },
    {
      "epoch": 10.684717208182912,
      "grad_norm": 26.129091262817383,
      "learning_rate": 0.00019498546649810935,
      "loss": 7.2915,
      "step": 88790
    },
    {
      "epoch": 10.685920577617328,
      "grad_norm": 56.78028869628906,
      "learning_rate": 0.00019498427686567791,
      "loss": 7.3406,
      "step": 88800
    },
    {
      "epoch": 10.687123947051745,
      "grad_norm": 64.2637939453125,
      "learning_rate": 0.00019498308709578078,
      "loss": 7.3238,
      "step": 88810
    },
    {
      "epoch": 10.688327316486161,
      "grad_norm": 36.57625961303711,
      "learning_rate": 0.0001949818971884197,
      "loss": 7.3383,
      "step": 88820
    },
    {
      "epoch": 10.689530685920577,
      "grad_norm": 265.4883117675781,
      "learning_rate": 0.00019498070714359632,
      "loss": 7.3633,
      "step": 88830
    },
    {
      "epoch": 10.690734055354994,
      "grad_norm": 93.23102569580078,
      "learning_rate": 0.0001949795169613124,
      "loss": 7.3179,
      "step": 88840
    },
    {
      "epoch": 10.69193742478941,
      "grad_norm": 205.05117797851562,
      "learning_rate": 0.00019497832664156972,
      "loss": 7.4237,
      "step": 88850
    },
    {
      "epoch": 10.693140794223826,
      "grad_norm": 145.75830078125,
      "learning_rate": 0.00019497713618436992,
      "loss": 7.3747,
      "step": 88860
    },
    {
      "epoch": 10.694344163658243,
      "grad_norm": 68.23468017578125,
      "learning_rate": 0.00019497594558971475,
      "loss": 7.393,
      "step": 88870
    },
    {
      "epoch": 10.69554753309266,
      "grad_norm": 116.01805114746094,
      "learning_rate": 0.00019497475485760594,
      "loss": 7.4107,
      "step": 88880
    },
    {
      "epoch": 10.696750902527075,
      "grad_norm": 62.533447265625,
      "learning_rate": 0.00019497356398804524,
      "loss": 7.3809,
      "step": 88890
    },
    {
      "epoch": 10.697954271961493,
      "grad_norm": 168.87171936035156,
      "learning_rate": 0.0001949723729810343,
      "loss": 7.4853,
      "step": 88900
    },
    {
      "epoch": 10.699157641395908,
      "grad_norm": 166.12615966796875,
      "learning_rate": 0.00019497118183657493,
      "loss": 7.5827,
      "step": 88910
    },
    {
      "epoch": 10.700361010830324,
      "grad_norm": 50.531700134277344,
      "learning_rate": 0.00019496999055466878,
      "loss": 7.4548,
      "step": 88920
    },
    {
      "epoch": 10.701564380264742,
      "grad_norm": 149.77862548828125,
      "learning_rate": 0.0001949687991353176,
      "loss": 7.426,
      "step": 88930
    },
    {
      "epoch": 10.702767749699158,
      "grad_norm": 94.55686950683594,
      "learning_rate": 0.00019496760757852316,
      "loss": 7.4463,
      "step": 88940
    },
    {
      "epoch": 10.703971119133573,
      "grad_norm": 62.388206481933594,
      "learning_rate": 0.00019496641588428714,
      "loss": 7.4493,
      "step": 88950
    },
    {
      "epoch": 10.705174488567991,
      "grad_norm": 66.34426879882812,
      "learning_rate": 0.00019496522405261125,
      "loss": 7.3723,
      "step": 88960
    },
    {
      "epoch": 10.706377858002407,
      "grad_norm": 40.63010787963867,
      "learning_rate": 0.0001949640320834972,
      "loss": 7.3687,
      "step": 88970
    },
    {
      "epoch": 10.707581227436823,
      "grad_norm": 97.47332763671875,
      "learning_rate": 0.0001949628399769468,
      "loss": 7.3967,
      "step": 88980
    },
    {
      "epoch": 10.70878459687124,
      "grad_norm": 52.6695556640625,
      "learning_rate": 0.0001949616477329617,
      "loss": 7.3656,
      "step": 88990
    },
    {
      "epoch": 10.709987966305656,
      "grad_norm": 87.3397445678711,
      "learning_rate": 0.00019496045535154368,
      "loss": 7.3819,
      "step": 89000
    },
    {
      "epoch": 10.711191335740072,
      "grad_norm": 225.9004669189453,
      "learning_rate": 0.00019495926283269444,
      "loss": 7.3577,
      "step": 89010
    },
    {
      "epoch": 10.71239470517449,
      "grad_norm": 85.98999786376953,
      "learning_rate": 0.00019495807017641566,
      "loss": 7.3868,
      "step": 89020
    },
    {
      "epoch": 10.713598074608905,
      "grad_norm": 143.83712768554688,
      "learning_rate": 0.00019495687738270914,
      "loss": 7.3105,
      "step": 89030
    },
    {
      "epoch": 10.71480144404332,
      "grad_norm": 78.22279357910156,
      "learning_rate": 0.00019495568445157657,
      "loss": 7.2924,
      "step": 89040
    },
    {
      "epoch": 10.716004813477738,
      "grad_norm": 248.8448944091797,
      "learning_rate": 0.00019495449138301967,
      "loss": 7.3742,
      "step": 89050
    },
    {
      "epoch": 10.717208182912154,
      "grad_norm": 63.11293411254883,
      "learning_rate": 0.00019495329817704018,
      "loss": 7.369,
      "step": 89060
    },
    {
      "epoch": 10.71841155234657,
      "grad_norm": 67.78326416015625,
      "learning_rate": 0.00019495210483363984,
      "loss": 7.343,
      "step": 89070
    },
    {
      "epoch": 10.719614921780988,
      "grad_norm": 101.28987121582031,
      "learning_rate": 0.00019495091135282033,
      "loss": 7.3267,
      "step": 89080
    },
    {
      "epoch": 10.720818291215403,
      "grad_norm": 61.997962951660156,
      "learning_rate": 0.00019494971773458342,
      "loss": 7.3417,
      "step": 89090
    },
    {
      "epoch": 10.722021660649819,
      "grad_norm": 349.41046142578125,
      "learning_rate": 0.00019494852397893085,
      "loss": 7.3135,
      "step": 89100
    },
    {
      "epoch": 10.723225030084237,
      "grad_norm": 105.47533416748047,
      "learning_rate": 0.0001949473300858643,
      "loss": 7.3145,
      "step": 89110
    },
    {
      "epoch": 10.724428399518652,
      "grad_norm": 106.47731018066406,
      "learning_rate": 0.00019494613605538553,
      "loss": 7.3453,
      "step": 89120
    },
    {
      "epoch": 10.725631768953068,
      "grad_norm": 189.0004425048828,
      "learning_rate": 0.00019494494188749626,
      "loss": 7.3526,
      "step": 89130
    },
    {
      "epoch": 10.726835138387486,
      "grad_norm": 245.7454071044922,
      "learning_rate": 0.0001949437475821982,
      "loss": 7.3629,
      "step": 89140
    },
    {
      "epoch": 10.728038507821902,
      "grad_norm": 319.6542053222656,
      "learning_rate": 0.00019494255313949312,
      "loss": 7.4202,
      "step": 89150
    },
    {
      "epoch": 10.729241877256317,
      "grad_norm": 200.45176696777344,
      "learning_rate": 0.00019494135855938272,
      "loss": 7.3789,
      "step": 89160
    },
    {
      "epoch": 10.730445246690735,
      "grad_norm": 246.00425720214844,
      "learning_rate": 0.00019494016384186873,
      "loss": 7.3848,
      "step": 89170
    },
    {
      "epoch": 10.73164861612515,
      "grad_norm": 587.3739624023438,
      "learning_rate": 0.00019493896898695287,
      "loss": 7.3729,
      "step": 89180
    },
    {
      "epoch": 10.732851985559567,
      "grad_norm": 169.81912231445312,
      "learning_rate": 0.00019493777399463692,
      "loss": 7.4477,
      "step": 89190
    },
    {
      "epoch": 10.734055354993982,
      "grad_norm": 132.12127685546875,
      "learning_rate": 0.00019493657886492254,
      "loss": 7.3375,
      "step": 89200
    },
    {
      "epoch": 10.7352587244284,
      "grad_norm": 114.60580444335938,
      "learning_rate": 0.0001949353835978115,
      "loss": 7.384,
      "step": 89210
    },
    {
      "epoch": 10.736462093862816,
      "grad_norm": 85.80974578857422,
      "learning_rate": 0.00019493418819330552,
      "loss": 7.4829,
      "step": 89220
    },
    {
      "epoch": 10.737665463297231,
      "grad_norm": 102.380126953125,
      "learning_rate": 0.00019493299265140634,
      "loss": 7.3055,
      "step": 89230
    },
    {
      "epoch": 10.738868832731649,
      "grad_norm": 155.44085693359375,
      "learning_rate": 0.00019493179697211566,
      "loss": 7.3606,
      "step": 89240
    },
    {
      "epoch": 10.740072202166065,
      "grad_norm": 312.52703857421875,
      "learning_rate": 0.00019493060115543528,
      "loss": 7.4591,
      "step": 89250
    },
    {
      "epoch": 10.74127557160048,
      "grad_norm": 82.09359741210938,
      "learning_rate": 0.00019492940520136683,
      "loss": 7.2659,
      "step": 89260
    },
    {
      "epoch": 10.742478941034898,
      "grad_norm": 33.064884185791016,
      "learning_rate": 0.00019492820910991209,
      "loss": 7.4064,
      "step": 89270
    },
    {
      "epoch": 10.743682310469314,
      "grad_norm": 230.91871643066406,
      "learning_rate": 0.00019492701288107282,
      "loss": 7.3389,
      "step": 89280
    },
    {
      "epoch": 10.74488567990373,
      "grad_norm": 211.80934143066406,
      "learning_rate": 0.00019492581651485072,
      "loss": 7.4542,
      "step": 89290
    },
    {
      "epoch": 10.746089049338147,
      "grad_norm": 206.44613647460938,
      "learning_rate": 0.00019492462001124755,
      "loss": 7.2896,
      "step": 89300
    },
    {
      "epoch": 10.747292418772563,
      "grad_norm": 56.54222869873047,
      "learning_rate": 0.00019492342337026498,
      "loss": 7.3159,
      "step": 89310
    },
    {
      "epoch": 10.748495788206979,
      "grad_norm": 46.79085922241211,
      "learning_rate": 0.0001949222265919048,
      "loss": 7.5051,
      "step": 89320
    },
    {
      "epoch": 10.749699157641396,
      "grad_norm": 68.00275421142578,
      "learning_rate": 0.00019492102967616872,
      "loss": 7.4421,
      "step": 89330
    },
    {
      "epoch": 10.750902527075812,
      "grad_norm": 23.74716567993164,
      "learning_rate": 0.0001949198326230585,
      "loss": 7.3478,
      "step": 89340
    },
    {
      "epoch": 10.752105896510228,
      "grad_norm": 83.93447875976562,
      "learning_rate": 0.00019491863543257579,
      "loss": 7.3559,
      "step": 89350
    },
    {
      "epoch": 10.753309265944646,
      "grad_norm": 33.42729187011719,
      "learning_rate": 0.00019491743810472244,
      "loss": 7.3887,
      "step": 89360
    },
    {
      "epoch": 10.754512635379061,
      "grad_norm": 50.58473205566406,
      "learning_rate": 0.00019491624063950008,
      "loss": 7.3563,
      "step": 89370
    },
    {
      "epoch": 10.755716004813477,
      "grad_norm": 51.0051155090332,
      "learning_rate": 0.00019491504303691047,
      "loss": 7.3379,
      "step": 89380
    },
    {
      "epoch": 10.756919374247895,
      "grad_norm": 22.731534957885742,
      "learning_rate": 0.0001949138452969554,
      "loss": 7.4411,
      "step": 89390
    },
    {
      "epoch": 10.75812274368231,
      "grad_norm": 11.859912872314453,
      "learning_rate": 0.00019491264741963656,
      "loss": 7.405,
      "step": 89400
    },
    {
      "epoch": 10.759326113116726,
      "grad_norm": 23.532920837402344,
      "learning_rate": 0.00019491144940495566,
      "loss": 7.3864,
      "step": 89410
    },
    {
      "epoch": 10.760529482551144,
      "grad_norm": 27.584901809692383,
      "learning_rate": 0.00019491025125291449,
      "loss": 7.419,
      "step": 89420
    },
    {
      "epoch": 10.76173285198556,
      "grad_norm": 44.958309173583984,
      "learning_rate": 0.00019490905296351473,
      "loss": 7.3588,
      "step": 89430
    },
    {
      "epoch": 10.762936221419976,
      "grad_norm": 32.18195724487305,
      "learning_rate": 0.00019490785453675816,
      "loss": 7.382,
      "step": 89440
    },
    {
      "epoch": 10.764139590854393,
      "grad_norm": 32.33243179321289,
      "learning_rate": 0.0001949066559726465,
      "loss": 7.4039,
      "step": 89450
    },
    {
      "epoch": 10.765342960288809,
      "grad_norm": 63.45510482788086,
      "learning_rate": 0.00019490545727118144,
      "loss": 7.4506,
      "step": 89460
    },
    {
      "epoch": 10.766546329723225,
      "grad_norm": 211.78750610351562,
      "learning_rate": 0.00019490425843236478,
      "loss": 7.459,
      "step": 89470
    },
    {
      "epoch": 10.76774969915764,
      "grad_norm": 1398.822509765625,
      "learning_rate": 0.00019490305945619823,
      "loss": 7.3793,
      "step": 89480
    },
    {
      "epoch": 10.768953068592058,
      "grad_norm": 277.9879455566406,
      "learning_rate": 0.0001949018603426835,
      "loss": 7.4114,
      "step": 89490
    },
    {
      "epoch": 10.770156438026474,
      "grad_norm": 571.7463989257812,
      "learning_rate": 0.00019490066109182234,
      "loss": 7.5029,
      "step": 89500
    },
    {
      "epoch": 10.77135980746089,
      "grad_norm": 409.6896057128906,
      "learning_rate": 0.00019489946170361653,
      "loss": 7.4473,
      "step": 89510
    },
    {
      "epoch": 10.772563176895307,
      "grad_norm": 375.6978759765625,
      "learning_rate": 0.00019489826217806774,
      "loss": 7.5655,
      "step": 89520
    },
    {
      "epoch": 10.773766546329723,
      "grad_norm": 652.9776000976562,
      "learning_rate": 0.00019489706251517777,
      "loss": 7.4944,
      "step": 89530
    },
    {
      "epoch": 10.774969915764139,
      "grad_norm": 430.00433349609375,
      "learning_rate": 0.00019489586271494828,
      "loss": 7.5482,
      "step": 89540
    },
    {
      "epoch": 10.776173285198556,
      "grad_norm": 24.84916114807129,
      "learning_rate": 0.00019489466277738105,
      "loss": 7.5287,
      "step": 89550
    },
    {
      "epoch": 10.777376654632972,
      "grad_norm": 27.26685905456543,
      "learning_rate": 0.00019489346270247783,
      "loss": 7.4141,
      "step": 89560
    },
    {
      "epoch": 10.778580024067388,
      "grad_norm": 44.02090072631836,
      "learning_rate": 0.00019489226249024035,
      "loss": 7.3955,
      "step": 89570
    },
    {
      "epoch": 10.779783393501805,
      "grad_norm": 64.54180908203125,
      "learning_rate": 0.0001948910621406703,
      "loss": 7.4451,
      "step": 89580
    },
    {
      "epoch": 10.780986762936221,
      "grad_norm": 53.844871520996094,
      "learning_rate": 0.0001948898616537695,
      "loss": 7.391,
      "step": 89590
    },
    {
      "epoch": 10.782190132370637,
      "grad_norm": 243.5305938720703,
      "learning_rate": 0.00019488866102953964,
      "loss": 7.4442,
      "step": 89600
    },
    {
      "epoch": 10.783393501805055,
      "grad_norm": 215.44296264648438,
      "learning_rate": 0.00019488746026798242,
      "loss": 7.3934,
      "step": 89610
    },
    {
      "epoch": 10.78459687123947,
      "grad_norm": 541.2142333984375,
      "learning_rate": 0.00019488625936909965,
      "loss": 7.4049,
      "step": 89620
    },
    {
      "epoch": 10.785800240673886,
      "grad_norm": 407.396484375,
      "learning_rate": 0.00019488505833289304,
      "loss": 7.3971,
      "step": 89630
    },
    {
      "epoch": 10.787003610108304,
      "grad_norm": 39.224212646484375,
      "learning_rate": 0.0001948838571593643,
      "loss": 7.3513,
      "step": 89640
    },
    {
      "epoch": 10.78820697954272,
      "grad_norm": 41.66058349609375,
      "learning_rate": 0.00019488265584851523,
      "loss": 7.352,
      "step": 89650
    },
    {
      "epoch": 10.789410348977135,
      "grad_norm": 74.9732666015625,
      "learning_rate": 0.00019488145440034747,
      "loss": 7.3182,
      "step": 89660
    },
    {
      "epoch": 10.790613718411553,
      "grad_norm": 168.19195556640625,
      "learning_rate": 0.00019488025281486288,
      "loss": 7.366,
      "step": 89670
    },
    {
      "epoch": 10.791817087845969,
      "grad_norm": 243.7516632080078,
      "learning_rate": 0.00019487905109206313,
      "loss": 7.428,
      "step": 89680
    },
    {
      "epoch": 10.793020457280385,
      "grad_norm": 474.8160705566406,
      "learning_rate": 0.00019487784923194994,
      "loss": 7.4098,
      "step": 89690
    },
    {
      "epoch": 10.794223826714802,
      "grad_norm": 416.8306884765625,
      "learning_rate": 0.0001948766472345251,
      "loss": 7.4483,
      "step": 89700
    },
    {
      "epoch": 10.795427196149218,
      "grad_norm": 85.0167236328125,
      "learning_rate": 0.0001948754450997903,
      "loss": 7.3781,
      "step": 89710
    },
    {
      "epoch": 10.796630565583634,
      "grad_norm": 142.3661651611328,
      "learning_rate": 0.00019487424282774732,
      "loss": 7.42,
      "step": 89720
    },
    {
      "epoch": 10.797833935018051,
      "grad_norm": 393.02716064453125,
      "learning_rate": 0.0001948730404183979,
      "loss": 7.3965,
      "step": 89730
    },
    {
      "epoch": 10.799037304452467,
      "grad_norm": 447.46881103515625,
      "learning_rate": 0.0001948718378717438,
      "loss": 7.4281,
      "step": 89740
    },
    {
      "epoch": 10.800240673886883,
      "grad_norm": 14.403511047363281,
      "learning_rate": 0.00019487063518778666,
      "loss": 7.3942,
      "step": 89750
    },
    {
      "epoch": 10.8014440433213,
      "grad_norm": 171.69520568847656,
      "learning_rate": 0.00019486943236652832,
      "loss": 7.4459,
      "step": 89760
    },
    {
      "epoch": 10.802647412755716,
      "grad_norm": 7661.4140625,
      "learning_rate": 0.0001948682294079705,
      "loss": 7.3851,
      "step": 89770
    },
    {
      "epoch": 10.803850782190132,
      "grad_norm": 2992.94482421875,
      "learning_rate": 0.00019486702631211488,
      "loss": 7.4832,
      "step": 89780
    },
    {
      "epoch": 10.80505415162455,
      "grad_norm": 295.5586853027344,
      "learning_rate": 0.00019486582307896331,
      "loss": 7.4184,
      "step": 89790
    },
    {
      "epoch": 10.806257521058965,
      "grad_norm": 355.16094970703125,
      "learning_rate": 0.00019486461970851746,
      "loss": 7.4807,
      "step": 89800
    },
    {
      "epoch": 10.807460890493381,
      "grad_norm": 10914.90625,
      "learning_rate": 0.00019486341620077906,
      "loss": 7.536,
      "step": 89810
    },
    {
      "epoch": 10.808664259927799,
      "grad_norm": 585.2021484375,
      "learning_rate": 0.0001948622125557499,
      "loss": 7.5429,
      "step": 89820
    },
    {
      "epoch": 10.809867629362214,
      "grad_norm": 1284.7327880859375,
      "learning_rate": 0.00019486100877343166,
      "loss": 7.4352,
      "step": 89830
    },
    {
      "epoch": 10.81107099879663,
      "grad_norm": 234.9237518310547,
      "learning_rate": 0.00019485980485382616,
      "loss": 7.3521,
      "step": 89840
    },
    {
      "epoch": 10.812274368231048,
      "grad_norm": 157.05502319335938,
      "learning_rate": 0.00019485860079693508,
      "loss": 7.4848,
      "step": 89850
    },
    {
      "epoch": 10.813477737665464,
      "grad_norm": 179.12625122070312,
      "learning_rate": 0.0001948573966027602,
      "loss": 7.4622,
      "step": 89860
    },
    {
      "epoch": 10.81468110709988,
      "grad_norm": 183.79322814941406,
      "learning_rate": 0.00019485619227130324,
      "loss": 7.3719,
      "step": 89870
    },
    {
      "epoch": 10.815884476534297,
      "grad_norm": 1028.42724609375,
      "learning_rate": 0.00019485498780256596,
      "loss": 7.3825,
      "step": 89880
    },
    {
      "epoch": 10.817087845968713,
      "grad_norm": 970.3973999023438,
      "learning_rate": 0.00019485378319655005,
      "loss": 7.4842,
      "step": 89890
    },
    {
      "epoch": 10.818291215403129,
      "grad_norm": 310.7757263183594,
      "learning_rate": 0.00019485257845325735,
      "loss": 7.4399,
      "step": 89900
    },
    {
      "epoch": 10.819494584837544,
      "grad_norm": 134.2033233642578,
      "learning_rate": 0.00019485137357268953,
      "loss": 7.3357,
      "step": 89910
    },
    {
      "epoch": 10.820697954271962,
      "grad_norm": 143.18923950195312,
      "learning_rate": 0.00019485016855484838,
      "loss": 7.4102,
      "step": 89920
    },
    {
      "epoch": 10.821901323706378,
      "grad_norm": 786.7396850585938,
      "learning_rate": 0.00019484896339973557,
      "loss": 7.3891,
      "step": 89930
    },
    {
      "epoch": 10.823104693140793,
      "grad_norm": 105.39263153076172,
      "learning_rate": 0.0001948477581073529,
      "loss": 7.4931,
      "step": 89940
    },
    {
      "epoch": 10.824308062575211,
      "grad_norm": 79.65034484863281,
      "learning_rate": 0.00019484655267770214,
      "loss": 7.4024,
      "step": 89950
    },
    {
      "epoch": 10.825511432009627,
      "grad_norm": 207.6728515625,
      "learning_rate": 0.000194845347110785,
      "loss": 7.408,
      "step": 89960
    },
    {
      "epoch": 10.826714801444043,
      "grad_norm": 291.0705261230469,
      "learning_rate": 0.0001948441414066032,
      "loss": 7.3783,
      "step": 89970
    },
    {
      "epoch": 10.82791817087846,
      "grad_norm": 257.7458190917969,
      "learning_rate": 0.00019484293556515852,
      "loss": 7.4086,
      "step": 89980
    },
    {
      "epoch": 10.829121540312876,
      "grad_norm": 600.5701904296875,
      "learning_rate": 0.0001948417295864527,
      "loss": 7.3214,
      "step": 89990
    },
    {
      "epoch": 10.830324909747292,
      "grad_norm": 126.4278564453125,
      "learning_rate": 0.0001948405234704875,
      "loss": 7.3987,
      "step": 90000
    },
    {
      "epoch": 10.83152827918171,
      "grad_norm": 197.2976837158203,
      "learning_rate": 0.00019483931721726462,
      "loss": 7.3567,
      "step": 90010
    },
    {
      "epoch": 10.832731648616125,
      "grad_norm": 1042.79052734375,
      "learning_rate": 0.0001948381108267858,
      "loss": 7.3748,
      "step": 90020
    },
    {
      "epoch": 10.833935018050541,
      "grad_norm": 205.5945587158203,
      "learning_rate": 0.0001948369042990529,
      "loss": 7.3638,
      "step": 90030
    },
    {
      "epoch": 10.835138387484958,
      "grad_norm": 491.60516357421875,
      "learning_rate": 0.00019483569763406754,
      "loss": 7.3998,
      "step": 90040
    },
    {
      "epoch": 10.836341756919374,
      "grad_norm": 111.6520004272461,
      "learning_rate": 0.00019483449083183152,
      "loss": 7.3486,
      "step": 90050
    },
    {
      "epoch": 10.83754512635379,
      "grad_norm": 99.41780090332031,
      "learning_rate": 0.00019483328389234656,
      "loss": 7.4199,
      "step": 90060
    },
    {
      "epoch": 10.838748495788208,
      "grad_norm": 171.64212036132812,
      "learning_rate": 0.00019483207681561445,
      "loss": 7.3448,
      "step": 90070
    },
    {
      "epoch": 10.839951865222623,
      "grad_norm": 199.79847717285156,
      "learning_rate": 0.0001948308696016369,
      "loss": 7.4762,
      "step": 90080
    },
    {
      "epoch": 10.84115523465704,
      "grad_norm": 270.7060241699219,
      "learning_rate": 0.00019482966225041568,
      "loss": 7.4044,
      "step": 90090
    },
    {
      "epoch": 10.842358604091457,
      "grad_norm": 739.8450927734375,
      "learning_rate": 0.0001948284547619525,
      "loss": 7.4325,
      "step": 90100
    },
    {
      "epoch": 10.843561973525873,
      "grad_norm": 216.46514892578125,
      "learning_rate": 0.00019482724713624917,
      "loss": 7.363,
      "step": 90110
    },
    {
      "epoch": 10.844765342960288,
      "grad_norm": 110.07520294189453,
      "learning_rate": 0.00019482603937330737,
      "loss": 7.2864,
      "step": 90120
    },
    {
      "epoch": 10.845968712394706,
      "grad_norm": 158.97850036621094,
      "learning_rate": 0.00019482483147312886,
      "loss": 7.4325,
      "step": 90130
    },
    {
      "epoch": 10.847172081829122,
      "grad_norm": 111.11646270751953,
      "learning_rate": 0.00019482362343571546,
      "loss": 7.3309,
      "step": 90140
    },
    {
      "epoch": 10.848375451263538,
      "grad_norm": 58.595584869384766,
      "learning_rate": 0.00019482241526106884,
      "loss": 7.5218,
      "step": 90150
    },
    {
      "epoch": 10.849578820697955,
      "grad_norm": 233.2845458984375,
      "learning_rate": 0.00019482120694919077,
      "loss": 7.4931,
      "step": 90160
    },
    {
      "epoch": 10.85078219013237,
      "grad_norm": 456.6014709472656,
      "learning_rate": 0.000194819998500083,
      "loss": 7.4513,
      "step": 90170
    },
    {
      "epoch": 10.851985559566787,
      "grad_norm": 902.7557373046875,
      "learning_rate": 0.0001948187899137473,
      "loss": 7.383,
      "step": 90180
    },
    {
      "epoch": 10.853188929001202,
      "grad_norm": 72.69772338867188,
      "learning_rate": 0.00019481758119018538,
      "loss": 7.4418,
      "step": 90190
    },
    {
      "epoch": 10.85439229843562,
      "grad_norm": 60.02667999267578,
      "learning_rate": 0.00019481637232939903,
      "loss": 7.3432,
      "step": 90200
    },
    {
      "epoch": 10.855595667870036,
      "grad_norm": 18.164566040039062,
      "learning_rate": 0.00019481516333138998,
      "loss": 7.3786,
      "step": 90210
    },
    {
      "epoch": 10.856799037304452,
      "grad_norm": 11.206013679504395,
      "learning_rate": 0.00019481395419615994,
      "loss": 7.3362,
      "step": 90220
    },
    {
      "epoch": 10.85800240673887,
      "grad_norm": 8.665871620178223,
      "learning_rate": 0.00019481274492371076,
      "loss": 7.3161,
      "step": 90230
    },
    {
      "epoch": 10.859205776173285,
      "grad_norm": 2.5264387130737305,
      "learning_rate": 0.0001948115355140441,
      "loss": 7.3466,
      "step": 90240
    },
    {
      "epoch": 10.8604091456077,
      "grad_norm": 20.93906021118164,
      "learning_rate": 0.00019481032596716172,
      "loss": 7.3447,
      "step": 90250
    },
    {
      "epoch": 10.861612515042118,
      "grad_norm": 6.975417613983154,
      "learning_rate": 0.00019480911628306542,
      "loss": 7.5147,
      "step": 90260
    },
    {
      "epoch": 10.862815884476534,
      "grad_norm": 2.2918078899383545,
      "learning_rate": 0.0001948079064617569,
      "loss": 7.3855,
      "step": 90270
    },
    {
      "epoch": 10.86401925391095,
      "grad_norm": 9.021227836608887,
      "learning_rate": 0.00019480669650323798,
      "loss": 7.3852,
      "step": 90280
    },
    {
      "epoch": 10.865222623345367,
      "grad_norm": 5.137051582336426,
      "learning_rate": 0.00019480548640751031,
      "loss": 7.5135,
      "step": 90290
    },
    {
      "epoch": 10.866425992779783,
      "grad_norm": 2.2990434169769287,
      "learning_rate": 0.00019480427617457572,
      "loss": 7.4849,
      "step": 90300
    },
    {
      "epoch": 10.867629362214199,
      "grad_norm": 24.73467445373535,
      "learning_rate": 0.00019480306580443591,
      "loss": 7.3778,
      "step": 90310
    },
    {
      "epoch": 10.868832731648617,
      "grad_norm": 5.181796073913574,
      "learning_rate": 0.0001948018552970927,
      "loss": 7.4085,
      "step": 90320
    },
    {
      "epoch": 10.870036101083032,
      "grad_norm": 4.720545768737793,
      "learning_rate": 0.00019480064465254778,
      "loss": 7.309,
      "step": 90330
    },
    {
      "epoch": 10.871239470517448,
      "grad_norm": 1.8978151082992554,
      "learning_rate": 0.0001947994338708029,
      "loss": 7.3723,
      "step": 90340
    },
    {
      "epoch": 10.872442839951866,
      "grad_norm": 6.551721096038818,
      "learning_rate": 0.00019479822295185984,
      "loss": 7.4731,
      "step": 90350
    },
    {
      "epoch": 10.873646209386282,
      "grad_norm": 2.412587881088257,
      "learning_rate": 0.00019479701189572038,
      "loss": 7.4649,
      "step": 90360
    },
    {
      "epoch": 10.874849578820697,
      "grad_norm": 4.760385513305664,
      "learning_rate": 0.0001947958007023862,
      "loss": 7.4933,
      "step": 90370
    },
    {
      "epoch": 10.876052948255115,
      "grad_norm": 5.1067585945129395,
      "learning_rate": 0.0001947945893718591,
      "loss": 7.4522,
      "step": 90380
    },
    {
      "epoch": 10.87725631768953,
      "grad_norm": 2.405296802520752,
      "learning_rate": 0.00019479337790414086,
      "loss": 7.4288,
      "step": 90390
    },
    {
      "epoch": 10.878459687123947,
      "grad_norm": 4.424656867980957,
      "learning_rate": 0.00019479216629923315,
      "loss": 7.3689,
      "step": 90400
    },
    {
      "epoch": 10.879663056558364,
      "grad_norm": 4.283668041229248,
      "learning_rate": 0.0001947909545571378,
      "loss": 7.417,
      "step": 90410
    },
    {
      "epoch": 10.88086642599278,
      "grad_norm": 2.591250419616699,
      "learning_rate": 0.00019478974267785653,
      "loss": 7.4594,
      "step": 90420
    },
    {
      "epoch": 10.882069795427196,
      "grad_norm": 3.5165891647338867,
      "learning_rate": 0.00019478853066139112,
      "loss": 7.5014,
      "step": 90430
    },
    {
      "epoch": 10.883273164861613,
      "grad_norm": 4.126473426818848,
      "learning_rate": 0.00019478731850774324,
      "loss": 7.4986,
      "step": 90440
    },
    {
      "epoch": 10.884476534296029,
      "grad_norm": 8.768253326416016,
      "learning_rate": 0.00019478610621691478,
      "loss": 7.4554,
      "step": 90450
    },
    {
      "epoch": 10.885679903730445,
      "grad_norm": 4.767611503601074,
      "learning_rate": 0.0001947848937889074,
      "loss": 7.3456,
      "step": 90460
    },
    {
      "epoch": 10.886883273164862,
      "grad_norm": 3.0895988941192627,
      "learning_rate": 0.00019478368122372285,
      "loss": 7.4833,
      "step": 90470
    },
    {
      "epoch": 10.888086642599278,
      "grad_norm": 4.713790416717529,
      "learning_rate": 0.00019478246852136293,
      "loss": 7.3336,
      "step": 90480
    },
    {
      "epoch": 10.889290012033694,
      "grad_norm": 2.3233847618103027,
      "learning_rate": 0.00019478125568182938,
      "loss": 7.4725,
      "step": 90490
    },
    {
      "epoch": 10.890493381468112,
      "grad_norm": 3.5966014862060547,
      "learning_rate": 0.00019478004270512394,
      "loss": 7.4315,
      "step": 90500
    },
    {
      "epoch": 10.891696750902527,
      "grad_norm": 3.502847909927368,
      "learning_rate": 0.0001947788295912484,
      "loss": 7.4436,
      "step": 90510
    },
    {
      "epoch": 10.892900120336943,
      "grad_norm": 5.073575496673584,
      "learning_rate": 0.0001947776163402045,
      "loss": 7.3108,
      "step": 90520
    },
    {
      "epoch": 10.89410348977136,
      "grad_norm": 2.306201457977295,
      "learning_rate": 0.00019477640295199398,
      "loss": 7.3425,
      "step": 90530
    },
    {
      "epoch": 10.895306859205776,
      "grad_norm": 1.9735738039016724,
      "learning_rate": 0.00019477518942661857,
      "loss": 7.3234,
      "step": 90540
    },
    {
      "epoch": 10.896510228640192,
      "grad_norm": 4.6471638679504395,
      "learning_rate": 0.0001947739757640801,
      "loss": 7.4421,
      "step": 90550
    },
    {
      "epoch": 10.89771359807461,
      "grad_norm": 2.2199184894561768,
      "learning_rate": 0.00019477276196438027,
      "loss": 7.3728,
      "step": 90560
    },
    {
      "epoch": 10.898916967509026,
      "grad_norm": 2.72428822517395,
      "learning_rate": 0.00019477154802752087,
      "loss": 7.3408,
      "step": 90570
    },
    {
      "epoch": 10.900120336943441,
      "grad_norm": 3.1605260372161865,
      "learning_rate": 0.00019477033395350364,
      "loss": 7.3783,
      "step": 90580
    },
    {
      "epoch": 10.901323706377857,
      "grad_norm": 2.2442264556884766,
      "learning_rate": 0.00019476911974233033,
      "loss": 7.3796,
      "step": 90590
    },
    {
      "epoch": 10.902527075812275,
      "grad_norm": 5.595448017120361,
      "learning_rate": 0.0001947679053940027,
      "loss": 7.3417,
      "step": 90600
    },
    {
      "epoch": 10.90373044524669,
      "grad_norm": 3.555079936981201,
      "learning_rate": 0.00019476669090852253,
      "loss": 7.3649,
      "step": 90610
    },
    {
      "epoch": 10.904933814681106,
      "grad_norm": 3.3862462043762207,
      "learning_rate": 0.00019476547628589154,
      "loss": 7.3889,
      "step": 90620
    },
    {
      "epoch": 10.906137184115524,
      "grad_norm": 3.648141622543335,
      "learning_rate": 0.00019476426152611152,
      "loss": 7.3131,
      "step": 90630
    },
    {
      "epoch": 10.90734055354994,
      "grad_norm": 3.0125887393951416,
      "learning_rate": 0.00019476304662918424,
      "loss": 7.3405,
      "step": 90640
    },
    {
      "epoch": 10.908543922984355,
      "grad_norm": 13.056954383850098,
      "learning_rate": 0.0001947618315951114,
      "loss": 7.3561,
      "step": 90650
    },
    {
      "epoch": 10.909747292418773,
      "grad_norm": 4.784685134887695,
      "learning_rate": 0.0001947606164238948,
      "loss": 7.4375,
      "step": 90660
    },
    {
      "epoch": 10.910950661853189,
      "grad_norm": 2.6147332191467285,
      "learning_rate": 0.0001947594011155362,
      "loss": 7.3464,
      "step": 90670
    },
    {
      "epoch": 10.912154031287605,
      "grad_norm": 7.623844623565674,
      "learning_rate": 0.00019475818567003735,
      "loss": 7.3462,
      "step": 90680
    },
    {
      "epoch": 10.913357400722022,
      "grad_norm": 4.65337610244751,
      "learning_rate": 0.0001947569700874,
      "loss": 7.3075,
      "step": 90690
    },
    {
      "epoch": 10.914560770156438,
      "grad_norm": 7.81140661239624,
      "learning_rate": 0.00019475575436762594,
      "loss": 7.385,
      "step": 90700
    },
    {
      "epoch": 10.915764139590854,
      "grad_norm": 7.787203311920166,
      "learning_rate": 0.00019475453851071692,
      "loss": 7.4105,
      "step": 90710
    },
    {
      "epoch": 10.916967509025271,
      "grad_norm": 9.700875282287598,
      "learning_rate": 0.00019475332251667467,
      "loss": 7.2566,
      "step": 90720
    },
    {
      "epoch": 10.918170878459687,
      "grad_norm": 5.274430751800537,
      "learning_rate": 0.00019475210638550094,
      "loss": 7.3838,
      "step": 90730
    },
    {
      "epoch": 10.919374247894103,
      "grad_norm": 5.20965576171875,
      "learning_rate": 0.00019475089011719755,
      "loss": 7.3184,
      "step": 90740
    },
    {
      "epoch": 10.92057761732852,
      "grad_norm": 20.575435638427734,
      "learning_rate": 0.00019474967371176623,
      "loss": 7.2909,
      "step": 90750
    },
    {
      "epoch": 10.921780986762936,
      "grad_norm": 4.916385173797607,
      "learning_rate": 0.00019474845716920873,
      "loss": 7.3813,
      "step": 90760
    },
    {
      "epoch": 10.922984356197352,
      "grad_norm": 5.026564598083496,
      "learning_rate": 0.00019474724048952684,
      "loss": 7.2943,
      "step": 90770
    },
    {
      "epoch": 10.92418772563177,
      "grad_norm": 9.916465759277344,
      "learning_rate": 0.00019474602367272227,
      "loss": 7.2248,
      "step": 90780
    },
    {
      "epoch": 10.925391095066185,
      "grad_norm": 6.919652938842773,
      "learning_rate": 0.00019474480671879683,
      "loss": 7.3152,
      "step": 90790
    },
    {
      "epoch": 10.926594464500601,
      "grad_norm": 8.888089179992676,
      "learning_rate": 0.00019474358962775227,
      "loss": 7.2641,
      "step": 90800
    },
    {
      "epoch": 10.927797833935019,
      "grad_norm": 18.766050338745117,
      "learning_rate": 0.00019474237239959033,
      "loss": 7.3283,
      "step": 90810
    },
    {
      "epoch": 10.929001203369435,
      "grad_norm": 14.107617378234863,
      "learning_rate": 0.0001947411550343128,
      "loss": 7.2375,
      "step": 90820
    },
    {
      "epoch": 10.93020457280385,
      "grad_norm": 18.338733673095703,
      "learning_rate": 0.00019473993753192138,
      "loss": 7.1868,
      "step": 90830
    },
    {
      "epoch": 10.931407942238268,
      "grad_norm": 9.514047622680664,
      "learning_rate": 0.00019473871989241794,
      "loss": 7.3133,
      "step": 90840
    },
    {
      "epoch": 10.932611311672684,
      "grad_norm": 11.194342613220215,
      "learning_rate": 0.00019473750211580418,
      "loss": 7.2377,
      "step": 90850
    },
    {
      "epoch": 10.9338146811071,
      "grad_norm": 11.65888500213623,
      "learning_rate": 0.0001947362842020818,
      "loss": 7.183,
      "step": 90860
    },
    {
      "epoch": 10.935018050541515,
      "grad_norm": 12.573893547058105,
      "learning_rate": 0.0001947350661512527,
      "loss": 7.2272,
      "step": 90870
    },
    {
      "epoch": 10.936221419975933,
      "grad_norm": 18.41805648803711,
      "learning_rate": 0.00019473384796331853,
      "loss": 7.2195,
      "step": 90880
    },
    {
      "epoch": 10.937424789410349,
      "grad_norm": 10.399399757385254,
      "learning_rate": 0.0001947326296382811,
      "loss": 7.1829,
      "step": 90890
    },
    {
      "epoch": 10.938628158844764,
      "grad_norm": 27.217512130737305,
      "learning_rate": 0.00019473141117614217,
      "loss": 7.2757,
      "step": 90900
    },
    {
      "epoch": 10.939831528279182,
      "grad_norm": 24.504770278930664,
      "learning_rate": 0.0001947301925769035,
      "loss": 7.2734,
      "step": 90910
    },
    {
      "epoch": 10.941034897713598,
      "grad_norm": 15.4624662399292,
      "learning_rate": 0.00019472897384056686,
      "loss": 7.267,
      "step": 90920
    },
    {
      "epoch": 10.942238267148014,
      "grad_norm": 23.55158233642578,
      "learning_rate": 0.00019472775496713398,
      "loss": 7.154,
      "step": 90930
    },
    {
      "epoch": 10.943441636582431,
      "grad_norm": 23.233240127563477,
      "learning_rate": 0.00019472653595660667,
      "loss": 7.3416,
      "step": 90940
    },
    {
      "epoch": 10.944645006016847,
      "grad_norm": 21.710960388183594,
      "learning_rate": 0.00019472531680898668,
      "loss": 7.1898,
      "step": 90950
    },
    {
      "epoch": 10.945848375451263,
      "grad_norm": 17.23990821838379,
      "learning_rate": 0.00019472409752427578,
      "loss": 7.2277,
      "step": 90960
    },
    {
      "epoch": 10.94705174488568,
      "grad_norm": 8.576688766479492,
      "learning_rate": 0.00019472287810247573,
      "loss": 7.2173,
      "step": 90970
    },
    {
      "epoch": 10.948255114320096,
      "grad_norm": 22.814760208129883,
      "learning_rate": 0.00019472165854358825,
      "loss": 7.3606,
      "step": 90980
    },
    {
      "epoch": 10.949458483754512,
      "grad_norm": 13.988065719604492,
      "learning_rate": 0.00019472043884761515,
      "loss": 7.1819,
      "step": 90990
    },
    {
      "epoch": 10.95066185318893,
      "grad_norm": 22.70740509033203,
      "learning_rate": 0.00019471921901455822,
      "loss": 7.3582,
      "step": 91000
    },
    {
      "epoch": 10.951865222623345,
      "grad_norm": 19.532262802124023,
      "learning_rate": 0.00019471799904441917,
      "loss": 7.1927,
      "step": 91010
    },
    {
      "epoch": 10.953068592057761,
      "grad_norm": 14.020984649658203,
      "learning_rate": 0.00019471677893719979,
      "loss": 7.3016,
      "step": 91020
    },
    {
      "epoch": 10.954271961492179,
      "grad_norm": 13.995887756347656,
      "learning_rate": 0.00019471555869290185,
      "loss": 7.1971,
      "step": 91030
    },
    {
      "epoch": 10.955475330926594,
      "grad_norm": 19.562938690185547,
      "learning_rate": 0.00019471433831152711,
      "loss": 7.2804,
      "step": 91040
    },
    {
      "epoch": 10.95667870036101,
      "grad_norm": 18.86977767944336,
      "learning_rate": 0.00019471311779307735,
      "loss": 7.3369,
      "step": 91050
    },
    {
      "epoch": 10.957882069795428,
      "grad_norm": 27.458768844604492,
      "learning_rate": 0.00019471189713755432,
      "loss": 7.2666,
      "step": 91060
    },
    {
      "epoch": 10.959085439229844,
      "grad_norm": 7.772292137145996,
      "learning_rate": 0.0001947106763449598,
      "loss": 7.2552,
      "step": 91070
    },
    {
      "epoch": 10.96028880866426,
      "grad_norm": 29.27229118347168,
      "learning_rate": 0.00019470945541529556,
      "loss": 7.2156,
      "step": 91080
    },
    {
      "epoch": 10.961492178098677,
      "grad_norm": 9.601334571838379,
      "learning_rate": 0.0001947082343485633,
      "loss": 7.1959,
      "step": 91090
    },
    {
      "epoch": 10.962695547533093,
      "grad_norm": 82.26222229003906,
      "learning_rate": 0.0001947070131447649,
      "loss": 7.2415,
      "step": 91100
    },
    {
      "epoch": 10.963898916967509,
      "grad_norm": 16.596284866333008,
      "learning_rate": 0.00019470579180390203,
      "loss": 7.2281,
      "step": 91110
    },
    {
      "epoch": 10.965102286401926,
      "grad_norm": 12.260496139526367,
      "learning_rate": 0.00019470457032597652,
      "loss": 7.2486,
      "step": 91120
    },
    {
      "epoch": 10.966305655836342,
      "grad_norm": 19.594247817993164,
      "learning_rate": 0.00019470334871099008,
      "loss": 7.1312,
      "step": 91130
    },
    {
      "epoch": 10.967509025270758,
      "grad_norm": 29.621244430541992,
      "learning_rate": 0.00019470212695894454,
      "loss": 7.1589,
      "step": 91140
    },
    {
      "epoch": 10.968712394705175,
      "grad_norm": 52.26381301879883,
      "learning_rate": 0.00019470090506984165,
      "loss": 7.269,
      "step": 91150
    },
    {
      "epoch": 10.969915764139591,
      "grad_norm": 44.431034088134766,
      "learning_rate": 0.00019469968304368317,
      "loss": 7.2556,
      "step": 91160
    },
    {
      "epoch": 10.971119133574007,
      "grad_norm": 10.630620002746582,
      "learning_rate": 0.00019469846088047084,
      "loss": 7.2296,
      "step": 91170
    },
    {
      "epoch": 10.972322503008424,
      "grad_norm": 50.92340850830078,
      "learning_rate": 0.00019469723858020648,
      "loss": 7.1534,
      "step": 91180
    },
    {
      "epoch": 10.97352587244284,
      "grad_norm": 14.209988594055176,
      "learning_rate": 0.00019469601614289183,
      "loss": 7.3142,
      "step": 91190
    },
    {
      "epoch": 10.974729241877256,
      "grad_norm": 36.25379943847656,
      "learning_rate": 0.00019469479356852866,
      "loss": 7.2598,
      "step": 91200
    },
    {
      "epoch": 10.975932611311674,
      "grad_norm": 36.42372131347656,
      "learning_rate": 0.00019469357085711875,
      "loss": 7.1667,
      "step": 91210
    },
    {
      "epoch": 10.97713598074609,
      "grad_norm": 15.982621192932129,
      "learning_rate": 0.00019469234800866388,
      "loss": 7.3051,
      "step": 91220
    },
    {
      "epoch": 10.978339350180505,
      "grad_norm": 35.26732635498047,
      "learning_rate": 0.00019469112502316577,
      "loss": 7.2107,
      "step": 91230
    },
    {
      "epoch": 10.979542719614923,
      "grad_norm": 110.99844360351562,
      "learning_rate": 0.00019468990190062626,
      "loss": 7.2238,
      "step": 91240
    },
    {
      "epoch": 10.980746089049338,
      "grad_norm": 76.19429016113281,
      "learning_rate": 0.00019468867864104705,
      "loss": 7.1451,
      "step": 91250
    },
    {
      "epoch": 10.981949458483754,
      "grad_norm": 59.430763244628906,
      "learning_rate": 0.00019468745524442996,
      "loss": 7.2429,
      "step": 91260
    },
    {
      "epoch": 10.98315282791817,
      "grad_norm": 42.25979995727539,
      "learning_rate": 0.00019468623171077674,
      "loss": 7.1314,
      "step": 91270
    },
    {
      "epoch": 10.984356197352588,
      "grad_norm": 90.5514144897461,
      "learning_rate": 0.00019468500804008918,
      "loss": 7.2055,
      "step": 91280
    },
    {
      "epoch": 10.985559566787003,
      "grad_norm": 17.978731155395508,
      "learning_rate": 0.00019468378423236903,
      "loss": 7.3261,
      "step": 91290
    },
    {
      "epoch": 10.98676293622142,
      "grad_norm": 53.58927917480469,
      "learning_rate": 0.00019468256028761806,
      "loss": 7.1953,
      "step": 91300
    },
    {
      "epoch": 10.987966305655837,
      "grad_norm": 53.382686614990234,
      "learning_rate": 0.00019468133620583803,
      "loss": 7.1197,
      "step": 91310
    },
    {
      "epoch": 10.989169675090253,
      "grad_norm": 16.70414161682129,
      "learning_rate": 0.00019468011198703079,
      "loss": 7.3798,
      "step": 91320
    },
    {
      "epoch": 10.990373044524668,
      "grad_norm": 54.25605392456055,
      "learning_rate": 0.000194678887631198,
      "loss": 7.2005,
      "step": 91330
    },
    {
      "epoch": 10.991576413959086,
      "grad_norm": 36.17827224731445,
      "learning_rate": 0.00019467766313834147,
      "loss": 7.1786,
      "step": 91340
    },
    {
      "epoch": 10.992779783393502,
      "grad_norm": 197.1502227783203,
      "learning_rate": 0.00019467643850846304,
      "loss": 7.1718,
      "step": 91350
    },
    {
      "epoch": 10.993983152827917,
      "grad_norm": 31.615577697753906,
      "learning_rate": 0.00019467521374156438,
      "loss": 7.2492,
      "step": 91360
    },
    {
      "epoch": 10.995186522262335,
      "grad_norm": 27.217073440551758,
      "learning_rate": 0.00019467398883764734,
      "loss": 7.1934,
      "step": 91370
    },
    {
      "epoch": 10.99638989169675,
      "grad_norm": 32.41168212890625,
      "learning_rate": 0.00019467276379671366,
      "loss": 7.0922,
      "step": 91380
    },
    {
      "epoch": 10.997593261131167,
      "grad_norm": 35.65819549560547,
      "learning_rate": 0.00019467153861876512,
      "loss": 7.1939,
      "step": 91390
    },
    {
      "epoch": 10.998796630565584,
      "grad_norm": 68.97451782226562,
      "learning_rate": 0.00019467031330380348,
      "loss": 7.203,
      "step": 91400
    },
    {
      "epoch": 11.0,
      "grad_norm": 52.4002685546875,
      "learning_rate": 0.00019466908785183056,
      "loss": 7.1735,
      "step": 91410
    },
    {
      "epoch": 11.0,
      "eval_loss": 7.210310459136963,
      "eval_runtime": 119.6216,
      "eval_samples_per_second": 61.753,
      "eval_steps_per_second": 7.724,
      "step": 91410
    },
    {
      "epoch": 11.001203369434416,
      "grad_norm": 88.97228240966797,
      "learning_rate": 0.00019466786226284802,
      "loss": 7.0587,
      "step": 91420
    },
    {
      "epoch": 11.002406738868833,
      "grad_norm": 31.14643096923828,
      "learning_rate": 0.00019466663653685778,
      "loss": 7.1077,
      "step": 91430
    },
    {
      "epoch": 11.00361010830325,
      "grad_norm": 57.134159088134766,
      "learning_rate": 0.00019466541067386152,
      "loss": 7.1847,
      "step": 91440
    },
    {
      "epoch": 11.004813477737665,
      "grad_norm": 23.970157623291016,
      "learning_rate": 0.00019466418467386105,
      "loss": 7.1429,
      "step": 91450
    },
    {
      "epoch": 11.006016847172083,
      "grad_norm": 87.09337615966797,
      "learning_rate": 0.0001946629585368581,
      "loss": 7.2082,
      "step": 91460
    },
    {
      "epoch": 11.007220216606498,
      "grad_norm": 78.77494049072266,
      "learning_rate": 0.00019466173226285452,
      "loss": 7.1348,
      "step": 91470
    },
    {
      "epoch": 11.008423586040914,
      "grad_norm": 43.038944244384766,
      "learning_rate": 0.00019466050585185203,
      "loss": 7.2466,
      "step": 91480
    },
    {
      "epoch": 11.009626955475332,
      "grad_norm": 19.317781448364258,
      "learning_rate": 0.0001946592793038524,
      "loss": 7.1249,
      "step": 91490
    },
    {
      "epoch": 11.010830324909747,
      "grad_norm": 47.15719223022461,
      "learning_rate": 0.0001946580526188574,
      "loss": 7.1508,
      "step": 91500
    },
    {
      "epoch": 11.012033694344163,
      "grad_norm": 45.80705642700195,
      "learning_rate": 0.00019465682579686888,
      "loss": 7.1475,
      "step": 91510
    },
    {
      "epoch": 11.01323706377858,
      "grad_norm": 34.837158203125,
      "learning_rate": 0.00019465559883788855,
      "loss": 7.2702,
      "step": 91520
    },
    {
      "epoch": 11.014440433212997,
      "grad_norm": 24.252323150634766,
      "learning_rate": 0.0001946543717419182,
      "loss": 7.1592,
      "step": 91530
    },
    {
      "epoch": 11.015643802647412,
      "grad_norm": 108.92929077148438,
      "learning_rate": 0.00019465314450895958,
      "loss": 7.1675,
      "step": 91540
    },
    {
      "epoch": 11.01684717208183,
      "grad_norm": 58.81249237060547,
      "learning_rate": 0.0001946519171390145,
      "loss": 7.1503,
      "step": 91550
    },
    {
      "epoch": 11.018050541516246,
      "grad_norm": 92.31916809082031,
      "learning_rate": 0.00019465068963208472,
      "loss": 7.2044,
      "step": 91560
    },
    {
      "epoch": 11.019253910950662,
      "grad_norm": 60.861717224121094,
      "learning_rate": 0.00019464946198817206,
      "loss": 7.28,
      "step": 91570
    },
    {
      "epoch": 11.020457280385079,
      "grad_norm": 34.50305938720703,
      "learning_rate": 0.00019464823420727823,
      "loss": 7.198,
      "step": 91580
    },
    {
      "epoch": 11.021660649819495,
      "grad_norm": 37.99361801147461,
      "learning_rate": 0.00019464700628940505,
      "loss": 7.2525,
      "step": 91590
    },
    {
      "epoch": 11.02286401925391,
      "grad_norm": 63.08331298828125,
      "learning_rate": 0.00019464577823455426,
      "loss": 7.1535,
      "step": 91600
    },
    {
      "epoch": 11.024067388688326,
      "grad_norm": 106.97830963134766,
      "learning_rate": 0.00019464455004272768,
      "loss": 7.1971,
      "step": 91610
    },
    {
      "epoch": 11.025270758122744,
      "grad_norm": 58.077816009521484,
      "learning_rate": 0.0001946433217139271,
      "loss": 7.1119,
      "step": 91620
    },
    {
      "epoch": 11.02647412755716,
      "grad_norm": 89.77117919921875,
      "learning_rate": 0.0001946420932481542,
      "loss": 7.1836,
      "step": 91630
    },
    {
      "epoch": 11.027677496991576,
      "grad_norm": 107.27037048339844,
      "learning_rate": 0.0001946408646454109,
      "loss": 7.1823,
      "step": 91640
    },
    {
      "epoch": 11.028880866425993,
      "grad_norm": 124.76302337646484,
      "learning_rate": 0.00019463963590569884,
      "loss": 7.2215,
      "step": 91650
    },
    {
      "epoch": 11.030084235860409,
      "grad_norm": 1184.3741455078125,
      "learning_rate": 0.00019463840702901988,
      "loss": 7.2242,
      "step": 91660
    },
    {
      "epoch": 11.031287605294825,
      "grad_norm": 253.3821258544922,
      "learning_rate": 0.00019463717801537577,
      "loss": 7.3282,
      "step": 91670
    },
    {
      "epoch": 11.032490974729242,
      "grad_norm": 155.58058166503906,
      "learning_rate": 0.00019463594886476836,
      "loss": 7.2619,
      "step": 91680
    },
    {
      "epoch": 11.033694344163658,
      "grad_norm": 281.6895751953125,
      "learning_rate": 0.00019463471957719927,
      "loss": 7.1515,
      "step": 91690
    },
    {
      "epoch": 11.034897713598074,
      "grad_norm": 182.16835021972656,
      "learning_rate": 0.00019463349015267045,
      "loss": 7.2305,
      "step": 91700
    },
    {
      "epoch": 11.036101083032491,
      "grad_norm": 102.86280822753906,
      "learning_rate": 0.0001946322605911836,
      "loss": 7.1943,
      "step": 91710
    },
    {
      "epoch": 11.037304452466907,
      "grad_norm": 72.52764892578125,
      "learning_rate": 0.00019463103089274047,
      "loss": 7.212,
      "step": 91720
    },
    {
      "epoch": 11.038507821901323,
      "grad_norm": 187.4462890625,
      "learning_rate": 0.0001946298010573429,
      "loss": 7.1589,
      "step": 91730
    },
    {
      "epoch": 11.03971119133574,
      "grad_norm": 91.68645477294922,
      "learning_rate": 0.00019462857108499263,
      "loss": 7.095,
      "step": 91740
    },
    {
      "epoch": 11.040914560770156,
      "grad_norm": 64.60287475585938,
      "learning_rate": 0.00019462734097569145,
      "loss": 7.1291,
      "step": 91750
    },
    {
      "epoch": 11.042117930204572,
      "grad_norm": 104.13249206542969,
      "learning_rate": 0.00019462611072944115,
      "loss": 7.1589,
      "step": 91760
    },
    {
      "epoch": 11.04332129963899,
      "grad_norm": 51.590476989746094,
      "learning_rate": 0.0001946248803462435,
      "loss": 7.1153,
      "step": 91770
    },
    {
      "epoch": 11.044524669073406,
      "grad_norm": 89.51387023925781,
      "learning_rate": 0.00019462364982610032,
      "loss": 7.2722,
      "step": 91780
    },
    {
      "epoch": 11.045728038507821,
      "grad_norm": 149.9066619873047,
      "learning_rate": 0.0001946224191690133,
      "loss": 7.1074,
      "step": 91790
    },
    {
      "epoch": 11.046931407942239,
      "grad_norm": 85.96741485595703,
      "learning_rate": 0.00019462118837498433,
      "loss": 7.165,
      "step": 91800
    },
    {
      "epoch": 11.048134777376655,
      "grad_norm": 389.3432312011719,
      "learning_rate": 0.00019461995744401512,
      "loss": 7.2343,
      "step": 91810
    },
    {
      "epoch": 11.04933814681107,
      "grad_norm": 109.79691314697266,
      "learning_rate": 0.00019461872637610748,
      "loss": 7.2251,
      "step": 91820
    },
    {
      "epoch": 11.050541516245488,
      "grad_norm": 67.6067123413086,
      "learning_rate": 0.00019461749517126316,
      "loss": 7.2376,
      "step": 91830
    },
    {
      "epoch": 11.051744885679904,
      "grad_norm": 136.72100830078125,
      "learning_rate": 0.000194616263829484,
      "loss": 7.2126,
      "step": 91840
    },
    {
      "epoch": 11.05294825511432,
      "grad_norm": 104.37311553955078,
      "learning_rate": 0.0001946150323507717,
      "loss": 7.204,
      "step": 91850
    },
    {
      "epoch": 11.054151624548737,
      "grad_norm": 229.95697021484375,
      "learning_rate": 0.00019461380073512813,
      "loss": 7.2275,
      "step": 91860
    },
    {
      "epoch": 11.055354993983153,
      "grad_norm": 110.85474395751953,
      "learning_rate": 0.000194612568982555,
      "loss": 7.2068,
      "step": 91870
    },
    {
      "epoch": 11.056558363417569,
      "grad_norm": 154.43856811523438,
      "learning_rate": 0.00019461133709305415,
      "loss": 7.1856,
      "step": 91880
    },
    {
      "epoch": 11.057761732851986,
      "grad_norm": 83.03573608398438,
      "learning_rate": 0.0001946101050666273,
      "loss": 7.1952,
      "step": 91890
    },
    {
      "epoch": 11.058965102286402,
      "grad_norm": 63.278724670410156,
      "learning_rate": 0.00019460887290327633,
      "loss": 7.2036,
      "step": 91900
    },
    {
      "epoch": 11.060168471720818,
      "grad_norm": 326.2492370605469,
      "learning_rate": 0.0001946076406030029,
      "loss": 7.209,
      "step": 91910
    },
    {
      "epoch": 11.061371841155236,
      "grad_norm": 133.4531707763672,
      "learning_rate": 0.00019460640816580888,
      "loss": 7.1174,
      "step": 91920
    },
    {
      "epoch": 11.062575210589651,
      "grad_norm": 59.3718147277832,
      "learning_rate": 0.00019460517559169603,
      "loss": 7.1888,
      "step": 91930
    },
    {
      "epoch": 11.063778580024067,
      "grad_norm": 418.29290771484375,
      "learning_rate": 0.00019460394288066613,
      "loss": 7.147,
      "step": 91940
    },
    {
      "epoch": 11.064981949458483,
      "grad_norm": 51.344417572021484,
      "learning_rate": 0.00019460271003272098,
      "loss": 7.1793,
      "step": 91950
    },
    {
      "epoch": 11.0661853188929,
      "grad_norm": 123.91556549072266,
      "learning_rate": 0.00019460147704786237,
      "loss": 7.2084,
      "step": 91960
    },
    {
      "epoch": 11.067388688327316,
      "grad_norm": 121.50829315185547,
      "learning_rate": 0.000194600243926092,
      "loss": 7.1258,
      "step": 91970
    },
    {
      "epoch": 11.068592057761732,
      "grad_norm": 125.03963470458984,
      "learning_rate": 0.00019459901066741178,
      "loss": 7.2939,
      "step": 91980
    },
    {
      "epoch": 11.06979542719615,
      "grad_norm": 142.17739868164062,
      "learning_rate": 0.00019459777727182342,
      "loss": 7.2213,
      "step": 91990
    },
    {
      "epoch": 11.070998796630565,
      "grad_norm": 94.00983428955078,
      "learning_rate": 0.00019459654373932871,
      "loss": 7.3204,
      "step": 92000
    },
    {
      "epoch": 11.072202166064981,
      "grad_norm": 88.48886108398438,
      "learning_rate": 0.00019459531006992947,
      "loss": 7.1866,
      "step": 92010
    },
    {
      "epoch": 11.073405535499399,
      "grad_norm": 127.03926849365234,
      "learning_rate": 0.00019459407626362744,
      "loss": 7.1556,
      "step": 92020
    },
    {
      "epoch": 11.074608904933815,
      "grad_norm": 45.86471176147461,
      "learning_rate": 0.00019459284232042443,
      "loss": 7.2194,
      "step": 92030
    },
    {
      "epoch": 11.07581227436823,
      "grad_norm": 164.46554565429688,
      "learning_rate": 0.0001945916082403222,
      "loss": 7.1532,
      "step": 92040
    },
    {
      "epoch": 11.077015643802648,
      "grad_norm": 212.33151245117188,
      "learning_rate": 0.00019459037402332261,
      "loss": 7.2713,
      "step": 92050
    },
    {
      "epoch": 11.078219013237064,
      "grad_norm": 58.923831939697266,
      "learning_rate": 0.00019458913966942737,
      "loss": 7.2077,
      "step": 92060
    },
    {
      "epoch": 11.07942238267148,
      "grad_norm": 66.29691314697266,
      "learning_rate": 0.0001945879051786383,
      "loss": 7.1387,
      "step": 92070
    },
    {
      "epoch": 11.080625752105897,
      "grad_norm": 48.843936920166016,
      "learning_rate": 0.00019458667055095716,
      "loss": 7.2679,
      "step": 92080
    },
    {
      "epoch": 11.081829121540313,
      "grad_norm": 42.04731369018555,
      "learning_rate": 0.00019458543578638579,
      "loss": 7.1973,
      "step": 92090
    },
    {
      "epoch": 11.083032490974729,
      "grad_norm": 94.57808685302734,
      "learning_rate": 0.00019458420088492591,
      "loss": 7.1809,
      "step": 92100
    },
    {
      "epoch": 11.084235860409146,
      "grad_norm": 196.99717712402344,
      "learning_rate": 0.00019458296584657934,
      "loss": 7.1951,
      "step": 92110
    },
    {
      "epoch": 11.085439229843562,
      "grad_norm": 124.04299926757812,
      "learning_rate": 0.00019458173067134787,
      "loss": 7.1782,
      "step": 92120
    },
    {
      "epoch": 11.086642599277978,
      "grad_norm": 51.40081024169922,
      "learning_rate": 0.00019458049535923328,
      "loss": 7.2055,
      "step": 92130
    },
    {
      "epoch": 11.087845968712395,
      "grad_norm": 96.16692352294922,
      "learning_rate": 0.00019457925991023736,
      "loss": 7.124,
      "step": 92140
    },
    {
      "epoch": 11.089049338146811,
      "grad_norm": 64.81087493896484,
      "learning_rate": 0.0001945780243243619,
      "loss": 7.1218,
      "step": 92150
    },
    {
      "epoch": 11.090252707581227,
      "grad_norm": 60.572288513183594,
      "learning_rate": 0.00019457678860160873,
      "loss": 7.2589,
      "step": 92160
    },
    {
      "epoch": 11.091456077015645,
      "grad_norm": 50.32078170776367,
      "learning_rate": 0.00019457555274197957,
      "loss": 7.1335,
      "step": 92170
    },
    {
      "epoch": 11.09265944645006,
      "grad_norm": 27.044208526611328,
      "learning_rate": 0.0001945743167454762,
      "loss": 7.219,
      "step": 92180
    },
    {
      "epoch": 11.093862815884476,
      "grad_norm": 138.8689422607422,
      "learning_rate": 0.00019457308061210049,
      "loss": 7.2221,
      "step": 92190
    },
    {
      "epoch": 11.095066185318894,
      "grad_norm": 25.8515682220459,
      "learning_rate": 0.00019457184434185416,
      "loss": 7.0875,
      "step": 92200
    },
    {
      "epoch": 11.09626955475331,
      "grad_norm": 50.15380859375,
      "learning_rate": 0.00019457060793473904,
      "loss": 7.2845,
      "step": 92210
    },
    {
      "epoch": 11.097472924187725,
      "grad_norm": 64.73393249511719,
      "learning_rate": 0.0001945693713907569,
      "loss": 7.1922,
      "step": 92220
    },
    {
      "epoch": 11.098676293622143,
      "grad_norm": 32.96710205078125,
      "learning_rate": 0.00019456813470990948,
      "loss": 7.1761,
      "step": 92230
    },
    {
      "epoch": 11.099879663056559,
      "grad_norm": 26.588193893432617,
      "learning_rate": 0.00019456689789219868,
      "loss": 7.2422,
      "step": 92240
    },
    {
      "epoch": 11.101083032490974,
      "grad_norm": 87.43901824951172,
      "learning_rate": 0.0001945656609376262,
      "loss": 7.243,
      "step": 92250
    },
    {
      "epoch": 11.102286401925392,
      "grad_norm": 43.802608489990234,
      "learning_rate": 0.0001945644238461939,
      "loss": 7.2159,
      "step": 92260
    },
    {
      "epoch": 11.103489771359808,
      "grad_norm": 79.26888275146484,
      "learning_rate": 0.0001945631866179035,
      "loss": 7.1176,
      "step": 92270
    },
    {
      "epoch": 11.104693140794224,
      "grad_norm": 61.184104919433594,
      "learning_rate": 0.00019456194925275683,
      "loss": 7.1025,
      "step": 92280
    },
    {
      "epoch": 11.10589651022864,
      "grad_norm": 90.06178283691406,
      "learning_rate": 0.00019456071175075564,
      "loss": 7.2312,
      "step": 92290
    },
    {
      "epoch": 11.107099879663057,
      "grad_norm": 78.83863830566406,
      "learning_rate": 0.0001945594741119018,
      "loss": 7.1369,
      "step": 92300
    },
    {
      "epoch": 11.108303249097473,
      "grad_norm": 249.10198974609375,
      "learning_rate": 0.00019455823633619706,
      "loss": 7.1614,
      "step": 92310
    },
    {
      "epoch": 11.109506618531888,
      "grad_norm": 81.26526641845703,
      "learning_rate": 0.00019455699842364319,
      "loss": 7.1176,
      "step": 92320
    },
    {
      "epoch": 11.110709987966306,
      "grad_norm": 56.093475341796875,
      "learning_rate": 0.00019455576037424196,
      "loss": 7.151,
      "step": 92330
    },
    {
      "epoch": 11.111913357400722,
      "grad_norm": 108.1319351196289,
      "learning_rate": 0.00019455452218799526,
      "loss": 7.1621,
      "step": 92340
    },
    {
      "epoch": 11.113116726835138,
      "grad_norm": 46.65833282470703,
      "learning_rate": 0.0001945532838649048,
      "loss": 7.1636,
      "step": 92350
    },
    {
      "epoch": 11.114320096269555,
      "grad_norm": 56.3868522644043,
      "learning_rate": 0.0001945520454049724,
      "loss": 7.1053,
      "step": 92360
    },
    {
      "epoch": 11.115523465703971,
      "grad_norm": 46.710933685302734,
      "learning_rate": 0.00019455080680819982,
      "loss": 7.1899,
      "step": 92370
    },
    {
      "epoch": 11.116726835138387,
      "grad_norm": 29.239248275756836,
      "learning_rate": 0.00019454956807458889,
      "loss": 7.1444,
      "step": 92380
    },
    {
      "epoch": 11.117930204572804,
      "grad_norm": 51.71601486206055,
      "learning_rate": 0.0001945483292041414,
      "loss": 7.1056,
      "step": 92390
    },
    {
      "epoch": 11.11913357400722,
      "grad_norm": 18.059215545654297,
      "learning_rate": 0.00019454709019685912,
      "loss": 7.1753,
      "step": 92400
    },
    {
      "epoch": 11.120336943441636,
      "grad_norm": 510.7772216796875,
      "learning_rate": 0.00019454585105274386,
      "loss": 7.2051,
      "step": 92410
    },
    {
      "epoch": 11.121540312876053,
      "grad_norm": 86.14214324951172,
      "learning_rate": 0.00019454461177179745,
      "loss": 7.2663,
      "step": 92420
    },
    {
      "epoch": 11.12274368231047,
      "grad_norm": 55.65878677368164,
      "learning_rate": 0.00019454337235402157,
      "loss": 7.2085,
      "step": 92430
    },
    {
      "epoch": 11.123947051744885,
      "grad_norm": 55.23828887939453,
      "learning_rate": 0.00019454213279941816,
      "loss": 7.1932,
      "step": 92440
    },
    {
      "epoch": 11.125150421179303,
      "grad_norm": 38.14577102661133,
      "learning_rate": 0.00019454089310798892,
      "loss": 7.2278,
      "step": 92450
    },
    {
      "epoch": 11.126353790613718,
      "grad_norm": 64.26361083984375,
      "learning_rate": 0.00019453965327973565,
      "loss": 7.0654,
      "step": 92460
    },
    {
      "epoch": 11.127557160048134,
      "grad_norm": 38.1277961730957,
      "learning_rate": 0.00019453841331466015,
      "loss": 7.2261,
      "step": 92470
    },
    {
      "epoch": 11.128760529482552,
      "grad_norm": 25.92961883544922,
      "learning_rate": 0.00019453717321276428,
      "loss": 7.19,
      "step": 92480
    },
    {
      "epoch": 11.129963898916968,
      "grad_norm": 30.9171085357666,
      "learning_rate": 0.00019453593297404974,
      "loss": 7.1756,
      "step": 92490
    },
    {
      "epoch": 11.131167268351383,
      "grad_norm": 23.84162712097168,
      "learning_rate": 0.00019453469259851836,
      "loss": 7.2164,
      "step": 92500
    },
    {
      "epoch": 11.132370637785801,
      "grad_norm": 64.93234252929688,
      "learning_rate": 0.00019453345208617195,
      "loss": 7.3184,
      "step": 92510
    },
    {
      "epoch": 11.133574007220217,
      "grad_norm": 39.70873260498047,
      "learning_rate": 0.00019453221143701227,
      "loss": 7.1402,
      "step": 92520
    },
    {
      "epoch": 11.134777376654633,
      "grad_norm": 53.39655303955078,
      "learning_rate": 0.0001945309706510412,
      "loss": 7.2207,
      "step": 92530
    },
    {
      "epoch": 11.13598074608905,
      "grad_norm": 41.56190872192383,
      "learning_rate": 0.00019452972972826044,
      "loss": 7.1307,
      "step": 92540
    },
    {
      "epoch": 11.137184115523466,
      "grad_norm": 57.11745071411133,
      "learning_rate": 0.00019452848866867182,
      "loss": 7.239,
      "step": 92550
    },
    {
      "epoch": 11.138387484957882,
      "grad_norm": 115.2613525390625,
      "learning_rate": 0.00019452724747227716,
      "loss": 7.1385,
      "step": 92560
    },
    {
      "epoch": 11.1395908543923,
      "grad_norm": 123.32693481445312,
      "learning_rate": 0.00019452600613907822,
      "loss": 7.1638,
      "step": 92570
    },
    {
      "epoch": 11.140794223826715,
      "grad_norm": 164.2078399658203,
      "learning_rate": 0.0001945247646690768,
      "loss": 7.2597,
      "step": 92580
    },
    {
      "epoch": 11.14199759326113,
      "grad_norm": 78.80862426757812,
      "learning_rate": 0.0001945235230622747,
      "loss": 7.2637,
      "step": 92590
    },
    {
      "epoch": 11.143200962695548,
      "grad_norm": 38.976924896240234,
      "learning_rate": 0.00019452228131867374,
      "loss": 7.2155,
      "step": 92600
    },
    {
      "epoch": 11.144404332129964,
      "grad_norm": 69.75531768798828,
      "learning_rate": 0.00019452103943827571,
      "loss": 7.2173,
      "step": 92610
    },
    {
      "epoch": 11.14560770156438,
      "grad_norm": 82.06482696533203,
      "learning_rate": 0.00019451979742108238,
      "loss": 7.1878,
      "step": 92620
    },
    {
      "epoch": 11.146811070998796,
      "grad_norm": 30.522260665893555,
      "learning_rate": 0.00019451855526709562,
      "loss": 7.1819,
      "step": 92630
    },
    {
      "epoch": 11.148014440433213,
      "grad_norm": 93.49755859375,
      "learning_rate": 0.0001945173129763171,
      "loss": 7.1299,
      "step": 92640
    },
    {
      "epoch": 11.14921780986763,
      "grad_norm": 29.52436637878418,
      "learning_rate": 0.00019451607054874875,
      "loss": 7.1796,
      "step": 92650
    },
    {
      "epoch": 11.150421179302045,
      "grad_norm": 143.0687713623047,
      "learning_rate": 0.0001945148279843923,
      "loss": 7.1951,
      "step": 92660
    },
    {
      "epoch": 11.151624548736462,
      "grad_norm": 59.0098762512207,
      "learning_rate": 0.00019451358528324955,
      "loss": 7.2471,
      "step": 92670
    },
    {
      "epoch": 11.152827918170878,
      "grad_norm": 39.90213394165039,
      "learning_rate": 0.00019451234244532227,
      "loss": 7.1435,
      "step": 92680
    },
    {
      "epoch": 11.154031287605294,
      "grad_norm": 80.75251007080078,
      "learning_rate": 0.00019451109947061235,
      "loss": 7.2116,
      "step": 92690
    },
    {
      "epoch": 11.155234657039712,
      "grad_norm": 37.435218811035156,
      "learning_rate": 0.00019450985635912152,
      "loss": 7.2105,
      "step": 92700
    },
    {
      "epoch": 11.156438026474127,
      "grad_norm": 73.86651611328125,
      "learning_rate": 0.00019450861311085161,
      "loss": 7.1904,
      "step": 92710
    },
    {
      "epoch": 11.157641395908543,
      "grad_norm": 50.3310546875,
      "learning_rate": 0.00019450736972580437,
      "loss": 7.0439,
      "step": 92720
    },
    {
      "epoch": 11.15884476534296,
      "grad_norm": 21.93097496032715,
      "learning_rate": 0.00019450612620398164,
      "loss": 7.1834,
      "step": 92730
    },
    {
      "epoch": 11.160048134777377,
      "grad_norm": 33.05723190307617,
      "learning_rate": 0.0001945048825453852,
      "loss": 7.178,
      "step": 92740
    },
    {
      "epoch": 11.161251504211792,
      "grad_norm": 25.512727737426758,
      "learning_rate": 0.0001945036387500169,
      "loss": 7.1151,
      "step": 92750
    },
    {
      "epoch": 11.16245487364621,
      "grad_norm": 40.864253997802734,
      "learning_rate": 0.0001945023948178785,
      "loss": 7.2379,
      "step": 92760
    },
    {
      "epoch": 11.163658243080626,
      "grad_norm": 25.75570297241211,
      "learning_rate": 0.00019450115074897176,
      "loss": 7.2645,
      "step": 92770
    },
    {
      "epoch": 11.164861612515042,
      "grad_norm": 13.679675102233887,
      "learning_rate": 0.00019449990654329854,
      "loss": 7.166,
      "step": 92780
    },
    {
      "epoch": 11.166064981949459,
      "grad_norm": 35.34743881225586,
      "learning_rate": 0.00019449866220086067,
      "loss": 7.1737,
      "step": 92790
    },
    {
      "epoch": 11.167268351383875,
      "grad_norm": 9.715303421020508,
      "learning_rate": 0.00019449741772165985,
      "loss": 7.282,
      "step": 92800
    },
    {
      "epoch": 11.16847172081829,
      "grad_norm": 18.321287155151367,
      "learning_rate": 0.00019449617310569796,
      "loss": 7.2068,
      "step": 92810
    },
    {
      "epoch": 11.169675090252708,
      "grad_norm": 20.16455841064453,
      "learning_rate": 0.0001944949283529768,
      "loss": 7.2858,
      "step": 92820
    },
    {
      "epoch": 11.170878459687124,
      "grad_norm": 12.81849479675293,
      "learning_rate": 0.0001944936834634981,
      "loss": 7.2756,
      "step": 92830
    },
    {
      "epoch": 11.17208182912154,
      "grad_norm": 51.767635345458984,
      "learning_rate": 0.00019449243843726373,
      "loss": 7.1577,
      "step": 92840
    },
    {
      "epoch": 11.173285198555957,
      "grad_norm": 31.77743911743164,
      "learning_rate": 0.00019449119327427547,
      "loss": 7.2155,
      "step": 92850
    },
    {
      "epoch": 11.174488567990373,
      "grad_norm": 26.9172420501709,
      "learning_rate": 0.00019448994797453513,
      "loss": 7.1221,
      "step": 92860
    },
    {
      "epoch": 11.175691937424789,
      "grad_norm": 129.29295349121094,
      "learning_rate": 0.0001944887025380445,
      "loss": 7.1493,
      "step": 92870
    },
    {
      "epoch": 11.176895306859207,
      "grad_norm": 38.593170166015625,
      "learning_rate": 0.0001944874569648054,
      "loss": 7.2299,
      "step": 92880
    },
    {
      "epoch": 11.178098676293622,
      "grad_norm": 65.74227905273438,
      "learning_rate": 0.00019448621125481963,
      "loss": 7.1211,
      "step": 92890
    },
    {
      "epoch": 11.179302045728038,
      "grad_norm": 39.74928283691406,
      "learning_rate": 0.00019448496540808895,
      "loss": 7.1957,
      "step": 92900
    },
    {
      "epoch": 11.180505415162456,
      "grad_norm": 29.60835838317871,
      "learning_rate": 0.00019448371942461522,
      "loss": 7.1436,
      "step": 92910
    },
    {
      "epoch": 11.181708784596871,
      "grad_norm": 28.809106826782227,
      "learning_rate": 0.0001944824733044002,
      "loss": 7.0376,
      "step": 92920
    },
    {
      "epoch": 11.182912154031287,
      "grad_norm": 33.554931640625,
      "learning_rate": 0.00019448122704744574,
      "loss": 7.2977,
      "step": 92930
    },
    {
      "epoch": 11.184115523465705,
      "grad_norm": 41.13526916503906,
      "learning_rate": 0.0001944799806537536,
      "loss": 7.1977,
      "step": 92940
    },
    {
      "epoch": 11.18531889290012,
      "grad_norm": 46.14151382446289,
      "learning_rate": 0.00019447873412332562,
      "loss": 7.1166,
      "step": 92950
    },
    {
      "epoch": 11.186522262334536,
      "grad_norm": 26.503915786743164,
      "learning_rate": 0.0001944774874561636,
      "loss": 7.2647,
      "step": 92960
    },
    {
      "epoch": 11.187725631768952,
      "grad_norm": 19.187028884887695,
      "learning_rate": 0.00019447624065226928,
      "loss": 7.1581,
      "step": 92970
    },
    {
      "epoch": 11.18892900120337,
      "grad_norm": 13.668183326721191,
      "learning_rate": 0.00019447499371164458,
      "loss": 7.2296,
      "step": 92980
    },
    {
      "epoch": 11.190132370637786,
      "grad_norm": 39.912803649902344,
      "learning_rate": 0.0001944737466342912,
      "loss": 7.1242,
      "step": 92990
    },
    {
      "epoch": 11.191335740072201,
      "grad_norm": 26.953168869018555,
      "learning_rate": 0.00019447249942021095,
      "loss": 7.1519,
      "step": 93000
    },
    {
      "epoch": 11.192539109506619,
      "grad_norm": 49.079139709472656,
      "learning_rate": 0.00019447125206940573,
      "loss": 7.1099,
      "step": 93010
    },
    {
      "epoch": 11.193742478941035,
      "grad_norm": 51.37486267089844,
      "learning_rate": 0.00019447000458187725,
      "loss": 7.0384,
      "step": 93020
    },
    {
      "epoch": 11.19494584837545,
      "grad_norm": 31.320575714111328,
      "learning_rate": 0.00019446875695762735,
      "loss": 7.2404,
      "step": 93030
    },
    {
      "epoch": 11.196149217809868,
      "grad_norm": 407.4568176269531,
      "learning_rate": 0.00019446750919665783,
      "loss": 7.1589,
      "step": 93040
    },
    {
      "epoch": 11.197352587244284,
      "grad_norm": 45.0624885559082,
      "learning_rate": 0.00019446626129897052,
      "loss": 7.2945,
      "step": 93050
    },
    {
      "epoch": 11.1985559566787,
      "grad_norm": 32.37380599975586,
      "learning_rate": 0.0001944650132645672,
      "loss": 7.1759,
      "step": 93060
    },
    {
      "epoch": 11.199759326113117,
      "grad_norm": 33.439395904541016,
      "learning_rate": 0.00019446376509344967,
      "loss": 7.2295,
      "step": 93070
    },
    {
      "epoch": 11.200962695547533,
      "grad_norm": 32.01109313964844,
      "learning_rate": 0.00019446251678561979,
      "loss": 7.1219,
      "step": 93080
    },
    {
      "epoch": 11.202166064981949,
      "grad_norm": 39.83729934692383,
      "learning_rate": 0.00019446126834107928,
      "loss": 7.1233,
      "step": 93090
    },
    {
      "epoch": 11.203369434416366,
      "grad_norm": 31.020111083984375,
      "learning_rate": 0.00019446001975983,
      "loss": 7.2814,
      "step": 93100
    },
    {
      "epoch": 11.204572803850782,
      "grad_norm": 74.25709533691406,
      "learning_rate": 0.00019445877104187377,
      "loss": 7.1696,
      "step": 93110
    },
    {
      "epoch": 11.205776173285198,
      "grad_norm": 50.41210174560547,
      "learning_rate": 0.00019445752218721237,
      "loss": 7.1742,
      "step": 93120
    },
    {
      "epoch": 11.206979542719615,
      "grad_norm": 13.669230461120605,
      "learning_rate": 0.00019445627319584761,
      "loss": 7.2549,
      "step": 93130
    },
    {
      "epoch": 11.208182912154031,
      "grad_norm": 56.16986846923828,
      "learning_rate": 0.0001944550240677813,
      "loss": 7.306,
      "step": 93140
    },
    {
      "epoch": 11.209386281588447,
      "grad_norm": 45.588157653808594,
      "learning_rate": 0.00019445377480301526,
      "loss": 6.9637,
      "step": 93150
    },
    {
      "epoch": 11.210589651022865,
      "grad_norm": 83.8860855102539,
      "learning_rate": 0.00019445252540155127,
      "loss": 7.1724,
      "step": 93160
    },
    {
      "epoch": 11.21179302045728,
      "grad_norm": 70.70026397705078,
      "learning_rate": 0.00019445127586339118,
      "loss": 7.1555,
      "step": 93170
    },
    {
      "epoch": 11.212996389891696,
      "grad_norm": 83.21636199951172,
      "learning_rate": 0.00019445002618853678,
      "loss": 7.0629,
      "step": 93180
    },
    {
      "epoch": 11.214199759326114,
      "grad_norm": 65.31185150146484,
      "learning_rate": 0.00019444877637698986,
      "loss": 7.2527,
      "step": 93190
    },
    {
      "epoch": 11.21540312876053,
      "grad_norm": 40.810035705566406,
      "learning_rate": 0.00019444752642875222,
      "loss": 7.1287,
      "step": 93200
    },
    {
      "epoch": 11.216606498194945,
      "grad_norm": 66.74649810791016,
      "learning_rate": 0.00019444627634382573,
      "loss": 7.2442,
      "step": 93210
    },
    {
      "epoch": 11.217809867629363,
      "grad_norm": 47.640777587890625,
      "learning_rate": 0.00019444502612221213,
      "loss": 7.1871,
      "step": 93220
    },
    {
      "epoch": 11.219013237063779,
      "grad_norm": 25.909526824951172,
      "learning_rate": 0.00019444377576391327,
      "loss": 7.0653,
      "step": 93230
    },
    {
      "epoch": 11.220216606498195,
      "grad_norm": 32.09661102294922,
      "learning_rate": 0.00019444252526893094,
      "loss": 7.0846,
      "step": 93240
    },
    {
      "epoch": 11.221419975932612,
      "grad_norm": 21.584688186645508,
      "learning_rate": 0.00019444127463726698,
      "loss": 7.123,
      "step": 93250
    },
    {
      "epoch": 11.222623345367028,
      "grad_norm": 42.743892669677734,
      "learning_rate": 0.00019444002386892316,
      "loss": 7.067,
      "step": 93260
    },
    {
      "epoch": 11.223826714801444,
      "grad_norm": 29.378700256347656,
      "learning_rate": 0.0001944387729639013,
      "loss": 7.1114,
      "step": 93270
    },
    {
      "epoch": 11.225030084235861,
      "grad_norm": 15.080081939697266,
      "learning_rate": 0.00019443752192220326,
      "loss": 7.174,
      "step": 93280
    },
    {
      "epoch": 11.226233453670277,
      "grad_norm": 21.681062698364258,
      "learning_rate": 0.00019443627074383076,
      "loss": 7.1851,
      "step": 93290
    },
    {
      "epoch": 11.227436823104693,
      "grad_norm": 19.04930877685547,
      "learning_rate": 0.0001944350194287857,
      "loss": 7.1523,
      "step": 93300
    },
    {
      "epoch": 11.22864019253911,
      "grad_norm": 52.01959228515625,
      "learning_rate": 0.00019443376797706984,
      "loss": 7.2094,
      "step": 93310
    },
    {
      "epoch": 11.229843561973526,
      "grad_norm": 57.18655014038086,
      "learning_rate": 0.000194432516388685,
      "loss": 7.1087,
      "step": 93320
    },
    {
      "epoch": 11.231046931407942,
      "grad_norm": 26.65498924255371,
      "learning_rate": 0.000194431264663633,
      "loss": 7.1064,
      "step": 93330
    },
    {
      "epoch": 11.232250300842358,
      "grad_norm": 79.77294158935547,
      "learning_rate": 0.00019443001280191564,
      "loss": 7.1444,
      "step": 93340
    },
    {
      "epoch": 11.233453670276775,
      "grad_norm": 26.291479110717773,
      "learning_rate": 0.00019442876080353472,
      "loss": 7.203,
      "step": 93350
    },
    {
      "epoch": 11.234657039711191,
      "grad_norm": 45.38020706176758,
      "learning_rate": 0.0001944275086684921,
      "loss": 7.1146,
      "step": 93360
    },
    {
      "epoch": 11.235860409145607,
      "grad_norm": 56.406150817871094,
      "learning_rate": 0.00019442625639678953,
      "loss": 7.1607,
      "step": 93370
    },
    {
      "epoch": 11.237063778580024,
      "grad_norm": 15.539941787719727,
      "learning_rate": 0.00019442500398842886,
      "loss": 7.1901,
      "step": 93380
    },
    {
      "epoch": 11.23826714801444,
      "grad_norm": 30.330307006835938,
      "learning_rate": 0.0001944237514434119,
      "loss": 7.1586,
      "step": 93390
    },
    {
      "epoch": 11.239470517448856,
      "grad_norm": 40.90454864501953,
      "learning_rate": 0.0001944224987617405,
      "loss": 7.2042,
      "step": 93400
    },
    {
      "epoch": 11.240673886883274,
      "grad_norm": 87.12039184570312,
      "learning_rate": 0.0001944212459434164,
      "loss": 7.1879,
      "step": 93410
    },
    {
      "epoch": 11.24187725631769,
      "grad_norm": 105.44353485107422,
      "learning_rate": 0.00019441999298844143,
      "loss": 7.0583,
      "step": 93420
    },
    {
      "epoch": 11.243080625752105,
      "grad_norm": 27.090097427368164,
      "learning_rate": 0.00019441873989681745,
      "loss": 7.1794,
      "step": 93430
    },
    {
      "epoch": 11.244283995186523,
      "grad_norm": 88.40074157714844,
      "learning_rate": 0.00019441748666854623,
      "loss": 7.1585,
      "step": 93440
    },
    {
      "epoch": 11.245487364620939,
      "grad_norm": 31.46632194519043,
      "learning_rate": 0.00019441623330362958,
      "loss": 7.1761,
      "step": 93450
    },
    {
      "epoch": 11.246690734055354,
      "grad_norm": 88.53533172607422,
      "learning_rate": 0.00019441497980206935,
      "loss": 7.1753,
      "step": 93460
    },
    {
      "epoch": 11.247894103489772,
      "grad_norm": 95.59032440185547,
      "learning_rate": 0.00019441372616386732,
      "loss": 7.1662,
      "step": 93470
    },
    {
      "epoch": 11.249097472924188,
      "grad_norm": 37.85075378417969,
      "learning_rate": 0.0001944124723890253,
      "loss": 7.2929,
      "step": 93480
    },
    {
      "epoch": 11.250300842358604,
      "grad_norm": 60.5465202331543,
      "learning_rate": 0.00019441121847754516,
      "loss": 7.2749,
      "step": 93490
    },
    {
      "epoch": 11.251504211793021,
      "grad_norm": 32.9837532043457,
      "learning_rate": 0.00019440996442942867,
      "loss": 7.0887,
      "step": 93500
    },
    {
      "epoch": 11.252707581227437,
      "grad_norm": 85.40311431884766,
      "learning_rate": 0.00019440871024467763,
      "loss": 7.0611,
      "step": 93510
    },
    {
      "epoch": 11.253910950661853,
      "grad_norm": 34.23643112182617,
      "learning_rate": 0.0001944074559232939,
      "loss": 7.1397,
      "step": 93520
    },
    {
      "epoch": 11.25511432009627,
      "grad_norm": 30.25879669189453,
      "learning_rate": 0.00019440620146527927,
      "loss": 7.1157,
      "step": 93530
    },
    {
      "epoch": 11.256317689530686,
      "grad_norm": 104.1557846069336,
      "learning_rate": 0.00019440494687063557,
      "loss": 7.0913,
      "step": 93540
    },
    {
      "epoch": 11.257521058965102,
      "grad_norm": 115.49676513671875,
      "learning_rate": 0.00019440369213936457,
      "loss": 7.141,
      "step": 93550
    },
    {
      "epoch": 11.25872442839952,
      "grad_norm": 47.08007049560547,
      "learning_rate": 0.00019440243727146815,
      "loss": 7.1111,
      "step": 93560
    },
    {
      "epoch": 11.259927797833935,
      "grad_norm": 102.65309143066406,
      "learning_rate": 0.0001944011822669481,
      "loss": 7.1391,
      "step": 93570
    },
    {
      "epoch": 11.261131167268351,
      "grad_norm": 121.46197509765625,
      "learning_rate": 0.0001943999271258062,
      "loss": 7.1205,
      "step": 93580
    },
    {
      "epoch": 11.262334536702769,
      "grad_norm": 747.8783569335938,
      "learning_rate": 0.0001943986718480443,
      "loss": 7.2145,
      "step": 93590
    },
    {
      "epoch": 11.263537906137184,
      "grad_norm": 416.4867858886719,
      "learning_rate": 0.00019439741643366426,
      "loss": 7.2525,
      "step": 93600
    },
    {
      "epoch": 11.2647412755716,
      "grad_norm": 273.11578369140625,
      "learning_rate": 0.0001943961608826678,
      "loss": 7.1684,
      "step": 93610
    },
    {
      "epoch": 11.265944645006018,
      "grad_norm": 452.9868469238281,
      "learning_rate": 0.00019439490519505682,
      "loss": 7.0907,
      "step": 93620
    },
    {
      "epoch": 11.267148014440433,
      "grad_norm": 214.64022827148438,
      "learning_rate": 0.0001943936493708331,
      "loss": 7.1935,
      "step": 93630
    },
    {
      "epoch": 11.26835138387485,
      "grad_norm": 475.05279541015625,
      "learning_rate": 0.00019439239340999847,
      "loss": 7.0981,
      "step": 93640
    },
    {
      "epoch": 11.269554753309265,
      "grad_norm": 269.6962890625,
      "learning_rate": 0.00019439113731255475,
      "loss": 7.2454,
      "step": 93650
    },
    {
      "epoch": 11.270758122743683,
      "grad_norm": 329.68585205078125,
      "learning_rate": 0.00019438988107850372,
      "loss": 7.1289,
      "step": 93660
    },
    {
      "epoch": 11.271961492178098,
      "grad_norm": 132.30197143554688,
      "learning_rate": 0.00019438862470784726,
      "loss": 7.211,
      "step": 93670
    },
    {
      "epoch": 11.273164861612514,
      "grad_norm": 134.85240173339844,
      "learning_rate": 0.00019438736820058715,
      "loss": 7.1663,
      "step": 93680
    },
    {
      "epoch": 11.274368231046932,
      "grad_norm": 176.08334350585938,
      "learning_rate": 0.00019438611155672518,
      "loss": 7.144,
      "step": 93690
    },
    {
      "epoch": 11.275571600481348,
      "grad_norm": 187.31689453125,
      "learning_rate": 0.00019438485477626323,
      "loss": 7.1344,
      "step": 93700
    },
    {
      "epoch": 11.276774969915763,
      "grad_norm": 184.18238830566406,
      "learning_rate": 0.0001943835978592031,
      "loss": 7.2313,
      "step": 93710
    },
    {
      "epoch": 11.277978339350181,
      "grad_norm": 214.82232666015625,
      "learning_rate": 0.00019438234080554657,
      "loss": 7.1117,
      "step": 93720
    },
    {
      "epoch": 11.279181708784597,
      "grad_norm": 186.7197265625,
      "learning_rate": 0.00019438108361529552,
      "loss": 7.0882,
      "step": 93730
    },
    {
      "epoch": 11.280385078219012,
      "grad_norm": 293.4224853515625,
      "learning_rate": 0.00019437982628845175,
      "loss": 7.26,
      "step": 93740
    },
    {
      "epoch": 11.28158844765343,
      "grad_norm": 258.84698486328125,
      "learning_rate": 0.00019437856882501702,
      "loss": 7.1738,
      "step": 93750
    },
    {
      "epoch": 11.282791817087846,
      "grad_norm": 309.6853332519531,
      "learning_rate": 0.00019437731122499326,
      "loss": 7.2102,
      "step": 93760
    },
    {
      "epoch": 11.283995186522262,
      "grad_norm": 645.7169189453125,
      "learning_rate": 0.0001943760534883822,
      "loss": 7.1867,
      "step": 93770
    },
    {
      "epoch": 11.28519855595668,
      "grad_norm": 587.5661010742188,
      "learning_rate": 0.0001943747956151857,
      "loss": 7.2186,
      "step": 93780
    },
    {
      "epoch": 11.286401925391095,
      "grad_norm": 586.8329467773438,
      "learning_rate": 0.00019437353760540555,
      "loss": 7.1943,
      "step": 93790
    },
    {
      "epoch": 11.28760529482551,
      "grad_norm": 709.7696533203125,
      "learning_rate": 0.00019437227945904358,
      "loss": 7.2064,
      "step": 93800
    },
    {
      "epoch": 11.288808664259928,
      "grad_norm": 522.105712890625,
      "learning_rate": 0.00019437102117610166,
      "loss": 7.2035,
      "step": 93810
    },
    {
      "epoch": 11.290012033694344,
      "grad_norm": 935.1792602539062,
      "learning_rate": 0.00019436976275658156,
      "loss": 7.2011,
      "step": 93820
    },
    {
      "epoch": 11.29121540312876,
      "grad_norm": 419.7920837402344,
      "learning_rate": 0.00019436850420048512,
      "loss": 7.1943,
      "step": 93830
    },
    {
      "epoch": 11.292418772563177,
      "grad_norm": 908.8665161132812,
      "learning_rate": 0.00019436724550781414,
      "loss": 7.2612,
      "step": 93840
    },
    {
      "epoch": 11.293622141997593,
      "grad_norm": 483.0259094238281,
      "learning_rate": 0.0001943659866785705,
      "loss": 7.3596,
      "step": 93850
    },
    {
      "epoch": 11.294825511432009,
      "grad_norm": 633.0369262695312,
      "learning_rate": 0.00019436472771275592,
      "loss": 7.2362,
      "step": 93860
    },
    {
      "epoch": 11.296028880866427,
      "grad_norm": 333.7626037597656,
      "learning_rate": 0.00019436346861037232,
      "loss": 7.2016,
      "step": 93870
    },
    {
      "epoch": 11.297232250300842,
      "grad_norm": 398.3389892578125,
      "learning_rate": 0.00019436220937142147,
      "loss": 7.2256,
      "step": 93880
    },
    {
      "epoch": 11.298435619735258,
      "grad_norm": 1115.1990966796875,
      "learning_rate": 0.0001943609499959052,
      "loss": 7.2736,
      "step": 93890
    },
    {
      "epoch": 11.299638989169676,
      "grad_norm": 346.9666748046875,
      "learning_rate": 0.00019435969048382532,
      "loss": 7.1469,
      "step": 93900
    },
    {
      "epoch": 11.300842358604092,
      "grad_norm": 2024.689453125,
      "learning_rate": 0.0001943584308351837,
      "loss": 7.0714,
      "step": 93910
    },
    {
      "epoch": 11.302045728038507,
      "grad_norm": 333.5049133300781,
      "learning_rate": 0.00019435717104998213,
      "loss": 7.0776,
      "step": 93920
    },
    {
      "epoch": 11.303249097472925,
      "grad_norm": 368.3515625,
      "learning_rate": 0.00019435591112822244,
      "loss": 7.0975,
      "step": 93930
    },
    {
      "epoch": 11.30445246690734,
      "grad_norm": 329.6982421875,
      "learning_rate": 0.00019435465106990645,
      "loss": 7.2583,
      "step": 93940
    },
    {
      "epoch": 11.305655836341757,
      "grad_norm": 252.72410583496094,
      "learning_rate": 0.00019435339087503598,
      "loss": 7.2367,
      "step": 93950
    },
    {
      "epoch": 11.306859205776174,
      "grad_norm": 304.77587890625,
      "learning_rate": 0.00019435213054361285,
      "loss": 7.1695,
      "step": 93960
    },
    {
      "epoch": 11.30806257521059,
      "grad_norm": 115.09303283691406,
      "learning_rate": 0.0001943508700756389,
      "loss": 7.2383,
      "step": 93970
    },
    {
      "epoch": 11.309265944645006,
      "grad_norm": 300.5790100097656,
      "learning_rate": 0.00019434960947111596,
      "loss": 7.1671,
      "step": 93980
    },
    {
      "epoch": 11.310469314079423,
      "grad_norm": 187.3275604248047,
      "learning_rate": 0.00019434834873004585,
      "loss": 7.1337,
      "step": 93990
    },
    {
      "epoch": 11.311672683513839,
      "grad_norm": 401.3988952636719,
      "learning_rate": 0.00019434708785243031,
      "loss": 7.1607,
      "step": 94000
    },
    {
      "epoch": 11.312876052948255,
      "grad_norm": 240.0775146484375,
      "learning_rate": 0.0001943458268382713,
      "loss": 7.1282,
      "step": 94010
    },
    {
      "epoch": 11.314079422382672,
      "grad_norm": 64.38675689697266,
      "learning_rate": 0.0001943445656875706,
      "loss": 7.1417,
      "step": 94020
    },
    {
      "epoch": 11.315282791817088,
      "grad_norm": 82.42398834228516,
      "learning_rate": 0.00019434330440033,
      "loss": 7.2038,
      "step": 94030
    },
    {
      "epoch": 11.316486161251504,
      "grad_norm": 88.60486602783203,
      "learning_rate": 0.00019434204297655137,
      "loss": 7.1203,
      "step": 94040
    },
    {
      "epoch": 11.31768953068592,
      "grad_norm": 91.94270324707031,
      "learning_rate": 0.00019434078141623647,
      "loss": 7.1417,
      "step": 94050
    },
    {
      "epoch": 11.318892900120337,
      "grad_norm": 465.58831787109375,
      "learning_rate": 0.00019433951971938718,
      "loss": 7.1706,
      "step": 94060
    },
    {
      "epoch": 11.320096269554753,
      "grad_norm": 281.7317810058594,
      "learning_rate": 0.00019433825788600534,
      "loss": 7.1066,
      "step": 94070
    },
    {
      "epoch": 11.321299638989169,
      "grad_norm": 342.6062316894531,
      "learning_rate": 0.0001943369959160927,
      "loss": 7.1243,
      "step": 94080
    },
    {
      "epoch": 11.322503008423586,
      "grad_norm": 2160.69140625,
      "learning_rate": 0.0001943357338096512,
      "loss": 7.1727,
      "step": 94090
    },
    {
      "epoch": 11.323706377858002,
      "grad_norm": 133.9144287109375,
      "learning_rate": 0.00019433447156668254,
      "loss": 7.1418,
      "step": 94100
    },
    {
      "epoch": 11.324909747292418,
      "grad_norm": 910.3568115234375,
      "learning_rate": 0.00019433320918718866,
      "loss": 7.2835,
      "step": 94110
    },
    {
      "epoch": 11.326113116726836,
      "grad_norm": 56116.01171875,
      "learning_rate": 0.00019433194667117128,
      "loss": 7.2748,
      "step": 94120
    },
    {
      "epoch": 11.327316486161251,
      "grad_norm": 2441.263671875,
      "learning_rate": 0.00019433068401863234,
      "loss": 7.2558,
      "step": 94130
    },
    {
      "epoch": 11.328519855595667,
      "grad_norm": 269.61370849609375,
      "learning_rate": 0.00019432942122957355,
      "loss": 7.2334,
      "step": 94140
    },
    {
      "epoch": 11.329723225030085,
      "grad_norm": 462.24847412109375,
      "learning_rate": 0.00019432815830399688,
      "loss": 7.1521,
      "step": 94150
    },
    {
      "epoch": 11.3309265944645,
      "grad_norm": 132.6151580810547,
      "learning_rate": 0.000194326895241904,
      "loss": 7.3506,
      "step": 94160
    },
    {
      "epoch": 11.332129963898916,
      "grad_norm": 253.41500854492188,
      "learning_rate": 0.00019432563204329684,
      "loss": 7.168,
      "step": 94170
    },
    {
      "epoch": 11.333333333333334,
      "grad_norm": 141.18609619140625,
      "learning_rate": 0.00019432436870817722,
      "loss": 7.3042,
      "step": 94180
    },
    {
      "epoch": 11.33453670276775,
      "grad_norm": 180.30938720703125,
      "learning_rate": 0.00019432310523654692,
      "loss": 7.092,
      "step": 94190
    },
    {
      "epoch": 11.335740072202166,
      "grad_norm": 721.1651611328125,
      "learning_rate": 0.00019432184162840782,
      "loss": 7.1546,
      "step": 94200
    },
    {
      "epoch": 11.336943441636583,
      "grad_norm": 254.870361328125,
      "learning_rate": 0.00019432057788376175,
      "loss": 7.1647,
      "step": 94210
    },
    {
      "epoch": 11.338146811070999,
      "grad_norm": 457.938232421875,
      "learning_rate": 0.0001943193140026105,
      "loss": 7.1007,
      "step": 94220
    },
    {
      "epoch": 11.339350180505415,
      "grad_norm": 87.42581939697266,
      "learning_rate": 0.0001943180499849559,
      "loss": 7.1604,
      "step": 94230
    },
    {
      "epoch": 11.340553549939832,
      "grad_norm": 91.93179321289062,
      "learning_rate": 0.0001943167858307998,
      "loss": 7.2931,
      "step": 94240
    },
    {
      "epoch": 11.341756919374248,
      "grad_norm": 85.75534057617188,
      "learning_rate": 0.00019431552154014402,
      "loss": 7.1948,
      "step": 94250
    },
    {
      "epoch": 11.342960288808664,
      "grad_norm": 154.4004364013672,
      "learning_rate": 0.0001943142571129904,
      "loss": 7.2252,
      "step": 94260
    },
    {
      "epoch": 11.344163658243081,
      "grad_norm": 84.31697082519531,
      "learning_rate": 0.00019431299254934076,
      "loss": 7.1564,
      "step": 94270
    },
    {
      "epoch": 11.345367027677497,
      "grad_norm": 127.80133819580078,
      "learning_rate": 0.00019431172784919695,
      "loss": 7.1801,
      "step": 94280
    },
    {
      "epoch": 11.346570397111913,
      "grad_norm": 113.86201477050781,
      "learning_rate": 0.00019431046301256077,
      "loss": 7.1007,
      "step": 94290
    },
    {
      "epoch": 11.34777376654633,
      "grad_norm": 289.74700927734375,
      "learning_rate": 0.0001943091980394341,
      "loss": 7.1574,
      "step": 94300
    },
    {
      "epoch": 11.348977135980746,
      "grad_norm": 104.62680053710938,
      "learning_rate": 0.0001943079329298187,
      "loss": 7.1413,
      "step": 94310
    },
    {
      "epoch": 11.350180505415162,
      "grad_norm": 136.5205841064453,
      "learning_rate": 0.0001943066676837164,
      "loss": 7.0677,
      "step": 94320
    },
    {
      "epoch": 11.35138387484958,
      "grad_norm": 109.92561340332031,
      "learning_rate": 0.00019430540230112913,
      "loss": 7.1886,
      "step": 94330
    },
    {
      "epoch": 11.352587244283995,
      "grad_norm": 150.95932006835938,
      "learning_rate": 0.00019430413678205865,
      "loss": 7.0972,
      "step": 94340
    },
    {
      "epoch": 11.353790613718411,
      "grad_norm": 191.3978271484375,
      "learning_rate": 0.00019430287112650677,
      "loss": 7.2288,
      "step": 94350
    },
    {
      "epoch": 11.354993983152827,
      "grad_norm": 7439.27978515625,
      "learning_rate": 0.0001943016053344754,
      "loss": 7.1949,
      "step": 94360
    },
    {
      "epoch": 11.356197352587245,
      "grad_norm": 14434.5625,
      "learning_rate": 0.0001943003394059663,
      "loss": 7.1926,
      "step": 94370
    },
    {
      "epoch": 11.35740072202166,
      "grad_norm": 7459.72607421875,
      "learning_rate": 0.00019429907334098134,
      "loss": 7.3478,
      "step": 94380
    },
    {
      "epoch": 11.358604091456076,
      "grad_norm": 47360.546875,
      "learning_rate": 0.00019429780713952233,
      "loss": 7.1775,
      "step": 94390
    },
    {
      "epoch": 11.359807460890494,
      "grad_norm": 6103.2099609375,
      "learning_rate": 0.0001942965408015911,
      "loss": 7.1571,
      "step": 94400
    },
    {
      "epoch": 11.36101083032491,
      "grad_norm": 5248.42041015625,
      "learning_rate": 0.0001942952743271895,
      "loss": 7.3057,
      "step": 94410
    },
    {
      "epoch": 11.362214199759325,
      "grad_norm": 8737.953125,
      "learning_rate": 0.00019429400771631936,
      "loss": 7.2353,
      "step": 94420
    },
    {
      "epoch": 11.363417569193743,
      "grad_norm": 15249.4755859375,
      "learning_rate": 0.0001942927409689825,
      "loss": 7.1192,
      "step": 94430
    },
    {
      "epoch": 11.364620938628159,
      "grad_norm": 24800.455078125,
      "learning_rate": 0.0001942914740851808,
      "loss": 7.1431,
      "step": 94440
    },
    {
      "epoch": 11.365824308062574,
      "grad_norm": 7497.79052734375,
      "learning_rate": 0.000194290207064916,
      "loss": 7.218,
      "step": 94450
    },
    {
      "epoch": 11.367027677496992,
      "grad_norm": 6490.017578125,
      "learning_rate": 0.00019428893990819005,
      "loss": 7.2677,
      "step": 94460
    },
    {
      "epoch": 11.368231046931408,
      "grad_norm": 12166.8515625,
      "learning_rate": 0.0001942876726150047,
      "loss": 7.2949,
      "step": 94470
    },
    {
      "epoch": 11.369434416365824,
      "grad_norm": 12129.544921875,
      "learning_rate": 0.0001942864051853618,
      "loss": 7.2637,
      "step": 94480
    },
    {
      "epoch": 11.370637785800241,
      "grad_norm": 10249.13671875,
      "learning_rate": 0.0001942851376192632,
      "loss": 7.3413,
      "step": 94490
    },
    {
      "epoch": 11.371841155234657,
      "grad_norm": 57884.32421875,
      "learning_rate": 0.00019428386991671072,
      "loss": 7.2272,
      "step": 94500
    },
    {
      "epoch": 11.373044524669073,
      "grad_norm": 16126.486328125,
      "learning_rate": 0.0001942826020777062,
      "loss": 7.4479,
      "step": 94510
    },
    {
      "epoch": 11.37424789410349,
      "grad_norm": 711.2689208984375,
      "learning_rate": 0.00019428133410225149,
      "loss": 7.1287,
      "step": 94520
    },
    {
      "epoch": 11.375451263537906,
      "grad_norm": 83732.4921875,
      "learning_rate": 0.00019428006599034843,
      "loss": 7.2605,
      "step": 94530
    },
    {
      "epoch": 11.376654632972322,
      "grad_norm": 235717.21875,
      "learning_rate": 0.00019427879774199883,
      "loss": 7.3628,
      "step": 94540
    },
    {
      "epoch": 11.37785800240674,
      "grad_norm": 5930.115234375,
      "learning_rate": 0.00019427752935720452,
      "loss": 7.3277,
      "step": 94550
    },
    {
      "epoch": 11.379061371841155,
      "grad_norm": 12438.8486328125,
      "learning_rate": 0.00019427626083596734,
      "loss": 7.2204,
      "step": 94560
    },
    {
      "epoch": 11.380264741275571,
      "grad_norm": 8256.85546875,
      "learning_rate": 0.00019427499217828913,
      "loss": 7.199,
      "step": 94570
    },
    {
      "epoch": 11.381468110709989,
      "grad_norm": 139658.09375,
      "learning_rate": 0.00019427372338417172,
      "loss": 7.182,
      "step": 94580
    },
    {
      "epoch": 11.382671480144404,
      "grad_norm": 771027.3125,
      "learning_rate": 0.00019427245445361702,
      "loss": 7.2765,
      "step": 94590
    },
    {
      "epoch": 11.38387484957882,
      "grad_norm": 4082.875,
      "learning_rate": 0.00019427118538662677,
      "loss": 7.2696,
      "step": 94600
    },
    {
      "epoch": 11.385078219013238,
      "grad_norm": 1302.040283203125,
      "learning_rate": 0.0001942699161832028,
      "loss": 7.3377,
      "step": 94610
    },
    {
      "epoch": 11.386281588447654,
      "grad_norm": 7915.7470703125,
      "learning_rate": 0.00019426864684334705,
      "loss": 7.1266,
      "step": 94620
    },
    {
      "epoch": 11.38748495788207,
      "grad_norm": 9391.2744140625,
      "learning_rate": 0.00019426737736706124,
      "loss": 7.1875,
      "step": 94630
    },
    {
      "epoch": 11.388688327316487,
      "grad_norm": 506.64593505859375,
      "learning_rate": 0.0001942661077543473,
      "loss": 7.1783,
      "step": 94640
    },
    {
      "epoch": 11.389891696750903,
      "grad_norm": 245.23056030273438,
      "learning_rate": 0.000194264838005207,
      "loss": 7.2504,
      "step": 94650
    },
    {
      "epoch": 11.391095066185319,
      "grad_norm": 223.31689453125,
      "learning_rate": 0.00019426356811964223,
      "loss": 7.2597,
      "step": 94660
    },
    {
      "epoch": 11.392298435619736,
      "grad_norm": 502.1722412109375,
      "learning_rate": 0.0001942622980976548,
      "loss": 7.2559,
      "step": 94670
    },
    {
      "epoch": 11.393501805054152,
      "grad_norm": 740.5872802734375,
      "learning_rate": 0.00019426102793924655,
      "loss": 7.2675,
      "step": 94680
    },
    {
      "epoch": 11.394705174488568,
      "grad_norm": 2435.196044921875,
      "learning_rate": 0.0001942597576444193,
      "loss": 7.1764,
      "step": 94690
    },
    {
      "epoch": 11.395908543922985,
      "grad_norm": 1092.825439453125,
      "learning_rate": 0.0001942584872131749,
      "loss": 7.4649,
      "step": 94700
    },
    {
      "epoch": 11.397111913357401,
      "grad_norm": 2141.05078125,
      "learning_rate": 0.00019425721664551526,
      "loss": 7.3611,
      "step": 94710
    },
    {
      "epoch": 11.398315282791817,
      "grad_norm": 468.9147644042969,
      "learning_rate": 0.0001942559459414421,
      "loss": 7.331,
      "step": 94720
    },
    {
      "epoch": 11.399518652226233,
      "grad_norm": 129.21363830566406,
      "learning_rate": 0.00019425467510095733,
      "loss": 7.2282,
      "step": 94730
    },
    {
      "epoch": 11.40072202166065,
      "grad_norm": 76.4381103515625,
      "learning_rate": 0.0001942534041240628,
      "loss": 7.2006,
      "step": 94740
    },
    {
      "epoch": 11.401925391095066,
      "grad_norm": 88.59284973144531,
      "learning_rate": 0.0001942521330107603,
      "loss": 7.1181,
      "step": 94750
    },
    {
      "epoch": 11.403128760529482,
      "grad_norm": 68.23381042480469,
      "learning_rate": 0.00019425086176105168,
      "loss": 7.1686,
      "step": 94760
    },
    {
      "epoch": 11.4043321299639,
      "grad_norm": 251.13429260253906,
      "learning_rate": 0.0001942495903749388,
      "loss": 7.0901,
      "step": 94770
    },
    {
      "epoch": 11.405535499398315,
      "grad_norm": 491.6206359863281,
      "learning_rate": 0.00019424831885242352,
      "loss": 7.2668,
      "step": 94780
    },
    {
      "epoch": 11.406738868832731,
      "grad_norm": 219.69728088378906,
      "learning_rate": 0.00019424704719350763,
      "loss": 7.239,
      "step": 94790
    },
    {
      "epoch": 11.407942238267148,
      "grad_norm": 483.8101806640625,
      "learning_rate": 0.000194245775398193,
      "loss": 7.1924,
      "step": 94800
    },
    {
      "epoch": 11.409145607701564,
      "grad_norm": 277.7901611328125,
      "learning_rate": 0.00019424450346648147,
      "loss": 7.2086,
      "step": 94810
    },
    {
      "epoch": 11.41034897713598,
      "grad_norm": 82.76998138427734,
      "learning_rate": 0.00019424323139837486,
      "loss": 7.2416,
      "step": 94820
    },
    {
      "epoch": 11.411552346570398,
      "grad_norm": 132.4668731689453,
      "learning_rate": 0.00019424195919387504,
      "loss": 7.2132,
      "step": 94830
    },
    {
      "epoch": 11.412755716004813,
      "grad_norm": 209.8936309814453,
      "learning_rate": 0.00019424068685298384,
      "loss": 7.1452,
      "step": 94840
    },
    {
      "epoch": 11.41395908543923,
      "grad_norm": 88.13241577148438,
      "learning_rate": 0.0001942394143757031,
      "loss": 7.1729,
      "step": 94850
    },
    {
      "epoch": 11.415162454873647,
      "grad_norm": 445.6883850097656,
      "learning_rate": 0.00019423814176203465,
      "loss": 7.1684,
      "step": 94860
    },
    {
      "epoch": 11.416365824308063,
      "grad_norm": 278.3733215332031,
      "learning_rate": 0.00019423686901198032,
      "loss": 7.0922,
      "step": 94870
    },
    {
      "epoch": 11.417569193742478,
      "grad_norm": 369.18939208984375,
      "learning_rate": 0.000194235596125542,
      "loss": 7.2681,
      "step": 94880
    },
    {
      "epoch": 11.418772563176896,
      "grad_norm": 1085.650634765625,
      "learning_rate": 0.00019423432310272152,
      "loss": 7.1861,
      "step": 94890
    },
    {
      "epoch": 11.419975932611312,
      "grad_norm": 620.5313110351562,
      "learning_rate": 0.00019423304994352068,
      "loss": 7.2098,
      "step": 94900
    },
    {
      "epoch": 11.421179302045728,
      "grad_norm": 240.4861297607422,
      "learning_rate": 0.00019423177664794138,
      "loss": 7.2149,
      "step": 94910
    },
    {
      "epoch": 11.422382671480145,
      "grad_norm": 105.56778717041016,
      "learning_rate": 0.00019423050321598542,
      "loss": 7.1391,
      "step": 94920
    },
    {
      "epoch": 11.42358604091456,
      "grad_norm": 49.94228744506836,
      "learning_rate": 0.00019422922964765465,
      "loss": 7.2401,
      "step": 94930
    },
    {
      "epoch": 11.424789410348977,
      "grad_norm": 541.0204467773438,
      "learning_rate": 0.0001942279559429509,
      "loss": 7.2546,
      "step": 94940
    },
    {
      "epoch": 11.425992779783394,
      "grad_norm": 1016.9796752929688,
      "learning_rate": 0.0001942266821018761,
      "loss": 7.2312,
      "step": 94950
    },
    {
      "epoch": 11.42719614921781,
      "grad_norm": 1361.0908203125,
      "learning_rate": 0.00019422540812443197,
      "loss": 7.2369,
      "step": 94960
    },
    {
      "epoch": 11.428399518652226,
      "grad_norm": 2959.397216796875,
      "learning_rate": 0.00019422413401062042,
      "loss": 7.1497,
      "step": 94970
    },
    {
      "epoch": 11.429602888086643,
      "grad_norm": 896.3115844726562,
      "learning_rate": 0.0001942228597604433,
      "loss": 7.259,
      "step": 94980
    },
    {
      "epoch": 11.43080625752106,
      "grad_norm": 622.7590942382812,
      "learning_rate": 0.00019422158537390244,
      "loss": 7.2424,
      "step": 94990
    },
    {
      "epoch": 11.432009626955475,
      "grad_norm": 891.4320678710938,
      "learning_rate": 0.00019422031085099967,
      "loss": 7.1982,
      "step": 95000
    },
    {
      "epoch": 11.433212996389893,
      "grad_norm": 184.54701232910156,
      "learning_rate": 0.00019421903619173687,
      "loss": 7.1738,
      "step": 95010
    },
    {
      "epoch": 11.434416365824308,
      "grad_norm": 208.2530975341797,
      "learning_rate": 0.00019421776139611583,
      "loss": 7.307,
      "step": 95020
    },
    {
      "epoch": 11.435619735258724,
      "grad_norm": 240.05751037597656,
      "learning_rate": 0.00019421648646413843,
      "loss": 7.1369,
      "step": 95030
    },
    {
      "epoch": 11.43682310469314,
      "grad_norm": 969.1519775390625,
      "learning_rate": 0.00019421521139580654,
      "loss": 7.2613,
      "step": 95040
    },
    {
      "epoch": 11.438026474127557,
      "grad_norm": 282.9772033691406,
      "learning_rate": 0.00019421393619112199,
      "loss": 7.227,
      "step": 95050
    },
    {
      "epoch": 11.439229843561973,
      "grad_norm": 157.27987670898438,
      "learning_rate": 0.00019421266085008657,
      "loss": 7.2907,
      "step": 95060
    },
    {
      "epoch": 11.440433212996389,
      "grad_norm": 596.8047485351562,
      "learning_rate": 0.0001942113853727022,
      "loss": 7.1922,
      "step": 95070
    },
    {
      "epoch": 11.441636582430807,
      "grad_norm": 307.0189208984375,
      "learning_rate": 0.00019421010975897068,
      "loss": 7.2553,
      "step": 95080
    },
    {
      "epoch": 11.442839951865222,
      "grad_norm": 469.5777893066406,
      "learning_rate": 0.00019420883400889384,
      "loss": 7.2223,
      "step": 95090
    },
    {
      "epoch": 11.444043321299638,
      "grad_norm": 287.67572021484375,
      "learning_rate": 0.0001942075581224736,
      "loss": 7.1608,
      "step": 95100
    },
    {
      "epoch": 11.445246690734056,
      "grad_norm": 19147.453125,
      "learning_rate": 0.00019420628209971177,
      "loss": 7.1531,
      "step": 95110
    },
    {
      "epoch": 11.446450060168472,
      "grad_norm": 323.05120849609375,
      "learning_rate": 0.00019420500594061017,
      "loss": 7.1275,
      "step": 95120
    },
    {
      "epoch": 11.447653429602887,
      "grad_norm": 121.5462646484375,
      "learning_rate": 0.0001942037296451707,
      "loss": 7.2484,
      "step": 95130
    },
    {
      "epoch": 11.448856799037305,
      "grad_norm": 149.90122985839844,
      "learning_rate": 0.00019420245321339513,
      "loss": 7.1881,
      "step": 95140
    },
    {
      "epoch": 11.45006016847172,
      "grad_norm": 189.19300842285156,
      "learning_rate": 0.00019420117664528536,
      "loss": 7.1153,
      "step": 95150
    },
    {
      "epoch": 11.451263537906136,
      "grad_norm": 119.75334930419922,
      "learning_rate": 0.00019419989994084324,
      "loss": 7.1951,
      "step": 95160
    },
    {
      "epoch": 11.452466907340554,
      "grad_norm": 118.05400085449219,
      "learning_rate": 0.00019419862310007062,
      "loss": 7.1437,
      "step": 95170
    },
    {
      "epoch": 11.45367027677497,
      "grad_norm": 37.96413040161133,
      "learning_rate": 0.00019419734612296933,
      "loss": 7.0621,
      "step": 95180
    },
    {
      "epoch": 11.454873646209386,
      "grad_norm": 267.73529052734375,
      "learning_rate": 0.00019419606900954124,
      "loss": 7.1646,
      "step": 95190
    },
    {
      "epoch": 11.456077015643803,
      "grad_norm": 177.4149169921875,
      "learning_rate": 0.00019419479175978818,
      "loss": 7.1765,
      "step": 95200
    },
    {
      "epoch": 11.457280385078219,
      "grad_norm": 141.97894287109375,
      "learning_rate": 0.00019419351437371195,
      "loss": 7.124,
      "step": 95210
    },
    {
      "epoch": 11.458483754512635,
      "grad_norm": 296.4188232421875,
      "learning_rate": 0.00019419223685131448,
      "loss": 7.1648,
      "step": 95220
    },
    {
      "epoch": 11.459687123947052,
      "grad_norm": 195.08131408691406,
      "learning_rate": 0.00019419095919259756,
      "loss": 7.1814,
      "step": 95230
    },
    {
      "epoch": 11.460890493381468,
      "grad_norm": 455.2322082519531,
      "learning_rate": 0.00019418968139756312,
      "loss": 7.2621,
      "step": 95240
    },
    {
      "epoch": 11.462093862815884,
      "grad_norm": 163.94717407226562,
      "learning_rate": 0.00019418840346621295,
      "loss": 7.2217,
      "step": 95250
    },
    {
      "epoch": 11.463297232250302,
      "grad_norm": 97.32848358154297,
      "learning_rate": 0.00019418712539854887,
      "loss": 7.2922,
      "step": 95260
    },
    {
      "epoch": 11.464500601684717,
      "grad_norm": 183.0526580810547,
      "learning_rate": 0.00019418584719457278,
      "loss": 7.1507,
      "step": 95270
    },
    {
      "epoch": 11.465703971119133,
      "grad_norm": 32.27583312988281,
      "learning_rate": 0.00019418456885428653,
      "loss": 7.1735,
      "step": 95280
    },
    {
      "epoch": 11.46690734055355,
      "grad_norm": 98.2454605102539,
      "learning_rate": 0.00019418329037769195,
      "loss": 7.1715,
      "step": 95290
    },
    {
      "epoch": 11.468110709987966,
      "grad_norm": 139.91546630859375,
      "learning_rate": 0.00019418201176479087,
      "loss": 7.0475,
      "step": 95300
    },
    {
      "epoch": 11.469314079422382,
      "grad_norm": 183.52671813964844,
      "learning_rate": 0.0001941807330155852,
      "loss": 7.2947,
      "step": 95310
    },
    {
      "epoch": 11.4705174488568,
      "grad_norm": 275.2720947265625,
      "learning_rate": 0.00019417945413007675,
      "loss": 7.1885,
      "step": 95320
    },
    {
      "epoch": 11.471720818291216,
      "grad_norm": 218.15499877929688,
      "learning_rate": 0.00019417817510826736,
      "loss": 7.2804,
      "step": 95330
    },
    {
      "epoch": 11.472924187725631,
      "grad_norm": 71.13890838623047,
      "learning_rate": 0.0001941768959501589,
      "loss": 7.2736,
      "step": 95340
    },
    {
      "epoch": 11.474127557160049,
      "grad_norm": 68.12261962890625,
      "learning_rate": 0.00019417561665575325,
      "loss": 7.2383,
      "step": 95350
    },
    {
      "epoch": 11.475330926594465,
      "grad_norm": 101.7570571899414,
      "learning_rate": 0.0001941743372250522,
      "loss": 7.2648,
      "step": 95360
    },
    {
      "epoch": 11.47653429602888,
      "grad_norm": 293.007080078125,
      "learning_rate": 0.00019417305765805765,
      "loss": 7.1892,
      "step": 95370
    },
    {
      "epoch": 11.477737665463298,
      "grad_norm": 74.36763000488281,
      "learning_rate": 0.0001941717779547714,
      "loss": 7.145,
      "step": 95380
    },
    {
      "epoch": 11.478941034897714,
      "grad_norm": 29.627981185913086,
      "learning_rate": 0.0001941704981151954,
      "loss": 7.2334,
      "step": 95390
    },
    {
      "epoch": 11.48014440433213,
      "grad_norm": 37.16349411010742,
      "learning_rate": 0.0001941692181393314,
      "loss": 7.2144,
      "step": 95400
    },
    {
      "epoch": 11.481347773766545,
      "grad_norm": 96.40779113769531,
      "learning_rate": 0.0001941679380271813,
      "loss": 7.3325,
      "step": 95410
    },
    {
      "epoch": 11.482551143200963,
      "grad_norm": 22.974285125732422,
      "learning_rate": 0.00019416665777874693,
      "loss": 7.3191,
      "step": 95420
    },
    {
      "epoch": 11.483754512635379,
      "grad_norm": 50.43960189819336,
      "learning_rate": 0.00019416537739403017,
      "loss": 7.23,
      "step": 95430
    },
    {
      "epoch": 11.484957882069795,
      "grad_norm": 23.416912078857422,
      "learning_rate": 0.00019416409687303285,
      "loss": 7.2831,
      "step": 95440
    },
    {
      "epoch": 11.486161251504212,
      "grad_norm": 207.31910705566406,
      "learning_rate": 0.00019416281621575685,
      "loss": 7.1682,
      "step": 95450
    },
    {
      "epoch": 11.487364620938628,
      "grad_norm": 25.098697662353516,
      "learning_rate": 0.000194161535422204,
      "loss": 7.3127,
      "step": 95460
    },
    {
      "epoch": 11.488567990373044,
      "grad_norm": 333.8545227050781,
      "learning_rate": 0.00019416025449237617,
      "loss": 7.1712,
      "step": 95470
    },
    {
      "epoch": 11.489771359807461,
      "grad_norm": 70.67992401123047,
      "learning_rate": 0.0001941589734262752,
      "loss": 7.1905,
      "step": 95480
    },
    {
      "epoch": 11.490974729241877,
      "grad_norm": 44.74904251098633,
      "learning_rate": 0.00019415769222390293,
      "loss": 7.2526,
      "step": 95490
    },
    {
      "epoch": 11.492178098676293,
      "grad_norm": 164.74118041992188,
      "learning_rate": 0.00019415641088526124,
      "loss": 7.201,
      "step": 95500
    },
    {
      "epoch": 11.49338146811071,
      "grad_norm": 30.1320743560791,
      "learning_rate": 0.00019415512941035198,
      "loss": 7.2024,
      "step": 95510
    },
    {
      "epoch": 11.494584837545126,
      "grad_norm": 2078.109375,
      "learning_rate": 0.00019415384779917702,
      "loss": 7.2331,
      "step": 95520
    },
    {
      "epoch": 11.495788206979542,
      "grad_norm": 5247.37939453125,
      "learning_rate": 0.00019415256605173818,
      "loss": 7.2696,
      "step": 95530
    },
    {
      "epoch": 11.49699157641396,
      "grad_norm": 19246.5703125,
      "learning_rate": 0.00019415128416803735,
      "loss": 7.3531,
      "step": 95540
    },
    {
      "epoch": 11.498194945848375,
      "grad_norm": 6615.9482421875,
      "learning_rate": 0.00019415000214807632,
      "loss": 7.3087,
      "step": 95550
    },
    {
      "epoch": 11.499398315282791,
      "grad_norm": 344.5341491699219,
      "learning_rate": 0.000194148719991857,
      "loss": 7.2626,
      "step": 95560
    },
    {
      "epoch": 11.500601684717209,
      "grad_norm": 391.76593017578125,
      "learning_rate": 0.00019414743769938128,
      "loss": 7.1514,
      "step": 95570
    },
    {
      "epoch": 11.501805054151625,
      "grad_norm": 4885.11474609375,
      "learning_rate": 0.00019414615527065093,
      "loss": 7.3166,
      "step": 95580
    },
    {
      "epoch": 11.50300842358604,
      "grad_norm": 3420.669921875,
      "learning_rate": 0.0001941448727056679,
      "loss": 7.259,
      "step": 95590
    },
    {
      "epoch": 11.504211793020458,
      "grad_norm": 4549.19873046875,
      "learning_rate": 0.00019414359000443395,
      "loss": 7.2412,
      "step": 95600
    },
    {
      "epoch": 11.505415162454874,
      "grad_norm": 162.44728088378906,
      "learning_rate": 0.000194142307166951,
      "loss": 7.2549,
      "step": 95610
    },
    {
      "epoch": 11.50661853188929,
      "grad_norm": 646.1400146484375,
      "learning_rate": 0.00019414102419322088,
      "loss": 7.3019,
      "step": 95620
    },
    {
      "epoch": 11.507821901323707,
      "grad_norm": 660.2918701171875,
      "learning_rate": 0.0001941397410832455,
      "loss": 7.2229,
      "step": 95630
    },
    {
      "epoch": 11.509025270758123,
      "grad_norm": 108.86157989501953,
      "learning_rate": 0.00019413845783702662,
      "loss": 7.1886,
      "step": 95640
    },
    {
      "epoch": 11.510228640192539,
      "grad_norm": 685.5956420898438,
      "learning_rate": 0.00019413717445456615,
      "loss": 7.212,
      "step": 95650
    },
    {
      "epoch": 11.511432009626956,
      "grad_norm": 1988.319580078125,
      "learning_rate": 0.00019413589093586597,
      "loss": 7.2999,
      "step": 95660
    },
    {
      "epoch": 11.512635379061372,
      "grad_norm": 2318.384033203125,
      "learning_rate": 0.0001941346072809279,
      "loss": 7.1294,
      "step": 95670
    },
    {
      "epoch": 11.513838748495788,
      "grad_norm": 308.8346252441406,
      "learning_rate": 0.00019413332348975382,
      "loss": 7.2871,
      "step": 95680
    },
    {
      "epoch": 11.515042117930205,
      "grad_norm": 178.97772216796875,
      "learning_rate": 0.00019413203956234558,
      "loss": 7.204,
      "step": 95690
    },
    {
      "epoch": 11.516245487364621,
      "grad_norm": 5297.19921875,
      "learning_rate": 0.00019413075549870503,
      "loss": 7.2808,
      "step": 95700
    },
    {
      "epoch": 11.517448856799037,
      "grad_norm": 1700379.125,
      "learning_rate": 0.00019412947129883405,
      "loss": 7.282,
      "step": 95710
    },
    {
      "epoch": 11.518652226233453,
      "grad_norm": 7690.8837890625,
      "learning_rate": 0.0001941281869627345,
      "loss": 7.2673,
      "step": 95720
    },
    {
      "epoch": 11.51985559566787,
      "grad_norm": 3003.178955078125,
      "learning_rate": 0.00019412690249040818,
      "loss": 7.2425,
      "step": 95730
    },
    {
      "epoch": 11.521058965102286,
      "grad_norm": 509.0173034667969,
      "learning_rate": 0.00019412561788185702,
      "loss": 7.2437,
      "step": 95740
    },
    {
      "epoch": 11.522262334536702,
      "grad_norm": 385.6251220703125,
      "learning_rate": 0.00019412433313708283,
      "loss": 7.3698,
      "step": 95750
    },
    {
      "epoch": 11.52346570397112,
      "grad_norm": 5005.37109375,
      "learning_rate": 0.0001941230482560875,
      "loss": 7.4122,
      "step": 95760
    },
    {
      "epoch": 11.524669073405535,
      "grad_norm": 13992.76171875,
      "learning_rate": 0.0001941217632388729,
      "loss": 7.4523,
      "step": 95770
    },
    {
      "epoch": 11.525872442839951,
      "grad_norm": 617.8475341796875,
      "learning_rate": 0.00019412047808544088,
      "loss": 7.4705,
      "step": 95780
    },
    {
      "epoch": 11.527075812274369,
      "grad_norm": 1462.038818359375,
      "learning_rate": 0.00019411919279579328,
      "loss": 7.4208,
      "step": 95790
    },
    {
      "epoch": 11.528279181708784,
      "grad_norm": 3715.39892578125,
      "learning_rate": 0.00019411790736993193,
      "loss": 7.3372,
      "step": 95800
    },
    {
      "epoch": 11.5294825511432,
      "grad_norm": 2403.631103515625,
      "learning_rate": 0.00019411662180785877,
      "loss": 7.3452,
      "step": 95810
    },
    {
      "epoch": 11.530685920577618,
      "grad_norm": 791.82958984375,
      "learning_rate": 0.00019411533610957563,
      "loss": 7.4349,
      "step": 95820
    },
    {
      "epoch": 11.531889290012034,
      "grad_norm": 471.4733581542969,
      "learning_rate": 0.00019411405027508434,
      "loss": 7.5038,
      "step": 95830
    },
    {
      "epoch": 11.53309265944645,
      "grad_norm": 306.58074951171875,
      "learning_rate": 0.0001941127643043868,
      "loss": 7.4147,
      "step": 95840
    },
    {
      "epoch": 11.534296028880867,
      "grad_norm": 707.7532958984375,
      "learning_rate": 0.00019411147819748482,
      "loss": 7.4396,
      "step": 95850
    },
    {
      "epoch": 11.535499398315283,
      "grad_norm": 649.7509765625,
      "learning_rate": 0.0001941101919543803,
      "loss": 7.3674,
      "step": 95860
    },
    {
      "epoch": 11.536702767749698,
      "grad_norm": 121.48490142822266,
      "learning_rate": 0.00019410890557507515,
      "loss": 7.3876,
      "step": 95870
    },
    {
      "epoch": 11.537906137184116,
      "grad_norm": 56.903682708740234,
      "learning_rate": 0.00019410761905957113,
      "loss": 7.4034,
      "step": 95880
    },
    {
      "epoch": 11.539109506618532,
      "grad_norm": 768.2057495117188,
      "learning_rate": 0.00019410633240787018,
      "loss": 7.4109,
      "step": 95890
    },
    {
      "epoch": 11.540312876052948,
      "grad_norm": 297.0628356933594,
      "learning_rate": 0.0001941050456199741,
      "loss": 7.3977,
      "step": 95900
    },
    {
      "epoch": 11.541516245487365,
      "grad_norm": 496.7565002441406,
      "learning_rate": 0.0001941037586958848,
      "loss": 7.5049,
      "step": 95910
    },
    {
      "epoch": 11.542719614921781,
      "grad_norm": 729.7393188476562,
      "learning_rate": 0.00019410247163560414,
      "loss": 7.3943,
      "step": 95920
    },
    {
      "epoch": 11.543922984356197,
      "grad_norm": 62.40972900390625,
      "learning_rate": 0.00019410118443913396,
      "loss": 7.4299,
      "step": 95930
    },
    {
      "epoch": 11.545126353790614,
      "grad_norm": 454.97454833984375,
      "learning_rate": 0.00019409989710647617,
      "loss": 7.4787,
      "step": 95940
    },
    {
      "epoch": 11.54632972322503,
      "grad_norm": 209.5741729736328,
      "learning_rate": 0.00019409860963763253,
      "loss": 7.4322,
      "step": 95950
    },
    {
      "epoch": 11.547533092659446,
      "grad_norm": 968.0933227539062,
      "learning_rate": 0.000194097322032605,
      "loss": 7.4145,
      "step": 95960
    },
    {
      "epoch": 11.548736462093864,
      "grad_norm": 4711.44921875,
      "learning_rate": 0.00019409603429139543,
      "loss": 7.9339,
      "step": 95970
    },
    {
      "epoch": 11.54993983152828,
      "grad_norm": 4932.09619140625,
      "learning_rate": 0.00019409474641400565,
      "loss": 8.1511,
      "step": 95980
    },
    {
      "epoch": 11.551143200962695,
      "grad_norm": 1032.34716796875,
      "learning_rate": 0.00019409345840043754,
      "loss": 8.1253,
      "step": 95990
    },
    {
      "epoch": 11.552346570397113,
      "grad_norm": 1271.8768310546875,
      "learning_rate": 0.00019409217025069295,
      "loss": 7.858,
      "step": 96000
    },
    {
      "epoch": 11.553549939831528,
      "grad_norm": 2195.34326171875,
      "learning_rate": 0.0001940908819647738,
      "loss": 7.6339,
      "step": 96010
    },
    {
      "epoch": 11.554753309265944,
      "grad_norm": 874.3984375,
      "learning_rate": 0.00019408959354268187,
      "loss": 7.4137,
      "step": 96020
    },
    {
      "epoch": 11.555956678700362,
      "grad_norm": 886.2830810546875,
      "learning_rate": 0.0001940883049844191,
      "loss": 7.4869,
      "step": 96030
    },
    {
      "epoch": 11.557160048134778,
      "grad_norm": 1172.5394287109375,
      "learning_rate": 0.0001940870162899873,
      "loss": 7.4175,
      "step": 96040
    },
    {
      "epoch": 11.558363417569193,
      "grad_norm": 815.3179321289062,
      "learning_rate": 0.00019408572745938836,
      "loss": 7.4108,
      "step": 96050
    },
    {
      "epoch": 11.559566787003611,
      "grad_norm": 748.6189575195312,
      "learning_rate": 0.00019408443849262418,
      "loss": 7.2286,
      "step": 96060
    },
    {
      "epoch": 11.560770156438027,
      "grad_norm": 2582.579833984375,
      "learning_rate": 0.00019408314938969653,
      "loss": 7.3017,
      "step": 96070
    },
    {
      "epoch": 11.561973525872443,
      "grad_norm": 1788.675537109375,
      "learning_rate": 0.0001940818601506074,
      "loss": 7.3826,
      "step": 96080
    },
    {
      "epoch": 11.56317689530686,
      "grad_norm": 710.4325561523438,
      "learning_rate": 0.00019408057077535852,
      "loss": 7.3341,
      "step": 96090
    },
    {
      "epoch": 11.564380264741276,
      "grad_norm": 6406.4833984375,
      "learning_rate": 0.0001940792812639519,
      "loss": 7.3452,
      "step": 96100
    },
    {
      "epoch": 11.565583634175692,
      "grad_norm": 539.2433471679688,
      "learning_rate": 0.00019407799161638928,
      "loss": 7.3412,
      "step": 96110
    },
    {
      "epoch": 11.566787003610107,
      "grad_norm": 505.794189453125,
      "learning_rate": 0.0001940767018326726,
      "loss": 7.3656,
      "step": 96120
    },
    {
      "epoch": 11.567990373044525,
      "grad_norm": 2025.591064453125,
      "learning_rate": 0.00019407541191280369,
      "loss": 7.3555,
      "step": 96130
    },
    {
      "epoch": 11.56919374247894,
      "grad_norm": 1910.954833984375,
      "learning_rate": 0.00019407412185678443,
      "loss": 7.3506,
      "step": 96140
    },
    {
      "epoch": 11.570397111913357,
      "grad_norm": 3230.227294921875,
      "learning_rate": 0.00019407283166461673,
      "loss": 7.3829,
      "step": 96150
    },
    {
      "epoch": 11.571600481347774,
      "grad_norm": 5494.11181640625,
      "learning_rate": 0.0001940715413363024,
      "loss": 7.4108,
      "step": 96160
    },
    {
      "epoch": 11.57280385078219,
      "grad_norm": 692.5964965820312,
      "learning_rate": 0.00019407025087184328,
      "loss": 7.4228,
      "step": 96170
    },
    {
      "epoch": 11.574007220216606,
      "grad_norm": 102.73624420166016,
      "learning_rate": 0.00019406896027124133,
      "loss": 7.3413,
      "step": 96180
    },
    {
      "epoch": 11.575210589651023,
      "grad_norm": 40.074222564697266,
      "learning_rate": 0.00019406766953449834,
      "loss": 7.3512,
      "step": 96190
    },
    {
      "epoch": 11.57641395908544,
      "grad_norm": 61.32003402709961,
      "learning_rate": 0.00019406637866161623,
      "loss": 7.2595,
      "step": 96200
    },
    {
      "epoch": 11.577617328519855,
      "grad_norm": 109.75365447998047,
      "learning_rate": 0.00019406508765259684,
      "loss": 7.4006,
      "step": 96210
    },
    {
      "epoch": 11.578820697954272,
      "grad_norm": 111.11814880371094,
      "learning_rate": 0.00019406379650744205,
      "loss": 7.3753,
      "step": 96220
    },
    {
      "epoch": 11.580024067388688,
      "grad_norm": 67.07972717285156,
      "learning_rate": 0.0001940625052261537,
      "loss": 7.2511,
      "step": 96230
    },
    {
      "epoch": 11.581227436823104,
      "grad_norm": 214.6697235107422,
      "learning_rate": 0.0001940612138087337,
      "loss": 7.3164,
      "step": 96240
    },
    {
      "epoch": 11.582430806257522,
      "grad_norm": 71.9437484741211,
      "learning_rate": 0.0001940599222551839,
      "loss": 7.3436,
      "step": 96250
    },
    {
      "epoch": 11.583634175691937,
      "grad_norm": 48.210391998291016,
      "learning_rate": 0.00019405863056550616,
      "loss": 7.3452,
      "step": 96260
    },
    {
      "epoch": 11.584837545126353,
      "grad_norm": 179.23489379882812,
      "learning_rate": 0.00019405733873970236,
      "loss": 7.2433,
      "step": 96270
    },
    {
      "epoch": 11.58604091456077,
      "grad_norm": 57.60596466064453,
      "learning_rate": 0.00019405604677777438,
      "loss": 7.2694,
      "step": 96280
    },
    {
      "epoch": 11.587244283995187,
      "grad_norm": 72.04112243652344,
      "learning_rate": 0.00019405475467972408,
      "loss": 7.2567,
      "step": 96290
    },
    {
      "epoch": 11.588447653429602,
      "grad_norm": 80.52898406982422,
      "learning_rate": 0.0001940534624455533,
      "loss": 7.2869,
      "step": 96300
    },
    {
      "epoch": 11.58965102286402,
      "grad_norm": 179.2220001220703,
      "learning_rate": 0.000194052170075264,
      "loss": 7.295,
      "step": 96310
    },
    {
      "epoch": 11.590854392298436,
      "grad_norm": 469.15692138671875,
      "learning_rate": 0.00019405087756885791,
      "loss": 7.2984,
      "step": 96320
    },
    {
      "epoch": 11.592057761732852,
      "grad_norm": 278.82354736328125,
      "learning_rate": 0.00019404958492633703,
      "loss": 7.2517,
      "step": 96330
    },
    {
      "epoch": 11.593261131167269,
      "grad_norm": 615.7554321289062,
      "learning_rate": 0.0001940482921477032,
      "loss": 7.3759,
      "step": 96340
    },
    {
      "epoch": 11.594464500601685,
      "grad_norm": 141.1912841796875,
      "learning_rate": 0.00019404699923295825,
      "loss": 7.3484,
      "step": 96350
    },
    {
      "epoch": 11.5956678700361,
      "grad_norm": 206.26466369628906,
      "learning_rate": 0.00019404570618210403,
      "loss": 7.2216,
      "step": 96360
    },
    {
      "epoch": 11.596871239470518,
      "grad_norm": 51.306732177734375,
      "learning_rate": 0.00019404441299514252,
      "loss": 7.225,
      "step": 96370
    },
    {
      "epoch": 11.598074608904934,
      "grad_norm": 50.382686614990234,
      "learning_rate": 0.00019404311967207547,
      "loss": 7.2798,
      "step": 96380
    },
    {
      "epoch": 11.59927797833935,
      "grad_norm": 76.64035034179688,
      "learning_rate": 0.00019404182621290484,
      "loss": 7.3731,
      "step": 96390
    },
    {
      "epoch": 11.600481347773766,
      "grad_norm": 70.0130615234375,
      "learning_rate": 0.00019404053261763243,
      "loss": 7.3046,
      "step": 96400
    },
    {
      "epoch": 11.601684717208183,
      "grad_norm": 44.97519302368164,
      "learning_rate": 0.00019403923888626018,
      "loss": 7.2776,
      "step": 96410
    },
    {
      "epoch": 11.602888086642599,
      "grad_norm": 79.91014862060547,
      "learning_rate": 0.00019403794501878996,
      "loss": 7.2753,
      "step": 96420
    },
    {
      "epoch": 11.604091456077015,
      "grad_norm": 88.37982177734375,
      "learning_rate": 0.00019403665101522358,
      "loss": 7.2956,
      "step": 96430
    },
    {
      "epoch": 11.605294825511432,
      "grad_norm": 53.483333587646484,
      "learning_rate": 0.000194035356875563,
      "loss": 7.2445,
      "step": 96440
    },
    {
      "epoch": 11.606498194945848,
      "grad_norm": 46.7051887512207,
      "learning_rate": 0.00019403406259980996,
      "loss": 7.2775,
      "step": 96450
    },
    {
      "epoch": 11.607701564380264,
      "grad_norm": 167.72657775878906,
      "learning_rate": 0.0001940327681879665,
      "loss": 7.2743,
      "step": 96460
    },
    {
      "epoch": 11.608904933814681,
      "grad_norm": 63.070072174072266,
      "learning_rate": 0.00019403147364003434,
      "loss": 7.2868,
      "step": 96470
    },
    {
      "epoch": 11.610108303249097,
      "grad_norm": 104.51396179199219,
      "learning_rate": 0.00019403017895601546,
      "loss": 7.2208,
      "step": 96480
    },
    {
      "epoch": 11.611311672683513,
      "grad_norm": 199.52304077148438,
      "learning_rate": 0.00019402888413591168,
      "loss": 7.2697,
      "step": 96490
    },
    {
      "epoch": 11.61251504211793,
      "grad_norm": 88.67973327636719,
      "learning_rate": 0.00019402758917972488,
      "loss": 7.3066,
      "step": 96500
    },
    {
      "epoch": 11.613718411552346,
      "grad_norm": 62.45795440673828,
      "learning_rate": 0.00019402629408745697,
      "loss": 7.281,
      "step": 96510
    },
    {
      "epoch": 11.614921780986762,
      "grad_norm": 64.20278930664062,
      "learning_rate": 0.0001940249988591098,
      "loss": 7.3778,
      "step": 96520
    },
    {
      "epoch": 11.61612515042118,
      "grad_norm": 196.82647705078125,
      "learning_rate": 0.0001940237034946852,
      "loss": 7.2641,
      "step": 96530
    },
    {
      "epoch": 11.617328519855596,
      "grad_norm": 95.7018051147461,
      "learning_rate": 0.00019402240799418511,
      "loss": 7.2731,
      "step": 96540
    },
    {
      "epoch": 11.618531889290011,
      "grad_norm": 155.0595245361328,
      "learning_rate": 0.0001940211123576114,
      "loss": 7.1003,
      "step": 96550
    },
    {
      "epoch": 11.619735258724429,
      "grad_norm": 117.70222473144531,
      "learning_rate": 0.0001940198165849659,
      "loss": 7.1122,
      "step": 96560
    },
    {
      "epoch": 11.620938628158845,
      "grad_norm": 295.2709045410156,
      "learning_rate": 0.00019401852067625055,
      "loss": 7.2431,
      "step": 96570
    },
    {
      "epoch": 11.62214199759326,
      "grad_norm": 125.61164855957031,
      "learning_rate": 0.00019401722463146718,
      "loss": 7.1774,
      "step": 96580
    },
    {
      "epoch": 11.623345367027678,
      "grad_norm": 274.9051208496094,
      "learning_rate": 0.00019401592845061765,
      "loss": 7.181,
      "step": 96590
    },
    {
      "epoch": 11.624548736462094,
      "grad_norm": 170.54946899414062,
      "learning_rate": 0.00019401463213370386,
      "loss": 7.3159,
      "step": 96600
    },
    {
      "epoch": 11.62575210589651,
      "grad_norm": 1383.941162109375,
      "learning_rate": 0.0001940133356807277,
      "loss": 7.4157,
      "step": 96610
    },
    {
      "epoch": 11.626955475330927,
      "grad_norm": 312.02886962890625,
      "learning_rate": 0.00019401203909169104,
      "loss": 7.2948,
      "step": 96620
    },
    {
      "epoch": 11.628158844765343,
      "grad_norm": 293.3307800292969,
      "learning_rate": 0.00019401074236659573,
      "loss": 7.2833,
      "step": 96630
    },
    {
      "epoch": 11.629362214199759,
      "grad_norm": 329.22216796875,
      "learning_rate": 0.00019400944550544366,
      "loss": 7.1939,
      "step": 96640
    },
    {
      "epoch": 11.630565583634176,
      "grad_norm": 107.05670166015625,
      "learning_rate": 0.00019400814850823673,
      "loss": 7.426,
      "step": 96650
    },
    {
      "epoch": 11.631768953068592,
      "grad_norm": 210.57260131835938,
      "learning_rate": 0.0001940068513749768,
      "loss": 7.3313,
      "step": 96660
    },
    {
      "epoch": 11.632972322503008,
      "grad_norm": 202.15170288085938,
      "learning_rate": 0.00019400555410566575,
      "loss": 7.2733,
      "step": 96670
    },
    {
      "epoch": 11.634175691937426,
      "grad_norm": 318.3314208984375,
      "learning_rate": 0.00019400425670030544,
      "loss": 7.268,
      "step": 96680
    },
    {
      "epoch": 11.635379061371841,
      "grad_norm": 259.2880554199219,
      "learning_rate": 0.00019400295915889774,
      "loss": 7.1044,
      "step": 96690
    },
    {
      "epoch": 11.636582430806257,
      "grad_norm": 308.6136779785156,
      "learning_rate": 0.0001940016614814446,
      "loss": 7.3616,
      "step": 96700
    },
    {
      "epoch": 11.637785800240675,
      "grad_norm": 105.45771789550781,
      "learning_rate": 0.0001940003636679478,
      "loss": 7.3933,
      "step": 96710
    },
    {
      "epoch": 11.63898916967509,
      "grad_norm": 121.55425262451172,
      "learning_rate": 0.00019399906571840932,
      "loss": 7.3401,
      "step": 96720
    },
    {
      "epoch": 11.640192539109506,
      "grad_norm": 191.5036163330078,
      "learning_rate": 0.00019399776763283092,
      "loss": 7.2856,
      "step": 96730
    },
    {
      "epoch": 11.641395908543924,
      "grad_norm": 176.63304138183594,
      "learning_rate": 0.0001939964694112146,
      "loss": 7.1382,
      "step": 96740
    },
    {
      "epoch": 11.64259927797834,
      "grad_norm": 114.03582000732422,
      "learning_rate": 0.00019399517105356216,
      "loss": 7.1845,
      "step": 96750
    },
    {
      "epoch": 11.643802647412755,
      "grad_norm": 489.2210998535156,
      "learning_rate": 0.00019399387255987548,
      "loss": 7.3281,
      "step": 96760
    },
    {
      "epoch": 11.645006016847173,
      "grad_norm": 102.50401306152344,
      "learning_rate": 0.0001939925739301565,
      "loss": 7.1657,
      "step": 96770
    },
    {
      "epoch": 11.646209386281589,
      "grad_norm": 61.25798034667969,
      "learning_rate": 0.000193991275164407,
      "loss": 7.3006,
      "step": 96780
    },
    {
      "epoch": 11.647412755716005,
      "grad_norm": 84.86211395263672,
      "learning_rate": 0.00019398997626262897,
      "loss": 7.232,
      "step": 96790
    },
    {
      "epoch": 11.648616125150422,
      "grad_norm": 103.94781494140625,
      "learning_rate": 0.00019398867722482423,
      "loss": 7.2801,
      "step": 96800
    },
    {
      "epoch": 11.649819494584838,
      "grad_norm": 139.27731323242188,
      "learning_rate": 0.00019398737805099468,
      "loss": 7.1559,
      "step": 96810
    },
    {
      "epoch": 11.651022864019254,
      "grad_norm": 275.8272705078125,
      "learning_rate": 0.00019398607874114218,
      "loss": 7.2544,
      "step": 96820
    },
    {
      "epoch": 11.65222623345367,
      "grad_norm": 184.69407653808594,
      "learning_rate": 0.0001939847792952686,
      "loss": 7.2527,
      "step": 96830
    },
    {
      "epoch": 11.653429602888087,
      "grad_norm": 66.01103973388672,
      "learning_rate": 0.00019398347971337585,
      "loss": 7.2175,
      "step": 96840
    },
    {
      "epoch": 11.654632972322503,
      "grad_norm": 142.17613220214844,
      "learning_rate": 0.0001939821799954658,
      "loss": 7.247,
      "step": 96850
    },
    {
      "epoch": 11.655836341756919,
      "grad_norm": 215.30686950683594,
      "learning_rate": 0.00019398088014154036,
      "loss": 7.1864,
      "step": 96860
    },
    {
      "epoch": 11.657039711191336,
      "grad_norm": 69.89537811279297,
      "learning_rate": 0.00019397958015160137,
      "loss": 7.2055,
      "step": 96870
    },
    {
      "epoch": 11.658243080625752,
      "grad_norm": 200.43955993652344,
      "learning_rate": 0.0001939782800256507,
      "loss": 7.2436,
      "step": 96880
    },
    {
      "epoch": 11.659446450060168,
      "grad_norm": 146.6250762939453,
      "learning_rate": 0.00019397697976369026,
      "loss": 7.1778,
      "step": 96890
    },
    {
      "epoch": 11.660649819494585,
      "grad_norm": 66.77310943603516,
      "learning_rate": 0.00019397567936572198,
      "loss": 7.2166,
      "step": 96900
    },
    {
      "epoch": 11.661853188929001,
      "grad_norm": 341.8200378417969,
      "learning_rate": 0.00019397437883174763,
      "loss": 7.2666,
      "step": 96910
    },
    {
      "epoch": 11.663056558363417,
      "grad_norm": 851.1067504882812,
      "learning_rate": 0.00019397307816176917,
      "loss": 7.2638,
      "step": 96920
    },
    {
      "epoch": 11.664259927797834,
      "grad_norm": 226.024658203125,
      "learning_rate": 0.00019397177735578847,
      "loss": 7.3104,
      "step": 96930
    },
    {
      "epoch": 11.66546329723225,
      "grad_norm": 277.4878845214844,
      "learning_rate": 0.0001939704764138074,
      "loss": 7.2567,
      "step": 96940
    },
    {
      "epoch": 11.666666666666666,
      "grad_norm": 306.32806396484375,
      "learning_rate": 0.00019396917533582783,
      "loss": 7.2011,
      "step": 96950
    },
    {
      "epoch": 11.667870036101084,
      "grad_norm": 230.68382263183594,
      "learning_rate": 0.00019396787412185172,
      "loss": 7.3227,
      "step": 96960
    },
    {
      "epoch": 11.6690734055355,
      "grad_norm": 606.9068603515625,
      "learning_rate": 0.00019396657277188087,
      "loss": 7.1708,
      "step": 96970
    },
    {
      "epoch": 11.670276774969915,
      "grad_norm": 732.1396484375,
      "learning_rate": 0.00019396527128591717,
      "loss": 7.2709,
      "step": 96980
    },
    {
      "epoch": 11.671480144404333,
      "grad_norm": 315.5829162597656,
      "learning_rate": 0.00019396396966396253,
      "loss": 7.1769,
      "step": 96990
    },
    {
      "epoch": 11.672683513838749,
      "grad_norm": 602.1509399414062,
      "learning_rate": 0.00019396266790601882,
      "loss": 7.1419,
      "step": 97000
    },
    {
      "epoch": 11.673886883273164,
      "grad_norm": 233.7982940673828,
      "learning_rate": 0.00019396136601208797,
      "loss": 7.249,
      "step": 97010
    },
    {
      "epoch": 11.675090252707582,
      "grad_norm": 234.9595184326172,
      "learning_rate": 0.0001939600639821718,
      "loss": 7.2539,
      "step": 97020
    },
    {
      "epoch": 11.676293622141998,
      "grad_norm": 443.2006530761719,
      "learning_rate": 0.00019395876181627222,
      "loss": 7.2811,
      "step": 97030
    },
    {
      "epoch": 11.677496991576414,
      "grad_norm": 337.4075927734375,
      "learning_rate": 0.0001939574595143911,
      "loss": 7.1453,
      "step": 97040
    },
    {
      "epoch": 11.678700361010831,
      "grad_norm": 519.0325927734375,
      "learning_rate": 0.00019395615707653034,
      "loss": 7.1625,
      "step": 97050
    },
    {
      "epoch": 11.679903730445247,
      "grad_norm": 447.83380126953125,
      "learning_rate": 0.00019395485450269186,
      "loss": 7.1652,
      "step": 97060
    },
    {
      "epoch": 11.681107099879663,
      "grad_norm": 695.3973999023438,
      "learning_rate": 0.0001939535517928775,
      "loss": 7.2403,
      "step": 97070
    },
    {
      "epoch": 11.68231046931408,
      "grad_norm": 718.5765380859375,
      "learning_rate": 0.00019395224894708913,
      "loss": 7.226,
      "step": 97080
    },
    {
      "epoch": 11.683513838748496,
      "grad_norm": 324.5961608886719,
      "learning_rate": 0.00019395094596532864,
      "loss": 7.2867,
      "step": 97090
    },
    {
      "epoch": 11.684717208182912,
      "grad_norm": 608.708984375,
      "learning_rate": 0.00019394964284759797,
      "loss": 7.2458,
      "step": 97100
    },
    {
      "epoch": 11.685920577617328,
      "grad_norm": 3069.66796875,
      "learning_rate": 0.00019394833959389897,
      "loss": 7.2379,
      "step": 97110
    },
    {
      "epoch": 11.687123947051745,
      "grad_norm": 1491.3551025390625,
      "learning_rate": 0.00019394703620423355,
      "loss": 7.2741,
      "step": 97120
    },
    {
      "epoch": 11.688327316486161,
      "grad_norm": 933.3352661132812,
      "learning_rate": 0.00019394573267860353,
      "loss": 7.2918,
      "step": 97130
    },
    {
      "epoch": 11.689530685920577,
      "grad_norm": 2231.718505859375,
      "learning_rate": 0.00019394442901701084,
      "loss": 7.1884,
      "step": 97140
    },
    {
      "epoch": 11.690734055354994,
      "grad_norm": 1130.867919921875,
      "learning_rate": 0.00019394312521945739,
      "loss": 7.2346,
      "step": 97150
    },
    {
      "epoch": 11.69193742478941,
      "grad_norm": 1030.5931396484375,
      "learning_rate": 0.00019394182128594506,
      "loss": 7.2024,
      "step": 97160
    },
    {
      "epoch": 11.693140794223826,
      "grad_norm": 627.8846435546875,
      "learning_rate": 0.00019394051721647565,
      "loss": 7.2279,
      "step": 97170
    },
    {
      "epoch": 11.694344163658243,
      "grad_norm": 735.602783203125,
      "learning_rate": 0.0001939392130110512,
      "loss": 7.347,
      "step": 97180
    },
    {
      "epoch": 11.69554753309266,
      "grad_norm": 1029.7615966796875,
      "learning_rate": 0.00019393790866967348,
      "loss": 7.1682,
      "step": 97190
    },
    {
      "epoch": 11.696750902527075,
      "grad_norm": 956.34716796875,
      "learning_rate": 0.0001939366041923444,
      "loss": 7.2551,
      "step": 97200
    },
    {
      "epoch": 11.697954271961493,
      "grad_norm": 1820.06103515625,
      "learning_rate": 0.00019393529957906588,
      "loss": 7.1317,
      "step": 97210
    },
    {
      "epoch": 11.699157641395908,
      "grad_norm": 957.2282104492188,
      "learning_rate": 0.0001939339948298398,
      "loss": 7.224,
      "step": 97220
    },
    {
      "epoch": 11.700361010830324,
      "grad_norm": 2351.061279296875,
      "learning_rate": 0.00019393268994466803,
      "loss": 7.1606,
      "step": 97230
    },
    {
      "epoch": 11.701564380264742,
      "grad_norm": 1216.29638671875,
      "learning_rate": 0.00019393138492355245,
      "loss": 7.269,
      "step": 97240
    },
    {
      "epoch": 11.702767749699158,
      "grad_norm": 1314.3826904296875,
      "learning_rate": 0.00019393007976649499,
      "loss": 7.2536,
      "step": 97250
    },
    {
      "epoch": 11.703971119133573,
      "grad_norm": 1006.3629760742188,
      "learning_rate": 0.0001939287744734975,
      "loss": 7.3008,
      "step": 97260
    },
    {
      "epoch": 11.705174488567991,
      "grad_norm": 2198.923828125,
      "learning_rate": 0.00019392746904456186,
      "loss": 7.2474,
      "step": 97270
    },
    {
      "epoch": 11.706377858002407,
      "grad_norm": 1371.2738037109375,
      "learning_rate": 0.00019392616347969002,
      "loss": 7.3,
      "step": 97280
    },
    {
      "epoch": 11.707581227436823,
      "grad_norm": 1478.193359375,
      "learning_rate": 0.00019392485777888383,
      "loss": 7.2453,
      "step": 97290
    },
    {
      "epoch": 11.70878459687124,
      "grad_norm": 2130.078369140625,
      "learning_rate": 0.00019392355194214518,
      "loss": 7.2719,
      "step": 97300
    },
    {
      "epoch": 11.709987966305656,
      "grad_norm": 572.8602294921875,
      "learning_rate": 0.00019392224596947592,
      "loss": 7.219,
      "step": 97310
    },
    {
      "epoch": 11.711191335740072,
      "grad_norm": 1772.4779052734375,
      "learning_rate": 0.00019392093986087802,
      "loss": 7.2102,
      "step": 97320
    },
    {
      "epoch": 11.71239470517449,
      "grad_norm": 2705.216064453125,
      "learning_rate": 0.00019391963361635334,
      "loss": 7.3427,
      "step": 97330
    },
    {
      "epoch": 11.713598074608905,
      "grad_norm": 824.8469848632812,
      "learning_rate": 0.00019391832723590375,
      "loss": 7.2595,
      "step": 97340
    },
    {
      "epoch": 11.71480144404332,
      "grad_norm": 357.5631103515625,
      "learning_rate": 0.00019391702071953115,
      "loss": 7.1763,
      "step": 97350
    },
    {
      "epoch": 11.716004813477738,
      "grad_norm": 1005.9691772460938,
      "learning_rate": 0.00019391571406723743,
      "loss": 7.3828,
      "step": 97360
    },
    {
      "epoch": 11.717208182912154,
      "grad_norm": 1013.7130126953125,
      "learning_rate": 0.00019391440727902447,
      "loss": 7.326,
      "step": 97370
    },
    {
      "epoch": 11.71841155234657,
      "grad_norm": 757.0460205078125,
      "learning_rate": 0.0001939131003548942,
      "loss": 7.3788,
      "step": 97380
    },
    {
      "epoch": 11.719614921780988,
      "grad_norm": 2146.37744140625,
      "learning_rate": 0.00019391179329484847,
      "loss": 7.2004,
      "step": 97390
    },
    {
      "epoch": 11.720818291215403,
      "grad_norm": 559.243408203125,
      "learning_rate": 0.00019391048609888921,
      "loss": 7.264,
      "step": 97400
    },
    {
      "epoch": 11.722021660649819,
      "grad_norm": 1754.53759765625,
      "learning_rate": 0.00019390917876701828,
      "loss": 7.5141,
      "step": 97410
    },
    {
      "epoch": 11.723225030084237,
      "grad_norm": 1784.84375,
      "learning_rate": 0.0001939078712992376,
      "loss": 7.4901,
      "step": 97420
    },
    {
      "epoch": 11.724428399518652,
      "grad_norm": 826.5912475585938,
      "learning_rate": 0.00019390656369554902,
      "loss": 7.263,
      "step": 97430
    },
    {
      "epoch": 11.725631768953068,
      "grad_norm": 1254.98583984375,
      "learning_rate": 0.00019390525595595448,
      "loss": 7.327,
      "step": 97440
    },
    {
      "epoch": 11.726835138387486,
      "grad_norm": 1729.592041015625,
      "learning_rate": 0.00019390394808045582,
      "loss": 7.3413,
      "step": 97450
    },
    {
      "epoch": 11.728038507821902,
      "grad_norm": 763.759765625,
      "learning_rate": 0.000193902640069055,
      "loss": 7.3255,
      "step": 97460
    },
    {
      "epoch": 11.729241877256317,
      "grad_norm": 629.5181884765625,
      "learning_rate": 0.00019390133192175385,
      "loss": 7.3258,
      "step": 97470
    },
    {
      "epoch": 11.730445246690735,
      "grad_norm": 1008.3265380859375,
      "learning_rate": 0.00019390002363855427,
      "loss": 7.2641,
      "step": 97480
    },
    {
      "epoch": 11.73164861612515,
      "grad_norm": 1309.145751953125,
      "learning_rate": 0.00019389871521945822,
      "loss": 7.2143,
      "step": 97490
    },
    {
      "epoch": 11.732851985559567,
      "grad_norm": 554.0267944335938,
      "learning_rate": 0.0001938974066644675,
      "loss": 7.1891,
      "step": 97500
    },
    {
      "epoch": 11.734055354993982,
      "grad_norm": 882.9625244140625,
      "learning_rate": 0.00019389609797358407,
      "loss": 7.3307,
      "step": 97510
    },
    {
      "epoch": 11.7352587244284,
      "grad_norm": 433.39190673828125,
      "learning_rate": 0.00019389478914680984,
      "loss": 7.2294,
      "step": 97520
    },
    {
      "epoch": 11.736462093862816,
      "grad_norm": 580.0017700195312,
      "learning_rate": 0.00019389348018414664,
      "loss": 7.2665,
      "step": 97530
    },
    {
      "epoch": 11.737665463297231,
      "grad_norm": 698.08837890625,
      "learning_rate": 0.00019389217108559635,
      "loss": 7.169,
      "step": 97540
    },
    {
      "epoch": 11.738868832731649,
      "grad_norm": 1450.7735595703125,
      "learning_rate": 0.00019389086185116097,
      "loss": 7.2314,
      "step": 97550
    },
    {
      "epoch": 11.740072202166065,
      "grad_norm": 343.9694519042969,
      "learning_rate": 0.0001938895524808423,
      "loss": 7.1547,
      "step": 97560
    },
    {
      "epoch": 11.74127557160048,
      "grad_norm": 2592.584228515625,
      "learning_rate": 0.0001938882429746423,
      "loss": 7.1933,
      "step": 97570
    },
    {
      "epoch": 11.742478941034898,
      "grad_norm": 623.029541015625,
      "learning_rate": 0.0001938869333325628,
      "loss": 7.2451,
      "step": 97580
    },
    {
      "epoch": 11.743682310469314,
      "grad_norm": 246.89083862304688,
      "learning_rate": 0.00019388562355460574,
      "loss": 7.2712,
      "step": 97590
    },
    {
      "epoch": 11.74488567990373,
      "grad_norm": 270.40576171875,
      "learning_rate": 0.000193884313640773,
      "loss": 7.2872,
      "step": 97600
    },
    {
      "epoch": 11.746089049338147,
      "grad_norm": 292.736572265625,
      "learning_rate": 0.00019388300359106646,
      "loss": 7.2095,
      "step": 97610
    },
    {
      "epoch": 11.747292418772563,
      "grad_norm": 374.883544921875,
      "learning_rate": 0.00019388169340548805,
      "loss": 7.2559,
      "step": 97620
    },
    {
      "epoch": 11.748495788206979,
      "grad_norm": 348.9725341796875,
      "learning_rate": 0.00019388038308403968,
      "loss": 7.174,
      "step": 97630
    },
    {
      "epoch": 11.749699157641396,
      "grad_norm": 557.6121826171875,
      "learning_rate": 0.00019387907262672318,
      "loss": 7.2309,
      "step": 97640
    },
    {
      "epoch": 11.750902527075812,
      "grad_norm": 592.2030029296875,
      "learning_rate": 0.00019387776203354048,
      "loss": 7.2064,
      "step": 97650
    },
    {
      "epoch": 11.752105896510228,
      "grad_norm": 807.6978149414062,
      "learning_rate": 0.0001938764513044935,
      "loss": 7.2876,
      "step": 97660
    },
    {
      "epoch": 11.753309265944646,
      "grad_norm": 578.1055297851562,
      "learning_rate": 0.00019387514043958413,
      "loss": 7.13,
      "step": 97670
    },
    {
      "epoch": 11.754512635379061,
      "grad_norm": 199.2294464111328,
      "learning_rate": 0.00019387382943881421,
      "loss": 7.2102,
      "step": 97680
    },
    {
      "epoch": 11.755716004813477,
      "grad_norm": 209.4580078125,
      "learning_rate": 0.00019387251830218572,
      "loss": 7.3086,
      "step": 97690
    },
    {
      "epoch": 11.756919374247895,
      "grad_norm": 212.68966674804688,
      "learning_rate": 0.00019387120702970049,
      "loss": 7.2825,
      "step": 97700
    },
    {
      "epoch": 11.75812274368231,
      "grad_norm": 342.11566162109375,
      "learning_rate": 0.0001938698956213605,
      "loss": 7.3014,
      "step": 97710
    },
    {
      "epoch": 11.759326113116726,
      "grad_norm": 373.3837890625,
      "learning_rate": 0.00019386858407716755,
      "loss": 7.2352,
      "step": 97720
    },
    {
      "epoch": 11.760529482551144,
      "grad_norm": 219.68666076660156,
      "learning_rate": 0.0001938672723971236,
      "loss": 7.2783,
      "step": 97730
    },
    {
      "epoch": 11.76173285198556,
      "grad_norm": 330.65557861328125,
      "learning_rate": 0.0001938659605812305,
      "loss": 7.2933,
      "step": 97740
    },
    {
      "epoch": 11.762936221419976,
      "grad_norm": 2246.142333984375,
      "learning_rate": 0.0001938646486294902,
      "loss": 7.3261,
      "step": 97750
    },
    {
      "epoch": 11.764139590854393,
      "grad_norm": 323.64886474609375,
      "learning_rate": 0.0001938633365419046,
      "loss": 7.2726,
      "step": 97760
    },
    {
      "epoch": 11.765342960288809,
      "grad_norm": 308.22125244140625,
      "learning_rate": 0.00019386202431847559,
      "loss": 7.1803,
      "step": 97770
    },
    {
      "epoch": 11.766546329723225,
      "grad_norm": 252.5232696533203,
      "learning_rate": 0.00019386071195920501,
      "loss": 7.2194,
      "step": 97780
    },
    {
      "epoch": 11.76774969915764,
      "grad_norm": 1257.066162109375,
      "learning_rate": 0.00019385939946409483,
      "loss": 7.2427,
      "step": 97790
    },
    {
      "epoch": 11.768953068592058,
      "grad_norm": 349.7895202636719,
      "learning_rate": 0.00019385808683314691,
      "loss": 7.2151,
      "step": 97800
    },
    {
      "epoch": 11.770156438026474,
      "grad_norm": 283.97357177734375,
      "learning_rate": 0.00019385677406636318,
      "loss": 7.2866,
      "step": 97810
    },
    {
      "epoch": 11.77135980746089,
      "grad_norm": 242.33619689941406,
      "learning_rate": 0.00019385546116374553,
      "loss": 7.2093,
      "step": 97820
    },
    {
      "epoch": 11.772563176895307,
      "grad_norm": 719.7299194335938,
      "learning_rate": 0.00019385414812529586,
      "loss": 7.2735,
      "step": 97830
    },
    {
      "epoch": 11.773766546329723,
      "grad_norm": 822.5176391601562,
      "learning_rate": 0.00019385283495101604,
      "loss": 7.246,
      "step": 97840
    },
    {
      "epoch": 11.774969915764139,
      "grad_norm": 784.7896728515625,
      "learning_rate": 0.00019385152164090802,
      "loss": 7.2197,
      "step": 97850
    },
    {
      "epoch": 11.776173285198556,
      "grad_norm": 191.56918334960938,
      "learning_rate": 0.00019385020819497365,
      "loss": 7.2684,
      "step": 97860
    },
    {
      "epoch": 11.777376654632972,
      "grad_norm": 508.599365234375,
      "learning_rate": 0.0001938488946132149,
      "loss": 7.2816,
      "step": 97870
    },
    {
      "epoch": 11.778580024067388,
      "grad_norm": 226.9486083984375,
      "learning_rate": 0.0001938475808956336,
      "loss": 7.2735,
      "step": 97880
    },
    {
      "epoch": 11.779783393501805,
      "grad_norm": 347.66680908203125,
      "learning_rate": 0.00019384626704223168,
      "loss": 7.1585,
      "step": 97890
    },
    {
      "epoch": 11.780986762936221,
      "grad_norm": 882.27490234375,
      "learning_rate": 0.00019384495305301106,
      "loss": 7.2276,
      "step": 97900
    },
    {
      "epoch": 11.782190132370637,
      "grad_norm": 262.778564453125,
      "learning_rate": 0.0001938436389279736,
      "loss": 7.2892,
      "step": 97910
    },
    {
      "epoch": 11.783393501805055,
      "grad_norm": 308.4969787597656,
      "learning_rate": 0.00019384232466712124,
      "loss": 7.2309,
      "step": 97920
    },
    {
      "epoch": 11.78459687123947,
      "grad_norm": 954.2713623046875,
      "learning_rate": 0.00019384101027045583,
      "loss": 7.273,
      "step": 97930
    },
    {
      "epoch": 11.785800240673886,
      "grad_norm": 184.59352111816406,
      "learning_rate": 0.00019383969573797937,
      "loss": 7.2294,
      "step": 97940
    },
    {
      "epoch": 11.787003610108304,
      "grad_norm": 824.8224487304688,
      "learning_rate": 0.00019383838106969367,
      "loss": 7.2572,
      "step": 97950
    },
    {
      "epoch": 11.78820697954272,
      "grad_norm": 721.7014770507812,
      "learning_rate": 0.00019383706626560068,
      "loss": 7.3616,
      "step": 97960
    },
    {
      "epoch": 11.789410348977135,
      "grad_norm": 201.83432006835938,
      "learning_rate": 0.00019383575132570226,
      "loss": 7.354,
      "step": 97970
    },
    {
      "epoch": 11.790613718411553,
      "grad_norm": 629.5089111328125,
      "learning_rate": 0.00019383443625000039,
      "loss": 7.2131,
      "step": 97980
    },
    {
      "epoch": 11.791817087845969,
      "grad_norm": 758.6402587890625,
      "learning_rate": 0.00019383312103849688,
      "loss": 7.2023,
      "step": 97990
    },
    {
      "epoch": 11.793020457280385,
      "grad_norm": 193.6214599609375,
      "learning_rate": 0.0001938318056911937,
      "loss": 7.2367,
      "step": 98000
    },
    {
      "epoch": 11.794223826714802,
      "grad_norm": 479.528076171875,
      "learning_rate": 0.00019383049020809268,
      "loss": 7.3421,
      "step": 98010
    },
    {
      "epoch": 11.795427196149218,
      "grad_norm": 507.7732238769531,
      "learning_rate": 0.00019382917458919585,
      "loss": 7.1746,
      "step": 98020
    },
    {
      "epoch": 11.796630565583634,
      "grad_norm": 186.1522674560547,
      "learning_rate": 0.00019382785883450502,
      "loss": 7.278,
      "step": 98030
    },
    {
      "epoch": 11.797833935018051,
      "grad_norm": 185.91578674316406,
      "learning_rate": 0.00019382654294402207,
      "loss": 7.2305,
      "step": 98040
    },
    {
      "epoch": 11.799037304452467,
      "grad_norm": 372.1086730957031,
      "learning_rate": 0.00019382522691774895,
      "loss": 7.1682,
      "step": 98050
    },
    {
      "epoch": 11.800240673886883,
      "grad_norm": 651.1045532226562,
      "learning_rate": 0.00019382391075568759,
      "loss": 7.2985,
      "step": 98060
    },
    {
      "epoch": 11.8014440433213,
      "grad_norm": 262.8721008300781,
      "learning_rate": 0.00019382259445783988,
      "loss": 7.202,
      "step": 98070
    },
    {
      "epoch": 11.802647412755716,
      "grad_norm": 188.15924072265625,
      "learning_rate": 0.0001938212780242077,
      "loss": 7.1035,
      "step": 98080
    },
    {
      "epoch": 11.803850782190132,
      "grad_norm": 330.3931579589844,
      "learning_rate": 0.00019381996145479295,
      "loss": 7.3127,
      "step": 98090
    },
    {
      "epoch": 11.80505415162455,
      "grad_norm": 227.94102478027344,
      "learning_rate": 0.00019381864474959758,
      "loss": 7.3039,
      "step": 98100
    },
    {
      "epoch": 11.806257521058965,
      "grad_norm": 140.84469604492188,
      "learning_rate": 0.00019381732790862345,
      "loss": 7.3337,
      "step": 98110
    },
    {
      "epoch": 11.807460890493381,
      "grad_norm": 340.9453430175781,
      "learning_rate": 0.00019381601093187246,
      "loss": 7.1657,
      "step": 98120
    },
    {
      "epoch": 11.808664259927799,
      "grad_norm": 183.7386932373047,
      "learning_rate": 0.0001938146938193466,
      "loss": 7.3006,
      "step": 98130
    },
    {
      "epoch": 11.809867629362214,
      "grad_norm": 779.0843505859375,
      "learning_rate": 0.00019381337657104766,
      "loss": 7.2035,
      "step": 98140
    },
    {
      "epoch": 11.81107099879663,
      "grad_norm": 237.23541259765625,
      "learning_rate": 0.0001938120591869776,
      "loss": 7.2857,
      "step": 98150
    },
    {
      "epoch": 11.812274368231048,
      "grad_norm": 181.39215087890625,
      "learning_rate": 0.00019381074166713837,
      "loss": 7.3467,
      "step": 98160
    },
    {
      "epoch": 11.813477737665464,
      "grad_norm": 236.60719299316406,
      "learning_rate": 0.0001938094240115318,
      "loss": 7.3021,
      "step": 98170
    },
    {
      "epoch": 11.81468110709988,
      "grad_norm": 175.98077392578125,
      "learning_rate": 0.00019380810622015985,
      "loss": 7.3032,
      "step": 98180
    },
    {
      "epoch": 11.815884476534297,
      "grad_norm": 531.7383422851562,
      "learning_rate": 0.00019380678829302442,
      "loss": 7.1955,
      "step": 98190
    },
    {
      "epoch": 11.817087845968713,
      "grad_norm": 257.8716735839844,
      "learning_rate": 0.0001938054702301274,
      "loss": 7.2312,
      "step": 98200
    },
    {
      "epoch": 11.818291215403129,
      "grad_norm": 214.31576538085938,
      "learning_rate": 0.00019380415203147072,
      "loss": 7.2404,
      "step": 98210
    },
    {
      "epoch": 11.819494584837544,
      "grad_norm": 1384.1571044921875,
      "learning_rate": 0.00019380283369705624,
      "loss": 7.2254,
      "step": 98220
    },
    {
      "epoch": 11.820697954271962,
      "grad_norm": 354.4250793457031,
      "learning_rate": 0.0001938015152268859,
      "loss": 7.3571,
      "step": 98230
    },
    {
      "epoch": 11.821901323706378,
      "grad_norm": 496.17333984375,
      "learning_rate": 0.00019380019662096164,
      "loss": 7.2166,
      "step": 98240
    },
    {
      "epoch": 11.823104693140793,
      "grad_norm": 468.5306091308594,
      "learning_rate": 0.00019379887787928531,
      "loss": 7.166,
      "step": 98250
    },
    {
      "epoch": 11.824308062575211,
      "grad_norm": 839.9148559570312,
      "learning_rate": 0.00019379755900185885,
      "loss": 7.2568,
      "step": 98260
    },
    {
      "epoch": 11.825511432009627,
      "grad_norm": 393.55718994140625,
      "learning_rate": 0.00019379623998868416,
      "loss": 7.2669,
      "step": 98270
    },
    {
      "epoch": 11.826714801444043,
      "grad_norm": 608.7523193359375,
      "learning_rate": 0.00019379492083976317,
      "loss": 7.2602,
      "step": 98280
    },
    {
      "epoch": 11.82791817087846,
      "grad_norm": 1107.749755859375,
      "learning_rate": 0.00019379360155509777,
      "loss": 7.1704,
      "step": 98290
    },
    {
      "epoch": 11.829121540312876,
      "grad_norm": 515.56591796875,
      "learning_rate": 0.00019379228213468988,
      "loss": 7.2147,
      "step": 98300
    },
    {
      "epoch": 11.830324909747292,
      "grad_norm": 355.8526306152344,
      "learning_rate": 0.0001937909625785414,
      "loss": 7.2304,
      "step": 98310
    },
    {
      "epoch": 11.83152827918171,
      "grad_norm": 751.2402954101562,
      "learning_rate": 0.00019378964288665421,
      "loss": 7.1255,
      "step": 98320
    },
    {
      "epoch": 11.832731648616125,
      "grad_norm": 215.8268585205078,
      "learning_rate": 0.0001937883230590303,
      "loss": 7.2136,
      "step": 98330
    },
    {
      "epoch": 11.833935018050541,
      "grad_norm": 377.1159973144531,
      "learning_rate": 0.00019378700309567149,
      "loss": 7.2863,
      "step": 98340
    },
    {
      "epoch": 11.835138387484958,
      "grad_norm": 536.8948364257812,
      "learning_rate": 0.00019378568299657973,
      "loss": 7.3514,
      "step": 98350
    },
    {
      "epoch": 11.836341756919374,
      "grad_norm": 192.38772583007812,
      "learning_rate": 0.00019378436276175695,
      "loss": 7.2993,
      "step": 98360
    },
    {
      "epoch": 11.83754512635379,
      "grad_norm": 391.0587463378906,
      "learning_rate": 0.00019378304239120505,
      "loss": 7.3251,
      "step": 98370
    },
    {
      "epoch": 11.838748495788208,
      "grad_norm": 128.91867065429688,
      "learning_rate": 0.00019378172188492588,
      "loss": 7.3763,
      "step": 98380
    },
    {
      "epoch": 11.839951865222623,
      "grad_norm": 147.3256378173828,
      "learning_rate": 0.00019378040124292146,
      "loss": 7.3197,
      "step": 98390
    },
    {
      "epoch": 11.84115523465704,
      "grad_norm": 141.4853973388672,
      "learning_rate": 0.00019377908046519363,
      "loss": 7.2579,
      "step": 98400
    },
    {
      "epoch": 11.842358604091457,
      "grad_norm": 309.0194091796875,
      "learning_rate": 0.00019377775955174432,
      "loss": 7.2703,
      "step": 98410
    },
    {
      "epoch": 11.843561973525873,
      "grad_norm": 235.58509826660156,
      "learning_rate": 0.00019377643850257545,
      "loss": 7.2928,
      "step": 98420
    },
    {
      "epoch": 11.844765342960288,
      "grad_norm": 116.53980255126953,
      "learning_rate": 0.00019377511731768888,
      "loss": 7.2744,
      "step": 98430
    },
    {
      "epoch": 11.845968712394706,
      "grad_norm": 85.77718353271484,
      "learning_rate": 0.0001937737959970866,
      "loss": 7.364,
      "step": 98440
    },
    {
      "epoch": 11.847172081829122,
      "grad_norm": 350.1137390136719,
      "learning_rate": 0.0001937724745407705,
      "loss": 7.3654,
      "step": 98450
    },
    {
      "epoch": 11.848375451263538,
      "grad_norm": 151.2226104736328,
      "learning_rate": 0.00019377115294874243,
      "loss": 7.3418,
      "step": 98460
    },
    {
      "epoch": 11.849578820697955,
      "grad_norm": 262.7079162597656,
      "learning_rate": 0.00019376983122100437,
      "loss": 7.2046,
      "step": 98470
    },
    {
      "epoch": 11.85078219013237,
      "grad_norm": 143.4476318359375,
      "learning_rate": 0.00019376850935755823,
      "loss": 7.3117,
      "step": 98480
    },
    {
      "epoch": 11.851985559566787,
      "grad_norm": 169.92245483398438,
      "learning_rate": 0.00019376718735840588,
      "loss": 7.3094,
      "step": 98490
    },
    {
      "epoch": 11.853188929001202,
      "grad_norm": 117.85218811035156,
      "learning_rate": 0.00019376586522354926,
      "loss": 7.3461,
      "step": 98500
    },
    {
      "epoch": 11.85439229843562,
      "grad_norm": 352.0733337402344,
      "learning_rate": 0.00019376454295299032,
      "loss": 7.1903,
      "step": 98510
    },
    {
      "epoch": 11.855595667870036,
      "grad_norm": 227.34487915039062,
      "learning_rate": 0.0001937632205467309,
      "loss": 7.2136,
      "step": 98520
    },
    {
      "epoch": 11.856799037304452,
      "grad_norm": 134.1216278076172,
      "learning_rate": 0.00019376189800477298,
      "loss": 7.4062,
      "step": 98530
    },
    {
      "epoch": 11.85800240673887,
      "grad_norm": 395.313232421875,
      "learning_rate": 0.00019376057532711843,
      "loss": 7.3388,
      "step": 98540
    },
    {
      "epoch": 11.859205776173285,
      "grad_norm": 257.0782775878906,
      "learning_rate": 0.00019375925251376917,
      "loss": 7.3002,
      "step": 98550
    },
    {
      "epoch": 11.8604091456077,
      "grad_norm": 271.50762939453125,
      "learning_rate": 0.0001937579295647271,
      "loss": 7.2972,
      "step": 98560
    },
    {
      "epoch": 11.861612515042118,
      "grad_norm": 158.8642120361328,
      "learning_rate": 0.00019375660647999422,
      "loss": 7.1253,
      "step": 98570
    },
    {
      "epoch": 11.862815884476534,
      "grad_norm": 130.6636505126953,
      "learning_rate": 0.00019375528325957234,
      "loss": 7.252,
      "step": 98580
    },
    {
      "epoch": 11.86401925391095,
      "grad_norm": 195.52801513671875,
      "learning_rate": 0.00019375395990346342,
      "loss": 7.3208,
      "step": 98590
    },
    {
      "epoch": 11.865222623345367,
      "grad_norm": 101.48219299316406,
      "learning_rate": 0.00019375263641166937,
      "loss": 7.2842,
      "step": 98600
    },
    {
      "epoch": 11.866425992779783,
      "grad_norm": 241.259765625,
      "learning_rate": 0.0001937513127841921,
      "loss": 7.3199,
      "step": 98610
    },
    {
      "epoch": 11.867629362214199,
      "grad_norm": 81.8074722290039,
      "learning_rate": 0.00019374998902103359,
      "loss": 7.326,
      "step": 98620
    },
    {
      "epoch": 11.868832731648617,
      "grad_norm": 143.58274841308594,
      "learning_rate": 0.00019374866512219564,
      "loss": 7.2454,
      "step": 98630
    },
    {
      "epoch": 11.870036101083032,
      "grad_norm": 199.72824096679688,
      "learning_rate": 0.00019374734108768026,
      "loss": 7.429,
      "step": 98640
    },
    {
      "epoch": 11.871239470517448,
      "grad_norm": 99.01521301269531,
      "learning_rate": 0.00019374601691748934,
      "loss": 7.4567,
      "step": 98650
    },
    {
      "epoch": 11.872442839951866,
      "grad_norm": 119.11502838134766,
      "learning_rate": 0.00019374469261162474,
      "loss": 7.5071,
      "step": 98660
    },
    {
      "epoch": 11.873646209386282,
      "grad_norm": 141.5318145751953,
      "learning_rate": 0.00019374336817008846,
      "loss": 7.2169,
      "step": 98670
    },
    {
      "epoch": 11.874849578820697,
      "grad_norm": 214.09669494628906,
      "learning_rate": 0.00019374204359288237,
      "loss": 7.3049,
      "step": 98680
    },
    {
      "epoch": 11.876052948255115,
      "grad_norm": 143.29026794433594,
      "learning_rate": 0.0001937407188800084,
      "loss": 7.2945,
      "step": 98690
    },
    {
      "epoch": 11.87725631768953,
      "grad_norm": 86.36083221435547,
      "learning_rate": 0.0001937393940314685,
      "loss": 7.2817,
      "step": 98700
    },
    {
      "epoch": 11.878459687123947,
      "grad_norm": 148.66464233398438,
      "learning_rate": 0.0001937380690472645,
      "loss": 7.3459,
      "step": 98710
    },
    {
      "epoch": 11.879663056558364,
      "grad_norm": 178.42825317382812,
      "learning_rate": 0.0001937367439273984,
      "loss": 7.3013,
      "step": 98720
    },
    {
      "epoch": 11.88086642599278,
      "grad_norm": 55.61753845214844,
      "learning_rate": 0.00019373541867187207,
      "loss": 7.3156,
      "step": 98730
    },
    {
      "epoch": 11.882069795427196,
      "grad_norm": 48.33180618286133,
      "learning_rate": 0.00019373409328068747,
      "loss": 7.28,
      "step": 98740
    },
    {
      "epoch": 11.883273164861613,
      "grad_norm": 83.62631225585938,
      "learning_rate": 0.00019373276775384648,
      "loss": 7.2481,
      "step": 98750
    },
    {
      "epoch": 11.884476534296029,
      "grad_norm": 98.91661834716797,
      "learning_rate": 0.00019373144209135107,
      "loss": 7.4269,
      "step": 98760
    },
    {
      "epoch": 11.885679903730445,
      "grad_norm": 201.1520538330078,
      "learning_rate": 0.00019373011629320304,
      "loss": 7.2154,
      "step": 98770
    },
    {
      "epoch": 11.886883273164862,
      "grad_norm": 70.19197082519531,
      "learning_rate": 0.00019372879035940444,
      "loss": 7.336,
      "step": 98780
    },
    {
      "epoch": 11.888086642599278,
      "grad_norm": 109.77552795410156,
      "learning_rate": 0.00019372746428995716,
      "loss": 7.348,
      "step": 98790
    },
    {
      "epoch": 11.889290012033694,
      "grad_norm": 116.01335906982422,
      "learning_rate": 0.0001937261380848631,
      "loss": 7.2029,
      "step": 98800
    },
    {
      "epoch": 11.890493381468112,
      "grad_norm": 1028.3544921875,
      "learning_rate": 0.00019372481174412414,
      "loss": 7.2875,
      "step": 98810
    },
    {
      "epoch": 11.891696750902527,
      "grad_norm": 103.023681640625,
      "learning_rate": 0.00019372348526774225,
      "loss": 7.3296,
      "step": 98820
    },
    {
      "epoch": 11.892900120336943,
      "grad_norm": 121.654052734375,
      "learning_rate": 0.00019372215865571933,
      "loss": 7.3729,
      "step": 98830
    },
    {
      "epoch": 11.89410348977136,
      "grad_norm": 69.78581237792969,
      "learning_rate": 0.00019372083190805732,
      "loss": 7.2538,
      "step": 98840
    },
    {
      "epoch": 11.895306859205776,
      "grad_norm": 74.58787536621094,
      "learning_rate": 0.00019371950502475812,
      "loss": 7.4778,
      "step": 98850
    },
    {
      "epoch": 11.896510228640192,
      "grad_norm": 164.1248016357422,
      "learning_rate": 0.00019371817800582365,
      "loss": 7.2292,
      "step": 98860
    },
    {
      "epoch": 11.89771359807461,
      "grad_norm": 118.36473846435547,
      "learning_rate": 0.00019371685085125585,
      "loss": 7.2678,
      "step": 98870
    },
    {
      "epoch": 11.898916967509026,
      "grad_norm": 113.45089721679688,
      "learning_rate": 0.00019371552356105664,
      "loss": 7.2654,
      "step": 98880
    },
    {
      "epoch": 11.900120336943441,
      "grad_norm": 152.16529846191406,
      "learning_rate": 0.00019371419613522793,
      "loss": 7.3207,
      "step": 98890
    },
    {
      "epoch": 11.901323706377857,
      "grad_norm": 128.1909942626953,
      "learning_rate": 0.00019371286857377165,
      "loss": 7.1912,
      "step": 98900
    },
    {
      "epoch": 11.902527075812275,
      "grad_norm": 126.4127197265625,
      "learning_rate": 0.00019371154087668968,
      "loss": 7.2401,
      "step": 98910
    },
    {
      "epoch": 11.90373044524669,
      "grad_norm": 92.52333068847656,
      "learning_rate": 0.00019371021304398397,
      "loss": 7.2159,
      "step": 98920
    },
    {
      "epoch": 11.904933814681106,
      "grad_norm": 97.5126953125,
      "learning_rate": 0.00019370888507565646,
      "loss": 7.3579,
      "step": 98930
    },
    {
      "epoch": 11.906137184115524,
      "grad_norm": 205.54931640625,
      "learning_rate": 0.00019370755697170906,
      "loss": 7.2454,
      "step": 98940
    },
    {
      "epoch": 11.90734055354994,
      "grad_norm": 161.52279663085938,
      "learning_rate": 0.0001937062287321437,
      "loss": 7.2965,
      "step": 98950
    },
    {
      "epoch": 11.908543922984355,
      "grad_norm": 141.312744140625,
      "learning_rate": 0.00019370490035696227,
      "loss": 7.2946,
      "step": 98960
    },
    {
      "epoch": 11.909747292418773,
      "grad_norm": 136.45433044433594,
      "learning_rate": 0.00019370357184616673,
      "loss": 7.2291,
      "step": 98970
    },
    {
      "epoch": 11.910950661853189,
      "grad_norm": 68.9397201538086,
      "learning_rate": 0.000193702243199759,
      "loss": 7.2818,
      "step": 98980
    },
    {
      "epoch": 11.912154031287605,
      "grad_norm": 148.0553436279297,
      "learning_rate": 0.00019370091441774095,
      "loss": 7.2469,
      "step": 98990
    },
    {
      "epoch": 11.913357400722022,
      "grad_norm": 115.45630645751953,
      "learning_rate": 0.00019369958550011457,
      "loss": 7.2465,
      "step": 99000
    },
    {
      "epoch": 11.914560770156438,
      "grad_norm": 190.38906860351562,
      "learning_rate": 0.00019369825644688176,
      "loss": 7.2955,
      "step": 99010
    },
    {
      "epoch": 11.915764139590854,
      "grad_norm": 52.35891342163086,
      "learning_rate": 0.0001936969272580444,
      "loss": 7.3681,
      "step": 99020
    },
    {
      "epoch": 11.916967509025271,
      "grad_norm": 60.705562591552734,
      "learning_rate": 0.0001936955979336045,
      "loss": 7.4014,
      "step": 99030
    },
    {
      "epoch": 11.918170878459687,
      "grad_norm": 150.88270568847656,
      "learning_rate": 0.00019369426847356392,
      "loss": 7.2546,
      "step": 99040
    },
    {
      "epoch": 11.919374247894103,
      "grad_norm": 72.51187896728516,
      "learning_rate": 0.0001936929388779246,
      "loss": 7.2991,
      "step": 99050
    },
    {
      "epoch": 11.92057761732852,
      "grad_norm": 157.69631958007812,
      "learning_rate": 0.00019369160914668845,
      "loss": 7.2478,
      "step": 99060
    },
    {
      "epoch": 11.921780986762936,
      "grad_norm": 106.79886627197266,
      "learning_rate": 0.00019369027927985743,
      "loss": 7.2584,
      "step": 99070
    },
    {
      "epoch": 11.922984356197352,
      "grad_norm": 94.18775939941406,
      "learning_rate": 0.00019368894927743345,
      "loss": 7.2719,
      "step": 99080
    },
    {
      "epoch": 11.92418772563177,
      "grad_norm": 180.44630432128906,
      "learning_rate": 0.00019368761913941842,
      "loss": 7.3039,
      "step": 99090
    },
    {
      "epoch": 11.925391095066185,
      "grad_norm": 116.6115951538086,
      "learning_rate": 0.00019368628886581429,
      "loss": 7.2578,
      "step": 99100
    },
    {
      "epoch": 11.926594464500601,
      "grad_norm": 189.6122283935547,
      "learning_rate": 0.00019368495845662294,
      "loss": 7.2478,
      "step": 99110
    },
    {
      "epoch": 11.927797833935019,
      "grad_norm": 122.07473754882812,
      "learning_rate": 0.00019368362791184632,
      "loss": 7.2349,
      "step": 99120
    },
    {
      "epoch": 11.929001203369435,
      "grad_norm": 53.1552734375,
      "learning_rate": 0.00019368229723148642,
      "loss": 7.2924,
      "step": 99130
    },
    {
      "epoch": 11.93020457280385,
      "grad_norm": 325.55657958984375,
      "learning_rate": 0.00019368096641554506,
      "loss": 7.2807,
      "step": 99140
    },
    {
      "epoch": 11.931407942238268,
      "grad_norm": 172.39874267578125,
      "learning_rate": 0.0001936796354640242,
      "loss": 7.2303,
      "step": 99150
    },
    {
      "epoch": 11.932611311672684,
      "grad_norm": 174.95828247070312,
      "learning_rate": 0.0001936783043769258,
      "loss": 7.3568,
      "step": 99160
    },
    {
      "epoch": 11.9338146811071,
      "grad_norm": 138.89820861816406,
      "learning_rate": 0.00019367697315425178,
      "loss": 7.2486,
      "step": 99170
    },
    {
      "epoch": 11.935018050541515,
      "grad_norm": 112.64092254638672,
      "learning_rate": 0.00019367564179600401,
      "loss": 7.3451,
      "step": 99180
    },
    {
      "epoch": 11.936221419975933,
      "grad_norm": 163.35719299316406,
      "learning_rate": 0.0001936743103021845,
      "loss": 7.4022,
      "step": 99190
    },
    {
      "epoch": 11.937424789410349,
      "grad_norm": 79.35734558105469,
      "learning_rate": 0.0001936729786727951,
      "loss": 7.228,
      "step": 99200
    },
    {
      "epoch": 11.938628158844764,
      "grad_norm": 171.12986755371094,
      "learning_rate": 0.0001936716469078378,
      "loss": 7.2654,
      "step": 99210
    },
    {
      "epoch": 11.939831528279182,
      "grad_norm": 117.14237213134766,
      "learning_rate": 0.00019367031500731448,
      "loss": 7.204,
      "step": 99220
    },
    {
      "epoch": 11.941034897713598,
      "grad_norm": 126.36323547363281,
      "learning_rate": 0.0001936689829712271,
      "loss": 7.3549,
      "step": 99230
    },
    {
      "epoch": 11.942238267148014,
      "grad_norm": 232.3984832763672,
      "learning_rate": 0.00019366765079957757,
      "loss": 7.327,
      "step": 99240
    },
    {
      "epoch": 11.943441636582431,
      "grad_norm": 150.30252075195312,
      "learning_rate": 0.00019366631849236784,
      "loss": 7.3605,
      "step": 99250
    },
    {
      "epoch": 11.944645006016847,
      "grad_norm": 161.02719116210938,
      "learning_rate": 0.00019366498604959982,
      "loss": 7.283,
      "step": 99260
    },
    {
      "epoch": 11.945848375451263,
      "grad_norm": 154.76730346679688,
      "learning_rate": 0.00019366365347127542,
      "loss": 7.2007,
      "step": 99270
    },
    {
      "epoch": 11.94705174488568,
      "grad_norm": 68.1380615234375,
      "learning_rate": 0.0001936623207573966,
      "loss": 7.1996,
      "step": 99280
    },
    {
      "epoch": 11.948255114320096,
      "grad_norm": 158.46258544921875,
      "learning_rate": 0.00019366098790796524,
      "loss": 7.2811,
      "step": 99290
    },
    {
      "epoch": 11.949458483754512,
      "grad_norm": 198.72340393066406,
      "learning_rate": 0.00019365965492298338,
      "loss": 7.2066,
      "step": 99300
    },
    {
      "epoch": 11.95066185318893,
      "grad_norm": 136.68829345703125,
      "learning_rate": 0.00019365832180245283,
      "loss": 7.2305,
      "step": 99310
    },
    {
      "epoch": 11.951865222623345,
      "grad_norm": 174.20272827148438,
      "learning_rate": 0.00019365698854637557,
      "loss": 7.1636,
      "step": 99320
    },
    {
      "epoch": 11.953068592057761,
      "grad_norm": 220.20640563964844,
      "learning_rate": 0.00019365565515475352,
      "loss": 7.2696,
      "step": 99330
    },
    {
      "epoch": 11.954271961492179,
      "grad_norm": 179.85169982910156,
      "learning_rate": 0.00019365432162758863,
      "loss": 7.2715,
      "step": 99340
    },
    {
      "epoch": 11.955475330926594,
      "grad_norm": 143.8240203857422,
      "learning_rate": 0.0001936529879648828,
      "loss": 7.2089,
      "step": 99350
    },
    {
      "epoch": 11.95667870036101,
      "grad_norm": 132.84027099609375,
      "learning_rate": 0.000193651654166638,
      "loss": 7.3723,
      "step": 99360
    },
    {
      "epoch": 11.957882069795428,
      "grad_norm": 75.92465209960938,
      "learning_rate": 0.0001936503202328561,
      "loss": 7.2857,
      "step": 99370
    },
    {
      "epoch": 11.959085439229844,
      "grad_norm": 108.4561538696289,
      "learning_rate": 0.00019364898616353905,
      "loss": 7.3754,
      "step": 99380
    },
    {
      "epoch": 11.96028880866426,
      "grad_norm": 236.8972625732422,
      "learning_rate": 0.00019364765195868886,
      "loss": 7.3046,
      "step": 99390
    },
    {
      "epoch": 11.961492178098677,
      "grad_norm": 74.1569595336914,
      "learning_rate": 0.00019364631761830734,
      "loss": 7.2137,
      "step": 99400
    },
    {
      "epoch": 11.962695547533093,
      "grad_norm": 178.64630126953125,
      "learning_rate": 0.0001936449831423965,
      "loss": 7.2537,
      "step": 99410
    },
    {
      "epoch": 11.963898916967509,
      "grad_norm": 95.76781463623047,
      "learning_rate": 0.00019364364853095825,
      "loss": 7.222,
      "step": 99420
    },
    {
      "epoch": 11.965102286401926,
      "grad_norm": 76.22486114501953,
      "learning_rate": 0.00019364231378399453,
      "loss": 7.2069,
      "step": 99430
    },
    {
      "epoch": 11.966305655836342,
      "grad_norm": 254.67642211914062,
      "learning_rate": 0.00019364097890150724,
      "loss": 7.2831,
      "step": 99440
    },
    {
      "epoch": 11.967509025270758,
      "grad_norm": 164.2748565673828,
      "learning_rate": 0.00019363964388349837,
      "loss": 7.275,
      "step": 99450
    },
    {
      "epoch": 11.968712394705175,
      "grad_norm": 57.14277648925781,
      "learning_rate": 0.00019363830872996978,
      "loss": 7.1939,
      "step": 99460
    },
    {
      "epoch": 11.969915764139591,
      "grad_norm": 68.4383316040039,
      "learning_rate": 0.00019363697344092345,
      "loss": 7.1941,
      "step": 99470
    },
    {
      "epoch": 11.971119133574007,
      "grad_norm": 69.10216522216797,
      "learning_rate": 0.0001936356380163613,
      "loss": 7.1493,
      "step": 99480
    },
    {
      "epoch": 11.972322503008424,
      "grad_norm": 105.06266021728516,
      "learning_rate": 0.00019363430245628526,
      "loss": 7.277,
      "step": 99490
    },
    {
      "epoch": 11.97352587244284,
      "grad_norm": 176.16969299316406,
      "learning_rate": 0.0001936329667606973,
      "loss": 7.3413,
      "step": 99500
    },
    {
      "epoch": 11.974729241877256,
      "grad_norm": 179.69448852539062,
      "learning_rate": 0.00019363163092959927,
      "loss": 7.2555,
      "step": 99510
    },
    {
      "epoch": 11.975932611311674,
      "grad_norm": 101.14362335205078,
      "learning_rate": 0.00019363029496299319,
      "loss": 7.179,
      "step": 99520
    },
    {
      "epoch": 11.97713598074609,
      "grad_norm": 46.72248458862305,
      "learning_rate": 0.00019362895886088094,
      "loss": 7.3154,
      "step": 99530
    },
    {
      "epoch": 11.978339350180505,
      "grad_norm": 100.50609588623047,
      "learning_rate": 0.00019362762262326445,
      "loss": 7.2748,
      "step": 99540
    },
    {
      "epoch": 11.979542719614923,
      "grad_norm": 209.98532104492188,
      "learning_rate": 0.0001936262862501457,
      "loss": 7.3112,
      "step": 99550
    },
    {
      "epoch": 11.980746089049338,
      "grad_norm": 172.79397583007812,
      "learning_rate": 0.0001936249497415266,
      "loss": 7.3209,
      "step": 99560
    },
    {
      "epoch": 11.981949458483754,
      "grad_norm": 171.19419860839844,
      "learning_rate": 0.00019362361309740908,
      "loss": 7.277,
      "step": 99570
    },
    {
      "epoch": 11.98315282791817,
      "grad_norm": 179.21743774414062,
      "learning_rate": 0.00019362227631779507,
      "loss": 7.3091,
      "step": 99580
    },
    {
      "epoch": 11.984356197352588,
      "grad_norm": 114.75621032714844,
      "learning_rate": 0.00019362093940268653,
      "loss": 7.1811,
      "step": 99590
    },
    {
      "epoch": 11.985559566787003,
      "grad_norm": 131.30796813964844,
      "learning_rate": 0.00019361960235208536,
      "loss": 7.196,
      "step": 99600
    },
    {
      "epoch": 11.98676293622142,
      "grad_norm": 104.64381408691406,
      "learning_rate": 0.0001936182651659935,
      "loss": 7.2974,
      "step": 99610
    },
    {
      "epoch": 11.987966305655837,
      "grad_norm": 113.8354263305664,
      "learning_rate": 0.00019361692784441293,
      "loss": 7.2838,
      "step": 99620
    },
    {
      "epoch": 11.989169675090253,
      "grad_norm": 88.38764953613281,
      "learning_rate": 0.00019361559038734552,
      "loss": 7.2407,
      "step": 99630
    },
    {
      "epoch": 11.990373044524668,
      "grad_norm": 144.893310546875,
      "learning_rate": 0.00019361425279479326,
      "loss": 7.1158,
      "step": 99640
    },
    {
      "epoch": 11.991576413959086,
      "grad_norm": 143.81150817871094,
      "learning_rate": 0.00019361291506675806,
      "loss": 7.2788,
      "step": 99650
    },
    {
      "epoch": 11.992779783393502,
      "grad_norm": 200.5011444091797,
      "learning_rate": 0.00019361157720324185,
      "loss": 7.2902,
      "step": 99660
    },
    {
      "epoch": 11.993983152827917,
      "grad_norm": 173.5289306640625,
      "learning_rate": 0.00019361023920424656,
      "loss": 7.1614,
      "step": 99670
    },
    {
      "epoch": 11.995186522262335,
      "grad_norm": 98.81207275390625,
      "learning_rate": 0.0001936089010697742,
      "loss": 7.1996,
      "step": 99680
    },
    {
      "epoch": 11.99638989169675,
      "grad_norm": 152.58604431152344,
      "learning_rate": 0.0001936075627998266,
      "loss": 7.2278,
      "step": 99690
    },
    {
      "epoch": 11.997593261131167,
      "grad_norm": 150.1853485107422,
      "learning_rate": 0.00019360622439440576,
      "loss": 7.2378,
      "step": 99700
    },
    {
      "epoch": 11.998796630565584,
      "grad_norm": 271.13079833984375,
      "learning_rate": 0.0001936048858535136,
      "loss": 7.2016,
      "step": 99710
    },
    {
      "epoch": 12.0,
      "grad_norm": 173.873779296875,
      "learning_rate": 0.00019360354717715207,
      "loss": 7.3304,
      "step": 99720
    },
    {
      "epoch": 12.0,
      "eval_loss": 7.209784507751465,
      "eval_runtime": 120.2166,
      "eval_samples_per_second": 61.447,
      "eval_steps_per_second": 7.686,
      "step": 99720
    },
    {
      "epoch": 12.001203369434416,
      "grad_norm": 187.25718688964844,
      "learning_rate": 0.00019360220836532306,
      "loss": 7.1957,
      "step": 99730
    },
    {
      "epoch": 12.002406738868833,
      "grad_norm": 103.36690521240234,
      "learning_rate": 0.00019360086941802858,
      "loss": 7.1717,
      "step": 99740
    },
    {
      "epoch": 12.00361010830325,
      "grad_norm": 268.524658203125,
      "learning_rate": 0.00019359953033527052,
      "loss": 7.3818,
      "step": 99750
    },
    {
      "epoch": 12.004813477737665,
      "grad_norm": 867.9302368164062,
      "learning_rate": 0.00019359819111705085,
      "loss": 7.1713,
      "step": 99760
    },
    {
      "epoch": 12.006016847172083,
      "grad_norm": 649.4356689453125,
      "learning_rate": 0.00019359685176337146,
      "loss": 7.2287,
      "step": 99770
    },
    {
      "epoch": 12.007220216606498,
      "grad_norm": 979.7839965820312,
      "learning_rate": 0.00019359551227423434,
      "loss": 7.256,
      "step": 99780
    },
    {
      "epoch": 12.008423586040914,
      "grad_norm": 512.5393676757812,
      "learning_rate": 0.00019359417264964139,
      "loss": 7.1805,
      "step": 99790
    },
    {
      "epoch": 12.009626955475332,
      "grad_norm": 587.2205810546875,
      "learning_rate": 0.00019359283288959459,
      "loss": 7.3021,
      "step": 99800
    },
    {
      "epoch": 12.010830324909747,
      "grad_norm": 1852.1031494140625,
      "learning_rate": 0.0001935914929940958,
      "loss": 7.1615,
      "step": 99810
    },
    {
      "epoch": 12.012033694344163,
      "grad_norm": 312.02886962890625,
      "learning_rate": 0.00019359015296314705,
      "loss": 7.1362,
      "step": 99820
    },
    {
      "epoch": 12.01323706377858,
      "grad_norm": 599.1365966796875,
      "learning_rate": 0.00019358881279675022,
      "loss": 7.2562,
      "step": 99830
    },
    {
      "epoch": 12.014440433212997,
      "grad_norm": 576.3143310546875,
      "learning_rate": 0.00019358747249490732,
      "loss": 7.2728,
      "step": 99840
    },
    {
      "epoch": 12.015643802647412,
      "grad_norm": 1972.549560546875,
      "learning_rate": 0.00019358613205762022,
      "loss": 7.334,
      "step": 99850
    },
    {
      "epoch": 12.01684717208183,
      "grad_norm": 861.9588012695312,
      "learning_rate": 0.00019358479148489085,
      "loss": 7.2409,
      "step": 99860
    },
    {
      "epoch": 12.018050541516246,
      "grad_norm": 539.2259521484375,
      "learning_rate": 0.00019358345077672123,
      "loss": 7.1917,
      "step": 99870
    },
    {
      "epoch": 12.019253910950662,
      "grad_norm": 384.87164306640625,
      "learning_rate": 0.0001935821099331132,
      "loss": 7.2393,
      "step": 99880
    },
    {
      "epoch": 12.020457280385079,
      "grad_norm": 732.364013671875,
      "learning_rate": 0.00019358076895406877,
      "loss": 7.297,
      "step": 99890
    },
    {
      "epoch": 12.021660649819495,
      "grad_norm": 1079.362060546875,
      "learning_rate": 0.00019357942783958988,
      "loss": 7.2927,
      "step": 99900
    },
    {
      "epoch": 12.02286401925391,
      "grad_norm": 1249.7060546875,
      "learning_rate": 0.00019357808658967843,
      "loss": 7.2572,
      "step": 99910
    },
    {
      "epoch": 12.024067388688326,
      "grad_norm": 925.38818359375,
      "learning_rate": 0.00019357674520433644,
      "loss": 7.3016,
      "step": 99920
    },
    {
      "epoch": 12.025270758122744,
      "grad_norm": 1079.5240478515625,
      "learning_rate": 0.00019357540368356574,
      "loss": 7.2744,
      "step": 99930
    },
    {
      "epoch": 12.02647412755716,
      "grad_norm": 1068.26806640625,
      "learning_rate": 0.00019357406202736836,
      "loss": 7.1954,
      "step": 99940
    },
    {
      "epoch": 12.027677496991576,
      "grad_norm": 489.1487731933594,
      "learning_rate": 0.0001935727202357462,
      "loss": 7.2742,
      "step": 99950
    },
    {
      "epoch": 12.028880866425993,
      "grad_norm": 2681.85498046875,
      "learning_rate": 0.0001935713783087012,
      "loss": 7.336,
      "step": 99960
    },
    {
      "epoch": 12.030084235860409,
      "grad_norm": 936.4141235351562,
      "learning_rate": 0.00019357003624623532,
      "loss": 7.215,
      "step": 99970
    },
    {
      "epoch": 12.031287605294825,
      "grad_norm": 440.8946228027344,
      "learning_rate": 0.0001935686940483505,
      "loss": 7.2278,
      "step": 99980
    },
    {
      "epoch": 12.032490974729242,
      "grad_norm": 240.0230255126953,
      "learning_rate": 0.00019356735171504864,
      "loss": 7.1908,
      "step": 99990
    },
    {
      "epoch": 12.033694344163658,
      "grad_norm": 1004.1493530273438,
      "learning_rate": 0.00019356600924633178,
      "loss": 7.2427,
      "step": 100000
    },
    {
      "epoch": 12.034897713598074,
      "grad_norm": 221.97073364257812,
      "learning_rate": 0.00019356466664220177,
      "loss": 7.2748,
      "step": 100010
    },
    {
      "epoch": 12.036101083032491,
      "grad_norm": 1387.8858642578125,
      "learning_rate": 0.00019356332390266058,
      "loss": 7.2564,
      "step": 100020
    },
    {
      "epoch": 12.037304452466907,
      "grad_norm": 332.2581787109375,
      "learning_rate": 0.00019356198102771018,
      "loss": 7.2014,
      "step": 100030
    },
    {
      "epoch": 12.038507821901323,
      "grad_norm": 72.22911071777344,
      "learning_rate": 0.0001935606380173525,
      "loss": 7.2356,
      "step": 100040
    },
    {
      "epoch": 12.03971119133574,
      "grad_norm": 109.0367202758789,
      "learning_rate": 0.00019355929487158946,
      "loss": 7.2605,
      "step": 100050
    },
    {
      "epoch": 12.040914560770156,
      "grad_norm": 100.6845703125,
      "learning_rate": 0.000193557951590423,
      "loss": 7.2765,
      "step": 100060
    },
    {
      "epoch": 12.042117930204572,
      "grad_norm": 136.1052703857422,
      "learning_rate": 0.00019355660817385514,
      "loss": 7.196,
      "step": 100070
    },
    {
      "epoch": 12.04332129963899,
      "grad_norm": 159.7888641357422,
      "learning_rate": 0.00019355526462188776,
      "loss": 7.2828,
      "step": 100080
    },
    {
      "epoch": 12.044524669073406,
      "grad_norm": 85.20870208740234,
      "learning_rate": 0.00019355392093452277,
      "loss": 7.118,
      "step": 100090
    },
    {
      "epoch": 12.045728038507821,
      "grad_norm": 94.80852508544922,
      "learning_rate": 0.0001935525771117622,
      "loss": 7.2206,
      "step": 100100
    },
    {
      "epoch": 12.046931407942239,
      "grad_norm": 150.93528747558594,
      "learning_rate": 0.00019355123315360792,
      "loss": 7.1637,
      "step": 100110
    },
    {
      "epoch": 12.048134777376655,
      "grad_norm": 135.34988403320312,
      "learning_rate": 0.00019354988906006195,
      "loss": 7.2427,
      "step": 100120
    },
    {
      "epoch": 12.04933814681107,
      "grad_norm": 148.75509643554688,
      "learning_rate": 0.00019354854483112616,
      "loss": 7.2306,
      "step": 100130
    },
    {
      "epoch": 12.050541516245488,
      "grad_norm": 80.59191131591797,
      "learning_rate": 0.00019354720046680255,
      "loss": 7.2597,
      "step": 100140
    },
    {
      "epoch": 12.051744885679904,
      "grad_norm": 373.69921875,
      "learning_rate": 0.000193545855967093,
      "loss": 7.2687,
      "step": 100150
    },
    {
      "epoch": 12.05294825511432,
      "grad_norm": 255.21372985839844,
      "learning_rate": 0.00019354451133199958,
      "loss": 7.2234,
      "step": 100160
    },
    {
      "epoch": 12.054151624548737,
      "grad_norm": 236.16403198242188,
      "learning_rate": 0.0001935431665615241,
      "loss": 7.1656,
      "step": 100170
    },
    {
      "epoch": 12.055354993983153,
      "grad_norm": 212.81130981445312,
      "learning_rate": 0.0001935418216556686,
      "loss": 7.1653,
      "step": 100180
    },
    {
      "epoch": 12.056558363417569,
      "grad_norm": 103.7719497680664,
      "learning_rate": 0.00019354047661443495,
      "loss": 7.2399,
      "step": 100190
    },
    {
      "epoch": 12.057761732851986,
      "grad_norm": 243.9726104736328,
      "learning_rate": 0.00019353913143782517,
      "loss": 7.3054,
      "step": 100200
    },
    {
      "epoch": 12.058965102286402,
      "grad_norm": 116.71979522705078,
      "learning_rate": 0.00019353778612584113,
      "loss": 7.2362,
      "step": 100210
    },
    {
      "epoch": 12.060168471720818,
      "grad_norm": 432.28131103515625,
      "learning_rate": 0.00019353644067848484,
      "loss": 7.2501,
      "step": 100220
    },
    {
      "epoch": 12.061371841155236,
      "grad_norm": 178.4130859375,
      "learning_rate": 0.00019353509509575822,
      "loss": 7.1561,
      "step": 100230
    },
    {
      "epoch": 12.062575210589651,
      "grad_norm": 277.750732421875,
      "learning_rate": 0.00019353374937766327,
      "loss": 7.1189,
      "step": 100240
    },
    {
      "epoch": 12.063778580024067,
      "grad_norm": 190.09104919433594,
      "learning_rate": 0.00019353240352420182,
      "loss": 7.2278,
      "step": 100250
    },
    {
      "epoch": 12.064981949458483,
      "grad_norm": 275.77935791015625,
      "learning_rate": 0.00019353105753537595,
      "loss": 7.1461,
      "step": 100260
    },
    {
      "epoch": 12.0661853188929,
      "grad_norm": 272.4902648925781,
      "learning_rate": 0.0001935297114111875,
      "loss": 7.1968,
      "step": 100270
    },
    {
      "epoch": 12.067388688327316,
      "grad_norm": 303.45550537109375,
      "learning_rate": 0.0001935283651516385,
      "loss": 7.1336,
      "step": 100280
    },
    {
      "epoch": 12.068592057761732,
      "grad_norm": 273.66064453125,
      "learning_rate": 0.00019352701875673085,
      "loss": 7.2506,
      "step": 100290
    },
    {
      "epoch": 12.06979542719615,
      "grad_norm": 228.72760009765625,
      "learning_rate": 0.00019352567222646655,
      "loss": 7.2459,
      "step": 100300
    },
    {
      "epoch": 12.070998796630565,
      "grad_norm": 195.4521026611328,
      "learning_rate": 0.00019352432556084747,
      "loss": 7.1851,
      "step": 100310
    },
    {
      "epoch": 12.072202166064981,
      "grad_norm": 322.5349426269531,
      "learning_rate": 0.00019352297875987562,
      "loss": 7.2004,
      "step": 100320
    },
    {
      "epoch": 12.073405535499399,
      "grad_norm": 181.39096069335938,
      "learning_rate": 0.0001935216318235529,
      "loss": 7.2689,
      "step": 100330
    },
    {
      "epoch": 12.074608904933815,
      "grad_norm": 185.29345703125,
      "learning_rate": 0.0001935202847518813,
      "loss": 7.2695,
      "step": 100340
    },
    {
      "epoch": 12.07581227436823,
      "grad_norm": 216.049560546875,
      "learning_rate": 0.00019351893754486276,
      "loss": 7.2004,
      "step": 100350
    },
    {
      "epoch": 12.077015643802648,
      "grad_norm": 200.36883544921875,
      "learning_rate": 0.00019351759020249924,
      "loss": 7.1085,
      "step": 100360
    },
    {
      "epoch": 12.078219013237064,
      "grad_norm": 336.3588562011719,
      "learning_rate": 0.0001935162427247927,
      "loss": 7.2686,
      "step": 100370
    },
    {
      "epoch": 12.07942238267148,
      "grad_norm": 160.20558166503906,
      "learning_rate": 0.000193514895111745,
      "loss": 7.1704,
      "step": 100380
    },
    {
      "epoch": 12.080625752105897,
      "grad_norm": 234.55931091308594,
      "learning_rate": 0.00019351354736335823,
      "loss": 7.1785,
      "step": 100390
    },
    {
      "epoch": 12.081829121540313,
      "grad_norm": 168.11973571777344,
      "learning_rate": 0.00019351219947963424,
      "loss": 7.2997,
      "step": 100400
    },
    {
      "epoch": 12.083032490974729,
      "grad_norm": 309.0727233886719,
      "learning_rate": 0.00019351085146057502,
      "loss": 7.2163,
      "step": 100410
    },
    {
      "epoch": 12.084235860409146,
      "grad_norm": 311.2032165527344,
      "learning_rate": 0.0001935095033061825,
      "loss": 7.2075,
      "step": 100420
    },
    {
      "epoch": 12.085439229843562,
      "grad_norm": 341.4127197265625,
      "learning_rate": 0.00019350815501645863,
      "loss": 7.2174,
      "step": 100430
    },
    {
      "epoch": 12.086642599277978,
      "grad_norm": 175.08729553222656,
      "learning_rate": 0.00019350680659140537,
      "loss": 7.1565,
      "step": 100440
    },
    {
      "epoch": 12.087845968712395,
      "grad_norm": 221.8728485107422,
      "learning_rate": 0.00019350545803102471,
      "loss": 7.1756,
      "step": 100450
    },
    {
      "epoch": 12.089049338146811,
      "grad_norm": 121.81482696533203,
      "learning_rate": 0.00019350410933531852,
      "loss": 7.2155,
      "step": 100460
    },
    {
      "epoch": 12.090252707581227,
      "grad_norm": 347.6531066894531,
      "learning_rate": 0.00019350276050428885,
      "loss": 7.2554,
      "step": 100470
    },
    {
      "epoch": 12.091456077015645,
      "grad_norm": 175.2806396484375,
      "learning_rate": 0.00019350141153793757,
      "loss": 7.2311,
      "step": 100480
    },
    {
      "epoch": 12.09265944645006,
      "grad_norm": 457.0815124511719,
      "learning_rate": 0.00019350006243626665,
      "loss": 7.1906,
      "step": 100490
    },
    {
      "epoch": 12.093862815884476,
      "grad_norm": 398.2767639160156,
      "learning_rate": 0.00019349871319927807,
      "loss": 7.1875,
      "step": 100500
    },
    {
      "epoch": 12.095066185318894,
      "grad_norm": 233.47964477539062,
      "learning_rate": 0.00019349736382697378,
      "loss": 7.2554,
      "step": 100510
    },
    {
      "epoch": 12.09626955475331,
      "grad_norm": 416.8374938964844,
      "learning_rate": 0.00019349601431935574,
      "loss": 7.1829,
      "step": 100520
    },
    {
      "epoch": 12.097472924187725,
      "grad_norm": 489.1484680175781,
      "learning_rate": 0.00019349466467642582,
      "loss": 7.2226,
      "step": 100530
    },
    {
      "epoch": 12.098676293622143,
      "grad_norm": 625.876220703125,
      "learning_rate": 0.00019349331489818607,
      "loss": 7.2729,
      "step": 100540
    },
    {
      "epoch": 12.099879663056559,
      "grad_norm": 301.9091491699219,
      "learning_rate": 0.00019349196498463843,
      "loss": 7.2179,
      "step": 100550
    },
    {
      "epoch": 12.101083032490974,
      "grad_norm": 621.7656860351562,
      "learning_rate": 0.00019349061493578482,
      "loss": 7.2261,
      "step": 100560
    },
    {
      "epoch": 12.102286401925392,
      "grad_norm": 604.3579711914062,
      "learning_rate": 0.0001934892647516272,
      "loss": 7.2722,
      "step": 100570
    },
    {
      "epoch": 12.103489771359808,
      "grad_norm": 460.4038391113281,
      "learning_rate": 0.00019348791443216757,
      "loss": 7.1147,
      "step": 100580
    },
    {
      "epoch": 12.104693140794224,
      "grad_norm": 158.8830108642578,
      "learning_rate": 0.00019348656397740778,
      "loss": 7.1657,
      "step": 100590
    },
    {
      "epoch": 12.10589651022864,
      "grad_norm": 313.7085876464844,
      "learning_rate": 0.00019348521338734989,
      "loss": 7.2749,
      "step": 100600
    },
    {
      "epoch": 12.107099879663057,
      "grad_norm": 282.0453186035156,
      "learning_rate": 0.00019348386266199584,
      "loss": 7.2465,
      "step": 100610
    },
    {
      "epoch": 12.108303249097473,
      "grad_norm": 323.4155578613281,
      "learning_rate": 0.0001934825118013475,
      "loss": 7.252,
      "step": 100620
    },
    {
      "epoch": 12.109506618531888,
      "grad_norm": 138.65968322753906,
      "learning_rate": 0.00019348116080540693,
      "loss": 7.222,
      "step": 100630
    },
    {
      "epoch": 12.110709987966306,
      "grad_norm": 186.51385498046875,
      "learning_rate": 0.00019347980967417602,
      "loss": 7.2035,
      "step": 100640
    },
    {
      "epoch": 12.111913357400722,
      "grad_norm": 289.2260437011719,
      "learning_rate": 0.00019347845840765677,
      "loss": 7.3423,
      "step": 100650
    },
    {
      "epoch": 12.113116726835138,
      "grad_norm": 274.5934753417969,
      "learning_rate": 0.00019347710700585107,
      "loss": 7.2769,
      "step": 100660
    },
    {
      "epoch": 12.114320096269555,
      "grad_norm": 499.8416442871094,
      "learning_rate": 0.00019347575546876095,
      "loss": 7.1917,
      "step": 100670
    },
    {
      "epoch": 12.115523465703971,
      "grad_norm": 274.7139892578125,
      "learning_rate": 0.00019347440379638832,
      "loss": 7.1797,
      "step": 100680
    },
    {
      "epoch": 12.116726835138387,
      "grad_norm": 140.1739501953125,
      "learning_rate": 0.00019347305198873514,
      "loss": 7.2113,
      "step": 100690
    },
    {
      "epoch": 12.117930204572804,
      "grad_norm": 273.6724548339844,
      "learning_rate": 0.0001934717000458034,
      "loss": 7.3191,
      "step": 100700
    },
    {
      "epoch": 12.11913357400722,
      "grad_norm": 349.076171875,
      "learning_rate": 0.00019347034796759503,
      "loss": 7.1546,
      "step": 100710
    },
    {
      "epoch": 12.120336943441636,
      "grad_norm": 627.3411865234375,
      "learning_rate": 0.000193468995754112,
      "loss": 7.2496,
      "step": 100720
    },
    {
      "epoch": 12.121540312876053,
      "grad_norm": 437.5904235839844,
      "learning_rate": 0.0001934676434053562,
      "loss": 7.1129,
      "step": 100730
    },
    {
      "epoch": 12.12274368231047,
      "grad_norm": 351.3270263671875,
      "learning_rate": 0.00019346629092132972,
      "loss": 7.2853,
      "step": 100740
    },
    {
      "epoch": 12.123947051744885,
      "grad_norm": 334.68109130859375,
      "learning_rate": 0.0001934649383020344,
      "loss": 7.2624,
      "step": 100750
    },
    {
      "epoch": 12.125150421179303,
      "grad_norm": 448.0818786621094,
      "learning_rate": 0.00019346358554747225,
      "loss": 7.1592,
      "step": 100760
    },
    {
      "epoch": 12.126353790613718,
      "grad_norm": 298.28668212890625,
      "learning_rate": 0.0001934622326576452,
      "loss": 7.2455,
      "step": 100770
    },
    {
      "epoch": 12.127557160048134,
      "grad_norm": 426.3834228515625,
      "learning_rate": 0.00019346087963255523,
      "loss": 7.266,
      "step": 100780
    },
    {
      "epoch": 12.128760529482552,
      "grad_norm": 598.8390502929688,
      "learning_rate": 0.00019345952647220428,
      "loss": 7.2482,
      "step": 100790
    },
    {
      "epoch": 12.129963898916968,
      "grad_norm": 531.2196044921875,
      "learning_rate": 0.00019345817317659435,
      "loss": 7.2136,
      "step": 100800
    },
    {
      "epoch": 12.131167268351383,
      "grad_norm": 497.2467041015625,
      "learning_rate": 0.00019345681974572735,
      "loss": 7.2716,
      "step": 100810
    },
    {
      "epoch": 12.132370637785801,
      "grad_norm": 453.32135009765625,
      "learning_rate": 0.00019345546617960526,
      "loss": 7.3065,
      "step": 100820
    },
    {
      "epoch": 12.133574007220217,
      "grad_norm": 289.6233825683594,
      "learning_rate": 0.00019345411247823004,
      "loss": 7.3029,
      "step": 100830
    },
    {
      "epoch": 12.134777376654633,
      "grad_norm": 155.31907653808594,
      "learning_rate": 0.00019345275864160366,
      "loss": 7.2986,
      "step": 100840
    },
    {
      "epoch": 12.13598074608905,
      "grad_norm": 772.6607055664062,
      "learning_rate": 0.00019345140466972804,
      "loss": 7.2191,
      "step": 100850
    },
    {
      "epoch": 12.137184115523466,
      "grad_norm": 429.3636474609375,
      "learning_rate": 0.0001934500505626052,
      "loss": 7.2323,
      "step": 100860
    },
    {
      "epoch": 12.138387484957882,
      "grad_norm": 345.03570556640625,
      "learning_rate": 0.00019344869632023702,
      "loss": 7.2898,
      "step": 100870
    },
    {
      "epoch": 12.1395908543923,
      "grad_norm": 236.43984985351562,
      "learning_rate": 0.00019344734194262552,
      "loss": 7.2594,
      "step": 100880
    },
    {
      "epoch": 12.140794223826715,
      "grad_norm": 84.82162475585938,
      "learning_rate": 0.00019344598742977267,
      "loss": 7.1923,
      "step": 100890
    },
    {
      "epoch": 12.14199759326113,
      "grad_norm": 192.3656768798828,
      "learning_rate": 0.00019344463278168038,
      "loss": 7.2424,
      "step": 100900
    },
    {
      "epoch": 12.143200962695548,
      "grad_norm": 189.2291259765625,
      "learning_rate": 0.00019344327799835064,
      "loss": 7.2318,
      "step": 100910
    },
    {
      "epoch": 12.144404332129964,
      "grad_norm": 371.7607116699219,
      "learning_rate": 0.00019344192307978544,
      "loss": 7.2003,
      "step": 100920
    },
    {
      "epoch": 12.14560770156438,
      "grad_norm": 241.46026611328125,
      "learning_rate": 0.00019344056802598667,
      "loss": 7.1725,
      "step": 100930
    },
    {
      "epoch": 12.146811070998796,
      "grad_norm": 252.74256896972656,
      "learning_rate": 0.00019343921283695631,
      "loss": 7.2456,
      "step": 100940
    },
    {
      "epoch": 12.148014440433213,
      "grad_norm": 334.9472351074219,
      "learning_rate": 0.00019343785751269637,
      "loss": 7.1778,
      "step": 100950
    },
    {
      "epoch": 12.14921780986763,
      "grad_norm": 311.9128723144531,
      "learning_rate": 0.00019343650205320878,
      "loss": 7.1967,
      "step": 100960
    },
    {
      "epoch": 12.150421179302045,
      "grad_norm": 291.40325927734375,
      "learning_rate": 0.0001934351464584955,
      "loss": 7.2297,
      "step": 100970
    },
    {
      "epoch": 12.151624548736462,
      "grad_norm": 414.8866271972656,
      "learning_rate": 0.0001934337907285585,
      "loss": 7.2476,
      "step": 100980
    },
    {
      "epoch": 12.152827918170878,
      "grad_norm": 274.6656494140625,
      "learning_rate": 0.0001934324348633997,
      "loss": 7.205,
      "step": 100990
    },
    {
      "epoch": 12.154031287605294,
      "grad_norm": 248.60829162597656,
      "learning_rate": 0.00019343107886302113,
      "loss": 7.1317,
      "step": 101000
    },
    {
      "epoch": 12.155234657039712,
      "grad_norm": 367.2349853515625,
      "learning_rate": 0.00019342972272742474,
      "loss": 7.1926,
      "step": 101010
    },
    {
      "epoch": 12.156438026474127,
      "grad_norm": 451.37103271484375,
      "learning_rate": 0.00019342836645661247,
      "loss": 7.1502,
      "step": 101020
    },
    {
      "epoch": 12.157641395908543,
      "grad_norm": 340.48223876953125,
      "learning_rate": 0.00019342701005058626,
      "loss": 7.1377,
      "step": 101030
    },
    {
      "epoch": 12.15884476534296,
      "grad_norm": 185.7396240234375,
      "learning_rate": 0.00019342565350934812,
      "loss": 7.2292,
      "step": 101040
    },
    {
      "epoch": 12.160048134777377,
      "grad_norm": 409.4861755371094,
      "learning_rate": 0.00019342429683289999,
      "loss": 7.0853,
      "step": 101050
    },
    {
      "epoch": 12.161251504211792,
      "grad_norm": 263.13800048828125,
      "learning_rate": 0.00019342294002124382,
      "loss": 7.2936,
      "step": 101060
    },
    {
      "epoch": 12.16245487364621,
      "grad_norm": 302.5745544433594,
      "learning_rate": 0.0001934215830743816,
      "loss": 7.2643,
      "step": 101070
    },
    {
      "epoch": 12.163658243080626,
      "grad_norm": 307.70635986328125,
      "learning_rate": 0.00019342022599231528,
      "loss": 7.1645,
      "step": 101080
    },
    {
      "epoch": 12.164861612515042,
      "grad_norm": 89.23799133300781,
      "learning_rate": 0.00019341886877504685,
      "loss": 7.2784,
      "step": 101090
    },
    {
      "epoch": 12.166064981949459,
      "grad_norm": 333.6704406738281,
      "learning_rate": 0.00019341751142257823,
      "loss": 7.2502,
      "step": 101100
    },
    {
      "epoch": 12.167268351383875,
      "grad_norm": 210.12353515625,
      "learning_rate": 0.0001934161539349114,
      "loss": 7.2462,
      "step": 101110
    },
    {
      "epoch": 12.16847172081829,
      "grad_norm": 221.42861938476562,
      "learning_rate": 0.00019341479631204835,
      "loss": 7.2906,
      "step": 101120
    },
    {
      "epoch": 12.169675090252708,
      "grad_norm": 214.8041229248047,
      "learning_rate": 0.000193413438553991,
      "loss": 7.2193,
      "step": 101130
    },
    {
      "epoch": 12.170878459687124,
      "grad_norm": 155.70309448242188,
      "learning_rate": 0.0001934120806607414,
      "loss": 7.1688,
      "step": 101140
    },
    {
      "epoch": 12.17208182912154,
      "grad_norm": 303.963623046875,
      "learning_rate": 0.00019341072263230142,
      "loss": 7.1714,
      "step": 101150
    },
    {
      "epoch": 12.173285198555957,
      "grad_norm": 256.7359619140625,
      "learning_rate": 0.00019340936446867306,
      "loss": 7.1868,
      "step": 101160
    },
    {
      "epoch": 12.174488567990373,
      "grad_norm": 306.3113708496094,
      "learning_rate": 0.00019340800616985827,
      "loss": 7.2854,
      "step": 101170
    },
    {
      "epoch": 12.175691937424789,
      "grad_norm": 240.64564514160156,
      "learning_rate": 0.00019340664773585906,
      "loss": 7.3645,
      "step": 101180
    },
    {
      "epoch": 12.176895306859207,
      "grad_norm": 235.72113037109375,
      "learning_rate": 0.00019340528916667737,
      "loss": 7.2431,
      "step": 101190
    },
    {
      "epoch": 12.178098676293622,
      "grad_norm": 256.3206787109375,
      "learning_rate": 0.00019340393046231517,
      "loss": 7.3116,
      "step": 101200
    },
    {
      "epoch": 12.179302045728038,
      "grad_norm": 387.47802734375,
      "learning_rate": 0.00019340257162277439,
      "loss": 7.2167,
      "step": 101210
    },
    {
      "epoch": 12.180505415162456,
      "grad_norm": 675.2747192382812,
      "learning_rate": 0.00019340121264805707,
      "loss": 7.162,
      "step": 101220
    },
    {
      "epoch": 12.181708784596871,
      "grad_norm": 583.4373168945312,
      "learning_rate": 0.0001933998535381651,
      "loss": 7.1362,
      "step": 101230
    },
    {
      "epoch": 12.182912154031287,
      "grad_norm": 471.8766174316406,
      "learning_rate": 0.0001933984942931005,
      "loss": 7.1989,
      "step": 101240
    },
    {
      "epoch": 12.184115523465705,
      "grad_norm": 375.63250732421875,
      "learning_rate": 0.0001933971349128652,
      "loss": 7.189,
      "step": 101250
    },
    {
      "epoch": 12.18531889290012,
      "grad_norm": 297.8193664550781,
      "learning_rate": 0.0001933957753974612,
      "loss": 7.2216,
      "step": 101260
    },
    {
      "epoch": 12.186522262334536,
      "grad_norm": 402.51312255859375,
      "learning_rate": 0.00019339441574689044,
      "loss": 7.2672,
      "step": 101270
    },
    {
      "epoch": 12.187725631768952,
      "grad_norm": 308.5905456542969,
      "learning_rate": 0.00019339305596115494,
      "loss": 7.2918,
      "step": 101280
    },
    {
      "epoch": 12.18892900120337,
      "grad_norm": 390.3688049316406,
      "learning_rate": 0.0001933916960402566,
      "loss": 7.1727,
      "step": 101290
    },
    {
      "epoch": 12.190132370637786,
      "grad_norm": 352.4194030761719,
      "learning_rate": 0.00019339033598419746,
      "loss": 7.2127,
      "step": 101300
    },
    {
      "epoch": 12.191335740072201,
      "grad_norm": 315.3782043457031,
      "learning_rate": 0.0001933889757929794,
      "loss": 7.2809,
      "step": 101310
    },
    {
      "epoch": 12.192539109506619,
      "grad_norm": 755.7091064453125,
      "learning_rate": 0.0001933876154666044,
      "loss": 7.3368,
      "step": 101320
    },
    {
      "epoch": 12.193742478941035,
      "grad_norm": 426.1267395019531,
      "learning_rate": 0.00019338625500507455,
      "loss": 7.2162,
      "step": 101330
    },
    {
      "epoch": 12.19494584837545,
      "grad_norm": 403.9010009765625,
      "learning_rate": 0.0001933848944083917,
      "loss": 7.2365,
      "step": 101340
    },
    {
      "epoch": 12.196149217809868,
      "grad_norm": 528.127197265625,
      "learning_rate": 0.00019338353367655785,
      "loss": 7.2191,
      "step": 101350
    },
    {
      "epoch": 12.197352587244284,
      "grad_norm": 413.1864013671875,
      "learning_rate": 0.00019338217280957494,
      "loss": 7.1863,
      "step": 101360
    },
    {
      "epoch": 12.1985559566787,
      "grad_norm": 758.5889892578125,
      "learning_rate": 0.00019338081180744502,
      "loss": 7.2286,
      "step": 101370
    },
    {
      "epoch": 12.199759326113117,
      "grad_norm": 689.4074096679688,
      "learning_rate": 0.00019337945067016997,
      "loss": 7.3487,
      "step": 101380
    },
    {
      "epoch": 12.200962695547533,
      "grad_norm": 675.9899291992188,
      "learning_rate": 0.00019337808939775183,
      "loss": 7.3461,
      "step": 101390
    },
    {
      "epoch": 12.202166064981949,
      "grad_norm": 1049.57373046875,
      "learning_rate": 0.00019337672799019253,
      "loss": 7.2437,
      "step": 101400
    },
    {
      "epoch": 12.203369434416366,
      "grad_norm": 855.0003662109375,
      "learning_rate": 0.00019337536644749404,
      "loss": 7.3677,
      "step": 101410
    },
    {
      "epoch": 12.204572803850782,
      "grad_norm": 713.3817749023438,
      "learning_rate": 0.00019337400476965834,
      "loss": 7.3115,
      "step": 101420
    },
    {
      "epoch": 12.205776173285198,
      "grad_norm": 505.4274597167969,
      "learning_rate": 0.00019337264295668743,
      "loss": 7.3864,
      "step": 101430
    },
    {
      "epoch": 12.206979542719615,
      "grad_norm": 361.9706726074219,
      "learning_rate": 0.00019337128100858323,
      "loss": 7.2781,
      "step": 101440
    },
    {
      "epoch": 12.208182912154031,
      "grad_norm": 553.4201049804688,
      "learning_rate": 0.00019336991892534772,
      "loss": 7.2834,
      "step": 101450
    },
    {
      "epoch": 12.209386281588447,
      "grad_norm": 306.5463562011719,
      "learning_rate": 0.00019336855670698292,
      "loss": 7.1579,
      "step": 101460
    },
    {
      "epoch": 12.210589651022865,
      "grad_norm": 526.7699584960938,
      "learning_rate": 0.00019336719435349075,
      "loss": 7.3386,
      "step": 101470
    },
    {
      "epoch": 12.21179302045728,
      "grad_norm": 292.1778869628906,
      "learning_rate": 0.0001933658318648732,
      "loss": 7.2308,
      "step": 101480
    },
    {
      "epoch": 12.212996389891696,
      "grad_norm": 220.00936889648438,
      "learning_rate": 0.00019336446924113224,
      "loss": 7.2684,
      "step": 101490
    },
    {
      "epoch": 12.214199759326114,
      "grad_norm": 407.5308837890625,
      "learning_rate": 0.00019336310648226986,
      "loss": 7.2218,
      "step": 101500
    },
    {
      "epoch": 12.21540312876053,
      "grad_norm": 273.404052734375,
      "learning_rate": 0.00019336174358828797,
      "loss": 7.256,
      "step": 101510
    },
    {
      "epoch": 12.216606498194945,
      "grad_norm": 380.9325866699219,
      "learning_rate": 0.00019336038055918863,
      "loss": 7.1894,
      "step": 101520
    },
    {
      "epoch": 12.217809867629363,
      "grad_norm": 295.1746826171875,
      "learning_rate": 0.00019335901739497374,
      "loss": 7.1889,
      "step": 101530
    },
    {
      "epoch": 12.219013237063779,
      "grad_norm": 235.63473510742188,
      "learning_rate": 0.0001933576540956453,
      "loss": 7.2149,
      "step": 101540
    },
    {
      "epoch": 12.220216606498195,
      "grad_norm": 471.75274658203125,
      "learning_rate": 0.00019335629066120533,
      "loss": 7.0959,
      "step": 101550
    },
    {
      "epoch": 12.221419975932612,
      "grad_norm": 499.1694030761719,
      "learning_rate": 0.00019335492709165572,
      "loss": 7.2216,
      "step": 101560
    },
    {
      "epoch": 12.222623345367028,
      "grad_norm": 415.7610168457031,
      "learning_rate": 0.0001933535633869985,
      "loss": 7.2801,
      "step": 101570
    },
    {
      "epoch": 12.223826714801444,
      "grad_norm": 603.1621704101562,
      "learning_rate": 0.0001933521995472356,
      "loss": 7.2372,
      "step": 101580
    },
    {
      "epoch": 12.225030084235861,
      "grad_norm": 375.14453125,
      "learning_rate": 0.00019335083557236904,
      "loss": 7.3356,
      "step": 101590
    },
    {
      "epoch": 12.226233453670277,
      "grad_norm": 581.4219970703125,
      "learning_rate": 0.00019334947146240077,
      "loss": 7.2836,
      "step": 101600
    },
    {
      "epoch": 12.227436823104693,
      "grad_norm": 443.1958312988281,
      "learning_rate": 0.00019334810721733273,
      "loss": 7.2507,
      "step": 101610
    },
    {
      "epoch": 12.22864019253911,
      "grad_norm": 709.861572265625,
      "learning_rate": 0.00019334674283716697,
      "loss": 7.1872,
      "step": 101620
    },
    {
      "epoch": 12.229843561973526,
      "grad_norm": 319.0143127441406,
      "learning_rate": 0.00019334537832190545,
      "loss": 7.1973,
      "step": 101630
    },
    {
      "epoch": 12.231046931407942,
      "grad_norm": 544.3501586914062,
      "learning_rate": 0.00019334401367155008,
      "loss": 7.2598,
      "step": 101640
    },
    {
      "epoch": 12.232250300842358,
      "grad_norm": 923.862548828125,
      "learning_rate": 0.00019334264888610288,
      "loss": 7.2714,
      "step": 101650
    },
    {
      "epoch": 12.233453670276775,
      "grad_norm": 584.833251953125,
      "learning_rate": 0.00019334128396556586,
      "loss": 7.2273,
      "step": 101660
    },
    {
      "epoch": 12.234657039711191,
      "grad_norm": 1729.77978515625,
      "learning_rate": 0.00019333991890994092,
      "loss": 7.3817,
      "step": 101670
    },
    {
      "epoch": 12.235860409145607,
      "grad_norm": 1678.5914306640625,
      "learning_rate": 0.0001933385537192301,
      "loss": 7.1393,
      "step": 101680
    },
    {
      "epoch": 12.237063778580024,
      "grad_norm": 598.308349609375,
      "learning_rate": 0.00019333718839343531,
      "loss": 7.2668,
      "step": 101690
    },
    {
      "epoch": 12.23826714801444,
      "grad_norm": 1732.816650390625,
      "learning_rate": 0.0001933358229325586,
      "loss": 7.2058,
      "step": 101700
    },
    {
      "epoch": 12.239470517448856,
      "grad_norm": 798.1570434570312,
      "learning_rate": 0.00019333445733660186,
      "loss": 7.3182,
      "step": 101710
    },
    {
      "epoch": 12.240673886883274,
      "grad_norm": 1631.985107421875,
      "learning_rate": 0.00019333309160556717,
      "loss": 7.3232,
      "step": 101720
    },
    {
      "epoch": 12.24187725631769,
      "grad_norm": 1210.0718994140625,
      "learning_rate": 0.00019333172573945644,
      "loss": 7.2411,
      "step": 101730
    },
    {
      "epoch": 12.243080625752105,
      "grad_norm": 449.6768493652344,
      "learning_rate": 0.00019333035973827164,
      "loss": 7.2817,
      "step": 101740
    },
    {
      "epoch": 12.244283995186523,
      "grad_norm": 648.34130859375,
      "learning_rate": 0.0001933289936020148,
      "loss": 7.1525,
      "step": 101750
    },
    {
      "epoch": 12.245487364620939,
      "grad_norm": 1073.5108642578125,
      "learning_rate": 0.00019332762733068783,
      "loss": 7.1361,
      "step": 101760
    },
    {
      "epoch": 12.246690734055354,
      "grad_norm": 2393.293701171875,
      "learning_rate": 0.0001933262609242928,
      "loss": 7.1835,
      "step": 101770
    },
    {
      "epoch": 12.247894103489772,
      "grad_norm": 778.3978881835938,
      "learning_rate": 0.00019332489438283156,
      "loss": 7.2649,
      "step": 101780
    },
    {
      "epoch": 12.249097472924188,
      "grad_norm": 674.7791137695312,
      "learning_rate": 0.00019332352770630616,
      "loss": 7.1656,
      "step": 101790
    },
    {
      "epoch": 12.250300842358604,
      "grad_norm": 1137.3094482421875,
      "learning_rate": 0.0001933221608947186,
      "loss": 7.248,
      "step": 101800
    },
    {
      "epoch": 12.251504211793021,
      "grad_norm": 695.7918090820312,
      "learning_rate": 0.00019332079394807083,
      "loss": 7.2165,
      "step": 101810
    },
    {
      "epoch": 12.252707581227437,
      "grad_norm": 1119.053955078125,
      "learning_rate": 0.00019331942686636483,
      "loss": 7.2189,
      "step": 101820
    },
    {
      "epoch": 12.253910950661853,
      "grad_norm": 1177.5511474609375,
      "learning_rate": 0.00019331805964960258,
      "loss": 7.195,
      "step": 101830
    },
    {
      "epoch": 12.25511432009627,
      "grad_norm": 691.4961547851562,
      "learning_rate": 0.0001933166922977861,
      "loss": 7.1818,
      "step": 101840
    },
    {
      "epoch": 12.256317689530686,
      "grad_norm": 742.1613159179688,
      "learning_rate": 0.00019331532481091726,
      "loss": 7.2151,
      "step": 101850
    },
    {
      "epoch": 12.257521058965102,
      "grad_norm": 1167.755859375,
      "learning_rate": 0.00019331395718899814,
      "loss": 7.2807,
      "step": 101860
    },
    {
      "epoch": 12.25872442839952,
      "grad_norm": 894.4489135742188,
      "learning_rate": 0.00019331258943203067,
      "loss": 7.18,
      "step": 101870
    },
    {
      "epoch": 12.259927797833935,
      "grad_norm": 713.478271484375,
      "learning_rate": 0.00019331122154001685,
      "loss": 7.3303,
      "step": 101880
    },
    {
      "epoch": 12.261131167268351,
      "grad_norm": 774.05419921875,
      "learning_rate": 0.00019330985351295865,
      "loss": 7.2863,
      "step": 101890
    },
    {
      "epoch": 12.262334536702769,
      "grad_norm": 800.697265625,
      "learning_rate": 0.00019330848535085806,
      "loss": 7.3505,
      "step": 101900
    },
    {
      "epoch": 12.263537906137184,
      "grad_norm": 600.5453491210938,
      "learning_rate": 0.00019330711705371707,
      "loss": 7.3349,
      "step": 101910
    },
    {
      "epoch": 12.2647412755716,
      "grad_norm": 1391.663818359375,
      "learning_rate": 0.00019330574862153762,
      "loss": 7.3021,
      "step": 101920
    },
    {
      "epoch": 12.265944645006018,
      "grad_norm": 1171.2723388671875,
      "learning_rate": 0.00019330438005432172,
      "loss": 7.2055,
      "step": 101930
    },
    {
      "epoch": 12.267148014440433,
      "grad_norm": 942.2772827148438,
      "learning_rate": 0.00019330301135207136,
      "loss": 7.189,
      "step": 101940
    },
    {
      "epoch": 12.26835138387485,
      "grad_norm": 712.9710083007812,
      "learning_rate": 0.0001933016425147885,
      "loss": 7.3585,
      "step": 101950
    },
    {
      "epoch": 12.269554753309265,
      "grad_norm": 568.570556640625,
      "learning_rate": 0.0001933002735424751,
      "loss": 7.2893,
      "step": 101960
    },
    {
      "epoch": 12.270758122743683,
      "grad_norm": 841.6047973632812,
      "learning_rate": 0.0001932989044351332,
      "loss": 7.2358,
      "step": 101970
    },
    {
      "epoch": 12.271961492178098,
      "grad_norm": 508.7265319824219,
      "learning_rate": 0.00019329753519276477,
      "loss": 7.2913,
      "step": 101980
    },
    {
      "epoch": 12.273164861612514,
      "grad_norm": 491.8988342285156,
      "learning_rate": 0.00019329616581537174,
      "loss": 7.2009,
      "step": 101990
    },
    {
      "epoch": 12.274368231046932,
      "grad_norm": 633.57861328125,
      "learning_rate": 0.00019329479630295616,
      "loss": 7.2609,
      "step": 102000
    },
    {
      "epoch": 12.275571600481348,
      "grad_norm": 722.3191528320312,
      "learning_rate": 0.00019329342665551992,
      "loss": 7.2123,
      "step": 102010
    },
    {
      "epoch": 12.276774969915763,
      "grad_norm": 1048.0599365234375,
      "learning_rate": 0.0001932920568730651,
      "loss": 7.2431,
      "step": 102020
    },
    {
      "epoch": 12.277978339350181,
      "grad_norm": 657.6622924804688,
      "learning_rate": 0.00019329068695559363,
      "loss": 7.2258,
      "step": 102030
    },
    {
      "epoch": 12.279181708784597,
      "grad_norm": 400.900634765625,
      "learning_rate": 0.0001932893169031075,
      "loss": 7.1728,
      "step": 102040
    },
    {
      "epoch": 12.280385078219012,
      "grad_norm": 454.8036193847656,
      "learning_rate": 0.00019328794671560868,
      "loss": 7.2718,
      "step": 102050
    },
    {
      "epoch": 12.28158844765343,
      "grad_norm": 300.41357421875,
      "learning_rate": 0.0001932865763930992,
      "loss": 7.2075,
      "step": 102060
    },
    {
      "epoch": 12.282791817087846,
      "grad_norm": 546.9795532226562,
      "learning_rate": 0.000193285205935581,
      "loss": 7.3232,
      "step": 102070
    },
    {
      "epoch": 12.283995186522262,
      "grad_norm": 420.5072326660156,
      "learning_rate": 0.00019328383534305608,
      "loss": 7.2453,
      "step": 102080
    },
    {
      "epoch": 12.28519855595668,
      "grad_norm": 276.8692932128906,
      "learning_rate": 0.0001932824646155264,
      "loss": 7.1813,
      "step": 102090
    },
    {
      "epoch": 12.286401925391095,
      "grad_norm": 634.6013793945312,
      "learning_rate": 0.00019328109375299396,
      "loss": 7.245,
      "step": 102100
    },
    {
      "epoch": 12.28760529482551,
      "grad_norm": 369.4891052246094,
      "learning_rate": 0.00019327972275546076,
      "loss": 7.1738,
      "step": 102110
    },
    {
      "epoch": 12.288808664259928,
      "grad_norm": 1377.2781982421875,
      "learning_rate": 0.0001932783516229288,
      "loss": 7.1727,
      "step": 102120
    },
    {
      "epoch": 12.290012033694344,
      "grad_norm": 358.9952087402344,
      "learning_rate": 0.00019327698035539998,
      "loss": 7.2774,
      "step": 102130
    },
    {
      "epoch": 12.29121540312876,
      "grad_norm": 245.9854278564453,
      "learning_rate": 0.0001932756089528764,
      "loss": 7.2062,
      "step": 102140
    },
    {
      "epoch": 12.292418772563177,
      "grad_norm": 559.222900390625,
      "learning_rate": 0.00019327423741535995,
      "loss": 7.2896,
      "step": 102150
    },
    {
      "epoch": 12.293622141997593,
      "grad_norm": 480.6968688964844,
      "learning_rate": 0.00019327286574285263,
      "loss": 7.1975,
      "step": 102160
    },
    {
      "epoch": 12.294825511432009,
      "grad_norm": 815.69580078125,
      "learning_rate": 0.0001932714939353565,
      "loss": 7.3247,
      "step": 102170
    },
    {
      "epoch": 12.296028880866427,
      "grad_norm": 522.746337890625,
      "learning_rate": 0.00019327012199287346,
      "loss": 7.2509,
      "step": 102180
    },
    {
      "epoch": 12.297232250300842,
      "grad_norm": 238.93533325195312,
      "learning_rate": 0.0001932687499154055,
      "loss": 7.2088,
      "step": 102190
    },
    {
      "epoch": 12.298435619735258,
      "grad_norm": 459.73004150390625,
      "learning_rate": 0.00019326737770295466,
      "loss": 7.1601,
      "step": 102200
    },
    {
      "epoch": 12.299638989169676,
      "grad_norm": 394.89263916015625,
      "learning_rate": 0.0001932660053555229,
      "loss": 7.2681,
      "step": 102210
    },
    {
      "epoch": 12.300842358604092,
      "grad_norm": 660.4417724609375,
      "learning_rate": 0.00019326463287311222,
      "loss": 7.2479,
      "step": 102220
    },
    {
      "epoch": 12.302045728038507,
      "grad_norm": 1098.3001708984375,
      "learning_rate": 0.00019326326025572455,
      "loss": 7.1862,
      "step": 102230
    },
    {
      "epoch": 12.303249097472925,
      "grad_norm": 410.9057312011719,
      "learning_rate": 0.00019326188750336192,
      "loss": 7.3611,
      "step": 102240
    },
    {
      "epoch": 12.30445246690734,
      "grad_norm": 919.73193359375,
      "learning_rate": 0.0001932605146160263,
      "loss": 7.2923,
      "step": 102250
    },
    {
      "epoch": 12.305655836341757,
      "grad_norm": 616.4190673828125,
      "learning_rate": 0.00019325914159371975,
      "loss": 7.2314,
      "step": 102260
    },
    {
      "epoch": 12.306859205776174,
      "grad_norm": 473.75244140625,
      "learning_rate": 0.00019325776843644418,
      "loss": 7.2387,
      "step": 102270
    },
    {
      "epoch": 12.30806257521059,
      "grad_norm": 595.4457397460938,
      "learning_rate": 0.0001932563951442016,
      "loss": 7.3135,
      "step": 102280
    },
    {
      "epoch": 12.309265944645006,
      "grad_norm": 315.10076904296875,
      "learning_rate": 0.00019325502171699393,
      "loss": 7.2885,
      "step": 102290
    },
    {
      "epoch": 12.310469314079423,
      "grad_norm": 889.8563232421875,
      "learning_rate": 0.00019325364815482327,
      "loss": 7.2235,
      "step": 102300
    },
    {
      "epoch": 12.311672683513839,
      "grad_norm": 542.8992309570312,
      "learning_rate": 0.00019325227445769153,
      "loss": 7.1828,
      "step": 102310
    },
    {
      "epoch": 12.312876052948255,
      "grad_norm": 529.4866943359375,
      "learning_rate": 0.00019325090062560077,
      "loss": 7.2545,
      "step": 102320
    },
    {
      "epoch": 12.314079422382672,
      "grad_norm": 463.8544616699219,
      "learning_rate": 0.00019324952665855292,
      "loss": 7.2028,
      "step": 102330
    },
    {
      "epoch": 12.315282791817088,
      "grad_norm": 737.2639770507812,
      "learning_rate": 0.00019324815255654994,
      "loss": 7.2929,
      "step": 102340
    },
    {
      "epoch": 12.316486161251504,
      "grad_norm": 253.27159118652344,
      "learning_rate": 0.00019324677831959391,
      "loss": 7.2794,
      "step": 102350
    },
    {
      "epoch": 12.31768953068592,
      "grad_norm": 407.6042175292969,
      "learning_rate": 0.00019324540394768675,
      "loss": 7.338,
      "step": 102360
    },
    {
      "epoch": 12.318892900120337,
      "grad_norm": 563.1797485351562,
      "learning_rate": 0.00019324402944083047,
      "loss": 7.2382,
      "step": 102370
    },
    {
      "epoch": 12.320096269554753,
      "grad_norm": 697.7218017578125,
      "learning_rate": 0.00019324265479902706,
      "loss": 7.225,
      "step": 102380
    },
    {
      "epoch": 12.321299638989169,
      "grad_norm": 351.2344055175781,
      "learning_rate": 0.0001932412800222785,
      "loss": 7.2622,
      "step": 102390
    },
    {
      "epoch": 12.322503008423586,
      "grad_norm": 358.47821044921875,
      "learning_rate": 0.0001932399051105868,
      "loss": 7.3145,
      "step": 102400
    },
    {
      "epoch": 12.323706377858002,
      "grad_norm": 526.3197631835938,
      "learning_rate": 0.00019323853006395392,
      "loss": 7.3057,
      "step": 102410
    },
    {
      "epoch": 12.324909747292418,
      "grad_norm": 659.393310546875,
      "learning_rate": 0.00019323715488238187,
      "loss": 7.2431,
      "step": 102420
    },
    {
      "epoch": 12.326113116726836,
      "grad_norm": 350.7294616699219,
      "learning_rate": 0.00019323577956587267,
      "loss": 7.2883,
      "step": 102430
    },
    {
      "epoch": 12.327316486161251,
      "grad_norm": 313.63714599609375,
      "learning_rate": 0.00019323440411442823,
      "loss": 7.236,
      "step": 102440
    },
    {
      "epoch": 12.328519855595667,
      "grad_norm": 428.5990295410156,
      "learning_rate": 0.00019323302852805062,
      "loss": 7.1618,
      "step": 102450
    },
    {
      "epoch": 12.329723225030085,
      "grad_norm": 334.8892517089844,
      "learning_rate": 0.0001932316528067418,
      "loss": 7.1929,
      "step": 102460
    },
    {
      "epoch": 12.3309265944645,
      "grad_norm": 439.4962158203125,
      "learning_rate": 0.00019323027695050375,
      "loss": 7.161,
      "step": 102470
    },
    {
      "epoch": 12.332129963898916,
      "grad_norm": 531.44140625,
      "learning_rate": 0.00019322890095933848,
      "loss": 7.1904,
      "step": 102480
    },
    {
      "epoch": 12.333333333333334,
      "grad_norm": 258.1150207519531,
      "learning_rate": 0.00019322752483324796,
      "loss": 7.2999,
      "step": 102490
    },
    {
      "epoch": 12.33453670276775,
      "grad_norm": 538.8759155273438,
      "learning_rate": 0.0001932261485722342,
      "loss": 7.2119,
      "step": 102500
    },
    {
      "epoch": 12.335740072202166,
      "grad_norm": 449.08526611328125,
      "learning_rate": 0.00019322477217629917,
      "loss": 7.1645,
      "step": 102510
    },
    {
      "epoch": 12.336943441636583,
      "grad_norm": 315.3735656738281,
      "learning_rate": 0.0001932233956454449,
      "loss": 7.3677,
      "step": 102520
    },
    {
      "epoch": 12.338146811070999,
      "grad_norm": 227.24887084960938,
      "learning_rate": 0.00019322201897967336,
      "loss": 7.2955,
      "step": 102530
    },
    {
      "epoch": 12.339350180505415,
      "grad_norm": 391.2577209472656,
      "learning_rate": 0.00019322064217898652,
      "loss": 7.2652,
      "step": 102540
    },
    {
      "epoch": 12.340553549939832,
      "grad_norm": 557.8499755859375,
      "learning_rate": 0.0001932192652433864,
      "loss": 7.3406,
      "step": 102550
    },
    {
      "epoch": 12.341756919374248,
      "grad_norm": 406.92327880859375,
      "learning_rate": 0.000193217888172875,
      "loss": 7.2197,
      "step": 102560
    },
    {
      "epoch": 12.342960288808664,
      "grad_norm": 759.5283813476562,
      "learning_rate": 0.00019321651096745432,
      "loss": 7.3051,
      "step": 102570
    },
    {
      "epoch": 12.344163658243081,
      "grad_norm": 615.8949584960938,
      "learning_rate": 0.00019321513362712632,
      "loss": 7.3135,
      "step": 102580
    },
    {
      "epoch": 12.345367027677497,
      "grad_norm": 174.0048370361328,
      "learning_rate": 0.00019321375615189302,
      "loss": 7.3849,
      "step": 102590
    },
    {
      "epoch": 12.346570397111913,
      "grad_norm": 479.3607482910156,
      "learning_rate": 0.00019321237854175638,
      "loss": 7.1912,
      "step": 102600
    },
    {
      "epoch": 12.34777376654633,
      "grad_norm": 296.47662353515625,
      "learning_rate": 0.00019321100079671842,
      "loss": 7.2004,
      "step": 102610
    },
    {
      "epoch": 12.348977135980746,
      "grad_norm": 891.5411987304688,
      "learning_rate": 0.0001932096229167811,
      "loss": 7.2093,
      "step": 102620
    },
    {
      "epoch": 12.350180505415162,
      "grad_norm": 608.9300537109375,
      "learning_rate": 0.00019320824490194646,
      "loss": 7.2044,
      "step": 102630
    },
    {
      "epoch": 12.35138387484958,
      "grad_norm": 472.94873046875,
      "learning_rate": 0.00019320686675221651,
      "loss": 7.2476,
      "step": 102640
    },
    {
      "epoch": 12.352587244283995,
      "grad_norm": 593.6053466796875,
      "learning_rate": 0.0001932054884675932,
      "loss": 7.1844,
      "step": 102650
    },
    {
      "epoch": 12.353790613718411,
      "grad_norm": 395.6983642578125,
      "learning_rate": 0.0001932041100480785,
      "loss": 7.2608,
      "step": 102660
    },
    {
      "epoch": 12.354993983152827,
      "grad_norm": 480.7305603027344,
      "learning_rate": 0.00019320273149367448,
      "loss": 7.3589,
      "step": 102670
    },
    {
      "epoch": 12.356197352587245,
      "grad_norm": 636.570556640625,
      "learning_rate": 0.0001932013528043831,
      "loss": 7.297,
      "step": 102680
    },
    {
      "epoch": 12.35740072202166,
      "grad_norm": 383.88409423828125,
      "learning_rate": 0.00019319997398020632,
      "loss": 7.2024,
      "step": 102690
    },
    {
      "epoch": 12.358604091456076,
      "grad_norm": 790.86181640625,
      "learning_rate": 0.00019319859502114618,
      "loss": 7.2429,
      "step": 102700
    },
    {
      "epoch": 12.359807460890494,
      "grad_norm": 771.8855590820312,
      "learning_rate": 0.00019319721592720468,
      "loss": 7.2434,
      "step": 102710
    },
    {
      "epoch": 12.36101083032491,
      "grad_norm": 784.812744140625,
      "learning_rate": 0.0001931958366983838,
      "loss": 7.2969,
      "step": 102720
    },
    {
      "epoch": 12.362214199759325,
      "grad_norm": 528.169677734375,
      "learning_rate": 0.0001931944573346855,
      "loss": 7.17,
      "step": 102730
    },
    {
      "epoch": 12.363417569193743,
      "grad_norm": 669.7982788085938,
      "learning_rate": 0.00019319307783611183,
      "loss": 7.3552,
      "step": 102740
    },
    {
      "epoch": 12.364620938628159,
      "grad_norm": 1540.96484375,
      "learning_rate": 0.0001931916982026648,
      "loss": 7.2878,
      "step": 102750
    },
    {
      "epoch": 12.365824308062574,
      "grad_norm": 625.98974609375,
      "learning_rate": 0.0001931903184343463,
      "loss": 7.1922,
      "step": 102760
    },
    {
      "epoch": 12.367027677496992,
      "grad_norm": 4020.57275390625,
      "learning_rate": 0.00019318893853115845,
      "loss": 7.1787,
      "step": 102770
    },
    {
      "epoch": 12.368231046931408,
      "grad_norm": 863.9078979492188,
      "learning_rate": 0.00019318755849310318,
      "loss": 7.2502,
      "step": 102780
    },
    {
      "epoch": 12.369434416365824,
      "grad_norm": 522.5355224609375,
      "learning_rate": 0.00019318617832018254,
      "loss": 7.2442,
      "step": 102790
    },
    {
      "epoch": 12.370637785800241,
      "grad_norm": 1180.03515625,
      "learning_rate": 0.00019318479801239846,
      "loss": 7.206,
      "step": 102800
    },
    {
      "epoch": 12.371841155234657,
      "grad_norm": 842.2874145507812,
      "learning_rate": 0.000193183417569753,
      "loss": 7.2195,
      "step": 102810
    },
    {
      "epoch": 12.373044524669073,
      "grad_norm": 1174.523193359375,
      "learning_rate": 0.0001931820369922481,
      "loss": 7.2102,
      "step": 102820
    },
    {
      "epoch": 12.37424789410349,
      "grad_norm": 1294.15966796875,
      "learning_rate": 0.0001931806562798858,
      "loss": 7.2565,
      "step": 102830
    },
    {
      "epoch": 12.375451263537906,
      "grad_norm": 615.298828125,
      "learning_rate": 0.0001931792754326681,
      "loss": 7.2556,
      "step": 102840
    },
    {
      "epoch": 12.376654632972322,
      "grad_norm": 622.953125,
      "learning_rate": 0.00019317789445059695,
      "loss": 7.2275,
      "step": 102850
    },
    {
      "epoch": 12.37785800240674,
      "grad_norm": 941.1856079101562,
      "learning_rate": 0.0001931765133336744,
      "loss": 7.2496,
      "step": 102860
    },
    {
      "epoch": 12.379061371841155,
      "grad_norm": 527.05712890625,
      "learning_rate": 0.00019317513208190243,
      "loss": 7.2863,
      "step": 102870
    },
    {
      "epoch": 12.380264741275571,
      "grad_norm": 848.1106567382812,
      "learning_rate": 0.00019317375069528305,
      "loss": 7.3015,
      "step": 102880
    },
    {
      "epoch": 12.381468110709989,
      "grad_norm": 1050.827880859375,
      "learning_rate": 0.00019317236917381822,
      "loss": 7.3324,
      "step": 102890
    },
    {
      "epoch": 12.382671480144404,
      "grad_norm": 1029.7017822265625,
      "learning_rate": 0.00019317098751750998,
      "loss": 7.2734,
      "step": 102900
    },
    {
      "epoch": 12.38387484957882,
      "grad_norm": 531.5472412109375,
      "learning_rate": 0.00019316960572636033,
      "loss": 7.3096,
      "step": 102910
    },
    {
      "epoch": 12.385078219013238,
      "grad_norm": 997.7796630859375,
      "learning_rate": 0.00019316822380037124,
      "loss": 7.2518,
      "step": 102920
    },
    {
      "epoch": 12.386281588447654,
      "grad_norm": 685.7658081054688,
      "learning_rate": 0.00019316684173954475,
      "loss": 7.2804,
      "step": 102930
    },
    {
      "epoch": 12.38748495788207,
      "grad_norm": 733.6983642578125,
      "learning_rate": 0.00019316545954388284,
      "loss": 7.1888,
      "step": 102940
    },
    {
      "epoch": 12.388688327316487,
      "grad_norm": 650.7233276367188,
      "learning_rate": 0.0001931640772133875,
      "loss": 7.2076,
      "step": 102950
    },
    {
      "epoch": 12.389891696750903,
      "grad_norm": 778.4987182617188,
      "learning_rate": 0.00019316269474806074,
      "loss": 7.1991,
      "step": 102960
    },
    {
      "epoch": 12.391095066185319,
      "grad_norm": 1474.708740234375,
      "learning_rate": 0.00019316131214790452,
      "loss": 7.1736,
      "step": 102970
    },
    {
      "epoch": 12.392298435619736,
      "grad_norm": 842.0376586914062,
      "learning_rate": 0.00019315992941292093,
      "loss": 7.1433,
      "step": 102980
    },
    {
      "epoch": 12.393501805054152,
      "grad_norm": 567.975341796875,
      "learning_rate": 0.0001931585465431119,
      "loss": 7.2068,
      "step": 102990
    },
    {
      "epoch": 12.394705174488568,
      "grad_norm": 362.0715637207031,
      "learning_rate": 0.00019315716353847945,
      "loss": 7.2032,
      "step": 103000
    },
    {
      "epoch": 12.395908543922985,
      "grad_norm": 372.596923828125,
      "learning_rate": 0.0001931557803990256,
      "loss": 7.2915,
      "step": 103010
    },
    {
      "epoch": 12.397111913357401,
      "grad_norm": 757.5379638671875,
      "learning_rate": 0.00019315439712475234,
      "loss": 7.1542,
      "step": 103020
    },
    {
      "epoch": 12.398315282791817,
      "grad_norm": 518.6209716796875,
      "learning_rate": 0.00019315301371566163,
      "loss": 7.2141,
      "step": 103030
    },
    {
      "epoch": 12.399518652226233,
      "grad_norm": 845.45947265625,
      "learning_rate": 0.00019315163017175557,
      "loss": 7.2308,
      "step": 103040
    },
    {
      "epoch": 12.40072202166065,
      "grad_norm": 781.7288208007812,
      "learning_rate": 0.00019315024649303606,
      "loss": 7.2957,
      "step": 103050
    },
    {
      "epoch": 12.401925391095066,
      "grad_norm": 515.6735229492188,
      "learning_rate": 0.00019314886267950514,
      "loss": 7.2819,
      "step": 103060
    },
    {
      "epoch": 12.403128760529482,
      "grad_norm": 637.827392578125,
      "learning_rate": 0.00019314747873116486,
      "loss": 7.2471,
      "step": 103070
    },
    {
      "epoch": 12.4043321299639,
      "grad_norm": 947.3207397460938,
      "learning_rate": 0.00019314609464801715,
      "loss": 7.1896,
      "step": 103080
    },
    {
      "epoch": 12.405535499398315,
      "grad_norm": 634.7215576171875,
      "learning_rate": 0.00019314471043006405,
      "loss": 7.2115,
      "step": 103090
    },
    {
      "epoch": 12.406738868832731,
      "grad_norm": 304.4258117675781,
      "learning_rate": 0.00019314332607730755,
      "loss": 7.2377,
      "step": 103100
    },
    {
      "epoch": 12.407942238267148,
      "grad_norm": 465.19647216796875,
      "learning_rate": 0.00019314194158974966,
      "loss": 7.2257,
      "step": 103110
    },
    {
      "epoch": 12.409145607701564,
      "grad_norm": 611.6575317382812,
      "learning_rate": 0.00019314055696739238,
      "loss": 7.2662,
      "step": 103120
    },
    {
      "epoch": 12.41034897713598,
      "grad_norm": 615.4658813476562,
      "learning_rate": 0.00019313917221023776,
      "loss": 7.1679,
      "step": 103130
    },
    {
      "epoch": 12.411552346570398,
      "grad_norm": 315.9479675292969,
      "learning_rate": 0.0001931377873182877,
      "loss": 7.1305,
      "step": 103140
    },
    {
      "epoch": 12.412755716004813,
      "grad_norm": 859.2803955078125,
      "learning_rate": 0.00019313640229154432,
      "loss": 7.1801,
      "step": 103150
    },
    {
      "epoch": 12.41395908543923,
      "grad_norm": 311.7611389160156,
      "learning_rate": 0.00019313501713000956,
      "loss": 7.3273,
      "step": 103160
    },
    {
      "epoch": 12.415162454873647,
      "grad_norm": 421.8462829589844,
      "learning_rate": 0.00019313363183368542,
      "loss": 7.2462,
      "step": 103170
    },
    {
      "epoch": 12.416365824308063,
      "grad_norm": 573.8762817382812,
      "learning_rate": 0.00019313224640257393,
      "loss": 7.2063,
      "step": 103180
    },
    {
      "epoch": 12.417569193742478,
      "grad_norm": 419.0594482421875,
      "learning_rate": 0.00019313086083667708,
      "loss": 7.2863,
      "step": 103190
    },
    {
      "epoch": 12.418772563176896,
      "grad_norm": 545.3675537109375,
      "learning_rate": 0.00019312947513599687,
      "loss": 7.3011,
      "step": 103200
    },
    {
      "epoch": 12.419975932611312,
      "grad_norm": 540.0492553710938,
      "learning_rate": 0.00019312808930053535,
      "loss": 7.1821,
      "step": 103210
    },
    {
      "epoch": 12.421179302045728,
      "grad_norm": 625.4568481445312,
      "learning_rate": 0.00019312670333029446,
      "loss": 7.3009,
      "step": 103220
    },
    {
      "epoch": 12.422382671480145,
      "grad_norm": 500.1476135253906,
      "learning_rate": 0.00019312531722527626,
      "loss": 7.2518,
      "step": 103230
    },
    {
      "epoch": 12.42358604091456,
      "grad_norm": 470.7789001464844,
      "learning_rate": 0.0001931239309854827,
      "loss": 7.2698,
      "step": 103240
    },
    {
      "epoch": 12.424789410348977,
      "grad_norm": 505.5385437011719,
      "learning_rate": 0.00019312254461091585,
      "loss": 7.3202,
      "step": 103250
    },
    {
      "epoch": 12.425992779783394,
      "grad_norm": 788.8880004882812,
      "learning_rate": 0.00019312115810157768,
      "loss": 7.2149,
      "step": 103260
    },
    {
      "epoch": 12.42719614921781,
      "grad_norm": 1218.793212890625,
      "learning_rate": 0.00019311977145747023,
      "loss": 7.206,
      "step": 103270
    },
    {
      "epoch": 12.428399518652226,
      "grad_norm": 405.4271240234375,
      "learning_rate": 0.00019311838467859545,
      "loss": 7.241,
      "step": 103280
    },
    {
      "epoch": 12.429602888086643,
      "grad_norm": 570.1797485351562,
      "learning_rate": 0.0001931169977649554,
      "loss": 7.2073,
      "step": 103290
    },
    {
      "epoch": 12.43080625752106,
      "grad_norm": 1340.482421875,
      "learning_rate": 0.00019311561071655203,
      "loss": 7.2762,
      "step": 103300
    },
    {
      "epoch": 12.432009626955475,
      "grad_norm": 872.6629028320312,
      "learning_rate": 0.0001931142235333874,
      "loss": 7.1806,
      "step": 103310
    },
    {
      "epoch": 12.433212996389893,
      "grad_norm": 923.6234130859375,
      "learning_rate": 0.0001931128362154635,
      "loss": 7.2311,
      "step": 103320
    },
    {
      "epoch": 12.434416365824308,
      "grad_norm": 1062.56591796875,
      "learning_rate": 0.00019311144876278232,
      "loss": 7.255,
      "step": 103330
    },
    {
      "epoch": 12.435619735258724,
      "grad_norm": 936.6660766601562,
      "learning_rate": 0.0001931100611753459,
      "loss": 7.2829,
      "step": 103340
    },
    {
      "epoch": 12.43682310469314,
      "grad_norm": 2858.281982421875,
      "learning_rate": 0.00019310867345315625,
      "loss": 7.2651,
      "step": 103350
    },
    {
      "epoch": 12.438026474127557,
      "grad_norm": 916.9486083984375,
      "learning_rate": 0.00019310728559621533,
      "loss": 7.2414,
      "step": 103360
    },
    {
      "epoch": 12.439229843561973,
      "grad_norm": 588.5291137695312,
      "learning_rate": 0.0001931058976045252,
      "loss": 7.2024,
      "step": 103370
    },
    {
      "epoch": 12.440433212996389,
      "grad_norm": 2572.08203125,
      "learning_rate": 0.00019310450947808785,
      "loss": 7.2622,
      "step": 103380
    },
    {
      "epoch": 12.441636582430807,
      "grad_norm": 749.9818115234375,
      "learning_rate": 0.00019310312121690527,
      "loss": 7.2346,
      "step": 103390
    },
    {
      "epoch": 12.442839951865222,
      "grad_norm": 857.43359375,
      "learning_rate": 0.0001931017328209795,
      "loss": 7.2685,
      "step": 103400
    },
    {
      "epoch": 12.444043321299638,
      "grad_norm": 1062.347412109375,
      "learning_rate": 0.00019310034429031255,
      "loss": 7.1894,
      "step": 103410
    },
    {
      "epoch": 12.445246690734056,
      "grad_norm": 978.841064453125,
      "learning_rate": 0.0001930989556249064,
      "loss": 7.2423,
      "step": 103420
    },
    {
      "epoch": 12.446450060168472,
      "grad_norm": 3225.428466796875,
      "learning_rate": 0.00019309756682476308,
      "loss": 7.1619,
      "step": 103430
    },
    {
      "epoch": 12.447653429602887,
      "grad_norm": 729.1671752929688,
      "learning_rate": 0.0001930961778898846,
      "loss": 7.3156,
      "step": 103440
    },
    {
      "epoch": 12.448856799037305,
      "grad_norm": 2013.3564453125,
      "learning_rate": 0.00019309478882027295,
      "loss": 7.2963,
      "step": 103450
    },
    {
      "epoch": 12.45006016847172,
      "grad_norm": 1602.622802734375,
      "learning_rate": 0.0001930933996159302,
      "loss": 7.3292,
      "step": 103460
    },
    {
      "epoch": 12.451263537906136,
      "grad_norm": 1088.54345703125,
      "learning_rate": 0.00019309201027685826,
      "loss": 7.2578,
      "step": 103470
    },
    {
      "epoch": 12.452466907340554,
      "grad_norm": 1920.3013916015625,
      "learning_rate": 0.00019309062080305922,
      "loss": 7.2533,
      "step": 103480
    },
    {
      "epoch": 12.45367027677497,
      "grad_norm": 1126.5413818359375,
      "learning_rate": 0.00019308923119453505,
      "loss": 7.1836,
      "step": 103490
    },
    {
      "epoch": 12.454873646209386,
      "grad_norm": 1206.23974609375,
      "learning_rate": 0.00019308784145128781,
      "loss": 7.2413,
      "step": 103500
    },
    {
      "epoch": 12.456077015643803,
      "grad_norm": 611.6064453125,
      "learning_rate": 0.00019308645157331948,
      "loss": 7.278,
      "step": 103510
    },
    {
      "epoch": 12.457280385078219,
      "grad_norm": 1355.9659423828125,
      "learning_rate": 0.00019308506156063207,
      "loss": 7.2932,
      "step": 103520
    },
    {
      "epoch": 12.458483754512635,
      "grad_norm": 1553.713134765625,
      "learning_rate": 0.0001930836714132276,
      "loss": 7.2575,
      "step": 103530
    },
    {
      "epoch": 12.459687123947052,
      "grad_norm": 876.7474975585938,
      "learning_rate": 0.00019308228113110802,
      "loss": 7.3791,
      "step": 103540
    },
    {
      "epoch": 12.460890493381468,
      "grad_norm": 2083.907470703125,
      "learning_rate": 0.00019308089071427546,
      "loss": 7.1998,
      "step": 103550
    },
    {
      "epoch": 12.462093862815884,
      "grad_norm": 1367.4674072265625,
      "learning_rate": 0.00019307950016273184,
      "loss": 7.2563,
      "step": 103560
    },
    {
      "epoch": 12.463297232250302,
      "grad_norm": 2078.96923828125,
      "learning_rate": 0.00019307810947647922,
      "loss": 7.2632,
      "step": 103570
    },
    {
      "epoch": 12.464500601684717,
      "grad_norm": 1232.34814453125,
      "learning_rate": 0.00019307671865551957,
      "loss": 7.2679,
      "step": 103580
    },
    {
      "epoch": 12.465703971119133,
      "grad_norm": 737.5986328125,
      "learning_rate": 0.00019307532769985498,
      "loss": 7.2929,
      "step": 103590
    },
    {
      "epoch": 12.46690734055355,
      "grad_norm": 1198.1258544921875,
      "learning_rate": 0.00019307393660948735,
      "loss": 7.2022,
      "step": 103600
    },
    {
      "epoch": 12.468110709987966,
      "grad_norm": 1645.317626953125,
      "learning_rate": 0.0001930725453844188,
      "loss": 7.2737,
      "step": 103610
    },
    {
      "epoch": 12.469314079422382,
      "grad_norm": 2021.9385986328125,
      "learning_rate": 0.00019307115402465125,
      "loss": 7.2181,
      "step": 103620
    },
    {
      "epoch": 12.4705174488568,
      "grad_norm": 1294.0675048828125,
      "learning_rate": 0.0001930697625301868,
      "loss": 7.299,
      "step": 103630
    },
    {
      "epoch": 12.471720818291216,
      "grad_norm": 2241.862548828125,
      "learning_rate": 0.00019306837090102744,
      "loss": 7.2652,
      "step": 103640
    },
    {
      "epoch": 12.472924187725631,
      "grad_norm": 5705.3740234375,
      "learning_rate": 0.00019306697913717513,
      "loss": 7.2608,
      "step": 103650
    },
    {
      "epoch": 12.474127557160049,
      "grad_norm": 3235.27734375,
      "learning_rate": 0.00019306558723863196,
      "loss": 7.2397,
      "step": 103660
    },
    {
      "epoch": 12.475330926594465,
      "grad_norm": 7455.26025390625,
      "learning_rate": 0.00019306419520539987,
      "loss": 7.35,
      "step": 103670
    },
    {
      "epoch": 12.47653429602888,
      "grad_norm": 6502.79052734375,
      "learning_rate": 0.00019306280303748094,
      "loss": 7.2179,
      "step": 103680
    },
    {
      "epoch": 12.477737665463298,
      "grad_norm": 7488.34228515625,
      "learning_rate": 0.00019306141073487714,
      "loss": 7.2402,
      "step": 103690
    },
    {
      "epoch": 12.478941034897714,
      "grad_norm": 18683.474609375,
      "learning_rate": 0.00019306001829759052,
      "loss": 7.3177,
      "step": 103700
    },
    {
      "epoch": 12.48014440433213,
      "grad_norm": 20434.6953125,
      "learning_rate": 0.00019305862572562306,
      "loss": 7.2047,
      "step": 103710
    },
    {
      "epoch": 12.481347773766545,
      "grad_norm": 7194.01513671875,
      "learning_rate": 0.0001930572330189768,
      "loss": 7.2306,
      "step": 103720
    },
    {
      "epoch": 12.482551143200963,
      "grad_norm": 8411.98046875,
      "learning_rate": 0.00019305584017765374,
      "loss": 7.2336,
      "step": 103730
    },
    {
      "epoch": 12.483754512635379,
      "grad_norm": 3715.859375,
      "learning_rate": 0.00019305444720165593,
      "loss": 7.3076,
      "step": 103740
    },
    {
      "epoch": 12.484957882069795,
      "grad_norm": 4028.9384765625,
      "learning_rate": 0.00019305305409098535,
      "loss": 7.2069,
      "step": 103750
    },
    {
      "epoch": 12.486161251504212,
      "grad_norm": 5455.505859375,
      "learning_rate": 0.00019305166084564401,
      "loss": 7.2597,
      "step": 103760
    },
    {
      "epoch": 12.487364620938628,
      "grad_norm": 1815.1351318359375,
      "learning_rate": 0.00019305026746563395,
      "loss": 7.3252,
      "step": 103770
    },
    {
      "epoch": 12.488567990373044,
      "grad_norm": 8051.52294921875,
      "learning_rate": 0.00019304887395095717,
      "loss": 7.2213,
      "step": 103780
    },
    {
      "epoch": 12.489771359807461,
      "grad_norm": 7441.04150390625,
      "learning_rate": 0.00019304748030161574,
      "loss": 7.1658,
      "step": 103790
    },
    {
      "epoch": 12.490974729241877,
      "grad_norm": 1232.86962890625,
      "learning_rate": 0.0001930460865176116,
      "loss": 7.3066,
      "step": 103800
    },
    {
      "epoch": 12.492178098676293,
      "grad_norm": 7903.53271484375,
      "learning_rate": 0.00019304469259894683,
      "loss": 7.2759,
      "step": 103810
    },
    {
      "epoch": 12.49338146811071,
      "grad_norm": 2103.48046875,
      "learning_rate": 0.00019304329854562336,
      "loss": 7.3269,
      "step": 103820
    },
    {
      "epoch": 12.494584837545126,
      "grad_norm": 1905.847900390625,
      "learning_rate": 0.00019304190435764332,
      "loss": 7.4008,
      "step": 103830
    },
    {
      "epoch": 12.495788206979542,
      "grad_norm": 1053.1986083984375,
      "learning_rate": 0.00019304051003500865,
      "loss": 7.2427,
      "step": 103840
    },
    {
      "epoch": 12.49699157641396,
      "grad_norm": 1760.149658203125,
      "learning_rate": 0.00019303911557772135,
      "loss": 7.4051,
      "step": 103850
    },
    {
      "epoch": 12.498194945848375,
      "grad_norm": 1173.190185546875,
      "learning_rate": 0.00019303772098578357,
      "loss": 7.2239,
      "step": 103860
    },
    {
      "epoch": 12.499398315282791,
      "grad_norm": 3636.263916015625,
      "learning_rate": 0.00019303632625919718,
      "loss": 7.2492,
      "step": 103870
    },
    {
      "epoch": 12.500601684717209,
      "grad_norm": 1471.09033203125,
      "learning_rate": 0.00019303493139796426,
      "loss": 7.2483,
      "step": 103880
    },
    {
      "epoch": 12.501805054151625,
      "grad_norm": 1033.12451171875,
      "learning_rate": 0.00019303353640208682,
      "loss": 7.1177,
      "step": 103890
    },
    {
      "epoch": 12.50300842358604,
      "grad_norm": 1721.246826171875,
      "learning_rate": 0.0001930321412715669,
      "loss": 7.1704,
      "step": 103900
    },
    {
      "epoch": 12.504211793020458,
      "grad_norm": 907.1747436523438,
      "learning_rate": 0.00019303074600640648,
      "loss": 7.3458,
      "step": 103910
    },
    {
      "epoch": 12.505415162454874,
      "grad_norm": 1607.6934814453125,
      "learning_rate": 0.00019302935060660763,
      "loss": 7.2169,
      "step": 103920
    },
    {
      "epoch": 12.50661853188929,
      "grad_norm": 1713.159423828125,
      "learning_rate": 0.00019302795507217233,
      "loss": 7.2595,
      "step": 103930
    },
    {
      "epoch": 12.507821901323707,
      "grad_norm": 784.2911376953125,
      "learning_rate": 0.00019302655940310262,
      "loss": 7.158,
      "step": 103940
    },
    {
      "epoch": 12.509025270758123,
      "grad_norm": 1609.2664794921875,
      "learning_rate": 0.00019302516359940048,
      "loss": 7.287,
      "step": 103950
    },
    {
      "epoch": 12.510228640192539,
      "grad_norm": 936.5604248046875,
      "learning_rate": 0.000193023767661068,
      "loss": 7.1432,
      "step": 103960
    },
    {
      "epoch": 12.511432009626956,
      "grad_norm": 1297.2315673828125,
      "learning_rate": 0.00019302237158810715,
      "loss": 7.1673,
      "step": 103970
    },
    {
      "epoch": 12.512635379061372,
      "grad_norm": 1857.13427734375,
      "learning_rate": 0.00019302097538051994,
      "loss": 7.2295,
      "step": 103980
    },
    {
      "epoch": 12.513838748495788,
      "grad_norm": 1274.822998046875,
      "learning_rate": 0.00019301957903830843,
      "loss": 7.3262,
      "step": 103990
    },
    {
      "epoch": 12.515042117930205,
      "grad_norm": 1373.2264404296875,
      "learning_rate": 0.0001930181825614746,
      "loss": 7.2863,
      "step": 104000
    },
    {
      "epoch": 12.516245487364621,
      "grad_norm": 1836.18359375,
      "learning_rate": 0.00019301678595002052,
      "loss": 7.1068,
      "step": 104010
    },
    {
      "epoch": 12.517448856799037,
      "grad_norm": 5573.10791015625,
      "learning_rate": 0.00019301538920394823,
      "loss": 7.3332,
      "step": 104020
    },
    {
      "epoch": 12.518652226233453,
      "grad_norm": 2214.633544921875,
      "learning_rate": 0.00019301399232325963,
      "loss": 7.1898,
      "step": 104030
    },
    {
      "epoch": 12.51985559566787,
      "grad_norm": 846.73681640625,
      "learning_rate": 0.00019301259530795685,
      "loss": 7.2471,
      "step": 104040
    },
    {
      "epoch": 12.521058965102286,
      "grad_norm": 2576.935546875,
      "learning_rate": 0.00019301119815804188,
      "loss": 7.269,
      "step": 104050
    },
    {
      "epoch": 12.522262334536702,
      "grad_norm": 982.31884765625,
      "learning_rate": 0.00019300980087351674,
      "loss": 7.1956,
      "step": 104060
    },
    {
      "epoch": 12.52346570397112,
      "grad_norm": 1140.760498046875,
      "learning_rate": 0.00019300840345438345,
      "loss": 7.2854,
      "step": 104070
    },
    {
      "epoch": 12.524669073405535,
      "grad_norm": 2254.50390625,
      "learning_rate": 0.00019300700590064405,
      "loss": 7.2001,
      "step": 104080
    },
    {
      "epoch": 12.525872442839951,
      "grad_norm": 1678.689697265625,
      "learning_rate": 0.00019300560821230054,
      "loss": 7.2017,
      "step": 104090
    },
    {
      "epoch": 12.527075812274369,
      "grad_norm": 1556.8857421875,
      "learning_rate": 0.00019300421038935496,
      "loss": 7.2397,
      "step": 104100
    },
    {
      "epoch": 12.528279181708784,
      "grad_norm": 1105.6611328125,
      "learning_rate": 0.0001930028124318093,
      "loss": 7.1864,
      "step": 104110
    },
    {
      "epoch": 12.5294825511432,
      "grad_norm": 2138.776123046875,
      "learning_rate": 0.00019300141433966562,
      "loss": 7.3185,
      "step": 104120
    },
    {
      "epoch": 12.530685920577618,
      "grad_norm": 1353.1348876953125,
      "learning_rate": 0.00019300001611292596,
      "loss": 7.1513,
      "step": 104130
    },
    {
      "epoch": 12.531889290012034,
      "grad_norm": 757.1256103515625,
      "learning_rate": 0.00019299861775159228,
      "loss": 7.1477,
      "step": 104140
    },
    {
      "epoch": 12.53309265944645,
      "grad_norm": 2442.3095703125,
      "learning_rate": 0.00019299721925566666,
      "loss": 7.2617,
      "step": 104150
    },
    {
      "epoch": 12.534296028880867,
      "grad_norm": 1530.1597900390625,
      "learning_rate": 0.00019299582062515112,
      "loss": 7.3253,
      "step": 104160
    },
    {
      "epoch": 12.535499398315283,
      "grad_norm": 2273.302001953125,
      "learning_rate": 0.00019299442186004764,
      "loss": 7.1853,
      "step": 104170
    },
    {
      "epoch": 12.536702767749698,
      "grad_norm": 2154.60400390625,
      "learning_rate": 0.00019299302296035828,
      "loss": 7.2701,
      "step": 104180
    },
    {
      "epoch": 12.537906137184116,
      "grad_norm": 808.8679809570312,
      "learning_rate": 0.00019299162392608506,
      "loss": 7.2159,
      "step": 104190
    },
    {
      "epoch": 12.539109506618532,
      "grad_norm": 1620.1002197265625,
      "learning_rate": 0.00019299022475722998,
      "loss": 7.3142,
      "step": 104200
    },
    {
      "epoch": 12.540312876052948,
      "grad_norm": 898.46875,
      "learning_rate": 0.00019298882545379514,
      "loss": 7.2186,
      "step": 104210
    },
    {
      "epoch": 12.541516245487365,
      "grad_norm": 1303.6656494140625,
      "learning_rate": 0.00019298742601578245,
      "loss": 7.2974,
      "step": 104220
    },
    {
      "epoch": 12.542719614921781,
      "grad_norm": 852.11962890625,
      "learning_rate": 0.00019298602644319401,
      "loss": 7.2035,
      "step": 104230
    },
    {
      "epoch": 12.543922984356197,
      "grad_norm": 1093.3443603515625,
      "learning_rate": 0.00019298462673603185,
      "loss": 7.2162,
      "step": 104240
    },
    {
      "epoch": 12.545126353790614,
      "grad_norm": 2158.495849609375,
      "learning_rate": 0.00019298322689429798,
      "loss": 7.1457,
      "step": 104250
    },
    {
      "epoch": 12.54632972322503,
      "grad_norm": 1533.84423828125,
      "learning_rate": 0.00019298182691799442,
      "loss": 7.2129,
      "step": 104260
    },
    {
      "epoch": 12.547533092659446,
      "grad_norm": 1018.7616577148438,
      "learning_rate": 0.00019298042680712318,
      "loss": 7.1869,
      "step": 104270
    },
    {
      "epoch": 12.548736462093864,
      "grad_norm": 1718.45947265625,
      "learning_rate": 0.00019297902656168633,
      "loss": 7.2355,
      "step": 104280
    },
    {
      "epoch": 12.54993983152828,
      "grad_norm": 925.1983032226562,
      "learning_rate": 0.00019297762618168587,
      "loss": 7.2473,
      "step": 104290
    },
    {
      "epoch": 12.551143200962695,
      "grad_norm": 1960.1783447265625,
      "learning_rate": 0.00019297622566712383,
      "loss": 7.3286,
      "step": 104300
    },
    {
      "epoch": 12.552346570397113,
      "grad_norm": 1889.014404296875,
      "learning_rate": 0.0001929748250180022,
      "loss": 7.2456,
      "step": 104310
    },
    {
      "epoch": 12.553549939831528,
      "grad_norm": 1755.573974609375,
      "learning_rate": 0.0001929734242343231,
      "loss": 7.2467,
      "step": 104320
    },
    {
      "epoch": 12.554753309265944,
      "grad_norm": 1215.699951171875,
      "learning_rate": 0.00019297202331608845,
      "loss": 7.2599,
      "step": 104330
    },
    {
      "epoch": 12.555956678700362,
      "grad_norm": 1160.3714599609375,
      "learning_rate": 0.00019297062226330037,
      "loss": 7.101,
      "step": 104340
    },
    {
      "epoch": 12.557160048134778,
      "grad_norm": 1271.2740478515625,
      "learning_rate": 0.00019296922107596082,
      "loss": 7.2709,
      "step": 104350
    },
    {
      "epoch": 12.558363417569193,
      "grad_norm": 619.2705078125,
      "learning_rate": 0.00019296781975407189,
      "loss": 7.1826,
      "step": 104360
    },
    {
      "epoch": 12.559566787003611,
      "grad_norm": 739.95654296875,
      "learning_rate": 0.00019296641829763552,
      "loss": 7.1994,
      "step": 104370
    },
    {
      "epoch": 12.560770156438027,
      "grad_norm": 1045.4051513671875,
      "learning_rate": 0.00019296501670665382,
      "loss": 7.2761,
      "step": 104380
    },
    {
      "epoch": 12.561973525872443,
      "grad_norm": 870.4738159179688,
      "learning_rate": 0.00019296361498112878,
      "loss": 7.1822,
      "step": 104390
    },
    {
      "epoch": 12.56317689530686,
      "grad_norm": 959.38623046875,
      "learning_rate": 0.00019296221312106244,
      "loss": 7.2169,
      "step": 104400
    },
    {
      "epoch": 12.564380264741276,
      "grad_norm": 885.3505859375,
      "learning_rate": 0.00019296081112645682,
      "loss": 7.1756,
      "step": 104410
    },
    {
      "epoch": 12.565583634175692,
      "grad_norm": 988.3435668945312,
      "learning_rate": 0.00019295940899731397,
      "loss": 7.2232,
      "step": 104420
    },
    {
      "epoch": 12.566787003610107,
      "grad_norm": 777.8799438476562,
      "learning_rate": 0.00019295800673363588,
      "loss": 7.1958,
      "step": 104430
    },
    {
      "epoch": 12.567990373044525,
      "grad_norm": 521.8305053710938,
      "learning_rate": 0.00019295660433542465,
      "loss": 7.3149,
      "step": 104440
    },
    {
      "epoch": 12.56919374247894,
      "grad_norm": 587.714111328125,
      "learning_rate": 0.0001929552018026822,
      "loss": 7.2738,
      "step": 104450
    },
    {
      "epoch": 12.570397111913357,
      "grad_norm": 717.1371459960938,
      "learning_rate": 0.00019295379913541067,
      "loss": 7.2753,
      "step": 104460
    },
    {
      "epoch": 12.571600481347774,
      "grad_norm": 1174.6915283203125,
      "learning_rate": 0.00019295239633361204,
      "loss": 7.1544,
      "step": 104470
    },
    {
      "epoch": 12.57280385078219,
      "grad_norm": 1600.67236328125,
      "learning_rate": 0.00019295099339728834,
      "loss": 7.2503,
      "step": 104480
    },
    {
      "epoch": 12.574007220216606,
      "grad_norm": 1140.218994140625,
      "learning_rate": 0.0001929495903264416,
      "loss": 7.2818,
      "step": 104490
    },
    {
      "epoch": 12.575210589651023,
      "grad_norm": 426.3902282714844,
      "learning_rate": 0.00019294818712107384,
      "loss": 7.192,
      "step": 104500
    },
    {
      "epoch": 12.57641395908544,
      "grad_norm": 553.1784057617188,
      "learning_rate": 0.0001929467837811871,
      "loss": 7.1825,
      "step": 104510
    },
    {
      "epoch": 12.577617328519855,
      "grad_norm": 1348.63671875,
      "learning_rate": 0.00019294538030678344,
      "loss": 7.2816,
      "step": 104520
    },
    {
      "epoch": 12.578820697954272,
      "grad_norm": 550.9815063476562,
      "learning_rate": 0.00019294397669786486,
      "loss": 7.2044,
      "step": 104530
    },
    {
      "epoch": 12.580024067388688,
      "grad_norm": 659.5072021484375,
      "learning_rate": 0.00019294257295443342,
      "loss": 7.2081,
      "step": 104540
    },
    {
      "epoch": 12.581227436823104,
      "grad_norm": 1185.632080078125,
      "learning_rate": 0.00019294116907649106,
      "loss": 7.174,
      "step": 104550
    },
    {
      "epoch": 12.582430806257522,
      "grad_norm": 668.927734375,
      "learning_rate": 0.00019293976506403994,
      "loss": 7.1641,
      "step": 104560
    },
    {
      "epoch": 12.583634175691937,
      "grad_norm": 335.18792724609375,
      "learning_rate": 0.00019293836091708203,
      "loss": 7.1545,
      "step": 104570
    },
    {
      "epoch": 12.584837545126353,
      "grad_norm": 526.4817504882812,
      "learning_rate": 0.00019293695663561935,
      "loss": 7.1394,
      "step": 104580
    },
    {
      "epoch": 12.58604091456077,
      "grad_norm": 451.1551818847656,
      "learning_rate": 0.00019293555221965396,
      "loss": 7.1956,
      "step": 104590
    },
    {
      "epoch": 12.587244283995187,
      "grad_norm": 378.8392333984375,
      "learning_rate": 0.0001929341476691879,
      "loss": 7.2655,
      "step": 104600
    },
    {
      "epoch": 12.588447653429602,
      "grad_norm": 864.7603149414062,
      "learning_rate": 0.00019293274298422315,
      "loss": 7.2422,
      "step": 104610
    },
    {
      "epoch": 12.58965102286402,
      "grad_norm": 1140.2222900390625,
      "learning_rate": 0.0001929313381647618,
      "loss": 7.2524,
      "step": 104620
    },
    {
      "epoch": 12.590854392298436,
      "grad_norm": 321.66156005859375,
      "learning_rate": 0.00019292993321080583,
      "loss": 7.261,
      "step": 104630
    },
    {
      "epoch": 12.592057761732852,
      "grad_norm": 260.8164978027344,
      "learning_rate": 0.00019292852812235732,
      "loss": 7.2221,
      "step": 104640
    },
    {
      "epoch": 12.593261131167269,
      "grad_norm": 399.3897705078125,
      "learning_rate": 0.00019292712289941828,
      "loss": 7.2745,
      "step": 104650
    },
    {
      "epoch": 12.594464500601685,
      "grad_norm": 671.0757446289062,
      "learning_rate": 0.0001929257175419908,
      "loss": 7.2062,
      "step": 104660
    },
    {
      "epoch": 12.5956678700361,
      "grad_norm": 741.1159057617188,
      "learning_rate": 0.0001929243120500768,
      "loss": 7.2583,
      "step": 104670
    },
    {
      "epoch": 12.596871239470518,
      "grad_norm": 679.1481323242188,
      "learning_rate": 0.00019292290642367838,
      "loss": 7.1983,
      "step": 104680
    },
    {
      "epoch": 12.598074608904934,
      "grad_norm": 378.79522705078125,
      "learning_rate": 0.00019292150066279763,
      "loss": 7.3115,
      "step": 104690
    },
    {
      "epoch": 12.59927797833935,
      "grad_norm": 686.9435424804688,
      "learning_rate": 0.00019292009476743645,
      "loss": 7.1957,
      "step": 104700
    },
    {
      "epoch": 12.600481347773766,
      "grad_norm": 463.1709289550781,
      "learning_rate": 0.00019291868873759702,
      "loss": 7.2124,
      "step": 104710
    },
    {
      "epoch": 12.601684717208183,
      "grad_norm": 753.973876953125,
      "learning_rate": 0.00019291728257328128,
      "loss": 7.2216,
      "step": 104720
    },
    {
      "epoch": 12.602888086642599,
      "grad_norm": 511.06182861328125,
      "learning_rate": 0.00019291587627449127,
      "loss": 7.1613,
      "step": 104730
    },
    {
      "epoch": 12.604091456077015,
      "grad_norm": 201.756103515625,
      "learning_rate": 0.00019291446984122908,
      "loss": 7.1919,
      "step": 104740
    },
    {
      "epoch": 12.605294825511432,
      "grad_norm": 546.0944213867188,
      "learning_rate": 0.0001929130632734967,
      "loss": 7.2802,
      "step": 104750
    },
    {
      "epoch": 12.606498194945848,
      "grad_norm": 257.6075439453125,
      "learning_rate": 0.00019291165657129615,
      "loss": 7.2511,
      "step": 104760
    },
    {
      "epoch": 12.607701564380264,
      "grad_norm": 505.5705871582031,
      "learning_rate": 0.0001929102497346295,
      "loss": 7.2642,
      "step": 104770
    },
    {
      "epoch": 12.608904933814681,
      "grad_norm": 505.3547668457031,
      "learning_rate": 0.00019290884276349884,
      "loss": 7.1775,
      "step": 104780
    },
    {
      "epoch": 12.610108303249097,
      "grad_norm": 350.80865478515625,
      "learning_rate": 0.0001929074356579061,
      "loss": 7.2038,
      "step": 104790
    },
    {
      "epoch": 12.611311672683513,
      "grad_norm": 426.80780029296875,
      "learning_rate": 0.00019290602841785337,
      "loss": 7.2106,
      "step": 104800
    },
    {
      "epoch": 12.61251504211793,
      "grad_norm": 414.563232421875,
      "learning_rate": 0.00019290462104334268,
      "loss": 7.1431,
      "step": 104810
    },
    {
      "epoch": 12.613718411552346,
      "grad_norm": 480.8339538574219,
      "learning_rate": 0.00019290321353437605,
      "loss": 7.2138,
      "step": 104820
    },
    {
      "epoch": 12.614921780986762,
      "grad_norm": 470.5589599609375,
      "learning_rate": 0.00019290180589095557,
      "loss": 7.2506,
      "step": 104830
    },
    {
      "epoch": 12.61612515042118,
      "grad_norm": 571.3959350585938,
      "learning_rate": 0.0001929003981130832,
      "loss": 7.2327,
      "step": 104840
    },
    {
      "epoch": 12.617328519855596,
      "grad_norm": 1099.705810546875,
      "learning_rate": 0.00019289899020076104,
      "loss": 7.2126,
      "step": 104850
    },
    {
      "epoch": 12.618531889290011,
      "grad_norm": 587.909423828125,
      "learning_rate": 0.00019289758215399106,
      "loss": 7.233,
      "step": 104860
    },
    {
      "epoch": 12.619735258724429,
      "grad_norm": 929.6836547851562,
      "learning_rate": 0.00019289617397277539,
      "loss": 7.1646,
      "step": 104870
    },
    {
      "epoch": 12.620938628158845,
      "grad_norm": 789.2809448242188,
      "learning_rate": 0.00019289476565711604,
      "loss": 7.2197,
      "step": 104880
    },
    {
      "epoch": 12.62214199759326,
      "grad_norm": 526.9064331054688,
      "learning_rate": 0.00019289335720701497,
      "loss": 7.1982,
      "step": 104890
    },
    {
      "epoch": 12.623345367027678,
      "grad_norm": 775.0690307617188,
      "learning_rate": 0.00019289194862247432,
      "loss": 7.3218,
      "step": 104900
    },
    {
      "epoch": 12.624548736462094,
      "grad_norm": 640.3621215820312,
      "learning_rate": 0.0001928905399034961,
      "loss": 7.22,
      "step": 104910
    },
    {
      "epoch": 12.62575210589651,
      "grad_norm": 918.5899658203125,
      "learning_rate": 0.00019288913105008231,
      "loss": 7.2355,
      "step": 104920
    },
    {
      "epoch": 12.626955475330927,
      "grad_norm": 644.662109375,
      "learning_rate": 0.00019288772206223502,
      "loss": 7.1904,
      "step": 104930
    },
    {
      "epoch": 12.628158844765343,
      "grad_norm": 442.900146484375,
      "learning_rate": 0.00019288631293995625,
      "loss": 7.1846,
      "step": 104940
    },
    {
      "epoch": 12.629362214199759,
      "grad_norm": 376.1397399902344,
      "learning_rate": 0.00019288490368324804,
      "loss": 7.2564,
      "step": 104950
    },
    {
      "epoch": 12.630565583634176,
      "grad_norm": 332.4659118652344,
      "learning_rate": 0.00019288349429211248,
      "loss": 7.2008,
      "step": 104960
    },
    {
      "epoch": 12.631768953068592,
      "grad_norm": 511.4075622558594,
      "learning_rate": 0.00019288208476655157,
      "loss": 7.2184,
      "step": 104970
    },
    {
      "epoch": 12.632972322503008,
      "grad_norm": 344.4521484375,
      "learning_rate": 0.0001928806751065673,
      "loss": 7.1591,
      "step": 104980
    },
    {
      "epoch": 12.634175691937426,
      "grad_norm": 397.17681884765625,
      "learning_rate": 0.0001928792653121618,
      "loss": 7.2057,
      "step": 104990
    },
    {
      "epoch": 12.635379061371841,
      "grad_norm": 436.052490234375,
      "learning_rate": 0.00019287785538333707,
      "loss": 7.1754,
      "step": 105000
    },
    {
      "epoch": 12.636582430806257,
      "grad_norm": 525.3685913085938,
      "learning_rate": 0.00019287644532009517,
      "loss": 7.1748,
      "step": 105010
    },
    {
      "epoch": 12.637785800240675,
      "grad_norm": 969.0641479492188,
      "learning_rate": 0.00019287503512243807,
      "loss": 7.231,
      "step": 105020
    },
    {
      "epoch": 12.63898916967509,
      "grad_norm": 754.4326782226562,
      "learning_rate": 0.00019287362479036792,
      "loss": 7.1445,
      "step": 105030
    },
    {
      "epoch": 12.640192539109506,
      "grad_norm": 291.4302978515625,
      "learning_rate": 0.00019287221432388667,
      "loss": 7.1647,
      "step": 105040
    },
    {
      "epoch": 12.641395908543924,
      "grad_norm": 686.8651733398438,
      "learning_rate": 0.00019287080372299643,
      "loss": 7.3095,
      "step": 105050
    },
    {
      "epoch": 12.64259927797834,
      "grad_norm": 501.2530212402344,
      "learning_rate": 0.00019286939298769919,
      "loss": 7.2156,
      "step": 105060
    },
    {
      "epoch": 12.643802647412755,
      "grad_norm": 726.4334716796875,
      "learning_rate": 0.000192867982117997,
      "loss": 7.1564,
      "step": 105070
    },
    {
      "epoch": 12.645006016847173,
      "grad_norm": 427.7022705078125,
      "learning_rate": 0.00019286657111389192,
      "loss": 7.1946,
      "step": 105080
    },
    {
      "epoch": 12.646209386281589,
      "grad_norm": 403.608154296875,
      "learning_rate": 0.00019286515997538598,
      "loss": 7.176,
      "step": 105090
    },
    {
      "epoch": 12.647412755716005,
      "grad_norm": 723.849853515625,
      "learning_rate": 0.00019286374870248124,
      "loss": 7.1783,
      "step": 105100
    },
    {
      "epoch": 12.648616125150422,
      "grad_norm": 1141.3013916015625,
      "learning_rate": 0.0001928623372951797,
      "loss": 7.2682,
      "step": 105110
    },
    {
      "epoch": 12.649819494584838,
      "grad_norm": 1122.19580078125,
      "learning_rate": 0.00019286092575348344,
      "loss": 7.2814,
      "step": 105120
    },
    {
      "epoch": 12.651022864019254,
      "grad_norm": 1010.5200805664062,
      "learning_rate": 0.0001928595140773945,
      "loss": 7.3321,
      "step": 105130
    },
    {
      "epoch": 12.65222623345367,
      "grad_norm": 524.7896118164062,
      "learning_rate": 0.00019285810226691492,
      "loss": 7.3271,
      "step": 105140
    },
    {
      "epoch": 12.653429602888087,
      "grad_norm": 1147.3658447265625,
      "learning_rate": 0.00019285669032204674,
      "loss": 7.2735,
      "step": 105150
    },
    {
      "epoch": 12.654632972322503,
      "grad_norm": 14950.4248046875,
      "learning_rate": 0.000192855278242792,
      "loss": 7.2107,
      "step": 105160
    },
    {
      "epoch": 12.655836341756919,
      "grad_norm": 1618.273681640625,
      "learning_rate": 0.00019285386602915273,
      "loss": 7.2349,
      "step": 105170
    },
    {
      "epoch": 12.657039711191336,
      "grad_norm": 1051.3424072265625,
      "learning_rate": 0.00019285245368113102,
      "loss": 7.1108,
      "step": 105180
    },
    {
      "epoch": 12.658243080625752,
      "grad_norm": 1006.135009765625,
      "learning_rate": 0.0001928510411987289,
      "loss": 7.1417,
      "step": 105190
    },
    {
      "epoch": 12.659446450060168,
      "grad_norm": 886.9829711914062,
      "learning_rate": 0.00019284962858194838,
      "loss": 7.2302,
      "step": 105200
    },
    {
      "epoch": 12.660649819494585,
      "grad_norm": 1091.4029541015625,
      "learning_rate": 0.00019284821583079152,
      "loss": 7.2558,
      "step": 105210
    },
    {
      "epoch": 12.661853188929001,
      "grad_norm": 1300.4217529296875,
      "learning_rate": 0.00019284680294526035,
      "loss": 7.2654,
      "step": 105220
    },
    {
      "epoch": 12.663056558363417,
      "grad_norm": 915.7379150390625,
      "learning_rate": 0.00019284538992535695,
      "loss": 7.1368,
      "step": 105230
    },
    {
      "epoch": 12.664259927797834,
      "grad_norm": 637.56396484375,
      "learning_rate": 0.00019284397677108334,
      "loss": 7.1809,
      "step": 105240
    },
    {
      "epoch": 12.66546329723225,
      "grad_norm": 1624.16455078125,
      "learning_rate": 0.0001928425634824416,
      "loss": 7.245,
      "step": 105250
    },
    {
      "epoch": 12.666666666666666,
      "grad_norm": 1183.37451171875,
      "learning_rate": 0.0001928411500594337,
      "loss": 7.2306,
      "step": 105260
    },
    {
      "epoch": 12.667870036101084,
      "grad_norm": 1145.4989013671875,
      "learning_rate": 0.00019283973650206177,
      "loss": 7.2561,
      "step": 105270
    },
    {
      "epoch": 12.6690734055355,
      "grad_norm": 1351.666748046875,
      "learning_rate": 0.0001928383228103278,
      "loss": 7.2238,
      "step": 105280
    },
    {
      "epoch": 12.670276774969915,
      "grad_norm": 717.142822265625,
      "learning_rate": 0.00019283690898423388,
      "loss": 7.2542,
      "step": 105290
    },
    {
      "epoch": 12.671480144404333,
      "grad_norm": 2237.205322265625,
      "learning_rate": 0.000192835495023782,
      "loss": 7.2325,
      "step": 105300
    },
    {
      "epoch": 12.672683513838749,
      "grad_norm": 1041.7303466796875,
      "learning_rate": 0.0001928340809289743,
      "loss": 7.2826,
      "step": 105310
    },
    {
      "epoch": 12.673886883273164,
      "grad_norm": 1921.87646484375,
      "learning_rate": 0.0001928326666998127,
      "loss": 7.2943,
      "step": 105320
    },
    {
      "epoch": 12.675090252707582,
      "grad_norm": 1226.3826904296875,
      "learning_rate": 0.00019283125233629934,
      "loss": 7.1878,
      "step": 105330
    },
    {
      "epoch": 12.676293622141998,
      "grad_norm": 1008.4230346679688,
      "learning_rate": 0.00019282983783843624,
      "loss": 7.2731,
      "step": 105340
    },
    {
      "epoch": 12.677496991576414,
      "grad_norm": 2094.8876953125,
      "learning_rate": 0.00019282842320622544,
      "loss": 7.3077,
      "step": 105350
    },
    {
      "epoch": 12.678700361010831,
      "grad_norm": 1533.685302734375,
      "learning_rate": 0.00019282700843966898,
      "loss": 7.3553,
      "step": 105360
    },
    {
      "epoch": 12.679903730445247,
      "grad_norm": 2363.348388671875,
      "learning_rate": 0.00019282559353876895,
      "loss": 7.3194,
      "step": 105370
    },
    {
      "epoch": 12.681107099879663,
      "grad_norm": 1206.56005859375,
      "learning_rate": 0.00019282417850352733,
      "loss": 7.2537,
      "step": 105380
    },
    {
      "epoch": 12.68231046931408,
      "grad_norm": 1111.983642578125,
      "learning_rate": 0.00019282276333394625,
      "loss": 7.2669,
      "step": 105390
    },
    {
      "epoch": 12.683513838748496,
      "grad_norm": 1521.7628173828125,
      "learning_rate": 0.0001928213480300277,
      "loss": 7.2765,
      "step": 105400
    },
    {
      "epoch": 12.684717208182912,
      "grad_norm": 1030.261962890625,
      "learning_rate": 0.00019281993259177371,
      "loss": 7.2829,
      "step": 105410
    },
    {
      "epoch": 12.685920577617328,
      "grad_norm": 2103.438720703125,
      "learning_rate": 0.0001928185170191864,
      "loss": 7.1891,
      "step": 105420
    },
    {
      "epoch": 12.687123947051745,
      "grad_norm": 1855.0755615234375,
      "learning_rate": 0.00019281710131226776,
      "loss": 7.2707,
      "step": 105430
    },
    {
      "epoch": 12.688327316486161,
      "grad_norm": 1581.441162109375,
      "learning_rate": 0.00019281568547101986,
      "loss": 7.2825,
      "step": 105440
    },
    {
      "epoch": 12.689530685920577,
      "grad_norm": 1499.044921875,
      "learning_rate": 0.00019281426949544473,
      "loss": 7.2123,
      "step": 105450
    },
    {
      "epoch": 12.690734055354994,
      "grad_norm": 2257.961181640625,
      "learning_rate": 0.0001928128533855445,
      "loss": 7.2567,
      "step": 105460
    },
    {
      "epoch": 12.69193742478941,
      "grad_norm": 1788.625244140625,
      "learning_rate": 0.0001928114371413211,
      "loss": 7.2857,
      "step": 105470
    },
    {
      "epoch": 12.693140794223826,
      "grad_norm": 1475.48486328125,
      "learning_rate": 0.00019281002076277664,
      "loss": 7.0721,
      "step": 105480
    },
    {
      "epoch": 12.694344163658243,
      "grad_norm": 1593.0787353515625,
      "learning_rate": 0.0001928086042499132,
      "loss": 7.3759,
      "step": 105490
    },
    {
      "epoch": 12.69554753309266,
      "grad_norm": 1411.5899658203125,
      "learning_rate": 0.00019280718760273275,
      "loss": 7.2307,
      "step": 105500
    },
    {
      "epoch": 12.696750902527075,
      "grad_norm": 1971.3603515625,
      "learning_rate": 0.0001928057708212374,
      "loss": 7.3171,
      "step": 105510
    },
    {
      "epoch": 12.697954271961493,
      "grad_norm": 1886.3671875,
      "learning_rate": 0.00019280435390542922,
      "loss": 7.29,
      "step": 105520
    },
    {
      "epoch": 12.699157641395908,
      "grad_norm": 1795.4234619140625,
      "learning_rate": 0.00019280293685531018,
      "loss": 7.3123,
      "step": 105530
    },
    {
      "epoch": 12.700361010830324,
      "grad_norm": 2612.332763671875,
      "learning_rate": 0.0001928015196708824,
      "loss": 7.3679,
      "step": 105540
    },
    {
      "epoch": 12.701564380264742,
      "grad_norm": 2954.5986328125,
      "learning_rate": 0.0001928001023521479,
      "loss": 7.2931,
      "step": 105550
    },
    {
      "epoch": 12.702767749699158,
      "grad_norm": 1535.2064208984375,
      "learning_rate": 0.00019279868489910877,
      "loss": 7.3075,
      "step": 105560
    },
    {
      "epoch": 12.703971119133573,
      "grad_norm": 2086.252197265625,
      "learning_rate": 0.00019279726731176702,
      "loss": 7.2187,
      "step": 105570
    },
    {
      "epoch": 12.705174488567991,
      "grad_norm": 2196.475830078125,
      "learning_rate": 0.0001927958495901247,
      "loss": 7.3416,
      "step": 105580
    },
    {
      "epoch": 12.706377858002407,
      "grad_norm": 1500.63232421875,
      "learning_rate": 0.00019279443173418387,
      "loss": 7.2577,
      "step": 105590
    },
    {
      "epoch": 12.707581227436823,
      "grad_norm": 2814.72607421875,
      "learning_rate": 0.0001927930137439466,
      "loss": 7.3348,
      "step": 105600
    },
    {
      "epoch": 12.70878459687124,
      "grad_norm": 1926.3851318359375,
      "learning_rate": 0.00019279159561941495,
      "loss": 7.3492,
      "step": 105610
    },
    {
      "epoch": 12.709987966305656,
      "grad_norm": 2156.497314453125,
      "learning_rate": 0.0001927901773605909,
      "loss": 7.2432,
      "step": 105620
    },
    {
      "epoch": 12.711191335740072,
      "grad_norm": 3704.6865234375,
      "learning_rate": 0.0001927887589674766,
      "loss": 7.1535,
      "step": 105630
    },
    {
      "epoch": 12.71239470517449,
      "grad_norm": 1651.7120361328125,
      "learning_rate": 0.00019278734044007404,
      "loss": 7.2742,
      "step": 105640
    },
    {
      "epoch": 12.713598074608905,
      "grad_norm": 3680.71875,
      "learning_rate": 0.00019278592177838526,
      "loss": 7.2761,
      "step": 105650
    },
    {
      "epoch": 12.71480144404332,
      "grad_norm": 2541.768310546875,
      "learning_rate": 0.00019278450298241238,
      "loss": 7.2813,
      "step": 105660
    },
    {
      "epoch": 12.716004813477738,
      "grad_norm": 2693.757080078125,
      "learning_rate": 0.0001927830840521574,
      "loss": 7.2498,
      "step": 105670
    },
    {
      "epoch": 12.717208182912154,
      "grad_norm": 1763.8775634765625,
      "learning_rate": 0.0001927816649876224,
      "loss": 7.2485,
      "step": 105680
    },
    {
      "epoch": 12.71841155234657,
      "grad_norm": 1683.4327392578125,
      "learning_rate": 0.00019278024578880945,
      "loss": 7.1773,
      "step": 105690
    },
    {
      "epoch": 12.719614921780988,
      "grad_norm": 2728.01611328125,
      "learning_rate": 0.00019277882645572054,
      "loss": 7.3171,
      "step": 105700
    },
    {
      "epoch": 12.720818291215403,
      "grad_norm": 1501.5855712890625,
      "learning_rate": 0.00019277740698835774,
      "loss": 7.2085,
      "step": 105710
    },
    {
      "epoch": 12.722021660649819,
      "grad_norm": 2660.85302734375,
      "learning_rate": 0.0001927759873867232,
      "loss": 7.3015,
      "step": 105720
    },
    {
      "epoch": 12.723225030084237,
      "grad_norm": 1528.508544921875,
      "learning_rate": 0.00019277456765081884,
      "loss": 7.3208,
      "step": 105730
    },
    {
      "epoch": 12.724428399518652,
      "grad_norm": 1237.6644287109375,
      "learning_rate": 0.0001927731477806468,
      "loss": 7.2579,
      "step": 105740
    },
    {
      "epoch": 12.725631768953068,
      "grad_norm": 1447.3583984375,
      "learning_rate": 0.00019277172777620912,
      "loss": 7.2394,
      "step": 105750
    },
    {
      "epoch": 12.726835138387486,
      "grad_norm": 1225.0308837890625,
      "learning_rate": 0.00019277030763750782,
      "loss": 7.2694,
      "step": 105760
    },
    {
      "epoch": 12.728038507821902,
      "grad_norm": 2391.46533203125,
      "learning_rate": 0.00019276888736454497,
      "loss": 7.2474,
      "step": 105770
    },
    {
      "epoch": 12.729241877256317,
      "grad_norm": 1786.611328125,
      "learning_rate": 0.00019276746695732268,
      "loss": 7.1899,
      "step": 105780
    },
    {
      "epoch": 12.730445246690735,
      "grad_norm": 1737.0618896484375,
      "learning_rate": 0.00019276604641584293,
      "loss": 7.2423,
      "step": 105790
    },
    {
      "epoch": 12.73164861612515,
      "grad_norm": 4655.181640625,
      "learning_rate": 0.00019276462574010781,
      "loss": 7.2267,
      "step": 105800
    },
    {
      "epoch": 12.732851985559567,
      "grad_norm": 1269.7513427734375,
      "learning_rate": 0.00019276320493011938,
      "loss": 7.2454,
      "step": 105810
    },
    {
      "epoch": 12.734055354993982,
      "grad_norm": 1343.595458984375,
      "learning_rate": 0.0001927617839858797,
      "loss": 7.248,
      "step": 105820
    },
    {
      "epoch": 12.7352587244284,
      "grad_norm": 1270.7225341796875,
      "learning_rate": 0.00019276036290739082,
      "loss": 7.2976,
      "step": 105830
    },
    {
      "epoch": 12.736462093862816,
      "grad_norm": 1309.854736328125,
      "learning_rate": 0.00019275894169465475,
      "loss": 7.1258,
      "step": 105840
    },
    {
      "epoch": 12.737665463297231,
      "grad_norm": 1100.3641357421875,
      "learning_rate": 0.00019275752034767364,
      "loss": 7.2197,
      "step": 105850
    },
    {
      "epoch": 12.738868832731649,
      "grad_norm": 753.73974609375,
      "learning_rate": 0.00019275609886644946,
      "loss": 7.1745,
      "step": 105860
    },
    {
      "epoch": 12.740072202166065,
      "grad_norm": 741.9046020507812,
      "learning_rate": 0.00019275467725098433,
      "loss": 7.2588,
      "step": 105870
    },
    {
      "epoch": 12.74127557160048,
      "grad_norm": 1046.6353759765625,
      "learning_rate": 0.00019275325550128028,
      "loss": 7.2604,
      "step": 105880
    },
    {
      "epoch": 12.742478941034898,
      "grad_norm": 1046.2142333984375,
      "learning_rate": 0.00019275183361733934,
      "loss": 7.3747,
      "step": 105890
    },
    {
      "epoch": 12.743682310469314,
      "grad_norm": 1912.5325927734375,
      "learning_rate": 0.00019275041159916363,
      "loss": 7.2507,
      "step": 105900
    },
    {
      "epoch": 12.74488567990373,
      "grad_norm": 1162.363037109375,
      "learning_rate": 0.00019274898944675515,
      "loss": 7.2649,
      "step": 105910
    },
    {
      "epoch": 12.746089049338147,
      "grad_norm": 1675.916015625,
      "learning_rate": 0.000192747567160116,
      "loss": 7.3377,
      "step": 105920
    },
    {
      "epoch": 12.747292418772563,
      "grad_norm": 1460.8504638671875,
      "learning_rate": 0.00019274614473924822,
      "loss": 7.1886,
      "step": 105930
    },
    {
      "epoch": 12.748495788206979,
      "grad_norm": 566.094482421875,
      "learning_rate": 0.00019274472218415385,
      "loss": 7.2528,
      "step": 105940
    },
    {
      "epoch": 12.749699157641396,
      "grad_norm": 681.6715087890625,
      "learning_rate": 0.00019274329949483501,
      "loss": 7.284,
      "step": 105950
    },
    {
      "epoch": 12.750902527075812,
      "grad_norm": 891.517822265625,
      "learning_rate": 0.00019274187667129368,
      "loss": 7.3392,
      "step": 105960
    },
    {
      "epoch": 12.752105896510228,
      "grad_norm": 1218.25390625,
      "learning_rate": 0.00019274045371353197,
      "loss": 7.1866,
      "step": 105970
    },
    {
      "epoch": 12.753309265944646,
      "grad_norm": 1238.319580078125,
      "learning_rate": 0.0001927390306215519,
      "loss": 7.3285,
      "step": 105980
    },
    {
      "epoch": 12.754512635379061,
      "grad_norm": 529.3389282226562,
      "learning_rate": 0.00019273760739535562,
      "loss": 7.2722,
      "step": 105990
    },
    {
      "epoch": 12.755716004813477,
      "grad_norm": 1807.80615234375,
      "learning_rate": 0.0001927361840349451,
      "loss": 7.3601,
      "step": 106000
    },
    {
      "epoch": 12.756919374247895,
      "grad_norm": 506.76751708984375,
      "learning_rate": 0.0001927347605403224,
      "loss": 7.2907,
      "step": 106010
    },
    {
      "epoch": 12.75812274368231,
      "grad_norm": 1444.1904296875,
      "learning_rate": 0.00019273333691148965,
      "loss": 7.2053,
      "step": 106020
    },
    {
      "epoch": 12.759326113116726,
      "grad_norm": 771.5220947265625,
      "learning_rate": 0.0001927319131484488,
      "loss": 7.2297,
      "step": 106030
    },
    {
      "epoch": 12.760529482551144,
      "grad_norm": 797.7838134765625,
      "learning_rate": 0.00019273048925120203,
      "loss": 7.2448,
      "step": 106040
    },
    {
      "epoch": 12.76173285198556,
      "grad_norm": 795.5725708007812,
      "learning_rate": 0.00019272906521975133,
      "loss": 7.1719,
      "step": 106050
    },
    {
      "epoch": 12.762936221419976,
      "grad_norm": 946.47216796875,
      "learning_rate": 0.0001927276410540988,
      "loss": 7.2363,
      "step": 106060
    },
    {
      "epoch": 12.764139590854393,
      "grad_norm": 1349.8101806640625,
      "learning_rate": 0.00019272621675424647,
      "loss": 7.2639,
      "step": 106070
    },
    {
      "epoch": 12.765342960288809,
      "grad_norm": 911.401611328125,
      "learning_rate": 0.00019272479232019637,
      "loss": 7.2361,
      "step": 106080
    },
    {
      "epoch": 12.766546329723225,
      "grad_norm": 820.5132446289062,
      "learning_rate": 0.00019272336775195067,
      "loss": 7.2526,
      "step": 106090
    },
    {
      "epoch": 12.76774969915764,
      "grad_norm": 1301.506591796875,
      "learning_rate": 0.0001927219430495113,
      "loss": 7.2908,
      "step": 106100
    },
    {
      "epoch": 12.768953068592058,
      "grad_norm": 391.9560546875,
      "learning_rate": 0.00019272051821288044,
      "loss": 7.131,
      "step": 106110
    },
    {
      "epoch": 12.770156438026474,
      "grad_norm": 898.4758911132812,
      "learning_rate": 0.00019271909324206006,
      "loss": 7.2503,
      "step": 106120
    },
    {
      "epoch": 12.77135980746089,
      "grad_norm": 616.5748901367188,
      "learning_rate": 0.00019271766813705227,
      "loss": 7.1981,
      "step": 106130
    },
    {
      "epoch": 12.772563176895307,
      "grad_norm": 451.3939208984375,
      "learning_rate": 0.00019271624289785914,
      "loss": 7.2903,
      "step": 106140
    },
    {
      "epoch": 12.773766546329723,
      "grad_norm": 863.1996459960938,
      "learning_rate": 0.0001927148175244827,
      "loss": 7.1782,
      "step": 106150
    },
    {
      "epoch": 12.774969915764139,
      "grad_norm": 399.4698181152344,
      "learning_rate": 0.000192713392016925,
      "loss": 7.2379,
      "step": 106160
    },
    {
      "epoch": 12.776173285198556,
      "grad_norm": 548.5631713867188,
      "learning_rate": 0.00019271196637518818,
      "loss": 7.2288,
      "step": 106170
    },
    {
      "epoch": 12.777376654632972,
      "grad_norm": 600.9078979492188,
      "learning_rate": 0.00019271054059927424,
      "loss": 7.139,
      "step": 106180
    },
    {
      "epoch": 12.778580024067388,
      "grad_norm": 303.4968566894531,
      "learning_rate": 0.00019270911468918526,
      "loss": 7.279,
      "step": 106190
    },
    {
      "epoch": 12.779783393501805,
      "grad_norm": 1112.216796875,
      "learning_rate": 0.0001927076886449233,
      "loss": 7.2589,
      "step": 106200
    },
    {
      "epoch": 12.780986762936221,
      "grad_norm": 928.738037109375,
      "learning_rate": 0.00019270626246649042,
      "loss": 7.2635,
      "step": 106210
    },
    {
      "epoch": 12.782190132370637,
      "grad_norm": 775.9215698242188,
      "learning_rate": 0.0001927048361538887,
      "loss": 7.1795,
      "step": 106220
    },
    {
      "epoch": 12.783393501805055,
      "grad_norm": 621.3668823242188,
      "learning_rate": 0.0001927034097071202,
      "loss": 7.271,
      "step": 106230
    },
    {
      "epoch": 12.78459687123947,
      "grad_norm": 463.00164794921875,
      "learning_rate": 0.00019270198312618697,
      "loss": 7.236,
      "step": 106240
    },
    {
      "epoch": 12.785800240673886,
      "grad_norm": 1090.231689453125,
      "learning_rate": 0.00019270055641109108,
      "loss": 7.3159,
      "step": 106250
    },
    {
      "epoch": 12.787003610108304,
      "grad_norm": 494.4300537109375,
      "learning_rate": 0.00019269912956183457,
      "loss": 7.2348,
      "step": 106260
    },
    {
      "epoch": 12.78820697954272,
      "grad_norm": 773.3367919921875,
      "learning_rate": 0.00019269770257841957,
      "loss": 7.3009,
      "step": 106270
    },
    {
      "epoch": 12.789410348977135,
      "grad_norm": 508.0045166015625,
      "learning_rate": 0.0001926962754608481,
      "loss": 7.259,
      "step": 106280
    },
    {
      "epoch": 12.790613718411553,
      "grad_norm": 497.3860778808594,
      "learning_rate": 0.00019269484820912225,
      "loss": 7.2769,
      "step": 106290
    },
    {
      "epoch": 12.791817087845969,
      "grad_norm": 914.2775268554688,
      "learning_rate": 0.00019269342082324403,
      "loss": 7.2492,
      "step": 106300
    },
    {
      "epoch": 12.793020457280385,
      "grad_norm": 297.7761535644531,
      "learning_rate": 0.00019269199330321557,
      "loss": 7.1343,
      "step": 106310
    },
    {
      "epoch": 12.794223826714802,
      "grad_norm": 1316.914794921875,
      "learning_rate": 0.0001926905656490389,
      "loss": 7.1067,
      "step": 106320
    },
    {
      "epoch": 12.795427196149218,
      "grad_norm": 536.9522705078125,
      "learning_rate": 0.00019268913786071613,
      "loss": 7.2784,
      "step": 106330
    },
    {
      "epoch": 12.796630565583634,
      "grad_norm": 504.66485595703125,
      "learning_rate": 0.00019268770993824923,
      "loss": 7.3043,
      "step": 106340
    },
    {
      "epoch": 12.797833935018051,
      "grad_norm": 834.7509155273438,
      "learning_rate": 0.00019268628188164036,
      "loss": 7.2565,
      "step": 106350
    },
    {
      "epoch": 12.799037304452467,
      "grad_norm": 791.4818115234375,
      "learning_rate": 0.00019268485369089156,
      "loss": 7.2705,
      "step": 106360
    },
    {
      "epoch": 12.800240673886883,
      "grad_norm": 1112.0408935546875,
      "learning_rate": 0.00019268342536600492,
      "loss": 7.2604,
      "step": 106370
    },
    {
      "epoch": 12.8014440433213,
      "grad_norm": 687.0901489257812,
      "learning_rate": 0.00019268199690698244,
      "loss": 7.3188,
      "step": 106380
    },
    {
      "epoch": 12.802647412755716,
      "grad_norm": 517.177734375,
      "learning_rate": 0.00019268056831382626,
      "loss": 7.2657,
      "step": 106390
    },
    {
      "epoch": 12.803850782190132,
      "grad_norm": 1602.3511962890625,
      "learning_rate": 0.0001926791395865384,
      "loss": 7.1756,
      "step": 106400
    },
    {
      "epoch": 12.80505415162455,
      "grad_norm": 652.8690185546875,
      "learning_rate": 0.00019267771072512094,
      "loss": 7.3113,
      "step": 106410
    },
    {
      "epoch": 12.806257521058965,
      "grad_norm": 1281.319580078125,
      "learning_rate": 0.00019267628172957597,
      "loss": 7.2228,
      "step": 106420
    },
    {
      "epoch": 12.807460890493381,
      "grad_norm": 848.3667602539062,
      "learning_rate": 0.0001926748525999055,
      "loss": 7.3012,
      "step": 106430
    },
    {
      "epoch": 12.808664259927799,
      "grad_norm": 713.812255859375,
      "learning_rate": 0.00019267342333611167,
      "loss": 7.202,
      "step": 106440
    },
    {
      "epoch": 12.809867629362214,
      "grad_norm": 896.2886352539062,
      "learning_rate": 0.0001926719939381965,
      "loss": 7.27,
      "step": 106450
    },
    {
      "epoch": 12.81107099879663,
      "grad_norm": 929.7326049804688,
      "learning_rate": 0.0001926705644061621,
      "loss": 7.171,
      "step": 106460
    },
    {
      "epoch": 12.812274368231048,
      "grad_norm": 1868.1378173828125,
      "learning_rate": 0.00019266913474001048,
      "loss": 7.2998,
      "step": 106470
    },
    {
      "epoch": 12.813477737665464,
      "grad_norm": 1212.1624755859375,
      "learning_rate": 0.00019266770493974378,
      "loss": 7.122,
      "step": 106480
    },
    {
      "epoch": 12.81468110709988,
      "grad_norm": 1813.029541015625,
      "learning_rate": 0.000192666275005364,
      "loss": 7.2796,
      "step": 106490
    },
    {
      "epoch": 12.815884476534297,
      "grad_norm": 1303.044677734375,
      "learning_rate": 0.00019266484493687326,
      "loss": 7.2759,
      "step": 106500
    },
    {
      "epoch": 12.817087845968713,
      "grad_norm": 732.6553344726562,
      "learning_rate": 0.00019266341473427357,
      "loss": 7.2141,
      "step": 106510
    },
    {
      "epoch": 12.818291215403129,
      "grad_norm": 2084.2119140625,
      "learning_rate": 0.0001926619843975671,
      "loss": 7.4055,
      "step": 106520
    },
    {
      "epoch": 12.819494584837544,
      "grad_norm": 1265.0379638671875,
      "learning_rate": 0.00019266055392675583,
      "loss": 7.292,
      "step": 106530
    },
    {
      "epoch": 12.820697954271962,
      "grad_norm": 743.56884765625,
      "learning_rate": 0.00019265912332184186,
      "loss": 7.3046,
      "step": 106540
    },
    {
      "epoch": 12.821901323706378,
      "grad_norm": 1051.9405517578125,
      "learning_rate": 0.0001926576925828273,
      "loss": 7.2897,
      "step": 106550
    },
    {
      "epoch": 12.823104693140793,
      "grad_norm": 870.7982177734375,
      "learning_rate": 0.00019265626170971413,
      "loss": 7.2685,
      "step": 106560
    },
    {
      "epoch": 12.824308062575211,
      "grad_norm": 930.4187622070312,
      "learning_rate": 0.0001926548307025045,
      "loss": 7.2323,
      "step": 106570
    },
    {
      "epoch": 12.825511432009627,
      "grad_norm": 598.90869140625,
      "learning_rate": 0.00019265339956120046,
      "loss": 7.1915,
      "step": 106580
    },
    {
      "epoch": 12.826714801444043,
      "grad_norm": 783.8077392578125,
      "learning_rate": 0.00019265196828580406,
      "loss": 7.3256,
      "step": 106590
    },
    {
      "epoch": 12.82791817087846,
      "grad_norm": 1008.6307983398438,
      "learning_rate": 0.0001926505368763174,
      "loss": 7.3393,
      "step": 106600
    },
    {
      "epoch": 12.829121540312876,
      "grad_norm": 788.7634887695312,
      "learning_rate": 0.0001926491053327425,
      "loss": 7.2716,
      "step": 106610
    },
    {
      "epoch": 12.830324909747292,
      "grad_norm": 885.7045288085938,
      "learning_rate": 0.0001926476736550815,
      "loss": 7.252,
      "step": 106620
    },
    {
      "epoch": 12.83152827918171,
      "grad_norm": 793.7576904296875,
      "learning_rate": 0.00019264624184333647,
      "loss": 7.2465,
      "step": 106630
    },
    {
      "epoch": 12.832731648616125,
      "grad_norm": 1067.9049072265625,
      "learning_rate": 0.00019264480989750942,
      "loss": 7.2384,
      "step": 106640
    },
    {
      "epoch": 12.833935018050541,
      "grad_norm": 546.8511352539062,
      "learning_rate": 0.00019264337781760245,
      "loss": 7.187,
      "step": 106650
    },
    {
      "epoch": 12.835138387484958,
      "grad_norm": 746.3120727539062,
      "learning_rate": 0.00019264194560361767,
      "loss": 7.2499,
      "step": 106660
    },
    {
      "epoch": 12.836341756919374,
      "grad_norm": 852.8392944335938,
      "learning_rate": 0.00019264051325555709,
      "loss": 7.2268,
      "step": 106670
    },
    {
      "epoch": 12.83754512635379,
      "grad_norm": 671.7431640625,
      "learning_rate": 0.00019263908077342285,
      "loss": 7.3018,
      "step": 106680
    },
    {
      "epoch": 12.838748495788208,
      "grad_norm": 670.9478149414062,
      "learning_rate": 0.00019263764815721698,
      "loss": 7.1596,
      "step": 106690
    },
    {
      "epoch": 12.839951865222623,
      "grad_norm": 919.4144897460938,
      "learning_rate": 0.00019263621540694151,
      "loss": 7.2458,
      "step": 106700
    },
    {
      "epoch": 12.84115523465704,
      "grad_norm": 1038.4052734375,
      "learning_rate": 0.0001926347825225986,
      "loss": 7.19,
      "step": 106710
    },
    {
      "epoch": 12.842358604091457,
      "grad_norm": 798.1454467773438,
      "learning_rate": 0.00019263334950419032,
      "loss": 7.2504,
      "step": 106720
    },
    {
      "epoch": 12.843561973525873,
      "grad_norm": 849.4981079101562,
      "learning_rate": 0.0001926319163517187,
      "loss": 7.2519,
      "step": 106730
    },
    {
      "epoch": 12.844765342960288,
      "grad_norm": 380.7121887207031,
      "learning_rate": 0.0001926304830651858,
      "loss": 7.2666,
      "step": 106740
    },
    {
      "epoch": 12.845968712394706,
      "grad_norm": 838.4188842773438,
      "learning_rate": 0.0001926290496445937,
      "loss": 7.248,
      "step": 106750
    },
    {
      "epoch": 12.847172081829122,
      "grad_norm": 1152.8275146484375,
      "learning_rate": 0.00019262761608994456,
      "loss": 7.2972,
      "step": 106760
    },
    {
      "epoch": 12.848375451263538,
      "grad_norm": 1013.2494506835938,
      "learning_rate": 0.00019262618240124035,
      "loss": 7.271,
      "step": 106770
    },
    {
      "epoch": 12.849578820697955,
      "grad_norm": 621.736083984375,
      "learning_rate": 0.0001926247485784832,
      "loss": 7.2184,
      "step": 106780
    },
    {
      "epoch": 12.85078219013237,
      "grad_norm": 1140.8487548828125,
      "learning_rate": 0.00019262331462167518,
      "loss": 7.2565,
      "step": 106790
    },
    {
      "epoch": 12.851985559566787,
      "grad_norm": 1151.6893310546875,
      "learning_rate": 0.00019262188053081834,
      "loss": 7.2355,
      "step": 106800
    },
    {
      "epoch": 12.853188929001202,
      "grad_norm": 901.2284545898438,
      "learning_rate": 0.0001926204463059148,
      "loss": 7.2758,
      "step": 106810
    },
    {
      "epoch": 12.85439229843562,
      "grad_norm": 1705.5452880859375,
      "learning_rate": 0.00019261901194696655,
      "loss": 7.2953,
      "step": 106820
    },
    {
      "epoch": 12.855595667870036,
      "grad_norm": 1389.90869140625,
      "learning_rate": 0.00019261757745397574,
      "loss": 7.2572,
      "step": 106830
    },
    {
      "epoch": 12.856799037304452,
      "grad_norm": 848.8786010742188,
      "learning_rate": 0.00019261614282694444,
      "loss": 7.1925,
      "step": 106840
    },
    {
      "epoch": 12.85800240673887,
      "grad_norm": 1665.571044921875,
      "learning_rate": 0.00019261470806587473,
      "loss": 7.2532,
      "step": 106850
    },
    {
      "epoch": 12.859205776173285,
      "grad_norm": 1021.503662109375,
      "learning_rate": 0.00019261327317076867,
      "loss": 7.3014,
      "step": 106860
    },
    {
      "epoch": 12.8604091456077,
      "grad_norm": 1767.217529296875,
      "learning_rate": 0.00019261183814162834,
      "loss": 7.2869,
      "step": 106870
    },
    {
      "epoch": 12.861612515042118,
      "grad_norm": 1115.7940673828125,
      "learning_rate": 0.0001926104029784558,
      "loss": 7.387,
      "step": 106880
    },
    {
      "epoch": 12.862815884476534,
      "grad_norm": 815.3505249023438,
      "learning_rate": 0.00019260896768125314,
      "loss": 7.2325,
      "step": 106890
    },
    {
      "epoch": 12.86401925391095,
      "grad_norm": 3455.2578125,
      "learning_rate": 0.00019260753225002242,
      "loss": 7.3216,
      "step": 106900
    },
    {
      "epoch": 12.865222623345367,
      "grad_norm": 1894.0472412109375,
      "learning_rate": 0.00019260609668476578,
      "loss": 7.2956,
      "step": 106910
    },
    {
      "epoch": 12.866425992779783,
      "grad_norm": 2075.967529296875,
      "learning_rate": 0.00019260466098548523,
      "loss": 7.3369,
      "step": 106920
    },
    {
      "epoch": 12.867629362214199,
      "grad_norm": 1722.5640869140625,
      "learning_rate": 0.0001926032251521829,
      "loss": 7.2593,
      "step": 106930
    },
    {
      "epoch": 12.868832731648617,
      "grad_norm": 1012.143310546875,
      "learning_rate": 0.0001926017891848608,
      "loss": 7.2522,
      "step": 106940
    },
    {
      "epoch": 12.870036101083032,
      "grad_norm": 2170.95361328125,
      "learning_rate": 0.00019260035308352107,
      "loss": 7.2281,
      "step": 106950
    },
    {
      "epoch": 12.871239470517448,
      "grad_norm": 1574.2196044921875,
      "learning_rate": 0.00019259891684816576,
      "loss": 7.2903,
      "step": 106960
    },
    {
      "epoch": 12.872442839951866,
      "grad_norm": 4061.8955078125,
      "learning_rate": 0.00019259748047879697,
      "loss": 7.395,
      "step": 106970
    },
    {
      "epoch": 12.873646209386282,
      "grad_norm": 2118.121337890625,
      "learning_rate": 0.00019259604397541674,
      "loss": 7.3463,
      "step": 106980
    },
    {
      "epoch": 12.874849578820697,
      "grad_norm": 1463.2431640625,
      "learning_rate": 0.00019259460733802718,
      "loss": 7.4214,
      "step": 106990
    },
    {
      "epoch": 12.876052948255115,
      "grad_norm": 3004.537353515625,
      "learning_rate": 0.0001925931705666304,
      "loss": 7.1198,
      "step": 107000
    },
    {
      "epoch": 12.87725631768953,
      "grad_norm": 2270.33154296875,
      "learning_rate": 0.0001925917336612284,
      "loss": 7.2168,
      "step": 107010
    },
    {
      "epoch": 12.878459687123947,
      "grad_norm": 1603.310791015625,
      "learning_rate": 0.00019259029662182332,
      "loss": 7.3277,
      "step": 107020
    },
    {
      "epoch": 12.879663056558364,
      "grad_norm": 1873.32275390625,
      "learning_rate": 0.00019258885944841719,
      "loss": 7.2713,
      "step": 107030
    },
    {
      "epoch": 12.88086642599278,
      "grad_norm": 1999.0050048828125,
      "learning_rate": 0.00019258742214101215,
      "loss": 7.2877,
      "step": 107040
    },
    {
      "epoch": 12.882069795427196,
      "grad_norm": 2248.563232421875,
      "learning_rate": 0.00019258598469961024,
      "loss": 7.267,
      "step": 107050
    },
    {
      "epoch": 12.883273164861613,
      "grad_norm": 1957.7547607421875,
      "learning_rate": 0.00019258454712421354,
      "loss": 7.2537,
      "step": 107060
    },
    {
      "epoch": 12.884476534296029,
      "grad_norm": 2966.833740234375,
      "learning_rate": 0.00019258310941482417,
      "loss": 7.275,
      "step": 107070
    },
    {
      "epoch": 12.885679903730445,
      "grad_norm": 2127.40966796875,
      "learning_rate": 0.00019258167157144417,
      "loss": 7.1983,
      "step": 107080
    },
    {
      "epoch": 12.886883273164862,
      "grad_norm": 1940.19677734375,
      "learning_rate": 0.00019258023359407564,
      "loss": 7.3223,
      "step": 107090
    },
    {
      "epoch": 12.888086642599278,
      "grad_norm": 1240.020263671875,
      "learning_rate": 0.00019257879548272065,
      "loss": 7.2978,
      "step": 107100
    },
    {
      "epoch": 12.889290012033694,
      "grad_norm": 1843.514892578125,
      "learning_rate": 0.0001925773572373813,
      "loss": 7.2528,
      "step": 107110
    },
    {
      "epoch": 12.890493381468112,
      "grad_norm": 2873.846435546875,
      "learning_rate": 0.00019257591885805963,
      "loss": 7.2442,
      "step": 107120
    },
    {
      "epoch": 12.891696750902527,
      "grad_norm": 1514.1368408203125,
      "learning_rate": 0.00019257448034475775,
      "loss": 7.2172,
      "step": 107130
    },
    {
      "epoch": 12.892900120336943,
      "grad_norm": 1113.5157470703125,
      "learning_rate": 0.00019257304169747777,
      "loss": 7.2373,
      "step": 107140
    },
    {
      "epoch": 12.89410348977136,
      "grad_norm": 1617.1826171875,
      "learning_rate": 0.00019257160291622172,
      "loss": 7.1976,
      "step": 107150
    },
    {
      "epoch": 12.895306859205776,
      "grad_norm": 1258.438232421875,
      "learning_rate": 0.0001925701640009917,
      "loss": 7.1783,
      "step": 107160
    },
    {
      "epoch": 12.896510228640192,
      "grad_norm": 2233.916015625,
      "learning_rate": 0.00019256872495178982,
      "loss": 7.0803,
      "step": 107170
    },
    {
      "epoch": 12.89771359807461,
      "grad_norm": 1309.814697265625,
      "learning_rate": 0.00019256728576861812,
      "loss": 7.2011,
      "step": 107180
    },
    {
      "epoch": 12.898916967509026,
      "grad_norm": 922.4718627929688,
      "learning_rate": 0.00019256584645147872,
      "loss": 7.2048,
      "step": 107190
    },
    {
      "epoch": 12.900120336943441,
      "grad_norm": 1547.85302734375,
      "learning_rate": 0.0001925644070003737,
      "loss": 7.0259,
      "step": 107200
    },
    {
      "epoch": 12.901323706377857,
      "grad_norm": 1193.3726806640625,
      "learning_rate": 0.00019256296741530508,
      "loss": 7.2371,
      "step": 107210
    },
    {
      "epoch": 12.902527075812275,
      "grad_norm": 1601.9708251953125,
      "learning_rate": 0.00019256152769627502,
      "loss": 7.2353,
      "step": 107220
    },
    {
      "epoch": 12.90373044524669,
      "grad_norm": 2380.31591796875,
      "learning_rate": 0.00019256008784328558,
      "loss": 7.1135,
      "step": 107230
    },
    {
      "epoch": 12.904933814681106,
      "grad_norm": 1205.4659423828125,
      "learning_rate": 0.00019255864785633887,
      "loss": 7.105,
      "step": 107240
    },
    {
      "epoch": 12.906137184115524,
      "grad_norm": 4662.0458984375,
      "learning_rate": 0.0001925572077354369,
      "loss": 7.2564,
      "step": 107250
    },
    {
      "epoch": 12.90734055354994,
      "grad_norm": 2169.8310546875,
      "learning_rate": 0.0001925557674805818,
      "loss": 7.2418,
      "step": 107260
    },
    {
      "epoch": 12.908543922984355,
      "grad_norm": 8630.0146484375,
      "learning_rate": 0.00019255432709177565,
      "loss": 7.1595,
      "step": 107270
    },
    {
      "epoch": 12.909747292418773,
      "grad_norm": 3923.92919921875,
      "learning_rate": 0.00019255288656902059,
      "loss": 7.2695,
      "step": 107280
    },
    {
      "epoch": 12.910950661853189,
      "grad_norm": 2810.24267578125,
      "learning_rate": 0.0001925514459123186,
      "loss": 7.2671,
      "step": 107290
    },
    {
      "epoch": 12.912154031287605,
      "grad_norm": 5679.7978515625,
      "learning_rate": 0.00019255000512167182,
      "loss": 7.2389,
      "step": 107300
    },
    {
      "epoch": 12.913357400722022,
      "grad_norm": 2105.82763671875,
      "learning_rate": 0.00019254856419708235,
      "loss": 7.2707,
      "step": 107310
    },
    {
      "epoch": 12.914560770156438,
      "grad_norm": 2365.353515625,
      "learning_rate": 0.00019254712313855225,
      "loss": 7.218,
      "step": 107320
    },
    {
      "epoch": 12.915764139590854,
      "grad_norm": 1215.5013427734375,
      "learning_rate": 0.00019254568194608364,
      "loss": 7.2295,
      "step": 107330
    },
    {
      "epoch": 12.916967509025271,
      "grad_norm": 999.1481323242188,
      "learning_rate": 0.00019254424061967854,
      "loss": 7.2461,
      "step": 107340
    },
    {
      "epoch": 12.918170878459687,
      "grad_norm": 1257.3634033203125,
      "learning_rate": 0.0001925427991593391,
      "loss": 7.3096,
      "step": 107350
    },
    {
      "epoch": 12.919374247894103,
      "grad_norm": 857.899658203125,
      "learning_rate": 0.00019254135756506738,
      "loss": 7.3217,
      "step": 107360
    },
    {
      "epoch": 12.92057761732852,
      "grad_norm": 3409.077392578125,
      "learning_rate": 0.00019253991583686545,
      "loss": 7.3375,
      "step": 107370
    },
    {
      "epoch": 12.921780986762936,
      "grad_norm": 2719.557861328125,
      "learning_rate": 0.00019253847397473543,
      "loss": 7.333,
      "step": 107380
    },
    {
      "epoch": 12.922984356197352,
      "grad_norm": 3630.5634765625,
      "learning_rate": 0.0001925370319786794,
      "loss": 7.2898,
      "step": 107390
    },
    {
      "epoch": 12.92418772563177,
      "grad_norm": 4198.22705078125,
      "learning_rate": 0.0001925355898486994,
      "loss": 7.1538,
      "step": 107400
    },
    {
      "epoch": 12.925391095066185,
      "grad_norm": 3447.393310546875,
      "learning_rate": 0.00019253414758479758,
      "loss": 7.2984,
      "step": 107410
    },
    {
      "epoch": 12.926594464500601,
      "grad_norm": 2648.252685546875,
      "learning_rate": 0.000192532705186976,
      "loss": 7.2454,
      "step": 107420
    },
    {
      "epoch": 12.927797833935019,
      "grad_norm": 5186.08154296875,
      "learning_rate": 0.00019253126265523677,
      "loss": 7.3981,
      "step": 107430
    },
    {
      "epoch": 12.929001203369435,
      "grad_norm": 11985.9013671875,
      "learning_rate": 0.00019252981998958192,
      "loss": 7.4484,
      "step": 107440
    },
    {
      "epoch": 12.93020457280385,
      "grad_norm": 11013.3349609375,
      "learning_rate": 0.00019252837719001362,
      "loss": 7.3707,
      "step": 107450
    },
    {
      "epoch": 12.931407942238268,
      "grad_norm": 5295.51708984375,
      "learning_rate": 0.0001925269342565339,
      "loss": 7.3675,
      "step": 107460
    },
    {
      "epoch": 12.932611311672684,
      "grad_norm": 1148.791259765625,
      "learning_rate": 0.00019252549118914484,
      "loss": 7.3161,
      "step": 107470
    },
    {
      "epoch": 12.9338146811071,
      "grad_norm": 4863.2080078125,
      "learning_rate": 0.00019252404798784855,
      "loss": 7.2384,
      "step": 107480
    },
    {
      "epoch": 12.935018050541515,
      "grad_norm": 1197.8677978515625,
      "learning_rate": 0.00019252260465264716,
      "loss": 7.2969,
      "step": 107490
    },
    {
      "epoch": 12.936221419975933,
      "grad_norm": 3278.40185546875,
      "learning_rate": 0.00019252116118354266,
      "loss": 7.2178,
      "step": 107500
    },
    {
      "epoch": 12.937424789410349,
      "grad_norm": 1501.00390625,
      "learning_rate": 0.00019251971758053724,
      "loss": 7.232,
      "step": 107510
    },
    {
      "epoch": 12.938628158844764,
      "grad_norm": 2276.19580078125,
      "learning_rate": 0.00019251827384363294,
      "loss": 7.2032,
      "step": 107520
    },
    {
      "epoch": 12.939831528279182,
      "grad_norm": 1260.967041015625,
      "learning_rate": 0.00019251682997283186,
      "loss": 7.1716,
      "step": 107530
    },
    {
      "epoch": 12.941034897713598,
      "grad_norm": 2488.078125,
      "learning_rate": 0.00019251538596813604,
      "loss": 7.3369,
      "step": 107540
    },
    {
      "epoch": 12.942238267148014,
      "grad_norm": 1627.5501708984375,
      "learning_rate": 0.00019251394182954768,
      "loss": 7.1778,
      "step": 107550
    },
    {
      "epoch": 12.943441636582431,
      "grad_norm": 1441.1497802734375,
      "learning_rate": 0.00019251249755706878,
      "loss": 7.2897,
      "step": 107560
    },
    {
      "epoch": 12.944645006016847,
      "grad_norm": 2034.059326171875,
      "learning_rate": 0.00019251105315070142,
      "loss": 7.3516,
      "step": 107570
    },
    {
      "epoch": 12.945848375451263,
      "grad_norm": 2380.8642578125,
      "learning_rate": 0.0001925096086104478,
      "loss": 7.2541,
      "step": 107580
    },
    {
      "epoch": 12.94705174488568,
      "grad_norm": 1377.181884765625,
      "learning_rate": 0.00019250816393630985,
      "loss": 7.2594,
      "step": 107590
    },
    {
      "epoch": 12.948255114320096,
      "grad_norm": 2636.31005859375,
      "learning_rate": 0.0001925067191282898,
      "loss": 7.2899,
      "step": 107600
    },
    {
      "epoch": 12.949458483754512,
      "grad_norm": 904.372314453125,
      "learning_rate": 0.00019250527418638967,
      "loss": 7.2828,
      "step": 107610
    },
    {
      "epoch": 12.95066185318893,
      "grad_norm": 2921.962646484375,
      "learning_rate": 0.00019250382911061158,
      "loss": 7.3664,
      "step": 107620
    },
    {
      "epoch": 12.951865222623345,
      "grad_norm": 6842.02099609375,
      "learning_rate": 0.0001925023839009576,
      "loss": 7.3008,
      "step": 107630
    },
    {
      "epoch": 12.953068592057761,
      "grad_norm": 2014.3673095703125,
      "learning_rate": 0.00019250093855742984,
      "loss": 7.3621,
      "step": 107640
    },
    {
      "epoch": 12.954271961492179,
      "grad_norm": 1539.3853759765625,
      "learning_rate": 0.0001924994930800304,
      "loss": 7.2573,
      "step": 107650
    },
    {
      "epoch": 12.955475330926594,
      "grad_norm": 1351.1629638671875,
      "learning_rate": 0.00019249804746876136,
      "loss": 7.3223,
      "step": 107660
    },
    {
      "epoch": 12.95667870036101,
      "grad_norm": 2710.90478515625,
      "learning_rate": 0.00019249660172362477,
      "loss": 7.3821,
      "step": 107670
    },
    {
      "epoch": 12.957882069795428,
      "grad_norm": 1781.0064697265625,
      "learning_rate": 0.00019249515584462276,
      "loss": 7.285,
      "step": 107680
    },
    {
      "epoch": 12.959085439229844,
      "grad_norm": 1480.45654296875,
      "learning_rate": 0.00019249370983175746,
      "loss": 7.3548,
      "step": 107690
    },
    {
      "epoch": 12.96028880866426,
      "grad_norm": 3168.09130859375,
      "learning_rate": 0.0001924922636850309,
      "loss": 7.2683,
      "step": 107700
    },
    {
      "epoch": 12.961492178098677,
      "grad_norm": 1672.503173828125,
      "learning_rate": 0.0001924908174044452,
      "loss": 7.2961,
      "step": 107710
    },
    {
      "epoch": 12.962695547533093,
      "grad_norm": 3248.193115234375,
      "learning_rate": 0.00019248937099000245,
      "loss": 7.4289,
      "step": 107720
    },
    {
      "epoch": 12.963898916967509,
      "grad_norm": 2357.509521484375,
      "learning_rate": 0.00019248792444170477,
      "loss": 7.3711,
      "step": 107730
    },
    {
      "epoch": 12.965102286401926,
      "grad_norm": 4426.1787109375,
      "learning_rate": 0.0001924864777595542,
      "loss": 7.4848,
      "step": 107740
    },
    {
      "epoch": 12.966305655836342,
      "grad_norm": 3416.599365234375,
      "learning_rate": 0.0001924850309435529,
      "loss": 7.3628,
      "step": 107750
    },
    {
      "epoch": 12.967509025270758,
      "grad_norm": 2189.18115234375,
      "learning_rate": 0.0001924835839937029,
      "loss": 7.3817,
      "step": 107760
    },
    {
      "epoch": 12.968712394705175,
      "grad_norm": 3998.75048828125,
      "learning_rate": 0.0001924821369100063,
      "loss": 7.4406,
      "step": 107770
    },
    {
      "epoch": 12.969915764139591,
      "grad_norm": 4420.34423828125,
      "learning_rate": 0.00019248068969246528,
      "loss": 7.4438,
      "step": 107780
    },
    {
      "epoch": 12.971119133574007,
      "grad_norm": 15506.3779296875,
      "learning_rate": 0.0001924792423410818,
      "loss": 7.434,
      "step": 107790
    },
    {
      "epoch": 12.972322503008424,
      "grad_norm": 14481.087890625,
      "learning_rate": 0.00019247779485585806,
      "loss": 7.4374,
      "step": 107800
    },
    {
      "epoch": 12.97352587244284,
      "grad_norm": 18825.759765625,
      "learning_rate": 0.0001924763472367961,
      "loss": 7.6036,
      "step": 107810
    },
    {
      "epoch": 12.974729241877256,
      "grad_norm": 23447.19140625,
      "learning_rate": 0.00019247489948389805,
      "loss": 7.6959,
      "step": 107820
    },
    {
      "epoch": 12.975932611311674,
      "grad_norm": 8680.255859375,
      "learning_rate": 0.000192473451597166,
      "loss": 7.7892,
      "step": 107830
    },
    {
      "epoch": 12.97713598074609,
      "grad_norm": 8544.8154296875,
      "learning_rate": 0.000192472003576602,
      "loss": 7.6692,
      "step": 107840
    },
    {
      "epoch": 12.978339350180505,
      "grad_norm": 10029.8583984375,
      "learning_rate": 0.0001924705554222082,
      "loss": 7.4473,
      "step": 107850
    },
    {
      "epoch": 12.979542719614923,
      "grad_norm": 5717.8291015625,
      "learning_rate": 0.00019246910713398668,
      "loss": 7.4975,
      "step": 107860
    },
    {
      "epoch": 12.980746089049338,
      "grad_norm": 59259.328125,
      "learning_rate": 0.00019246765871193953,
      "loss": 7.5435,
      "step": 107870
    },
    {
      "epoch": 12.981949458483754,
      "grad_norm": 11020.2177734375,
      "learning_rate": 0.00019246621015606884,
      "loss": 7.362,
      "step": 107880
    },
    {
      "epoch": 12.98315282791817,
      "grad_norm": 8665.4013671875,
      "learning_rate": 0.00019246476146637673,
      "loss": 7.4455,
      "step": 107890
    },
    {
      "epoch": 12.984356197352588,
      "grad_norm": 26855.755859375,
      "learning_rate": 0.00019246331264286527,
      "loss": 7.4505,
      "step": 107900
    },
    {
      "epoch": 12.985559566787003,
      "grad_norm": 28965.345703125,
      "learning_rate": 0.00019246186368553655,
      "loss": 7.3503,
      "step": 107910
    },
    {
      "epoch": 12.98676293622142,
      "grad_norm": 15996.9560546875,
      "learning_rate": 0.00019246041459439274,
      "loss": 7.4335,
      "step": 107920
    },
    {
      "epoch": 12.987966305655837,
      "grad_norm": 31387.595703125,
      "learning_rate": 0.00019245896536943582,
      "loss": 7.3639,
      "step": 107930
    },
    {
      "epoch": 12.989169675090253,
      "grad_norm": 11455.1064453125,
      "learning_rate": 0.00019245751601066797,
      "loss": 7.5035,
      "step": 107940
    },
    {
      "epoch": 12.990373044524668,
      "grad_norm": 20698.126953125,
      "learning_rate": 0.00019245606651809128,
      "loss": 7.6349,
      "step": 107950
    },
    {
      "epoch": 12.991576413959086,
      "grad_norm": 4148.4462890625,
      "learning_rate": 0.0001924546168917078,
      "loss": 7.5947,
      "step": 107960
    },
    {
      "epoch": 12.992779783393502,
      "grad_norm": 5056.8125,
      "learning_rate": 0.00019245316713151973,
      "loss": 7.6983,
      "step": 107970
    },
    {
      "epoch": 12.993983152827917,
      "grad_norm": 3186.49609375,
      "learning_rate": 0.00019245171723752903,
      "loss": 7.6045,
      "step": 107980
    },
    {
      "epoch": 12.995186522262335,
      "grad_norm": 1980.4691162109375,
      "learning_rate": 0.0001924502672097379,
      "loss": 7.5403,
      "step": 107990
    },
    {
      "epoch": 12.99638989169675,
      "grad_norm": 4110.58984375,
      "learning_rate": 0.00019244881704814844,
      "loss": 7.5637,
      "step": 108000
    },
    {
      "epoch": 12.997593261131167,
      "grad_norm": 21176.83203125,
      "learning_rate": 0.0001924473667527627,
      "loss": 7.5598,
      "step": 108010
    },
    {
      "epoch": 12.998796630565584,
      "grad_norm": 39221.96875,
      "learning_rate": 0.00019244591632358276,
      "loss": 7.5556,
      "step": 108020
    },
    {
      "epoch": 13.0,
      "grad_norm": 8619.4814453125,
      "learning_rate": 0.00019244446576061077,
      "loss": 7.5775,
      "step": 108030
    },
    {
      "epoch": 13.0,
      "eval_loss": 7.555514335632324,
      "eval_runtime": 120.208,
      "eval_samples_per_second": 61.452,
      "eval_steps_per_second": 7.687,
      "step": 108030
    },
    {
      "epoch": 13.001203369434416,
      "grad_norm": 15074.41015625,
      "learning_rate": 0.00019244301506384884,
      "loss": 7.5985,
      "step": 108040
    },
    {
      "epoch": 13.002406738868833,
      "grad_norm": 12033.8212890625,
      "learning_rate": 0.000192441564233299,
      "loss": 7.4701,
      "step": 108050
    },
    {
      "epoch": 13.00361010830325,
      "grad_norm": 15881.751953125,
      "learning_rate": 0.00019244011326896343,
      "loss": 7.5534,
      "step": 108060
    },
    {
      "epoch": 13.004813477737665,
      "grad_norm": 25642.037109375,
      "learning_rate": 0.0001924386621708442,
      "loss": 7.6858,
      "step": 108070
    },
    {
      "epoch": 13.006016847172083,
      "grad_norm": 5778.74951171875,
      "learning_rate": 0.00019243721093894337,
      "loss": 7.5327,
      "step": 108080
    },
    {
      "epoch": 13.007220216606498,
      "grad_norm": 12361.5380859375,
      "learning_rate": 0.0001924357595732631,
      "loss": 7.5847,
      "step": 108090
    },
    {
      "epoch": 13.008423586040914,
      "grad_norm": 7474.11962890625,
      "learning_rate": 0.00019243430807380546,
      "loss": 7.5545,
      "step": 108100
    },
    {
      "epoch": 13.009626955475332,
      "grad_norm": 35978.73046875,
      "learning_rate": 0.00019243285644057254,
      "loss": 7.5959,
      "step": 108110
    },
    {
      "epoch": 13.010830324909747,
      "grad_norm": 4408.5224609375,
      "learning_rate": 0.00019243140467356645,
      "loss": 7.5043,
      "step": 108120
    },
    {
      "epoch": 13.012033694344163,
      "grad_norm": 10178.572265625,
      "learning_rate": 0.00019242995277278933,
      "loss": 7.5143,
      "step": 108130
    },
    {
      "epoch": 13.01323706377858,
      "grad_norm": 12298.19921875,
      "learning_rate": 0.00019242850073824324,
      "loss": 7.4687,
      "step": 108140
    },
    {
      "epoch": 13.014440433212997,
      "grad_norm": 4582.2744140625,
      "learning_rate": 0.00019242704856993026,
      "loss": 7.5413,
      "step": 108150
    },
    {
      "epoch": 13.015643802647412,
      "grad_norm": 21745.123046875,
      "learning_rate": 0.00019242559626785256,
      "loss": 7.4886,
      "step": 108160
    },
    {
      "epoch": 13.01684717208183,
      "grad_norm": 13244.1435546875,
      "learning_rate": 0.00019242414383201215,
      "loss": 7.597,
      "step": 108170
    },
    {
      "epoch": 13.018050541516246,
      "grad_norm": 3957.949462890625,
      "learning_rate": 0.00019242269126241123,
      "loss": 7.6059,
      "step": 108180
    },
    {
      "epoch": 13.019253910950662,
      "grad_norm": 3317.078369140625,
      "learning_rate": 0.00019242123855905185,
      "loss": 7.4801,
      "step": 108190
    },
    {
      "epoch": 13.020457280385079,
      "grad_norm": 2736.24072265625,
      "learning_rate": 0.00019241978572193612,
      "loss": 7.5018,
      "step": 108200
    },
    {
      "epoch": 13.021660649819495,
      "grad_norm": 2209.564208984375,
      "learning_rate": 0.00019241833275106612,
      "loss": 7.5323,
      "step": 108210
    },
    {
      "epoch": 13.02286401925391,
      "grad_norm": 3912.98974609375,
      "learning_rate": 0.000192416879646444,
      "loss": 7.4479,
      "step": 108220
    },
    {
      "epoch": 13.024067388688326,
      "grad_norm": 9202.1728515625,
      "learning_rate": 0.00019241542640807184,
      "loss": 7.5079,
      "step": 108230
    },
    {
      "epoch": 13.025270758122744,
      "grad_norm": 4686.619140625,
      "learning_rate": 0.00019241397303595173,
      "loss": 7.547,
      "step": 108240
    },
    {
      "epoch": 13.02647412755716,
      "grad_norm": 7604.71435546875,
      "learning_rate": 0.0001924125195300858,
      "loss": 7.5015,
      "step": 108250
    },
    {
      "epoch": 13.027677496991576,
      "grad_norm": 10323.5732421875,
      "learning_rate": 0.00019241106589047612,
      "loss": 7.4948,
      "step": 108260
    },
    {
      "epoch": 13.028880866425993,
      "grad_norm": 5556.9228515625,
      "learning_rate": 0.0001924096121171248,
      "loss": 7.6201,
      "step": 108270
    },
    {
      "epoch": 13.030084235860409,
      "grad_norm": 1971.686767578125,
      "learning_rate": 0.00019240815821003397,
      "loss": 7.6097,
      "step": 108280
    },
    {
      "epoch": 13.031287605294825,
      "grad_norm": 4989.94921875,
      "learning_rate": 0.00019240670416920575,
      "loss": 7.4809,
      "step": 108290
    },
    {
      "epoch": 13.032490974729242,
      "grad_norm": 6852.31494140625,
      "learning_rate": 0.0001924052499946422,
      "loss": 7.6369,
      "step": 108300
    },
    {
      "epoch": 13.033694344163658,
      "grad_norm": 21070.107421875,
      "learning_rate": 0.0001924037956863454,
      "loss": 7.4799,
      "step": 108310
    },
    {
      "epoch": 13.034897713598074,
      "grad_norm": 4888.40478515625,
      "learning_rate": 0.00019240234124431752,
      "loss": 7.4902,
      "step": 108320
    },
    {
      "epoch": 13.036101083032491,
      "grad_norm": 5162.17578125,
      "learning_rate": 0.00019240088666856066,
      "loss": 7.5887,
      "step": 108330
    },
    {
      "epoch": 13.037304452466907,
      "grad_norm": 2335.8583984375,
      "learning_rate": 0.00019239943195907685,
      "loss": 7.4685,
      "step": 108340
    },
    {
      "epoch": 13.038507821901323,
      "grad_norm": 2831.58935546875,
      "learning_rate": 0.0001923979771158683,
      "loss": 7.4267,
      "step": 108350
    },
    {
      "epoch": 13.03971119133574,
      "grad_norm": 4768.30517578125,
      "learning_rate": 0.00019239652213893705,
      "loss": 7.4714,
      "step": 108360
    },
    {
      "epoch": 13.040914560770156,
      "grad_norm": 2200.756103515625,
      "learning_rate": 0.00019239506702828523,
      "loss": 7.451,
      "step": 108370
    },
    {
      "epoch": 13.042117930204572,
      "grad_norm": 4867.78369140625,
      "learning_rate": 0.0001923936117839149,
      "loss": 7.4994,
      "step": 108380
    },
    {
      "epoch": 13.04332129963899,
      "grad_norm": 7237.81396484375,
      "learning_rate": 0.00019239215640582823,
      "loss": 7.4377,
      "step": 108390
    },
    {
      "epoch": 13.044524669073406,
      "grad_norm": 2619.4345703125,
      "learning_rate": 0.00019239070089402733,
      "loss": 7.521,
      "step": 108400
    },
    {
      "epoch": 13.045728038507821,
      "grad_norm": 7368.5830078125,
      "learning_rate": 0.00019238924524851424,
      "loss": 7.537,
      "step": 108410
    },
    {
      "epoch": 13.046931407942239,
      "grad_norm": 4348.572265625,
      "learning_rate": 0.0001923877894692911,
      "loss": 7.4566,
      "step": 108420
    },
    {
      "epoch": 13.048134777376655,
      "grad_norm": 5014.33154296875,
      "learning_rate": 0.00019238633355636,
      "loss": 7.5919,
      "step": 108430
    },
    {
      "epoch": 13.04933814681107,
      "grad_norm": 4388.39697265625,
      "learning_rate": 0.0001923848775097231,
      "loss": 7.5423,
      "step": 108440
    },
    {
      "epoch": 13.050541516245488,
      "grad_norm": 8234.9833984375,
      "learning_rate": 0.00019238342132938248,
      "loss": 7.6377,
      "step": 108450
    },
    {
      "epoch": 13.051744885679904,
      "grad_norm": 14437.505859375,
      "learning_rate": 0.0001923819650153402,
      "loss": 7.5116,
      "step": 108460
    },
    {
      "epoch": 13.05294825511432,
      "grad_norm": 10338.55078125,
      "learning_rate": 0.00019238050856759843,
      "loss": 7.5523,
      "step": 108470
    },
    {
      "epoch": 13.054151624548737,
      "grad_norm": 29504.134765625,
      "learning_rate": 0.00019237905198615926,
      "loss": 7.5164,
      "step": 108480
    },
    {
      "epoch": 13.055354993983153,
      "grad_norm": 10035.2490234375,
      "learning_rate": 0.0001923775952710248,
      "loss": 7.5909,
      "step": 108490
    },
    {
      "epoch": 13.056558363417569,
      "grad_norm": 6084.50146484375,
      "learning_rate": 0.00019237613842219714,
      "loss": 7.5066,
      "step": 108500
    },
    {
      "epoch": 13.057761732851986,
      "grad_norm": 8544.9130859375,
      "learning_rate": 0.0001923746814396784,
      "loss": 7.3935,
      "step": 108510
    },
    {
      "epoch": 13.058965102286402,
      "grad_norm": 5593.34765625,
      "learning_rate": 0.00019237322432347068,
      "loss": 7.4424,
      "step": 108520
    },
    {
      "epoch": 13.060168471720818,
      "grad_norm": 12463.994140625,
      "learning_rate": 0.00019237176707357612,
      "loss": 7.5306,
      "step": 108530
    },
    {
      "epoch": 13.061371841155236,
      "grad_norm": 7964.5947265625,
      "learning_rate": 0.00019237030968999682,
      "loss": 7.4792,
      "step": 108540
    },
    {
      "epoch": 13.062575210589651,
      "grad_norm": 8108.3310546875,
      "learning_rate": 0.00019236885217273483,
      "loss": 7.4017,
      "step": 108550
    },
    {
      "epoch": 13.063778580024067,
      "grad_norm": 11556.1376953125,
      "learning_rate": 0.00019236739452179236,
      "loss": 7.443,
      "step": 108560
    },
    {
      "epoch": 13.064981949458483,
      "grad_norm": 10719.9599609375,
      "learning_rate": 0.00019236593673717142,
      "loss": 7.4535,
      "step": 108570
    },
    {
      "epoch": 13.0661853188929,
      "grad_norm": 8270.806640625,
      "learning_rate": 0.0001923644788188742,
      "loss": 7.4508,
      "step": 108580
    },
    {
      "epoch": 13.067388688327316,
      "grad_norm": 9861.3232421875,
      "learning_rate": 0.00019236302076690277,
      "loss": 7.487,
      "step": 108590
    },
    {
      "epoch": 13.068592057761732,
      "grad_norm": 16110.0537109375,
      "learning_rate": 0.0001923615625812592,
      "loss": 7.4765,
      "step": 108600
    },
    {
      "epoch": 13.06979542719615,
      "grad_norm": 25546.724609375,
      "learning_rate": 0.00019236010426194568,
      "loss": 7.3592,
      "step": 108610
    },
    {
      "epoch": 13.070998796630565,
      "grad_norm": 13784.078125,
      "learning_rate": 0.0001923586458089643,
      "loss": 7.4452,
      "step": 108620
    },
    {
      "epoch": 13.072202166064981,
      "grad_norm": 5039.65673828125,
      "learning_rate": 0.00019235718722231712,
      "loss": 7.5369,
      "step": 108630
    },
    {
      "epoch": 13.073405535499399,
      "grad_norm": 17726.580078125,
      "learning_rate": 0.00019235572850200632,
      "loss": 7.5048,
      "step": 108640
    },
    {
      "epoch": 13.074608904933815,
      "grad_norm": 24905.537109375,
      "learning_rate": 0.00019235426964803396,
      "loss": 7.4813,
      "step": 108650
    },
    {
      "epoch": 13.07581227436823,
      "grad_norm": 20696.16015625,
      "learning_rate": 0.0001923528106604022,
      "loss": 7.4328,
      "step": 108660
    },
    {
      "epoch": 13.077015643802648,
      "grad_norm": 14115.568359375,
      "learning_rate": 0.0001923513515391131,
      "loss": 7.4074,
      "step": 108670
    },
    {
      "epoch": 13.078219013237064,
      "grad_norm": 37487.0390625,
      "learning_rate": 0.00019234989228416878,
      "loss": 7.4354,
      "step": 108680
    },
    {
      "epoch": 13.07942238267148,
      "grad_norm": 7489.62646484375,
      "learning_rate": 0.00019234843289557137,
      "loss": 7.3867,
      "step": 108690
    },
    {
      "epoch": 13.080625752105897,
      "grad_norm": 4249.4619140625,
      "learning_rate": 0.000192346973373323,
      "loss": 7.4146,
      "step": 108700
    },
    {
      "epoch": 13.081829121540313,
      "grad_norm": 10376.52734375,
      "learning_rate": 0.00019234551371742574,
      "loss": 7.4164,
      "step": 108710
    },
    {
      "epoch": 13.083032490974729,
      "grad_norm": 22127.861328125,
      "learning_rate": 0.00019234405392788172,
      "loss": 7.3075,
      "step": 108720
    },
    {
      "epoch": 13.084235860409146,
      "grad_norm": 2832.487060546875,
      "learning_rate": 0.00019234259400469306,
      "loss": 7.3684,
      "step": 108730
    },
    {
      "epoch": 13.085439229843562,
      "grad_norm": 7236.1982421875,
      "learning_rate": 0.0001923411339478619,
      "loss": 7.3607,
      "step": 108740
    },
    {
      "epoch": 13.086642599277978,
      "grad_norm": 1548.1239013671875,
      "learning_rate": 0.00019233967375739027,
      "loss": 7.2633,
      "step": 108750
    },
    {
      "epoch": 13.087845968712395,
      "grad_norm": 1546.273193359375,
      "learning_rate": 0.00019233821343328037,
      "loss": 7.4365,
      "step": 108760
    },
    {
      "epoch": 13.089049338146811,
      "grad_norm": 2337.670654296875,
      "learning_rate": 0.00019233675297553426,
      "loss": 7.4905,
      "step": 108770
    },
    {
      "epoch": 13.090252707581227,
      "grad_norm": 1819.724609375,
      "learning_rate": 0.00019233529238415407,
      "loss": 7.5188,
      "step": 108780
    },
    {
      "epoch": 13.091456077015645,
      "grad_norm": 1907.8560791015625,
      "learning_rate": 0.0001923338316591419,
      "loss": 7.3461,
      "step": 108790
    },
    {
      "epoch": 13.09265944645006,
      "grad_norm": 1162.979736328125,
      "learning_rate": 0.0001923323708004999,
      "loss": 7.38,
      "step": 108800
    },
    {
      "epoch": 13.093862815884476,
      "grad_norm": 1256.3233642578125,
      "learning_rate": 0.00019233090980823016,
      "loss": 7.4264,
      "step": 108810
    },
    {
      "epoch": 13.095066185318894,
      "grad_norm": 1566.36962890625,
      "learning_rate": 0.0001923294486823348,
      "loss": 7.4459,
      "step": 108820
    },
    {
      "epoch": 13.09626955475331,
      "grad_norm": 1580.012451171875,
      "learning_rate": 0.00019232798742281596,
      "loss": 7.4325,
      "step": 108830
    },
    {
      "epoch": 13.097472924187725,
      "grad_norm": 1637.6971435546875,
      "learning_rate": 0.0001923265260296757,
      "loss": 7.618,
      "step": 108840
    },
    {
      "epoch": 13.098676293622143,
      "grad_norm": 1655.1015625,
      "learning_rate": 0.00019232506450291616,
      "loss": 7.5967,
      "step": 108850
    },
    {
      "epoch": 13.099879663056559,
      "grad_norm": 1030.7801513671875,
      "learning_rate": 0.00019232360284253942,
      "loss": 7.5514,
      "step": 108860
    },
    {
      "epoch": 13.101083032490974,
      "grad_norm": 2341.02392578125,
      "learning_rate": 0.0001923221410485477,
      "loss": 7.4855,
      "step": 108870
    },
    {
      "epoch": 13.102286401925392,
      "grad_norm": 2831.236572265625,
      "learning_rate": 0.000192320679120943,
      "loss": 7.5784,
      "step": 108880
    },
    {
      "epoch": 13.103489771359808,
      "grad_norm": 1810.7940673828125,
      "learning_rate": 0.00019231921705972752,
      "loss": 7.5255,
      "step": 108890
    },
    {
      "epoch": 13.104693140794224,
      "grad_norm": 1143.26904296875,
      "learning_rate": 0.00019231775486490334,
      "loss": 7.5552,
      "step": 108900
    },
    {
      "epoch": 13.10589651022864,
      "grad_norm": 2173.7587890625,
      "learning_rate": 0.00019231629253647255,
      "loss": 7.59,
      "step": 108910
    },
    {
      "epoch": 13.107099879663057,
      "grad_norm": 3471.625244140625,
      "learning_rate": 0.0001923148300744373,
      "loss": 7.5861,
      "step": 108920
    },
    {
      "epoch": 13.108303249097473,
      "grad_norm": 1325.9033203125,
      "learning_rate": 0.00019231336747879972,
      "loss": 7.5344,
      "step": 108930
    },
    {
      "epoch": 13.109506618531888,
      "grad_norm": 1272.190673828125,
      "learning_rate": 0.00019231190474956186,
      "loss": 7.5464,
      "step": 108940
    },
    {
      "epoch": 13.110709987966306,
      "grad_norm": 1664.0906982421875,
      "learning_rate": 0.00019231044188672592,
      "loss": 7.5406,
      "step": 108950
    },
    {
      "epoch": 13.111913357400722,
      "grad_norm": 2486.56689453125,
      "learning_rate": 0.00019230897889029397,
      "loss": 7.642,
      "step": 108960
    },
    {
      "epoch": 13.113116726835138,
      "grad_norm": 854.6378784179688,
      "learning_rate": 0.00019230751576026817,
      "loss": 7.5285,
      "step": 108970
    },
    {
      "epoch": 13.114320096269555,
      "grad_norm": 1075.2103271484375,
      "learning_rate": 0.00019230605249665055,
      "loss": 7.6098,
      "step": 108980
    },
    {
      "epoch": 13.115523465703971,
      "grad_norm": 1015.71630859375,
      "learning_rate": 0.00019230458909944332,
      "loss": 7.4277,
      "step": 108990
    },
    {
      "epoch": 13.116726835138387,
      "grad_norm": 1571.4444580078125,
      "learning_rate": 0.00019230312556864853,
      "loss": 7.473,
      "step": 109000
    },
    {
      "epoch": 13.117930204572804,
      "grad_norm": 1367.8033447265625,
      "learning_rate": 0.00019230166190426835,
      "loss": 7.5912,
      "step": 109010
    },
    {
      "epoch": 13.11913357400722,
      "grad_norm": 520.5321655273438,
      "learning_rate": 0.00019230019810630484,
      "loss": 7.5585,
      "step": 109020
    },
    {
      "epoch": 13.120336943441636,
      "grad_norm": 888.4727172851562,
      "learning_rate": 0.0001922987341747602,
      "loss": 7.5432,
      "step": 109030
    },
    {
      "epoch": 13.121540312876053,
      "grad_norm": 1325.9951171875,
      "learning_rate": 0.00019229727010963647,
      "loss": 7.583,
      "step": 109040
    },
    {
      "epoch": 13.12274368231047,
      "grad_norm": 1547.339111328125,
      "learning_rate": 0.00019229580591093585,
      "loss": 7.5742,
      "step": 109050
    },
    {
      "epoch": 13.123947051744885,
      "grad_norm": 3020.97998046875,
      "learning_rate": 0.00019229434157866038,
      "loss": 7.5616,
      "step": 109060
    },
    {
      "epoch": 13.125150421179303,
      "grad_norm": 1091.6090087890625,
      "learning_rate": 0.0001922928771128122,
      "loss": 7.5141,
      "step": 109070
    },
    {
      "epoch": 13.126353790613718,
      "grad_norm": 1790.71728515625,
      "learning_rate": 0.00019229141251339346,
      "loss": 7.4928,
      "step": 109080
    },
    {
      "epoch": 13.127557160048134,
      "grad_norm": 1158.9244384765625,
      "learning_rate": 0.00019228994778040624,
      "loss": 7.4885,
      "step": 109090
    },
    {
      "epoch": 13.128760529482552,
      "grad_norm": 1233.3555908203125,
      "learning_rate": 0.00019228848291385272,
      "loss": 7.465,
      "step": 109100
    },
    {
      "epoch": 13.129963898916968,
      "grad_norm": 1639.1177978515625,
      "learning_rate": 0.00019228701791373495,
      "loss": 7.4926,
      "step": 109110
    },
    {
      "epoch": 13.131167268351383,
      "grad_norm": 808.8528442382812,
      "learning_rate": 0.0001922855527800551,
      "loss": 7.584,
      "step": 109120
    },
    {
      "epoch": 13.132370637785801,
      "grad_norm": 2141.6220703125,
      "learning_rate": 0.00019228408751281525,
      "loss": 7.4634,
      "step": 109130
    },
    {
      "epoch": 13.133574007220217,
      "grad_norm": 1800.8280029296875,
      "learning_rate": 0.00019228262211201753,
      "loss": 7.4745,
      "step": 109140
    },
    {
      "epoch": 13.134777376654633,
      "grad_norm": 1815.041748046875,
      "learning_rate": 0.0001922811565776641,
      "loss": 7.5555,
      "step": 109150
    },
    {
      "epoch": 13.13598074608905,
      "grad_norm": 5168.802734375,
      "learning_rate": 0.00019227969090975704,
      "loss": 7.4652,
      "step": 109160
    },
    {
      "epoch": 13.137184115523466,
      "grad_norm": 4394.5966796875,
      "learning_rate": 0.0001922782251082985,
      "loss": 7.3334,
      "step": 109170
    },
    {
      "epoch": 13.138387484957882,
      "grad_norm": 1770.221923828125,
      "learning_rate": 0.00019227675917329054,
      "loss": 7.4452,
      "step": 109180
    },
    {
      "epoch": 13.1395908543923,
      "grad_norm": 2380.8955078125,
      "learning_rate": 0.00019227529310473539,
      "loss": 7.3853,
      "step": 109190
    },
    {
      "epoch": 13.140794223826715,
      "grad_norm": 5343.65087890625,
      "learning_rate": 0.00019227382690263504,
      "loss": 7.4249,
      "step": 109200
    },
    {
      "epoch": 13.14199759326113,
      "grad_norm": 8766.2265625,
      "learning_rate": 0.00019227236056699174,
      "loss": 7.3178,
      "step": 109210
    },
    {
      "epoch": 13.143200962695548,
      "grad_norm": 18961.572265625,
      "learning_rate": 0.0001922708940978075,
      "loss": 7.2531,
      "step": 109220
    },
    {
      "epoch": 13.144404332129964,
      "grad_norm": 6364.12939453125,
      "learning_rate": 0.00019226942749508456,
      "loss": 7.3507,
      "step": 109230
    },
    {
      "epoch": 13.14560770156438,
      "grad_norm": 10984.939453125,
      "learning_rate": 0.00019226796075882494,
      "loss": 7.2775,
      "step": 109240
    },
    {
      "epoch": 13.146811070998796,
      "grad_norm": 9057.078125,
      "learning_rate": 0.0001922664938890308,
      "loss": 7.3545,
      "step": 109250
    },
    {
      "epoch": 13.148014440433213,
      "grad_norm": 16218.24609375,
      "learning_rate": 0.00019226502688570428,
      "loss": 7.3798,
      "step": 109260
    },
    {
      "epoch": 13.14921780986763,
      "grad_norm": 8272.513671875,
      "learning_rate": 0.00019226355974884748,
      "loss": 7.4395,
      "step": 109270
    },
    {
      "epoch": 13.150421179302045,
      "grad_norm": 5024.8720703125,
      "learning_rate": 0.0001922620924784625,
      "loss": 7.3874,
      "step": 109280
    },
    {
      "epoch": 13.151624548736462,
      "grad_norm": 4208.353515625,
      "learning_rate": 0.0001922606250745515,
      "loss": 7.3606,
      "step": 109290
    },
    {
      "epoch": 13.152827918170878,
      "grad_norm": 5596.2080078125,
      "learning_rate": 0.00019225915753711664,
      "loss": 7.3807,
      "step": 109300
    },
    {
      "epoch": 13.154031287605294,
      "grad_norm": 3162.18603515625,
      "learning_rate": 0.00019225768986615998,
      "loss": 7.3208,
      "step": 109310
    },
    {
      "epoch": 13.155234657039712,
      "grad_norm": 2683.37890625,
      "learning_rate": 0.00019225622206168365,
      "loss": 7.3207,
      "step": 109320
    },
    {
      "epoch": 13.156438026474127,
      "grad_norm": 3708.72216796875,
      "learning_rate": 0.00019225475412368984,
      "loss": 7.1981,
      "step": 109330
    },
    {
      "epoch": 13.157641395908543,
      "grad_norm": 7212.5322265625,
      "learning_rate": 0.00019225328605218056,
      "loss": 7.255,
      "step": 109340
    },
    {
      "epoch": 13.15884476534296,
      "grad_norm": 1769.2431640625,
      "learning_rate": 0.00019225181784715802,
      "loss": 7.4158,
      "step": 109350
    },
    {
      "epoch": 13.160048134777377,
      "grad_norm": 1829.830810546875,
      "learning_rate": 0.00019225034950862434,
      "loss": 7.2682,
      "step": 109360
    },
    {
      "epoch": 13.161251504211792,
      "grad_norm": 1371.1326904296875,
      "learning_rate": 0.0001922488810365816,
      "loss": 7.3645,
      "step": 109370
    },
    {
      "epoch": 13.16245487364621,
      "grad_norm": 3305.714111328125,
      "learning_rate": 0.00019224741243103197,
      "loss": 7.3996,
      "step": 109380
    },
    {
      "epoch": 13.163658243080626,
      "grad_norm": 1587.72119140625,
      "learning_rate": 0.00019224594369197755,
      "loss": 7.328,
      "step": 109390
    },
    {
      "epoch": 13.164861612515042,
      "grad_norm": 2476.06298828125,
      "learning_rate": 0.0001922444748194205,
      "loss": 7.3211,
      "step": 109400
    },
    {
      "epoch": 13.166064981949459,
      "grad_norm": 3601.108642578125,
      "learning_rate": 0.0001922430058133629,
      "loss": 7.27,
      "step": 109410
    },
    {
      "epoch": 13.167268351383875,
      "grad_norm": 1661.8583984375,
      "learning_rate": 0.00019224153667380688,
      "loss": 7.322,
      "step": 109420
    },
    {
      "epoch": 13.16847172081829,
      "grad_norm": 2693.05419921875,
      "learning_rate": 0.00019224006740075462,
      "loss": 7.4785,
      "step": 109430
    },
    {
      "epoch": 13.169675090252708,
      "grad_norm": 1941.9661865234375,
      "learning_rate": 0.00019223859799420817,
      "loss": 7.3159,
      "step": 109440
    },
    {
      "epoch": 13.170878459687124,
      "grad_norm": 1575.8074951171875,
      "learning_rate": 0.00019223712845416974,
      "loss": 7.2925,
      "step": 109450
    },
    {
      "epoch": 13.17208182912154,
      "grad_norm": 7722.078125,
      "learning_rate": 0.00019223565878064139,
      "loss": 7.2437,
      "step": 109460
    },
    {
      "epoch": 13.173285198555957,
      "grad_norm": 2231.897705078125,
      "learning_rate": 0.00019223418897362527,
      "loss": 7.3116,
      "step": 109470
    },
    {
      "epoch": 13.174488567990373,
      "grad_norm": 5219.56787109375,
      "learning_rate": 0.00019223271903312349,
      "loss": 7.3412,
      "step": 109480
    },
    {
      "epoch": 13.175691937424789,
      "grad_norm": 2755.5380859375,
      "learning_rate": 0.0001922312489591382,
      "loss": 7.27,
      "step": 109490
    },
    {
      "epoch": 13.176895306859207,
      "grad_norm": 18974.353515625,
      "learning_rate": 0.00019222977875167155,
      "loss": 7.2698,
      "step": 109500
    },
    {
      "epoch": 13.178098676293622,
      "grad_norm": 2879.1416015625,
      "learning_rate": 0.0001922283084107256,
      "loss": 7.3253,
      "step": 109510
    },
    {
      "epoch": 13.179302045728038,
      "grad_norm": 1610.7001953125,
      "learning_rate": 0.00019222683793630255,
      "loss": 7.3353,
      "step": 109520
    },
    {
      "epoch": 13.180505415162456,
      "grad_norm": 2331.283935546875,
      "learning_rate": 0.0001922253673284045,
      "loss": 7.2884,
      "step": 109530
    },
    {
      "epoch": 13.181708784596871,
      "grad_norm": 1315.8941650390625,
      "learning_rate": 0.00019222389658703356,
      "loss": 7.4174,
      "step": 109540
    },
    {
      "epoch": 13.182912154031287,
      "grad_norm": 1656.6331787109375,
      "learning_rate": 0.00019222242571219185,
      "loss": 7.2555,
      "step": 109550
    },
    {
      "epoch": 13.184115523465705,
      "grad_norm": 1284.9022216796875,
      "learning_rate": 0.00019222095470388152,
      "loss": 7.3754,
      "step": 109560
    },
    {
      "epoch": 13.18531889290012,
      "grad_norm": 5103.41259765625,
      "learning_rate": 0.00019221948356210473,
      "loss": 7.294,
      "step": 109570
    },
    {
      "epoch": 13.186522262334536,
      "grad_norm": 10232.8310546875,
      "learning_rate": 0.00019221801228686358,
      "loss": 7.4663,
      "step": 109580
    },
    {
      "epoch": 13.187725631768952,
      "grad_norm": 13162.70703125,
      "learning_rate": 0.00019221654087816018,
      "loss": 7.3562,
      "step": 109590
    },
    {
      "epoch": 13.18892900120337,
      "grad_norm": 5513.1787109375,
      "learning_rate": 0.00019221506933599672,
      "loss": 7.4971,
      "step": 109600
    },
    {
      "epoch": 13.190132370637786,
      "grad_norm": 5229.48681640625,
      "learning_rate": 0.00019221359766037523,
      "loss": 7.5271,
      "step": 109610
    },
    {
      "epoch": 13.191335740072201,
      "grad_norm": 7014.1669921875,
      "learning_rate": 0.00019221212585129792,
      "loss": 7.5002,
      "step": 109620
    },
    {
      "epoch": 13.192539109506619,
      "grad_norm": 2516.60205078125,
      "learning_rate": 0.0001922106539087669,
      "loss": 7.5269,
      "step": 109630
    },
    {
      "epoch": 13.193742478941035,
      "grad_norm": 3174.449462890625,
      "learning_rate": 0.00019220918183278428,
      "loss": 7.3531,
      "step": 109640
    },
    {
      "epoch": 13.19494584837545,
      "grad_norm": 5468.00048828125,
      "learning_rate": 0.00019220770962335225,
      "loss": 7.3733,
      "step": 109650
    },
    {
      "epoch": 13.196149217809868,
      "grad_norm": 10192.5185546875,
      "learning_rate": 0.00019220623728047285,
      "loss": 7.4507,
      "step": 109660
    },
    {
      "epoch": 13.197352587244284,
      "grad_norm": 5235.59423828125,
      "learning_rate": 0.00019220476480414828,
      "loss": 7.4021,
      "step": 109670
    },
    {
      "epoch": 13.1985559566787,
      "grad_norm": 25926.591796875,
      "learning_rate": 0.00019220329219438067,
      "loss": 7.6072,
      "step": 109680
    },
    {
      "epoch": 13.199759326113117,
      "grad_norm": 6222.349609375,
      "learning_rate": 0.0001922018194511721,
      "loss": 7.466,
      "step": 109690
    },
    {
      "epoch": 13.200962695547533,
      "grad_norm": 5446.16162109375,
      "learning_rate": 0.00019220034657452476,
      "loss": 7.3945,
      "step": 109700
    },
    {
      "epoch": 13.202166064981949,
      "grad_norm": 5760.71044921875,
      "learning_rate": 0.00019219887356444076,
      "loss": 7.3501,
      "step": 109710
    },
    {
      "epoch": 13.203369434416366,
      "grad_norm": 5378.56689453125,
      "learning_rate": 0.0001921974004209222,
      "loss": 7.4397,
      "step": 109720
    },
    {
      "epoch": 13.204572803850782,
      "grad_norm": 3318.55029296875,
      "learning_rate": 0.00019219592714397123,
      "loss": 7.4323,
      "step": 109730
    },
    {
      "epoch": 13.205776173285198,
      "grad_norm": 13129.595703125,
      "learning_rate": 0.00019219445373359003,
      "loss": 7.3708,
      "step": 109740
    },
    {
      "epoch": 13.206979542719615,
      "grad_norm": 8539.7080078125,
      "learning_rate": 0.00019219298018978065,
      "loss": 7.4311,
      "step": 109750
    },
    {
      "epoch": 13.208182912154031,
      "grad_norm": 3719.565673828125,
      "learning_rate": 0.0001921915065125453,
      "loss": 7.2877,
      "step": 109760
    },
    {
      "epoch": 13.209386281588447,
      "grad_norm": 20094.443359375,
      "learning_rate": 0.00019219003270188608,
      "loss": 7.4069,
      "step": 109770
    },
    {
      "epoch": 13.210589651022865,
      "grad_norm": 3689.990234375,
      "learning_rate": 0.0001921885587578051,
      "loss": 7.3464,
      "step": 109780
    },
    {
      "epoch": 13.21179302045728,
      "grad_norm": 5972.1689453125,
      "learning_rate": 0.00019218708468030454,
      "loss": 7.2594,
      "step": 109790
    },
    {
      "epoch": 13.212996389891696,
      "grad_norm": 4884.939453125,
      "learning_rate": 0.00019218561046938648,
      "loss": 7.3278,
      "step": 109800
    },
    {
      "epoch": 13.214199759326114,
      "grad_norm": 13642.798828125,
      "learning_rate": 0.0001921841361250531,
      "loss": 7.2183,
      "step": 109810
    },
    {
      "epoch": 13.21540312876053,
      "grad_norm": 6999.47900390625,
      "learning_rate": 0.0001921826616473065,
      "loss": 7.2478,
      "step": 109820
    },
    {
      "epoch": 13.216606498194945,
      "grad_norm": 8606.3095703125,
      "learning_rate": 0.0001921811870361488,
      "loss": 7.4009,
      "step": 109830
    },
    {
      "epoch": 13.217809867629363,
      "grad_norm": 5449.6181640625,
      "learning_rate": 0.00019217971229158222,
      "loss": 7.3963,
      "step": 109840
    },
    {
      "epoch": 13.219013237063779,
      "grad_norm": 7508.28466796875,
      "learning_rate": 0.00019217823741360883,
      "loss": 7.3927,
      "step": 109850
    },
    {
      "epoch": 13.220216606498195,
      "grad_norm": 2850.408447265625,
      "learning_rate": 0.00019217676240223077,
      "loss": 7.2463,
      "step": 109860
    },
    {
      "epoch": 13.221419975932612,
      "grad_norm": 1229.3658447265625,
      "learning_rate": 0.00019217528725745016,
      "loss": 7.3286,
      "step": 109870
    },
    {
      "epoch": 13.222623345367028,
      "grad_norm": 1293.100830078125,
      "learning_rate": 0.00019217381197926919,
      "loss": 7.3012,
      "step": 109880
    },
    {
      "epoch": 13.223826714801444,
      "grad_norm": 3768.919189453125,
      "learning_rate": 0.00019217233656768993,
      "loss": 7.3584,
      "step": 109890
    },
    {
      "epoch": 13.225030084235861,
      "grad_norm": 1835.58447265625,
      "learning_rate": 0.00019217086102271452,
      "loss": 7.3239,
      "step": 109900
    },
    {
      "epoch": 13.226233453670277,
      "grad_norm": 4949.21826171875,
      "learning_rate": 0.00019216938534434515,
      "loss": 7.3144,
      "step": 109910
    },
    {
      "epoch": 13.227436823104693,
      "grad_norm": 3693.646484375,
      "learning_rate": 0.0001921679095325839,
      "loss": 7.2849,
      "step": 109920
    },
    {
      "epoch": 13.22864019253911,
      "grad_norm": 2925.319091796875,
      "learning_rate": 0.00019216643358743296,
      "loss": 7.3033,
      "step": 109930
    },
    {
      "epoch": 13.229843561973526,
      "grad_norm": 8444.2021484375,
      "learning_rate": 0.0001921649575088944,
      "loss": 7.2008,
      "step": 109940
    },
    {
      "epoch": 13.231046931407942,
      "grad_norm": 7526.40966796875,
      "learning_rate": 0.00019216348129697043,
      "loss": 7.385,
      "step": 109950
    },
    {
      "epoch": 13.232250300842358,
      "grad_norm": 10518.126953125,
      "learning_rate": 0.0001921620049516631,
      "loss": 7.3023,
      "step": 109960
    },
    {
      "epoch": 13.233453670276775,
      "grad_norm": 5515.12255859375,
      "learning_rate": 0.00019216052847297464,
      "loss": 7.3565,
      "step": 109970
    },
    {
      "epoch": 13.234657039711191,
      "grad_norm": 4395.20166015625,
      "learning_rate": 0.00019215905186090713,
      "loss": 7.3707,
      "step": 109980
    },
    {
      "epoch": 13.235860409145607,
      "grad_norm": 3720.353515625,
      "learning_rate": 0.0001921575751154627,
      "loss": 7.3712,
      "step": 109990
    },
    {
      "epoch": 13.237063778580024,
      "grad_norm": 2171.3388671875,
      "learning_rate": 0.00019215609823664353,
      "loss": 7.3181,
      "step": 110000
    },
    {
      "epoch": 13.23826714801444,
      "grad_norm": 3307.187744140625,
      "learning_rate": 0.00019215462122445173,
      "loss": 7.3334,
      "step": 110010
    },
    {
      "epoch": 13.239470517448856,
      "grad_norm": 3297.223388671875,
      "learning_rate": 0.00019215314407888942,
      "loss": 7.379,
      "step": 110020
    },
    {
      "epoch": 13.240673886883274,
      "grad_norm": 2438.4609375,
      "learning_rate": 0.00019215166679995878,
      "loss": 7.4286,
      "step": 110030
    },
    {
      "epoch": 13.24187725631769,
      "grad_norm": 2479.4765625,
      "learning_rate": 0.00019215018938766192,
      "loss": 7.3698,
      "step": 110040
    },
    {
      "epoch": 13.243080625752105,
      "grad_norm": 3579.760986328125,
      "learning_rate": 0.00019214871184200098,
      "loss": 7.3216,
      "step": 110050
    },
    {
      "epoch": 13.244283995186523,
      "grad_norm": 14938.8564453125,
      "learning_rate": 0.0001921472341629781,
      "loss": 7.3142,
      "step": 110060
    },
    {
      "epoch": 13.245487364620939,
      "grad_norm": 7436.0517578125,
      "learning_rate": 0.00019214575635059542,
      "loss": 7.2936,
      "step": 110070
    },
    {
      "epoch": 13.246690734055354,
      "grad_norm": 467193.0,
      "learning_rate": 0.0001921442784048551,
      "loss": 7.356,
      "step": 110080
    },
    {
      "epoch": 13.247894103489772,
      "grad_norm": 347034.1875,
      "learning_rate": 0.00019214280032575924,
      "loss": 7.3351,
      "step": 110090
    },
    {
      "epoch": 13.249097472924188,
      "grad_norm": 376508.1875,
      "learning_rate": 0.00019214132211331,
      "loss": 7.3825,
      "step": 110100
    },
    {
      "epoch": 13.250300842358604,
      "grad_norm": 199689.65625,
      "learning_rate": 0.00019213984376750953,
      "loss": 7.2715,
      "step": 110110
    },
    {
      "epoch": 13.251504211793021,
      "grad_norm": 187651.8125,
      "learning_rate": 0.00019213836528835995,
      "loss": 7.4711,
      "step": 110120
    },
    {
      "epoch": 13.252707581227437,
      "grad_norm": 221268.75,
      "learning_rate": 0.0001921368866758634,
      "loss": 7.5331,
      "step": 110130
    },
    {
      "epoch": 13.253910950661853,
      "grad_norm": 123181.625,
      "learning_rate": 0.00019213540793002205,
      "loss": 7.725,
      "step": 110140
    },
    {
      "epoch": 13.25511432009627,
      "grad_norm": 198592.984375,
      "learning_rate": 0.000192133929050838,
      "loss": 7.5445,
      "step": 110150
    },
    {
      "epoch": 13.256317689530686,
      "grad_norm": 532332.25,
      "learning_rate": 0.0001921324500383134,
      "loss": 7.9431,
      "step": 110160
    },
    {
      "epoch": 13.257521058965102,
      "grad_norm": 371910.6875,
      "learning_rate": 0.0001921309708924504,
      "loss": 9.2739,
      "step": 110170
    },
    {
      "epoch": 13.25872442839952,
      "grad_norm": 1474643.0,
      "learning_rate": 0.00019212949161325117,
      "loss": 9.9371,
      "step": 110180
    },
    {
      "epoch": 13.259927797833935,
      "grad_norm": 3103805.0,
      "learning_rate": 0.0001921280122007178,
      "loss": 9.8847,
      "step": 110190
    },
    {
      "epoch": 13.261131167268351,
      "grad_norm": 3739055.0,
      "learning_rate": 0.00019212653265485242,
      "loss": 10.0419,
      "step": 110200
    },
    {
      "epoch": 13.262334536702769,
      "grad_norm": 818148.875,
      "learning_rate": 0.00019212505297565725,
      "loss": 9.682,
      "step": 110210
    },
    {
      "epoch": 13.263537906137184,
      "grad_norm": 441856.84375,
      "learning_rate": 0.00019212357316313438,
      "loss": 8.9333,
      "step": 110220
    },
    {
      "epoch": 13.2647412755716,
      "grad_norm": 351513.40625,
      "learning_rate": 0.00019212209321728592,
      "loss": 8.4022,
      "step": 110230
    },
    {
      "epoch": 13.265944645006018,
      "grad_norm": 222066.546875,
      "learning_rate": 0.00019212061313811406,
      "loss": 8.3944,
      "step": 110240
    },
    {
      "epoch": 13.267148014440433,
      "grad_norm": 523585.0625,
      "learning_rate": 0.00019211913292562093,
      "loss": 8.3361,
      "step": 110250
    },
    {
      "epoch": 13.26835138387485,
      "grad_norm": 152219.90625,
      "learning_rate": 0.0001921176525798087,
      "loss": 8.4178,
      "step": 110260
    },
    {
      "epoch": 13.269554753309265,
      "grad_norm": 168611.453125,
      "learning_rate": 0.00019211617210067945,
      "loss": 8.2823,
      "step": 110270
    },
    {
      "epoch": 13.270758122743683,
      "grad_norm": 168448.765625,
      "learning_rate": 0.00019211469148823536,
      "loss": 8.0514,
      "step": 110280
    },
    {
      "epoch": 13.271961492178098,
      "grad_norm": 289148.90625,
      "learning_rate": 0.00019211321074247856,
      "loss": 7.751,
      "step": 110290
    },
    {
      "epoch": 13.273164861612514,
      "grad_norm": 235207.921875,
      "learning_rate": 0.00019211172986341124,
      "loss": 7.7347,
      "step": 110300
    },
    {
      "epoch": 13.274368231046932,
      "grad_norm": 175265.390625,
      "learning_rate": 0.00019211024885103548,
      "loss": 7.8259,
      "step": 110310
    },
    {
      "epoch": 13.275571600481348,
      "grad_norm": 92630.265625,
      "learning_rate": 0.00019210876770535344,
      "loss": 7.6984,
      "step": 110320
    },
    {
      "epoch": 13.276774969915763,
      "grad_norm": 192967.0,
      "learning_rate": 0.00019210728642636728,
      "loss": 7.769,
      "step": 110330
    },
    {
      "epoch": 13.277978339350181,
      "grad_norm": 93588.078125,
      "learning_rate": 0.00019210580501407916,
      "loss": 7.6669,
      "step": 110340
    },
    {
      "epoch": 13.279181708784597,
      "grad_norm": 44452.171875,
      "learning_rate": 0.00019210432346849116,
      "loss": 7.6325,
      "step": 110350
    },
    {
      "epoch": 13.280385078219012,
      "grad_norm": 52937.7578125,
      "learning_rate": 0.00019210284178960548,
      "loss": 7.6443,
      "step": 110360
    },
    {
      "epoch": 13.28158844765343,
      "grad_norm": 123180.0234375,
      "learning_rate": 0.00019210135997742428,
      "loss": 7.4855,
      "step": 110370
    },
    {
      "epoch": 13.282791817087846,
      "grad_norm": 186309.625,
      "learning_rate": 0.00019209987803194963,
      "loss": 7.6418,
      "step": 110380
    },
    {
      "epoch": 13.283995186522262,
      "grad_norm": 144785.140625,
      "learning_rate": 0.00019209839595318374,
      "loss": 7.5307,
      "step": 110390
    },
    {
      "epoch": 13.28519855595668,
      "grad_norm": 57480.515625,
      "learning_rate": 0.00019209691374112868,
      "loss": 7.4946,
      "step": 110400
    },
    {
      "epoch": 13.286401925391095,
      "grad_norm": 87574.5078125,
      "learning_rate": 0.0001920954313957867,
      "loss": 7.4765,
      "step": 110410
    },
    {
      "epoch": 13.28760529482551,
      "grad_norm": 223329.765625,
      "learning_rate": 0.00019209394891715991,
      "loss": 7.5298,
      "step": 110420
    },
    {
      "epoch": 13.288808664259928,
      "grad_norm": 367378.78125,
      "learning_rate": 0.0001920924663052504,
      "loss": 7.5591,
      "step": 110430
    },
    {
      "epoch": 13.290012033694344,
      "grad_norm": 141020.21875,
      "learning_rate": 0.00019209098356006037,
      "loss": 7.5163,
      "step": 110440
    },
    {
      "epoch": 13.29121540312876,
      "grad_norm": 101079.3203125,
      "learning_rate": 0.00019208950068159194,
      "loss": 7.6223,
      "step": 110450
    },
    {
      "epoch": 13.292418772563177,
      "grad_norm": 41622.0078125,
      "learning_rate": 0.00019208801766984728,
      "loss": 7.572,
      "step": 110460
    },
    {
      "epoch": 13.293622141997593,
      "grad_norm": 124686.2734375,
      "learning_rate": 0.0001920865345248285,
      "loss": 7.6329,
      "step": 110470
    },
    {
      "epoch": 13.294825511432009,
      "grad_norm": 259451.703125,
      "learning_rate": 0.0001920850512465378,
      "loss": 7.7731,
      "step": 110480
    },
    {
      "epoch": 13.296028880866427,
      "grad_norm": 108624.3125,
      "learning_rate": 0.00019208356783497726,
      "loss": 7.5644,
      "step": 110490
    },
    {
      "epoch": 13.297232250300842,
      "grad_norm": 101247.6875,
      "learning_rate": 0.0001920820842901491,
      "loss": 7.4012,
      "step": 110500
    },
    {
      "epoch": 13.298435619735258,
      "grad_norm": 193138.0625,
      "learning_rate": 0.00019208060061205542,
      "loss": 7.5445,
      "step": 110510
    },
    {
      "epoch": 13.299638989169676,
      "grad_norm": 167540.546875,
      "learning_rate": 0.0001920791168006984,
      "loss": 7.4594,
      "step": 110520
    },
    {
      "epoch": 13.300842358604092,
      "grad_norm": 55895.1484375,
      "learning_rate": 0.00019207763285608012,
      "loss": 7.4328,
      "step": 110530
    },
    {
      "epoch": 13.302045728038507,
      "grad_norm": 197037.609375,
      "learning_rate": 0.0001920761487782028,
      "loss": 7.3331,
      "step": 110540
    },
    {
      "epoch": 13.303249097472925,
      "grad_norm": 66672.2578125,
      "learning_rate": 0.00019207466456706854,
      "loss": 7.4436,
      "step": 110550
    },
    {
      "epoch": 13.30445246690734,
      "grad_norm": 230686.03125,
      "learning_rate": 0.00019207318022267952,
      "loss": 7.4088,
      "step": 110560
    },
    {
      "epoch": 13.305655836341757,
      "grad_norm": 91527.140625,
      "learning_rate": 0.00019207169574503787,
      "loss": 7.3945,
      "step": 110570
    },
    {
      "epoch": 13.306859205776174,
      "grad_norm": 306368.1875,
      "learning_rate": 0.00019207021113414573,
      "loss": 7.4848,
      "step": 110580
    },
    {
      "epoch": 13.30806257521059,
      "grad_norm": 96212.015625,
      "learning_rate": 0.0001920687263900053,
      "loss": 7.4783,
      "step": 110590
    },
    {
      "epoch": 13.309265944645006,
      "grad_norm": 148426.03125,
      "learning_rate": 0.00019206724151261866,
      "loss": 7.4722,
      "step": 110600
    },
    {
      "epoch": 13.310469314079423,
      "grad_norm": 515727.4375,
      "learning_rate": 0.000192065756501988,
      "loss": 7.4529,
      "step": 110610
    },
    {
      "epoch": 13.311672683513839,
      "grad_norm": 91867.6796875,
      "learning_rate": 0.00019206427135811546,
      "loss": 7.4579,
      "step": 110620
    },
    {
      "epoch": 13.312876052948255,
      "grad_norm": 179382.359375,
      "learning_rate": 0.0001920627860810032,
      "loss": 7.426,
      "step": 110630
    },
    {
      "epoch": 13.314079422382672,
      "grad_norm": 162694.515625,
      "learning_rate": 0.00019206130067065335,
      "loss": 7.3658,
      "step": 110640
    },
    {
      "epoch": 13.315282791817088,
      "grad_norm": 34279.328125,
      "learning_rate": 0.00019205981512706804,
      "loss": 7.464,
      "step": 110650
    },
    {
      "epoch": 13.316486161251504,
      "grad_norm": 81591.9140625,
      "learning_rate": 0.00019205832945024952,
      "loss": 7.4596,
      "step": 110660
    },
    {
      "epoch": 13.31768953068592,
      "grad_norm": 125334.453125,
      "learning_rate": 0.0001920568436401998,
      "loss": 7.434,
      "step": 110670
    },
    {
      "epoch": 13.318892900120337,
      "grad_norm": 174087.625,
      "learning_rate": 0.0001920553576969211,
      "loss": 7.3845,
      "step": 110680
    },
    {
      "epoch": 13.320096269554753,
      "grad_norm": 73718.3359375,
      "learning_rate": 0.00019205387162041563,
      "loss": 7.3848,
      "step": 110690
    },
    {
      "epoch": 13.321299638989169,
      "grad_norm": 68589.8359375,
      "learning_rate": 0.00019205238541068545,
      "loss": 7.3469,
      "step": 110700
    },
    {
      "epoch": 13.322503008423586,
      "grad_norm": 1782040.875,
      "learning_rate": 0.00019205089906773273,
      "loss": 7.4777,
      "step": 110710
    },
    {
      "epoch": 13.323706377858002,
      "grad_norm": 95544.1953125,
      "learning_rate": 0.00019204941259155965,
      "loss": 7.5217,
      "step": 110720
    },
    {
      "epoch": 13.324909747292418,
      "grad_norm": 1522073.375,
      "learning_rate": 0.00019204792598216832,
      "loss": 8.0024,
      "step": 110730
    },
    {
      "epoch": 13.326113116726836,
      "grad_norm": 324821.5625,
      "learning_rate": 0.00019204643923956093,
      "loss": 8.1135,
      "step": 110740
    },
    {
      "epoch": 13.327316486161251,
      "grad_norm": 203893.78125,
      "learning_rate": 0.00019204495236373962,
      "loss": 8.0937,
      "step": 110750
    },
    {
      "epoch": 13.328519855595667,
      "grad_norm": 2383935.75,
      "learning_rate": 0.00019204346535470652,
      "loss": 8.3211,
      "step": 110760
    },
    {
      "epoch": 13.329723225030085,
      "grad_norm": 1821319.625,
      "learning_rate": 0.00019204197821246383,
      "loss": 8.4965,
      "step": 110770
    },
    {
      "epoch": 13.3309265944645,
      "grad_norm": 334876.78125,
      "learning_rate": 0.00019204049093701366,
      "loss": 9.1062,
      "step": 110780
    },
    {
      "epoch": 13.332129963898916,
      "grad_norm": 1427789.0,
      "learning_rate": 0.00019203900352835818,
      "loss": 9.3586,
      "step": 110790
    },
    {
      "epoch": 13.333333333333334,
      "grad_norm": 910.2598266601562,
      "learning_rate": 0.0001920375159864995,
      "loss": 9.1072,
      "step": 110800
    },
    {
      "epoch": 13.33453670276775,
      "grad_norm": 2685.69775390625,
      "learning_rate": 0.00019203602831143986,
      "loss": 8.4476,
      "step": 110810
    },
    {
      "epoch": 13.335740072202166,
      "grad_norm": 604.9318237304688,
      "learning_rate": 0.00019203454050318135,
      "loss": 8.3137,
      "step": 110820
    },
    {
      "epoch": 13.336943441636583,
      "grad_norm": 1511.6759033203125,
      "learning_rate": 0.0001920330525617261,
      "loss": 8.0222,
      "step": 110830
    },
    {
      "epoch": 13.338146811070999,
      "grad_norm": 182.51419067382812,
      "learning_rate": 0.00019203156448707636,
      "loss": 7.6407,
      "step": 110840
    },
    {
      "epoch": 13.339350180505415,
      "grad_norm": 157.84323120117188,
      "learning_rate": 0.00019203007627923422,
      "loss": 7.8411,
      "step": 110850
    },
    {
      "epoch": 13.340553549939832,
      "grad_norm": 200.6873321533203,
      "learning_rate": 0.00019202858793820182,
      "loss": 7.7679,
      "step": 110860
    },
    {
      "epoch": 13.341756919374248,
      "grad_norm": 167.75271606445312,
      "learning_rate": 0.00019202709946398132,
      "loss": 7.5541,
      "step": 110870
    },
    {
      "epoch": 13.342960288808664,
      "grad_norm": 471.394775390625,
      "learning_rate": 0.0001920256108565749,
      "loss": 7.6278,
      "step": 110880
    },
    {
      "epoch": 13.344163658243081,
      "grad_norm": 273.3186340332031,
      "learning_rate": 0.00019202412211598467,
      "loss": 7.5743,
      "step": 110890
    },
    {
      "epoch": 13.345367027677497,
      "grad_norm": 202.0496826171875,
      "learning_rate": 0.00019202263324221287,
      "loss": 7.6374,
      "step": 110900
    },
    {
      "epoch": 13.346570397111913,
      "grad_norm": 247.0529327392578,
      "learning_rate": 0.00019202114423526158,
      "loss": 7.6715,
      "step": 110910
    },
    {
      "epoch": 13.34777376654633,
      "grad_norm": 96.9984130859375,
      "learning_rate": 0.00019201965509513296,
      "loss": 7.5571,
      "step": 110920
    },
    {
      "epoch": 13.348977135980746,
      "grad_norm": 844.7772827148438,
      "learning_rate": 0.00019201816582182919,
      "loss": 7.6822,
      "step": 110930
    },
    {
      "epoch": 13.350180505415162,
      "grad_norm": 1959.7215576171875,
      "learning_rate": 0.00019201667641535243,
      "loss": 7.6507,
      "step": 110940
    },
    {
      "epoch": 13.35138387484958,
      "grad_norm": 387.8429870605469,
      "learning_rate": 0.0001920151868757048,
      "loss": 7.7109,
      "step": 110950
    },
    {
      "epoch": 13.352587244283995,
      "grad_norm": 142.9046630859375,
      "learning_rate": 0.00019201369720288843,
      "loss": 7.534,
      "step": 110960
    },
    {
      "epoch": 13.353790613718411,
      "grad_norm": 35.90800094604492,
      "learning_rate": 0.0001920122073969056,
      "loss": 7.6489,
      "step": 110970
    },
    {
      "epoch": 13.354993983152827,
      "grad_norm": 46.38391876220703,
      "learning_rate": 0.00019201071745775835,
      "loss": 7.7431,
      "step": 110980
    },
    {
      "epoch": 13.356197352587245,
      "grad_norm": 31.96817398071289,
      "learning_rate": 0.0001920092273854489,
      "loss": 7.742,
      "step": 110990
    },
    {
      "epoch": 13.35740072202166,
      "grad_norm": 27.3577880859375,
      "learning_rate": 0.00019200773717997937,
      "loss": 7.6949,
      "step": 111000
    },
    {
      "epoch": 13.358604091456076,
      "grad_norm": 131.8791961669922,
      "learning_rate": 0.00019200624684135193,
      "loss": 7.5673,
      "step": 111010
    },
    {
      "epoch": 13.359807460890494,
      "grad_norm": 125.35891723632812,
      "learning_rate": 0.00019200475636956874,
      "loss": 7.5904,
      "step": 111020
    },
    {
      "epoch": 13.36101083032491,
      "grad_norm": 845.0186767578125,
      "learning_rate": 0.00019200326576463192,
      "loss": 7.6362,
      "step": 111030
    },
    {
      "epoch": 13.362214199759325,
      "grad_norm": 556.4752197265625,
      "learning_rate": 0.0001920017750265437,
      "loss": 7.5838,
      "step": 111040
    },
    {
      "epoch": 13.363417569193743,
      "grad_norm": 97.72586059570312,
      "learning_rate": 0.00019200028415530617,
      "loss": 7.4948,
      "step": 111050
    },
    {
      "epoch": 13.364620938628159,
      "grad_norm": 263.8709411621094,
      "learning_rate": 0.00019199879315092154,
      "loss": 7.5156,
      "step": 111060
    },
    {
      "epoch": 13.365824308062574,
      "grad_norm": 67.794921875,
      "learning_rate": 0.00019199730201339193,
      "loss": 7.5486,
      "step": 111070
    },
    {
      "epoch": 13.367027677496992,
      "grad_norm": 224.88714599609375,
      "learning_rate": 0.0001919958107427195,
      "loss": 7.6127,
      "step": 111080
    },
    {
      "epoch": 13.368231046931408,
      "grad_norm": 96.0893783569336,
      "learning_rate": 0.00019199431933890644,
      "loss": 7.5068,
      "step": 111090
    },
    {
      "epoch": 13.369434416365824,
      "grad_norm": 71.08943176269531,
      "learning_rate": 0.00019199282780195487,
      "loss": 7.6105,
      "step": 111100
    },
    {
      "epoch": 13.370637785800241,
      "grad_norm": 113.27473449707031,
      "learning_rate": 0.00019199133613186694,
      "loss": 7.4517,
      "step": 111110
    },
    {
      "epoch": 13.371841155234657,
      "grad_norm": 148.3744659423828,
      "learning_rate": 0.0001919898443286449,
      "loss": 7.5555,
      "step": 111120
    },
    {
      "epoch": 13.373044524669073,
      "grad_norm": 294.1316833496094,
      "learning_rate": 0.0001919883523922908,
      "loss": 7.5281,
      "step": 111130
    },
    {
      "epoch": 13.37424789410349,
      "grad_norm": 236.04330444335938,
      "learning_rate": 0.00019198686032280687,
      "loss": 7.4634,
      "step": 111140
    },
    {
      "epoch": 13.375451263537906,
      "grad_norm": 174.94091796875,
      "learning_rate": 0.0001919853681201952,
      "loss": 7.4227,
      "step": 111150
    },
    {
      "epoch": 13.376654632972322,
      "grad_norm": 337.4954528808594,
      "learning_rate": 0.000191983875784458,
      "loss": 7.4553,
      "step": 111160
    },
    {
      "epoch": 13.37785800240674,
      "grad_norm": 214.25624084472656,
      "learning_rate": 0.00019198238331559746,
      "loss": 7.4461,
      "step": 111170
    },
    {
      "epoch": 13.379061371841155,
      "grad_norm": 605.7977294921875,
      "learning_rate": 0.00019198089071361567,
      "loss": 7.5649,
      "step": 111180
    },
    {
      "epoch": 13.380264741275571,
      "grad_norm": 407.91314697265625,
      "learning_rate": 0.00019197939797851486,
      "loss": 7.4917,
      "step": 111190
    },
    {
      "epoch": 13.381468110709989,
      "grad_norm": 456.61358642578125,
      "learning_rate": 0.0001919779051102971,
      "loss": 7.4187,
      "step": 111200
    },
    {
      "epoch": 13.382671480144404,
      "grad_norm": 1044.808349609375,
      "learning_rate": 0.0001919764121089646,
      "loss": 7.5112,
      "step": 111210
    },
    {
      "epoch": 13.38387484957882,
      "grad_norm": 1758.309326171875,
      "learning_rate": 0.00019197491897451958,
      "loss": 7.4937,
      "step": 111220
    },
    {
      "epoch": 13.385078219013238,
      "grad_norm": 2519.55224609375,
      "learning_rate": 0.00019197342570696413,
      "loss": 7.4275,
      "step": 111230
    },
    {
      "epoch": 13.386281588447654,
      "grad_norm": 1460.6614990234375,
      "learning_rate": 0.0001919719323063004,
      "loss": 7.458,
      "step": 111240
    },
    {
      "epoch": 13.38748495788207,
      "grad_norm": 1221.3485107421875,
      "learning_rate": 0.0001919704387725306,
      "loss": 7.4563,
      "step": 111250
    },
    {
      "epoch": 13.388688327316487,
      "grad_norm": 1648.83203125,
      "learning_rate": 0.00019196894510565684,
      "loss": 7.4569,
      "step": 111260
    },
    {
      "epoch": 13.389891696750903,
      "grad_norm": 1689.3546142578125,
      "learning_rate": 0.00019196745130568133,
      "loss": 7.442,
      "step": 111270
    },
    {
      "epoch": 13.391095066185319,
      "grad_norm": 2027.291748046875,
      "learning_rate": 0.0001919659573726062,
      "loss": 7.447,
      "step": 111280
    },
    {
      "epoch": 13.392298435619736,
      "grad_norm": 1643.912841796875,
      "learning_rate": 0.00019196446330643362,
      "loss": 7.3493,
      "step": 111290
    },
    {
      "epoch": 13.393501805054152,
      "grad_norm": 580.79345703125,
      "learning_rate": 0.00019196296910716578,
      "loss": 7.4129,
      "step": 111300
    },
    {
      "epoch": 13.394705174488568,
      "grad_norm": 1131.4515380859375,
      "learning_rate": 0.00019196147477480483,
      "loss": 7.3378,
      "step": 111310
    },
    {
      "epoch": 13.395908543922985,
      "grad_norm": 915.0283203125,
      "learning_rate": 0.00019195998030935288,
      "loss": 7.3997,
      "step": 111320
    },
    {
      "epoch": 13.397111913357401,
      "grad_norm": 3438.982177734375,
      "learning_rate": 0.00019195848571081214,
      "loss": 7.3809,
      "step": 111330
    },
    {
      "epoch": 13.398315282791817,
      "grad_norm": 2424.0498046875,
      "learning_rate": 0.00019195699097918477,
      "loss": 7.4452,
      "step": 111340
    },
    {
      "epoch": 13.399518652226233,
      "grad_norm": 1284.0897216796875,
      "learning_rate": 0.00019195549611447294,
      "loss": 7.468,
      "step": 111350
    },
    {
      "epoch": 13.40072202166065,
      "grad_norm": 1931.1319580078125,
      "learning_rate": 0.0001919540011166788,
      "loss": 7.6212,
      "step": 111360
    },
    {
      "epoch": 13.401925391095066,
      "grad_norm": 10985.525390625,
      "learning_rate": 0.00019195250598580453,
      "loss": 7.4169,
      "step": 111370
    },
    {
      "epoch": 13.403128760529482,
      "grad_norm": 2739.75927734375,
      "learning_rate": 0.0001919510107218523,
      "loss": 7.4372,
      "step": 111380
    },
    {
      "epoch": 13.4043321299639,
      "grad_norm": 2312.317138671875,
      "learning_rate": 0.00019194951532482422,
      "loss": 7.434,
      "step": 111390
    },
    {
      "epoch": 13.405535499398315,
      "grad_norm": 810.5154418945312,
      "learning_rate": 0.0001919480197947225,
      "loss": 7.4419,
      "step": 111400
    },
    {
      "epoch": 13.406738868832731,
      "grad_norm": 2130.062744140625,
      "learning_rate": 0.00019194652413154927,
      "loss": 7.4068,
      "step": 111410
    },
    {
      "epoch": 13.407942238267148,
      "grad_norm": 570.1299438476562,
      "learning_rate": 0.00019194502833530676,
      "loss": 7.4127,
      "step": 111420
    },
    {
      "epoch": 13.409145607701564,
      "grad_norm": 2961.287841796875,
      "learning_rate": 0.00019194353240599706,
      "loss": 7.4312,
      "step": 111430
    },
    {
      "epoch": 13.41034897713598,
      "grad_norm": 1396.0400390625,
      "learning_rate": 0.0001919420363436224,
      "loss": 7.3678,
      "step": 111440
    },
    {
      "epoch": 13.411552346570398,
      "grad_norm": 470.570556640625,
      "learning_rate": 0.00019194054014818487,
      "loss": 7.3416,
      "step": 111450
    },
    {
      "epoch": 13.412755716004813,
      "grad_norm": 940.7001342773438,
      "learning_rate": 0.00019193904381968672,
      "loss": 7.5045,
      "step": 111460
    },
    {
      "epoch": 13.41395908543923,
      "grad_norm": 336.78741455078125,
      "learning_rate": 0.00019193754735813006,
      "loss": 7.3757,
      "step": 111470
    },
    {
      "epoch": 13.415162454873647,
      "grad_norm": 1625.3330078125,
      "learning_rate": 0.00019193605076351706,
      "loss": 7.3302,
      "step": 111480
    },
    {
      "epoch": 13.416365824308063,
      "grad_norm": 641.3401489257812,
      "learning_rate": 0.00019193455403584994,
      "loss": 7.4026,
      "step": 111490
    },
    {
      "epoch": 13.417569193742478,
      "grad_norm": 847.099853515625,
      "learning_rate": 0.00019193305717513077,
      "loss": 7.5379,
      "step": 111500
    },
    {
      "epoch": 13.418772563176896,
      "grad_norm": 1427.4117431640625,
      "learning_rate": 0.00019193156018136178,
      "loss": 7.5689,
      "step": 111510
    },
    {
      "epoch": 13.419975932611312,
      "grad_norm": 986.011962890625,
      "learning_rate": 0.00019193006305454513,
      "loss": 7.5583,
      "step": 111520
    },
    {
      "epoch": 13.421179302045728,
      "grad_norm": 1134.5452880859375,
      "learning_rate": 0.00019192856579468298,
      "loss": 7.4819,
      "step": 111530
    },
    {
      "epoch": 13.422382671480145,
      "grad_norm": 962.0313110351562,
      "learning_rate": 0.0001919270684017775,
      "loss": 7.5342,
      "step": 111540
    },
    {
      "epoch": 13.42358604091456,
      "grad_norm": 445.82147216796875,
      "learning_rate": 0.00019192557087583086,
      "loss": 7.5181,
      "step": 111550
    },
    {
      "epoch": 13.424789410348977,
      "grad_norm": 1068.888671875,
      "learning_rate": 0.00019192407321684524,
      "loss": 7.4808,
      "step": 111560
    },
    {
      "epoch": 13.425992779783394,
      "grad_norm": 613.2035522460938,
      "learning_rate": 0.00019192257542482278,
      "loss": 7.5092,
      "step": 111570
    },
    {
      "epoch": 13.42719614921781,
      "grad_norm": 1166.583740234375,
      "learning_rate": 0.00019192107749976565,
      "loss": 7.5399,
      "step": 111580
    },
    {
      "epoch": 13.428399518652226,
      "grad_norm": 837.2355346679688,
      "learning_rate": 0.00019191957944167601,
      "loss": 7.5028,
      "step": 111590
    },
    {
      "epoch": 13.429602888086643,
      "grad_norm": 297.49188232421875,
      "learning_rate": 0.00019191808125055608,
      "loss": 7.4061,
      "step": 111600
    },
    {
      "epoch": 13.43080625752106,
      "grad_norm": 1156.8074951171875,
      "learning_rate": 0.00019191658292640797,
      "loss": 7.3911,
      "step": 111610
    },
    {
      "epoch": 13.432009626955475,
      "grad_norm": 2390.753662109375,
      "learning_rate": 0.00019191508446923388,
      "loss": 7.396,
      "step": 111620
    },
    {
      "epoch": 13.433212996389893,
      "grad_norm": 26288.8203125,
      "learning_rate": 0.00019191358587903595,
      "loss": 7.4404,
      "step": 111630
    },
    {
      "epoch": 13.434416365824308,
      "grad_norm": 91465.8203125,
      "learning_rate": 0.0001919120871558164,
      "loss": 7.3824,
      "step": 111640
    },
    {
      "epoch": 13.435619735258724,
      "grad_norm": 514493.34375,
      "learning_rate": 0.00019191058829957734,
      "loss": 7.4483,
      "step": 111650
    },
    {
      "epoch": 13.43682310469314,
      "grad_norm": 30260.517578125,
      "learning_rate": 0.000191909089310321,
      "loss": 7.6983,
      "step": 111660
    },
    {
      "epoch": 13.438026474127557,
      "grad_norm": 298628.625,
      "learning_rate": 0.00019190759018804948,
      "loss": 7.3443,
      "step": 111670
    },
    {
      "epoch": 13.439229843561973,
      "grad_norm": 345494.8125,
      "learning_rate": 0.00019190609093276502,
      "loss": 7.423,
      "step": 111680
    },
    {
      "epoch": 13.440433212996389,
      "grad_norm": 60102.81640625,
      "learning_rate": 0.00019190459154446972,
      "loss": 7.6221,
      "step": 111690
    },
    {
      "epoch": 13.441636582430807,
      "grad_norm": 13738.0263671875,
      "learning_rate": 0.0001919030920231658,
      "loss": 7.6564,
      "step": 111700
    },
    {
      "epoch": 13.442839951865222,
      "grad_norm": 36233.39453125,
      "learning_rate": 0.00019190159236885543,
      "loss": 7.4552,
      "step": 111710
    },
    {
      "epoch": 13.444043321299638,
      "grad_norm": 27194894.0,
      "learning_rate": 0.00019190009258154077,
      "loss": 7.6795,
      "step": 111720
    },
    {
      "epoch": 13.445246690734056,
      "grad_norm": 431800.4375,
      "learning_rate": 0.00019189859266122397,
      "loss": 8.6444,
      "step": 111730
    },
    {
      "epoch": 13.446450060168472,
      "grad_norm": 3136103.0,
      "learning_rate": 0.0001918970926079072,
      "loss": 9.2939,
      "step": 111740
    },
    {
      "epoch": 13.447653429602887,
      "grad_norm": 6191241.0,
      "learning_rate": 0.00019189559242159266,
      "loss": 9.0118,
      "step": 111750
    },
    {
      "epoch": 13.448856799037305,
      "grad_norm": 2778721.5,
      "learning_rate": 0.00019189409210228252,
      "loss": 8.9635,
      "step": 111760
    },
    {
      "epoch": 13.45006016847172,
      "grad_norm": 2456772.25,
      "learning_rate": 0.0001918925916499789,
      "loss": 9.1585,
      "step": 111770
    },
    {
      "epoch": 13.451263537906136,
      "grad_norm": 125611504.0,
      "learning_rate": 0.00019189109106468408,
      "loss": 8.8731,
      "step": 111780
    },
    {
      "epoch": 13.452466907340554,
      "grad_norm": 1990498.75,
      "learning_rate": 0.00019188959034640015,
      "loss": 8.7496,
      "step": 111790
    },
    {
      "epoch": 13.45367027677497,
      "grad_norm": 11776526.0,
      "learning_rate": 0.00019188808949512927,
      "loss": 8.527,
      "step": 111800
    },
    {
      "epoch": 13.454873646209386,
      "grad_norm": 1119731.625,
      "learning_rate": 0.00019188658851087363,
      "loss": 8.5386,
      "step": 111810
    },
    {
      "epoch": 13.456077015643803,
      "grad_norm": 9190955.0,
      "learning_rate": 0.00019188508739363544,
      "loss": 8.6631,
      "step": 111820
    },
    {
      "epoch": 13.457280385078219,
      "grad_norm": 926190.8125,
      "learning_rate": 0.00019188358614341682,
      "loss": 8.5332,
      "step": 111830
    },
    {
      "epoch": 13.458483754512635,
      "grad_norm": 2818753.5,
      "learning_rate": 0.00019188208476022,
      "loss": 8.2075,
      "step": 111840
    },
    {
      "epoch": 13.459687123947052,
      "grad_norm": 3279688.75,
      "learning_rate": 0.00019188058324404707,
      "loss": 8.1707,
      "step": 111850
    },
    {
      "epoch": 13.460890493381468,
      "grad_norm": 3798531.75,
      "learning_rate": 0.0001918790815949003,
      "loss": 8.2591,
      "step": 111860
    },
    {
      "epoch": 13.462093862815884,
      "grad_norm": 3740412.5,
      "learning_rate": 0.0001918775798127818,
      "loss": 8.6526,
      "step": 111870
    },
    {
      "epoch": 13.463297232250302,
      "grad_norm": 4504479.5,
      "learning_rate": 0.00019187607789769372,
      "loss": 9.0411,
      "step": 111880
    },
    {
      "epoch": 13.464500601684717,
      "grad_norm": 6027376.0,
      "learning_rate": 0.00019187457584963832,
      "loss": 8.9381,
      "step": 111890
    },
    {
      "epoch": 13.465703971119133,
      "grad_norm": 16627999.0,
      "learning_rate": 0.00019187307366861769,
      "loss": 8.9557,
      "step": 111900
    },
    {
      "epoch": 13.46690734055355,
      "grad_norm": 68241200.0,
      "learning_rate": 0.00019187157135463405,
      "loss": 9.025,
      "step": 111910
    },
    {
      "epoch": 13.468110709987966,
      "grad_norm": 6391292.5,
      "learning_rate": 0.00019187006890768958,
      "loss": 8.7331,
      "step": 111920
    },
    {
      "epoch": 13.469314079422382,
      "grad_norm": 1137912.875,
      "learning_rate": 0.00019186856632778644,
      "loss": 8.2902,
      "step": 111930
    },
    {
      "epoch": 13.4705174488568,
      "grad_norm": 5648679.0,
      "learning_rate": 0.00019186706361492677,
      "loss": 8.3575,
      "step": 111940
    },
    {
      "epoch": 13.471720818291216,
      "grad_norm": 2155786.25,
      "learning_rate": 0.0001918655607691128,
      "loss": 8.7709,
      "step": 111950
    },
    {
      "epoch": 13.472924187725631,
      "grad_norm": 4373881.0,
      "learning_rate": 0.00019186405779034668,
      "loss": 8.9272,
      "step": 111960
    },
    {
      "epoch": 13.474127557160049,
      "grad_norm": 8810573.0,
      "learning_rate": 0.00019186255467863058,
      "loss": 8.9189,
      "step": 111970
    },
    {
      "epoch": 13.475330926594465,
      "grad_norm": 116230408.0,
      "learning_rate": 0.00019186105143396672,
      "loss": 8.6474,
      "step": 111980
    },
    {
      "epoch": 13.47653429602888,
      "grad_norm": 1655787.125,
      "learning_rate": 0.0001918595480563572,
      "loss": 8.5065,
      "step": 111990
    },
    {
      "epoch": 13.477737665463298,
      "grad_norm": 868290.6875,
      "learning_rate": 0.00019185804454580424,
      "loss": 8.2941,
      "step": 112000
    },
    {
      "epoch": 13.478941034897714,
      "grad_norm": 847596.375,
      "learning_rate": 0.00019185654090231002,
      "loss": 8.1891,
      "step": 112010
    },
    {
      "epoch": 13.48014440433213,
      "grad_norm": 837659.0625,
      "learning_rate": 0.00019185503712587673,
      "loss": 8.1248,
      "step": 112020
    },
    {
      "epoch": 13.481347773766545,
      "grad_norm": 1641022.0,
      "learning_rate": 0.00019185353321650648,
      "loss": 8.0722,
      "step": 112030
    },
    {
      "epoch": 13.482551143200963,
      "grad_norm": 621701.4375,
      "learning_rate": 0.0001918520291742015,
      "loss": 8.183,
      "step": 112040
    },
    {
      "epoch": 13.483754512635379,
      "grad_norm": 562886.3125,
      "learning_rate": 0.00019185052499896395,
      "loss": 8.0529,
      "step": 112050
    },
    {
      "epoch": 13.484957882069795,
      "grad_norm": 1825469.875,
      "learning_rate": 0.00019184902069079604,
      "loss": 8.073,
      "step": 112060
    },
    {
      "epoch": 13.486161251504212,
      "grad_norm": 1084234.25,
      "learning_rate": 0.0001918475162496999,
      "loss": 8.0609,
      "step": 112070
    },
    {
      "epoch": 13.487364620938628,
      "grad_norm": 429773.15625,
      "learning_rate": 0.00019184601167567772,
      "loss": 8.099,
      "step": 112080
    },
    {
      "epoch": 13.488567990373044,
      "grad_norm": 349600.46875,
      "learning_rate": 0.0001918445069687317,
      "loss": 7.9518,
      "step": 112090
    },
    {
      "epoch": 13.489771359807461,
      "grad_norm": 463701.96875,
      "learning_rate": 0.000191843002128864,
      "loss": 7.974,
      "step": 112100
    },
    {
      "epoch": 13.490974729241877,
      "grad_norm": 622646528.0,
      "learning_rate": 0.0001918414971560768,
      "loss": 8.7608,
      "step": 112110
    },
    {
      "epoch": 13.492178098676293,
      "grad_norm": 2490747392.0,
      "learning_rate": 0.0001918399920503723,
      "loss": 9.3857,
      "step": 112120
    },
    {
      "epoch": 13.49338146811071,
      "grad_norm": 1269773184.0,
      "learning_rate": 0.00019183848681175261,
      "loss": 9.4415,
      "step": 112130
    },
    {
      "epoch": 13.494584837545126,
      "grad_norm": 115942656.0,
      "learning_rate": 0.00019183698144022,
      "loss": 9.3894,
      "step": 112140
    },
    {
      "epoch": 13.495788206979542,
      "grad_norm": 3249985792.0,
      "learning_rate": 0.00019183547593577658,
      "loss": 8.934,
      "step": 112150
    },
    {
      "epoch": 13.49699157641396,
      "grad_norm": 330774.625,
      "learning_rate": 0.00019183397029842457,
      "loss": 8.6737,
      "step": 112160
    },
    {
      "epoch": 13.498194945848375,
      "grad_norm": 557643.8125,
      "learning_rate": 0.0001918324645281661,
      "loss": 9.1788,
      "step": 112170
    },
    {
      "epoch": 13.499398315282791,
      "grad_norm": 2705413.5,
      "learning_rate": 0.00019183095862500342,
      "loss": 9.439,
      "step": 112180
    },
    {
      "epoch": 13.500601684717209,
      "grad_norm": 1079514.25,
      "learning_rate": 0.00019182945258893863,
      "loss": 9.5887,
      "step": 112190
    },
    {
      "epoch": 13.501805054151625,
      "grad_norm": 3137683.5,
      "learning_rate": 0.00019182794641997396,
      "loss": 9.093,
      "step": 112200
    },
    {
      "epoch": 13.50300842358604,
      "grad_norm": 4004074.5,
      "learning_rate": 0.0001918264401181116,
      "loss": 8.9457,
      "step": 112210
    },
    {
      "epoch": 13.504211793020458,
      "grad_norm": 682983.3125,
      "learning_rate": 0.00019182493368335372,
      "loss": 8.7,
      "step": 112220
    },
    {
      "epoch": 13.505415162454874,
      "grad_norm": 1760792.375,
      "learning_rate": 0.00019182342711570246,
      "loss": 8.5762,
      "step": 112230
    },
    {
      "epoch": 13.50661853188929,
      "grad_norm": 919176.3125,
      "learning_rate": 0.00019182192041516008,
      "loss": 8.562,
      "step": 112240
    },
    {
      "epoch": 13.507821901323707,
      "grad_norm": 881351.125,
      "learning_rate": 0.00019182041358172867,
      "loss": 8.4858,
      "step": 112250
    },
    {
      "epoch": 13.509025270758123,
      "grad_norm": 3324286.75,
      "learning_rate": 0.00019181890661541044,
      "loss": 8.4438,
      "step": 112260
    },
    {
      "epoch": 13.510228640192539,
      "grad_norm": 1041977.375,
      "learning_rate": 0.00019181739951620758,
      "loss": 8.4676,
      "step": 112270
    },
    {
      "epoch": 13.511432009626956,
      "grad_norm": 633026.8125,
      "learning_rate": 0.00019181589228412231,
      "loss": 8.3156,
      "step": 112280
    },
    {
      "epoch": 13.512635379061372,
      "grad_norm": 1533722.25,
      "learning_rate": 0.00019181438491915675,
      "loss": 8.3899,
      "step": 112290
    },
    {
      "epoch": 13.513838748495788,
      "grad_norm": 200570.53125,
      "learning_rate": 0.00019181287742131312,
      "loss": 8.2837,
      "step": 112300
    },
    {
      "epoch": 13.515042117930205,
      "grad_norm": 362316.71875,
      "learning_rate": 0.00019181136979059358,
      "loss": 8.3123,
      "step": 112310
    },
    {
      "epoch": 13.516245487364621,
      "grad_norm": 1786150.75,
      "learning_rate": 0.00019180986202700032,
      "loss": 8.1862,
      "step": 112320
    },
    {
      "epoch": 13.517448856799037,
      "grad_norm": 608451.375,
      "learning_rate": 0.00019180835413053551,
      "loss": 8.182,
      "step": 112330
    },
    {
      "epoch": 13.518652226233453,
      "grad_norm": 736070.6875,
      "learning_rate": 0.00019180684610120138,
      "loss": 8.2476,
      "step": 112340
    },
    {
      "epoch": 13.51985559566787,
      "grad_norm": 435168.6875,
      "learning_rate": 0.00019180533793900005,
      "loss": 8.3545,
      "step": 112350
    },
    {
      "epoch": 13.521058965102286,
      "grad_norm": 333991.34375,
      "learning_rate": 0.00019180382964393376,
      "loss": 8.2604,
      "step": 112360
    },
    {
      "epoch": 13.522262334536702,
      "grad_norm": 128969.703125,
      "learning_rate": 0.00019180232121600465,
      "loss": 7.995,
      "step": 112370
    },
    {
      "epoch": 13.52346570397112,
      "grad_norm": 303082.875,
      "learning_rate": 0.00019180081265521488,
      "loss": 8.0015,
      "step": 112380
    },
    {
      "epoch": 13.524669073405535,
      "grad_norm": 776147.625,
      "learning_rate": 0.0001917993039615667,
      "loss": 8.0454,
      "step": 112390
    },
    {
      "epoch": 13.525872442839951,
      "grad_norm": 380622.28125,
      "learning_rate": 0.00019179779513506226,
      "loss": 8.0309,
      "step": 112400
    },
    {
      "epoch": 13.527075812274369,
      "grad_norm": 335196.6875,
      "learning_rate": 0.00019179628617570376,
      "loss": 8.0057,
      "step": 112410
    },
    {
      "epoch": 13.528279181708784,
      "grad_norm": 147368.046875,
      "learning_rate": 0.00019179477708349335,
      "loss": 8.0296,
      "step": 112420
    },
    {
      "epoch": 13.5294825511432,
      "grad_norm": 1779401.5,
      "learning_rate": 0.00019179326785843324,
      "loss": 8.0364,
      "step": 112430
    },
    {
      "epoch": 13.530685920577618,
      "grad_norm": 574697.5,
      "learning_rate": 0.0001917917585005256,
      "loss": 7.9876,
      "step": 112440
    },
    {
      "epoch": 13.531889290012034,
      "grad_norm": 971916.1875,
      "learning_rate": 0.00019179024900977263,
      "loss": 7.9857,
      "step": 112450
    },
    {
      "epoch": 13.53309265944645,
      "grad_norm": 2491465.5,
      "learning_rate": 0.0001917887393861765,
      "loss": 7.9302,
      "step": 112460
    },
    {
      "epoch": 13.534296028880867,
      "grad_norm": 1735155.625,
      "learning_rate": 0.00019178722962973942,
      "loss": 7.858,
      "step": 112470
    },
    {
      "epoch": 13.535499398315283,
      "grad_norm": 1431859.375,
      "learning_rate": 0.00019178571974046355,
      "loss": 7.872,
      "step": 112480
    },
    {
      "epoch": 13.536702767749698,
      "grad_norm": 370191.375,
      "learning_rate": 0.0001917842097183511,
      "loss": 7.8488,
      "step": 112490
    },
    {
      "epoch": 13.537906137184116,
      "grad_norm": 745230.4375,
      "learning_rate": 0.0001917826995634042,
      "loss": 7.9449,
      "step": 112500
    },
    {
      "epoch": 13.539109506618532,
      "grad_norm": 698564.4375,
      "learning_rate": 0.00019178118927562507,
      "loss": 7.8787,
      "step": 112510
    },
    {
      "epoch": 13.540312876052948,
      "grad_norm": 492286.28125,
      "learning_rate": 0.00019177967885501595,
      "loss": 7.9291,
      "step": 112520
    },
    {
      "epoch": 13.541516245487365,
      "grad_norm": 299212.96875,
      "learning_rate": 0.00019177816830157892,
      "loss": 7.9905,
      "step": 112530
    },
    {
      "epoch": 13.542719614921781,
      "grad_norm": 533764.8125,
      "learning_rate": 0.00019177665761531623,
      "loss": 8.0556,
      "step": 112540
    },
    {
      "epoch": 13.543922984356197,
      "grad_norm": 572797.0,
      "learning_rate": 0.0001917751467962301,
      "loss": 8.0768,
      "step": 112550
    },
    {
      "epoch": 13.545126353790614,
      "grad_norm": 784374.0,
      "learning_rate": 0.00019177363584432265,
      "loss": 8.1136,
      "step": 112560
    },
    {
      "epoch": 13.54632972322503,
      "grad_norm": 310809.3125,
      "learning_rate": 0.0001917721247595961,
      "loss": 8.3211,
      "step": 112570
    },
    {
      "epoch": 13.547533092659446,
      "grad_norm": 709662.75,
      "learning_rate": 0.0001917706135420526,
      "loss": 8.2857,
      "step": 112580
    },
    {
      "epoch": 13.548736462093864,
      "grad_norm": 612764.25,
      "learning_rate": 0.0001917691021916944,
      "loss": 8.1536,
      "step": 112590
    },
    {
      "epoch": 13.54993983152828,
      "grad_norm": 1159474.0,
      "learning_rate": 0.00019176759070852362,
      "loss": 8.3407,
      "step": 112600
    },
    {
      "epoch": 13.551143200962695,
      "grad_norm": 1587835.375,
      "learning_rate": 0.0001917660790925425,
      "loss": 8.6622,
      "step": 112610
    },
    {
      "epoch": 13.552346570397113,
      "grad_norm": 470549.40625,
      "learning_rate": 0.0001917645673437532,
      "loss": 9.0315,
      "step": 112620
    },
    {
      "epoch": 13.553549939831528,
      "grad_norm": 335479.0,
      "learning_rate": 0.00019176305546215788,
      "loss": 9.6403,
      "step": 112630
    },
    {
      "epoch": 13.554753309265944,
      "grad_norm": 1567790.375,
      "learning_rate": 0.00019176154344775884,
      "loss": 9.6195,
      "step": 112640
    },
    {
      "epoch": 13.555956678700362,
      "grad_norm": 169770.359375,
      "learning_rate": 0.00019176003130055814,
      "loss": 8.9873,
      "step": 112650
    },
    {
      "epoch": 13.557160048134778,
      "grad_norm": 142464.234375,
      "learning_rate": 0.00019175851902055802,
      "loss": 8.1042,
      "step": 112660
    },
    {
      "epoch": 13.558363417569193,
      "grad_norm": 17675.6640625,
      "learning_rate": 0.00019175700660776067,
      "loss": 7.9578,
      "step": 112670
    },
    {
      "epoch": 13.559566787003611,
      "grad_norm": 162659.609375,
      "learning_rate": 0.0001917554940621683,
      "loss": 7.8636,
      "step": 112680
    },
    {
      "epoch": 13.560770156438027,
      "grad_norm": 26223.21875,
      "learning_rate": 0.00019175398138378307,
      "loss": 7.9636,
      "step": 112690
    },
    {
      "epoch": 13.561973525872443,
      "grad_norm": 198680.3125,
      "learning_rate": 0.00019175246857260715,
      "loss": 7.9908,
      "step": 112700
    },
    {
      "epoch": 13.56317689530686,
      "grad_norm": 181411.140625,
      "learning_rate": 0.00019175095562864279,
      "loss": 8.0474,
      "step": 112710
    },
    {
      "epoch": 13.564380264741276,
      "grad_norm": 142860.9375,
      "learning_rate": 0.00019174944255189212,
      "loss": 8.0895,
      "step": 112720
    },
    {
      "epoch": 13.565583634175692,
      "grad_norm": 298246.0625,
      "learning_rate": 0.00019174792934235736,
      "loss": 7.9235,
      "step": 112730
    },
    {
      "epoch": 13.566787003610107,
      "grad_norm": 609628.5625,
      "learning_rate": 0.00019174641600004072,
      "loss": 7.9122,
      "step": 112740
    },
    {
      "epoch": 13.567990373044525,
      "grad_norm": 172673.015625,
      "learning_rate": 0.0001917449025249443,
      "loss": 7.9096,
      "step": 112750
    },
    {
      "epoch": 13.56919374247894,
      "grad_norm": 156694.53125,
      "learning_rate": 0.00019174338891707042,
      "loss": 7.8797,
      "step": 112760
    },
    {
      "epoch": 13.570397111913357,
      "grad_norm": 398692.3125,
      "learning_rate": 0.00019174187517642118,
      "loss": 7.9444,
      "step": 112770
    },
    {
      "epoch": 13.571600481347774,
      "grad_norm": 148620.859375,
      "learning_rate": 0.0001917403613029988,
      "loss": 8.0473,
      "step": 112780
    },
    {
      "epoch": 13.57280385078219,
      "grad_norm": 147429.3125,
      "learning_rate": 0.00019173884729680544,
      "loss": 7.9269,
      "step": 112790
    },
    {
      "epoch": 13.574007220216606,
      "grad_norm": 72187.125,
      "learning_rate": 0.00019173733315784336,
      "loss": 7.9013,
      "step": 112800
    },
    {
      "epoch": 13.575210589651023,
      "grad_norm": 189529.171875,
      "learning_rate": 0.00019173581888611469,
      "loss": 7.886,
      "step": 112810
    },
    {
      "epoch": 13.57641395908544,
      "grad_norm": 381347.59375,
      "learning_rate": 0.00019173430448162163,
      "loss": 7.9838,
      "step": 112820
    },
    {
      "epoch": 13.577617328519855,
      "grad_norm": 395882.90625,
      "learning_rate": 0.00019173278994436642,
      "loss": 7.9936,
      "step": 112830
    },
    {
      "epoch": 13.578820697954272,
      "grad_norm": 143029.53125,
      "learning_rate": 0.0001917312752743512,
      "loss": 7.8788,
      "step": 112840
    },
    {
      "epoch": 13.580024067388688,
      "grad_norm": 146582.78125,
      "learning_rate": 0.00019172976047157815,
      "loss": 7.8896,
      "step": 112850
    },
    {
      "epoch": 13.581227436823104,
      "grad_norm": 513577.9375,
      "learning_rate": 0.0001917282455360495,
      "loss": 7.8231,
      "step": 112860
    },
    {
      "epoch": 13.582430806257522,
      "grad_norm": 313743.125,
      "learning_rate": 0.00019172673046776746,
      "loss": 7.9447,
      "step": 112870
    },
    {
      "epoch": 13.583634175691937,
      "grad_norm": 1286934.0,
      "learning_rate": 0.00019172521526673414,
      "loss": 7.9083,
      "step": 112880
    },
    {
      "epoch": 13.584837545126353,
      "grad_norm": 816960.625,
      "learning_rate": 0.00019172369993295183,
      "loss": 7.9728,
      "step": 112890
    },
    {
      "epoch": 13.58604091456077,
      "grad_norm": 271149.84375,
      "learning_rate": 0.00019172218446642268,
      "loss": 7.8854,
      "step": 112900
    },
    {
      "epoch": 13.587244283995187,
      "grad_norm": 416902.96875,
      "learning_rate": 0.00019172066886714888,
      "loss": 7.8649,
      "step": 112910
    },
    {
      "epoch": 13.588447653429602,
      "grad_norm": 338114.21875,
      "learning_rate": 0.0001917191531351326,
      "loss": 7.8158,
      "step": 112920
    },
    {
      "epoch": 13.58965102286402,
      "grad_norm": 249585.328125,
      "learning_rate": 0.0001917176372703761,
      "loss": 7.9567,
      "step": 112930
    },
    {
      "epoch": 13.590854392298436,
      "grad_norm": 621158.875,
      "learning_rate": 0.00019171612127288153,
      "loss": 7.9621,
      "step": 112940
    },
    {
      "epoch": 13.592057761732852,
      "grad_norm": 276559.34375,
      "learning_rate": 0.00019171460514265108,
      "loss": 8.0345,
      "step": 112950
    },
    {
      "epoch": 13.593261131167269,
      "grad_norm": 377018.6875,
      "learning_rate": 0.00019171308887968693,
      "loss": 8.1667,
      "step": 112960
    },
    {
      "epoch": 13.594464500601685,
      "grad_norm": 168727.6875,
      "learning_rate": 0.00019171157248399132,
      "loss": 7.9832,
      "step": 112970
    },
    {
      "epoch": 13.5956678700361,
      "grad_norm": 882711.375,
      "learning_rate": 0.00019171005595556643,
      "loss": 8.0051,
      "step": 112980
    },
    {
      "epoch": 13.596871239470518,
      "grad_norm": 329854.875,
      "learning_rate": 0.00019170853929441442,
      "loss": 7.9144,
      "step": 112990
    },
    {
      "epoch": 13.598074608904934,
      "grad_norm": 416692.65625,
      "learning_rate": 0.0001917070225005375,
      "loss": 7.8785,
      "step": 113000
    },
    {
      "epoch": 13.59927797833935,
      "grad_norm": 130376.7265625,
      "learning_rate": 0.00019170550557393793,
      "loss": 7.8839,
      "step": 113010
    },
    {
      "epoch": 13.600481347773766,
      "grad_norm": 184268.09375,
      "learning_rate": 0.0001917039885146178,
      "loss": 7.8423,
      "step": 113020
    },
    {
      "epoch": 13.601684717208183,
      "grad_norm": 384845.84375,
      "learning_rate": 0.0001917024713225794,
      "loss": 7.8866,
      "step": 113030
    },
    {
      "epoch": 13.602888086642599,
      "grad_norm": 130104.640625,
      "learning_rate": 0.00019170095399782485,
      "loss": 7.8785,
      "step": 113040
    },
    {
      "epoch": 13.604091456077015,
      "grad_norm": 120734.328125,
      "learning_rate": 0.00019169943654035642,
      "loss": 7.7937,
      "step": 113050
    },
    {
      "epoch": 13.605294825511432,
      "grad_norm": 255052.390625,
      "learning_rate": 0.00019169791895017624,
      "loss": 7.7194,
      "step": 113060
    },
    {
      "epoch": 13.606498194945848,
      "grad_norm": 124873.4765625,
      "learning_rate": 0.00019169640122728655,
      "loss": 7.7463,
      "step": 113070
    },
    {
      "epoch": 13.607701564380264,
      "grad_norm": 98920.671875,
      "learning_rate": 0.00019169488337168948,
      "loss": 7.8596,
      "step": 113080
    },
    {
      "epoch": 13.608904933814681,
      "grad_norm": 42414.5078125,
      "learning_rate": 0.0001916933653833873,
      "loss": 7.7009,
      "step": 113090
    },
    {
      "epoch": 13.610108303249097,
      "grad_norm": 14710.7763671875,
      "learning_rate": 0.0001916918472623822,
      "loss": 7.81,
      "step": 113100
    },
    {
      "epoch": 13.611311672683513,
      "grad_norm": 27084.69921875,
      "learning_rate": 0.00019169032900867637,
      "loss": 7.8299,
      "step": 113110
    },
    {
      "epoch": 13.61251504211793,
      "grad_norm": 9788.5009765625,
      "learning_rate": 0.00019168881062227194,
      "loss": 7.707,
      "step": 113120
    },
    {
      "epoch": 13.613718411552346,
      "grad_norm": 25628.39453125,
      "learning_rate": 0.0001916872921031712,
      "loss": 7.8188,
      "step": 113130
    },
    {
      "epoch": 13.614921780986762,
      "grad_norm": 134475.671875,
      "learning_rate": 0.0001916857734513763,
      "loss": 7.7533,
      "step": 113140
    },
    {
      "epoch": 13.61612515042118,
      "grad_norm": 11464.96484375,
      "learning_rate": 0.00019168425466688945,
      "loss": 7.7125,
      "step": 113150
    },
    {
      "epoch": 13.617328519855596,
      "grad_norm": 18534.837890625,
      "learning_rate": 0.0001916827357497129,
      "loss": 7.658,
      "step": 113160
    },
    {
      "epoch": 13.618531889290011,
      "grad_norm": 15738.4111328125,
      "learning_rate": 0.00019168121669984874,
      "loss": 7.6425,
      "step": 113170
    },
    {
      "epoch": 13.619735258724429,
      "grad_norm": 66036.796875,
      "learning_rate": 0.00019167969751729923,
      "loss": 7.6592,
      "step": 113180
    },
    {
      "epoch": 13.620938628158845,
      "grad_norm": 6695.29541015625,
      "learning_rate": 0.00019167817820206655,
      "loss": 7.4731,
      "step": 113190
    },
    {
      "epoch": 13.62214199759326,
      "grad_norm": 8706.462890625,
      "learning_rate": 0.00019167665875415292,
      "loss": 7.7305,
      "step": 113200
    },
    {
      "epoch": 13.623345367027678,
      "grad_norm": 17654.146484375,
      "learning_rate": 0.00019167513917356054,
      "loss": 7.6584,
      "step": 113210
    },
    {
      "epoch": 13.624548736462094,
      "grad_norm": 21089.666015625,
      "learning_rate": 0.00019167361946029163,
      "loss": 7.6102,
      "step": 113220
    },
    {
      "epoch": 13.62575210589651,
      "grad_norm": 37830.56640625,
      "learning_rate": 0.0001916720996143483,
      "loss": 7.6818,
      "step": 113230
    },
    {
      "epoch": 13.626955475330927,
      "grad_norm": 18966.365234375,
      "learning_rate": 0.00019167057963573284,
      "loss": 7.6295,
      "step": 113240
    },
    {
      "epoch": 13.628158844765343,
      "grad_norm": 14995.208984375,
      "learning_rate": 0.00019166905952444743,
      "loss": 7.6847,
      "step": 113250
    },
    {
      "epoch": 13.629362214199759,
      "grad_norm": 96365.71875,
      "learning_rate": 0.00019166753928049425,
      "loss": 7.6703,
      "step": 113260
    },
    {
      "epoch": 13.630565583634176,
      "grad_norm": 15477.8720703125,
      "learning_rate": 0.0001916660189038755,
      "loss": 7.6484,
      "step": 113270
    },
    {
      "epoch": 13.631768953068592,
      "grad_norm": 93922.109375,
      "learning_rate": 0.0001916644983945934,
      "loss": 7.611,
      "step": 113280
    },
    {
      "epoch": 13.632972322503008,
      "grad_norm": 79700.671875,
      "learning_rate": 0.0001916629777526501,
      "loss": 7.6818,
      "step": 113290
    },
    {
      "epoch": 13.634175691937426,
      "grad_norm": 35336.71875,
      "learning_rate": 0.00019166145697804788,
      "loss": 7.7733,
      "step": 113300
    },
    {
      "epoch": 13.635379061371841,
      "grad_norm": 45406.3515625,
      "learning_rate": 0.00019165993607078892,
      "loss": 7.7625,
      "step": 113310
    },
    {
      "epoch": 13.636582430806257,
      "grad_norm": 50273.1015625,
      "learning_rate": 0.00019165841503087535,
      "loss": 7.7031,
      "step": 113320
    },
    {
      "epoch": 13.637785800240675,
      "grad_norm": 22194.9296875,
      "learning_rate": 0.00019165689385830948,
      "loss": 7.6699,
      "step": 113330
    },
    {
      "epoch": 13.63898916967509,
      "grad_norm": 18881.408203125,
      "learning_rate": 0.0001916553725530934,
      "loss": 7.5607,
      "step": 113340
    },
    {
      "epoch": 13.640192539109506,
      "grad_norm": 8510.826171875,
      "learning_rate": 0.0001916538511152294,
      "loss": 7.6266,
      "step": 113350
    },
    {
      "epoch": 13.641395908543924,
      "grad_norm": 6942.92138671875,
      "learning_rate": 0.00019165232954471964,
      "loss": 7.5286,
      "step": 113360
    },
    {
      "epoch": 13.64259927797834,
      "grad_norm": 6047.9248046875,
      "learning_rate": 0.00019165080784156632,
      "loss": 7.4819,
      "step": 113370
    },
    {
      "epoch": 13.643802647412755,
      "grad_norm": 13872.33984375,
      "learning_rate": 0.00019164928600577167,
      "loss": 7.6924,
      "step": 113380
    },
    {
      "epoch": 13.645006016847173,
      "grad_norm": 23888.787109375,
      "learning_rate": 0.00019164776403733788,
      "loss": 7.7363,
      "step": 113390
    },
    {
      "epoch": 13.646209386281589,
      "grad_norm": 23308.720703125,
      "learning_rate": 0.00019164624193626712,
      "loss": 7.7116,
      "step": 113400
    },
    {
      "epoch": 13.647412755716005,
      "grad_norm": 2657.74462890625,
      "learning_rate": 0.00019164471970256166,
      "loss": 7.7341,
      "step": 113410
    },
    {
      "epoch": 13.648616125150422,
      "grad_norm": 520.08154296875,
      "learning_rate": 0.00019164319733622364,
      "loss": 7.6925,
      "step": 113420
    },
    {
      "epoch": 13.649819494584838,
      "grad_norm": 530.255126953125,
      "learning_rate": 0.00019164167483725527,
      "loss": 7.511,
      "step": 113430
    },
    {
      "epoch": 13.651022864019254,
      "grad_norm": 288.64007568359375,
      "learning_rate": 0.0001916401522056588,
      "loss": 7.5524,
      "step": 113440
    },
    {
      "epoch": 13.65222623345367,
      "grad_norm": 462.2852478027344,
      "learning_rate": 0.0001916386294414364,
      "loss": 7.4896,
      "step": 113450
    },
    {
      "epoch": 13.653429602888087,
      "grad_norm": 277.65899658203125,
      "learning_rate": 0.00019163710654459027,
      "loss": 7.5753,
      "step": 113460
    },
    {
      "epoch": 13.654632972322503,
      "grad_norm": 100.58748626708984,
      "learning_rate": 0.00019163558351512262,
      "loss": 7.4679,
      "step": 113470
    },
    {
      "epoch": 13.655836341756919,
      "grad_norm": 245.92759704589844,
      "learning_rate": 0.00019163406035303568,
      "loss": 7.5388,
      "step": 113480
    },
    {
      "epoch": 13.657039711191336,
      "grad_norm": 321.40283203125,
      "learning_rate": 0.00019163253705833163,
      "loss": 7.5197,
      "step": 113490
    },
    {
      "epoch": 13.658243080625752,
      "grad_norm": 293.3218994140625,
      "learning_rate": 0.00019163101363101266,
      "loss": 7.577,
      "step": 113500
    },
    {
      "epoch": 13.659446450060168,
      "grad_norm": 46.85881805419922,
      "learning_rate": 0.00019162949007108098,
      "loss": 7.5217,
      "step": 113510
    },
    {
      "epoch": 13.660649819494585,
      "grad_norm": 29.74823570251465,
      "learning_rate": 0.0001916279663785388,
      "loss": 7.6511,
      "step": 113520
    },
    {
      "epoch": 13.661853188929001,
      "grad_norm": 25.847501754760742,
      "learning_rate": 0.00019162644255338836,
      "loss": 7.6641,
      "step": 113530
    },
    {
      "epoch": 13.663056558363417,
      "grad_norm": 24.84147834777832,
      "learning_rate": 0.00019162491859563182,
      "loss": 7.5375,
      "step": 113540
    },
    {
      "epoch": 13.664259927797834,
      "grad_norm": 13.050982475280762,
      "learning_rate": 0.0001916233945052714,
      "loss": 7.6707,
      "step": 113550
    },
    {
      "epoch": 13.66546329723225,
      "grad_norm": 149.4770965576172,
      "learning_rate": 0.00019162187028230933,
      "loss": 7.5944,
      "step": 113560
    },
    {
      "epoch": 13.666666666666666,
      "grad_norm": 92.45903015136719,
      "learning_rate": 0.00019162034592674776,
      "loss": 7.6368,
      "step": 113570
    },
    {
      "epoch": 13.667870036101084,
      "grad_norm": 135.0058135986328,
      "learning_rate": 0.00019161882143858892,
      "loss": 7.681,
      "step": 113580
    },
    {
      "epoch": 13.6690734055355,
      "grad_norm": 29.964786529541016,
      "learning_rate": 0.00019161729681783506,
      "loss": 7.5944,
      "step": 113590
    },
    {
      "epoch": 13.670276774969915,
      "grad_norm": 65.10269165039062,
      "learning_rate": 0.00019161577206448836,
      "loss": 7.6076,
      "step": 113600
    },
    {
      "epoch": 13.671480144404333,
      "grad_norm": 21.363481521606445,
      "learning_rate": 0.00019161424717855098,
      "loss": 7.6193,
      "step": 113610
    },
    {
      "epoch": 13.672683513838749,
      "grad_norm": 11.367424964904785,
      "learning_rate": 0.0001916127221600252,
      "loss": 7.6415,
      "step": 113620
    },
    {
      "epoch": 13.673886883273164,
      "grad_norm": 32.594451904296875,
      "learning_rate": 0.00019161119700891318,
      "loss": 7.6179,
      "step": 113630
    },
    {
      "epoch": 13.675090252707582,
      "grad_norm": 61.654598236083984,
      "learning_rate": 0.00019160967172521712,
      "loss": 7.5527,
      "step": 113640
    },
    {
      "epoch": 13.676293622141998,
      "grad_norm": 33.25052261352539,
      "learning_rate": 0.00019160814630893926,
      "loss": 7.5598,
      "step": 113650
    },
    {
      "epoch": 13.677496991576414,
      "grad_norm": 22.977548599243164,
      "learning_rate": 0.0001916066207600818,
      "loss": 7.5235,
      "step": 113660
    },
    {
      "epoch": 13.678700361010831,
      "grad_norm": 18.656898498535156,
      "learning_rate": 0.00019160509507864693,
      "loss": 7.6432,
      "step": 113670
    },
    {
      "epoch": 13.679903730445247,
      "grad_norm": 61.93690872192383,
      "learning_rate": 0.00019160356926463688,
      "loss": 7.6145,
      "step": 113680
    },
    {
      "epoch": 13.681107099879663,
      "grad_norm": 121.2342300415039,
      "learning_rate": 0.00019160204331805387,
      "loss": 7.5778,
      "step": 113690
    },
    {
      "epoch": 13.68231046931408,
      "grad_norm": 40.884281158447266,
      "learning_rate": 0.00019160051723890006,
      "loss": 7.6093,
      "step": 113700
    },
    {
      "epoch": 13.683513838748496,
      "grad_norm": 108.30182647705078,
      "learning_rate": 0.0001915989910271777,
      "loss": 7.573,
      "step": 113710
    },
    {
      "epoch": 13.684717208182912,
      "grad_norm": 122.6958999633789,
      "learning_rate": 0.00019159746468288899,
      "loss": 7.5289,
      "step": 113720
    },
    {
      "epoch": 13.685920577617328,
      "grad_norm": 58.39677429199219,
      "learning_rate": 0.0001915959382060361,
      "loss": 7.629,
      "step": 113730
    },
    {
      "epoch": 13.687123947051745,
      "grad_norm": 35.375911712646484,
      "learning_rate": 0.00019159441159662132,
      "loss": 7.4844,
      "step": 113740
    },
    {
      "epoch": 13.688327316486161,
      "grad_norm": 45.86326217651367,
      "learning_rate": 0.00019159288485464678,
      "loss": 7.5992,
      "step": 113750
    },
    {
      "epoch": 13.689530685920577,
      "grad_norm": 835.2339477539062,
      "learning_rate": 0.00019159135798011475,
      "loss": 7.5361,
      "step": 113760
    },
    {
      "epoch": 13.690734055354994,
      "grad_norm": 714.7633666992188,
      "learning_rate": 0.00019158983097302738,
      "loss": 7.5489,
      "step": 113770
    },
    {
      "epoch": 13.69193742478941,
      "grad_norm": 3187.551025390625,
      "learning_rate": 0.00019158830383338694,
      "loss": 7.4862,
      "step": 113780
    },
    {
      "epoch": 13.693140794223826,
      "grad_norm": 2072.183349609375,
      "learning_rate": 0.00019158677656119557,
      "loss": 7.5505,
      "step": 113790
    },
    {
      "epoch": 13.694344163658243,
      "grad_norm": 1155.9842529296875,
      "learning_rate": 0.00019158524915645556,
      "loss": 7.4884,
      "step": 113800
    },
    {
      "epoch": 13.69554753309266,
      "grad_norm": 8257.2919921875,
      "learning_rate": 0.00019158372161916907,
      "loss": 7.5835,
      "step": 113810
    },
    {
      "epoch": 13.696750902527075,
      "grad_norm": 3663.29638671875,
      "learning_rate": 0.00019158219394933834,
      "loss": 7.5884,
      "step": 113820
    },
    {
      "epoch": 13.697954271961493,
      "grad_norm": 2061.91650390625,
      "learning_rate": 0.00019158066614696556,
      "loss": 7.5723,
      "step": 113830
    },
    {
      "epoch": 13.699157641395908,
      "grad_norm": 1974.542724609375,
      "learning_rate": 0.0001915791382120529,
      "loss": 7.561,
      "step": 113840
    },
    {
      "epoch": 13.700361010830324,
      "grad_norm": 1014.097412109375,
      "learning_rate": 0.00019157761014460266,
      "loss": 7.5404,
      "step": 113850
    },
    {
      "epoch": 13.701564380264742,
      "grad_norm": 5040.04150390625,
      "learning_rate": 0.00019157608194461702,
      "loss": 7.5522,
      "step": 113860
    },
    {
      "epoch": 13.702767749699158,
      "grad_norm": 4629.54736328125,
      "learning_rate": 0.00019157455361209816,
      "loss": 7.5626,
      "step": 113870
    },
    {
      "epoch": 13.703971119133573,
      "grad_norm": 3905.244140625,
      "learning_rate": 0.00019157302514704832,
      "loss": 7.5662,
      "step": 113880
    },
    {
      "epoch": 13.705174488567991,
      "grad_norm": 2505.708251953125,
      "learning_rate": 0.00019157149654946967,
      "loss": 7.4498,
      "step": 113890
    },
    {
      "epoch": 13.706377858002407,
      "grad_norm": 2081.571044921875,
      "learning_rate": 0.0001915699678193645,
      "loss": 7.6297,
      "step": 113900
    },
    {
      "epoch": 13.707581227436823,
      "grad_norm": 5621.26806640625,
      "learning_rate": 0.00019156843895673498,
      "loss": 7.5617,
      "step": 113910
    },
    {
      "epoch": 13.70878459687124,
      "grad_norm": 2439.014404296875,
      "learning_rate": 0.00019156690996158327,
      "loss": 7.5824,
      "step": 113920
    },
    {
      "epoch": 13.709987966305656,
      "grad_norm": 289.02557373046875,
      "learning_rate": 0.0001915653808339117,
      "loss": 7.561,
      "step": 113930
    },
    {
      "epoch": 13.711191335740072,
      "grad_norm": 971.802978515625,
      "learning_rate": 0.00019156385157372235,
      "loss": 7.5163,
      "step": 113940
    },
    {
      "epoch": 13.71239470517449,
      "grad_norm": 1331.414306640625,
      "learning_rate": 0.00019156232218101756,
      "loss": 7.4725,
      "step": 113950
    },
    {
      "epoch": 13.713598074608905,
      "grad_norm": 449.71319580078125,
      "learning_rate": 0.00019156079265579944,
      "loss": 7.5403,
      "step": 113960
    },
    {
      "epoch": 13.71480144404332,
      "grad_norm": 2657.580322265625,
      "learning_rate": 0.00019155926299807026,
      "loss": 7.521,
      "step": 113970
    },
    {
      "epoch": 13.716004813477738,
      "grad_norm": 2584.49560546875,
      "learning_rate": 0.00019155773320783223,
      "loss": 7.6036,
      "step": 113980
    },
    {
      "epoch": 13.717208182912154,
      "grad_norm": 381.4786376953125,
      "learning_rate": 0.00019155620328508756,
      "loss": 7.4513,
      "step": 113990
    },
    {
      "epoch": 13.71841155234657,
      "grad_norm": 219.59832763671875,
      "learning_rate": 0.00019155467322983847,
      "loss": 7.5127,
      "step": 114000
    },
    {
      "epoch": 13.719614921780988,
      "grad_norm": 1460.5084228515625,
      "learning_rate": 0.00019155314304208711,
      "loss": 7.621,
      "step": 114010
    },
    {
      "epoch": 13.720818291215403,
      "grad_norm": 521.77099609375,
      "learning_rate": 0.0001915516127218358,
      "loss": 7.5442,
      "step": 114020
    },
    {
      "epoch": 13.722021660649819,
      "grad_norm": 272.19732666015625,
      "learning_rate": 0.0001915500822690867,
      "loss": 7.5867,
      "step": 114030
    },
    {
      "epoch": 13.723225030084237,
      "grad_norm": 953.1600341796875,
      "learning_rate": 0.000191548551683842,
      "loss": 7.5439,
      "step": 114040
    },
    {
      "epoch": 13.724428399518652,
      "grad_norm": 1730.5391845703125,
      "learning_rate": 0.00019154702096610396,
      "loss": 7.5285,
      "step": 114050
    },
    {
      "epoch": 13.725631768953068,
      "grad_norm": 3452.368896484375,
      "learning_rate": 0.00019154549011587477,
      "loss": 7.5289,
      "step": 114060
    },
    {
      "epoch": 13.726835138387486,
      "grad_norm": 1592.57275390625,
      "learning_rate": 0.00019154395913315665,
      "loss": 7.5607,
      "step": 114070
    },
    {
      "epoch": 13.728038507821902,
      "grad_norm": 3693.570068359375,
      "learning_rate": 0.00019154242801795187,
      "loss": 7.6056,
      "step": 114080
    },
    {
      "epoch": 13.729241877256317,
      "grad_norm": 3637.03564453125,
      "learning_rate": 0.00019154089677026254,
      "loss": 7.5423,
      "step": 114090
    },
    {
      "epoch": 13.730445246690735,
      "grad_norm": 1049.1793212890625,
      "learning_rate": 0.00019153936539009094,
      "loss": 7.5813,
      "step": 114100
    },
    {
      "epoch": 13.73164861612515,
      "grad_norm": 2903.21484375,
      "learning_rate": 0.0001915378338774393,
      "loss": 7.5154,
      "step": 114110
    },
    {
      "epoch": 13.732851985559567,
      "grad_norm": 3569.20703125,
      "learning_rate": 0.00019153630223230978,
      "loss": 7.558,
      "step": 114120
    },
    {
      "epoch": 13.734055354993982,
      "grad_norm": 2672.241455078125,
      "learning_rate": 0.00019153477045470466,
      "loss": 7.544,
      "step": 114130
    },
    {
      "epoch": 13.7352587244284,
      "grad_norm": 991.072509765625,
      "learning_rate": 0.0001915332385446261,
      "loss": 7.6041,
      "step": 114140
    },
    {
      "epoch": 13.736462093862816,
      "grad_norm": 881.470947265625,
      "learning_rate": 0.0001915317065020764,
      "loss": 7.4928,
      "step": 114150
    },
    {
      "epoch": 13.737665463297231,
      "grad_norm": 504.5856018066406,
      "learning_rate": 0.00019153017432705768,
      "loss": 7.6152,
      "step": 114160
    },
    {
      "epoch": 13.738868832731649,
      "grad_norm": 155.9636688232422,
      "learning_rate": 0.0001915286420195722,
      "loss": 7.6489,
      "step": 114170
    },
    {
      "epoch": 13.740072202166065,
      "grad_norm": 163597.453125,
      "learning_rate": 0.00019152710957962216,
      "loss": 7.5399,
      "step": 114180
    },
    {
      "epoch": 13.74127557160048,
      "grad_norm": 1000418.0,
      "learning_rate": 0.00019152557700720984,
      "loss": 7.5673,
      "step": 114190
    },
    {
      "epoch": 13.742478941034898,
      "grad_norm": 664559.9375,
      "learning_rate": 0.0001915240443023374,
      "loss": 7.5518,
      "step": 114200
    },
    {
      "epoch": 13.743682310469314,
      "grad_norm": 857.570068359375,
      "learning_rate": 0.00019152251146500703,
      "loss": 7.6212,
      "step": 114210
    },
    {
      "epoch": 13.74488567990373,
      "grad_norm": 463.45947265625,
      "learning_rate": 0.00019152097849522103,
      "loss": 7.5444,
      "step": 114220
    },
    {
      "epoch": 13.746089049338147,
      "grad_norm": 629.0192260742188,
      "learning_rate": 0.00019151944539298158,
      "loss": 7.5369,
      "step": 114230
    },
    {
      "epoch": 13.747292418772563,
      "grad_norm": 354.37371826171875,
      "learning_rate": 0.0001915179121582909,
      "loss": 7.5589,
      "step": 114240
    },
    {
      "epoch": 13.748495788206979,
      "grad_norm": 378.7722473144531,
      "learning_rate": 0.0001915163787911512,
      "loss": 7.5461,
      "step": 114250
    },
    {
      "epoch": 13.749699157641396,
      "grad_norm": 288.4242248535156,
      "learning_rate": 0.00019151484529156466,
      "loss": 7.4922,
      "step": 114260
    },
    {
      "epoch": 13.750902527075812,
      "grad_norm": 1222.2769775390625,
      "learning_rate": 0.0001915133116595336,
      "loss": 7.6345,
      "step": 114270
    },
    {
      "epoch": 13.752105896510228,
      "grad_norm": 1110.3900146484375,
      "learning_rate": 0.00019151177789506016,
      "loss": 7.4965,
      "step": 114280
    },
    {
      "epoch": 13.753309265944646,
      "grad_norm": 1740.095947265625,
      "learning_rate": 0.0001915102439981466,
      "loss": 7.6424,
      "step": 114290
    },
    {
      "epoch": 13.754512635379061,
      "grad_norm": 2457.517822265625,
      "learning_rate": 0.0001915087099687951,
      "loss": 7.479,
      "step": 114300
    },
    {
      "epoch": 13.755716004813477,
      "grad_norm": 521.0994262695312,
      "learning_rate": 0.00019150717580700794,
      "loss": 7.5619,
      "step": 114310
    },
    {
      "epoch": 13.756919374247895,
      "grad_norm": 742.9874877929688,
      "learning_rate": 0.00019150564151278727,
      "loss": 7.5571,
      "step": 114320
    },
    {
      "epoch": 13.75812274368231,
      "grad_norm": 497.9338684082031,
      "learning_rate": 0.00019150410708613537,
      "loss": 7.6108,
      "step": 114330
    },
    {
      "epoch": 13.759326113116726,
      "grad_norm": 247.5394744873047,
      "learning_rate": 0.0001915025725270544,
      "loss": 7.5334,
      "step": 114340
    },
    {
      "epoch": 13.760529482551144,
      "grad_norm": 554.1671752929688,
      "learning_rate": 0.00019150103783554666,
      "loss": 7.5137,
      "step": 114350
    },
    {
      "epoch": 13.76173285198556,
      "grad_norm": 335.91912841796875,
      "learning_rate": 0.0001914995030116143,
      "loss": 7.5443,
      "step": 114360
    },
    {
      "epoch": 13.762936221419976,
      "grad_norm": 219.8412628173828,
      "learning_rate": 0.0001914979680552596,
      "loss": 7.5378,
      "step": 114370
    },
    {
      "epoch": 13.764139590854393,
      "grad_norm": 204.30767822265625,
      "learning_rate": 0.0001914964329664847,
      "loss": 7.6104,
      "step": 114380
    },
    {
      "epoch": 13.765342960288809,
      "grad_norm": 1333.4427490234375,
      "learning_rate": 0.0001914948977452919,
      "loss": 7.5312,
      "step": 114390
    },
    {
      "epoch": 13.766546329723225,
      "grad_norm": 1243.517822265625,
      "learning_rate": 0.00019149336239168338,
      "loss": 7.4446,
      "step": 114400
    },
    {
      "epoch": 13.76774969915764,
      "grad_norm": 274.4715881347656,
      "learning_rate": 0.00019149182690566142,
      "loss": 7.5787,
      "step": 114410
    },
    {
      "epoch": 13.768953068592058,
      "grad_norm": 319.8570251464844,
      "learning_rate": 0.00019149029128722814,
      "loss": 7.5531,
      "step": 114420
    },
    {
      "epoch": 13.770156438026474,
      "grad_norm": 184.08749389648438,
      "learning_rate": 0.00019148875553638586,
      "loss": 7.6061,
      "step": 114430
    },
    {
      "epoch": 13.77135980746089,
      "grad_norm": 981.418212890625,
      "learning_rate": 0.00019148721965313675,
      "loss": 7.555,
      "step": 114440
    },
    {
      "epoch": 13.772563176895307,
      "grad_norm": 1492.6351318359375,
      "learning_rate": 0.00019148568363748302,
      "loss": 7.6056,
      "step": 114450
    },
    {
      "epoch": 13.773766546329723,
      "grad_norm": 186.94186401367188,
      "learning_rate": 0.00019148414748942696,
      "loss": 7.5459,
      "step": 114460
    },
    {
      "epoch": 13.774969915764139,
      "grad_norm": 2787.349365234375,
      "learning_rate": 0.00019148261120897073,
      "loss": 7.4659,
      "step": 114470
    },
    {
      "epoch": 13.776173285198556,
      "grad_norm": 651.6773681640625,
      "learning_rate": 0.00019148107479611656,
      "loss": 7.5889,
      "step": 114480
    },
    {
      "epoch": 13.777376654632972,
      "grad_norm": 680.3724975585938,
      "learning_rate": 0.00019147953825086674,
      "loss": 7.6009,
      "step": 114490
    },
    {
      "epoch": 13.778580024067388,
      "grad_norm": 5606.00048828125,
      "learning_rate": 0.0001914780015732234,
      "loss": 7.6182,
      "step": 114500
    },
    {
      "epoch": 13.779783393501805,
      "grad_norm": 10631.271484375,
      "learning_rate": 0.0001914764647631888,
      "loss": 7.5175,
      "step": 114510
    },
    {
      "epoch": 13.780986762936221,
      "grad_norm": 1343.87890625,
      "learning_rate": 0.00019147492782076522,
      "loss": 7.5946,
      "step": 114520
    },
    {
      "epoch": 13.782190132370637,
      "grad_norm": 9865.0048828125,
      "learning_rate": 0.00019147339074595477,
      "loss": 7.5153,
      "step": 114530
    },
    {
      "epoch": 13.783393501805055,
      "grad_norm": 3485.634033203125,
      "learning_rate": 0.0001914718535387598,
      "loss": 7.5806,
      "step": 114540
    },
    {
      "epoch": 13.78459687123947,
      "grad_norm": 2424.786376953125,
      "learning_rate": 0.00019147031619918243,
      "loss": 7.5224,
      "step": 114550
    },
    {
      "epoch": 13.785800240673886,
      "grad_norm": 1291.46875,
      "learning_rate": 0.00019146877872722496,
      "loss": 7.4227,
      "step": 114560
    },
    {
      "epoch": 13.787003610108304,
      "grad_norm": 2239.263427734375,
      "learning_rate": 0.00019146724112288957,
      "loss": 7.5755,
      "step": 114570
    },
    {
      "epoch": 13.78820697954272,
      "grad_norm": 20331.4296875,
      "learning_rate": 0.0001914657033861785,
      "loss": 7.5735,
      "step": 114580
    },
    {
      "epoch": 13.789410348977135,
      "grad_norm": 10993.927734375,
      "learning_rate": 0.00019146416551709398,
      "loss": 7.5049,
      "step": 114590
    },
    {
      "epoch": 13.790613718411553,
      "grad_norm": 10267.681640625,
      "learning_rate": 0.0001914626275156382,
      "loss": 7.6153,
      "step": 114600
    },
    {
      "epoch": 13.791817087845969,
      "grad_norm": 21958.6171875,
      "learning_rate": 0.00019146108938181345,
      "loss": 7.562,
      "step": 114610
    },
    {
      "epoch": 13.793020457280385,
      "grad_norm": 8434.3837890625,
      "learning_rate": 0.0001914595511156219,
      "loss": 7.4472,
      "step": 114620
    },
    {
      "epoch": 13.794223826714802,
      "grad_norm": 2945.95263671875,
      "learning_rate": 0.00019145801271706585,
      "loss": 7.4919,
      "step": 114630
    },
    {
      "epoch": 13.795427196149218,
      "grad_norm": 1485.851318359375,
      "learning_rate": 0.00019145647418614744,
      "loss": 7.5163,
      "step": 114640
    },
    {
      "epoch": 13.796630565583634,
      "grad_norm": 897.9228515625,
      "learning_rate": 0.00019145493552286892,
      "loss": 7.4508,
      "step": 114650
    },
    {
      "epoch": 13.797833935018051,
      "grad_norm": 2290.776611328125,
      "learning_rate": 0.00019145339672723254,
      "loss": 7.5969,
      "step": 114660
    },
    {
      "epoch": 13.799037304452467,
      "grad_norm": 709.6951904296875,
      "learning_rate": 0.0001914518577992405,
      "loss": 7.4705,
      "step": 114670
    },
    {
      "epoch": 13.800240673886883,
      "grad_norm": 2647.744384765625,
      "learning_rate": 0.00019145031873889508,
      "loss": 7.6015,
      "step": 114680
    },
    {
      "epoch": 13.8014440433213,
      "grad_norm": 4433.423828125,
      "learning_rate": 0.00019144877954619844,
      "loss": 7.5197,
      "step": 114690
    },
    {
      "epoch": 13.802647412755716,
      "grad_norm": 992.7537231445312,
      "learning_rate": 0.00019144724022115286,
      "loss": 7.5472,
      "step": 114700
    },
    {
      "epoch": 13.803850782190132,
      "grad_norm": 3379.71044921875,
      "learning_rate": 0.00019144570076376052,
      "loss": 7.7093,
      "step": 114710
    },
    {
      "epoch": 13.80505415162455,
      "grad_norm": 2955.573974609375,
      "learning_rate": 0.00019144416117402368,
      "loss": 7.4811,
      "step": 114720
    },
    {
      "epoch": 13.806257521058965,
      "grad_norm": 1883.2606201171875,
      "learning_rate": 0.00019144262145194457,
      "loss": 7.591,
      "step": 114730
    },
    {
      "epoch": 13.807460890493381,
      "grad_norm": 1220.7481689453125,
      "learning_rate": 0.00019144108159752543,
      "loss": 7.4673,
      "step": 114740
    },
    {
      "epoch": 13.808664259927799,
      "grad_norm": 714.267333984375,
      "learning_rate": 0.00019143954161076842,
      "loss": 7.5356,
      "step": 114750
    },
    {
      "epoch": 13.809867629362214,
      "grad_norm": 1203.8299560546875,
      "learning_rate": 0.00019143800149167586,
      "loss": 7.6293,
      "step": 114760
    },
    {
      "epoch": 13.81107099879663,
      "grad_norm": 352.29852294921875,
      "learning_rate": 0.00019143646124024992,
      "loss": 7.575,
      "step": 114770
    },
    {
      "epoch": 13.812274368231048,
      "grad_norm": 176.0352020263672,
      "learning_rate": 0.00019143492085649285,
      "loss": 7.5293,
      "step": 114780
    },
    {
      "epoch": 13.813477737665464,
      "grad_norm": 75.21821594238281,
      "learning_rate": 0.00019143338034040687,
      "loss": 7.5387,
      "step": 114790
    },
    {
      "epoch": 13.81468110709988,
      "grad_norm": 54.11989212036133,
      "learning_rate": 0.0001914318396919942,
      "loss": 7.5445,
      "step": 114800
    },
    {
      "epoch": 13.815884476534297,
      "grad_norm": 66.36608123779297,
      "learning_rate": 0.00019143029891125713,
      "loss": 7.5884,
      "step": 114810
    },
    {
      "epoch": 13.817087845968713,
      "grad_norm": 59.86701965332031,
      "learning_rate": 0.0001914287579981978,
      "loss": 7.5834,
      "step": 114820
    },
    {
      "epoch": 13.818291215403129,
      "grad_norm": 149.61434936523438,
      "learning_rate": 0.00019142721695281849,
      "loss": 7.6047,
      "step": 114830
    },
    {
      "epoch": 13.819494584837544,
      "grad_norm": 139.30702209472656,
      "learning_rate": 0.00019142567577512144,
      "loss": 7.5179,
      "step": 114840
    },
    {
      "epoch": 13.820697954271962,
      "grad_norm": 547.0332641601562,
      "learning_rate": 0.00019142413446510888,
      "loss": 7.446,
      "step": 114850
    },
    {
      "epoch": 13.821901323706378,
      "grad_norm": 131.2542266845703,
      "learning_rate": 0.000191422593022783,
      "loss": 7.5526,
      "step": 114860
    },
    {
      "epoch": 13.823104693140793,
      "grad_norm": 49.47675323486328,
      "learning_rate": 0.00019142105144814603,
      "loss": 7.5435,
      "step": 114870
    },
    {
      "epoch": 13.824308062575211,
      "grad_norm": 311121.65625,
      "learning_rate": 0.00019141950974120028,
      "loss": 7.5234,
      "step": 114880
    },
    {
      "epoch": 13.825511432009627,
      "grad_norm": 1522005.75,
      "learning_rate": 0.0001914179679019479,
      "loss": 7.5947,
      "step": 114890
    },
    {
      "epoch": 13.826714801444043,
      "grad_norm": 277930.375,
      "learning_rate": 0.00019141642593039116,
      "loss": 7.7185,
      "step": 114900
    },
    {
      "epoch": 13.82791817087846,
      "grad_norm": 365228.09375,
      "learning_rate": 0.00019141488382653227,
      "loss": 7.6736,
      "step": 114910
    },
    {
      "epoch": 13.829121540312876,
      "grad_norm": 777802.75,
      "learning_rate": 0.0001914133415903735,
      "loss": 7.6472,
      "step": 114920
    },
    {
      "epoch": 13.830324909747292,
      "grad_norm": 40078.71875,
      "learning_rate": 0.00019141179922191703,
      "loss": 7.7741,
      "step": 114930
    },
    {
      "epoch": 13.83152827918171,
      "grad_norm": 392143.34375,
      "learning_rate": 0.00019141025672116512,
      "loss": 7.701,
      "step": 114940
    },
    {
      "epoch": 13.832731648616125,
      "grad_norm": 375842.65625,
      "learning_rate": 0.00019140871408812,
      "loss": 7.7547,
      "step": 114950
    },
    {
      "epoch": 13.833935018050541,
      "grad_norm": 1742897.0,
      "learning_rate": 0.00019140717132278391,
      "loss": 7.6706,
      "step": 114960
    },
    {
      "epoch": 13.835138387484958,
      "grad_norm": 363935.90625,
      "learning_rate": 0.00019140562842515907,
      "loss": 7.7104,
      "step": 114970
    },
    {
      "epoch": 13.836341756919374,
      "grad_norm": 650831.125,
      "learning_rate": 0.00019140408539524773,
      "loss": 7.7497,
      "step": 114980
    },
    {
      "epoch": 13.83754512635379,
      "grad_norm": 386689.5,
      "learning_rate": 0.0001914025422330521,
      "loss": 7.5964,
      "step": 114990
    },
    {
      "epoch": 13.838748495788208,
      "grad_norm": 383262.0,
      "learning_rate": 0.00019140099893857443,
      "loss": 7.6103,
      "step": 115000
    },
    {
      "epoch": 13.839951865222623,
      "grad_norm": 1188258.5,
      "learning_rate": 0.00019139945551181696,
      "loss": 7.7531,
      "step": 115010
    },
    {
      "epoch": 13.84115523465704,
      "grad_norm": 383898.875,
      "learning_rate": 0.0001913979119527819,
      "loss": 7.7964,
      "step": 115020
    },
    {
      "epoch": 13.842358604091457,
      "grad_norm": 1139049.25,
      "learning_rate": 0.00019139636826147148,
      "loss": 7.7614,
      "step": 115030
    },
    {
      "epoch": 13.843561973525873,
      "grad_norm": 1116222.25,
      "learning_rate": 0.000191394824437888,
      "loss": 7.6898,
      "step": 115040
    },
    {
      "epoch": 13.844765342960288,
      "grad_norm": 327987.90625,
      "learning_rate": 0.00019139328048203358,
      "loss": 7.6877,
      "step": 115050
    },
    {
      "epoch": 13.845968712394706,
      "grad_norm": 590856.0625,
      "learning_rate": 0.00019139173639391054,
      "loss": 7.6416,
      "step": 115060
    },
    {
      "epoch": 13.847172081829122,
      "grad_norm": 109.004638671875,
      "learning_rate": 0.0001913901921735211,
      "loss": 7.6984,
      "step": 115070
    },
    {
      "epoch": 13.848375451263538,
      "grad_norm": 123.90349578857422,
      "learning_rate": 0.0001913886478208675,
      "loss": 7.5357,
      "step": 115080
    },
    {
      "epoch": 13.849578820697955,
      "grad_norm": 151.775634765625,
      "learning_rate": 0.00019138710333595197,
      "loss": 7.5502,
      "step": 115090
    },
    {
      "epoch": 13.85078219013237,
      "grad_norm": 162.43109130859375,
      "learning_rate": 0.0001913855587187767,
      "loss": 7.56,
      "step": 115100
    },
    {
      "epoch": 13.851985559566787,
      "grad_norm": 177.98968505859375,
      "learning_rate": 0.000191384013969344,
      "loss": 7.571,
      "step": 115110
    },
    {
      "epoch": 13.853188929001202,
      "grad_norm": 435.82769775390625,
      "learning_rate": 0.00019138246908765604,
      "loss": 7.4176,
      "step": 115120
    },
    {
      "epoch": 13.85439229843562,
      "grad_norm": 114.03434753417969,
      "learning_rate": 0.00019138092407371512,
      "loss": 7.6695,
      "step": 115130
    },
    {
      "epoch": 13.855595667870036,
      "grad_norm": 912.5336303710938,
      "learning_rate": 0.00019137937892752343,
      "loss": 7.6092,
      "step": 115140
    },
    {
      "epoch": 13.856799037304452,
      "grad_norm": 618.1643676757812,
      "learning_rate": 0.0001913778336490832,
      "loss": 7.5304,
      "step": 115150
    },
    {
      "epoch": 13.85800240673887,
      "grad_norm": 1248.12353515625,
      "learning_rate": 0.00019137628823839668,
      "loss": 7.5448,
      "step": 115160
    },
    {
      "epoch": 13.859205776173285,
      "grad_norm": 2512.384521484375,
      "learning_rate": 0.00019137474269546612,
      "loss": 7.5656,
      "step": 115170
    },
    {
      "epoch": 13.8604091456077,
      "grad_norm": 1786.19482421875,
      "learning_rate": 0.00019137319702029377,
      "loss": 7.5498,
      "step": 115180
    },
    {
      "epoch": 13.861612515042118,
      "grad_norm": 3869.5947265625,
      "learning_rate": 0.00019137165121288182,
      "loss": 7.4658,
      "step": 115190
    },
    {
      "epoch": 13.862815884476534,
      "grad_norm": 1402.50390625,
      "learning_rate": 0.00019137010527323255,
      "loss": 7.6734,
      "step": 115200
    },
    {
      "epoch": 13.86401925391095,
      "grad_norm": 3173.96923828125,
      "learning_rate": 0.00019136855920134815,
      "loss": 7.5028,
      "step": 115210
    },
    {
      "epoch": 13.865222623345367,
      "grad_norm": 2175.4541015625,
      "learning_rate": 0.0001913670129972309,
      "loss": 7.5096,
      "step": 115220
    },
    {
      "epoch": 13.866425992779783,
      "grad_norm": 4703.65869140625,
      "learning_rate": 0.00019136546666088307,
      "loss": 7.4453,
      "step": 115230
    },
    {
      "epoch": 13.867629362214199,
      "grad_norm": 2822.893798828125,
      "learning_rate": 0.0001913639201923068,
      "loss": 7.5953,
      "step": 115240
    },
    {
      "epoch": 13.868832731648617,
      "grad_norm": 1708.392333984375,
      "learning_rate": 0.0001913623735915044,
      "loss": 7.5144,
      "step": 115250
    },
    {
      "epoch": 13.870036101083032,
      "grad_norm": 362.4938659667969,
      "learning_rate": 0.00019136082685847807,
      "loss": 7.6342,
      "step": 115260
    },
    {
      "epoch": 13.871239470517448,
      "grad_norm": 78.56669616699219,
      "learning_rate": 0.0001913592799932301,
      "loss": 7.5313,
      "step": 115270
    },
    {
      "epoch": 13.872442839951866,
      "grad_norm": 878.3430786132812,
      "learning_rate": 0.00019135773299576268,
      "loss": 7.5766,
      "step": 115280
    },
    {
      "epoch": 13.873646209386282,
      "grad_norm": 604.4127197265625,
      "learning_rate": 0.00019135618586607804,
      "loss": 7.4555,
      "step": 115290
    },
    {
      "epoch": 13.874849578820697,
      "grad_norm": 322.37274169921875,
      "learning_rate": 0.00019135463860417846,
      "loss": 7.4844,
      "step": 115300
    },
    {
      "epoch": 13.876052948255115,
      "grad_norm": 1616.5966796875,
      "learning_rate": 0.00019135309121006617,
      "loss": 7.5152,
      "step": 115310
    },
    {
      "epoch": 13.87725631768953,
      "grad_norm": 684.5773315429688,
      "learning_rate": 0.0001913515436837434,
      "loss": 7.5278,
      "step": 115320
    },
    {
      "epoch": 13.878459687123947,
      "grad_norm": 454.45574951171875,
      "learning_rate": 0.0001913499960252124,
      "loss": 7.5819,
      "step": 115330
    },
    {
      "epoch": 13.879663056558364,
      "grad_norm": 1045.1630859375,
      "learning_rate": 0.0001913484482344754,
      "loss": 7.6103,
      "step": 115340
    },
    {
      "epoch": 13.88086642599278,
      "grad_norm": 768.3417358398438,
      "learning_rate": 0.00019134690031153464,
      "loss": 7.5073,
      "step": 115350
    },
    {
      "epoch": 13.882069795427196,
      "grad_norm": 1216.54150390625,
      "learning_rate": 0.00019134535225639234,
      "loss": 7.65,
      "step": 115360
    },
    {
      "epoch": 13.883273164861613,
      "grad_norm": 2441.576171875,
      "learning_rate": 0.00019134380406905078,
      "loss": 7.5967,
      "step": 115370
    },
    {
      "epoch": 13.884476534296029,
      "grad_norm": 726.3614501953125,
      "learning_rate": 0.0001913422557495122,
      "loss": 7.6242,
      "step": 115380
    },
    {
      "epoch": 13.885679903730445,
      "grad_norm": 946.9081420898438,
      "learning_rate": 0.0001913407072977788,
      "loss": 7.4688,
      "step": 115390
    },
    {
      "epoch": 13.886883273164862,
      "grad_norm": 415.5739440917969,
      "learning_rate": 0.00019133915871385283,
      "loss": 7.5077,
      "step": 115400
    },
    {
      "epoch": 13.888086642599278,
      "grad_norm": 1068.4891357421875,
      "learning_rate": 0.0001913376099977366,
      "loss": 7.4925,
      "step": 115410
    },
    {
      "epoch": 13.889290012033694,
      "grad_norm": 582.84326171875,
      "learning_rate": 0.00019133606114943226,
      "loss": 7.4997,
      "step": 115420
    },
    {
      "epoch": 13.890493381468112,
      "grad_norm": 614.4970703125,
      "learning_rate": 0.00019133451216894209,
      "loss": 7.5631,
      "step": 115430
    },
    {
      "epoch": 13.891696750902527,
      "grad_norm": 226.0276336669922,
      "learning_rate": 0.00019133296305626836,
      "loss": 7.4974,
      "step": 115440
    },
    {
      "epoch": 13.892900120336943,
      "grad_norm": 241.5906982421875,
      "learning_rate": 0.00019133141381141324,
      "loss": 7.5471,
      "step": 115450
    },
    {
      "epoch": 13.89410348977136,
      "grad_norm": 314.6260070800781,
      "learning_rate": 0.00019132986443437907,
      "loss": 7.3999,
      "step": 115460
    },
    {
      "epoch": 13.895306859205776,
      "grad_norm": 388.9868469238281,
      "learning_rate": 0.000191328314925168,
      "loss": 7.3893,
      "step": 115470
    },
    {
      "epoch": 13.896510228640192,
      "grad_norm": 722.2169189453125,
      "learning_rate": 0.0001913267652837823,
      "loss": 7.3604,
      "step": 115480
    },
    {
      "epoch": 13.89771359807461,
      "grad_norm": 557.0263061523438,
      "learning_rate": 0.00019132521551022425,
      "loss": 7.3636,
      "step": 115490
    },
    {
      "epoch": 13.898916967509026,
      "grad_norm": 252.8494873046875,
      "learning_rate": 0.00019132366560449605,
      "loss": 7.4054,
      "step": 115500
    },
    {
      "epoch": 13.900120336943441,
      "grad_norm": 1011.9021606445312,
      "learning_rate": 0.00019132211556659995,
      "loss": 7.3604,
      "step": 115510
    },
    {
      "epoch": 13.901323706377857,
      "grad_norm": 579.222412109375,
      "learning_rate": 0.0001913205653965382,
      "loss": 7.3747,
      "step": 115520
    },
    {
      "epoch": 13.902527075812275,
      "grad_norm": 1107.37353515625,
      "learning_rate": 0.00019131901509431308,
      "loss": 7.3347,
      "step": 115530
    },
    {
      "epoch": 13.90373044524669,
      "grad_norm": 1647.1949462890625,
      "learning_rate": 0.00019131746465992676,
      "loss": 7.2693,
      "step": 115540
    },
    {
      "epoch": 13.904933814681106,
      "grad_norm": 1018.6296997070312,
      "learning_rate": 0.00019131591409338156,
      "loss": 7.3063,
      "step": 115550
    },
    {
      "epoch": 13.906137184115524,
      "grad_norm": 1787.6634521484375,
      "learning_rate": 0.00019131436339467968,
      "loss": 7.3459,
      "step": 115560
    },
    {
      "epoch": 13.90734055354994,
      "grad_norm": 663.8572998046875,
      "learning_rate": 0.00019131281256382337,
      "loss": 7.2402,
      "step": 115570
    },
    {
      "epoch": 13.908543922984355,
      "grad_norm": 1712.150634765625,
      "learning_rate": 0.00019131126160081485,
      "loss": 7.3927,
      "step": 115580
    },
    {
      "epoch": 13.909747292418773,
      "grad_norm": 1224.6893310546875,
      "learning_rate": 0.0001913097105056564,
      "loss": 7.3018,
      "step": 115590
    },
    {
      "epoch": 13.910950661853189,
      "grad_norm": 624.0105590820312,
      "learning_rate": 0.00019130815927835028,
      "loss": 7.2882,
      "step": 115600
    },
    {
      "epoch": 13.912154031287605,
      "grad_norm": 935.8486938476562,
      "learning_rate": 0.00019130660791889868,
      "loss": 7.424,
      "step": 115610
    },
    {
      "epoch": 13.913357400722022,
      "grad_norm": 2166.057373046875,
      "learning_rate": 0.0001913050564273039,
      "loss": 7.3981,
      "step": 115620
    },
    {
      "epoch": 13.914560770156438,
      "grad_norm": 1311.1768798828125,
      "learning_rate": 0.00019130350480356815,
      "loss": 7.279,
      "step": 115630
    },
    {
      "epoch": 13.915764139590854,
      "grad_norm": 623.0220947265625,
      "learning_rate": 0.0001913019530476937,
      "loss": 7.2919,
      "step": 115640
    },
    {
      "epoch": 13.916967509025271,
      "grad_norm": 7811.54296875,
      "learning_rate": 0.00019130040115968274,
      "loss": 7.3636,
      "step": 115650
    },
    {
      "epoch": 13.918170878459687,
      "grad_norm": 4457.65869140625,
      "learning_rate": 0.00019129884913953758,
      "loss": 7.3314,
      "step": 115660
    },
    {
      "epoch": 13.919374247894103,
      "grad_norm": 741759.125,
      "learning_rate": 0.00019129729698726046,
      "loss": 7.3351,
      "step": 115670
    },
    {
      "epoch": 13.92057761732852,
      "grad_norm": 44654788.0,
      "learning_rate": 0.0001912957447028536,
      "loss": 7.4368,
      "step": 115680
    },
    {
      "epoch": 13.921780986762936,
      "grad_norm": 220440.046875,
      "learning_rate": 0.00019129419228631927,
      "loss": 7.5926,
      "step": 115690
    },
    {
      "epoch": 13.922984356197352,
      "grad_norm": 183853.40625,
      "learning_rate": 0.00019129263973765968,
      "loss": 7.4413,
      "step": 115700
    },
    {
      "epoch": 13.92418772563177,
      "grad_norm": 120519.4140625,
      "learning_rate": 0.00019129108705687712,
      "loss": 7.392,
      "step": 115710
    },
    {
      "epoch": 13.925391095066185,
      "grad_norm": 148056.796875,
      "learning_rate": 0.00019128953424397383,
      "loss": 7.4608,
      "step": 115720
    },
    {
      "epoch": 13.926594464500601,
      "grad_norm": 71594.375,
      "learning_rate": 0.000191287981298952,
      "loss": 7.4792,
      "step": 115730
    },
    {
      "epoch": 13.927797833935019,
      "grad_norm": 110738.96875,
      "learning_rate": 0.00019128642822181395,
      "loss": 7.4789,
      "step": 115740
    },
    {
      "epoch": 13.929001203369435,
      "grad_norm": 477835.84375,
      "learning_rate": 0.0001912848750125619,
      "loss": 7.449,
      "step": 115750
    },
    {
      "epoch": 13.93020457280385,
      "grad_norm": 73396.6875,
      "learning_rate": 0.0001912833216711981,
      "loss": 7.4704,
      "step": 115760
    },
    {
      "epoch": 13.931407942238268,
      "grad_norm": 3557499.75,
      "learning_rate": 0.00019128176819772482,
      "loss": 7.4118,
      "step": 115770
    },
    {
      "epoch": 13.932611311672684,
      "grad_norm": 213500.859375,
      "learning_rate": 0.00019128021459214424,
      "loss": 7.5003,
      "step": 115780
    },
    {
      "epoch": 13.9338146811071,
      "grad_norm": 258351.4375,
      "learning_rate": 0.00019127866085445866,
      "loss": 7.4014,
      "step": 115790
    },
    {
      "epoch": 13.935018050541515,
      "grad_norm": 38978.03515625,
      "learning_rate": 0.00019127710698467034,
      "loss": 7.3462,
      "step": 115800
    },
    {
      "epoch": 13.936221419975933,
      "grad_norm": 306324.0,
      "learning_rate": 0.00019127555298278153,
      "loss": 7.3709,
      "step": 115810
    },
    {
      "epoch": 13.937424789410349,
      "grad_norm": 87539.3515625,
      "learning_rate": 0.00019127399884879442,
      "loss": 7.2118,
      "step": 115820
    },
    {
      "epoch": 13.938628158844764,
      "grad_norm": 101658.6015625,
      "learning_rate": 0.00019127244458271128,
      "loss": 7.3367,
      "step": 115830
    },
    {
      "epoch": 13.939831528279182,
      "grad_norm": 237027.84375,
      "learning_rate": 0.00019127089018453446,
      "loss": 7.3105,
      "step": 115840
    },
    {
      "epoch": 13.941034897713598,
      "grad_norm": 491532.09375,
      "learning_rate": 0.00019126933565426606,
      "loss": 7.4238,
      "step": 115850
    },
    {
      "epoch": 13.942238267148014,
      "grad_norm": 306450.78125,
      "learning_rate": 0.00019126778099190842,
      "loss": 7.4432,
      "step": 115860
    },
    {
      "epoch": 13.943441636582431,
      "grad_norm": 176997.78125,
      "learning_rate": 0.00019126622619746378,
      "loss": 7.4159,
      "step": 115870
    },
    {
      "epoch": 13.944645006016847,
      "grad_norm": 339021.1875,
      "learning_rate": 0.00019126467127093434,
      "loss": 7.3195,
      "step": 115880
    },
    {
      "epoch": 13.945848375451263,
      "grad_norm": 1361273.0,
      "learning_rate": 0.0001912631162123224,
      "loss": 7.3144,
      "step": 115890
    },
    {
      "epoch": 13.94705174488568,
      "grad_norm": 1290467.125,
      "learning_rate": 0.00019126156102163022,
      "loss": 7.4041,
      "step": 115900
    },
    {
      "epoch": 13.948255114320096,
      "grad_norm": 747153.25,
      "learning_rate": 0.00019126000569886,
      "loss": 7.4325,
      "step": 115910
    },
    {
      "epoch": 13.949458483754512,
      "grad_norm": 86327400.0,
      "learning_rate": 0.00019125845024401406,
      "loss": 7.5069,
      "step": 115920
    },
    {
      "epoch": 13.95066185318893,
      "grad_norm": 736983.8125,
      "learning_rate": 0.0001912568946570946,
      "loss": 7.4164,
      "step": 115930
    },
    {
      "epoch": 13.951865222623345,
      "grad_norm": 736444.875,
      "learning_rate": 0.00019125533893810386,
      "loss": 7.4977,
      "step": 115940
    },
    {
      "epoch": 13.953068592057761,
      "grad_norm": 425886.09375,
      "learning_rate": 0.00019125378308704414,
      "loss": 7.448,
      "step": 115950
    },
    {
      "epoch": 13.954271961492179,
      "grad_norm": 210765.0625,
      "learning_rate": 0.00019125222710391767,
      "loss": 7.391,
      "step": 115960
    },
    {
      "epoch": 13.955475330926594,
      "grad_norm": 164297.4375,
      "learning_rate": 0.00019125067098872664,
      "loss": 7.2417,
      "step": 115970
    },
    {
      "epoch": 13.95667870036101,
      "grad_norm": 416952.03125,
      "learning_rate": 0.00019124911474147343,
      "loss": 7.3721,
      "step": 115980
    },
    {
      "epoch": 13.957882069795428,
      "grad_norm": 180548.59375,
      "learning_rate": 0.00019124755836216019,
      "loss": 7.3208,
      "step": 115990
    },
    {
      "epoch": 13.959085439229844,
      "grad_norm": 125330.5546875,
      "learning_rate": 0.0001912460018507892,
      "loss": 7.4576,
      "step": 116000
    },
    {
      "epoch": 13.96028880866426,
      "grad_norm": 109254.2578125,
      "learning_rate": 0.00019124444520736273,
      "loss": 7.3524,
      "step": 116010
    },
    {
      "epoch": 13.961492178098677,
      "grad_norm": 110185.8359375,
      "learning_rate": 0.00019124288843188302,
      "loss": 7.3794,
      "step": 116020
    },
    {
      "epoch": 13.962695547533093,
      "grad_norm": 297316.21875,
      "learning_rate": 0.00019124133152435233,
      "loss": 7.455,
      "step": 116030
    },
    {
      "epoch": 13.963898916967509,
      "grad_norm": 335070.09375,
      "learning_rate": 0.0001912397744847729,
      "loss": 7.2952,
      "step": 116040
    },
    {
      "epoch": 13.965102286401926,
      "grad_norm": 1116348.75,
      "learning_rate": 0.00019123821731314698,
      "loss": 7.4031,
      "step": 116050
    },
    {
      "epoch": 13.966305655836342,
      "grad_norm": 90652.84375,
      "learning_rate": 0.00019123666000947682,
      "loss": 7.3023,
      "step": 116060
    },
    {
      "epoch": 13.967509025270758,
      "grad_norm": 1236206.125,
      "learning_rate": 0.00019123510257376472,
      "loss": 7.4441,
      "step": 116070
    },
    {
      "epoch": 13.968712394705175,
      "grad_norm": 518104.8125,
      "learning_rate": 0.00019123354500601288,
      "loss": 7.3606,
      "step": 116080
    },
    {
      "epoch": 13.969915764139591,
      "grad_norm": 57724.90234375,
      "learning_rate": 0.0001912319873062236,
      "loss": 7.4245,
      "step": 116090
    },
    {
      "epoch": 13.971119133574007,
      "grad_norm": 1966861.125,
      "learning_rate": 0.00019123042947439907,
      "loss": 7.3486,
      "step": 116100
    },
    {
      "epoch": 13.972322503008424,
      "grad_norm": 207407.09375,
      "learning_rate": 0.00019122887151054162,
      "loss": 7.4521,
      "step": 116110
    },
    {
      "epoch": 13.97352587244284,
      "grad_norm": 1851230.375,
      "learning_rate": 0.00019122731341465347,
      "loss": 7.3208,
      "step": 116120
    },
    {
      "epoch": 13.974729241877256,
      "grad_norm": 99642.921875,
      "learning_rate": 0.00019122575518673687,
      "loss": 7.2795,
      "step": 116130
    },
    {
      "epoch": 13.975932611311674,
      "grad_norm": 169405.84375,
      "learning_rate": 0.00019122419682679405,
      "loss": 7.4647,
      "step": 116140
    },
    {
      "epoch": 13.97713598074609,
      "grad_norm": 683880.6875,
      "learning_rate": 0.00019122263833482733,
      "loss": 7.399,
      "step": 116150
    },
    {
      "epoch": 13.978339350180505,
      "grad_norm": 93851.8515625,
      "learning_rate": 0.0001912210797108389,
      "loss": 7.3239,
      "step": 116160
    },
    {
      "epoch": 13.979542719614923,
      "grad_norm": 630520.75,
      "learning_rate": 0.00019121952095483107,
      "loss": 7.3098,
      "step": 116170
    },
    {
      "epoch": 13.980746089049338,
      "grad_norm": 58073.99609375,
      "learning_rate": 0.00019121796206680606,
      "loss": 7.4691,
      "step": 116180
    },
    {
      "epoch": 13.981949458483754,
      "grad_norm": 264698.6875,
      "learning_rate": 0.00019121640304676615,
      "loss": 7.3899,
      "step": 116190
    },
    {
      "epoch": 13.98315282791817,
      "grad_norm": 612826.6875,
      "learning_rate": 0.0001912148438947136,
      "loss": 7.3602,
      "step": 116200
    },
    {
      "epoch": 13.984356197352588,
      "grad_norm": 274209.15625,
      "learning_rate": 0.0001912132846106506,
      "loss": 7.3731,
      "step": 116210
    },
    {
      "epoch": 13.985559566787003,
      "grad_norm": 411674.96875,
      "learning_rate": 0.00019121172519457949,
      "loss": 7.3124,
      "step": 116220
    },
    {
      "epoch": 13.98676293622142,
      "grad_norm": 573140.625,
      "learning_rate": 0.0001912101656465025,
      "loss": 7.371,
      "step": 116230
    },
    {
      "epoch": 13.987966305655837,
      "grad_norm": 275947.03125,
      "learning_rate": 0.00019120860596642185,
      "loss": 7.396,
      "step": 116240
    },
    {
      "epoch": 13.989169675090253,
      "grad_norm": 1235720.25,
      "learning_rate": 0.00019120704615433983,
      "loss": 7.3123,
      "step": 116250
    },
    {
      "epoch": 13.990373044524668,
      "grad_norm": 270274.875,
      "learning_rate": 0.00019120548621025872,
      "loss": 7.2705,
      "step": 116260
    },
    {
      "epoch": 13.991576413959086,
      "grad_norm": 1281890.75,
      "learning_rate": 0.00019120392613418074,
      "loss": 7.4206,
      "step": 116270
    },
    {
      "epoch": 13.992779783393502,
      "grad_norm": 128000.0078125,
      "learning_rate": 0.0001912023659261082,
      "loss": 7.381,
      "step": 116280
    },
    {
      "epoch": 13.993983152827917,
      "grad_norm": 445059.4375,
      "learning_rate": 0.00019120080558604328,
      "loss": 7.3548,
      "step": 116290
    },
    {
      "epoch": 13.995186522262335,
      "grad_norm": 12816711.0,
      "learning_rate": 0.00019119924511398827,
      "loss": 7.385,
      "step": 116300
    },
    {
      "epoch": 13.99638989169675,
      "grad_norm": 733852.875,
      "learning_rate": 0.00019119768450994546,
      "loss": 7.2864,
      "step": 116310
    },
    {
      "epoch": 13.997593261131167,
      "grad_norm": 254457.140625,
      "learning_rate": 0.0001911961237739171,
      "loss": 7.3542,
      "step": 116320
    },
    {
      "epoch": 13.998796630565584,
      "grad_norm": 250920.65625,
      "learning_rate": 0.00019119456290590537,
      "loss": 7.501,
      "step": 116330
    },
    {
      "epoch": 14.0,
      "grad_norm": 757592.3125,
      "learning_rate": 0.00019119300190591264,
      "loss": 7.4478,
      "step": 116340
    },
    {
      "epoch": 14.0,
      "eval_loss": 7.410001277923584,
      "eval_runtime": 119.0415,
      "eval_samples_per_second": 62.054,
      "eval_steps_per_second": 7.762,
      "step": 116340
    },
    {
      "epoch": 14.001203369434416,
      "grad_norm": 803210.9375,
      "learning_rate": 0.00019119144077394112,
      "loss": 7.4929,
      "step": 116350
    },
    {
      "epoch": 14.002406738868833,
      "grad_norm": 492713.84375,
      "learning_rate": 0.00019118987950999308,
      "loss": 7.3895,
      "step": 116360
    },
    {
      "epoch": 14.00361010830325,
      "grad_norm": 211750.4375,
      "learning_rate": 0.00019118831811407074,
      "loss": 7.3983,
      "step": 116370
    },
    {
      "epoch": 14.004813477737665,
      "grad_norm": 1076696.375,
      "learning_rate": 0.00019118675658617642,
      "loss": 7.4769,
      "step": 116380
    },
    {
      "epoch": 14.006016847172083,
      "grad_norm": 326243.5,
      "learning_rate": 0.00019118519492631233,
      "loss": 7.4759,
      "step": 116390
    },
    {
      "epoch": 14.007220216606498,
      "grad_norm": 404986.9375,
      "learning_rate": 0.00019118363313448073,
      "loss": 7.3638,
      "step": 116400
    },
    {
      "epoch": 14.008423586040914,
      "grad_norm": 1638931.375,
      "learning_rate": 0.00019118207121068394,
      "loss": 7.3761,
      "step": 116410
    },
    {
      "epoch": 14.009626955475332,
      "grad_norm": 356602.65625,
      "learning_rate": 0.00019118050915492416,
      "loss": 7.3779,
      "step": 116420
    },
    {
      "epoch": 14.010830324909747,
      "grad_norm": 879317.875,
      "learning_rate": 0.00019117894696720367,
      "loss": 7.3589,
      "step": 116430
    },
    {
      "epoch": 14.012033694344163,
      "grad_norm": 176502.515625,
      "learning_rate": 0.00019117738464752474,
      "loss": 7.3981,
      "step": 116440
    },
    {
      "epoch": 14.01323706377858,
      "grad_norm": 1440828.375,
      "learning_rate": 0.00019117582219588963,
      "loss": 7.3956,
      "step": 116450
    },
    {
      "epoch": 14.014440433212997,
      "grad_norm": 368090.78125,
      "learning_rate": 0.0001911742596123006,
      "loss": 7.4667,
      "step": 116460
    },
    {
      "epoch": 14.015643802647412,
      "grad_norm": 837236.625,
      "learning_rate": 0.0001911726968967599,
      "loss": 7.3579,
      "step": 116470
    },
    {
      "epoch": 14.01684717208183,
      "grad_norm": 1544272.375,
      "learning_rate": 0.00019117113404926978,
      "loss": 7.3264,
      "step": 116480
    },
    {
      "epoch": 14.018050541516246,
      "grad_norm": 377169.53125,
      "learning_rate": 0.00019116957106983252,
      "loss": 7.3992,
      "step": 116490
    },
    {
      "epoch": 14.019253910950662,
      "grad_norm": 212784.9375,
      "learning_rate": 0.0001911680079584504,
      "loss": 7.4495,
      "step": 116500
    },
    {
      "epoch": 14.020457280385079,
      "grad_norm": 1739744.375,
      "learning_rate": 0.00019116644471512567,
      "loss": 7.4042,
      "step": 116510
    },
    {
      "epoch": 14.021660649819495,
      "grad_norm": 1461934.0,
      "learning_rate": 0.00019116488133986054,
      "loss": 7.4079,
      "step": 116520
    },
    {
      "epoch": 14.02286401925391,
      "grad_norm": 71283.9375,
      "learning_rate": 0.00019116331783265738,
      "loss": 7.5439,
      "step": 116530
    },
    {
      "epoch": 14.024067388688326,
      "grad_norm": 401461.125,
      "learning_rate": 0.00019116175419351837,
      "loss": 7.3666,
      "step": 116540
    },
    {
      "epoch": 14.025270758122744,
      "grad_norm": 8769769.0,
      "learning_rate": 0.00019116019042244578,
      "loss": 7.2829,
      "step": 116550
    },
    {
      "epoch": 14.02647412755716,
      "grad_norm": 962703.1875,
      "learning_rate": 0.00019115862651944188,
      "loss": 7.4646,
      "step": 116560
    },
    {
      "epoch": 14.027677496991576,
      "grad_norm": 628370.5625,
      "learning_rate": 0.00019115706248450898,
      "loss": 7.4724,
      "step": 116570
    },
    {
      "epoch": 14.028880866425993,
      "grad_norm": 1131946.875,
      "learning_rate": 0.00019115549831764925,
      "loss": 7.4846,
      "step": 116580
    },
    {
      "epoch": 14.030084235860409,
      "grad_norm": 205639.09375,
      "learning_rate": 0.00019115393401886505,
      "loss": 7.6485,
      "step": 116590
    },
    {
      "epoch": 14.031287605294825,
      "grad_norm": 695927.1875,
      "learning_rate": 0.0001911523695881586,
      "loss": 7.4882,
      "step": 116600
    },
    {
      "epoch": 14.032490974729242,
      "grad_norm": 1315748.25,
      "learning_rate": 0.00019115080502553214,
      "loss": 7.4432,
      "step": 116610
    },
    {
      "epoch": 14.033694344163658,
      "grad_norm": 1552780.75,
      "learning_rate": 0.000191149240330988,
      "loss": 7.4332,
      "step": 116620
    },
    {
      "epoch": 14.034897713598074,
      "grad_norm": 1138868.875,
      "learning_rate": 0.00019114767550452836,
      "loss": 7.5017,
      "step": 116630
    },
    {
      "epoch": 14.036101083032491,
      "grad_norm": 63607.9453125,
      "learning_rate": 0.00019114611054615557,
      "loss": 7.5391,
      "step": 116640
    },
    {
      "epoch": 14.037304452466907,
      "grad_norm": 779430.625,
      "learning_rate": 0.00019114454545587185,
      "loss": 7.5726,
      "step": 116650
    },
    {
      "epoch": 14.038507821901323,
      "grad_norm": 8863215.0,
      "learning_rate": 0.00019114298023367946,
      "loss": 7.3929,
      "step": 116660
    },
    {
      "epoch": 14.03971119133574,
      "grad_norm": 305410.1875,
      "learning_rate": 0.00019114141487958067,
      "loss": 7.3741,
      "step": 116670
    },
    {
      "epoch": 14.040914560770156,
      "grad_norm": 5025197.5,
      "learning_rate": 0.00019113984939357774,
      "loss": 7.5043,
      "step": 116680
    },
    {
      "epoch": 14.042117930204572,
      "grad_norm": 183266.390625,
      "learning_rate": 0.00019113828377567295,
      "loss": 7.4493,
      "step": 116690
    },
    {
      "epoch": 14.04332129963899,
      "grad_norm": 1260241.125,
      "learning_rate": 0.0001911367180258686,
      "loss": 7.3701,
      "step": 116700
    },
    {
      "epoch": 14.044524669073406,
      "grad_norm": 1585346.75,
      "learning_rate": 0.00019113515214416689,
      "loss": 7.6099,
      "step": 116710
    },
    {
      "epoch": 14.045728038507821,
      "grad_norm": 483882.375,
      "learning_rate": 0.00019113358613057012,
      "loss": 7.4502,
      "step": 116720
    },
    {
      "epoch": 14.046931407942239,
      "grad_norm": 4439001.0,
      "learning_rate": 0.0001911320199850805,
      "loss": 7.5441,
      "step": 116730
    },
    {
      "epoch": 14.048134777376655,
      "grad_norm": 598037.875,
      "learning_rate": 0.00019113045370770042,
      "loss": 7.4766,
      "step": 116740
    },
    {
      "epoch": 14.04933814681107,
      "grad_norm": 267443.03125,
      "learning_rate": 0.00019112888729843204,
      "loss": 7.5264,
      "step": 116750
    },
    {
      "epoch": 14.050541516245488,
      "grad_norm": 2042944.75,
      "learning_rate": 0.00019112732075727767,
      "loss": 7.4372,
      "step": 116760
    },
    {
      "epoch": 14.051744885679904,
      "grad_norm": 8590250.0,
      "learning_rate": 0.00019112575408423957,
      "loss": 7.3768,
      "step": 116770
    },
    {
      "epoch": 14.05294825511432,
      "grad_norm": 778564.25,
      "learning_rate": 0.00019112418727932,
      "loss": 7.4346,
      "step": 116780
    },
    {
      "epoch": 14.054151624548737,
      "grad_norm": 438709.09375,
      "learning_rate": 0.00019112262034252125,
      "loss": 7.4166,
      "step": 116790
    },
    {
      "epoch": 14.055354993983153,
      "grad_norm": 168011.28125,
      "learning_rate": 0.00019112105327384556,
      "loss": 7.4851,
      "step": 116800
    },
    {
      "epoch": 14.056558363417569,
      "grad_norm": 5476587.0,
      "learning_rate": 0.0001911194860732952,
      "loss": 7.4535,
      "step": 116810
    },
    {
      "epoch": 14.057761732851986,
      "grad_norm": 350191.34375,
      "learning_rate": 0.00019111791874087245,
      "loss": 7.4771,
      "step": 116820
    },
    {
      "epoch": 14.058965102286402,
      "grad_norm": 518645.3125,
      "learning_rate": 0.0001911163512765796,
      "loss": 7.5144,
      "step": 116830
    },
    {
      "epoch": 14.060168471720818,
      "grad_norm": 966435.1875,
      "learning_rate": 0.00019111478368041883,
      "loss": 7.5035,
      "step": 116840
    },
    {
      "epoch": 14.061371841155236,
      "grad_norm": 763140.375,
      "learning_rate": 0.00019111321595239252,
      "loss": 7.4865,
      "step": 116850
    },
    {
      "epoch": 14.062575210589651,
      "grad_norm": 1528244.625,
      "learning_rate": 0.00019111164809250292,
      "loss": 7.425,
      "step": 116860
    },
    {
      "epoch": 14.063778580024067,
      "grad_norm": 937681.5,
      "learning_rate": 0.0001911100801007522,
      "loss": 7.5801,
      "step": 116870
    },
    {
      "epoch": 14.064981949458483,
      "grad_norm": 632960.375,
      "learning_rate": 0.00019110851197714276,
      "loss": 7.5401,
      "step": 116880
    },
    {
      "epoch": 14.0661853188929,
      "grad_norm": 430858.0625,
      "learning_rate": 0.00019110694372167676,
      "loss": 7.5046,
      "step": 116890
    },
    {
      "epoch": 14.067388688327316,
      "grad_norm": 567233.125,
      "learning_rate": 0.00019110537533435654,
      "loss": 7.4265,
      "step": 116900
    },
    {
      "epoch": 14.068592057761732,
      "grad_norm": 732851.25,
      "learning_rate": 0.00019110380681518438,
      "loss": 7.3545,
      "step": 116910
    },
    {
      "epoch": 14.06979542719615,
      "grad_norm": 460917.28125,
      "learning_rate": 0.00019110223816416248,
      "loss": 7.3706,
      "step": 116920
    },
    {
      "epoch": 14.070998796630565,
      "grad_norm": 186629.015625,
      "learning_rate": 0.00019110066938129316,
      "loss": 7.4142,
      "step": 116930
    },
    {
      "epoch": 14.072202166064981,
      "grad_norm": 269234.15625,
      "learning_rate": 0.00019109910046657865,
      "loss": 7.5336,
      "step": 116940
    },
    {
      "epoch": 14.073405535499399,
      "grad_norm": 1003651.625,
      "learning_rate": 0.0001910975314200213,
      "loss": 7.5019,
      "step": 116950
    },
    {
      "epoch": 14.074608904933815,
      "grad_norm": 52844284.0,
      "learning_rate": 0.0001910959622416233,
      "loss": 7.5406,
      "step": 116960
    },
    {
      "epoch": 14.07581227436823,
      "grad_norm": 703983.3125,
      "learning_rate": 0.00019109439293138697,
      "loss": 7.3688,
      "step": 116970
    },
    {
      "epoch": 14.077015643802648,
      "grad_norm": 379927.46875,
      "learning_rate": 0.00019109282348931457,
      "loss": 7.5265,
      "step": 116980
    },
    {
      "epoch": 14.078219013237064,
      "grad_norm": 590824.375,
      "learning_rate": 0.00019109125391540834,
      "loss": 7.5557,
      "step": 116990
    },
    {
      "epoch": 14.07942238267148,
      "grad_norm": 223986.796875,
      "learning_rate": 0.00019108968420967058,
      "loss": 7.5759,
      "step": 117000
    },
    {
      "epoch": 14.080625752105897,
      "grad_norm": 5330962.0,
      "learning_rate": 0.00019108811437210357,
      "loss": 7.6068,
      "step": 117010
    },
    {
      "epoch": 14.081829121540313,
      "grad_norm": 4413083.0,
      "learning_rate": 0.00019108654440270958,
      "loss": 7.6588,
      "step": 117020
    },
    {
      "epoch": 14.083032490974729,
      "grad_norm": 2584871.25,
      "learning_rate": 0.00019108497430149083,
      "loss": 7.6518,
      "step": 117030
    },
    {
      "epoch": 14.084235860409146,
      "grad_norm": 779955.9375,
      "learning_rate": 0.00019108340406844967,
      "loss": 7.705,
      "step": 117040
    },
    {
      "epoch": 14.085439229843562,
      "grad_norm": 5292306.5,
      "learning_rate": 0.00019108183370358833,
      "loss": 7.71,
      "step": 117050
    },
    {
      "epoch": 14.086642599277978,
      "grad_norm": 250159.6875,
      "learning_rate": 0.0001910802632069091,
      "loss": 7.714,
      "step": 117060
    },
    {
      "epoch": 14.087845968712395,
      "grad_norm": 1570434.375,
      "learning_rate": 0.0001910786925784142,
      "loss": 7.7099,
      "step": 117070
    },
    {
      "epoch": 14.089049338146811,
      "grad_norm": 3275709.0,
      "learning_rate": 0.00019107712181810595,
      "loss": 7.7099,
      "step": 117080
    },
    {
      "epoch": 14.090252707581227,
      "grad_norm": 613445.9375,
      "learning_rate": 0.00019107555092598666,
      "loss": 7.7203,
      "step": 117090
    },
    {
      "epoch": 14.091456077015645,
      "grad_norm": 10168207.0,
      "learning_rate": 0.00019107397990205852,
      "loss": 7.7154,
      "step": 117100
    },
    {
      "epoch": 14.09265944645006,
      "grad_norm": 296258.9375,
      "learning_rate": 0.00019107240874632387,
      "loss": 7.6818,
      "step": 117110
    },
    {
      "epoch": 14.093862815884476,
      "grad_norm": 8713289.0,
      "learning_rate": 0.000191070837458785,
      "loss": 7.7367,
      "step": 117120
    },
    {
      "epoch": 14.095066185318894,
      "grad_norm": 448273.75,
      "learning_rate": 0.00019106926603944407,
      "loss": 7.7558,
      "step": 117130
    },
    {
      "epoch": 14.09626955475331,
      "grad_norm": 2434391.75,
      "learning_rate": 0.00019106769448830346,
      "loss": 7.665,
      "step": 117140
    },
    {
      "epoch": 14.097472924187725,
      "grad_norm": 1744259.0,
      "learning_rate": 0.0001910661228053654,
      "loss": 7.7213,
      "step": 117150
    },
    {
      "epoch": 14.098676293622143,
      "grad_norm": 870823.75,
      "learning_rate": 0.00019106455099063217,
      "loss": 7.7306,
      "step": 117160
    },
    {
      "epoch": 14.099879663056559,
      "grad_norm": 268853.03125,
      "learning_rate": 0.00019106297904410608,
      "loss": 7.7649,
      "step": 117170
    },
    {
      "epoch": 14.101083032490974,
      "grad_norm": 316112.84375,
      "learning_rate": 0.00019106140696578938,
      "loss": 7.7016,
      "step": 117180
    },
    {
      "epoch": 14.102286401925392,
      "grad_norm": 855629.75,
      "learning_rate": 0.00019105983475568432,
      "loss": 7.7779,
      "step": 117190
    },
    {
      "epoch": 14.103489771359808,
      "grad_norm": 11373207.0,
      "learning_rate": 0.00019105826241379322,
      "loss": 7.7289,
      "step": 117200
    },
    {
      "epoch": 14.104693140794224,
      "grad_norm": 547612.75,
      "learning_rate": 0.00019105668994011832,
      "loss": 7.7554,
      "step": 117210
    },
    {
      "epoch": 14.10589651022864,
      "grad_norm": 7882332.0,
      "learning_rate": 0.00019105511733466188,
      "loss": 7.7567,
      "step": 117220
    },
    {
      "epoch": 14.107099879663057,
      "grad_norm": 1851580.25,
      "learning_rate": 0.00019105354459742624,
      "loss": 7.7383,
      "step": 117230
    },
    {
      "epoch": 14.108303249097473,
      "grad_norm": 7409177.5,
      "learning_rate": 0.00019105197172841366,
      "loss": 7.841,
      "step": 117240
    },
    {
      "epoch": 14.109506618531888,
      "grad_norm": 926833.1875,
      "learning_rate": 0.00019105039872762634,
      "loss": 7.7346,
      "step": 117250
    },
    {
      "epoch": 14.110709987966306,
      "grad_norm": 916795.875,
      "learning_rate": 0.00019104882559506665,
      "loss": 7.7149,
      "step": 117260
    },
    {
      "epoch": 14.111913357400722,
      "grad_norm": 2947412.25,
      "learning_rate": 0.0001910472523307368,
      "loss": 7.7044,
      "step": 117270
    },
    {
      "epoch": 14.113116726835138,
      "grad_norm": 2368365.75,
      "learning_rate": 0.00019104567893463915,
      "loss": 7.7467,
      "step": 117280
    },
    {
      "epoch": 14.114320096269555,
      "grad_norm": 257529.765625,
      "learning_rate": 0.0001910441054067759,
      "loss": 7.7203,
      "step": 117290
    },
    {
      "epoch": 14.115523465703971,
      "grad_norm": 7703754.0,
      "learning_rate": 0.00019104253174714932,
      "loss": 7.762,
      "step": 117300
    },
    {
      "epoch": 14.116726835138387,
      "grad_norm": 1125570.375,
      "learning_rate": 0.00019104095795576174,
      "loss": 7.7003,
      "step": 117310
    },
    {
      "epoch": 14.117930204572804,
      "grad_norm": 4180567.75,
      "learning_rate": 0.0001910393840326154,
      "loss": 7.748,
      "step": 117320
    },
    {
      "epoch": 14.11913357400722,
      "grad_norm": 15315138.0,
      "learning_rate": 0.00019103780997771264,
      "loss": 7.7816,
      "step": 117330
    },
    {
      "epoch": 14.120336943441636,
      "grad_norm": 770472.75,
      "learning_rate": 0.00019103623579105566,
      "loss": 7.7377,
      "step": 117340
    },
    {
      "epoch": 14.121540312876053,
      "grad_norm": 2066392.125,
      "learning_rate": 0.00019103466147264676,
      "loss": 7.7626,
      "step": 117350
    },
    {
      "epoch": 14.12274368231047,
      "grad_norm": 2314799.0,
      "learning_rate": 0.00019103308702248827,
      "loss": 7.7147,
      "step": 117360
    },
    {
      "epoch": 14.123947051744885,
      "grad_norm": 3735358.75,
      "learning_rate": 0.0001910315124405824,
      "loss": 7.6771,
      "step": 117370
    },
    {
      "epoch": 14.125150421179303,
      "grad_norm": 2781628.5,
      "learning_rate": 0.00019102993772693145,
      "loss": 7.7394,
      "step": 117380
    },
    {
      "epoch": 14.126353790613718,
      "grad_norm": 167057.25,
      "learning_rate": 0.0001910283628815377,
      "loss": 7.7124,
      "step": 117390
    },
    {
      "epoch": 14.127557160048134,
      "grad_norm": 13759909.0,
      "learning_rate": 0.00019102678790440346,
      "loss": 7.6901,
      "step": 117400
    },
    {
      "epoch": 14.128760529482552,
      "grad_norm": 3111263.5,
      "learning_rate": 0.00019102521279553097,
      "loss": 7.6916,
      "step": 117410
    },
    {
      "epoch": 14.129963898916968,
      "grad_norm": 302472.0,
      "learning_rate": 0.00019102363755492253,
      "loss": 7.6121,
      "step": 117420
    },
    {
      "epoch": 14.131167268351383,
      "grad_norm": 3485300.0,
      "learning_rate": 0.00019102206218258043,
      "loss": 7.6994,
      "step": 117430
    },
    {
      "epoch": 14.132370637785801,
      "grad_norm": 5885566.5,
      "learning_rate": 0.0001910204866785069,
      "loss": 7.6486,
      "step": 117440
    },
    {
      "epoch": 14.133574007220217,
      "grad_norm": 1157274.125,
      "learning_rate": 0.00019101891104270427,
      "loss": 7.6777,
      "step": 117450
    },
    {
      "epoch": 14.134777376654633,
      "grad_norm": 1057398.0,
      "learning_rate": 0.00019101733527517481,
      "loss": 7.6866,
      "step": 117460
    },
    {
      "epoch": 14.13598074608905,
      "grad_norm": 1628617.25,
      "learning_rate": 0.0001910157593759208,
      "loss": 7.6428,
      "step": 117470
    },
    {
      "epoch": 14.137184115523466,
      "grad_norm": 919845.0625,
      "learning_rate": 0.0001910141833449445,
      "loss": 7.7262,
      "step": 117480
    },
    {
      "epoch": 14.138387484957882,
      "grad_norm": 32634906.0,
      "learning_rate": 0.00019101260718224817,
      "loss": 7.6889,
      "step": 117490
    },
    {
      "epoch": 14.1395908543923,
      "grad_norm": 1332457.375,
      "learning_rate": 0.0001910110308878342,
      "loss": 7.7526,
      "step": 117500
    },
    {
      "epoch": 14.140794223826715,
      "grad_norm": 688719.75,
      "learning_rate": 0.00019100945446170475,
      "loss": 7.7073,
      "step": 117510
    },
    {
      "epoch": 14.14199759326113,
      "grad_norm": 3635474.25,
      "learning_rate": 0.00019100787790386216,
      "loss": 7.6832,
      "step": 117520
    },
    {
      "epoch": 14.143200962695548,
      "grad_norm": 3694973.5,
      "learning_rate": 0.00019100630121430872,
      "loss": 7.7528,
      "step": 117530
    },
    {
      "epoch": 14.144404332129964,
      "grad_norm": 612005.6875,
      "learning_rate": 0.00019100472439304668,
      "loss": 7.6511,
      "step": 117540
    },
    {
      "epoch": 14.14560770156438,
      "grad_norm": 310124.40625,
      "learning_rate": 0.00019100314744007832,
      "loss": 7.6766,
      "step": 117550
    },
    {
      "epoch": 14.146811070998796,
      "grad_norm": 472426.40625,
      "learning_rate": 0.00019100157035540597,
      "loss": 7.6752,
      "step": 117560
    },
    {
      "epoch": 14.148014440433213,
      "grad_norm": 1072434.125,
      "learning_rate": 0.00019099999313903183,
      "loss": 7.6697,
      "step": 117570
    },
    {
      "epoch": 14.14921780986763,
      "grad_norm": 390492.5625,
      "learning_rate": 0.00019099841579095828,
      "loss": 7.7356,
      "step": 117580
    },
    {
      "epoch": 14.150421179302045,
      "grad_norm": 500926.78125,
      "learning_rate": 0.00019099683831118753,
      "loss": 7.7001,
      "step": 117590
    },
    {
      "epoch": 14.151624548736462,
      "grad_norm": 617192.5,
      "learning_rate": 0.0001909952606997219,
      "loss": 7.6278,
      "step": 117600
    },
    {
      "epoch": 14.152827918170878,
      "grad_norm": 1324842.375,
      "learning_rate": 0.00019099368295656368,
      "loss": 7.595,
      "step": 117610
    },
    {
      "epoch": 14.154031287605294,
      "grad_norm": 308666.59375,
      "learning_rate": 0.0001909921050817151,
      "loss": 7.6461,
      "step": 117620
    },
    {
      "epoch": 14.155234657039712,
      "grad_norm": 2842516.75,
      "learning_rate": 0.0001909905270751785,
      "loss": 7.5856,
      "step": 117630
    },
    {
      "epoch": 14.156438026474127,
      "grad_norm": 84979.5625,
      "learning_rate": 0.00019098894893695615,
      "loss": 7.6255,
      "step": 117640
    },
    {
      "epoch": 14.157641395908543,
      "grad_norm": 841324.625,
      "learning_rate": 0.00019098737066705027,
      "loss": 7.5809,
      "step": 117650
    },
    {
      "epoch": 14.15884476534296,
      "grad_norm": 1796559.0,
      "learning_rate": 0.00019098579226546326,
      "loss": 7.6321,
      "step": 117660
    },
    {
      "epoch": 14.160048134777377,
      "grad_norm": 532219.5,
      "learning_rate": 0.00019098421373219733,
      "loss": 7.5754,
      "step": 117670
    },
    {
      "epoch": 14.161251504211792,
      "grad_norm": 1030640.5,
      "learning_rate": 0.00019098263506725477,
      "loss": 7.6161,
      "step": 117680
    },
    {
      "epoch": 14.16245487364621,
      "grad_norm": 60111.03125,
      "learning_rate": 0.0001909810562706379,
      "loss": 7.682,
      "step": 117690
    },
    {
      "epoch": 14.163658243080626,
      "grad_norm": 1406366.875,
      "learning_rate": 0.00019097947734234895,
      "loss": 7.695,
      "step": 117700
    },
    {
      "epoch": 14.164861612515042,
      "grad_norm": 1101489.5,
      "learning_rate": 0.00019097789828239026,
      "loss": 7.6959,
      "step": 117710
    },
    {
      "epoch": 14.166064981949459,
      "grad_norm": 325357.65625,
      "learning_rate": 0.00019097631909076407,
      "loss": 7.6261,
      "step": 117720
    },
    {
      "epoch": 14.167268351383875,
      "grad_norm": 229152.828125,
      "learning_rate": 0.00019097473976747268,
      "loss": 7.5822,
      "step": 117730
    },
    {
      "epoch": 14.16847172081829,
      "grad_norm": 55116.89453125,
      "learning_rate": 0.0001909731603125184,
      "loss": 7.569,
      "step": 117740
    },
    {
      "epoch": 14.169675090252708,
      "grad_norm": 276046.9375,
      "learning_rate": 0.0001909715807259035,
      "loss": 7.6175,
      "step": 117750
    },
    {
      "epoch": 14.170878459687124,
      "grad_norm": 2432009.5,
      "learning_rate": 0.00019097000100763025,
      "loss": 7.6564,
      "step": 117760
    },
    {
      "epoch": 14.17208182912154,
      "grad_norm": 5763792.0,
      "learning_rate": 0.00019096842115770094,
      "loss": 7.5945,
      "step": 117770
    },
    {
      "epoch": 14.173285198555957,
      "grad_norm": 715934.125,
      "learning_rate": 0.00019096684117611787,
      "loss": 7.6371,
      "step": 117780
    },
    {
      "epoch": 14.174488567990373,
      "grad_norm": 104462.046875,
      "learning_rate": 0.00019096526106288333,
      "loss": 7.5974,
      "step": 117790
    },
    {
      "epoch": 14.175691937424789,
      "grad_norm": 172284.640625,
      "learning_rate": 0.00019096368081799957,
      "loss": 7.5797,
      "step": 117800
    },
    {
      "epoch": 14.176895306859207,
      "grad_norm": 839931.125,
      "learning_rate": 0.00019096210044146895,
      "loss": 7.6661,
      "step": 117810
    },
    {
      "epoch": 14.178098676293622,
      "grad_norm": 103322.25,
      "learning_rate": 0.00019096051993329369,
      "loss": 7.5413,
      "step": 117820
    },
    {
      "epoch": 14.179302045728038,
      "grad_norm": 347017.21875,
      "learning_rate": 0.0001909589392934761,
      "loss": 7.6953,
      "step": 117830
    },
    {
      "epoch": 14.180505415162456,
      "grad_norm": 82645.4453125,
      "learning_rate": 0.00019095735852201848,
      "loss": 7.5759,
      "step": 117840
    },
    {
      "epoch": 14.181708784596871,
      "grad_norm": 741268.0625,
      "learning_rate": 0.00019095577761892306,
      "loss": 7.6107,
      "step": 117850
    },
    {
      "epoch": 14.182912154031287,
      "grad_norm": 1132008.0,
      "learning_rate": 0.00019095419658419225,
      "loss": 7.6534,
      "step": 117860
    },
    {
      "epoch": 14.184115523465705,
      "grad_norm": 7227692.5,
      "learning_rate": 0.0001909526154178282,
      "loss": 7.6384,
      "step": 117870
    },
    {
      "epoch": 14.18531889290012,
      "grad_norm": 2999032.0,
      "learning_rate": 0.00019095103411983328,
      "loss": 7.6321,
      "step": 117880
    },
    {
      "epoch": 14.186522262334536,
      "grad_norm": 352842.65625,
      "learning_rate": 0.00019094945269020977,
      "loss": 7.5941,
      "step": 117890
    },
    {
      "epoch": 14.187725631768952,
      "grad_norm": 1475735.375,
      "learning_rate": 0.00019094787112895994,
      "loss": 7.5975,
      "step": 117900
    },
    {
      "epoch": 14.18892900120337,
      "grad_norm": 1614576.5,
      "learning_rate": 0.00019094628943608607,
      "loss": 7.6226,
      "step": 117910
    },
    {
      "epoch": 14.190132370637786,
      "grad_norm": 1642992.375,
      "learning_rate": 0.0001909447076115905,
      "loss": 7.7208,
      "step": 117920
    },
    {
      "epoch": 14.191335740072201,
      "grad_norm": 136119.203125,
      "learning_rate": 0.00019094312565547545,
      "loss": 7.6313,
      "step": 117930
    },
    {
      "epoch": 14.192539109506619,
      "grad_norm": 83723.78125,
      "learning_rate": 0.00019094154356774325,
      "loss": 7.667,
      "step": 117940
    },
    {
      "epoch": 14.193742478941035,
      "grad_norm": 142090.59375,
      "learning_rate": 0.0001909399613483962,
      "loss": 7.6637,
      "step": 117950
    },
    {
      "epoch": 14.19494584837545,
      "grad_norm": 1624011.75,
      "learning_rate": 0.00019093837899743655,
      "loss": 7.5476,
      "step": 117960
    },
    {
      "epoch": 14.196149217809868,
      "grad_norm": 165817.359375,
      "learning_rate": 0.00019093679651486665,
      "loss": 7.6582,
      "step": 117970
    },
    {
      "epoch": 14.197352587244284,
      "grad_norm": 205383.234375,
      "learning_rate": 0.0001909352139006887,
      "loss": 7.6367,
      "step": 117980
    },
    {
      "epoch": 14.1985559566787,
      "grad_norm": 78171.171875,
      "learning_rate": 0.00019093363115490507,
      "loss": 7.5957,
      "step": 117990
    },
    {
      "epoch": 14.199759326113117,
      "grad_norm": 116647.828125,
      "learning_rate": 0.00019093204827751805,
      "loss": 7.5047,
      "step": 118000
    },
    {
      "epoch": 14.200962695547533,
      "grad_norm": 291826.875,
      "learning_rate": 0.00019093046526852988,
      "loss": 7.5897,
      "step": 118010
    },
    {
      "epoch": 14.202166064981949,
      "grad_norm": 178334.6875,
      "learning_rate": 0.00019092888212794292,
      "loss": 7.5844,
      "step": 118020
    },
    {
      "epoch": 14.203369434416366,
      "grad_norm": 109844.4140625,
      "learning_rate": 0.00019092729885575936,
      "loss": 7.5493,
      "step": 118030
    },
    {
      "epoch": 14.204572803850782,
      "grad_norm": 92265.140625,
      "learning_rate": 0.00019092571545198156,
      "loss": 7.6348,
      "step": 118040
    },
    {
      "epoch": 14.205776173285198,
      "grad_norm": 487097.03125,
      "learning_rate": 0.00019092413191661183,
      "loss": 7.5966,
      "step": 118050
    },
    {
      "epoch": 14.206979542719615,
      "grad_norm": 344763.875,
      "learning_rate": 0.0001909225482496524,
      "loss": 7.6262,
      "step": 118060
    },
    {
      "epoch": 14.208182912154031,
      "grad_norm": 224788.859375,
      "learning_rate": 0.00019092096445110562,
      "loss": 7.5832,
      "step": 118070
    },
    {
      "epoch": 14.209386281588447,
      "grad_norm": 72774.234375,
      "learning_rate": 0.00019091938052097373,
      "loss": 7.6131,
      "step": 118080
    },
    {
      "epoch": 14.210589651022865,
      "grad_norm": 148898.328125,
      "learning_rate": 0.0001909177964592591,
      "loss": 7.6359,
      "step": 118090
    },
    {
      "epoch": 14.21179302045728,
      "grad_norm": 136293.4375,
      "learning_rate": 0.00019091621226596393,
      "loss": 7.6048,
      "step": 118100
    },
    {
      "epoch": 14.212996389891696,
      "grad_norm": 124320.875,
      "learning_rate": 0.00019091462794109055,
      "loss": 7.6171,
      "step": 118110
    },
    {
      "epoch": 14.214199759326114,
      "grad_norm": 119047.1328125,
      "learning_rate": 0.00019091304348464126,
      "loss": 7.5196,
      "step": 118120
    },
    {
      "epoch": 14.21540312876053,
      "grad_norm": 725180.8125,
      "learning_rate": 0.00019091145889661837,
      "loss": 7.6205,
      "step": 118130
    },
    {
      "epoch": 14.216606498194945,
      "grad_norm": 100127.4296875,
      "learning_rate": 0.00019090987417702413,
      "loss": 7.5324,
      "step": 118140
    },
    {
      "epoch": 14.217809867629363,
      "grad_norm": 185144.703125,
      "learning_rate": 0.00019090828932586087,
      "loss": 7.6019,
      "step": 118150
    },
    {
      "epoch": 14.219013237063779,
      "grad_norm": 3687808.5,
      "learning_rate": 0.00019090670434313088,
      "loss": 7.6518,
      "step": 118160
    },
    {
      "epoch": 14.220216606498195,
      "grad_norm": 104892.1953125,
      "learning_rate": 0.00019090511922883641,
      "loss": 7.5703,
      "step": 118170
    },
    {
      "epoch": 14.221419975932612,
      "grad_norm": 37905.5546875,
      "learning_rate": 0.00019090353398297985,
      "loss": 7.6244,
      "step": 118180
    },
    {
      "epoch": 14.222623345367028,
      "grad_norm": 34009.3671875,
      "learning_rate": 0.0001909019486055634,
      "loss": 7.6278,
      "step": 118190
    },
    {
      "epoch": 14.223826714801444,
      "grad_norm": 143581.0,
      "learning_rate": 0.00019090036309658937,
      "loss": 7.5578,
      "step": 118200
    },
    {
      "epoch": 14.225030084235861,
      "grad_norm": 6221499.0,
      "learning_rate": 0.0001908987774560601,
      "loss": 7.5906,
      "step": 118210
    },
    {
      "epoch": 14.226233453670277,
      "grad_norm": 75237.9140625,
      "learning_rate": 0.0001908971916839778,
      "loss": 7.5605,
      "step": 118220
    },
    {
      "epoch": 14.227436823104693,
      "grad_norm": 1089059.5,
      "learning_rate": 0.0001908956057803449,
      "loss": 7.6115,
      "step": 118230
    },
    {
      "epoch": 14.22864019253911,
      "grad_norm": 412480.71875,
      "learning_rate": 0.00019089401974516356,
      "loss": 7.6311,
      "step": 118240
    },
    {
      "epoch": 14.229843561973526,
      "grad_norm": 482569.3125,
      "learning_rate": 0.00019089243357843616,
      "loss": 7.6057,
      "step": 118250
    },
    {
      "epoch": 14.231046931407942,
      "grad_norm": 226829.71875,
      "learning_rate": 0.00019089084728016498,
      "loss": 7.6026,
      "step": 118260
    },
    {
      "epoch": 14.232250300842358,
      "grad_norm": 164830.5,
      "learning_rate": 0.00019088926085035225,
      "loss": 7.639,
      "step": 118270
    },
    {
      "epoch": 14.233453670276775,
      "grad_norm": 5754952.5,
      "learning_rate": 0.00019088767428900036,
      "loss": 7.5471,
      "step": 118280
    },
    {
      "epoch": 14.234657039711191,
      "grad_norm": 191127.46875,
      "learning_rate": 0.00019088608759611157,
      "loss": 7.7126,
      "step": 118290
    },
    {
      "epoch": 14.235860409145607,
      "grad_norm": 4961701.0,
      "learning_rate": 0.0001908845007716882,
      "loss": 7.6296,
      "step": 118300
    },
    {
      "epoch": 14.237063778580024,
      "grad_norm": 250321.875,
      "learning_rate": 0.00019088291381573247,
      "loss": 7.6384,
      "step": 118310
    },
    {
      "epoch": 14.23826714801444,
      "grad_norm": 125380.984375,
      "learning_rate": 0.00019088132672824673,
      "loss": 7.6243,
      "step": 118320
    },
    {
      "epoch": 14.239470517448856,
      "grad_norm": 346098.59375,
      "learning_rate": 0.00019087973950923325,
      "loss": 7.5883,
      "step": 118330
    },
    {
      "epoch": 14.240673886883274,
      "grad_norm": 197186.375,
      "learning_rate": 0.0001908781521586944,
      "loss": 7.6825,
      "step": 118340
    },
    {
      "epoch": 14.24187725631769,
      "grad_norm": 198040.796875,
      "learning_rate": 0.0001908765646766324,
      "loss": 7.6941,
      "step": 118350
    },
    {
      "epoch": 14.243080625752105,
      "grad_norm": 398055.09375,
      "learning_rate": 0.00019087497706304958,
      "loss": 7.58,
      "step": 118360
    },
    {
      "epoch": 14.244283995186523,
      "grad_norm": 284740.65625,
      "learning_rate": 0.00019087338931794823,
      "loss": 7.4917,
      "step": 118370
    },
    {
      "epoch": 14.245487364620939,
      "grad_norm": 151272.890625,
      "learning_rate": 0.00019087180144133065,
      "loss": 7.6321,
      "step": 118380
    },
    {
      "epoch": 14.246690734055354,
      "grad_norm": 427842.3125,
      "learning_rate": 0.00019087021343319914,
      "loss": 7.5916,
      "step": 118390
    },
    {
      "epoch": 14.247894103489772,
      "grad_norm": 1271051.125,
      "learning_rate": 0.000190868625293556,
      "loss": 7.5778,
      "step": 118400
    },
    {
      "epoch": 14.249097472924188,
      "grad_norm": 1155579.125,
      "learning_rate": 0.00019086703702240354,
      "loss": 7.5796,
      "step": 118410
    },
    {
      "epoch": 14.250300842358604,
      "grad_norm": 688704.5,
      "learning_rate": 0.000190865448619744,
      "loss": 7.5713,
      "step": 118420
    },
    {
      "epoch": 14.251504211793021,
      "grad_norm": 2906671.5,
      "learning_rate": 0.00019086386008557976,
      "loss": 7.6762,
      "step": 118430
    },
    {
      "epoch": 14.252707581227437,
      "grad_norm": 2189865.0,
      "learning_rate": 0.00019086227141991305,
      "loss": 7.6628,
      "step": 118440
    },
    {
      "epoch": 14.253910950661853,
      "grad_norm": 262524.40625,
      "learning_rate": 0.00019086068262274622,
      "loss": 7.5942,
      "step": 118450
    },
    {
      "epoch": 14.25511432009627,
      "grad_norm": 2990643.5,
      "learning_rate": 0.00019085909369408156,
      "loss": 7.6678,
      "step": 118460
    },
    {
      "epoch": 14.256317689530686,
      "grad_norm": 769595.125,
      "learning_rate": 0.00019085750463392133,
      "loss": 7.6162,
      "step": 118470
    },
    {
      "epoch": 14.257521058965102,
      "grad_norm": 525261.9375,
      "learning_rate": 0.0001908559154422679,
      "loss": 7.6189,
      "step": 118480
    },
    {
      "epoch": 14.25872442839952,
      "grad_norm": 268710.96875,
      "learning_rate": 0.00019085432611912347,
      "loss": 7.6095,
      "step": 118490
    },
    {
      "epoch": 14.259927797833935,
      "grad_norm": 555324.25,
      "learning_rate": 0.00019085273666449046,
      "loss": 7.6909,
      "step": 118500
    },
    {
      "epoch": 14.261131167268351,
      "grad_norm": 384020.65625,
      "learning_rate": 0.00019085114707837107,
      "loss": 7.5423,
      "step": 118510
    },
    {
      "epoch": 14.262334536702769,
      "grad_norm": 3127925.0,
      "learning_rate": 0.00019084955736076763,
      "loss": 7.596,
      "step": 118520
    },
    {
      "epoch": 14.263537906137184,
      "grad_norm": 8664014.0,
      "learning_rate": 0.0001908479675116825,
      "loss": 7.6043,
      "step": 118530
    },
    {
      "epoch": 14.2647412755716,
      "grad_norm": 192653.078125,
      "learning_rate": 0.00019084637753111788,
      "loss": 7.5048,
      "step": 118540
    },
    {
      "epoch": 14.265944645006018,
      "grad_norm": 480439.0,
      "learning_rate": 0.00019084478741907614,
      "loss": 7.6826,
      "step": 118550
    },
    {
      "epoch": 14.267148014440433,
      "grad_norm": 1125248.0,
      "learning_rate": 0.00019084319717555957,
      "loss": 7.6078,
      "step": 118560
    },
    {
      "epoch": 14.26835138387485,
      "grad_norm": 2509802.5,
      "learning_rate": 0.00019084160680057048,
      "loss": 7.5447,
      "step": 118570
    },
    {
      "epoch": 14.269554753309265,
      "grad_norm": 1077061.875,
      "learning_rate": 0.00019084001629411114,
      "loss": 7.5356,
      "step": 118580
    },
    {
      "epoch": 14.270758122743683,
      "grad_norm": 182195.296875,
      "learning_rate": 0.0001908384256561839,
      "loss": 7.5576,
      "step": 118590
    },
    {
      "epoch": 14.271961492178098,
      "grad_norm": 2173710.75,
      "learning_rate": 0.00019083683488679097,
      "loss": 7.5693,
      "step": 118600
    },
    {
      "epoch": 14.273164861612514,
      "grad_norm": 3091036.0,
      "learning_rate": 0.00019083524398593477,
      "loss": 7.5223,
      "step": 118610
    },
    {
      "epoch": 14.274368231046932,
      "grad_norm": 1276458.625,
      "learning_rate": 0.00019083365295361748,
      "loss": 7.6413,
      "step": 118620
    },
    {
      "epoch": 14.275571600481348,
      "grad_norm": 657346.9375,
      "learning_rate": 0.00019083206178984152,
      "loss": 7.487,
      "step": 118630
    },
    {
      "epoch": 14.276774969915763,
      "grad_norm": 271066.5,
      "learning_rate": 0.0001908304704946091,
      "loss": 7.5473,
      "step": 118640
    },
    {
      "epoch": 14.277978339350181,
      "grad_norm": 1240744.5,
      "learning_rate": 0.0001908288790679226,
      "loss": 7.5293,
      "step": 118650
    },
    {
      "epoch": 14.279181708784597,
      "grad_norm": 2022690.625,
      "learning_rate": 0.00019082728750978429,
      "loss": 7.6134,
      "step": 118660
    },
    {
      "epoch": 14.280385078219012,
      "grad_norm": 432702.25,
      "learning_rate": 0.00019082569582019648,
      "loss": 7.605,
      "step": 118670
    },
    {
      "epoch": 14.28158844765343,
      "grad_norm": 2926700.25,
      "learning_rate": 0.00019082410399916144,
      "loss": 7.5629,
      "step": 118680
    },
    {
      "epoch": 14.282791817087846,
      "grad_norm": 207177.15625,
      "learning_rate": 0.0001908225120466815,
      "loss": 7.587,
      "step": 118690
    },
    {
      "epoch": 14.283995186522262,
      "grad_norm": 1329514.0,
      "learning_rate": 0.00019082091996275896,
      "loss": 7.5791,
      "step": 118700
    },
    {
      "epoch": 14.28519855595668,
      "grad_norm": 2284441.75,
      "learning_rate": 0.00019081932774739612,
      "loss": 7.6031,
      "step": 118710
    },
    {
      "epoch": 14.286401925391095,
      "grad_norm": 594683.125,
      "learning_rate": 0.0001908177354005953,
      "loss": 7.584,
      "step": 118720
    },
    {
      "epoch": 14.28760529482551,
      "grad_norm": 1597930.875,
      "learning_rate": 0.0001908161429223588,
      "loss": 7.6239,
      "step": 118730
    },
    {
      "epoch": 14.288808664259928,
      "grad_norm": 108540.125,
      "learning_rate": 0.0001908145503126889,
      "loss": 7.5947,
      "step": 118740
    },
    {
      "epoch": 14.290012033694344,
      "grad_norm": 17705258.0,
      "learning_rate": 0.00019081295757158794,
      "loss": 7.5908,
      "step": 118750
    },
    {
      "epoch": 14.29121540312876,
      "grad_norm": 1176973.0,
      "learning_rate": 0.00019081136469905824,
      "loss": 7.5876,
      "step": 118760
    },
    {
      "epoch": 14.292418772563177,
      "grad_norm": 1378779.5,
      "learning_rate": 0.000190809771695102,
      "loss": 7.5621,
      "step": 118770
    },
    {
      "epoch": 14.293622141997593,
      "grad_norm": 680724.9375,
      "learning_rate": 0.00019080817855972164,
      "loss": 7.5074,
      "step": 118780
    },
    {
      "epoch": 14.294825511432009,
      "grad_norm": 521541.3125,
      "learning_rate": 0.00019080658529291943,
      "loss": 7.6662,
      "step": 118790
    },
    {
      "epoch": 14.296028880866427,
      "grad_norm": 754615.75,
      "learning_rate": 0.00019080499189469768,
      "loss": 7.5437,
      "step": 118800
    },
    {
      "epoch": 14.297232250300842,
      "grad_norm": 500591.46875,
      "learning_rate": 0.00019080339836505868,
      "loss": 7.5233,
      "step": 118810
    },
    {
      "epoch": 14.298435619735258,
      "grad_norm": 1726237.75,
      "learning_rate": 0.00019080180470400473,
      "loss": 7.6256,
      "step": 118820
    },
    {
      "epoch": 14.299638989169676,
      "grad_norm": 2642596.25,
      "learning_rate": 0.00019080021091153816,
      "loss": 7.5715,
      "step": 118830
    },
    {
      "epoch": 14.300842358604092,
      "grad_norm": 89558.421875,
      "learning_rate": 0.00019079861698766126,
      "loss": 7.6221,
      "step": 118840
    },
    {
      "epoch": 14.302045728038507,
      "grad_norm": 1838512.75,
      "learning_rate": 0.00019079702293237639,
      "loss": 7.5489,
      "step": 118850
    },
    {
      "epoch": 14.303249097472925,
      "grad_norm": 500271.28125,
      "learning_rate": 0.00019079542874568574,
      "loss": 7.5761,
      "step": 118860
    },
    {
      "epoch": 14.30445246690734,
      "grad_norm": 425299.125,
      "learning_rate": 0.00019079383442759172,
      "loss": 7.5812,
      "step": 118870
    },
    {
      "epoch": 14.305655836341757,
      "grad_norm": 181704.6875,
      "learning_rate": 0.0001907922399780966,
      "loss": 7.6447,
      "step": 118880
    },
    {
      "epoch": 14.306859205776174,
      "grad_norm": 908107.375,
      "learning_rate": 0.0001907906453972027,
      "loss": 7.45,
      "step": 118890
    },
    {
      "epoch": 14.30806257521059,
      "grad_norm": 325110.0,
      "learning_rate": 0.00019078905068491234,
      "loss": 7.5869,
      "step": 118900
    },
    {
      "epoch": 14.309265944645006,
      "grad_norm": 1366996.25,
      "learning_rate": 0.0001907874558412278,
      "loss": 7.5595,
      "step": 118910
    },
    {
      "epoch": 14.310469314079423,
      "grad_norm": 400215.5625,
      "learning_rate": 0.00019078586086615138,
      "loss": 7.639,
      "step": 118920
    },
    {
      "epoch": 14.311672683513839,
      "grad_norm": 97284.28125,
      "learning_rate": 0.0001907842657596854,
      "loss": 7.5516,
      "step": 118930
    },
    {
      "epoch": 14.312876052948255,
      "grad_norm": 736925.9375,
      "learning_rate": 0.00019078267052183217,
      "loss": 7.5797,
      "step": 118940
    },
    {
      "epoch": 14.314079422382672,
      "grad_norm": 3637919.25,
      "learning_rate": 0.00019078107515259403,
      "loss": 7.6444,
      "step": 118950
    },
    {
      "epoch": 14.315282791817088,
      "grad_norm": 223660.40625,
      "learning_rate": 0.00019077947965197325,
      "loss": 7.6054,
      "step": 118960
    },
    {
      "epoch": 14.316486161251504,
      "grad_norm": 1043933.5625,
      "learning_rate": 0.00019077788401997215,
      "loss": 7.5201,
      "step": 118970
    },
    {
      "epoch": 14.31768953068592,
      "grad_norm": 298401.21875,
      "learning_rate": 0.00019077628825659302,
      "loss": 7.5479,
      "step": 118980
    },
    {
      "epoch": 14.318892900120337,
      "grad_norm": 295401.25,
      "learning_rate": 0.0001907746923618382,
      "loss": 7.5631,
      "step": 118990
    },
    {
      "epoch": 14.320096269554753,
      "grad_norm": 2421919.25,
      "learning_rate": 0.00019077309633570998,
      "loss": 7.6099,
      "step": 119000
    },
    {
      "epoch": 14.321299638989169,
      "grad_norm": 1209339.875,
      "learning_rate": 0.0001907715001782107,
      "loss": 7.5773,
      "step": 119010
    },
    {
      "epoch": 14.322503008423586,
      "grad_norm": 158389.015625,
      "learning_rate": 0.00019076990388934264,
      "loss": 7.5278,
      "step": 119020
    },
    {
      "epoch": 14.323706377858002,
      "grad_norm": 141192.375,
      "learning_rate": 0.00019076830746910813,
      "loss": 7.6169,
      "step": 119030
    },
    {
      "epoch": 14.324909747292418,
      "grad_norm": 71631.265625,
      "learning_rate": 0.00019076671091750944,
      "loss": 7.5682,
      "step": 119040
    },
    {
      "epoch": 14.326113116726836,
      "grad_norm": 174709.296875,
      "learning_rate": 0.00019076511423454895,
      "loss": 7.5451,
      "step": 119050
    },
    {
      "epoch": 14.327316486161251,
      "grad_norm": 228096.328125,
      "learning_rate": 0.0001907635174202289,
      "loss": 7.6053,
      "step": 119060
    },
    {
      "epoch": 14.328519855595667,
      "grad_norm": 202864.921875,
      "learning_rate": 0.00019076192047455163,
      "loss": 7.6161,
      "step": 119070
    },
    {
      "epoch": 14.329723225030085,
      "grad_norm": 255449.78125,
      "learning_rate": 0.00019076032339751948,
      "loss": 7.5969,
      "step": 119080
    },
    {
      "epoch": 14.3309265944645,
      "grad_norm": 870253.25,
      "learning_rate": 0.0001907587261891347,
      "loss": 7.5154,
      "step": 119090
    },
    {
      "epoch": 14.332129963898916,
      "grad_norm": 311781.03125,
      "learning_rate": 0.00019075712884939966,
      "loss": 7.5915,
      "step": 119100
    },
    {
      "epoch": 14.333333333333334,
      "grad_norm": 782481.75,
      "learning_rate": 0.00019075553137831665,
      "loss": 7.6192,
      "step": 119110
    },
    {
      "epoch": 14.33453670276775,
      "grad_norm": 1564174.0,
      "learning_rate": 0.00019075393377588796,
      "loss": 7.5312,
      "step": 119120
    },
    {
      "epoch": 14.335740072202166,
      "grad_norm": 99209.40625,
      "learning_rate": 0.00019075233604211593,
      "loss": 7.595,
      "step": 119130
    },
    {
      "epoch": 14.336943441636583,
      "grad_norm": 31667.69140625,
      "learning_rate": 0.00019075073817700287,
      "loss": 7.6006,
      "step": 119140
    },
    {
      "epoch": 14.338146811070999,
      "grad_norm": 8151985.0,
      "learning_rate": 0.00019074914018055107,
      "loss": 7.5678,
      "step": 119150
    },
    {
      "epoch": 14.339350180505415,
      "grad_norm": 2709748.75,
      "learning_rate": 0.00019074754205276287,
      "loss": 7.574,
      "step": 119160
    },
    {
      "epoch": 14.340553549939832,
      "grad_norm": 1194706.875,
      "learning_rate": 0.00019074594379364057,
      "loss": 7.5266,
      "step": 119170
    },
    {
      "epoch": 14.341756919374248,
      "grad_norm": 1466402.625,
      "learning_rate": 0.00019074434540318648,
      "loss": 7.5475,
      "step": 119180
    },
    {
      "epoch": 14.342960288808664,
      "grad_norm": 4018239.25,
      "learning_rate": 0.00019074274688140293,
      "loss": 7.5761,
      "step": 119190
    },
    {
      "epoch": 14.344163658243081,
      "grad_norm": 109044.6484375,
      "learning_rate": 0.00019074114822829224,
      "loss": 7.5674,
      "step": 119200
    },
    {
      "epoch": 14.345367027677497,
      "grad_norm": 430628.3125,
      "learning_rate": 0.00019073954944385667,
      "loss": 7.6097,
      "step": 119210
    },
    {
      "epoch": 14.346570397111913,
      "grad_norm": 133542.625,
      "learning_rate": 0.00019073795052809857,
      "loss": 7.5421,
      "step": 119220
    },
    {
      "epoch": 14.34777376654633,
      "grad_norm": 69039.859375,
      "learning_rate": 0.00019073635148102026,
      "loss": 7.6095,
      "step": 119230
    },
    {
      "epoch": 14.348977135980746,
      "grad_norm": 19254.169921875,
      "learning_rate": 0.00019073475230262405,
      "loss": 7.5676,
      "step": 119240
    },
    {
      "epoch": 14.350180505415162,
      "grad_norm": 23850.453125,
      "learning_rate": 0.00019073315299291226,
      "loss": 7.5821,
      "step": 119250
    },
    {
      "epoch": 14.35138387484958,
      "grad_norm": 2005147.25,
      "learning_rate": 0.0001907315535518872,
      "loss": 7.5356,
      "step": 119260
    },
    {
      "epoch": 14.352587244283995,
      "grad_norm": 62143.3046875,
      "learning_rate": 0.00019072995397955117,
      "loss": 7.5563,
      "step": 119270
    },
    {
      "epoch": 14.353790613718411,
      "grad_norm": 132732.40625,
      "learning_rate": 0.00019072835427590647,
      "loss": 7.5333,
      "step": 119280
    },
    {
      "epoch": 14.354993983152827,
      "grad_norm": 409001.1875,
      "learning_rate": 0.00019072675444095549,
      "loss": 7.5244,
      "step": 119290
    },
    {
      "epoch": 14.356197352587245,
      "grad_norm": 612200.6875,
      "learning_rate": 0.00019072515447470046,
      "loss": 7.6445,
      "step": 119300
    },
    {
      "epoch": 14.35740072202166,
      "grad_norm": 60181.74609375,
      "learning_rate": 0.00019072355437714373,
      "loss": 7.6717,
      "step": 119310
    },
    {
      "epoch": 14.358604091456076,
      "grad_norm": 206314.375,
      "learning_rate": 0.00019072195414828765,
      "loss": 7.6011,
      "step": 119320
    },
    {
      "epoch": 14.359807460890494,
      "grad_norm": 59793.54296875,
      "learning_rate": 0.0001907203537881345,
      "loss": 7.565,
      "step": 119330
    },
    {
      "epoch": 14.36101083032491,
      "grad_norm": 13689.9794921875,
      "learning_rate": 0.00019071875329668657,
      "loss": 7.5052,
      "step": 119340
    },
    {
      "epoch": 14.362214199759325,
      "grad_norm": 71346.2109375,
      "learning_rate": 0.00019071715267394624,
      "loss": 7.5693,
      "step": 119350
    },
    {
      "epoch": 14.363417569193743,
      "grad_norm": 139831.796875,
      "learning_rate": 0.00019071555191991574,
      "loss": 7.5218,
      "step": 119360
    },
    {
      "epoch": 14.364620938628159,
      "grad_norm": 329947.125,
      "learning_rate": 0.00019071395103459747,
      "loss": 7.6216,
      "step": 119370
    },
    {
      "epoch": 14.365824308062574,
      "grad_norm": 574487.875,
      "learning_rate": 0.00019071235001799376,
      "loss": 7.6086,
      "step": 119380
    },
    {
      "epoch": 14.367027677496992,
      "grad_norm": 21661.830078125,
      "learning_rate": 0.00019071074887010684,
      "loss": 7.5997,
      "step": 119390
    },
    {
      "epoch": 14.368231046931408,
      "grad_norm": 12835.291015625,
      "learning_rate": 0.00019070914759093907,
      "loss": 7.563,
      "step": 119400
    },
    {
      "epoch": 14.369434416365824,
      "grad_norm": 560535.125,
      "learning_rate": 0.0001907075461804928,
      "loss": 7.5462,
      "step": 119410
    },
    {
      "epoch": 14.370637785800241,
      "grad_norm": 17859.89453125,
      "learning_rate": 0.00019070594463877028,
      "loss": 7.5947,
      "step": 119420
    },
    {
      "epoch": 14.371841155234657,
      "grad_norm": 153670.6875,
      "learning_rate": 0.00019070434296577387,
      "loss": 7.5563,
      "step": 119430
    },
    {
      "epoch": 14.373044524669073,
      "grad_norm": 51824.6796875,
      "learning_rate": 0.00019070274116150587,
      "loss": 7.5882,
      "step": 119440
    },
    {
      "epoch": 14.37424789410349,
      "grad_norm": 70184.84375,
      "learning_rate": 0.00019070113922596863,
      "loss": 7.5978,
      "step": 119450
    },
    {
      "epoch": 14.375451263537906,
      "grad_norm": 26208.642578125,
      "learning_rate": 0.00019069953715916444,
      "loss": 7.6352,
      "step": 119460
    },
    {
      "epoch": 14.376654632972322,
      "grad_norm": 6632.07373046875,
      "learning_rate": 0.0001906979349610956,
      "loss": 7.5943,
      "step": 119470
    },
    {
      "epoch": 14.37785800240674,
      "grad_norm": 21841.8359375,
      "learning_rate": 0.00019069633263176452,
      "loss": 7.6114,
      "step": 119480
    },
    {
      "epoch": 14.379061371841155,
      "grad_norm": 3858.576904296875,
      "learning_rate": 0.0001906947301711734,
      "loss": 7.5277,
      "step": 119490
    },
    {
      "epoch": 14.380264741275571,
      "grad_norm": 4729.509765625,
      "learning_rate": 0.00019069312757932465,
      "loss": 7.4972,
      "step": 119500
    },
    {
      "epoch": 14.381468110709989,
      "grad_norm": 19417.59375,
      "learning_rate": 0.00019069152485622052,
      "loss": 7.609,
      "step": 119510
    },
    {
      "epoch": 14.382671480144404,
      "grad_norm": 17170.166015625,
      "learning_rate": 0.0001906899220018634,
      "loss": 7.6092,
      "step": 119520
    },
    {
      "epoch": 14.38387484957882,
      "grad_norm": 3339.51953125,
      "learning_rate": 0.00019068831901625555,
      "loss": 7.5038,
      "step": 119530
    },
    {
      "epoch": 14.385078219013238,
      "grad_norm": 5656.55078125,
      "learning_rate": 0.0001906867158993993,
      "loss": 7.5383,
      "step": 119540
    },
    {
      "epoch": 14.386281588447654,
      "grad_norm": 5327.35791015625,
      "learning_rate": 0.00019068511265129702,
      "loss": 7.533,
      "step": 119550
    },
    {
      "epoch": 14.38748495788207,
      "grad_norm": 5790.3759765625,
      "learning_rate": 0.00019068350927195095,
      "loss": 7.5724,
      "step": 119560
    },
    {
      "epoch": 14.388688327316487,
      "grad_norm": 3798.71435546875,
      "learning_rate": 0.00019068190576136345,
      "loss": 7.5458,
      "step": 119570
    },
    {
      "epoch": 14.389891696750903,
      "grad_norm": 6133.72314453125,
      "learning_rate": 0.0001906803021195369,
      "loss": 7.5392,
      "step": 119580
    },
    {
      "epoch": 14.391095066185319,
      "grad_norm": 2831.714111328125,
      "learning_rate": 0.0001906786983464735,
      "loss": 7.4844,
      "step": 119590
    },
    {
      "epoch": 14.392298435619736,
      "grad_norm": 1254.361083984375,
      "learning_rate": 0.00019067709444217567,
      "loss": 7.592,
      "step": 119600
    },
    {
      "epoch": 14.393501805054152,
      "grad_norm": 156516.84375,
      "learning_rate": 0.0001906754904066457,
      "loss": 7.5673,
      "step": 119610
    },
    {
      "epoch": 14.394705174488568,
      "grad_norm": 870.5654907226562,
      "learning_rate": 0.0001906738862398859,
      "loss": 7.4808,
      "step": 119620
    },
    {
      "epoch": 14.395908543922985,
      "grad_norm": 9966.8095703125,
      "learning_rate": 0.00019067228194189859,
      "loss": 7.4807,
      "step": 119630
    },
    {
      "epoch": 14.397111913357401,
      "grad_norm": 2594.8515625,
      "learning_rate": 0.00019067067751268613,
      "loss": 7.505,
      "step": 119640
    },
    {
      "epoch": 14.398315282791817,
      "grad_norm": 12772.3779296875,
      "learning_rate": 0.0001906690729522508,
      "loss": 7.5407,
      "step": 119650
    },
    {
      "epoch": 14.399518652226233,
      "grad_norm": 58876.48046875,
      "learning_rate": 0.00019066746826059493,
      "loss": 7.5641,
      "step": 119660
    },
    {
      "epoch": 14.40072202166065,
      "grad_norm": 11685.4697265625,
      "learning_rate": 0.00019066586343772084,
      "loss": 7.5972,
      "step": 119670
    },
    {
      "epoch": 14.401925391095066,
      "grad_norm": 17592.52734375,
      "learning_rate": 0.00019066425848363088,
      "loss": 7.5208,
      "step": 119680
    },
    {
      "epoch": 14.403128760529482,
      "grad_norm": 16205.251953125,
      "learning_rate": 0.00019066265339832733,
      "loss": 7.5903,
      "step": 119690
    },
    {
      "epoch": 14.4043321299639,
      "grad_norm": 20633.93359375,
      "learning_rate": 0.00019066104818181255,
      "loss": 7.5114,
      "step": 119700
    },
    {
      "epoch": 14.405535499398315,
      "grad_norm": 38214.70703125,
      "learning_rate": 0.00019065944283408885,
      "loss": 7.5092,
      "step": 119710
    },
    {
      "epoch": 14.406738868832731,
      "grad_norm": 8815.671875,
      "learning_rate": 0.00019065783735515857,
      "loss": 7.4565,
      "step": 119720
    },
    {
      "epoch": 14.407942238267148,
      "grad_norm": 8079.5439453125,
      "learning_rate": 0.000190656231745024,
      "loss": 7.553,
      "step": 119730
    },
    {
      "epoch": 14.409145607701564,
      "grad_norm": 9402.7861328125,
      "learning_rate": 0.00019065462600368748,
      "loss": 7.5981,
      "step": 119740
    },
    {
      "epoch": 14.41034897713598,
      "grad_norm": 26985.255859375,
      "learning_rate": 0.00019065302013115135,
      "loss": 7.5289,
      "step": 119750
    },
    {
      "epoch": 14.411552346570398,
      "grad_norm": 15124.28125,
      "learning_rate": 0.00019065141412741787,
      "loss": 7.5674,
      "step": 119760
    },
    {
      "epoch": 14.412755716004813,
      "grad_norm": 17110.228515625,
      "learning_rate": 0.00019064980799248947,
      "loss": 7.5469,
      "step": 119770
    },
    {
      "epoch": 14.41395908543923,
      "grad_norm": 9252.294921875,
      "learning_rate": 0.00019064820172636838,
      "loss": 7.6501,
      "step": 119780
    },
    {
      "epoch": 14.415162454873647,
      "grad_norm": 11879.595703125,
      "learning_rate": 0.00019064659532905698,
      "loss": 7.5487,
      "step": 119790
    },
    {
      "epoch": 14.416365824308063,
      "grad_norm": 20099.056640625,
      "learning_rate": 0.00019064498880055756,
      "loss": 7.5602,
      "step": 119800
    },
    {
      "epoch": 14.417569193742478,
      "grad_norm": 34142.984375,
      "learning_rate": 0.0001906433821408725,
      "loss": 7.6255,
      "step": 119810
    },
    {
      "epoch": 14.418772563176896,
      "grad_norm": 6366.70751953125,
      "learning_rate": 0.00019064177535000406,
      "loss": 7.5919,
      "step": 119820
    },
    {
      "epoch": 14.419975932611312,
      "grad_norm": 25508.58984375,
      "learning_rate": 0.00019064016842795458,
      "loss": 7.6212,
      "step": 119830
    },
    {
      "epoch": 14.421179302045728,
      "grad_norm": 6327.87158203125,
      "learning_rate": 0.0001906385613747264,
      "loss": 7.5441,
      "step": 119840
    },
    {
      "epoch": 14.422382671480145,
      "grad_norm": 7255.55322265625,
      "learning_rate": 0.00019063695419032186,
      "loss": 7.5433,
      "step": 119850
    },
    {
      "epoch": 14.42358604091456,
      "grad_norm": 5273.5888671875,
      "learning_rate": 0.00019063534687474326,
      "loss": 7.5321,
      "step": 119860
    },
    {
      "epoch": 14.424789410348977,
      "grad_norm": 2021.651123046875,
      "learning_rate": 0.00019063373942799293,
      "loss": 7.5048,
      "step": 119870
    },
    {
      "epoch": 14.425992779783394,
      "grad_norm": 1237.4359130859375,
      "learning_rate": 0.00019063213185007322,
      "loss": 7.622,
      "step": 119880
    },
    {
      "epoch": 14.42719614921781,
      "grad_norm": 2026.2740478515625,
      "learning_rate": 0.00019063052414098643,
      "loss": 7.5603,
      "step": 119890
    },
    {
      "epoch": 14.428399518652226,
      "grad_norm": 4687.3798828125,
      "learning_rate": 0.00019062891630073492,
      "loss": 7.4505,
      "step": 119900
    },
    {
      "epoch": 14.429602888086643,
      "grad_norm": 3074.654052734375,
      "learning_rate": 0.00019062730832932097,
      "loss": 7.5527,
      "step": 119910
    },
    {
      "epoch": 14.43080625752106,
      "grad_norm": 1889.7685546875,
      "learning_rate": 0.0001906257002267469,
      "loss": 7.566,
      "step": 119920
    },
    {
      "epoch": 14.432009626955475,
      "grad_norm": 3460.92236328125,
      "learning_rate": 0.00019062409199301512,
      "loss": 7.5607,
      "step": 119930
    },
    {
      "epoch": 14.433212996389893,
      "grad_norm": 3774.3955078125,
      "learning_rate": 0.00019062248362812788,
      "loss": 7.5595,
      "step": 119940
    },
    {
      "epoch": 14.434416365824308,
      "grad_norm": 3955.254150390625,
      "learning_rate": 0.00019062087513208753,
      "loss": 7.4998,
      "step": 119950
    },
    {
      "epoch": 14.435619735258724,
      "grad_norm": 65530.7109375,
      "learning_rate": 0.0001906192665048964,
      "loss": 7.5635,
      "step": 119960
    },
    {
      "epoch": 14.43682310469314,
      "grad_norm": 2011.41845703125,
      "learning_rate": 0.0001906176577465568,
      "loss": 7.5687,
      "step": 119970
    },
    {
      "epoch": 14.438026474127557,
      "grad_norm": 3497.224853515625,
      "learning_rate": 0.00019061604885707115,
      "loss": 7.568,
      "step": 119980
    },
    {
      "epoch": 14.439229843561973,
      "grad_norm": 2236.07373046875,
      "learning_rate": 0.00019061443983644163,
      "loss": 7.6512,
      "step": 119990
    },
    {
      "epoch": 14.440433212996389,
      "grad_norm": 3275.374755859375,
      "learning_rate": 0.00019061283068467065,
      "loss": 7.5944,
      "step": 120000
    },
    {
      "epoch": 14.441636582430807,
      "grad_norm": 2291.748779296875,
      "learning_rate": 0.00019061122140176057,
      "loss": 7.5395,
      "step": 120010
    },
    {
      "epoch": 14.442839951865222,
      "grad_norm": 3766.966064453125,
      "learning_rate": 0.00019060961198771366,
      "loss": 7.5335,
      "step": 120020
    },
    {
      "epoch": 14.444043321299638,
      "grad_norm": 1537.174560546875,
      "learning_rate": 0.00019060800244253225,
      "loss": 7.6213,
      "step": 120030
    },
    {
      "epoch": 14.445246690734056,
      "grad_norm": 1786.9052734375,
      "learning_rate": 0.00019060639276621873,
      "loss": 7.5077,
      "step": 120040
    },
    {
      "epoch": 14.446450060168472,
      "grad_norm": 4729.9560546875,
      "learning_rate": 0.00019060478295877538,
      "loss": 7.467,
      "step": 120050
    },
    {
      "epoch": 14.447653429602887,
      "grad_norm": 70892.3671875,
      "learning_rate": 0.00019060317302020452,
      "loss": 7.5656,
      "step": 120060
    },
    {
      "epoch": 14.448856799037305,
      "grad_norm": 11207.615234375,
      "learning_rate": 0.00019060156295050852,
      "loss": 7.509,
      "step": 120070
    },
    {
      "epoch": 14.45006016847172,
      "grad_norm": 25483.3671875,
      "learning_rate": 0.00019059995274968964,
      "loss": 7.4925,
      "step": 120080
    },
    {
      "epoch": 14.451263537906136,
      "grad_norm": 9200.310546875,
      "learning_rate": 0.00019059834241775032,
      "loss": 7.5617,
      "step": 120090
    },
    {
      "epoch": 14.452466907340554,
      "grad_norm": 5665.3662109375,
      "learning_rate": 0.0001905967319546928,
      "loss": 7.5211,
      "step": 120100
    },
    {
      "epoch": 14.45367027677497,
      "grad_norm": 5527.3173828125,
      "learning_rate": 0.00019059512136051945,
      "loss": 7.5932,
      "step": 120110
    },
    {
      "epoch": 14.454873646209386,
      "grad_norm": 11837.158203125,
      "learning_rate": 0.0001905935106352326,
      "loss": 7.6151,
      "step": 120120
    },
    {
      "epoch": 14.456077015643803,
      "grad_norm": 20097.900390625,
      "learning_rate": 0.00019059189977883455,
      "loss": 7.5961,
      "step": 120130
    },
    {
      "epoch": 14.457280385078219,
      "grad_norm": 10696.216796875,
      "learning_rate": 0.0001905902887913277,
      "loss": 7.5447,
      "step": 120140
    },
    {
      "epoch": 14.458483754512635,
      "grad_norm": 13723.712890625,
      "learning_rate": 0.00019058867767271428,
      "loss": 7.5686,
      "step": 120150
    },
    {
      "epoch": 14.459687123947052,
      "grad_norm": 26544.7734375,
      "learning_rate": 0.00019058706642299672,
      "loss": 7.5666,
      "step": 120160
    },
    {
      "epoch": 14.460890493381468,
      "grad_norm": 11483.302734375,
      "learning_rate": 0.00019058545504217728,
      "loss": 7.6048,
      "step": 120170
    },
    {
      "epoch": 14.462093862815884,
      "grad_norm": 24804.689453125,
      "learning_rate": 0.00019058384353025837,
      "loss": 7.6328,
      "step": 120180
    },
    {
      "epoch": 14.463297232250302,
      "grad_norm": 11350.1083984375,
      "learning_rate": 0.00019058223188724222,
      "loss": 7.5416,
      "step": 120190
    },
    {
      "epoch": 14.464500601684717,
      "grad_norm": 7787.87548828125,
      "learning_rate": 0.00019058062011313124,
      "loss": 7.5303,
      "step": 120200
    },
    {
      "epoch": 14.465703971119133,
      "grad_norm": 129329.0703125,
      "learning_rate": 0.00019057900820792774,
      "loss": 7.5924,
      "step": 120210
    },
    {
      "epoch": 14.46690734055355,
      "grad_norm": 23617.984375,
      "learning_rate": 0.00019057739617163403,
      "loss": 7.5598,
      "step": 120220
    },
    {
      "epoch": 14.468110709987966,
      "grad_norm": 13259.3095703125,
      "learning_rate": 0.00019057578400425248,
      "loss": 7.54,
      "step": 120230
    },
    {
      "epoch": 14.469314079422382,
      "grad_norm": 8982.37109375,
      "learning_rate": 0.0001905741717057854,
      "loss": 7.4943,
      "step": 120240
    },
    {
      "epoch": 14.4705174488568,
      "grad_norm": 9860.9189453125,
      "learning_rate": 0.00019057255927623515,
      "loss": 7.593,
      "step": 120250
    },
    {
      "epoch": 14.471720818291216,
      "grad_norm": 33967.8984375,
      "learning_rate": 0.00019057094671560404,
      "loss": 7.5468,
      "step": 120260
    },
    {
      "epoch": 14.472924187725631,
      "grad_norm": 9480.3759765625,
      "learning_rate": 0.0001905693340238944,
      "loss": 7.6913,
      "step": 120270
    },
    {
      "epoch": 14.474127557160049,
      "grad_norm": 11778.146484375,
      "learning_rate": 0.0001905677212011086,
      "loss": 7.5879,
      "step": 120280
    },
    {
      "epoch": 14.475330926594465,
      "grad_norm": 10220.3251953125,
      "learning_rate": 0.00019056610824724892,
      "loss": 7.6608,
      "step": 120290
    },
    {
      "epoch": 14.47653429602888,
      "grad_norm": 6985.7392578125,
      "learning_rate": 0.00019056449516231772,
      "loss": 7.5828,
      "step": 120300
    },
    {
      "epoch": 14.477737665463298,
      "grad_norm": 29242.396484375,
      "learning_rate": 0.00019056288194631735,
      "loss": 7.564,
      "step": 120310
    },
    {
      "epoch": 14.478941034897714,
      "grad_norm": 8868.1533203125,
      "learning_rate": 0.0001905612685992501,
      "loss": 7.5048,
      "step": 120320
    },
    {
      "epoch": 14.48014440433213,
      "grad_norm": 5269.60400390625,
      "learning_rate": 0.0001905596551211184,
      "loss": 7.5593,
      "step": 120330
    },
    {
      "epoch": 14.481347773766545,
      "grad_norm": 37762.30859375,
      "learning_rate": 0.00019055804151192447,
      "loss": 7.4889,
      "step": 120340
    },
    {
      "epoch": 14.482551143200963,
      "grad_norm": 41898.69921875,
      "learning_rate": 0.00019055642777167072,
      "loss": 7.5272,
      "step": 120350
    },
    {
      "epoch": 14.483754512635379,
      "grad_norm": 15239.4765625,
      "learning_rate": 0.00019055481390035944,
      "loss": 7.6083,
      "step": 120360
    },
    {
      "epoch": 14.484957882069795,
      "grad_norm": 6722.9072265625,
      "learning_rate": 0.00019055319989799298,
      "loss": 7.5189,
      "step": 120370
    },
    {
      "epoch": 14.486161251504212,
      "grad_norm": 4932.267578125,
      "learning_rate": 0.0001905515857645737,
      "loss": 7.5753,
      "step": 120380
    },
    {
      "epoch": 14.487364620938628,
      "grad_norm": 5543.306640625,
      "learning_rate": 0.00019054997150010392,
      "loss": 7.5521,
      "step": 120390
    },
    {
      "epoch": 14.488567990373044,
      "grad_norm": 24199.45703125,
      "learning_rate": 0.000190548357104586,
      "loss": 7.5241,
      "step": 120400
    },
    {
      "epoch": 14.489771359807461,
      "grad_norm": 145155.375,
      "learning_rate": 0.0001905467425780222,
      "loss": 7.6052,
      "step": 120410
    },
    {
      "epoch": 14.490974729241877,
      "grad_norm": 35324.0234375,
      "learning_rate": 0.00019054512792041493,
      "loss": 7.4526,
      "step": 120420
    },
    {
      "epoch": 14.492178098676293,
      "grad_norm": 17105.146484375,
      "learning_rate": 0.0001905435131317665,
      "loss": 7.5068,
      "step": 120430
    },
    {
      "epoch": 14.49338146811071,
      "grad_norm": 12936.751953125,
      "learning_rate": 0.00019054189821207925,
      "loss": 7.5293,
      "step": 120440
    },
    {
      "epoch": 14.494584837545126,
      "grad_norm": 4785.9609375,
      "learning_rate": 0.00019054028316135556,
      "loss": 7.5097,
      "step": 120450
    },
    {
      "epoch": 14.495788206979542,
      "grad_norm": 2668.955810546875,
      "learning_rate": 0.00019053866797959766,
      "loss": 7.4817,
      "step": 120460
    },
    {
      "epoch": 14.49699157641396,
      "grad_norm": 4131.68310546875,
      "learning_rate": 0.000190537052666808,
      "loss": 7.5655,
      "step": 120470
    },
    {
      "epoch": 14.498194945848375,
      "grad_norm": 915.166015625,
      "learning_rate": 0.00019053543722298886,
      "loss": 7.4781,
      "step": 120480
    },
    {
      "epoch": 14.499398315282791,
      "grad_norm": 3766.930908203125,
      "learning_rate": 0.00019053382164814256,
      "loss": 7.6013,
      "step": 120490
    },
    {
      "epoch": 14.500601684717209,
      "grad_norm": 598.1075439453125,
      "learning_rate": 0.00019053220594227152,
      "loss": 7.4729,
      "step": 120500
    },
    {
      "epoch": 14.501805054151625,
      "grad_norm": 94520.7109375,
      "learning_rate": 0.00019053059010537798,
      "loss": 7.6828,
      "step": 120510
    },
    {
      "epoch": 14.50300842358604,
      "grad_norm": 2203.223876953125,
      "learning_rate": 0.00019052897413746438,
      "loss": 7.5277,
      "step": 120520
    },
    {
      "epoch": 14.504211793020458,
      "grad_norm": 4880.93896484375,
      "learning_rate": 0.00019052735803853297,
      "loss": 7.7061,
      "step": 120530
    },
    {
      "epoch": 14.505415162454874,
      "grad_norm": 7914.91259765625,
      "learning_rate": 0.00019052574180858614,
      "loss": 7.617,
      "step": 120540
    },
    {
      "epoch": 14.50661853188929,
      "grad_norm": 3421.029541015625,
      "learning_rate": 0.00019052412544762616,
      "loss": 7.5142,
      "step": 120550
    },
    {
      "epoch": 14.507821901323707,
      "grad_norm": 12922.5166015625,
      "learning_rate": 0.0001905225089556555,
      "loss": 7.447,
      "step": 120560
    },
    {
      "epoch": 14.509025270758123,
      "grad_norm": 1617.4580078125,
      "learning_rate": 0.00019052089233267635,
      "loss": 7.5523,
      "step": 120570
    },
    {
      "epoch": 14.510228640192539,
      "grad_norm": 8773.5224609375,
      "learning_rate": 0.00019051927557869113,
      "loss": 7.5148,
      "step": 120580
    },
    {
      "epoch": 14.511432009626956,
      "grad_norm": 5880.3486328125,
      "learning_rate": 0.00019051765869370222,
      "loss": 7.5477,
      "step": 120590
    },
    {
      "epoch": 14.512635379061372,
      "grad_norm": 2873.053466796875,
      "learning_rate": 0.00019051604167771188,
      "loss": 7.549,
      "step": 120600
    },
    {
      "epoch": 14.513838748495788,
      "grad_norm": 22872.376953125,
      "learning_rate": 0.00019051442453072245,
      "loss": 7.4861,
      "step": 120610
    },
    {
      "epoch": 14.515042117930205,
      "grad_norm": 4340.60498046875,
      "learning_rate": 0.00019051280725273633,
      "loss": 7.5089,
      "step": 120620
    },
    {
      "epoch": 14.516245487364621,
      "grad_norm": 3579.42626953125,
      "learning_rate": 0.00019051118984375585,
      "loss": 7.5295,
      "step": 120630
    },
    {
      "epoch": 14.517448856799037,
      "grad_norm": 4048.6767578125,
      "learning_rate": 0.0001905095723037833,
      "loss": 7.4164,
      "step": 120640
    },
    {
      "epoch": 14.518652226233453,
      "grad_norm": 4062.582275390625,
      "learning_rate": 0.00019050795463282106,
      "loss": 7.4519,
      "step": 120650
    },
    {
      "epoch": 14.51985559566787,
      "grad_norm": 3027.69189453125,
      "learning_rate": 0.00019050633683087147,
      "loss": 7.5603,
      "step": 120660
    },
    {
      "epoch": 14.521058965102286,
      "grad_norm": 9249.0751953125,
      "learning_rate": 0.00019050471889793684,
      "loss": 7.5184,
      "step": 120670
    },
    {
      "epoch": 14.522262334536702,
      "grad_norm": 7461.44189453125,
      "learning_rate": 0.0001905031008340196,
      "loss": 7.5324,
      "step": 120680
    },
    {
      "epoch": 14.52346570397112,
      "grad_norm": 4505.70166015625,
      "learning_rate": 0.00019050148263912196,
      "loss": 7.5441,
      "step": 120690
    },
    {
      "epoch": 14.524669073405535,
      "grad_norm": 1619.5029296875,
      "learning_rate": 0.00019049986431324636,
      "loss": 7.5592,
      "step": 120700
    },
    {
      "epoch": 14.525872442839951,
      "grad_norm": 4418.81591796875,
      "learning_rate": 0.0001904982458563951,
      "loss": 7.5768,
      "step": 120710
    },
    {
      "epoch": 14.527075812274369,
      "grad_norm": 2280.780029296875,
      "learning_rate": 0.00019049662726857057,
      "loss": 7.4688,
      "step": 120720
    },
    {
      "epoch": 14.528279181708784,
      "grad_norm": 1257.7911376953125,
      "learning_rate": 0.00019049500854977504,
      "loss": 7.4916,
      "step": 120730
    },
    {
      "epoch": 14.5294825511432,
      "grad_norm": 8332.865234375,
      "learning_rate": 0.00019049338970001092,
      "loss": 7.524,
      "step": 120740
    },
    {
      "epoch": 14.530685920577618,
      "grad_norm": 7059.63037109375,
      "learning_rate": 0.0001904917707192805,
      "loss": 7.5303,
      "step": 120750
    },
    {
      "epoch": 14.531889290012034,
      "grad_norm": 1096.492919921875,
      "learning_rate": 0.00019049015160758613,
      "loss": 7.4917,
      "step": 120760
    },
    {
      "epoch": 14.53309265944645,
      "grad_norm": 869.6895141601562,
      "learning_rate": 0.0001904885323649302,
      "loss": 7.5133,
      "step": 120770
    },
    {
      "epoch": 14.534296028880867,
      "grad_norm": 2343.33544921875,
      "learning_rate": 0.00019048691299131503,
      "loss": 7.6167,
      "step": 120780
    },
    {
      "epoch": 14.535499398315283,
      "grad_norm": 3463.762939453125,
      "learning_rate": 0.00019048529348674294,
      "loss": 7.6773,
      "step": 120790
    },
    {
      "epoch": 14.536702767749698,
      "grad_norm": 1520.4482421875,
      "learning_rate": 0.00019048367385121628,
      "loss": 7.7014,
      "step": 120800
    },
    {
      "epoch": 14.537906137184116,
      "grad_norm": 1333.6510009765625,
      "learning_rate": 0.0001904820540847374,
      "loss": 7.6907,
      "step": 120810
    },
    {
      "epoch": 14.539109506618532,
      "grad_norm": 671.68505859375,
      "learning_rate": 0.00019048043418730867,
      "loss": 7.5611,
      "step": 120820
    },
    {
      "epoch": 14.540312876052948,
      "grad_norm": 1106.837158203125,
      "learning_rate": 0.00019047881415893244,
      "loss": 7.5656,
      "step": 120830
    },
    {
      "epoch": 14.541516245487365,
      "grad_norm": 3465.642822265625,
      "learning_rate": 0.00019047719399961096,
      "loss": 7.6345,
      "step": 120840
    },
    {
      "epoch": 14.542719614921781,
      "grad_norm": 710.587890625,
      "learning_rate": 0.0001904755737093467,
      "loss": 7.5027,
      "step": 120850
    },
    {
      "epoch": 14.543922984356197,
      "grad_norm": 510.3826904296875,
      "learning_rate": 0.0001904739532881419,
      "loss": 7.8557,
      "step": 120860
    },
    {
      "epoch": 14.545126353790614,
      "grad_norm": 1216.6356201171875,
      "learning_rate": 0.00019047233273599898,
      "loss": 7.6288,
      "step": 120870
    },
    {
      "epoch": 14.54632972322503,
      "grad_norm": 320.2340393066406,
      "learning_rate": 0.00019047071205292025,
      "loss": 7.5113,
      "step": 120880
    },
    {
      "epoch": 14.547533092659446,
      "grad_norm": 1047.402099609375,
      "learning_rate": 0.00019046909123890808,
      "loss": 7.5235,
      "step": 120890
    },
    {
      "epoch": 14.548736462093864,
      "grad_norm": 507.4156188964844,
      "learning_rate": 0.0001904674702939648,
      "loss": 7.5328,
      "step": 120900
    },
    {
      "epoch": 14.54993983152828,
      "grad_norm": 3783.83837890625,
      "learning_rate": 0.00019046584921809272,
      "loss": 7.5166,
      "step": 120910
    },
    {
      "epoch": 14.551143200962695,
      "grad_norm": 516.8233642578125,
      "learning_rate": 0.00019046422801129425,
      "loss": 7.5846,
      "step": 120920
    },
    {
      "epoch": 14.552346570397113,
      "grad_norm": 513927.84375,
      "learning_rate": 0.0001904626066735717,
      "loss": 7.5438,
      "step": 120930
    },
    {
      "epoch": 14.553549939831528,
      "grad_norm": 1209770.0,
      "learning_rate": 0.00019046098520492742,
      "loss": 7.6522,
      "step": 120940
    },
    {
      "epoch": 14.554753309265944,
      "grad_norm": 1310284.0,
      "learning_rate": 0.00019045936360536375,
      "loss": 7.5744,
      "step": 120950
    },
    {
      "epoch": 14.555956678700362,
      "grad_norm": 5488427.5,
      "learning_rate": 0.00019045774187488305,
      "loss": 7.5469,
      "step": 120960
    },
    {
      "epoch": 14.557160048134778,
      "grad_norm": 2373651.0,
      "learning_rate": 0.00019045612001348768,
      "loss": 7.4876,
      "step": 120970
    },
    {
      "epoch": 14.558363417569193,
      "grad_norm": 467700.625,
      "learning_rate": 0.00019045449802118,
      "loss": 7.5919,
      "step": 120980
    },
    {
      "epoch": 14.559566787003611,
      "grad_norm": 779538.6875,
      "learning_rate": 0.00019045287589796226,
      "loss": 7.5552,
      "step": 120990
    },
    {
      "epoch": 14.560770156438027,
      "grad_norm": 2413643.5,
      "learning_rate": 0.0001904512536438369,
      "loss": 7.3939,
      "step": 121000
    },
    {
      "epoch": 14.561973525872443,
      "grad_norm": 2278255.5,
      "learning_rate": 0.00019044963125880627,
      "loss": 7.5379,
      "step": 121010
    },
    {
      "epoch": 14.56317689530686,
      "grad_norm": 639653.75,
      "learning_rate": 0.00019044800874287268,
      "loss": 7.6289,
      "step": 121020
    },
    {
      "epoch": 14.564380264741276,
      "grad_norm": 168440.578125,
      "learning_rate": 0.00019044638609603845,
      "loss": 7.5448,
      "step": 121030
    },
    {
      "epoch": 14.565583634175692,
      "grad_norm": 222035.640625,
      "learning_rate": 0.00019044476331830606,
      "loss": 7.6814,
      "step": 121040
    },
    {
      "epoch": 14.566787003610107,
      "grad_norm": 582913.375,
      "learning_rate": 0.00019044314040967767,
      "loss": 7.5201,
      "step": 121050
    },
    {
      "epoch": 14.567990373044525,
      "grad_norm": 35121.67578125,
      "learning_rate": 0.0001904415173701558,
      "loss": 7.5092,
      "step": 121060
    },
    {
      "epoch": 14.56919374247894,
      "grad_norm": 107542.0390625,
      "learning_rate": 0.0001904398941997427,
      "loss": 7.5003,
      "step": 121070
    },
    {
      "epoch": 14.570397111913357,
      "grad_norm": 314492.96875,
      "learning_rate": 0.00019043827089844072,
      "loss": 7.5012,
      "step": 121080
    },
    {
      "epoch": 14.571600481347774,
      "grad_norm": 84995.375,
      "learning_rate": 0.00019043664746625226,
      "loss": 7.6043,
      "step": 121090
    },
    {
      "epoch": 14.57280385078219,
      "grad_norm": 100890.8125,
      "learning_rate": 0.00019043502390317962,
      "loss": 7.5499,
      "step": 121100
    },
    {
      "epoch": 14.574007220216606,
      "grad_norm": 246601.78125,
      "learning_rate": 0.0001904334002092252,
      "loss": 7.5159,
      "step": 121110
    },
    {
      "epoch": 14.575210589651023,
      "grad_norm": 45118.39453125,
      "learning_rate": 0.0001904317763843913,
      "loss": 7.5484,
      "step": 121120
    },
    {
      "epoch": 14.57641395908544,
      "grad_norm": 303728.75,
      "learning_rate": 0.00019043015242868033,
      "loss": 7.6251,
      "step": 121130
    },
    {
      "epoch": 14.577617328519855,
      "grad_norm": 88892.3359375,
      "learning_rate": 0.00019042852834209458,
      "loss": 7.5392,
      "step": 121140
    },
    {
      "epoch": 14.578820697954272,
      "grad_norm": 35062.8203125,
      "learning_rate": 0.0001904269041246364,
      "loss": 7.5078,
      "step": 121150
    },
    {
      "epoch": 14.580024067388688,
      "grad_norm": 78435.4609375,
      "learning_rate": 0.0001904252797763082,
      "loss": 7.4399,
      "step": 121160
    },
    {
      "epoch": 14.581227436823104,
      "grad_norm": 203744.515625,
      "learning_rate": 0.00019042365529711228,
      "loss": 7.6061,
      "step": 121170
    },
    {
      "epoch": 14.582430806257522,
      "grad_norm": 49817.32421875,
      "learning_rate": 0.000190422030687051,
      "loss": 7.5163,
      "step": 121180
    },
    {
      "epoch": 14.583634175691937,
      "grad_norm": 73829.75,
      "learning_rate": 0.0001904204059461267,
      "loss": 7.5219,
      "step": 121190
    },
    {
      "epoch": 14.584837545126353,
      "grad_norm": 46404.96875,
      "learning_rate": 0.00019041878107434177,
      "loss": 7.6124,
      "step": 121200
    },
    {
      "epoch": 14.58604091456077,
      "grad_norm": 110398.5234375,
      "learning_rate": 0.00019041715607169858,
      "loss": 7.5484,
      "step": 121210
    },
    {
      "epoch": 14.587244283995187,
      "grad_norm": 339699.03125,
      "learning_rate": 0.00019041553093819937,
      "loss": 7.5913,
      "step": 121220
    },
    {
      "epoch": 14.588447653429602,
      "grad_norm": 32041.23046875,
      "learning_rate": 0.00019041390567384658,
      "loss": 7.6101,
      "step": 121230
    },
    {
      "epoch": 14.58965102286402,
      "grad_norm": 432425.84375,
      "learning_rate": 0.0001904122802786426,
      "loss": 7.5679,
      "step": 121240
    },
    {
      "epoch": 14.590854392298436,
      "grad_norm": 2484734.25,
      "learning_rate": 0.00019041065475258967,
      "loss": 7.48,
      "step": 121250
    },
    {
      "epoch": 14.592057761732852,
      "grad_norm": 3148994.25,
      "learning_rate": 0.00019040902909569024,
      "loss": 7.5393,
      "step": 121260
    },
    {
      "epoch": 14.593261131167269,
      "grad_norm": 5117872.0,
      "learning_rate": 0.0001904074033079466,
      "loss": 7.7275,
      "step": 121270
    },
    {
      "epoch": 14.594464500601685,
      "grad_norm": 1033.40185546875,
      "learning_rate": 0.00019040577738936113,
      "loss": 7.7333,
      "step": 121280
    },
    {
      "epoch": 14.5956678700361,
      "grad_norm": 1416.0164794921875,
      "learning_rate": 0.00019040415133993616,
      "loss": 7.639,
      "step": 121290
    },
    {
      "epoch": 14.596871239470518,
      "grad_norm": 1054.4320068359375,
      "learning_rate": 0.0001904025251596741,
      "loss": 7.6402,
      "step": 121300
    },
    {
      "epoch": 14.598074608904934,
      "grad_norm": 2897.607421875,
      "learning_rate": 0.00019040089884857727,
      "loss": 7.6028,
      "step": 121310
    },
    {
      "epoch": 14.59927797833935,
      "grad_norm": 1400.635498046875,
      "learning_rate": 0.000190399272406648,
      "loss": 7.6035,
      "step": 121320
    },
    {
      "epoch": 14.600481347773766,
      "grad_norm": 11341.0,
      "learning_rate": 0.00019039764583388865,
      "loss": 7.5357,
      "step": 121330
    },
    {
      "epoch": 14.601684717208183,
      "grad_norm": 1359.5899658203125,
      "learning_rate": 0.00019039601913030158,
      "loss": 7.5346,
      "step": 121340
    },
    {
      "epoch": 14.602888086642599,
      "grad_norm": 5838.73681640625,
      "learning_rate": 0.0001903943922958892,
      "loss": 7.5998,
      "step": 121350
    },
    {
      "epoch": 14.604091456077015,
      "grad_norm": 8564.9609375,
      "learning_rate": 0.00019039276533065378,
      "loss": 7.5731,
      "step": 121360
    },
    {
      "epoch": 14.605294825511432,
      "grad_norm": 9540.6455078125,
      "learning_rate": 0.00019039113823459772,
      "loss": 7.5266,
      "step": 121370
    },
    {
      "epoch": 14.606498194945848,
      "grad_norm": 20517.2109375,
      "learning_rate": 0.00019038951100772337,
      "loss": 7.6627,
      "step": 121380
    },
    {
      "epoch": 14.607701564380264,
      "grad_norm": 4013.63671875,
      "learning_rate": 0.0001903878836500331,
      "loss": 7.4729,
      "step": 121390
    },
    {
      "epoch": 14.608904933814681,
      "grad_norm": 11828.546875,
      "learning_rate": 0.00019038625616152923,
      "loss": 7.5315,
      "step": 121400
    },
    {
      "epoch": 14.610108303249097,
      "grad_norm": 10380.02734375,
      "learning_rate": 0.0001903846285422141,
      "loss": 7.5812,
      "step": 121410
    },
    {
      "epoch": 14.611311672683513,
      "grad_norm": 9079.84375,
      "learning_rate": 0.00019038300079209016,
      "loss": 7.629,
      "step": 121420
    },
    {
      "epoch": 14.61251504211793,
      "grad_norm": 25292.38671875,
      "learning_rate": 0.00019038137291115965,
      "loss": 7.6044,
      "step": 121430
    },
    {
      "epoch": 14.613718411552346,
      "grad_norm": 15534.5,
      "learning_rate": 0.00019037974489942498,
      "loss": 7.5668,
      "step": 121440
    },
    {
      "epoch": 14.614921780986762,
      "grad_norm": 364.77783203125,
      "learning_rate": 0.00019037811675688855,
      "loss": 7.5552,
      "step": 121450
    },
    {
      "epoch": 14.61612515042118,
      "grad_norm": 91.00682830810547,
      "learning_rate": 0.00019037648848355262,
      "loss": 8.0238,
      "step": 121460
    },
    {
      "epoch": 14.617328519855596,
      "grad_norm": 171.2364501953125,
      "learning_rate": 0.00019037486007941965,
      "loss": 8.359,
      "step": 121470
    },
    {
      "epoch": 14.618531889290011,
      "grad_norm": 85.66725158691406,
      "learning_rate": 0.0001903732315444919,
      "loss": 7.9111,
      "step": 121480
    },
    {
      "epoch": 14.619735258724429,
      "grad_norm": 252.13705444335938,
      "learning_rate": 0.00019037160287877178,
      "loss": 7.7282,
      "step": 121490
    },
    {
      "epoch": 14.620938628158845,
      "grad_norm": 53.998146057128906,
      "learning_rate": 0.00019036997408226168,
      "loss": 7.8245,
      "step": 121500
    },
    {
      "epoch": 14.62214199759326,
      "grad_norm": 43.97968292236328,
      "learning_rate": 0.00019036834515496386,
      "loss": 7.6265,
      "step": 121510
    },
    {
      "epoch": 14.623345367027678,
      "grad_norm": 280.2363586425781,
      "learning_rate": 0.00019036671609688074,
      "loss": 7.7256,
      "step": 121520
    },
    {
      "epoch": 14.624548736462094,
      "grad_norm": 77.4672622680664,
      "learning_rate": 0.00019036508690801468,
      "loss": 7.6647,
      "step": 121530
    },
    {
      "epoch": 14.62575210589651,
      "grad_norm": 126.6371078491211,
      "learning_rate": 0.00019036345758836806,
      "loss": 7.6212,
      "step": 121540
    },
    {
      "epoch": 14.626955475330927,
      "grad_norm": 327.92376708984375,
      "learning_rate": 0.00019036182813794318,
      "loss": 7.5839,
      "step": 121550
    },
    {
      "epoch": 14.628158844765343,
      "grad_norm": 93.58558654785156,
      "learning_rate": 0.00019036019855674243,
      "loss": 7.5643,
      "step": 121560
    },
    {
      "epoch": 14.629362214199759,
      "grad_norm": 1769.89013671875,
      "learning_rate": 0.00019035856884476817,
      "loss": 7.5869,
      "step": 121570
    },
    {
      "epoch": 14.630565583634176,
      "grad_norm": 18687.419921875,
      "learning_rate": 0.00019035693900202274,
      "loss": 7.535,
      "step": 121580
    },
    {
      "epoch": 14.631768953068592,
      "grad_norm": 13755.064453125,
      "learning_rate": 0.0001903553090285085,
      "loss": 7.5482,
      "step": 121590
    },
    {
      "epoch": 14.632972322503008,
      "grad_norm": 6644.853515625,
      "learning_rate": 0.00019035367892422783,
      "loss": 7.433,
      "step": 121600
    },
    {
      "epoch": 14.634175691937426,
      "grad_norm": 6483.65673828125,
      "learning_rate": 0.00019035204868918308,
      "loss": 7.5786,
      "step": 121610
    },
    {
      "epoch": 14.635379061371841,
      "grad_norm": 18597.09375,
      "learning_rate": 0.00019035041832337664,
      "loss": 7.5576,
      "step": 121620
    },
    {
      "epoch": 14.636582430806257,
      "grad_norm": 19700.56640625,
      "learning_rate": 0.00019034878782681077,
      "loss": 7.5908,
      "step": 121630
    },
    {
      "epoch": 14.637785800240675,
      "grad_norm": 5978.9208984375,
      "learning_rate": 0.00019034715719948796,
      "loss": 7.5647,
      "step": 121640
    },
    {
      "epoch": 14.63898916967509,
      "grad_norm": 15456.568359375,
      "learning_rate": 0.00019034552644141048,
      "loss": 7.513,
      "step": 121650
    },
    {
      "epoch": 14.640192539109506,
      "grad_norm": 12615.375,
      "learning_rate": 0.00019034389555258073,
      "loss": 7.5143,
      "step": 121660
    },
    {
      "epoch": 14.641395908543924,
      "grad_norm": 14405.9013671875,
      "learning_rate": 0.00019034226453300103,
      "loss": 7.5429,
      "step": 121670
    },
    {
      "epoch": 14.64259927797834,
      "grad_norm": 11867.283203125,
      "learning_rate": 0.0001903406333826738,
      "loss": 7.5883,
      "step": 121680
    },
    {
      "epoch": 14.643802647412755,
      "grad_norm": 23185.244140625,
      "learning_rate": 0.00019033900210160137,
      "loss": 7.5519,
      "step": 121690
    },
    {
      "epoch": 14.645006016847173,
      "grad_norm": 7817.38134765625,
      "learning_rate": 0.00019033737068978608,
      "loss": 7.6064,
      "step": 121700
    },
    {
      "epoch": 14.646209386281589,
      "grad_norm": 8757.47265625,
      "learning_rate": 0.00019033573914723032,
      "loss": 7.4861,
      "step": 121710
    },
    {
      "epoch": 14.647412755716005,
      "grad_norm": 30533.787109375,
      "learning_rate": 0.00019033410747393642,
      "loss": 7.579,
      "step": 121720
    },
    {
      "epoch": 14.648616125150422,
      "grad_norm": 12339.4365234375,
      "learning_rate": 0.0001903324756699068,
      "loss": 7.5106,
      "step": 121730
    },
    {
      "epoch": 14.649819494584838,
      "grad_norm": 11227.1005859375,
      "learning_rate": 0.0001903308437351438,
      "loss": 7.6391,
      "step": 121740
    },
    {
      "epoch": 14.651022864019254,
      "grad_norm": 10967.966796875,
      "learning_rate": 0.0001903292116696497,
      "loss": 7.582,
      "step": 121750
    },
    {
      "epoch": 14.65222623345367,
      "grad_norm": 16365.79296875,
      "learning_rate": 0.000190327579473427,
      "loss": 7.5544,
      "step": 121760
    },
    {
      "epoch": 14.653429602888087,
      "grad_norm": 26277.84765625,
      "learning_rate": 0.00019032594714647792,
      "loss": 7.6138,
      "step": 121770
    },
    {
      "epoch": 14.654632972322503,
      "grad_norm": 10200.408203125,
      "learning_rate": 0.00019032431468880495,
      "loss": 7.5447,
      "step": 121780
    },
    {
      "epoch": 14.655836341756919,
      "grad_norm": 29018.0078125,
      "learning_rate": 0.00019032268210041035,
      "loss": 7.4466,
      "step": 121790
    },
    {
      "epoch": 14.657039711191336,
      "grad_norm": 9675.4892578125,
      "learning_rate": 0.00019032104938129657,
      "loss": 7.5781,
      "step": 121800
    },
    {
      "epoch": 14.658243080625752,
      "grad_norm": 13699.0322265625,
      "learning_rate": 0.0001903194165314659,
      "loss": 7.6039,
      "step": 121810
    },
    {
      "epoch": 14.659446450060168,
      "grad_norm": 23537.6328125,
      "learning_rate": 0.00019031778355092078,
      "loss": 7.5662,
      "step": 121820
    },
    {
      "epoch": 14.660649819494585,
      "grad_norm": 5751.908203125,
      "learning_rate": 0.00019031615043966352,
      "loss": 7.6278,
      "step": 121830
    },
    {
      "epoch": 14.661853188929001,
      "grad_norm": 30593.689453125,
      "learning_rate": 0.00019031451719769648,
      "loss": 7.5632,
      "step": 121840
    },
    {
      "epoch": 14.663056558363417,
      "grad_norm": 6277.75244140625,
      "learning_rate": 0.000190312883825022,
      "loss": 7.5859,
      "step": 121850
    },
    {
      "epoch": 14.664259927797834,
      "grad_norm": 18195.185546875,
      "learning_rate": 0.0001903112503216425,
      "loss": 7.5196,
      "step": 121860
    },
    {
      "epoch": 14.66546329723225,
      "grad_norm": 45156.15625,
      "learning_rate": 0.00019030961668756034,
      "loss": 7.553,
      "step": 121870
    },
    {
      "epoch": 14.666666666666666,
      "grad_norm": 11432.529296875,
      "learning_rate": 0.00019030798292277788,
      "loss": 7.5454,
      "step": 121880
    },
    {
      "epoch": 14.667870036101084,
      "grad_norm": 10046.5576171875,
      "learning_rate": 0.00019030634902729744,
      "loss": 7.6115,
      "step": 121890
    },
    {
      "epoch": 14.6690734055355,
      "grad_norm": 26655.6875,
      "learning_rate": 0.00019030471500112143,
      "loss": 7.5587,
      "step": 121900
    },
    {
      "epoch": 14.670276774969915,
      "grad_norm": 14060.3193359375,
      "learning_rate": 0.0001903030808442522,
      "loss": 7.6529,
      "step": 121910
    },
    {
      "epoch": 14.671480144404333,
      "grad_norm": 2856.0517578125,
      "learning_rate": 0.00019030144655669213,
      "loss": 7.5603,
      "step": 121920
    },
    {
      "epoch": 14.672683513838749,
      "grad_norm": 8363.81640625,
      "learning_rate": 0.00019029981213844357,
      "loss": 7.4968,
      "step": 121930
    },
    {
      "epoch": 14.673886883273164,
      "grad_norm": 1043.9971923828125,
      "learning_rate": 0.00019029817758950887,
      "loss": 7.6176,
      "step": 121940
    },
    {
      "epoch": 14.675090252707582,
      "grad_norm": 824.0234375,
      "learning_rate": 0.00019029654290989042,
      "loss": 7.4648,
      "step": 121950
    },
    {
      "epoch": 14.676293622141998,
      "grad_norm": 12782.1513671875,
      "learning_rate": 0.00019029490809959056,
      "loss": 7.5828,
      "step": 121960
    },
    {
      "epoch": 14.677496991576414,
      "grad_norm": 11996.8740234375,
      "learning_rate": 0.0001902932731586117,
      "loss": 7.5493,
      "step": 121970
    },
    {
      "epoch": 14.678700361010831,
      "grad_norm": 7799.84716796875,
      "learning_rate": 0.0001902916380869562,
      "loss": 7.586,
      "step": 121980
    },
    {
      "epoch": 14.679903730445247,
      "grad_norm": 15158.6318359375,
      "learning_rate": 0.00019029000288462637,
      "loss": 7.5302,
      "step": 121990
    },
    {
      "epoch": 14.681107099879663,
      "grad_norm": 11101.453125,
      "learning_rate": 0.00019028836755162463,
      "loss": 7.4953,
      "step": 122000
    },
    {
      "epoch": 14.68231046931408,
      "grad_norm": 11440.546875,
      "learning_rate": 0.00019028673208795334,
      "loss": 7.6015,
      "step": 122010
    },
    {
      "epoch": 14.683513838748496,
      "grad_norm": 9446.1123046875,
      "learning_rate": 0.00019028509649361483,
      "loss": 7.4875,
      "step": 122020
    },
    {
      "epoch": 14.684717208182912,
      "grad_norm": 4520.98828125,
      "learning_rate": 0.00019028346076861152,
      "loss": 7.5343,
      "step": 122030
    },
    {
      "epoch": 14.685920577617328,
      "grad_norm": 29015.2890625,
      "learning_rate": 0.00019028182491294575,
      "loss": 7.6106,
      "step": 122040
    },
    {
      "epoch": 14.687123947051745,
      "grad_norm": 5525.0478515625,
      "learning_rate": 0.00019028018892661987,
      "loss": 7.5342,
      "step": 122050
    },
    {
      "epoch": 14.688327316486161,
      "grad_norm": 6324.21142578125,
      "learning_rate": 0.00019027855280963628,
      "loss": 7.637,
      "step": 122060
    },
    {
      "epoch": 14.689530685920577,
      "grad_norm": 30115.294921875,
      "learning_rate": 0.00019027691656199735,
      "loss": 7.5702,
      "step": 122070
    },
    {
      "epoch": 14.690734055354994,
      "grad_norm": 9524.673828125,
      "learning_rate": 0.00019027528018370542,
      "loss": 7.596,
      "step": 122080
    },
    {
      "epoch": 14.69193742478941,
      "grad_norm": 12366.0908203125,
      "learning_rate": 0.00019027364367476285,
      "loss": 7.6289,
      "step": 122090
    },
    {
      "epoch": 14.693140794223826,
      "grad_norm": 21210.775390625,
      "learning_rate": 0.00019027200703517204,
      "loss": 7.5118,
      "step": 122100
    },
    {
      "epoch": 14.694344163658243,
      "grad_norm": 13548.4619140625,
      "learning_rate": 0.00019027037026493536,
      "loss": 7.6133,
      "step": 122110
    },
    {
      "epoch": 14.69554753309266,
      "grad_norm": 11168.794921875,
      "learning_rate": 0.00019026873336405517,
      "loss": 7.6082,
      "step": 122120
    },
    {
      "epoch": 14.696750902527075,
      "grad_norm": 50473.8046875,
      "learning_rate": 0.00019026709633253384,
      "loss": 7.4926,
      "step": 122130
    },
    {
      "epoch": 14.697954271961493,
      "grad_norm": 30738.1171875,
      "learning_rate": 0.00019026545917037372,
      "loss": 7.6124,
      "step": 122140
    },
    {
      "epoch": 14.699157641395908,
      "grad_norm": 8852.40234375,
      "learning_rate": 0.0001902638218775772,
      "loss": 7.4859,
      "step": 122150
    },
    {
      "epoch": 14.700361010830324,
      "grad_norm": 4939.70166015625,
      "learning_rate": 0.00019026218445414666,
      "loss": 7.4696,
      "step": 122160
    },
    {
      "epoch": 14.701564380264742,
      "grad_norm": 21111.3046875,
      "learning_rate": 0.00019026054690008445,
      "loss": 7.5586,
      "step": 122170
    },
    {
      "epoch": 14.702767749699158,
      "grad_norm": 17790.203125,
      "learning_rate": 0.00019025890921539292,
      "loss": 7.5835,
      "step": 122180
    },
    {
      "epoch": 14.703971119133573,
      "grad_norm": 10519.9755859375,
      "learning_rate": 0.00019025727140007447,
      "loss": 7.5517,
      "step": 122190
    },
    {
      "epoch": 14.705174488567991,
      "grad_norm": 14167.900390625,
      "learning_rate": 0.00019025563345413148,
      "loss": 7.5415,
      "step": 122200
    },
    {
      "epoch": 14.706377858002407,
      "grad_norm": 2617.821533203125,
      "learning_rate": 0.0001902539953775663,
      "loss": 7.5788,
      "step": 122210
    },
    {
      "epoch": 14.707581227436823,
      "grad_norm": 23949.509765625,
      "learning_rate": 0.0001902523571703813,
      "loss": 7.5168,
      "step": 122220
    },
    {
      "epoch": 14.70878459687124,
      "grad_norm": 35940.30859375,
      "learning_rate": 0.00019025071883257885,
      "loss": 7.5534,
      "step": 122230
    },
    {
      "epoch": 14.709987966305656,
      "grad_norm": 55285.6875,
      "learning_rate": 0.00019024908036416134,
      "loss": 7.5196,
      "step": 122240
    },
    {
      "epoch": 14.711191335740072,
      "grad_norm": 9608.3037109375,
      "learning_rate": 0.00019024744176513112,
      "loss": 7.5453,
      "step": 122250
    },
    {
      "epoch": 14.71239470517449,
      "grad_norm": 6294.59912109375,
      "learning_rate": 0.0001902458030354906,
      "loss": 7.504,
      "step": 122260
    },
    {
      "epoch": 14.713598074608905,
      "grad_norm": 7079.0673828125,
      "learning_rate": 0.00019024416417524206,
      "loss": 7.5439,
      "step": 122270
    },
    {
      "epoch": 14.71480144404332,
      "grad_norm": 4539.9521484375,
      "learning_rate": 0.000190242525184388,
      "loss": 7.5062,
      "step": 122280
    },
    {
      "epoch": 14.716004813477738,
      "grad_norm": 6838.11767578125,
      "learning_rate": 0.00019024088606293068,
      "loss": 7.4395,
      "step": 122290
    },
    {
      "epoch": 14.717208182912154,
      "grad_norm": 17486.287109375,
      "learning_rate": 0.00019023924681087255,
      "loss": 7.5497,
      "step": 122300
    },
    {
      "epoch": 14.71841155234657,
      "grad_norm": 7714.32958984375,
      "learning_rate": 0.00019023760742821594,
      "loss": 7.5659,
      "step": 122310
    },
    {
      "epoch": 14.719614921780988,
      "grad_norm": 19279.427734375,
      "learning_rate": 0.00019023596791496324,
      "loss": 7.502,
      "step": 122320
    },
    {
      "epoch": 14.720818291215403,
      "grad_norm": 4168.82861328125,
      "learning_rate": 0.00019023432827111681,
      "loss": 7.6056,
      "step": 122330
    },
    {
      "epoch": 14.722021660649819,
      "grad_norm": 11071.7626953125,
      "learning_rate": 0.00019023268849667903,
      "loss": 7.5612,
      "step": 122340
    },
    {
      "epoch": 14.723225030084237,
      "grad_norm": 4132.884765625,
      "learning_rate": 0.00019023104859165223,
      "loss": 7.4394,
      "step": 122350
    },
    {
      "epoch": 14.724428399518652,
      "grad_norm": 486.6014404296875,
      "learning_rate": 0.0001902294085560389,
      "loss": 7.6269,
      "step": 122360
    },
    {
      "epoch": 14.725631768953068,
      "grad_norm": 1091.424072265625,
      "learning_rate": 0.0001902277683898413,
      "loss": 7.5794,
      "step": 122370
    },
    {
      "epoch": 14.726835138387486,
      "grad_norm": 82.9935073852539,
      "learning_rate": 0.0001902261280930618,
      "loss": 7.523,
      "step": 122380
    },
    {
      "epoch": 14.728038507821902,
      "grad_norm": 547.2711791992188,
      "learning_rate": 0.0001902244876657029,
      "loss": 7.555,
      "step": 122390
    },
    {
      "epoch": 14.729241877256317,
      "grad_norm": 233.89939880371094,
      "learning_rate": 0.00019022284710776684,
      "loss": 7.5725,
      "step": 122400
    },
    {
      "epoch": 14.730445246690735,
      "grad_norm": 1547.3026123046875,
      "learning_rate": 0.00019022120641925606,
      "loss": 7.5564,
      "step": 122410
    },
    {
      "epoch": 14.73164861612515,
      "grad_norm": 264.3048400878906,
      "learning_rate": 0.0001902195656001729,
      "loss": 7.4457,
      "step": 122420
    },
    {
      "epoch": 14.732851985559567,
      "grad_norm": 8613589.0,
      "learning_rate": 0.00019021792465051976,
      "loss": 7.5391,
      "step": 122430
    },
    {
      "epoch": 14.734055354993982,
      "grad_norm": 515.7554321289062,
      "learning_rate": 0.00019021628357029902,
      "loss": 7.5001,
      "step": 122440
    },
    {
      "epoch": 14.7352587244284,
      "grad_norm": 5424.0078125,
      "learning_rate": 0.00019021464235951304,
      "loss": 7.4824,
      "step": 122450
    },
    {
      "epoch": 14.736462093862816,
      "grad_norm": 405.7532653808594,
      "learning_rate": 0.0001902130010181642,
      "loss": 7.5818,
      "step": 122460
    },
    {
      "epoch": 14.737665463297231,
      "grad_norm": 4180.47998046875,
      "learning_rate": 0.00019021135954625488,
      "loss": 7.6316,
      "step": 122470
    },
    {
      "epoch": 14.738868832731649,
      "grad_norm": 3046.92431640625,
      "learning_rate": 0.00019020971794378745,
      "loss": 7.5179,
      "step": 122480
    },
    {
      "epoch": 14.740072202166065,
      "grad_norm": 5158.8134765625,
      "learning_rate": 0.0001902080762107643,
      "loss": 7.5426,
      "step": 122490
    },
    {
      "epoch": 14.74127557160048,
      "grad_norm": 7532.86669921875,
      "learning_rate": 0.00019020643434718775,
      "loss": 7.5693,
      "step": 122500
    },
    {
      "epoch": 14.742478941034898,
      "grad_norm": 5827.73486328125,
      "learning_rate": 0.00019020479235306025,
      "loss": 7.6041,
      "step": 122510
    },
    {
      "epoch": 14.743682310469314,
      "grad_norm": 6403.52685546875,
      "learning_rate": 0.00019020315022838413,
      "loss": 7.56,
      "step": 122520
    },
    {
      "epoch": 14.74488567990373,
      "grad_norm": 6356.53955078125,
      "learning_rate": 0.00019020150797316177,
      "loss": 7.5203,
      "step": 122530
    },
    {
      "epoch": 14.746089049338147,
      "grad_norm": 155.31362915039062,
      "learning_rate": 0.0001901998655873956,
      "loss": 7.5035,
      "step": 122540
    },
    {
      "epoch": 14.747292418772563,
      "grad_norm": 28.775062561035156,
      "learning_rate": 0.00019019822307108793,
      "loss": 7.5288,
      "step": 122550
    },
    {
      "epoch": 14.748495788206979,
      "grad_norm": 49.77052307128906,
      "learning_rate": 0.00019019658042424117,
      "loss": 7.5085,
      "step": 122560
    },
    {
      "epoch": 14.749699157641396,
      "grad_norm": 58.89057922363281,
      "learning_rate": 0.0001901949376468577,
      "loss": 7.6573,
      "step": 122570
    },
    {
      "epoch": 14.750902527075812,
      "grad_norm": 37.69395446777344,
      "learning_rate": 0.00019019329473893987,
      "loss": 7.524,
      "step": 122580
    },
    {
      "epoch": 14.752105896510228,
      "grad_norm": 132.66085815429688,
      "learning_rate": 0.00019019165170049008,
      "loss": 7.4882,
      "step": 122590
    },
    {
      "epoch": 14.753309265944646,
      "grad_norm": 153.02452087402344,
      "learning_rate": 0.00019019000853151067,
      "loss": 7.5685,
      "step": 122600
    },
    {
      "epoch": 14.754512635379061,
      "grad_norm": 53.82183074951172,
      "learning_rate": 0.0001901883652320041,
      "loss": 7.5153,
      "step": 122610
    },
    {
      "epoch": 14.755716004813477,
      "grad_norm": 133.5232391357422,
      "learning_rate": 0.00019018672180197267,
      "loss": 7.4755,
      "step": 122620
    },
    {
      "epoch": 14.756919374247895,
      "grad_norm": 37.387062072753906,
      "learning_rate": 0.0001901850782414188,
      "loss": 7.6259,
      "step": 122630
    },
    {
      "epoch": 14.75812274368231,
      "grad_norm": 112.40479278564453,
      "learning_rate": 0.0001901834345503448,
      "loss": 7.4955,
      "step": 122640
    },
    {
      "epoch": 14.759326113116726,
      "grad_norm": 57.584815979003906,
      "learning_rate": 0.00019018179072875316,
      "loss": 7.4981,
      "step": 122650
    },
    {
      "epoch": 14.760529482551144,
      "grad_norm": 55.004913330078125,
      "learning_rate": 0.00019018014677664623,
      "loss": 7.4647,
      "step": 122660
    },
    {
      "epoch": 14.76173285198556,
      "grad_norm": 55.24393081665039,
      "learning_rate": 0.00019017850269402632,
      "loss": 7.5824,
      "step": 122670
    },
    {
      "epoch": 14.762936221419976,
      "grad_norm": 158.1719970703125,
      "learning_rate": 0.00019017685848089587,
      "loss": 7.5682,
      "step": 122680
    },
    {
      "epoch": 14.764139590854393,
      "grad_norm": 41.28356170654297,
      "learning_rate": 0.00019017521413725722,
      "loss": 7.5727,
      "step": 122690
    },
    {
      "epoch": 14.765342960288809,
      "grad_norm": 33.73509216308594,
      "learning_rate": 0.0001901735696631128,
      "loss": 7.5237,
      "step": 122700
    },
    {
      "epoch": 14.766546329723225,
      "grad_norm": 53.68198776245117,
      "learning_rate": 0.00019017192505846496,
      "loss": 7.501,
      "step": 122710
    },
    {
      "epoch": 14.76774969915764,
      "grad_norm": 91.94418334960938,
      "learning_rate": 0.00019017028032331602,
      "loss": 7.5169,
      "step": 122720
    },
    {
      "epoch": 14.768953068592058,
      "grad_norm": 34.075313568115234,
      "learning_rate": 0.0001901686354576685,
      "loss": 7.5402,
      "step": 122730
    },
    {
      "epoch": 14.770156438026474,
      "grad_norm": 85.29915618896484,
      "learning_rate": 0.00019016699046152467,
      "loss": 7.5127,
      "step": 122740
    },
    {
      "epoch": 14.77135980746089,
      "grad_norm": 68.47270965576172,
      "learning_rate": 0.00019016534533488694,
      "loss": 7.5958,
      "step": 122750
    },
    {
      "epoch": 14.772563176895307,
      "grad_norm": 33.10287857055664,
      "learning_rate": 0.0001901637000777577,
      "loss": 7.5908,
      "step": 122760
    },
    {
      "epoch": 14.773766546329723,
      "grad_norm": 101.12022399902344,
      "learning_rate": 0.0001901620546901393,
      "loss": 7.4801,
      "step": 122770
    },
    {
      "epoch": 14.774969915764139,
      "grad_norm": 61.49049377441406,
      "learning_rate": 0.00019016040917203418,
      "loss": 7.609,
      "step": 122780
    },
    {
      "epoch": 14.776173285198556,
      "grad_norm": 159.17449951171875,
      "learning_rate": 0.0001901587635234447,
      "loss": 7.5843,
      "step": 122790
    },
    {
      "epoch": 14.777376654632972,
      "grad_norm": 135.71945190429688,
      "learning_rate": 0.00019015711774437321,
      "loss": 7.4823,
      "step": 122800
    },
    {
      "epoch": 14.778580024067388,
      "grad_norm": 120.03449249267578,
      "learning_rate": 0.00019015547183482212,
      "loss": 7.4968,
      "step": 122810
    },
    {
      "epoch": 14.779783393501805,
      "grad_norm": 195.04974365234375,
      "learning_rate": 0.0001901538257947938,
      "loss": 7.5601,
      "step": 122820
    },
    {
      "epoch": 14.780986762936221,
      "grad_norm": 165.19677734375,
      "learning_rate": 0.0001901521796242906,
      "loss": 7.486,
      "step": 122830
    },
    {
      "epoch": 14.782190132370637,
      "grad_norm": 152.77638244628906,
      "learning_rate": 0.000190150533323315,
      "loss": 7.4967,
      "step": 122840
    },
    {
      "epoch": 14.783393501805055,
      "grad_norm": 78.87261962890625,
      "learning_rate": 0.00019014888689186928,
      "loss": 7.564,
      "step": 122850
    },
    {
      "epoch": 14.78459687123947,
      "grad_norm": 178.48228454589844,
      "learning_rate": 0.00019014724032995588,
      "loss": 7.5698,
      "step": 122860
    },
    {
      "epoch": 14.785800240673886,
      "grad_norm": 244.239501953125,
      "learning_rate": 0.00019014559363757715,
      "loss": 7.521,
      "step": 122870
    },
    {
      "epoch": 14.787003610108304,
      "grad_norm": 193.8568572998047,
      "learning_rate": 0.0001901439468147355,
      "loss": 7.598,
      "step": 122880
    },
    {
      "epoch": 14.78820697954272,
      "grad_norm": 220.69061279296875,
      "learning_rate": 0.00019014229986143331,
      "loss": 7.5123,
      "step": 122890
    },
    {
      "epoch": 14.789410348977135,
      "grad_norm": 331.3968200683594,
      "learning_rate": 0.00019014065277767295,
      "loss": 7.4906,
      "step": 122900
    },
    {
      "epoch": 14.790613718411553,
      "grad_norm": 604.4608154296875,
      "learning_rate": 0.00019013900556345682,
      "loss": 7.4675,
      "step": 122910
    },
    {
      "epoch": 14.791817087845969,
      "grad_norm": 231.6101531982422,
      "learning_rate": 0.00019013735821878726,
      "loss": 7.4957,
      "step": 122920
    },
    {
      "epoch": 14.793020457280385,
      "grad_norm": 300.3955383300781,
      "learning_rate": 0.00019013571074366672,
      "loss": 7.4884,
      "step": 122930
    },
    {
      "epoch": 14.794223826714802,
      "grad_norm": 274.4205322265625,
      "learning_rate": 0.00019013406313809757,
      "loss": 7.5254,
      "step": 122940
    },
    {
      "epoch": 14.795427196149218,
      "grad_norm": 591.6887817382812,
      "learning_rate": 0.00019013241540208212,
      "loss": 7.4035,
      "step": 122950
    },
    {
      "epoch": 14.796630565583634,
      "grad_norm": 427.78155517578125,
      "learning_rate": 0.00019013076753562287,
      "loss": 7.5022,
      "step": 122960
    },
    {
      "epoch": 14.797833935018051,
      "grad_norm": 392.9446105957031,
      "learning_rate": 0.0001901291195387221,
      "loss": 7.4787,
      "step": 122970
    },
    {
      "epoch": 14.799037304452467,
      "grad_norm": 508.8036193847656,
      "learning_rate": 0.00019012747141138227,
      "loss": 7.4821,
      "step": 122980
    },
    {
      "epoch": 14.800240673886883,
      "grad_norm": 471.197265625,
      "learning_rate": 0.00019012582315360574,
      "loss": 7.5001,
      "step": 122990
    },
    {
      "epoch": 14.8014440433213,
      "grad_norm": 411.7552490234375,
      "learning_rate": 0.00019012417476539488,
      "loss": 7.4503,
      "step": 123000
    },
    {
      "epoch": 14.802647412755716,
      "grad_norm": 167.80752563476562,
      "learning_rate": 0.00019012252624675207,
      "loss": 7.508,
      "step": 123010
    },
    {
      "epoch": 14.803850782190132,
      "grad_norm": 489.46759033203125,
      "learning_rate": 0.00019012087759767975,
      "loss": 7.4661,
      "step": 123020
    },
    {
      "epoch": 14.80505415162455,
      "grad_norm": 495.8715515136719,
      "learning_rate": 0.00019011922881818026,
      "loss": 7.5215,
      "step": 123030
    },
    {
      "epoch": 14.806257521058965,
      "grad_norm": 316.3471984863281,
      "learning_rate": 0.000190117579908256,
      "loss": 7.4855,
      "step": 123040
    },
    {
      "epoch": 14.807460890493381,
      "grad_norm": 1219.4678955078125,
      "learning_rate": 0.00019011593086790934,
      "loss": 7.4835,
      "step": 123050
    },
    {
      "epoch": 14.808664259927799,
      "grad_norm": 726.0543823242188,
      "learning_rate": 0.0001901142816971427,
      "loss": 7.5142,
      "step": 123060
    },
    {
      "epoch": 14.809867629362214,
      "grad_norm": 244.98731994628906,
      "learning_rate": 0.00019011263239595842,
      "loss": 7.4415,
      "step": 123070
    },
    {
      "epoch": 14.81107099879663,
      "grad_norm": 278.6109619140625,
      "learning_rate": 0.00019011098296435892,
      "loss": 7.429,
      "step": 123080
    },
    {
      "epoch": 14.812274368231048,
      "grad_norm": 236.35256958007812,
      "learning_rate": 0.00019010933340234658,
      "loss": 7.4471,
      "step": 123090
    },
    {
      "epoch": 14.813477737665464,
      "grad_norm": 1024.350830078125,
      "learning_rate": 0.0001901076837099238,
      "loss": 7.5124,
      "step": 123100
    },
    {
      "epoch": 14.81468110709988,
      "grad_norm": 249.34996032714844,
      "learning_rate": 0.00019010603388709295,
      "loss": 7.5433,
      "step": 123110
    },
    {
      "epoch": 14.815884476534297,
      "grad_norm": 288.5848083496094,
      "learning_rate": 0.0001901043839338564,
      "loss": 7.4241,
      "step": 123120
    },
    {
      "epoch": 14.817087845968713,
      "grad_norm": 552.9564208984375,
      "learning_rate": 0.0001901027338502166,
      "loss": 7.4428,
      "step": 123130
    },
    {
      "epoch": 14.818291215403129,
      "grad_norm": 517.490966796875,
      "learning_rate": 0.00019010108363617584,
      "loss": 7.4797,
      "step": 123140
    },
    {
      "epoch": 14.819494584837544,
      "grad_norm": 1194.057373046875,
      "learning_rate": 0.0001900994332917366,
      "loss": 7.5048,
      "step": 123150
    },
    {
      "epoch": 14.820697954271962,
      "grad_norm": 290.843505859375,
      "learning_rate": 0.00019009778281690127,
      "loss": 7.4732,
      "step": 123160
    },
    {
      "epoch": 14.821901323706378,
      "grad_norm": 631.5757446289062,
      "learning_rate": 0.00019009613221167214,
      "loss": 7.4528,
      "step": 123170
    },
    {
      "epoch": 14.823104693140793,
      "grad_norm": 631.85107421875,
      "learning_rate": 0.0001900944814760517,
      "loss": 7.4764,
      "step": 123180
    },
    {
      "epoch": 14.824308062575211,
      "grad_norm": 532.7399291992188,
      "learning_rate": 0.00019009283061004233,
      "loss": 7.4767,
      "step": 123190
    },
    {
      "epoch": 14.825511432009627,
      "grad_norm": 604.7129516601562,
      "learning_rate": 0.00019009117961364634,
      "loss": 7.3929,
      "step": 123200
    },
    {
      "epoch": 14.826714801444043,
      "grad_norm": 5542.78515625,
      "learning_rate": 0.0001900895284868662,
      "loss": 7.4148,
      "step": 123210
    },
    {
      "epoch": 14.82791817087846,
      "grad_norm": 1393.5919189453125,
      "learning_rate": 0.00019008787722970424,
      "loss": 7.4536,
      "step": 123220
    },
    {
      "epoch": 14.829121540312876,
      "grad_norm": 1129.9512939453125,
      "learning_rate": 0.0001900862258421629,
      "loss": 7.5028,
      "step": 123230
    },
    {
      "epoch": 14.830324909747292,
      "grad_norm": 419.77069091796875,
      "learning_rate": 0.00019008457432424453,
      "loss": 7.4902,
      "step": 123240
    },
    {
      "epoch": 14.83152827918171,
      "grad_norm": 1766.2742919921875,
      "learning_rate": 0.0001900829226759516,
      "loss": 7.4646,
      "step": 123250
    },
    {
      "epoch": 14.832731648616125,
      "grad_norm": 596.582275390625,
      "learning_rate": 0.00019008127089728639,
      "loss": 7.4202,
      "step": 123260
    },
    {
      "epoch": 14.833935018050541,
      "grad_norm": 1147.4884033203125,
      "learning_rate": 0.00019007961898825134,
      "loss": 7.4772,
      "step": 123270
    },
    {
      "epoch": 14.835138387484958,
      "grad_norm": 646.4959716796875,
      "learning_rate": 0.00019007796694884882,
      "loss": 7.5686,
      "step": 123280
    },
    {
      "epoch": 14.836341756919374,
      "grad_norm": 753.332763671875,
      "learning_rate": 0.0001900763147790813,
      "loss": 7.4752,
      "step": 123290
    },
    {
      "epoch": 14.83754512635379,
      "grad_norm": 317.5640869140625,
      "learning_rate": 0.00019007466247895107,
      "loss": 7.4868,
      "step": 123300
    },
    {
      "epoch": 14.838748495788208,
      "grad_norm": 258.5713806152344,
      "learning_rate": 0.0001900730100484606,
      "loss": 7.4888,
      "step": 123310
    },
    {
      "epoch": 14.839951865222623,
      "grad_norm": 394.9584045410156,
      "learning_rate": 0.0001900713574876122,
      "loss": 7.5689,
      "step": 123320
    },
    {
      "epoch": 14.84115523465704,
      "grad_norm": 613.81689453125,
      "learning_rate": 0.00019006970479640834,
      "loss": 7.5011,
      "step": 123330
    },
    {
      "epoch": 14.842358604091457,
      "grad_norm": 834.0196533203125,
      "learning_rate": 0.00019006805197485137,
      "loss": 7.4924,
      "step": 123340
    },
    {
      "epoch": 14.843561973525873,
      "grad_norm": 323.1937561035156,
      "learning_rate": 0.0001900663990229437,
      "loss": 7.4384,
      "step": 123350
    },
    {
      "epoch": 14.844765342960288,
      "grad_norm": 575.8717041015625,
      "learning_rate": 0.0001900647459406877,
      "loss": 7.3928,
      "step": 123360
    },
    {
      "epoch": 14.845968712394706,
      "grad_norm": 664.209228515625,
      "learning_rate": 0.0001900630927280858,
      "loss": 7.4093,
      "step": 123370
    },
    {
      "epoch": 14.847172081829122,
      "grad_norm": 885.2428588867188,
      "learning_rate": 0.00019006143938514033,
      "loss": 7.4908,
      "step": 123380
    },
    {
      "epoch": 14.848375451263538,
      "grad_norm": 2975.583984375,
      "learning_rate": 0.00019005978591185376,
      "loss": 7.37,
      "step": 123390
    },
    {
      "epoch": 14.849578820697955,
      "grad_norm": 548.8884887695312,
      "learning_rate": 0.00019005813230822845,
      "loss": 7.3167,
      "step": 123400
    },
    {
      "epoch": 14.85078219013237,
      "grad_norm": 523.2142333984375,
      "learning_rate": 0.00019005647857426675,
      "loss": 7.4647,
      "step": 123410
    },
    {
      "epoch": 14.851985559566787,
      "grad_norm": 1123.9825439453125,
      "learning_rate": 0.00019005482470997112,
      "loss": 7.4199,
      "step": 123420
    },
    {
      "epoch": 14.853188929001202,
      "grad_norm": 382.41973876953125,
      "learning_rate": 0.0001900531707153439,
      "loss": 7.433,
      "step": 123430
    },
    {
      "epoch": 14.85439229843562,
      "grad_norm": 781.0701293945312,
      "learning_rate": 0.00019005151659038752,
      "loss": 7.3746,
      "step": 123440
    },
    {
      "epoch": 14.855595667870036,
      "grad_norm": 3915.824951171875,
      "learning_rate": 0.00019004986233510436,
      "loss": 7.4461,
      "step": 123450
    },
    {
      "epoch": 14.856799037304452,
      "grad_norm": 821.8715209960938,
      "learning_rate": 0.00019004820794949682,
      "loss": 7.5793,
      "step": 123460
    },
    {
      "epoch": 14.85800240673887,
      "grad_norm": 17140.52734375,
      "learning_rate": 0.0001900465534335673,
      "loss": 7.6551,
      "step": 123470
    },
    {
      "epoch": 14.859205776173285,
      "grad_norm": 10236.552734375,
      "learning_rate": 0.00019004489878731817,
      "loss": 7.5299,
      "step": 123480
    },
    {
      "epoch": 14.8604091456077,
      "grad_norm": 3837.97607421875,
      "learning_rate": 0.00019004324401075185,
      "loss": 7.4197,
      "step": 123490
    },
    {
      "epoch": 14.861612515042118,
      "grad_norm": 592.1118774414062,
      "learning_rate": 0.00019004158910387073,
      "loss": 7.4037,
      "step": 123500
    },
    {
      "epoch": 14.862815884476534,
      "grad_norm": 2852.675537109375,
      "learning_rate": 0.0001900399340666772,
      "loss": 7.4502,
      "step": 123510
    },
    {
      "epoch": 14.86401925391095,
      "grad_norm": 1133.818359375,
      "learning_rate": 0.00019003827889917365,
      "loss": 7.4708,
      "step": 123520
    },
    {
      "epoch": 14.865222623345367,
      "grad_norm": 1052.6903076171875,
      "learning_rate": 0.00019003662360136248,
      "loss": 7.4866,
      "step": 123530
    },
    {
      "epoch": 14.866425992779783,
      "grad_norm": 317.1869812011719,
      "learning_rate": 0.00019003496817324606,
      "loss": 7.5149,
      "step": 123540
    },
    {
      "epoch": 14.867629362214199,
      "grad_norm": 1005.131103515625,
      "learning_rate": 0.00019003331261482686,
      "loss": 7.4323,
      "step": 123550
    },
    {
      "epoch": 14.868832731648617,
      "grad_norm": 814.2579345703125,
      "learning_rate": 0.0001900316569261072,
      "loss": 7.4878,
      "step": 123560
    },
    {
      "epoch": 14.870036101083032,
      "grad_norm": 246.00350952148438,
      "learning_rate": 0.0001900300011070895,
      "loss": 7.4267,
      "step": 123570
    },
    {
      "epoch": 14.871239470517448,
      "grad_norm": 546.3649291992188,
      "learning_rate": 0.00019002834515777618,
      "loss": 7.3432,
      "step": 123580
    },
    {
      "epoch": 14.872442839951866,
      "grad_norm": 2349.000732421875,
      "learning_rate": 0.00019002668907816962,
      "loss": 7.4714,
      "step": 123590
    },
    {
      "epoch": 14.873646209386282,
      "grad_norm": 2980.470703125,
      "learning_rate": 0.00019002503286827217,
      "loss": 7.3986,
      "step": 123600
    },
    {
      "epoch": 14.874849578820697,
      "grad_norm": 724.5343627929688,
      "learning_rate": 0.00019002337652808632,
      "loss": 7.4567,
      "step": 123610
    },
    {
      "epoch": 14.876052948255115,
      "grad_norm": 4227.9267578125,
      "learning_rate": 0.0001900217200576144,
      "loss": 7.5499,
      "step": 123620
    },
    {
      "epoch": 14.87725631768953,
      "grad_norm": 1813.550537109375,
      "learning_rate": 0.00019002006345685883,
      "loss": 7.5066,
      "step": 123630
    },
    {
      "epoch": 14.878459687123947,
      "grad_norm": 3634.90966796875,
      "learning_rate": 0.000190018406725822,
      "loss": 7.4871,
      "step": 123640
    },
    {
      "epoch": 14.879663056558364,
      "grad_norm": 1739.0926513671875,
      "learning_rate": 0.0001900167498645063,
      "loss": 7.4252,
      "step": 123650
    },
    {
      "epoch": 14.88086642599278,
      "grad_norm": 1023.6815795898438,
      "learning_rate": 0.00019001509287291417,
      "loss": 7.4732,
      "step": 123660
    },
    {
      "epoch": 14.882069795427196,
      "grad_norm": 920.9478149414062,
      "learning_rate": 0.00019001343575104795,
      "loss": 7.4512,
      "step": 123670
    },
    {
      "epoch": 14.883273164861613,
      "grad_norm": 3632.940185546875,
      "learning_rate": 0.00019001177849891008,
      "loss": 7.4817,
      "step": 123680
    },
    {
      "epoch": 14.884476534296029,
      "grad_norm": 3801.51318359375,
      "learning_rate": 0.00019001012111650293,
      "loss": 7.3781,
      "step": 123690
    },
    {
      "epoch": 14.885679903730445,
      "grad_norm": 2609.939697265625,
      "learning_rate": 0.00019000846360382888,
      "loss": 7.3926,
      "step": 123700
    },
    {
      "epoch": 14.886883273164862,
      "grad_norm": 638.1207275390625,
      "learning_rate": 0.00019000680596089042,
      "loss": 7.4244,
      "step": 123710
    },
    {
      "epoch": 14.888086642599278,
      "grad_norm": 1193.732666015625,
      "learning_rate": 0.00019000514818768986,
      "loss": 7.4428,
      "step": 123720
    },
    {
      "epoch": 14.889290012033694,
      "grad_norm": 1122.576171875,
      "learning_rate": 0.00019000349028422964,
      "loss": 7.4545,
      "step": 123730
    },
    {
      "epoch": 14.890493381468112,
      "grad_norm": 4699.59033203125,
      "learning_rate": 0.00019000183225051213,
      "loss": 7.4537,
      "step": 123740
    },
    {
      "epoch": 14.891696750902527,
      "grad_norm": 5353.478515625,
      "learning_rate": 0.00019000017408653976,
      "loss": 7.334,
      "step": 123750
    },
    {
      "epoch": 14.892900120336943,
      "grad_norm": 1078.3751220703125,
      "learning_rate": 0.00018999851579231493,
      "loss": 7.4892,
      "step": 123760
    },
    {
      "epoch": 14.89410348977136,
      "grad_norm": 1494.1815185546875,
      "learning_rate": 0.00018999685736784003,
      "loss": 7.4796,
      "step": 123770
    },
    {
      "epoch": 14.895306859205776,
      "grad_norm": 2306.15380859375,
      "learning_rate": 0.00018999519881311746,
      "loss": 7.6021,
      "step": 123780
    },
    {
      "epoch": 14.896510228640192,
      "grad_norm": 6194.015625,
      "learning_rate": 0.00018999354012814957,
      "loss": 7.4274,
      "step": 123790
    },
    {
      "epoch": 14.89771359807461,
      "grad_norm": 659.0177612304688,
      "learning_rate": 0.00018999188131293884,
      "loss": 7.3333,
      "step": 123800
    },
    {
      "epoch": 14.898916967509026,
      "grad_norm": 408.4515380859375,
      "learning_rate": 0.00018999022236748764,
      "loss": 7.3332,
      "step": 123810
    },
    {
      "epoch": 14.900120336943441,
      "grad_norm": 362.9180908203125,
      "learning_rate": 0.00018998856329179836,
      "loss": 7.4628,
      "step": 123820
    },
    {
      "epoch": 14.901323706377857,
      "grad_norm": 586.2081298828125,
      "learning_rate": 0.00018998690408587345,
      "loss": 7.4983,
      "step": 123830
    },
    {
      "epoch": 14.902527075812275,
      "grad_norm": 986.05419921875,
      "learning_rate": 0.00018998524474971525,
      "loss": 7.3288,
      "step": 123840
    },
    {
      "epoch": 14.90373044524669,
      "grad_norm": 456.545654296875,
      "learning_rate": 0.00018998358528332619,
      "loss": 7.4967,
      "step": 123850
    },
    {
      "epoch": 14.904933814681106,
      "grad_norm": 218.33396911621094,
      "learning_rate": 0.00018998192568670864,
      "loss": 7.4225,
      "step": 123860
    },
    {
      "epoch": 14.906137184115524,
      "grad_norm": 914.9017944335938,
      "learning_rate": 0.00018998026595986502,
      "loss": 7.4261,
      "step": 123870
    },
    {
      "epoch": 14.90734055354994,
      "grad_norm": 942.8607788085938,
      "learning_rate": 0.0001899786061027978,
      "loss": 7.4334,
      "step": 123880
    },
    {
      "epoch": 14.908543922984355,
      "grad_norm": 588.4299926757812,
      "learning_rate": 0.00018997694611550927,
      "loss": 7.4604,
      "step": 123890
    },
    {
      "epoch": 14.909747292418773,
      "grad_norm": 175.57318115234375,
      "learning_rate": 0.0001899752859980019,
      "loss": 7.4446,
      "step": 123900
    },
    {
      "epoch": 14.910950661853189,
      "grad_norm": 643.1263427734375,
      "learning_rate": 0.00018997362575027808,
      "loss": 7.4946,
      "step": 123910
    },
    {
      "epoch": 14.912154031287605,
      "grad_norm": 616.7755737304688,
      "learning_rate": 0.0001899719653723402,
      "loss": 7.2834,
      "step": 123920
    },
    {
      "epoch": 14.913357400722022,
      "grad_norm": 164.27496337890625,
      "learning_rate": 0.00018997030486419067,
      "loss": 7.3486,
      "step": 123930
    },
    {
      "epoch": 14.914560770156438,
      "grad_norm": 619.3140258789062,
      "learning_rate": 0.0001899686442258319,
      "loss": 7.2992,
      "step": 123940
    },
    {
      "epoch": 14.915764139590854,
      "grad_norm": 183.71238708496094,
      "learning_rate": 0.00018996698345726632,
      "loss": 7.3425,
      "step": 123950
    },
    {
      "epoch": 14.916967509025271,
      "grad_norm": 113.04483032226562,
      "learning_rate": 0.00018996532255849627,
      "loss": 7.263,
      "step": 123960
    },
    {
      "epoch": 14.918170878459687,
      "grad_norm": 142.35398864746094,
      "learning_rate": 0.00018996366152952422,
      "loss": 7.3863,
      "step": 123970
    },
    {
      "epoch": 14.919374247894103,
      "grad_norm": 96.34976959228516,
      "learning_rate": 0.00018996200037035247,
      "loss": 7.4361,
      "step": 123980
    },
    {
      "epoch": 14.92057761732852,
      "grad_norm": 408.62603759765625,
      "learning_rate": 0.00018996033908098356,
      "loss": 7.3373,
      "step": 123990
    },
    {
      "epoch": 14.921780986762936,
      "grad_norm": 147.46519470214844,
      "learning_rate": 0.00018995867766141982,
      "loss": 7.3312,
      "step": 124000
    },
    {
      "epoch": 14.922984356197352,
      "grad_norm": 178.59030151367188,
      "learning_rate": 0.00018995701611166364,
      "loss": 7.3884,
      "step": 124010
    },
    {
      "epoch": 14.92418772563177,
      "grad_norm": 722.7620239257812,
      "learning_rate": 0.00018995535443171746,
      "loss": 7.293,
      "step": 124020
    },
    {
      "epoch": 14.925391095066185,
      "grad_norm": 154.78280639648438,
      "learning_rate": 0.00018995369262158368,
      "loss": 7.4163,
      "step": 124030
    },
    {
      "epoch": 14.926594464500601,
      "grad_norm": 142.9695587158203,
      "learning_rate": 0.00018995203068126467,
      "loss": 7.3729,
      "step": 124040
    },
    {
      "epoch": 14.927797833935019,
      "grad_norm": 449.0110778808594,
      "learning_rate": 0.00018995036861076288,
      "loss": 7.3684,
      "step": 124050
    },
    {
      "epoch": 14.929001203369435,
      "grad_norm": 150.65570068359375,
      "learning_rate": 0.00018994870641008072,
      "loss": 7.302,
      "step": 124060
    },
    {
      "epoch": 14.93020457280385,
      "grad_norm": 751.1250610351562,
      "learning_rate": 0.00018994704407922054,
      "loss": 7.3357,
      "step": 124070
    },
    {
      "epoch": 14.931407942238268,
      "grad_norm": 525.4304809570312,
      "learning_rate": 0.0001899453816181848,
      "loss": 7.2665,
      "step": 124080
    },
    {
      "epoch": 14.932611311672684,
      "grad_norm": 292.63311767578125,
      "learning_rate": 0.00018994371902697588,
      "loss": 7.4467,
      "step": 124090
    },
    {
      "epoch": 14.9338146811071,
      "grad_norm": 164.7617645263672,
      "learning_rate": 0.00018994205630559622,
      "loss": 7.2848,
      "step": 124100
    },
    {
      "epoch": 14.935018050541515,
      "grad_norm": 789.33349609375,
      "learning_rate": 0.00018994039345404814,
      "loss": 7.3403,
      "step": 124110
    },
    {
      "epoch": 14.936221419975933,
      "grad_norm": 412.9779357910156,
      "learning_rate": 0.00018993873047233416,
      "loss": 7.3437,
      "step": 124120
    },
    {
      "epoch": 14.937424789410349,
      "grad_norm": 211.40426635742188,
      "learning_rate": 0.0001899370673604566,
      "loss": 7.2826,
      "step": 124130
    },
    {
      "epoch": 14.938628158844764,
      "grad_norm": 798.9019775390625,
      "learning_rate": 0.0001899354041184179,
      "loss": 7.3176,
      "step": 124140
    },
    {
      "epoch": 14.939831528279182,
      "grad_norm": 475.837890625,
      "learning_rate": 0.00018993374074622048,
      "loss": 7.1957,
      "step": 124150
    },
    {
      "epoch": 14.941034897713598,
      "grad_norm": 414.1307678222656,
      "learning_rate": 0.00018993207724386674,
      "loss": 7.3616,
      "step": 124160
    },
    {
      "epoch": 14.942238267148014,
      "grad_norm": 243.13662719726562,
      "learning_rate": 0.00018993041361135905,
      "loss": 7.3629,
      "step": 124170
    },
    {
      "epoch": 14.943441636582431,
      "grad_norm": 89.33318328857422,
      "learning_rate": 0.00018992874984869986,
      "loss": 7.2794,
      "step": 124180
    },
    {
      "epoch": 14.944645006016847,
      "grad_norm": 486.8153381347656,
      "learning_rate": 0.00018992708595589157,
      "loss": 7.2819,
      "step": 124190
    },
    {
      "epoch": 14.945848375451263,
      "grad_norm": 128.45579528808594,
      "learning_rate": 0.00018992542193293658,
      "loss": 7.279,
      "step": 124200
    },
    {
      "epoch": 14.94705174488568,
      "grad_norm": 136.5570831298828,
      "learning_rate": 0.00018992375777983732,
      "loss": 7.2401,
      "step": 124210
    },
    {
      "epoch": 14.948255114320096,
      "grad_norm": 544.4719848632812,
      "learning_rate": 0.00018992209349659618,
      "loss": 7.2996,
      "step": 124220
    },
    {
      "epoch": 14.949458483754512,
      "grad_norm": 807.21044921875,
      "learning_rate": 0.00018992042908321554,
      "loss": 7.3783,
      "step": 124230
    },
    {
      "epoch": 14.95066185318893,
      "grad_norm": 732.8168334960938,
      "learning_rate": 0.00018991876453969787,
      "loss": 7.3139,
      "step": 124240
    },
    {
      "epoch": 14.951865222623345,
      "grad_norm": 722.955322265625,
      "learning_rate": 0.00018991709986604556,
      "loss": 7.3174,
      "step": 124250
    },
    {
      "epoch": 14.953068592057761,
      "grad_norm": 178.8930206298828,
      "learning_rate": 0.00018991543506226098,
      "loss": 7.3048,
      "step": 124260
    },
    {
      "epoch": 14.954271961492179,
      "grad_norm": 419.10394287109375,
      "learning_rate": 0.00018991377012834656,
      "loss": 7.2753,
      "step": 124270
    },
    {
      "epoch": 14.955475330926594,
      "grad_norm": 396.1692199707031,
      "learning_rate": 0.00018991210506430473,
      "loss": 7.3006,
      "step": 124280
    },
    {
      "epoch": 14.95667870036101,
      "grad_norm": 255.86659240722656,
      "learning_rate": 0.00018991043987013785,
      "loss": 7.2862,
      "step": 124290
    },
    {
      "epoch": 14.957882069795428,
      "grad_norm": 118.07201385498047,
      "learning_rate": 0.0001899087745458484,
      "loss": 7.2756,
      "step": 124300
    },
    {
      "epoch": 14.959085439229844,
      "grad_norm": 149.11395263671875,
      "learning_rate": 0.00018990710909143877,
      "loss": 7.2947,
      "step": 124310
    },
    {
      "epoch": 14.96028880866426,
      "grad_norm": 4135.94873046875,
      "learning_rate": 0.00018990544350691132,
      "loss": 7.2493,
      "step": 124320
    },
    {
      "epoch": 14.961492178098677,
      "grad_norm": 2144.12060546875,
      "learning_rate": 0.00018990377779226853,
      "loss": 7.4241,
      "step": 124330
    },
    {
      "epoch": 14.962695547533093,
      "grad_norm": 1090.0350341796875,
      "learning_rate": 0.00018990211194751274,
      "loss": 7.29,
      "step": 124340
    },
    {
      "epoch": 14.963898916967509,
      "grad_norm": 300.70428466796875,
      "learning_rate": 0.00018990044597264643,
      "loss": 7.2437,
      "step": 124350
    },
    {
      "epoch": 14.965102286401926,
      "grad_norm": 395.0734558105469,
      "learning_rate": 0.00018989877986767196,
      "loss": 7.297,
      "step": 124360
    },
    {
      "epoch": 14.966305655836342,
      "grad_norm": 218.81089782714844,
      "learning_rate": 0.00018989711363259177,
      "loss": 7.1761,
      "step": 124370
    },
    {
      "epoch": 14.967509025270758,
      "grad_norm": 116.31067657470703,
      "learning_rate": 0.00018989544726740826,
      "loss": 7.2163,
      "step": 124380
    },
    {
      "epoch": 14.968712394705175,
      "grad_norm": 694.2081909179688,
      "learning_rate": 0.00018989378077212385,
      "loss": 7.133,
      "step": 124390
    },
    {
      "epoch": 14.969915764139591,
      "grad_norm": 417.2356262207031,
      "learning_rate": 0.00018989211414674096,
      "loss": 7.385,
      "step": 124400
    },
    {
      "epoch": 14.971119133574007,
      "grad_norm": 157.76431274414062,
      "learning_rate": 0.00018989044739126194,
      "loss": 7.2909,
      "step": 124410
    },
    {
      "epoch": 14.972322503008424,
      "grad_norm": 244.60145568847656,
      "learning_rate": 0.00018988878050568928,
      "loss": 7.2631,
      "step": 124420
    },
    {
      "epoch": 14.97352587244284,
      "grad_norm": 106.88508605957031,
      "learning_rate": 0.00018988711349002534,
      "loss": 7.3083,
      "step": 124430
    },
    {
      "epoch": 14.974729241877256,
      "grad_norm": 333.1745300292969,
      "learning_rate": 0.00018988544634427257,
      "loss": 7.3695,
      "step": 124440
    },
    {
      "epoch": 14.975932611311674,
      "grad_norm": 2485.152099609375,
      "learning_rate": 0.00018988377906843337,
      "loss": 7.3444,
      "step": 124450
    },
    {
      "epoch": 14.97713598074609,
      "grad_norm": 5089.16796875,
      "learning_rate": 0.00018988211166251015,
      "loss": 7.3382,
      "step": 124460
    },
    {
      "epoch": 14.978339350180505,
      "grad_norm": 5350.455078125,
      "learning_rate": 0.0001898804441265053,
      "loss": 7.2943,
      "step": 124470
    },
    {
      "epoch": 14.979542719614923,
      "grad_norm": 15196.720703125,
      "learning_rate": 0.00018987877646042128,
      "loss": 7.3234,
      "step": 124480
    },
    {
      "epoch": 14.980746089049338,
      "grad_norm": 11147.7783203125,
      "learning_rate": 0.0001898771086642605,
      "loss": 7.3459,
      "step": 124490
    },
    {
      "epoch": 14.981949458483754,
      "grad_norm": 760.6731567382812,
      "learning_rate": 0.00018987544073802533,
      "loss": 7.4078,
      "step": 124500
    },
    {
      "epoch": 14.98315282791817,
      "grad_norm": 145.38235473632812,
      "learning_rate": 0.0001898737726817182,
      "loss": 7.5813,
      "step": 124510
    },
    {
      "epoch": 14.984356197352588,
      "grad_norm": 398.988525390625,
      "learning_rate": 0.00018987210449534155,
      "loss": 7.4103,
      "step": 124520
    },
    {
      "epoch": 14.985559566787003,
      "grad_norm": 181.80856323242188,
      "learning_rate": 0.00018987043617889777,
      "loss": 7.4287,
      "step": 124530
    },
    {
      "epoch": 14.98676293622142,
      "grad_norm": 9107.564453125,
      "learning_rate": 0.0001898687677323893,
      "loss": 7.3466,
      "step": 124540
    },
    {
      "epoch": 14.987966305655837,
      "grad_norm": 9322.19140625,
      "learning_rate": 0.00018986709915581851,
      "loss": 7.2363,
      "step": 124550
    },
    {
      "epoch": 14.989169675090253,
      "grad_norm": 1166.1513671875,
      "learning_rate": 0.00018986543044918782,
      "loss": 7.3837,
      "step": 124560
    },
    {
      "epoch": 14.990373044524668,
      "grad_norm": 337.0623779296875,
      "learning_rate": 0.00018986376161249968,
      "loss": 7.3834,
      "step": 124570
    },
    {
      "epoch": 14.991576413959086,
      "grad_norm": 270.37750244140625,
      "learning_rate": 0.00018986209264575652,
      "loss": 7.3116,
      "step": 124580
    },
    {
      "epoch": 14.992779783393502,
      "grad_norm": 167.70469665527344,
      "learning_rate": 0.0001898604235489607,
      "loss": 7.4191,
      "step": 124590
    },
    {
      "epoch": 14.993983152827917,
      "grad_norm": 221.2666778564453,
      "learning_rate": 0.00018985875432211467,
      "loss": 7.3555,
      "step": 124600
    },
    {
      "epoch": 14.995186522262335,
      "grad_norm": 138.39874267578125,
      "learning_rate": 0.00018985708496522082,
      "loss": 7.4054,
      "step": 124610
    },
    {
      "epoch": 14.99638989169675,
      "grad_norm": 1753.5211181640625,
      "learning_rate": 0.00018985541547828158,
      "loss": 7.3489,
      "step": 124620
    },
    {
      "epoch": 14.997593261131167,
      "grad_norm": 1907.968017578125,
      "learning_rate": 0.0001898537458612994,
      "loss": 7.2738,
      "step": 124630
    },
    {
      "epoch": 14.998796630565584,
      "grad_norm": 353.17498779296875,
      "learning_rate": 0.00018985207611427668,
      "loss": 7.2584,
      "step": 124640
    },
    {
      "epoch": 15.0,
      "grad_norm": 1369.762451171875,
      "learning_rate": 0.00018985040623721578,
      "loss": 7.367,
      "step": 124650
    },
    {
      "epoch": 15.0,
      "eval_loss": 7.337878227233887,
      "eval_runtime": 118.6149,
      "eval_samples_per_second": 62.277,
      "eval_steps_per_second": 7.79,
      "step": 124650
    },
    {
      "epoch": 15.001203369434416,
      "grad_norm": 576.1705932617188,
      "learning_rate": 0.00018984873623011916,
      "loss": 7.3064,
      "step": 124660
    },
    {
      "epoch": 15.002406738868833,
      "grad_norm": 588.5545654296875,
      "learning_rate": 0.00018984706609298922,
      "loss": 7.319,
      "step": 124670
    },
    {
      "epoch": 15.00361010830325,
      "grad_norm": 5784.7509765625,
      "learning_rate": 0.00018984539582582842,
      "loss": 7.4283,
      "step": 124680
    },
    {
      "epoch": 15.004813477737665,
      "grad_norm": 3700.33544921875,
      "learning_rate": 0.00018984372542863917,
      "loss": 7.3646,
      "step": 124690
    },
    {
      "epoch": 15.006016847172083,
      "grad_norm": 6456.92431640625,
      "learning_rate": 0.00018984205490142384,
      "loss": 7.382,
      "step": 124700
    },
    {
      "epoch": 15.007220216606498,
      "grad_norm": 5785.44287109375,
      "learning_rate": 0.00018984038424418488,
      "loss": 7.3675,
      "step": 124710
    },
    {
      "epoch": 15.008423586040914,
      "grad_norm": 6390.42431640625,
      "learning_rate": 0.0001898387134569247,
      "loss": 7.4846,
      "step": 124720
    },
    {
      "epoch": 15.009626955475332,
      "grad_norm": 95375.96875,
      "learning_rate": 0.00018983704253964573,
      "loss": 7.4069,
      "step": 124730
    },
    {
      "epoch": 15.010830324909747,
      "grad_norm": 157604.671875,
      "learning_rate": 0.00018983537149235038,
      "loss": 7.3716,
      "step": 124740
    },
    {
      "epoch": 15.012033694344163,
      "grad_norm": 20311.841796875,
      "learning_rate": 0.00018983370031504108,
      "loss": 7.402,
      "step": 124750
    },
    {
      "epoch": 15.01323706377858,
      "grad_norm": 159118.9375,
      "learning_rate": 0.0001898320290077202,
      "loss": 7.3826,
      "step": 124760
    },
    {
      "epoch": 15.014440433212997,
      "grad_norm": 42688.9296875,
      "learning_rate": 0.00018983035757039022,
      "loss": 7.5085,
      "step": 124770
    },
    {
      "epoch": 15.015643802647412,
      "grad_norm": 203326.390625,
      "learning_rate": 0.00018982868600305356,
      "loss": 7.614,
      "step": 124780
    },
    {
      "epoch": 15.01684717208183,
      "grad_norm": 71533.9453125,
      "learning_rate": 0.00018982701430571254,
      "loss": 7.6447,
      "step": 124790
    },
    {
      "epoch": 15.018050541516246,
      "grad_norm": 381853.0,
      "learning_rate": 0.0001898253424783697,
      "loss": 7.4278,
      "step": 124800
    },
    {
      "epoch": 15.019253910950662,
      "grad_norm": 345558.40625,
      "learning_rate": 0.00018982367052102742,
      "loss": 7.4888,
      "step": 124810
    },
    {
      "epoch": 15.020457280385079,
      "grad_norm": 30896.861328125,
      "learning_rate": 0.00018982199843368811,
      "loss": 7.4484,
      "step": 124820
    },
    {
      "epoch": 15.021660649819495,
      "grad_norm": 27468.91015625,
      "learning_rate": 0.00018982032621635416,
      "loss": 7.5939,
      "step": 124830
    },
    {
      "epoch": 15.02286401925391,
      "grad_norm": 88471.640625,
      "learning_rate": 0.00018981865386902806,
      "loss": 7.5852,
      "step": 124840
    },
    {
      "epoch": 15.024067388688326,
      "grad_norm": 121201.0859375,
      "learning_rate": 0.00018981698139171219,
      "loss": 7.6381,
      "step": 124850
    },
    {
      "epoch": 15.025270758122744,
      "grad_norm": 268526.6875,
      "learning_rate": 0.00018981530878440893,
      "loss": 7.5442,
      "step": 124860
    },
    {
      "epoch": 15.02647412755716,
      "grad_norm": 38136.21875,
      "learning_rate": 0.00018981363604712078,
      "loss": 7.6097,
      "step": 124870
    },
    {
      "epoch": 15.027677496991576,
      "grad_norm": 92757.8203125,
      "learning_rate": 0.00018981196317985013,
      "loss": 7.5753,
      "step": 124880
    },
    {
      "epoch": 15.028880866425993,
      "grad_norm": 46574.66015625,
      "learning_rate": 0.00018981029018259936,
      "loss": 7.5345,
      "step": 124890
    },
    {
      "epoch": 15.030084235860409,
      "grad_norm": 23084.111328125,
      "learning_rate": 0.00018980861705537096,
      "loss": 7.6102,
      "step": 124900
    },
    {
      "epoch": 15.031287605294825,
      "grad_norm": 11719.412109375,
      "learning_rate": 0.00018980694379816732,
      "loss": 7.5387,
      "step": 124910
    },
    {
      "epoch": 15.032490974729242,
      "grad_norm": 17841.978515625,
      "learning_rate": 0.00018980527041099086,
      "loss": 7.6421,
      "step": 124920
    },
    {
      "epoch": 15.033694344163658,
      "grad_norm": 3395.980712890625,
      "learning_rate": 0.00018980359689384397,
      "loss": 7.6997,
      "step": 124930
    },
    {
      "epoch": 15.034897713598074,
      "grad_norm": 40489.2734375,
      "learning_rate": 0.00018980192324672912,
      "loss": 7.6845,
      "step": 124940
    },
    {
      "epoch": 15.036101083032491,
      "grad_norm": 31771.244140625,
      "learning_rate": 0.00018980024946964872,
      "loss": 7.5917,
      "step": 124950
    },
    {
      "epoch": 15.037304452466907,
      "grad_norm": 15742.474609375,
      "learning_rate": 0.00018979857556260517,
      "loss": 7.5517,
      "step": 124960
    },
    {
      "epoch": 15.038507821901323,
      "grad_norm": 25554.30078125,
      "learning_rate": 0.00018979690152560094,
      "loss": 7.595,
      "step": 124970
    },
    {
      "epoch": 15.03971119133574,
      "grad_norm": 10457.7578125,
      "learning_rate": 0.0001897952273586384,
      "loss": 7.5729,
      "step": 124980
    },
    {
      "epoch": 15.040914560770156,
      "grad_norm": 52429.6484375,
      "learning_rate": 0.00018979355306172001,
      "loss": 7.5176,
      "step": 124990
    },
    {
      "epoch": 15.042117930204572,
      "grad_norm": 15750.4541015625,
      "learning_rate": 0.00018979187863484817,
      "loss": 7.5973,
      "step": 125000
    },
    {
      "epoch": 15.04332129963899,
      "grad_norm": 46397.9453125,
      "learning_rate": 0.00018979020407802532,
      "loss": 7.5389,
      "step": 125010
    },
    {
      "epoch": 15.044524669073406,
      "grad_norm": 5923.57275390625,
      "learning_rate": 0.00018978852939125388,
      "loss": 7.6101,
      "step": 125020
    },
    {
      "epoch": 15.045728038507821,
      "grad_norm": 5345.36767578125,
      "learning_rate": 0.00018978685457453627,
      "loss": 7.6017,
      "step": 125030
    },
    {
      "epoch": 15.046931407942239,
      "grad_norm": 26302.05078125,
      "learning_rate": 0.0001897851796278749,
      "loss": 7.5601,
      "step": 125040
    },
    {
      "epoch": 15.048134777376655,
      "grad_norm": 9897.3740234375,
      "learning_rate": 0.00018978350455127222,
      "loss": 7.5551,
      "step": 125050
    },
    {
      "epoch": 15.04933814681107,
      "grad_norm": 19237822.0,
      "learning_rate": 0.00018978182934473064,
      "loss": 7.5993,
      "step": 125060
    },
    {
      "epoch": 15.050541516245488,
      "grad_norm": 6017680.5,
      "learning_rate": 0.00018978015400825256,
      "loss": 7.5672,
      "step": 125070
    },
    {
      "epoch": 15.051744885679904,
      "grad_norm": 42625164.0,
      "learning_rate": 0.00018977847854184047,
      "loss": 7.6082,
      "step": 125080
    },
    {
      "epoch": 15.05294825511432,
      "grad_norm": 41006348.0,
      "learning_rate": 0.00018977680294549675,
      "loss": 7.5719,
      "step": 125090
    },
    {
      "epoch": 15.054151624548737,
      "grad_norm": 6016879.5,
      "learning_rate": 0.0001897751272192238,
      "loss": 7.6385,
      "step": 125100
    },
    {
      "epoch": 15.055354993983153,
      "grad_norm": 5245343.5,
      "learning_rate": 0.00018977345136302415,
      "loss": 7.6785,
      "step": 125110
    },
    {
      "epoch": 15.056558363417569,
      "grad_norm": 1546680.5,
      "learning_rate": 0.00018977177537690007,
      "loss": 7.6086,
      "step": 125120
    },
    {
      "epoch": 15.057761732851986,
      "grad_norm": 3506114.75,
      "learning_rate": 0.00018977009926085411,
      "loss": 7.7258,
      "step": 125130
    },
    {
      "epoch": 15.058965102286402,
      "grad_norm": 13028783.0,
      "learning_rate": 0.00018976842301488865,
      "loss": 7.6379,
      "step": 125140
    },
    {
      "epoch": 15.060168471720818,
      "grad_norm": 38439512.0,
      "learning_rate": 0.0001897667466390061,
      "loss": 7.6399,
      "step": 125150
    },
    {
      "epoch": 15.061371841155236,
      "grad_norm": 5304360.5,
      "learning_rate": 0.0001897650701332089,
      "loss": 7.6453,
      "step": 125160
    },
    {
      "epoch": 15.062575210589651,
      "grad_norm": 114621320.0,
      "learning_rate": 0.0001897633934974995,
      "loss": 7.6827,
      "step": 125170
    },
    {
      "epoch": 15.063778580024067,
      "grad_norm": 38171608.0,
      "learning_rate": 0.0001897617167318803,
      "loss": 7.6337,
      "step": 125180
    },
    {
      "epoch": 15.064981949458483,
      "grad_norm": 56016476.0,
      "learning_rate": 0.00018976003983635375,
      "loss": 7.63,
      "step": 125190
    },
    {
      "epoch": 15.0661853188929,
      "grad_norm": 30925518.0,
      "learning_rate": 0.00018975836281092225,
      "loss": 7.7324,
      "step": 125200
    },
    {
      "epoch": 15.067388688327316,
      "grad_norm": 68823728.0,
      "learning_rate": 0.0001897566856555882,
      "loss": 7.6188,
      "step": 125210
    },
    {
      "epoch": 15.068592057761732,
      "grad_norm": 3008264448.0,
      "learning_rate": 0.00018975500837035413,
      "loss": 7.6448,
      "step": 125220
    },
    {
      "epoch": 15.06979542719615,
      "grad_norm": 61769200.0,
      "learning_rate": 0.00018975333095522237,
      "loss": 7.5929,
      "step": 125230
    },
    {
      "epoch": 15.070998796630565,
      "grad_norm": 33129292.0,
      "learning_rate": 0.00018975165341019538,
      "loss": 7.468,
      "step": 125240
    },
    {
      "epoch": 15.072202166064981,
      "grad_norm": 403965568.0,
      "learning_rate": 0.0001897499757352756,
      "loss": 7.5362,
      "step": 125250
    },
    {
      "epoch": 15.073405535499399,
      "grad_norm": 67818064.0,
      "learning_rate": 0.00018974829793046545,
      "loss": 7.4803,
      "step": 125260
    },
    {
      "epoch": 15.074608904933815,
      "grad_norm": 91769424.0,
      "learning_rate": 0.00018974661999576734,
      "loss": 7.5615,
      "step": 125270
    },
    {
      "epoch": 15.07581227436823,
      "grad_norm": 11050623.0,
      "learning_rate": 0.00018974494193118373,
      "loss": 7.5256,
      "step": 125280
    },
    {
      "epoch": 15.077015643802648,
      "grad_norm": 124145520.0,
      "learning_rate": 0.000189743263736717,
      "loss": 7.5175,
      "step": 125290
    },
    {
      "epoch": 15.078219013237064,
      "grad_norm": 265549664.0,
      "learning_rate": 0.00018974158541236965,
      "loss": 7.4579,
      "step": 125300
    },
    {
      "epoch": 15.07942238267148,
      "grad_norm": 617909696.0,
      "learning_rate": 0.00018973990695814405,
      "loss": 7.5923,
      "step": 125310
    },
    {
      "epoch": 15.080625752105897,
      "grad_norm": 245739968.0,
      "learning_rate": 0.00018973822837404264,
      "loss": 7.5682,
      "step": 125320
    },
    {
      "epoch": 15.081829121540313,
      "grad_norm": 87915488.0,
      "learning_rate": 0.00018973654966006787,
      "loss": 7.6359,
      "step": 125330
    },
    {
      "epoch": 15.083032490974729,
      "grad_norm": 55169520.0,
      "learning_rate": 0.00018973487081622216,
      "loss": 7.5563,
      "step": 125340
    },
    {
      "epoch": 15.084235860409146,
      "grad_norm": 916735040.0,
      "learning_rate": 0.00018973319184250795,
      "loss": 7.5715,
      "step": 125350
    },
    {
      "epoch": 15.085439229843562,
      "grad_norm": 11494315.0,
      "learning_rate": 0.00018973151273892763,
      "loss": 7.459,
      "step": 125360
    },
    {
      "epoch": 15.086642599277978,
      "grad_norm": 362630496.0,
      "learning_rate": 0.00018972983350548365,
      "loss": 7.5424,
      "step": 125370
    },
    {
      "epoch": 15.087845968712395,
      "grad_norm": 256943984.0,
      "learning_rate": 0.00018972815414217847,
      "loss": 7.5612,
      "step": 125380
    },
    {
      "epoch": 15.089049338146811,
      "grad_norm": 39302696.0,
      "learning_rate": 0.0001897264746490145,
      "loss": 7.5001,
      "step": 125390
    },
    {
      "epoch": 15.090252707581227,
      "grad_norm": 50235272.0,
      "learning_rate": 0.00018972479502599415,
      "loss": 7.485,
      "step": 125400
    },
    {
      "epoch": 15.091456077015645,
      "grad_norm": 2478366.5,
      "learning_rate": 0.00018972311527311987,
      "loss": 7.4595,
      "step": 125410
    },
    {
      "epoch": 15.09265944645006,
      "grad_norm": 25049776.0,
      "learning_rate": 0.00018972143539039413,
      "loss": 7.4178,
      "step": 125420
    },
    {
      "epoch": 15.093862815884476,
      "grad_norm": 44553648.0,
      "learning_rate": 0.00018971975537781927,
      "loss": 7.3734,
      "step": 125430
    },
    {
      "epoch": 15.095066185318894,
      "grad_norm": 25317616.0,
      "learning_rate": 0.0001897180752353978,
      "loss": 7.3609,
      "step": 125440
    },
    {
      "epoch": 15.09626955475331,
      "grad_norm": 43974652.0,
      "learning_rate": 0.00018971639496313212,
      "loss": 7.321,
      "step": 125450
    },
    {
      "epoch": 15.097472924187725,
      "grad_norm": 76693800.0,
      "learning_rate": 0.00018971471456102464,
      "loss": 7.452,
      "step": 125460
    },
    {
      "epoch": 15.098676293622143,
      "grad_norm": 48645816.0,
      "learning_rate": 0.00018971303402907786,
      "loss": 7.3009,
      "step": 125470
    },
    {
      "epoch": 15.099879663056559,
      "grad_norm": 185690368.0,
      "learning_rate": 0.00018971135336729415,
      "loss": 7.4491,
      "step": 125480
    },
    {
      "epoch": 15.101083032490974,
      "grad_norm": 32901132.0,
      "learning_rate": 0.00018970967257567596,
      "loss": 7.4528,
      "step": 125490
    },
    {
      "epoch": 15.102286401925392,
      "grad_norm": 124543952.0,
      "learning_rate": 0.00018970799165422575,
      "loss": 7.4281,
      "step": 125500
    },
    {
      "epoch": 15.103489771359808,
      "grad_norm": 51466192.0,
      "learning_rate": 0.0001897063106029459,
      "loss": 7.3982,
      "step": 125510
    },
    {
      "epoch": 15.104693140794224,
      "grad_norm": 66691996.0,
      "learning_rate": 0.00018970462942183888,
      "loss": 7.5264,
      "step": 125520
    },
    {
      "epoch": 15.10589651022864,
      "grad_norm": 123348960.0,
      "learning_rate": 0.00018970294811090712,
      "loss": 7.3115,
      "step": 125530
    },
    {
      "epoch": 15.107099879663057,
      "grad_norm": 127885784.0,
      "learning_rate": 0.00018970126667015302,
      "loss": 7.5694,
      "step": 125540
    },
    {
      "epoch": 15.108303249097473,
      "grad_norm": 100609040.0,
      "learning_rate": 0.00018969958509957907,
      "loss": 7.3544,
      "step": 125550
    },
    {
      "epoch": 15.109506618531888,
      "grad_norm": 95557008.0,
      "learning_rate": 0.00018969790339918766,
      "loss": 7.4631,
      "step": 125560
    },
    {
      "epoch": 15.110709987966306,
      "grad_norm": 20857170.0,
      "learning_rate": 0.00018969622156898126,
      "loss": 7.4274,
      "step": 125570
    },
    {
      "epoch": 15.111913357400722,
      "grad_norm": 24841854.0,
      "learning_rate": 0.00018969453960896227,
      "loss": 7.3387,
      "step": 125580
    },
    {
      "epoch": 15.113116726835138,
      "grad_norm": 75238168.0,
      "learning_rate": 0.00018969285751913313,
      "loss": 7.4576,
      "step": 125590
    },
    {
      "epoch": 15.114320096269555,
      "grad_norm": 4234657.5,
      "learning_rate": 0.00018969117529949627,
      "loss": 7.5172,
      "step": 125600
    },
    {
      "epoch": 15.115523465703971,
      "grad_norm": 12823890.0,
      "learning_rate": 0.00018968949295005417,
      "loss": 7.4722,
      "step": 125610
    },
    {
      "epoch": 15.116726835138387,
      "grad_norm": 89470008.0,
      "learning_rate": 0.0001896878104708092,
      "loss": 7.5282,
      "step": 125620
    },
    {
      "epoch": 15.117930204572804,
      "grad_norm": 10607944.0,
      "learning_rate": 0.00018968612786176382,
      "loss": 7.6361,
      "step": 125630
    },
    {
      "epoch": 15.11913357400722,
      "grad_norm": 6230450.5,
      "learning_rate": 0.0001896844451229205,
      "loss": 7.6768,
      "step": 125640
    },
    {
      "epoch": 15.120336943441636,
      "grad_norm": 113595872.0,
      "learning_rate": 0.0001896827622542816,
      "loss": 7.6624,
      "step": 125650
    },
    {
      "epoch": 15.121540312876053,
      "grad_norm": 25938984.0,
      "learning_rate": 0.00018968107925584966,
      "loss": 7.4947,
      "step": 125660
    },
    {
      "epoch": 15.12274368231047,
      "grad_norm": 35611652.0,
      "learning_rate": 0.000189679396127627,
      "loss": 7.5701,
      "step": 125670
    },
    {
      "epoch": 15.123947051744885,
      "grad_norm": 33624144.0,
      "learning_rate": 0.00018967771286961614,
      "loss": 7.6416,
      "step": 125680
    },
    {
      "epoch": 15.125150421179303,
      "grad_norm": 92578088.0,
      "learning_rate": 0.0001896760294818195,
      "loss": 7.6108,
      "step": 125690
    },
    {
      "epoch": 15.126353790613718,
      "grad_norm": 8069914.5,
      "learning_rate": 0.00018967434596423946,
      "loss": 7.4654,
      "step": 125700
    },
    {
      "epoch": 15.127557160048134,
      "grad_norm": 98823008.0,
      "learning_rate": 0.00018967266231687856,
      "loss": 7.4354,
      "step": 125710
    },
    {
      "epoch": 15.128760529482552,
      "grad_norm": 238709248.0,
      "learning_rate": 0.00018967097853973917,
      "loss": 7.4948,
      "step": 125720
    },
    {
      "epoch": 15.129963898916968,
      "grad_norm": 194297568.0,
      "learning_rate": 0.0001896692946328237,
      "loss": 7.4762,
      "step": 125730
    },
    {
      "epoch": 15.131167268351383,
      "grad_norm": 198150320.0,
      "learning_rate": 0.00018966761059613466,
      "loss": 7.5298,
      "step": 125740
    },
    {
      "epoch": 15.132370637785801,
      "grad_norm": 2562157.25,
      "learning_rate": 0.00018966592642967442,
      "loss": 7.6936,
      "step": 125750
    },
    {
      "epoch": 15.133574007220217,
      "grad_norm": 496502944.0,
      "learning_rate": 0.00018966424213344543,
      "loss": 7.6467,
      "step": 125760
    },
    {
      "epoch": 15.134777376654633,
      "grad_norm": 4172826.25,
      "learning_rate": 0.00018966255770745018,
      "loss": 7.6105,
      "step": 125770
    },
    {
      "epoch": 15.13598074608905,
      "grad_norm": 3981231.75,
      "learning_rate": 0.00018966087315169108,
      "loss": 7.6741,
      "step": 125780
    },
    {
      "epoch": 15.137184115523466,
      "grad_norm": 15576460.0,
      "learning_rate": 0.00018965918846617052,
      "loss": 7.5744,
      "step": 125790
    },
    {
      "epoch": 15.138387484957882,
      "grad_norm": 1738140.375,
      "learning_rate": 0.00018965750365089102,
      "loss": 7.6001,
      "step": 125800
    },
    {
      "epoch": 15.1395908543923,
      "grad_norm": 17191982.0,
      "learning_rate": 0.00018965581870585493,
      "loss": 7.6273,
      "step": 125810
    },
    {
      "epoch": 15.140794223826715,
      "grad_norm": 91394256.0,
      "learning_rate": 0.0001896541336310648,
      "loss": 7.6291,
      "step": 125820
    },
    {
      "epoch": 15.14199759326113,
      "grad_norm": 9186479.0,
      "learning_rate": 0.00018965244842652294,
      "loss": 7.6816,
      "step": 125830
    },
    {
      "epoch": 15.143200962695548,
      "grad_norm": 20866930.0,
      "learning_rate": 0.00018965076309223188,
      "loss": 7.6359,
      "step": 125840
    },
    {
      "epoch": 15.144404332129964,
      "grad_norm": 1660808.25,
      "learning_rate": 0.00018964907762819403,
      "loss": 7.63,
      "step": 125850
    },
    {
      "epoch": 15.14560770156438,
      "grad_norm": 2959039.5,
      "learning_rate": 0.00018964739203441183,
      "loss": 7.7153,
      "step": 125860
    },
    {
      "epoch": 15.146811070998796,
      "grad_norm": 56636160.0,
      "learning_rate": 0.0001896457063108877,
      "loss": 7.7589,
      "step": 125870
    },
    {
      "epoch": 15.148014440433213,
      "grad_norm": 880030.6875,
      "learning_rate": 0.00018964402045762416,
      "loss": 7.6986,
      "step": 125880
    },
    {
      "epoch": 15.14921780986763,
      "grad_norm": 2088721.625,
      "learning_rate": 0.00018964233447462355,
      "loss": 7.6734,
      "step": 125890
    },
    {
      "epoch": 15.150421179302045,
      "grad_norm": 282244.65625,
      "learning_rate": 0.00018964064836188835,
      "loss": 7.5968,
      "step": 125900
    },
    {
      "epoch": 15.151624548736462,
      "grad_norm": 1778353.75,
      "learning_rate": 0.000189638962119421,
      "loss": 7.5881,
      "step": 125910
    },
    {
      "epoch": 15.152827918170878,
      "grad_norm": 81155880.0,
      "learning_rate": 0.00018963727574722392,
      "loss": 7.5837,
      "step": 125920
    },
    {
      "epoch": 15.154031287605294,
      "grad_norm": 1443482.0,
      "learning_rate": 0.00018963558924529962,
      "loss": 7.6416,
      "step": 125930
    },
    {
      "epoch": 15.155234657039712,
      "grad_norm": 371085.125,
      "learning_rate": 0.00018963390261365044,
      "loss": 7.6189,
      "step": 125940
    },
    {
      "epoch": 15.156438026474127,
      "grad_norm": 1904537.25,
      "learning_rate": 0.00018963221585227894,
      "loss": 7.6868,
      "step": 125950
    },
    {
      "epoch": 15.157641395908543,
      "grad_norm": 398652.34375,
      "learning_rate": 0.00018963052896118742,
      "loss": 7.7223,
      "step": 125960
    },
    {
      "epoch": 15.15884476534296,
      "grad_norm": 7054542.0,
      "learning_rate": 0.00018962884194037845,
      "loss": 7.5471,
      "step": 125970
    },
    {
      "epoch": 15.160048134777377,
      "grad_norm": 548993.375,
      "learning_rate": 0.00018962715478985437,
      "loss": 7.6031,
      "step": 125980
    },
    {
      "epoch": 15.161251504211792,
      "grad_norm": 6219262.0,
      "learning_rate": 0.0001896254675096177,
      "loss": 7.6288,
      "step": 125990
    },
    {
      "epoch": 15.16245487364621,
      "grad_norm": 13239046.0,
      "learning_rate": 0.00018962378009967085,
      "loss": 7.6348,
      "step": 126000
    },
    {
      "epoch": 15.163658243080626,
      "grad_norm": 3554142.5,
      "learning_rate": 0.00018962209256001623,
      "loss": 7.6106,
      "step": 126010
    },
    {
      "epoch": 15.164861612515042,
      "grad_norm": 6099709.5,
      "learning_rate": 0.00018962040489065637,
      "loss": 7.5621,
      "step": 126020
    },
    {
      "epoch": 15.166064981949459,
      "grad_norm": 7998756.5,
      "learning_rate": 0.00018961871709159364,
      "loss": 7.6405,
      "step": 126030
    },
    {
      "epoch": 15.167268351383875,
      "grad_norm": 3567050.75,
      "learning_rate": 0.0001896170291628305,
      "loss": 7.6215,
      "step": 126040
    },
    {
      "epoch": 15.16847172081829,
      "grad_norm": 6089238.0,
      "learning_rate": 0.00018961534110436938,
      "loss": 7.5543,
      "step": 126050
    },
    {
      "epoch": 15.169675090252708,
      "grad_norm": 2044088.625,
      "learning_rate": 0.00018961365291621273,
      "loss": 7.6081,
      "step": 126060
    },
    {
      "epoch": 15.170878459687124,
      "grad_norm": 3017435.0,
      "learning_rate": 0.000189611964598363,
      "loss": 7.625,
      "step": 126070
    },
    {
      "epoch": 15.17208182912154,
      "grad_norm": 765081.5,
      "learning_rate": 0.00018961027615082266,
      "loss": 7.7568,
      "step": 126080
    },
    {
      "epoch": 15.173285198555957,
      "grad_norm": 1368187.875,
      "learning_rate": 0.0001896085875735941,
      "loss": 7.5899,
      "step": 126090
    },
    {
      "epoch": 15.174488567990373,
      "grad_norm": 3677093.25,
      "learning_rate": 0.0001896068988666798,
      "loss": 7.6117,
      "step": 126100
    },
    {
      "epoch": 15.175691937424789,
      "grad_norm": 51560796.0,
      "learning_rate": 0.0001896052100300822,
      "loss": 7.6965,
      "step": 126110
    },
    {
      "epoch": 15.176895306859207,
      "grad_norm": 29220676.0,
      "learning_rate": 0.00018960352106380372,
      "loss": 7.6417,
      "step": 126120
    },
    {
      "epoch": 15.178098676293622,
      "grad_norm": 115824040.0,
      "learning_rate": 0.00018960183196784684,
      "loss": 7.6137,
      "step": 126130
    },
    {
      "epoch": 15.179302045728038,
      "grad_norm": 67169680.0,
      "learning_rate": 0.00018960014274221396,
      "loss": 7.6809,
      "step": 126140
    },
    {
      "epoch": 15.180505415162456,
      "grad_norm": 52605844.0,
      "learning_rate": 0.00018959845338690758,
      "loss": 7.6398,
      "step": 126150
    },
    {
      "epoch": 15.181708784596871,
      "grad_norm": 51805612.0,
      "learning_rate": 0.00018959676390193012,
      "loss": 7.5682,
      "step": 126160
    },
    {
      "epoch": 15.182912154031287,
      "grad_norm": 121982488.0,
      "learning_rate": 0.000189595074287284,
      "loss": 7.6199,
      "step": 126170
    },
    {
      "epoch": 15.184115523465705,
      "grad_norm": 17981904.0,
      "learning_rate": 0.0001895933845429717,
      "loss": 7.7416,
      "step": 126180
    },
    {
      "epoch": 15.18531889290012,
      "grad_norm": 165518880.0,
      "learning_rate": 0.00018959169466899563,
      "loss": 7.4758,
      "step": 126190
    },
    {
      "epoch": 15.186522262334536,
      "grad_norm": 61010684.0,
      "learning_rate": 0.0001895900046653583,
      "loss": 7.6222,
      "step": 126200
    },
    {
      "epoch": 15.187725631768952,
      "grad_norm": 86460704.0,
      "learning_rate": 0.0001895883145320621,
      "loss": 7.6164,
      "step": 126210
    },
    {
      "epoch": 15.18892900120337,
      "grad_norm": 150151440.0,
      "learning_rate": 0.00018958662426910947,
      "loss": 7.6218,
      "step": 126220
    },
    {
      "epoch": 15.190132370637786,
      "grad_norm": 54037864.0,
      "learning_rate": 0.00018958493387650292,
      "loss": 7.6237,
      "step": 126230
    },
    {
      "epoch": 15.191335740072201,
      "grad_norm": 43249288.0,
      "learning_rate": 0.0001895832433542448,
      "loss": 7.6659,
      "step": 126240
    },
    {
      "epoch": 15.192539109506619,
      "grad_norm": 571147008.0,
      "learning_rate": 0.00018958155270233765,
      "loss": 7.6348,
      "step": 126250
    },
    {
      "epoch": 15.193742478941035,
      "grad_norm": 90332880.0,
      "learning_rate": 0.00018957986192078385,
      "loss": 7.6374,
      "step": 126260
    },
    {
      "epoch": 15.19494584837545,
      "grad_norm": 32835612.0,
      "learning_rate": 0.0001895781710095859,
      "loss": 7.6133,
      "step": 126270
    },
    {
      "epoch": 15.196149217809868,
      "grad_norm": 45888972.0,
      "learning_rate": 0.0001895764799687462,
      "loss": 7.655,
      "step": 126280
    },
    {
      "epoch": 15.197352587244284,
      "grad_norm": 211303104.0,
      "learning_rate": 0.0001895747887982672,
      "loss": 7.7339,
      "step": 126290
    },
    {
      "epoch": 15.1985559566787,
      "grad_norm": 15292181.0,
      "learning_rate": 0.00018957309749815138,
      "loss": 7.6366,
      "step": 126300
    },
    {
      "epoch": 15.199759326113117,
      "grad_norm": 227544224.0,
      "learning_rate": 0.0001895714060684012,
      "loss": 7.6085,
      "step": 126310
    },
    {
      "epoch": 15.200962695547533,
      "grad_norm": 140794672.0,
      "learning_rate": 0.00018956971450901904,
      "loss": 7.6487,
      "step": 126320
    },
    {
      "epoch": 15.202166064981949,
      "grad_norm": 278590528.0,
      "learning_rate": 0.0001895680228200074,
      "loss": 7.6761,
      "step": 126330
    },
    {
      "epoch": 15.203369434416366,
      "grad_norm": 299309792.0,
      "learning_rate": 0.00018956633100136874,
      "loss": 7.6734,
      "step": 126340
    },
    {
      "epoch": 15.204572803850782,
      "grad_norm": 52255880.0,
      "learning_rate": 0.0001895646390531055,
      "loss": 7.6526,
      "step": 126350
    },
    {
      "epoch": 15.205776173285198,
      "grad_norm": 25389330.0,
      "learning_rate": 0.00018956294697522005,
      "loss": 7.6007,
      "step": 126360
    },
    {
      "epoch": 15.206979542719615,
      "grad_norm": 62796488.0,
      "learning_rate": 0.00018956125476771494,
      "loss": 7.6383,
      "step": 126370
    },
    {
      "epoch": 15.208182912154031,
      "grad_norm": 22353084.0,
      "learning_rate": 0.00018955956243059258,
      "loss": 7.5952,
      "step": 126380
    },
    {
      "epoch": 15.209386281588447,
      "grad_norm": 62999420.0,
      "learning_rate": 0.00018955786996385543,
      "loss": 7.5965,
      "step": 126390
    },
    {
      "epoch": 15.210589651022865,
      "grad_norm": 5550995.0,
      "learning_rate": 0.00018955617736750592,
      "loss": 7.6796,
      "step": 126400
    },
    {
      "epoch": 15.21179302045728,
      "grad_norm": 198051696.0,
      "learning_rate": 0.00018955448464154652,
      "loss": 7.603,
      "step": 126410
    },
    {
      "epoch": 15.212996389891696,
      "grad_norm": 142097216.0,
      "learning_rate": 0.00018955279178597967,
      "loss": 7.706,
      "step": 126420
    },
    {
      "epoch": 15.214199759326114,
      "grad_norm": 152284144.0,
      "learning_rate": 0.0001895510988008078,
      "loss": 7.5868,
      "step": 126430
    },
    {
      "epoch": 15.21540312876053,
      "grad_norm": 81759360.0,
      "learning_rate": 0.00018954940568603342,
      "loss": 7.6344,
      "step": 126440
    },
    {
      "epoch": 15.216606498194945,
      "grad_norm": 108964240.0,
      "learning_rate": 0.00018954771244165892,
      "loss": 7.6994,
      "step": 126450
    },
    {
      "epoch": 15.217809867629363,
      "grad_norm": 180551568.0,
      "learning_rate": 0.00018954601906768676,
      "loss": 7.5111,
      "step": 126460
    },
    {
      "epoch": 15.219013237063779,
      "grad_norm": 62206476.0,
      "learning_rate": 0.00018954432556411943,
      "loss": 7.6055,
      "step": 126470
    },
    {
      "epoch": 15.220216606498195,
      "grad_norm": 66520216.0,
      "learning_rate": 0.0001895426319309593,
      "loss": 7.5669,
      "step": 126480
    },
    {
      "epoch": 15.221419975932612,
      "grad_norm": 427407872.0,
      "learning_rate": 0.00018954093816820892,
      "loss": 7.5638,
      "step": 126490
    },
    {
      "epoch": 15.222623345367028,
      "grad_norm": 51074128.0,
      "learning_rate": 0.00018953924427587068,
      "loss": 7.7063,
      "step": 126500
    },
    {
      "epoch": 15.223826714801444,
      "grad_norm": 97677584.0,
      "learning_rate": 0.00018953755025394707,
      "loss": 7.4796,
      "step": 126510
    },
    {
      "epoch": 15.225030084235861,
      "grad_norm": 85726640.0,
      "learning_rate": 0.0001895358561024405,
      "loss": 7.6195,
      "step": 126520
    },
    {
      "epoch": 15.226233453670277,
      "grad_norm": 89334944.0,
      "learning_rate": 0.00018953416182135343,
      "loss": 7.6106,
      "step": 126530
    },
    {
      "epoch": 15.227436823104693,
      "grad_norm": 138687424.0,
      "learning_rate": 0.00018953246741068832,
      "loss": 7.6594,
      "step": 126540
    },
    {
      "epoch": 15.22864019253911,
      "grad_norm": 139341248.0,
      "learning_rate": 0.00018953077287044766,
      "loss": 7.5984,
      "step": 126550
    },
    {
      "epoch": 15.229843561973526,
      "grad_norm": 11463859.0,
      "learning_rate": 0.00018952907820063387,
      "loss": 7.5688,
      "step": 126560
    },
    {
      "epoch": 15.231046931407942,
      "grad_norm": 198660160.0,
      "learning_rate": 0.00018952738340124932,
      "loss": 7.6082,
      "step": 126570
    },
    {
      "epoch": 15.232250300842358,
      "grad_norm": 138143184.0,
      "learning_rate": 0.00018952568847229663,
      "loss": 7.6043,
      "step": 126580
    },
    {
      "epoch": 15.233453670276775,
      "grad_norm": 40182768.0,
      "learning_rate": 0.0001895239934137781,
      "loss": 7.6157,
      "step": 126590
    },
    {
      "epoch": 15.234657039711191,
      "grad_norm": 7230538.0,
      "learning_rate": 0.0001895222982256963,
      "loss": 7.6768,
      "step": 126600
    },
    {
      "epoch": 15.235860409145607,
      "grad_norm": 99642928.0,
      "learning_rate": 0.00018952060290805363,
      "loss": 7.7098,
      "step": 126610
    },
    {
      "epoch": 15.237063778580024,
      "grad_norm": 144365696.0,
      "learning_rate": 0.0001895189074608525,
      "loss": 7.6389,
      "step": 126620
    },
    {
      "epoch": 15.23826714801444,
      "grad_norm": 46300184.0,
      "learning_rate": 0.00018951721188409547,
      "loss": 7.5654,
      "step": 126630
    },
    {
      "epoch": 15.239470517448856,
      "grad_norm": 36155432.0,
      "learning_rate": 0.00018951551617778489,
      "loss": 7.539,
      "step": 126640
    },
    {
      "epoch": 15.240673886883274,
      "grad_norm": 118549752.0,
      "learning_rate": 0.00018951382034192328,
      "loss": 7.6694,
      "step": 126650
    },
    {
      "epoch": 15.24187725631769,
      "grad_norm": 202069824.0,
      "learning_rate": 0.00018951212437651305,
      "loss": 7.6656,
      "step": 126660
    },
    {
      "epoch": 15.243080625752105,
      "grad_norm": 165244304.0,
      "learning_rate": 0.0001895104282815567,
      "loss": 7.6104,
      "step": 126670
    },
    {
      "epoch": 15.244283995186523,
      "grad_norm": 61142300.0,
      "learning_rate": 0.00018950873205705665,
      "loss": 7.6946,
      "step": 126680
    },
    {
      "epoch": 15.245487364620939,
      "grad_norm": 40359160.0,
      "learning_rate": 0.00018950703570301536,
      "loss": 7.5679,
      "step": 126690
    },
    {
      "epoch": 15.246690734055354,
      "grad_norm": 32732284.0,
      "learning_rate": 0.0001895053392194353,
      "loss": 7.6494,
      "step": 126700
    },
    {
      "epoch": 15.247894103489772,
      "grad_norm": 8230265.5,
      "learning_rate": 0.0001895036426063189,
      "loss": 7.6714,
      "step": 126710
    },
    {
      "epoch": 15.249097472924188,
      "grad_norm": 193257040.0,
      "learning_rate": 0.00018950194586366865,
      "loss": 7.6969,
      "step": 126720
    },
    {
      "epoch": 15.250300842358604,
      "grad_norm": 29375584.0,
      "learning_rate": 0.00018950024899148696,
      "loss": 7.6407,
      "step": 126730
    },
    {
      "epoch": 15.251504211793021,
      "grad_norm": 83153360.0,
      "learning_rate": 0.00018949855198977636,
      "loss": 7.6445,
      "step": 126740
    },
    {
      "epoch": 15.252707581227437,
      "grad_norm": 211772352.0,
      "learning_rate": 0.00018949685485853922,
      "loss": 7.5814,
      "step": 126750
    },
    {
      "epoch": 15.253910950661853,
      "grad_norm": 50353728.0,
      "learning_rate": 0.00018949515759777806,
      "loss": 7.66,
      "step": 126760
    },
    {
      "epoch": 15.25511432009627,
      "grad_norm": 14870720.0,
      "learning_rate": 0.00018949346020749526,
      "loss": 7.5546,
      "step": 126770
    },
    {
      "epoch": 15.256317689530686,
      "grad_norm": 62450724.0,
      "learning_rate": 0.0001894917626876934,
      "loss": 7.5365,
      "step": 126780
    },
    {
      "epoch": 15.257521058965102,
      "grad_norm": 76659632.0,
      "learning_rate": 0.0001894900650383748,
      "loss": 7.5909,
      "step": 126790
    },
    {
      "epoch": 15.25872442839952,
      "grad_norm": 125586104.0,
      "learning_rate": 0.00018948836725954203,
      "loss": 7.6316,
      "step": 126800
    },
    {
      "epoch": 15.259927797833935,
      "grad_norm": 18692324.0,
      "learning_rate": 0.00018948666935119748,
      "loss": 7.5532,
      "step": 126810
    },
    {
      "epoch": 15.261131167268351,
      "grad_norm": 36259796.0,
      "learning_rate": 0.00018948497131334363,
      "loss": 7.5931,
      "step": 126820
    },
    {
      "epoch": 15.262334536702769,
      "grad_norm": 29815936.0,
      "learning_rate": 0.00018948327314598294,
      "loss": 7.5578,
      "step": 126830
    },
    {
      "epoch": 15.263537906137184,
      "grad_norm": 27945626.0,
      "learning_rate": 0.00018948157484911785,
      "loss": 7.5871,
      "step": 126840
    },
    {
      "epoch": 15.2647412755716,
      "grad_norm": 49709876.0,
      "learning_rate": 0.00018947987642275083,
      "loss": 7.6183,
      "step": 126850
    },
    {
      "epoch": 15.265944645006018,
      "grad_norm": 22007600.0,
      "learning_rate": 0.00018947817786688434,
      "loss": 7.5946,
      "step": 126860
    },
    {
      "epoch": 15.267148014440433,
      "grad_norm": 122649344.0,
      "learning_rate": 0.0001894764791815208,
      "loss": 7.6612,
      "step": 126870
    },
    {
      "epoch": 15.26835138387485,
      "grad_norm": 23468648.0,
      "learning_rate": 0.00018947478036666275,
      "loss": 7.6735,
      "step": 126880
    },
    {
      "epoch": 15.269554753309265,
      "grad_norm": 130342416.0,
      "learning_rate": 0.00018947308142231258,
      "loss": 7.6228,
      "step": 126890
    },
    {
      "epoch": 15.270758122743683,
      "grad_norm": 50178788.0,
      "learning_rate": 0.0001894713823484728,
      "loss": 7.6629,
      "step": 126900
    },
    {
      "epoch": 15.271961492178098,
      "grad_norm": 18792854.0,
      "learning_rate": 0.0001894696831451458,
      "loss": 7.5434,
      "step": 126910
    },
    {
      "epoch": 15.273164861612514,
      "grad_norm": 136559232.0,
      "learning_rate": 0.0001894679838123341,
      "loss": 7.6019,
      "step": 126920
    },
    {
      "epoch": 15.274368231046932,
      "grad_norm": 75619864.0,
      "learning_rate": 0.00018946628435004016,
      "loss": 7.631,
      "step": 126930
    },
    {
      "epoch": 15.275571600481348,
      "grad_norm": 54553956.0,
      "learning_rate": 0.00018946458475826639,
      "loss": 7.6471,
      "step": 126940
    },
    {
      "epoch": 15.276774969915763,
      "grad_norm": 143264192.0,
      "learning_rate": 0.00018946288503701527,
      "loss": 7.6097,
      "step": 126950
    },
    {
      "epoch": 15.277978339350181,
      "grad_norm": 17656644.0,
      "learning_rate": 0.00018946118518628927,
      "loss": 7.6484,
      "step": 126960
    },
    {
      "epoch": 15.279181708784597,
      "grad_norm": 39571772.0,
      "learning_rate": 0.00018945948520609086,
      "loss": 7.5435,
      "step": 126970
    },
    {
      "epoch": 15.280385078219012,
      "grad_norm": 18116824.0,
      "learning_rate": 0.0001894577850964225,
      "loss": 7.6014,
      "step": 126980
    },
    {
      "epoch": 15.28158844765343,
      "grad_norm": 35965420.0,
      "learning_rate": 0.0001894560848572866,
      "loss": 7.6067,
      "step": 126990
    },
    {
      "epoch": 15.282791817087846,
      "grad_norm": 7533833.0,
      "learning_rate": 0.0001894543844886857,
      "loss": 7.6144,
      "step": 127000
    },
    {
      "epoch": 15.283995186522262,
      "grad_norm": 13021408.0,
      "learning_rate": 0.0001894526839906222,
      "loss": 7.5807,
      "step": 127010
    },
    {
      "epoch": 15.28519855595668,
      "grad_norm": 6979057.5,
      "learning_rate": 0.00018945098336309858,
      "loss": 7.5642,
      "step": 127020
    },
    {
      "epoch": 15.286401925391095,
      "grad_norm": 27475448.0,
      "learning_rate": 0.0001894492826061173,
      "loss": 7.6356,
      "step": 127030
    },
    {
      "epoch": 15.28760529482551,
      "grad_norm": 32513116.0,
      "learning_rate": 0.00018944758171968084,
      "loss": 7.6351,
      "step": 127040
    },
    {
      "epoch": 15.288808664259928,
      "grad_norm": 27836212.0,
      "learning_rate": 0.00018944588070379163,
      "loss": 7.6687,
      "step": 127050
    },
    {
      "epoch": 15.290012033694344,
      "grad_norm": 23459752.0,
      "learning_rate": 0.00018944417955845216,
      "loss": 7.6519,
      "step": 127060
    },
    {
      "epoch": 15.29121540312876,
      "grad_norm": 32514826.0,
      "learning_rate": 0.00018944247828366488,
      "loss": 7.6131,
      "step": 127070
    },
    {
      "epoch": 15.292418772563177,
      "grad_norm": 13687643.0,
      "learning_rate": 0.00018944077687943223,
      "loss": 7.5811,
      "step": 127080
    },
    {
      "epoch": 15.293622141997593,
      "grad_norm": 12742487.0,
      "learning_rate": 0.00018943907534575672,
      "loss": 7.6129,
      "step": 127090
    },
    {
      "epoch": 15.294825511432009,
      "grad_norm": 56113888.0,
      "learning_rate": 0.00018943737368264078,
      "loss": 7.6463,
      "step": 127100
    },
    {
      "epoch": 15.296028880866427,
      "grad_norm": 67657248.0,
      "learning_rate": 0.00018943567189008684,
      "loss": 7.6239,
      "step": 127110
    },
    {
      "epoch": 15.297232250300842,
      "grad_norm": 20921442.0,
      "learning_rate": 0.00018943396996809746,
      "loss": 7.6075,
      "step": 127120
    },
    {
      "epoch": 15.298435619735258,
      "grad_norm": 7145852.5,
      "learning_rate": 0.00018943226791667502,
      "loss": 7.6048,
      "step": 127130
    },
    {
      "epoch": 15.299638989169676,
      "grad_norm": 9042083.0,
      "learning_rate": 0.00018943056573582202,
      "loss": 7.5822,
      "step": 127140
    },
    {
      "epoch": 15.300842358604092,
      "grad_norm": 42575684.0,
      "learning_rate": 0.00018942886342554087,
      "loss": 7.642,
      "step": 127150
    },
    {
      "epoch": 15.302045728038507,
      "grad_norm": 27993732.0,
      "learning_rate": 0.0001894271609858341,
      "loss": 7.5568,
      "step": 127160
    },
    {
      "epoch": 15.303249097472925,
      "grad_norm": 23491200.0,
      "learning_rate": 0.00018942545841670416,
      "loss": 7.5542,
      "step": 127170
    },
    {
      "epoch": 15.30445246690734,
      "grad_norm": 44398652.0,
      "learning_rate": 0.0001894237557181535,
      "loss": 7.6108,
      "step": 127180
    },
    {
      "epoch": 15.305655836341757,
      "grad_norm": 14377230.0,
      "learning_rate": 0.00018942205289018457,
      "loss": 7.5937,
      "step": 127190
    },
    {
      "epoch": 15.306859205776174,
      "grad_norm": 45386232.0,
      "learning_rate": 0.00018942034993279987,
      "loss": 7.6048,
      "step": 127200
    },
    {
      "epoch": 15.30806257521059,
      "grad_norm": 13268388.0,
      "learning_rate": 0.00018941864684600185,
      "loss": 7.5048,
      "step": 127210
    },
    {
      "epoch": 15.309265944645006,
      "grad_norm": 22896958.0,
      "learning_rate": 0.00018941694362979296,
      "loss": 7.6437,
      "step": 127220
    },
    {
      "epoch": 15.310469314079423,
      "grad_norm": 12896894.0,
      "learning_rate": 0.00018941524028417568,
      "loss": 7.6753,
      "step": 127230
    },
    {
      "epoch": 15.311672683513839,
      "grad_norm": 33137488.0,
      "learning_rate": 0.00018941353680915248,
      "loss": 7.7099,
      "step": 127240
    },
    {
      "epoch": 15.312876052948255,
      "grad_norm": 28891676.0,
      "learning_rate": 0.0001894118332047258,
      "loss": 7.6516,
      "step": 127250
    },
    {
      "epoch": 15.314079422382672,
      "grad_norm": 46271324.0,
      "learning_rate": 0.00018941012947089813,
      "loss": 7.583,
      "step": 127260
    },
    {
      "epoch": 15.315282791817088,
      "grad_norm": 62318632.0,
      "learning_rate": 0.00018940842560767195,
      "loss": 7.6459,
      "step": 127270
    },
    {
      "epoch": 15.316486161251504,
      "grad_norm": 33538558.0,
      "learning_rate": 0.00018940672161504965,
      "loss": 7.6399,
      "step": 127280
    },
    {
      "epoch": 15.31768953068592,
      "grad_norm": 22386000.0,
      "learning_rate": 0.0001894050174930338,
      "loss": 7.5709,
      "step": 127290
    },
    {
      "epoch": 15.318892900120337,
      "grad_norm": 19844442.0,
      "learning_rate": 0.0001894033132416268,
      "loss": 7.6348,
      "step": 127300
    },
    {
      "epoch": 15.320096269554753,
      "grad_norm": 78055824.0,
      "learning_rate": 0.00018940160886083112,
      "loss": 7.6563,
      "step": 127310
    },
    {
      "epoch": 15.321299638989169,
      "grad_norm": 32610438.0,
      "learning_rate": 0.00018939990435064928,
      "loss": 7.7332,
      "step": 127320
    },
    {
      "epoch": 15.322503008423586,
      "grad_norm": 22330076.0,
      "learning_rate": 0.00018939819971108363,
      "loss": 7.6675,
      "step": 127330
    },
    {
      "epoch": 15.323706377858002,
      "grad_norm": 75460280.0,
      "learning_rate": 0.00018939649494213677,
      "loss": 7.7185,
      "step": 127340
    },
    {
      "epoch": 15.324909747292418,
      "grad_norm": 75515872.0,
      "learning_rate": 0.00018939479004381112,
      "loss": 7.6141,
      "step": 127350
    },
    {
      "epoch": 15.326113116726836,
      "grad_norm": 178747952.0,
      "learning_rate": 0.0001893930850161091,
      "loss": 7.6413,
      "step": 127360
    },
    {
      "epoch": 15.327316486161251,
      "grad_norm": 57875368.0,
      "learning_rate": 0.00018939137985903327,
      "loss": 7.6971,
      "step": 127370
    },
    {
      "epoch": 15.328519855595667,
      "grad_norm": 1167977.25,
      "learning_rate": 0.00018938967457258598,
      "loss": 7.6191,
      "step": 127380
    },
    {
      "epoch": 15.329723225030085,
      "grad_norm": 51320640.0,
      "learning_rate": 0.0001893879691567698,
      "loss": 7.6472,
      "step": 127390
    },
    {
      "epoch": 15.3309265944645,
      "grad_norm": 106197.671875,
      "learning_rate": 0.00018938626361158714,
      "loss": 7.6916,
      "step": 127400
    },
    {
      "epoch": 15.332129963898916,
      "grad_norm": 1414810435584.0,
      "learning_rate": 0.0001893845579370405,
      "loss": 7.5789,
      "step": 127410
    },
    {
      "epoch": 15.333333333333334,
      "grad_norm": 2504443035648.0,
      "learning_rate": 0.00018938285213313232,
      "loss": 7.8044,
      "step": 127420
    },
    {
      "epoch": 15.33453670276775,
      "grad_norm": 3258781532160.0,
      "learning_rate": 0.0001893811461998651,
      "loss": 7.7567,
      "step": 127430
    },
    {
      "epoch": 15.335740072202166,
      "grad_norm": 1773208600576.0,
      "learning_rate": 0.0001893794401372413,
      "loss": 7.9074,
      "step": 127440
    },
    {
      "epoch": 15.336943441636583,
      "grad_norm": 3807304744960.0,
      "learning_rate": 0.00018937773394526336,
      "loss": 7.8283,
      "step": 127450
    },
    {
      "epoch": 15.338146811070999,
      "grad_norm": 1970687705088.0,
      "learning_rate": 0.00018937602762393382,
      "loss": 7.7362,
      "step": 127460
    },
    {
      "epoch": 15.339350180505415,
      "grad_norm": 2991477489664.0,
      "learning_rate": 0.00018937432117325503,
      "loss": 7.8072,
      "step": 127470
    },
    {
      "epoch": 15.340553549939832,
      "grad_norm": 8188653469696.0,
      "learning_rate": 0.0001893726145932296,
      "loss": 7.8177,
      "step": 127480
    },
    {
      "epoch": 15.341756919374248,
      "grad_norm": 3692498518016.0,
      "learning_rate": 0.0001893709078838599,
      "loss": 7.7759,
      "step": 127490
    },
    {
      "epoch": 15.342960288808664,
      "grad_norm": 813133398016.0,
      "learning_rate": 0.00018936920104514843,
      "loss": 7.8943,
      "step": 127500
    },
    {
      "epoch": 15.344163658243081,
      "grad_norm": 2547295977472.0,
      "learning_rate": 0.00018936749407709767,
      "loss": 7.8693,
      "step": 127510
    },
    {
      "epoch": 15.345367027677497,
      "grad_norm": 1573141.875,
      "learning_rate": 0.0001893657869797101,
      "loss": 7.908,
      "step": 127520
    },
    {
      "epoch": 15.346570397111913,
      "grad_norm": 9042798.0,
      "learning_rate": 0.00018936407975298812,
      "loss": 7.9104,
      "step": 127530
    },
    {
      "epoch": 15.34777376654633,
      "grad_norm": 19842218.0,
      "learning_rate": 0.00018936237239693427,
      "loss": 7.9035,
      "step": 127540
    },
    {
      "epoch": 15.348977135980746,
      "grad_norm": 19952302.0,
      "learning_rate": 0.00018936066491155102,
      "loss": 7.9823,
      "step": 127550
    },
    {
      "epoch": 15.350180505415162,
      "grad_norm": 7535808.0,
      "learning_rate": 0.00018935895729684082,
      "loss": 7.9795,
      "step": 127560
    },
    {
      "epoch": 15.35138387484958,
      "grad_norm": 2834054.75,
      "learning_rate": 0.00018935724955280615,
      "loss": 7.8004,
      "step": 127570
    },
    {
      "epoch": 15.352587244283995,
      "grad_norm": 7588425.0,
      "learning_rate": 0.00018935554167944947,
      "loss": 7.9261,
      "step": 127580
    },
    {
      "epoch": 15.353790613718411,
      "grad_norm": 10952721.0,
      "learning_rate": 0.00018935383367677327,
      "loss": 7.9108,
      "step": 127590
    },
    {
      "epoch": 15.354993983152827,
      "grad_norm": 2211648.25,
      "learning_rate": 0.00018935212554478,
      "loss": 7.9316,
      "step": 127600
    },
    {
      "epoch": 15.356197352587245,
      "grad_norm": 1134086.625,
      "learning_rate": 0.00018935041728347218,
      "loss": 7.8866,
      "step": 127610
    },
    {
      "epoch": 15.35740072202166,
      "grad_norm": 6018494.5,
      "learning_rate": 0.0001893487088928522,
      "loss": 7.8616,
      "step": 127620
    },
    {
      "epoch": 15.358604091456076,
      "grad_norm": 6723438.5,
      "learning_rate": 0.00018934700037292261,
      "loss": 7.8405,
      "step": 127630
    },
    {
      "epoch": 15.359807460890494,
      "grad_norm": 5646049.5,
      "learning_rate": 0.00018934529172368584,
      "loss": 7.8521,
      "step": 127640
    },
    {
      "epoch": 15.36101083032491,
      "grad_norm": 1505657.25,
      "learning_rate": 0.00018934358294514438,
      "loss": 7.8557,
      "step": 127650
    },
    {
      "epoch": 15.362214199759325,
      "grad_norm": 8570042.0,
      "learning_rate": 0.0001893418740373007,
      "loss": 7.7837,
      "step": 127660
    },
    {
      "epoch": 15.363417569193743,
      "grad_norm": 926838.625,
      "learning_rate": 0.00018934016500015725,
      "loss": 7.8622,
      "step": 127670
    },
    {
      "epoch": 15.364620938628159,
      "grad_norm": 4178495.0,
      "learning_rate": 0.00018933845583371653,
      "loss": 7.8453,
      "step": 127680
    },
    {
      "epoch": 15.365824308062574,
      "grad_norm": 5984799.0,
      "learning_rate": 0.000189336746537981,
      "loss": 7.868,
      "step": 127690
    },
    {
      "epoch": 15.367027677496992,
      "grad_norm": 13215079.0,
      "learning_rate": 0.00018933503711295315,
      "loss": 7.9283,
      "step": 127700
    },
    {
      "epoch": 15.368231046931408,
      "grad_norm": 8529610.0,
      "learning_rate": 0.00018933332755863546,
      "loss": 7.8827,
      "step": 127710
    },
    {
      "epoch": 15.369434416365824,
      "grad_norm": 11834520.0,
      "learning_rate": 0.00018933161787503038,
      "loss": 7.9105,
      "step": 127720
    },
    {
      "epoch": 15.370637785800241,
      "grad_norm": 20368342.0,
      "learning_rate": 0.0001893299080621404,
      "loss": 7.896,
      "step": 127730
    },
    {
      "epoch": 15.371841155234657,
      "grad_norm": 11510814.0,
      "learning_rate": 0.00018932819811996797,
      "loss": 7.9498,
      "step": 127740
    },
    {
      "epoch": 15.373044524669073,
      "grad_norm": 6029462.5,
      "learning_rate": 0.00018932648804851558,
      "loss": 7.9647,
      "step": 127750
    },
    {
      "epoch": 15.37424789410349,
      "grad_norm": 62824980.0,
      "learning_rate": 0.00018932477784778574,
      "loss": 7.8644,
      "step": 127760
    },
    {
      "epoch": 15.375451263537906,
      "grad_norm": 6200466.0,
      "learning_rate": 0.0001893230675177809,
      "loss": 7.9306,
      "step": 127770
    },
    {
      "epoch": 15.376654632972322,
      "grad_norm": 31674432.0,
      "learning_rate": 0.00018932135705850348,
      "loss": 7.8448,
      "step": 127780
    },
    {
      "epoch": 15.37785800240674,
      "grad_norm": 17501450.0,
      "learning_rate": 0.000189319646469956,
      "loss": 7.9711,
      "step": 127790
    },
    {
      "epoch": 15.379061371841155,
      "grad_norm": 6270009.5,
      "learning_rate": 0.000189317935752141,
      "loss": 7.9291,
      "step": 127800
    },
    {
      "epoch": 15.380264741275571,
      "grad_norm": 15273657.0,
      "learning_rate": 0.00018931622490506086,
      "loss": 7.9161,
      "step": 127810
    },
    {
      "epoch": 15.381468110709989,
      "grad_norm": 9113166.0,
      "learning_rate": 0.00018931451392871807,
      "loss": 7.9603,
      "step": 127820
    },
    {
      "epoch": 15.382671480144404,
      "grad_norm": 11408706.0,
      "learning_rate": 0.00018931280282311516,
      "loss": 7.9122,
      "step": 127830
    },
    {
      "epoch": 15.38387484957882,
      "grad_norm": 32239620.0,
      "learning_rate": 0.00018931109158825456,
      "loss": 7.9983,
      "step": 127840
    },
    {
      "epoch": 15.385078219013238,
      "grad_norm": 7070946.0,
      "learning_rate": 0.00018930938022413877,
      "loss": 7.8207,
      "step": 127850
    },
    {
      "epoch": 15.386281588447654,
      "grad_norm": 29044062.0,
      "learning_rate": 0.00018930766873077024,
      "loss": 8.0038,
      "step": 127860
    },
    {
      "epoch": 15.38748495788207,
      "grad_norm": 18711484.0,
      "learning_rate": 0.00018930595710815146,
      "loss": 7.9539,
      "step": 127870
    },
    {
      "epoch": 15.388688327316487,
      "grad_norm": 24030608.0,
      "learning_rate": 0.00018930424535628496,
      "loss": 7.9431,
      "step": 127880
    },
    {
      "epoch": 15.389891696750903,
      "grad_norm": 20224566.0,
      "learning_rate": 0.0001893025334751731,
      "loss": 8.0645,
      "step": 127890
    },
    {
      "epoch": 15.391095066185319,
      "grad_norm": 18130146.0,
      "learning_rate": 0.00018930082146481847,
      "loss": 7.9791,
      "step": 127900
    },
    {
      "epoch": 15.392298435619736,
      "grad_norm": 16162494.0,
      "learning_rate": 0.00018929910932522347,
      "loss": 7.9798,
      "step": 127910
    },
    {
      "epoch": 15.393501805054152,
      "grad_norm": 11518512.0,
      "learning_rate": 0.00018929739705639067,
      "loss": 7.9174,
      "step": 127920
    },
    {
      "epoch": 15.394705174488568,
      "grad_norm": 10792011.0,
      "learning_rate": 0.00018929568465832244,
      "loss": 7.9288,
      "step": 127930
    },
    {
      "epoch": 15.395908543922985,
      "grad_norm": 4750258.0,
      "learning_rate": 0.00018929397213102132,
      "loss": 8.0279,
      "step": 127940
    },
    {
      "epoch": 15.397111913357401,
      "grad_norm": 10598107.0,
      "learning_rate": 0.00018929225947448978,
      "loss": 7.8804,
      "step": 127950
    },
    {
      "epoch": 15.398315282791817,
      "grad_norm": 21077748.0,
      "learning_rate": 0.0001892905466887303,
      "loss": 7.974,
      "step": 127960
    },
    {
      "epoch": 15.399518652226233,
      "grad_norm": 6635778.5,
      "learning_rate": 0.00018928883377374533,
      "loss": 7.803,
      "step": 127970
    },
    {
      "epoch": 15.40072202166065,
      "grad_norm": 16265248.0,
      "learning_rate": 0.0001892871207295374,
      "loss": 7.9542,
      "step": 127980
    },
    {
      "epoch": 15.401925391095066,
      "grad_norm": 11509844.0,
      "learning_rate": 0.00018928540755610895,
      "loss": 7.8735,
      "step": 127990
    },
    {
      "epoch": 15.403128760529482,
      "grad_norm": 2122466.75,
      "learning_rate": 0.00018928369425346247,
      "loss": 7.8465,
      "step": 128000
    },
    {
      "epoch": 15.4043321299639,
      "grad_norm": 7207854.0,
      "learning_rate": 0.00018928198082160044,
      "loss": 7.9605,
      "step": 128010
    },
    {
      "epoch": 15.405535499398315,
      "grad_norm": 28157966.0,
      "learning_rate": 0.00018928026726052535,
      "loss": 8.047,
      "step": 128020
    },
    {
      "epoch": 15.406738868832731,
      "grad_norm": 45350984.0,
      "learning_rate": 0.00018927855357023967,
      "loss": 7.9673,
      "step": 128030
    },
    {
      "epoch": 15.407942238267148,
      "grad_norm": 70852912.0,
      "learning_rate": 0.00018927683975074588,
      "loss": 7.9313,
      "step": 128040
    },
    {
      "epoch": 15.409145607701564,
      "grad_norm": 9853865.0,
      "learning_rate": 0.00018927512580204644,
      "loss": 7.9559,
      "step": 128050
    },
    {
      "epoch": 15.41034897713598,
      "grad_norm": 17501476.0,
      "learning_rate": 0.00018927341172414388,
      "loss": 7.9305,
      "step": 128060
    },
    {
      "epoch": 15.411552346570398,
      "grad_norm": 23052134.0,
      "learning_rate": 0.00018927169751704062,
      "loss": 7.9331,
      "step": 128070
    },
    {
      "epoch": 15.412755716004813,
      "grad_norm": 56401612.0,
      "learning_rate": 0.00018926998318073924,
      "loss": 7.9313,
      "step": 128080
    },
    {
      "epoch": 15.41395908543923,
      "grad_norm": 30189430.0,
      "learning_rate": 0.0001892682687152421,
      "loss": 8.0057,
      "step": 128090
    },
    {
      "epoch": 15.415162454873647,
      "grad_norm": 84842288.0,
      "learning_rate": 0.00018926655412055174,
      "loss": 7.9918,
      "step": 128100
    },
    {
      "epoch": 15.416365824308063,
      "grad_norm": 58338020.0,
      "learning_rate": 0.00018926483939667063,
      "loss": 8.0324,
      "step": 128110
    },
    {
      "epoch": 15.417569193742478,
      "grad_norm": 33805.7890625,
      "learning_rate": 0.00018926312454360127,
      "loss": 8.0076,
      "step": 128120
    },
    {
      "epoch": 15.418772563176896,
      "grad_norm": 71840.484375,
      "learning_rate": 0.00018926140956134613,
      "loss": 8.0435,
      "step": 128130
    },
    {
      "epoch": 15.419975932611312,
      "grad_norm": 43471.79296875,
      "learning_rate": 0.0001892596944499077,
      "loss": 7.7596,
      "step": 128140
    },
    {
      "epoch": 15.421179302045728,
      "grad_norm": 51212.53125,
      "learning_rate": 0.00018925797920928844,
      "loss": 7.7364,
      "step": 128150
    },
    {
      "epoch": 15.422382671480145,
      "grad_norm": 39187.37890625,
      "learning_rate": 0.00018925626383949084,
      "loss": 7.728,
      "step": 128160
    },
    {
      "epoch": 15.42358604091456,
      "grad_norm": 14409.4521484375,
      "learning_rate": 0.00018925454834051744,
      "loss": 7.7351,
      "step": 128170
    },
    {
      "epoch": 15.424789410348977,
      "grad_norm": 116253.59375,
      "learning_rate": 0.00018925283271237063,
      "loss": 7.6738,
      "step": 128180
    },
    {
      "epoch": 15.425992779783394,
      "grad_norm": 95810.2578125,
      "learning_rate": 0.00018925111695505294,
      "loss": 7.7214,
      "step": 128190
    },
    {
      "epoch": 15.42719614921781,
      "grad_norm": 106774.734375,
      "learning_rate": 0.00018924940106856684,
      "loss": 7.6666,
      "step": 128200
    },
    {
      "epoch": 15.428399518652226,
      "grad_norm": 44819.26171875,
      "learning_rate": 0.00018924768505291484,
      "loss": 7.686,
      "step": 128210
    },
    {
      "epoch": 15.429602888086643,
      "grad_norm": 32586.5625,
      "learning_rate": 0.0001892459689080994,
      "loss": 7.5878,
      "step": 128220
    },
    {
      "epoch": 15.43080625752106,
      "grad_norm": 62129.98828125,
      "learning_rate": 0.000189244252634123,
      "loss": 7.6324,
      "step": 128230
    },
    {
      "epoch": 15.432009626955475,
      "grad_norm": 40304.99609375,
      "learning_rate": 0.00018924253623098814,
      "loss": 7.6888,
      "step": 128240
    },
    {
      "epoch": 15.433212996389893,
      "grad_norm": 52973.91796875,
      "learning_rate": 0.0001892408196986973,
      "loss": 7.6634,
      "step": 128250
    },
    {
      "epoch": 15.434416365824308,
      "grad_norm": 69154.9140625,
      "learning_rate": 0.00018923910303725295,
      "loss": 7.6284,
      "step": 128260
    },
    {
      "epoch": 15.435619735258724,
      "grad_norm": 37306.50390625,
      "learning_rate": 0.0001892373862466576,
      "loss": 7.6516,
      "step": 128270
    },
    {
      "epoch": 15.43682310469314,
      "grad_norm": 129092.2578125,
      "learning_rate": 0.0001892356693269137,
      "loss": 7.5421,
      "step": 128280
    },
    {
      "epoch": 15.438026474127557,
      "grad_norm": 6688697.5,
      "learning_rate": 0.00018923395227802377,
      "loss": 7.6009,
      "step": 128290
    },
    {
      "epoch": 15.439229843561973,
      "grad_norm": 20396596.0,
      "learning_rate": 0.00018923223509999026,
      "loss": 7.6637,
      "step": 128300
    },
    {
      "epoch": 15.440433212996389,
      "grad_norm": 30236666.0,
      "learning_rate": 0.0001892305177928157,
      "loss": 7.5875,
      "step": 128310
    },
    {
      "epoch": 15.441636582430807,
      "grad_norm": 7551369.0,
      "learning_rate": 0.00018922880035650253,
      "loss": 7.5586,
      "step": 128320
    },
    {
      "epoch": 15.442839951865222,
      "grad_norm": 57299056.0,
      "learning_rate": 0.0001892270827910533,
      "loss": 7.6573,
      "step": 128330
    },
    {
      "epoch": 15.444043321299638,
      "grad_norm": 14870585.0,
      "learning_rate": 0.00018922536509647042,
      "loss": 7.565,
      "step": 128340
    },
    {
      "epoch": 15.445246690734056,
      "grad_norm": 33727032.0,
      "learning_rate": 0.0001892236472727564,
      "loss": 7.7003,
      "step": 128350
    },
    {
      "epoch": 15.446450060168472,
      "grad_norm": 6558484.5,
      "learning_rate": 0.00018922192931991374,
      "loss": 7.6156,
      "step": 128360
    },
    {
      "epoch": 15.447653429602887,
      "grad_norm": 49388.5,
      "learning_rate": 0.00018922021123794492,
      "loss": 7.589,
      "step": 128370
    },
    {
      "epoch": 15.448856799037305,
      "grad_norm": 70256.1796875,
      "learning_rate": 0.00018921849302685245,
      "loss": 7.5735,
      "step": 128380
    },
    {
      "epoch": 15.45006016847172,
      "grad_norm": 40877.8828125,
      "learning_rate": 0.00018921677468663876,
      "loss": 7.6634,
      "step": 128390
    },
    {
      "epoch": 15.451263537906136,
      "grad_norm": 109990.078125,
      "learning_rate": 0.00018921505621730637,
      "loss": 7.608,
      "step": 128400
    },
    {
      "epoch": 15.452466907340554,
      "grad_norm": 63311.01171875,
      "learning_rate": 0.0001892133376188578,
      "loss": 7.6643,
      "step": 128410
    },
    {
      "epoch": 15.45367027677497,
      "grad_norm": 17717.41015625,
      "learning_rate": 0.00018921161889129547,
      "loss": 7.5901,
      "step": 128420
    },
    {
      "epoch": 15.454873646209386,
      "grad_norm": 65983.3984375,
      "learning_rate": 0.00018920990003462192,
      "loss": 7.6224,
      "step": 128430
    },
    {
      "epoch": 15.456077015643803,
      "grad_norm": 37821.62890625,
      "learning_rate": 0.0001892081810488396,
      "loss": 7.5199,
      "step": 128440
    },
    {
      "epoch": 15.457280385078219,
      "grad_norm": 65781.1484375,
      "learning_rate": 0.00018920646193395103,
      "loss": 7.5692,
      "step": 128450
    },
    {
      "epoch": 15.458483754512635,
      "grad_norm": 75545.5078125,
      "learning_rate": 0.00018920474268995869,
      "loss": 7.5532,
      "step": 128460
    },
    {
      "epoch": 15.459687123947052,
      "grad_norm": 20159.916015625,
      "learning_rate": 0.00018920302331686508,
      "loss": 7.5708,
      "step": 128470
    },
    {
      "epoch": 15.460890493381468,
      "grad_norm": 65203.3515625,
      "learning_rate": 0.00018920130381467262,
      "loss": 7.5549,
      "step": 128480
    },
    {
      "epoch": 15.462093862815884,
      "grad_norm": 24130.107421875,
      "learning_rate": 0.0001891995841833839,
      "loss": 7.5831,
      "step": 128490
    },
    {
      "epoch": 15.463297232250302,
      "grad_norm": 13654795.0,
      "learning_rate": 0.00018919786442300133,
      "loss": 7.5671,
      "step": 128500
    },
    {
      "epoch": 15.464500601684717,
      "grad_norm": 21652528.0,
      "learning_rate": 0.00018919614453352746,
      "loss": 7.6134,
      "step": 128510
    },
    {
      "epoch": 15.465703971119133,
      "grad_norm": 7533261.5,
      "learning_rate": 0.00018919442451496472,
      "loss": 7.5694,
      "step": 128520
    },
    {
      "epoch": 15.46690734055355,
      "grad_norm": 18992542.0,
      "learning_rate": 0.00018919270436731561,
      "loss": 7.6579,
      "step": 128530
    },
    {
      "epoch": 15.468110709987966,
      "grad_norm": 6445664.5,
      "learning_rate": 0.0001891909840905827,
      "loss": 7.5958,
      "step": 128540
    },
    {
      "epoch": 15.469314079422382,
      "grad_norm": 41615272.0,
      "learning_rate": 0.00018918926368476834,
      "loss": 7.6084,
      "step": 128550
    },
    {
      "epoch": 15.4705174488568,
      "grad_norm": 8817272.0,
      "learning_rate": 0.00018918754314987515,
      "loss": 7.5406,
      "step": 128560
    },
    {
      "epoch": 15.471720818291216,
      "grad_norm": 4685589.5,
      "learning_rate": 0.00018918582248590555,
      "loss": 7.5583,
      "step": 128570
    },
    {
      "epoch": 15.472924187725631,
      "grad_norm": 1491868.625,
      "learning_rate": 0.00018918410169286201,
      "loss": 7.6372,
      "step": 128580
    },
    {
      "epoch": 15.474127557160049,
      "grad_norm": 12772666.0,
      "learning_rate": 0.0001891823807707471,
      "loss": 7.6059,
      "step": 128590
    },
    {
      "epoch": 15.475330926594465,
      "grad_norm": 4786470.5,
      "learning_rate": 0.00018918065971956324,
      "loss": 7.6642,
      "step": 128600
    },
    {
      "epoch": 15.47653429602888,
      "grad_norm": 2048514.625,
      "learning_rate": 0.00018917893853931297,
      "loss": 7.6305,
      "step": 128610
    },
    {
      "epoch": 15.477737665463298,
      "grad_norm": 4387916.0,
      "learning_rate": 0.00018917721722999878,
      "loss": 7.6003,
      "step": 128620
    },
    {
      "epoch": 15.478941034897714,
      "grad_norm": 3760863.5,
      "learning_rate": 0.0001891754957916231,
      "loss": 7.6605,
      "step": 128630
    },
    {
      "epoch": 15.48014440433213,
      "grad_norm": 9387811.0,
      "learning_rate": 0.00018917377422418845,
      "loss": 7.6452,
      "step": 128640
    },
    {
      "epoch": 15.481347773766545,
      "grad_norm": 3026744.25,
      "learning_rate": 0.00018917205252769737,
      "loss": 7.6199,
      "step": 128650
    },
    {
      "epoch": 15.482551143200963,
      "grad_norm": 8218311.0,
      "learning_rate": 0.0001891703307021523,
      "loss": 7.6202,
      "step": 128660
    },
    {
      "epoch": 15.483754512635379,
      "grad_norm": 10987606.0,
      "learning_rate": 0.00018916860874755573,
      "loss": 7.6613,
      "step": 128670
    },
    {
      "epoch": 15.484957882069795,
      "grad_norm": 13410976.0,
      "learning_rate": 0.00018916688666391018,
      "loss": 7.544,
      "step": 128680
    },
    {
      "epoch": 15.486161251504212,
      "grad_norm": 4956526.5,
      "learning_rate": 0.00018916516445121811,
      "loss": 7.5892,
      "step": 128690
    },
    {
      "epoch": 15.487364620938628,
      "grad_norm": 11888376.0,
      "learning_rate": 0.00018916344210948206,
      "loss": 7.6252,
      "step": 128700
    },
    {
      "epoch": 15.488567990373044,
      "grad_norm": 10391656.0,
      "learning_rate": 0.0001891617196387045,
      "loss": 7.5488,
      "step": 128710
    },
    {
      "epoch": 15.489771359807461,
      "grad_norm": 11783991.0,
      "learning_rate": 0.0001891599970388879,
      "loss": 7.5685,
      "step": 128720
    },
    {
      "epoch": 15.490974729241877,
      "grad_norm": 30776322.0,
      "learning_rate": 0.0001891582743100348,
      "loss": 7.6083,
      "step": 128730
    },
    {
      "epoch": 15.492178098676293,
      "grad_norm": 8654180.0,
      "learning_rate": 0.00018915655145214765,
      "loss": 7.5361,
      "step": 128740
    },
    {
      "epoch": 15.49338146811071,
      "grad_norm": 4086228.5,
      "learning_rate": 0.00018915482846522893,
      "loss": 7.6417,
      "step": 128750
    },
    {
      "epoch": 15.494584837545126,
      "grad_norm": 13028856.0,
      "learning_rate": 0.00018915310534928117,
      "loss": 7.6083,
      "step": 128760
    },
    {
      "epoch": 15.495788206979542,
      "grad_norm": 7764.36767578125,
      "learning_rate": 0.0001891513821043069,
      "loss": 7.5509,
      "step": 128770
    },
    {
      "epoch": 15.49699157641396,
      "grad_norm": 23228.30859375,
      "learning_rate": 0.00018914965873030853,
      "loss": 7.5357,
      "step": 128780
    },
    {
      "epoch": 15.498194945848375,
      "grad_norm": 14338.8427734375,
      "learning_rate": 0.0001891479352272886,
      "loss": 7.5521,
      "step": 128790
    },
    {
      "epoch": 15.499398315282791,
      "grad_norm": 84327.7890625,
      "learning_rate": 0.00018914621159524958,
      "loss": 7.682,
      "step": 128800
    },
    {
      "epoch": 15.500601684717209,
      "grad_norm": 9639.7568359375,
      "learning_rate": 0.000189144487834194,
      "loss": 7.6499,
      "step": 128810
    },
    {
      "epoch": 15.501805054151625,
      "grad_norm": 7120.83203125,
      "learning_rate": 0.00018914276394412432,
      "loss": 7.574,
      "step": 128820
    },
    {
      "epoch": 15.50300842358604,
      "grad_norm": 16379.93359375,
      "learning_rate": 0.0001891410399250431,
      "loss": 7.6814,
      "step": 128830
    },
    {
      "epoch": 15.504211793020458,
      "grad_norm": 5513.794921875,
      "learning_rate": 0.00018913931577695277,
      "loss": 7.6113,
      "step": 128840
    },
    {
      "epoch": 15.505415162454874,
      "grad_norm": 24535.748046875,
      "learning_rate": 0.00018913759149985582,
      "loss": 7.4889,
      "step": 128850
    },
    {
      "epoch": 15.50661853188929,
      "grad_norm": 7957.34375,
      "learning_rate": 0.00018913586709375476,
      "loss": 7.5974,
      "step": 128860
    },
    {
      "epoch": 15.507821901323707,
      "grad_norm": 10026.1513671875,
      "learning_rate": 0.00018913414255865212,
      "loss": 7.6244,
      "step": 128870
    },
    {
      "epoch": 15.509025270758123,
      "grad_norm": 21741.625,
      "learning_rate": 0.00018913241789455035,
      "loss": 7.5943,
      "step": 128880
    },
    {
      "epoch": 15.510228640192539,
      "grad_norm": 21506.59375,
      "learning_rate": 0.000189130693101452,
      "loss": 7.4769,
      "step": 128890
    },
    {
      "epoch": 15.511432009626956,
      "grad_norm": 14295.814453125,
      "learning_rate": 0.0001891289681793595,
      "loss": 7.6081,
      "step": 128900
    },
    {
      "epoch": 15.512635379061372,
      "grad_norm": 5405.8916015625,
      "learning_rate": 0.00018912724312827539,
      "loss": 7.5695,
      "step": 128910
    },
    {
      "epoch": 15.513838748495788,
      "grad_norm": 10172.646484375,
      "learning_rate": 0.00018912551794820215,
      "loss": 7.5735,
      "step": 128920
    },
    {
      "epoch": 15.515042117930205,
      "grad_norm": 12082.6923828125,
      "learning_rate": 0.00018912379263914225,
      "loss": 7.4674,
      "step": 128930
    },
    {
      "epoch": 15.516245487364621,
      "grad_norm": 9141.341796875,
      "learning_rate": 0.00018912206720109827,
      "loss": 7.5755,
      "step": 128940
    },
    {
      "epoch": 15.517448856799037,
      "grad_norm": 19292.23828125,
      "learning_rate": 0.00018912034163407263,
      "loss": 7.5052,
      "step": 128950
    },
    {
      "epoch": 15.518652226233453,
      "grad_norm": 10430.6025390625,
      "learning_rate": 0.00018911861593806786,
      "loss": 7.5322,
      "step": 128960
    },
    {
      "epoch": 15.51985559566787,
      "grad_norm": 17314.849609375,
      "learning_rate": 0.00018911689011308642,
      "loss": 7.5471,
      "step": 128970
    },
    {
      "epoch": 15.521058965102286,
      "grad_norm": 6586.01025390625,
      "learning_rate": 0.00018911516415913088,
      "loss": 7.5709,
      "step": 128980
    },
    {
      "epoch": 15.522262334536702,
      "grad_norm": 6877.32568359375,
      "learning_rate": 0.00018911343807620368,
      "loss": 7.5245,
      "step": 128990
    },
    {
      "epoch": 15.52346570397112,
      "grad_norm": 4980.56982421875,
      "learning_rate": 0.00018911171186430733,
      "loss": 7.57,
      "step": 129000
    },
    {
      "epoch": 15.524669073405535,
      "grad_norm": 5770.44287109375,
      "learning_rate": 0.00018910998552344434,
      "loss": 7.5478,
      "step": 129010
    },
    {
      "epoch": 15.525872442839951,
      "grad_norm": 4519.296875,
      "learning_rate": 0.0001891082590536172,
      "loss": 7.4952,
      "step": 129020
    },
    {
      "epoch": 15.527075812274369,
      "grad_norm": 16958.93359375,
      "learning_rate": 0.00018910653245482839,
      "loss": 7.5302,
      "step": 129030
    },
    {
      "epoch": 15.528279181708784,
      "grad_norm": 25330.244140625,
      "learning_rate": 0.00018910480572708046,
      "loss": 7.5386,
      "step": 129040
    },
    {
      "epoch": 15.5294825511432,
      "grad_norm": 5226.640625,
      "learning_rate": 0.00018910307887037587,
      "loss": 7.5301,
      "step": 129050
    },
    {
      "epoch": 15.530685920577618,
      "grad_norm": 9514.9990234375,
      "learning_rate": 0.0001891013518847171,
      "loss": 7.4962,
      "step": 129060
    },
    {
      "epoch": 15.531889290012034,
      "grad_norm": 3317.2431640625,
      "learning_rate": 0.0001890996247701067,
      "loss": 7.5716,
      "step": 129070
    },
    {
      "epoch": 15.53309265944645,
      "grad_norm": 8102.52880859375,
      "learning_rate": 0.0001890978975265471,
      "loss": 7.5251,
      "step": 129080
    },
    {
      "epoch": 15.534296028880867,
      "grad_norm": 6283.49169921875,
      "learning_rate": 0.00018909617015404087,
      "loss": 7.5218,
      "step": 129090
    },
    {
      "epoch": 15.535499398315283,
      "grad_norm": 3155.3505859375,
      "learning_rate": 0.0001890944426525905,
      "loss": 7.5427,
      "step": 129100
    },
    {
      "epoch": 15.536702767749698,
      "grad_norm": 2341.677978515625,
      "learning_rate": 0.00018909271502219848,
      "loss": 7.5057,
      "step": 129110
    },
    {
      "epoch": 15.537906137184116,
      "grad_norm": 15337.583984375,
      "learning_rate": 0.00018909098726286732,
      "loss": 7.5648,
      "step": 129120
    },
    {
      "epoch": 15.539109506618532,
      "grad_norm": 25096.537109375,
      "learning_rate": 0.00018908925937459947,
      "loss": 7.5205,
      "step": 129130
    },
    {
      "epoch": 15.540312876052948,
      "grad_norm": 7579.16015625,
      "learning_rate": 0.00018908753135739748,
      "loss": 7.5733,
      "step": 129140
    },
    {
      "epoch": 15.541516245487365,
      "grad_norm": 17752.521484375,
      "learning_rate": 0.0001890858032112638,
      "loss": 7.5869,
      "step": 129150
    },
    {
      "epoch": 15.542719614921781,
      "grad_norm": 8619.4423828125,
      "learning_rate": 0.00018908407493620102,
      "loss": 7.4921,
      "step": 129160
    },
    {
      "epoch": 15.543922984356197,
      "grad_norm": 302.3937683105469,
      "learning_rate": 0.00018908234653221157,
      "loss": 7.5666,
      "step": 129170
    },
    {
      "epoch": 15.545126353790614,
      "grad_norm": 673.1851806640625,
      "learning_rate": 0.00018908061799929798,
      "loss": 7.6795,
      "step": 129180
    },
    {
      "epoch": 15.54632972322503,
      "grad_norm": 344.5409851074219,
      "learning_rate": 0.00018907888933746274,
      "loss": 7.707,
      "step": 129190
    },
    {
      "epoch": 15.547533092659446,
      "grad_norm": 573.9432373046875,
      "learning_rate": 0.00018907716054670834,
      "loss": 7.5703,
      "step": 129200
    },
    {
      "epoch": 15.548736462093864,
      "grad_norm": 312.3635559082031,
      "learning_rate": 0.0001890754316270373,
      "loss": 7.7668,
      "step": 129210
    },
    {
      "epoch": 15.54993983152828,
      "grad_norm": 326.55804443359375,
      "learning_rate": 0.00018907370257845215,
      "loss": 7.5852,
      "step": 129220
    },
    {
      "epoch": 15.551143200962695,
      "grad_norm": 838.5820922851562,
      "learning_rate": 0.0001890719734009553,
      "loss": 7.6044,
      "step": 129230
    },
    {
      "epoch": 15.552346570397113,
      "grad_norm": 297.2216796875,
      "learning_rate": 0.00018907024409454937,
      "loss": 7.6065,
      "step": 129240
    },
    {
      "epoch": 15.553549939831528,
      "grad_norm": 314.0319519042969,
      "learning_rate": 0.00018906851465923675,
      "loss": 7.594,
      "step": 129250
    },
    {
      "epoch": 15.554753309265944,
      "grad_norm": 51.431846618652344,
      "learning_rate": 0.00018906678509502008,
      "loss": 7.5334,
      "step": 129260
    },
    {
      "epoch": 15.555956678700362,
      "grad_norm": 77.34065246582031,
      "learning_rate": 0.00018906505540190171,
      "loss": 7.6404,
      "step": 129270
    },
    {
      "epoch": 15.557160048134778,
      "grad_norm": 138.8236846923828,
      "learning_rate": 0.00018906332557988424,
      "loss": 7.704,
      "step": 129280
    },
    {
      "epoch": 15.558363417569193,
      "grad_norm": 44.31373596191406,
      "learning_rate": 0.00018906159562897016,
      "loss": 7.5354,
      "step": 129290
    },
    {
      "epoch": 15.559566787003611,
      "grad_norm": 50.43186950683594,
      "learning_rate": 0.00018905986554916196,
      "loss": 7.6112,
      "step": 129300
    },
    {
      "epoch": 15.560770156438027,
      "grad_norm": 54.885650634765625,
      "learning_rate": 0.00018905813534046212,
      "loss": 7.4474,
      "step": 129310
    },
    {
      "epoch": 15.561973525872443,
      "grad_norm": 24.321998596191406,
      "learning_rate": 0.00018905640500287317,
      "loss": 7.5927,
      "step": 129320
    },
    {
      "epoch": 15.56317689530686,
      "grad_norm": 46.89988327026367,
      "learning_rate": 0.00018905467453639765,
      "loss": 7.597,
      "step": 129330
    },
    {
      "epoch": 15.564380264741276,
      "grad_norm": 23.91341209411621,
      "learning_rate": 0.000189052943941038,
      "loss": 7.6583,
      "step": 129340
    },
    {
      "epoch": 15.565583634175692,
      "grad_norm": 24.96449089050293,
      "learning_rate": 0.0001890512132167968,
      "loss": 7.6456,
      "step": 129350
    },
    {
      "epoch": 15.566787003610107,
      "grad_norm": 21.215200424194336,
      "learning_rate": 0.00018904948236367645,
      "loss": 7.5274,
      "step": 129360
    },
    {
      "epoch": 15.567990373044525,
      "grad_norm": 20.48535919189453,
      "learning_rate": 0.00018904775138167953,
      "loss": 7.6008,
      "step": 129370
    },
    {
      "epoch": 15.56919374247894,
      "grad_norm": 39.43841552734375,
      "learning_rate": 0.00018904602027080856,
      "loss": 7.6322,
      "step": 129380
    },
    {
      "epoch": 15.570397111913357,
      "grad_norm": 37.140445709228516,
      "learning_rate": 0.00018904428903106598,
      "loss": 7.5993,
      "step": 129390
    },
    {
      "epoch": 15.571600481347774,
      "grad_norm": 46.02839279174805,
      "learning_rate": 0.00018904255766245432,
      "loss": 7.5191,
      "step": 129400
    },
    {
      "epoch": 15.57280385078219,
      "grad_norm": 44.908042907714844,
      "learning_rate": 0.00018904082616497613,
      "loss": 7.4756,
      "step": 129410
    },
    {
      "epoch": 15.574007220216606,
      "grad_norm": 32.054500579833984,
      "learning_rate": 0.00018903909453863387,
      "loss": 7.4928,
      "step": 129420
    },
    {
      "epoch": 15.575210589651023,
      "grad_norm": 64.82672119140625,
      "learning_rate": 0.00018903736278343006,
      "loss": 7.5339,
      "step": 129430
    },
    {
      "epoch": 15.57641395908544,
      "grad_norm": 29.251670837402344,
      "learning_rate": 0.00018903563089936717,
      "loss": 7.5726,
      "step": 129440
    },
    {
      "epoch": 15.577617328519855,
      "grad_norm": 109.84314727783203,
      "learning_rate": 0.00018903389888644777,
      "loss": 7.6259,
      "step": 129450
    },
    {
      "epoch": 15.578820697954272,
      "grad_norm": 46.81904983520508,
      "learning_rate": 0.00018903216674467433,
      "loss": 7.6572,
      "step": 129460
    },
    {
      "epoch": 15.580024067388688,
      "grad_norm": 1282.04541015625,
      "learning_rate": 0.00018903043447404937,
      "loss": 7.5675,
      "step": 129470
    },
    {
      "epoch": 15.581227436823104,
      "grad_norm": 706.8040771484375,
      "learning_rate": 0.0001890287020745754,
      "loss": 7.5486,
      "step": 129480
    },
    {
      "epoch": 15.582430806257522,
      "grad_norm": 740.2406005859375,
      "learning_rate": 0.0001890269695462549,
      "loss": 7.617,
      "step": 129490
    },
    {
      "epoch": 15.583634175691937,
      "grad_norm": 295.137939453125,
      "learning_rate": 0.0001890252368890904,
      "loss": 7.5754,
      "step": 129500
    },
    {
      "epoch": 15.584837545126353,
      "grad_norm": 116.53155517578125,
      "learning_rate": 0.0001890235041030844,
      "loss": 7.4607,
      "step": 129510
    },
    {
      "epoch": 15.58604091456077,
      "grad_norm": 641.5328369140625,
      "learning_rate": 0.00018902177118823943,
      "loss": 7.5049,
      "step": 129520
    },
    {
      "epoch": 15.587244283995187,
      "grad_norm": 1330.26416015625,
      "learning_rate": 0.00018902003814455794,
      "loss": 7.4866,
      "step": 129530
    },
    {
      "epoch": 15.588447653429602,
      "grad_norm": 317.4263916015625,
      "learning_rate": 0.00018901830497204252,
      "loss": 7.5133,
      "step": 129540
    },
    {
      "epoch": 15.58965102286402,
      "grad_norm": 3190.15087890625,
      "learning_rate": 0.00018901657167069558,
      "loss": 7.4912,
      "step": 129550
    },
    {
      "epoch": 15.590854392298436,
      "grad_norm": 581.2105102539062,
      "learning_rate": 0.00018901483824051974,
      "loss": 7.3774,
      "step": 129560
    },
    {
      "epoch": 15.592057761732852,
      "grad_norm": 403.0574035644531,
      "learning_rate": 0.00018901310468151742,
      "loss": 7.65,
      "step": 129570
    },
    {
      "epoch": 15.593261131167269,
      "grad_norm": 646.2369384765625,
      "learning_rate": 0.00018901137099369114,
      "loss": 7.4794,
      "step": 129580
    },
    {
      "epoch": 15.594464500601685,
      "grad_norm": 116.70044708251953,
      "learning_rate": 0.0001890096371770435,
      "loss": 7.4795,
      "step": 129590
    },
    {
      "epoch": 15.5956678700361,
      "grad_norm": 284.9908447265625,
      "learning_rate": 0.0001890079032315769,
      "loss": 7.5945,
      "step": 129600
    },
    {
      "epoch": 15.596871239470518,
      "grad_norm": 1824.6612548828125,
      "learning_rate": 0.00018900616915729388,
      "loss": 7.6665,
      "step": 129610
    },
    {
      "epoch": 15.598074608904934,
      "grad_norm": 61.20644760131836,
      "learning_rate": 0.00018900443495419695,
      "loss": 7.5661,
      "step": 129620
    },
    {
      "epoch": 15.59927797833935,
      "grad_norm": 75.7002944946289,
      "learning_rate": 0.00018900270062228864,
      "loss": 7.4868,
      "step": 129630
    },
    {
      "epoch": 15.600481347773766,
      "grad_norm": 52.21888732910156,
      "learning_rate": 0.00018900096616157144,
      "loss": 7.5386,
      "step": 129640
    },
    {
      "epoch": 15.601684717208183,
      "grad_norm": 97.52935791015625,
      "learning_rate": 0.00018899923157204788,
      "loss": 7.387,
      "step": 129650
    },
    {
      "epoch": 15.602888086642599,
      "grad_norm": 92.35443878173828,
      "learning_rate": 0.00018899749685372049,
      "loss": 7.5255,
      "step": 129660
    },
    {
      "epoch": 15.604091456077015,
      "grad_norm": 95.04878997802734,
      "learning_rate": 0.00018899576200659168,
      "loss": 7.4623,
      "step": 129670
    },
    {
      "epoch": 15.605294825511432,
      "grad_norm": 196.79928588867188,
      "learning_rate": 0.00018899402703066408,
      "loss": 7.4881,
      "step": 129680
    },
    {
      "epoch": 15.606498194945848,
      "grad_norm": 178.7354736328125,
      "learning_rate": 0.00018899229192594015,
      "loss": 7.5186,
      "step": 129690
    },
    {
      "epoch": 15.607701564380264,
      "grad_norm": 1128.239501953125,
      "learning_rate": 0.00018899055669242238,
      "loss": 7.4829,
      "step": 129700
    },
    {
      "epoch": 15.608904933814681,
      "grad_norm": 316.3932800292969,
      "learning_rate": 0.0001889888213301133,
      "loss": 7.5183,
      "step": 129710
    },
    {
      "epoch": 15.610108303249097,
      "grad_norm": 203.64212036132812,
      "learning_rate": 0.00018898708583901542,
      "loss": 7.4503,
      "step": 129720
    },
    {
      "epoch": 15.611311672683513,
      "grad_norm": 636.5574951171875,
      "learning_rate": 0.0001889853502191313,
      "loss": 7.5411,
      "step": 129730
    },
    {
      "epoch": 15.61251504211793,
      "grad_norm": 1181.5462646484375,
      "learning_rate": 0.0001889836144704634,
      "loss": 7.4925,
      "step": 129740
    },
    {
      "epoch": 15.613718411552346,
      "grad_norm": 218.50665283203125,
      "learning_rate": 0.0001889818785930142,
      "loss": 7.5558,
      "step": 129750
    },
    {
      "epoch": 15.614921780986762,
      "grad_norm": 116.60340118408203,
      "learning_rate": 0.00018898014258678626,
      "loss": 7.5185,
      "step": 129760
    },
    {
      "epoch": 15.61612515042118,
      "grad_norm": 131.9979705810547,
      "learning_rate": 0.00018897840645178214,
      "loss": 7.4183,
      "step": 129770
    },
    {
      "epoch": 15.617328519855596,
      "grad_norm": 318.1040954589844,
      "learning_rate": 0.00018897667018800426,
      "loss": 7.4749,
      "step": 129780
    },
    {
      "epoch": 15.618531889290011,
      "grad_norm": 216.2886962890625,
      "learning_rate": 0.00018897493379545517,
      "loss": 7.4415,
      "step": 129790
    },
    {
      "epoch": 15.619735258724429,
      "grad_norm": 5524.82861328125,
      "learning_rate": 0.00018897319727413737,
      "loss": 7.3992,
      "step": 129800
    },
    {
      "epoch": 15.620938628158845,
      "grad_norm": 3024.58349609375,
      "learning_rate": 0.0001889714606240534,
      "loss": 7.4304,
      "step": 129810
    },
    {
      "epoch": 15.62214199759326,
      "grad_norm": 462.4556884765625,
      "learning_rate": 0.00018896972384520577,
      "loss": 7.4408,
      "step": 129820
    },
    {
      "epoch": 15.623345367027678,
      "grad_norm": 363.1320495605469,
      "learning_rate": 0.00018896798693759695,
      "loss": 7.4105,
      "step": 129830
    },
    {
      "epoch": 15.624548736462094,
      "grad_norm": 505.0877990722656,
      "learning_rate": 0.00018896624990122953,
      "loss": 7.4509,
      "step": 129840
    },
    {
      "epoch": 15.62575210589651,
      "grad_norm": 347.98187255859375,
      "learning_rate": 0.00018896451273610595,
      "loss": 7.4938,
      "step": 129850
    },
    {
      "epoch": 15.626955475330927,
      "grad_norm": 400.0885314941406,
      "learning_rate": 0.00018896277544222878,
      "loss": 7.4287,
      "step": 129860
    },
    {
      "epoch": 15.628158844765343,
      "grad_norm": 325.1827087402344,
      "learning_rate": 0.0001889610380196005,
      "loss": 7.3676,
      "step": 129870
    },
    {
      "epoch": 15.629362214199759,
      "grad_norm": 237.39175415039062,
      "learning_rate": 0.00018895930046822364,
      "loss": 7.4048,
      "step": 129880
    },
    {
      "epoch": 15.630565583634176,
      "grad_norm": 135.3329620361328,
      "learning_rate": 0.0001889575627881007,
      "loss": 7.4627,
      "step": 129890
    },
    {
      "epoch": 15.631768953068592,
      "grad_norm": 311.0946350097656,
      "learning_rate": 0.00018895582497923421,
      "loss": 7.375,
      "step": 129900
    },
    {
      "epoch": 15.632972322503008,
      "grad_norm": 204.72337341308594,
      "learning_rate": 0.00018895408704162664,
      "loss": 7.4442,
      "step": 129910
    },
    {
      "epoch": 15.634175691937426,
      "grad_norm": 301.5276184082031,
      "learning_rate": 0.0001889523489752806,
      "loss": 7.3823,
      "step": 129920
    },
    {
      "epoch": 15.635379061371841,
      "grad_norm": 360.1788635253906,
      "learning_rate": 0.0001889506107801985,
      "loss": 7.4466,
      "step": 129930
    },
    {
      "epoch": 15.636582430806257,
      "grad_norm": 267.43511962890625,
      "learning_rate": 0.00018894887245638294,
      "loss": 7.4446,
      "step": 129940
    },
    {
      "epoch": 15.637785800240675,
      "grad_norm": 401.21533203125,
      "learning_rate": 0.00018894713400383642,
      "loss": 7.4477,
      "step": 129950
    },
    {
      "epoch": 15.63898916967509,
      "grad_norm": 379.4913635253906,
      "learning_rate": 0.0001889453954225614,
      "loss": 7.4691,
      "step": 129960
    },
    {
      "epoch": 15.640192539109506,
      "grad_norm": 447.0564270019531,
      "learning_rate": 0.00018894365671256045,
      "loss": 7.4469,
      "step": 129970
    },
    {
      "epoch": 15.641395908543924,
      "grad_norm": 616.8658447265625,
      "learning_rate": 0.00018894191787383603,
      "loss": 7.3652,
      "step": 129980
    },
    {
      "epoch": 15.64259927797834,
      "grad_norm": 367.65478515625,
      "learning_rate": 0.00018894017890639075,
      "loss": 7.393,
      "step": 129990
    },
    {
      "epoch": 15.643802647412755,
      "grad_norm": 820.666259765625,
      "learning_rate": 0.00018893843981022704,
      "loss": 7.4179,
      "step": 130000
    },
    {
      "epoch": 15.645006016847173,
      "grad_norm": 387.88372802734375,
      "learning_rate": 0.00018893670058534747,
      "loss": 7.4676,
      "step": 130010
    },
    {
      "epoch": 15.646209386281589,
      "grad_norm": 443.2305908203125,
      "learning_rate": 0.00018893496123175454,
      "loss": 7.4074,
      "step": 130020
    },
    {
      "epoch": 15.647412755716005,
      "grad_norm": 919.445556640625,
      "learning_rate": 0.00018893322174945074,
      "loss": 7.4859,
      "step": 130030
    },
    {
      "epoch": 15.648616125150422,
      "grad_norm": 524.7247924804688,
      "learning_rate": 0.00018893148213843861,
      "loss": 7.4252,
      "step": 130040
    },
    {
      "epoch": 15.649819494584838,
      "grad_norm": 730.0946044921875,
      "learning_rate": 0.00018892974239872068,
      "loss": 7.4358,
      "step": 130050
    },
    {
      "epoch": 15.651022864019254,
      "grad_norm": 849.2393798828125,
      "learning_rate": 0.00018892800253029947,
      "loss": 7.4084,
      "step": 130060
    },
    {
      "epoch": 15.65222623345367,
      "grad_norm": 1203.624267578125,
      "learning_rate": 0.00018892626253317748,
      "loss": 7.3498,
      "step": 130070
    },
    {
      "epoch": 15.653429602888087,
      "grad_norm": 819.9821166992188,
      "learning_rate": 0.0001889245224073572,
      "loss": 7.4228,
      "step": 130080
    },
    {
      "epoch": 15.654632972322503,
      "grad_norm": 465.1644592285156,
      "learning_rate": 0.0001889227821528412,
      "loss": 7.4745,
      "step": 130090
    },
    {
      "epoch": 15.655836341756919,
      "grad_norm": 1157.712158203125,
      "learning_rate": 0.000188921041769632,
      "loss": 7.4576,
      "step": 130100
    },
    {
      "epoch": 15.657039711191336,
      "grad_norm": 1139.73876953125,
      "learning_rate": 0.00018891930125773207,
      "loss": 7.3787,
      "step": 130110
    },
    {
      "epoch": 15.658243080625752,
      "grad_norm": 817.1405639648438,
      "learning_rate": 0.000188917560617144,
      "loss": 7.4338,
      "step": 130120
    },
    {
      "epoch": 15.659446450060168,
      "grad_norm": 750.9105834960938,
      "learning_rate": 0.0001889158198478702,
      "loss": 7.4513,
      "step": 130130
    },
    {
      "epoch": 15.660649819494585,
      "grad_norm": 550.6473999023438,
      "learning_rate": 0.0001889140789499133,
      "loss": 7.4223,
      "step": 130140
    },
    {
      "epoch": 15.661853188929001,
      "grad_norm": 494.1597900390625,
      "learning_rate": 0.0001889123379232758,
      "loss": 7.4624,
      "step": 130150
    },
    {
      "epoch": 15.663056558363417,
      "grad_norm": 545.972412109375,
      "learning_rate": 0.00018891059676796017,
      "loss": 7.4263,
      "step": 130160
    },
    {
      "epoch": 15.664259927797834,
      "grad_norm": 373.7424621582031,
      "learning_rate": 0.00018890885548396896,
      "loss": 7.3634,
      "step": 130170
    },
    {
      "epoch": 15.66546329723225,
      "grad_norm": 1047.8740234375,
      "learning_rate": 0.00018890711407130466,
      "loss": 7.4613,
      "step": 130180
    },
    {
      "epoch": 15.666666666666666,
      "grad_norm": 503.9117126464844,
      "learning_rate": 0.00018890537252996983,
      "loss": 7.4797,
      "step": 130190
    },
    {
      "epoch": 15.667870036101084,
      "grad_norm": 813.9482421875,
      "learning_rate": 0.00018890363085996697,
      "loss": 7.5782,
      "step": 130200
    },
    {
      "epoch": 15.6690734055355,
      "grad_norm": 656.7713623046875,
      "learning_rate": 0.00018890188906129862,
      "loss": 7.482,
      "step": 130210
    },
    {
      "epoch": 15.670276774969915,
      "grad_norm": 551.8267822265625,
      "learning_rate": 0.00018890014713396727,
      "loss": 7.4858,
      "step": 130220
    },
    {
      "epoch": 15.671480144404333,
      "grad_norm": 595.2601318359375,
      "learning_rate": 0.00018889840507797548,
      "loss": 7.4328,
      "step": 130230
    },
    {
      "epoch": 15.672683513838749,
      "grad_norm": 447.2945861816406,
      "learning_rate": 0.00018889666289332574,
      "loss": 7.419,
      "step": 130240
    },
    {
      "epoch": 15.673886883273164,
      "grad_norm": 1281.6627197265625,
      "learning_rate": 0.00018889492058002055,
      "loss": 7.4653,
      "step": 130250
    },
    {
      "epoch": 15.675090252707582,
      "grad_norm": 1769.7283935546875,
      "learning_rate": 0.0001888931781380625,
      "loss": 7.524,
      "step": 130260
    },
    {
      "epoch": 15.676293622141998,
      "grad_norm": 768.0513305664062,
      "learning_rate": 0.00018889143556745407,
      "loss": 7.4309,
      "step": 130270
    },
    {
      "epoch": 15.677496991576414,
      "grad_norm": 1498.276123046875,
      "learning_rate": 0.0001888896928681978,
      "loss": 7.3979,
      "step": 130280
    },
    {
      "epoch": 15.678700361010831,
      "grad_norm": 626.4533081054688,
      "learning_rate": 0.00018888795004029618,
      "loss": 7.4299,
      "step": 130290
    },
    {
      "epoch": 15.679903730445247,
      "grad_norm": 1042.447265625,
      "learning_rate": 0.00018888620708375177,
      "loss": 7.3081,
      "step": 130300
    },
    {
      "epoch": 15.681107099879663,
      "grad_norm": 898.1484375,
      "learning_rate": 0.00018888446399856703,
      "loss": 7.3163,
      "step": 130310
    },
    {
      "epoch": 15.68231046931408,
      "grad_norm": 586.4484252929688,
      "learning_rate": 0.00018888272078474455,
      "loss": 7.3992,
      "step": 130320
    },
    {
      "epoch": 15.683513838748496,
      "grad_norm": 432.5578308105469,
      "learning_rate": 0.00018888097744228683,
      "loss": 7.4685,
      "step": 130330
    },
    {
      "epoch": 15.684717208182912,
      "grad_norm": 705.9630737304688,
      "learning_rate": 0.0001888792339711964,
      "loss": 7.4727,
      "step": 130340
    },
    {
      "epoch": 15.685920577617328,
      "grad_norm": 1393.160400390625,
      "learning_rate": 0.00018887749037147574,
      "loss": 7.3867,
      "step": 130350
    },
    {
      "epoch": 15.687123947051745,
      "grad_norm": 931.7964477539062,
      "learning_rate": 0.00018887574664312745,
      "loss": 7.4873,
      "step": 130360
    },
    {
      "epoch": 15.688327316486161,
      "grad_norm": 2567.56201171875,
      "learning_rate": 0.00018887400278615402,
      "loss": 7.4288,
      "step": 130370
    },
    {
      "epoch": 15.689530685920577,
      "grad_norm": 1022.0443725585938,
      "learning_rate": 0.00018887225880055792,
      "loss": 7.3856,
      "step": 130380
    },
    {
      "epoch": 15.690734055354994,
      "grad_norm": 602.54736328125,
      "learning_rate": 0.00018887051468634175,
      "loss": 7.3619,
      "step": 130390
    },
    {
      "epoch": 15.69193742478941,
      "grad_norm": 816.98828125,
      "learning_rate": 0.000188868770443508,
      "loss": 7.392,
      "step": 130400
    },
    {
      "epoch": 15.693140794223826,
      "grad_norm": 587.8736572265625,
      "learning_rate": 0.00018886702607205918,
      "loss": 7.3641,
      "step": 130410
    },
    {
      "epoch": 15.694344163658243,
      "grad_norm": 735.8976440429688,
      "learning_rate": 0.00018886528157199784,
      "loss": 7.4425,
      "step": 130420
    },
    {
      "epoch": 15.69554753309266,
      "grad_norm": 2796.738525390625,
      "learning_rate": 0.00018886353694332652,
      "loss": 7.3884,
      "step": 130430
    },
    {
      "epoch": 15.696750902527075,
      "grad_norm": 1038.78125,
      "learning_rate": 0.00018886179218604768,
      "loss": 7.4072,
      "step": 130440
    },
    {
      "epoch": 15.697954271961493,
      "grad_norm": 1287.992919921875,
      "learning_rate": 0.00018886004730016393,
      "loss": 7.413,
      "step": 130450
    },
    {
      "epoch": 15.699157641395908,
      "grad_norm": 485.7673034667969,
      "learning_rate": 0.00018885830228567774,
      "loss": 7.4314,
      "step": 130460
    },
    {
      "epoch": 15.700361010830324,
      "grad_norm": 653.8545532226562,
      "learning_rate": 0.00018885655714259163,
      "loss": 7.4078,
      "step": 130470
    },
    {
      "epoch": 15.701564380264742,
      "grad_norm": 942.43017578125,
      "learning_rate": 0.00018885481187090817,
      "loss": 7.5279,
      "step": 130480
    },
    {
      "epoch": 15.702767749699158,
      "grad_norm": 1374.56298828125,
      "learning_rate": 0.00018885306647062984,
      "loss": 7.4221,
      "step": 130490
    },
    {
      "epoch": 15.703971119133573,
      "grad_norm": 1813.4957275390625,
      "learning_rate": 0.0001888513209417592,
      "loss": 7.5479,
      "step": 130500
    },
    {
      "epoch": 15.705174488567991,
      "grad_norm": 1242.5025634765625,
      "learning_rate": 0.00018884957528429873,
      "loss": 7.4474,
      "step": 130510
    },
    {
      "epoch": 15.706377858002407,
      "grad_norm": 841.1098022460938,
      "learning_rate": 0.00018884782949825103,
      "loss": 7.5788,
      "step": 130520
    },
    {
      "epoch": 15.707581227436823,
      "grad_norm": 750.1177978515625,
      "learning_rate": 0.00018884608358361857,
      "loss": 7.539,
      "step": 130530
    },
    {
      "epoch": 15.70878459687124,
      "grad_norm": 818.6088256835938,
      "learning_rate": 0.00018884433754040388,
      "loss": 7.4332,
      "step": 130540
    },
    {
      "epoch": 15.709987966305656,
      "grad_norm": 978.2344360351562,
      "learning_rate": 0.0001888425913686095,
      "loss": 7.3983,
      "step": 130550
    },
    {
      "epoch": 15.711191335740072,
      "grad_norm": 875.69482421875,
      "learning_rate": 0.000188840845068238,
      "loss": 7.4338,
      "step": 130560
    },
    {
      "epoch": 15.71239470517449,
      "grad_norm": 1089.465576171875,
      "learning_rate": 0.0001888390986392918,
      "loss": 7.3953,
      "step": 130570
    },
    {
      "epoch": 15.713598074608905,
      "grad_norm": 782.388916015625,
      "learning_rate": 0.00018883735208177353,
      "loss": 7.345,
      "step": 130580
    },
    {
      "epoch": 15.71480144404332,
      "grad_norm": 1078.1539306640625,
      "learning_rate": 0.00018883560539568567,
      "loss": 7.3685,
      "step": 130590
    },
    {
      "epoch": 15.716004813477738,
      "grad_norm": 1036.6904296875,
      "learning_rate": 0.00018883385858103075,
      "loss": 7.4876,
      "step": 130600
    },
    {
      "epoch": 15.717208182912154,
      "grad_norm": 854.8037109375,
      "learning_rate": 0.00018883211163781132,
      "loss": 7.2753,
      "step": 130610
    },
    {
      "epoch": 15.71841155234657,
      "grad_norm": 704.491943359375,
      "learning_rate": 0.00018883036456602986,
      "loss": 7.3449,
      "step": 130620
    },
    {
      "epoch": 15.719614921780988,
      "grad_norm": 2854.128173828125,
      "learning_rate": 0.00018882861736568897,
      "loss": 7.3086,
      "step": 130630
    },
    {
      "epoch": 15.720818291215403,
      "grad_norm": 1014.7636108398438,
      "learning_rate": 0.0001888268700367911,
      "loss": 7.3472,
      "step": 130640
    },
    {
      "epoch": 15.722021660649819,
      "grad_norm": 1005.1486206054688,
      "learning_rate": 0.00018882512257933885,
      "loss": 7.4366,
      "step": 130650
    },
    {
      "epoch": 15.723225030084237,
      "grad_norm": 940.9915771484375,
      "learning_rate": 0.00018882337499333472,
      "loss": 7.4396,
      "step": 130660
    },
    {
      "epoch": 15.724428399518652,
      "grad_norm": 690.8624267578125,
      "learning_rate": 0.00018882162727878118,
      "loss": 7.4627,
      "step": 130670
    },
    {
      "epoch": 15.725631768953068,
      "grad_norm": 647.682373046875,
      "learning_rate": 0.0001888198794356809,
      "loss": 7.2562,
      "step": 130680
    },
    {
      "epoch": 15.726835138387486,
      "grad_norm": 552.2103271484375,
      "learning_rate": 0.00018881813146403627,
      "loss": 7.3659,
      "step": 130690
    },
    {
      "epoch": 15.728038507821902,
      "grad_norm": 1565.328125,
      "learning_rate": 0.00018881638336384988,
      "loss": 7.4248,
      "step": 130700
    },
    {
      "epoch": 15.729241877256317,
      "grad_norm": 410.8542175292969,
      "learning_rate": 0.00018881463513512427,
      "loss": 7.3311,
      "step": 130710
    },
    {
      "epoch": 15.730445246690735,
      "grad_norm": 1234.8740234375,
      "learning_rate": 0.00018881288677786195,
      "loss": 7.3848,
      "step": 130720
    },
    {
      "epoch": 15.73164861612515,
      "grad_norm": 660.88720703125,
      "learning_rate": 0.00018881113829206548,
      "loss": 7.3114,
      "step": 130730
    },
    {
      "epoch": 15.732851985559567,
      "grad_norm": 337.2177734375,
      "learning_rate": 0.0001888093896777373,
      "loss": 7.4448,
      "step": 130740
    },
    {
      "epoch": 15.734055354993982,
      "grad_norm": 876.92431640625,
      "learning_rate": 0.00018880764093488005,
      "loss": 7.4242,
      "step": 130750
    },
    {
      "epoch": 15.7352587244284,
      "grad_norm": 1161.06689453125,
      "learning_rate": 0.00018880589206349624,
      "loss": 7.3932,
      "step": 130760
    },
    {
      "epoch": 15.736462093862816,
      "grad_norm": 919.97998046875,
      "learning_rate": 0.00018880414306358834,
      "loss": 7.4249,
      "step": 130770
    },
    {
      "epoch": 15.737665463297231,
      "grad_norm": 701.520263671875,
      "learning_rate": 0.00018880239393515895,
      "loss": 7.4571,
      "step": 130780
    },
    {
      "epoch": 15.738868832731649,
      "grad_norm": 633.79443359375,
      "learning_rate": 0.00018880064467821054,
      "loss": 7.4091,
      "step": 130790
    },
    {
      "epoch": 15.740072202166065,
      "grad_norm": 925.3358764648438,
      "learning_rate": 0.0001887988952927457,
      "loss": 7.3466,
      "step": 130800
    },
    {
      "epoch": 15.74127557160048,
      "grad_norm": 2348.89013671875,
      "learning_rate": 0.00018879714577876692,
      "loss": 7.2597,
      "step": 130810
    },
    {
      "epoch": 15.742478941034898,
      "grad_norm": 1451.4468994140625,
      "learning_rate": 0.00018879539613627676,
      "loss": 7.2688,
      "step": 130820
    },
    {
      "epoch": 15.743682310469314,
      "grad_norm": 519.1000366210938,
      "learning_rate": 0.00018879364636527771,
      "loss": 7.2851,
      "step": 130830
    },
    {
      "epoch": 15.74488567990373,
      "grad_norm": 489.55072021484375,
      "learning_rate": 0.00018879189646577236,
      "loss": 7.2179,
      "step": 130840
    },
    {
      "epoch": 15.746089049338147,
      "grad_norm": 962.341796875,
      "learning_rate": 0.00018879014643776321,
      "loss": 7.3851,
      "step": 130850
    },
    {
      "epoch": 15.747292418772563,
      "grad_norm": 1485.290771484375,
      "learning_rate": 0.0001887883962812528,
      "loss": 7.1963,
      "step": 130860
    },
    {
      "epoch": 15.748495788206979,
      "grad_norm": 507.83209228515625,
      "learning_rate": 0.00018878664599624364,
      "loss": 7.325,
      "step": 130870
    },
    {
      "epoch": 15.749699157641396,
      "grad_norm": 1357.3214111328125,
      "learning_rate": 0.0001887848955827383,
      "loss": 7.3346,
      "step": 130880
    },
    {
      "epoch": 15.750902527075812,
      "grad_norm": 641.6832275390625,
      "learning_rate": 0.00018878314504073928,
      "loss": 7.4197,
      "step": 130890
    },
    {
      "epoch": 15.752105896510228,
      "grad_norm": 1423.2657470703125,
      "learning_rate": 0.00018878139437024914,
      "loss": 7.3629,
      "step": 130900
    },
    {
      "epoch": 15.753309265944646,
      "grad_norm": 818.1052856445312,
      "learning_rate": 0.00018877964357127042,
      "loss": 7.3241,
      "step": 130910
    },
    {
      "epoch": 15.754512635379061,
      "grad_norm": 841.8325805664062,
      "learning_rate": 0.0001887778926438056,
      "loss": 7.3025,
      "step": 130920
    },
    {
      "epoch": 15.755716004813477,
      "grad_norm": 1068.5340576171875,
      "learning_rate": 0.00018877614158785728,
      "loss": 7.3987,
      "step": 130930
    },
    {
      "epoch": 15.756919374247895,
      "grad_norm": 1043.1324462890625,
      "learning_rate": 0.00018877439040342793,
      "loss": 7.3959,
      "step": 130940
    },
    {
      "epoch": 15.75812274368231,
      "grad_norm": 1346.5728759765625,
      "learning_rate": 0.00018877263909052014,
      "loss": 7.3352,
      "step": 130950
    },
    {
      "epoch": 15.759326113116726,
      "grad_norm": 1050.9190673828125,
      "learning_rate": 0.00018877088764913643,
      "loss": 7.2268,
      "step": 130960
    },
    {
      "epoch": 15.760529482551144,
      "grad_norm": 1043.7625732421875,
      "learning_rate": 0.00018876913607927934,
      "loss": 7.4861,
      "step": 130970
    },
    {
      "epoch": 15.76173285198556,
      "grad_norm": 1170.1287841796875,
      "learning_rate": 0.0001887673843809514,
      "loss": 7.3023,
      "step": 130980
    },
    {
      "epoch": 15.762936221419976,
      "grad_norm": 1080.5343017578125,
      "learning_rate": 0.0001887656325541551,
      "loss": 7.3434,
      "step": 130990
    },
    {
      "epoch": 15.764139590854393,
      "grad_norm": 1283.1015625,
      "learning_rate": 0.00018876388059889303,
      "loss": 7.2833,
      "step": 131000
    },
    {
      "epoch": 15.765342960288809,
      "grad_norm": 1127.6925048828125,
      "learning_rate": 0.0001887621285151677,
      "loss": 7.4026,
      "step": 131010
    },
    {
      "epoch": 15.766546329723225,
      "grad_norm": 1644.163818359375,
      "learning_rate": 0.00018876037630298167,
      "loss": 7.4056,
      "step": 131020
    },
    {
      "epoch": 15.76774969915764,
      "grad_norm": 1265.3726806640625,
      "learning_rate": 0.00018875862396233747,
      "loss": 7.2092,
      "step": 131030
    },
    {
      "epoch": 15.768953068592058,
      "grad_norm": 746.2817993164062,
      "learning_rate": 0.0001887568714932376,
      "loss": 7.3911,
      "step": 131040
    },
    {
      "epoch": 15.770156438026474,
      "grad_norm": 1599.3319091796875,
      "learning_rate": 0.00018875511889568464,
      "loss": 7.2268,
      "step": 131050
    },
    {
      "epoch": 15.77135980746089,
      "grad_norm": 1164.397216796875,
      "learning_rate": 0.0001887533661696811,
      "loss": 7.3815,
      "step": 131060
    },
    {
      "epoch": 15.772563176895307,
      "grad_norm": 900.9363403320312,
      "learning_rate": 0.00018875161331522957,
      "loss": 7.3425,
      "step": 131070
    },
    {
      "epoch": 15.773766546329723,
      "grad_norm": 1702.6241455078125,
      "learning_rate": 0.0001887498603323325,
      "loss": 7.3555,
      "step": 131080
    },
    {
      "epoch": 15.774969915764139,
      "grad_norm": 949.272705078125,
      "learning_rate": 0.00018874810722099248,
      "loss": 7.357,
      "step": 131090
    },
    {
      "epoch": 15.776173285198556,
      "grad_norm": 864.026123046875,
      "learning_rate": 0.00018874635398121204,
      "loss": 7.2819,
      "step": 131100
    },
    {
      "epoch": 15.777376654632972,
      "grad_norm": 1625.66162109375,
      "learning_rate": 0.00018874460061299371,
      "loss": 7.355,
      "step": 131110
    },
    {
      "epoch": 15.778580024067388,
      "grad_norm": 792.3026733398438,
      "learning_rate": 0.00018874284711634004,
      "loss": 7.3485,
      "step": 131120
    },
    {
      "epoch": 15.779783393501805,
      "grad_norm": 2529.9794921875,
      "learning_rate": 0.00018874109349125356,
      "loss": 7.2638,
      "step": 131130
    },
    {
      "epoch": 15.780986762936221,
      "grad_norm": 732.0496826171875,
      "learning_rate": 0.00018873933973773677,
      "loss": 7.3166,
      "step": 131140
    },
    {
      "epoch": 15.782190132370637,
      "grad_norm": 2237.266357421875,
      "learning_rate": 0.0001887375858557923,
      "loss": 7.3734,
      "step": 131150
    },
    {
      "epoch": 15.783393501805055,
      "grad_norm": 905.0309448242188,
      "learning_rate": 0.0001887358318454226,
      "loss": 7.4116,
      "step": 131160
    },
    {
      "epoch": 15.78459687123947,
      "grad_norm": 2812.263916015625,
      "learning_rate": 0.00018873407770663027,
      "loss": 7.3888,
      "step": 131170
    },
    {
      "epoch": 15.785800240673886,
      "grad_norm": 1823.0867919921875,
      "learning_rate": 0.00018873232343941782,
      "loss": 7.3182,
      "step": 131180
    },
    {
      "epoch": 15.787003610108304,
      "grad_norm": 1092.5263671875,
      "learning_rate": 0.00018873056904378778,
      "loss": 7.3221,
      "step": 131190
    },
    {
      "epoch": 15.78820697954272,
      "grad_norm": 1448.7960205078125,
      "learning_rate": 0.00018872881451974268,
      "loss": 7.3607,
      "step": 131200
    },
    {
      "epoch": 15.789410348977135,
      "grad_norm": 1206.72265625,
      "learning_rate": 0.00018872705986728512,
      "loss": 7.3801,
      "step": 131210
    },
    {
      "epoch": 15.790613718411553,
      "grad_norm": 1181.626708984375,
      "learning_rate": 0.00018872530508641758,
      "loss": 7.3637,
      "step": 131220
    },
    {
      "epoch": 15.791817087845969,
      "grad_norm": 522.133056640625,
      "learning_rate": 0.0001887235501771426,
      "loss": 7.3381,
      "step": 131230
    },
    {
      "epoch": 15.793020457280385,
      "grad_norm": 629.7036743164062,
      "learning_rate": 0.00018872179513946277,
      "loss": 7.2743,
      "step": 131240
    },
    {
      "epoch": 15.794223826714802,
      "grad_norm": 908.8665771484375,
      "learning_rate": 0.00018872003997338056,
      "loss": 7.3811,
      "step": 131250
    },
    {
      "epoch": 15.795427196149218,
      "grad_norm": 1087.2618408203125,
      "learning_rate": 0.00018871828467889858,
      "loss": 7.3002,
      "step": 131260
    },
    {
      "epoch": 15.796630565583634,
      "grad_norm": 652.273193359375,
      "learning_rate": 0.00018871652925601934,
      "loss": 7.2868,
      "step": 131270
    },
    {
      "epoch": 15.797833935018051,
      "grad_norm": 1055.8106689453125,
      "learning_rate": 0.00018871477370474535,
      "loss": 7.409,
      "step": 131280
    },
    {
      "epoch": 15.799037304452467,
      "grad_norm": 789.350830078125,
      "learning_rate": 0.00018871301802507923,
      "loss": 7.2364,
      "step": 131290
    },
    {
      "epoch": 15.800240673886883,
      "grad_norm": 1013.0385131835938,
      "learning_rate": 0.00018871126221702342,
      "loss": 7.346,
      "step": 131300
    },
    {
      "epoch": 15.8014440433213,
      "grad_norm": 527.7053833007812,
      "learning_rate": 0.00018870950628058051,
      "loss": 7.3471,
      "step": 131310
    },
    {
      "epoch": 15.802647412755716,
      "grad_norm": 586.3790893554688,
      "learning_rate": 0.00018870775021575305,
      "loss": 7.4011,
      "step": 131320
    },
    {
      "epoch": 15.803850782190132,
      "grad_norm": 776.3505859375,
      "learning_rate": 0.00018870599402254361,
      "loss": 7.3638,
      "step": 131330
    },
    {
      "epoch": 15.80505415162455,
      "grad_norm": 705.1693115234375,
      "learning_rate": 0.00018870423770095466,
      "loss": 7.2738,
      "step": 131340
    },
    {
      "epoch": 15.806257521058965,
      "grad_norm": 775.1263427734375,
      "learning_rate": 0.0001887024812509888,
      "loss": 7.4197,
      "step": 131350
    },
    {
      "epoch": 15.807460890493381,
      "grad_norm": 693.7694702148438,
      "learning_rate": 0.00018870072467264853,
      "loss": 7.288,
      "step": 131360
    },
    {
      "epoch": 15.808664259927799,
      "grad_norm": 1427.586181640625,
      "learning_rate": 0.00018869896796593644,
      "loss": 7.2597,
      "step": 131370
    },
    {
      "epoch": 15.809867629362214,
      "grad_norm": 505.89019775390625,
      "learning_rate": 0.000188697211130855,
      "loss": 7.2885,
      "step": 131380
    },
    {
      "epoch": 15.81107099879663,
      "grad_norm": 730.5121459960938,
      "learning_rate": 0.00018869545416740682,
      "loss": 7.2721,
      "step": 131390
    },
    {
      "epoch": 15.812274368231048,
      "grad_norm": 2388.80615234375,
      "learning_rate": 0.0001886936970755944,
      "loss": 7.2726,
      "step": 131400
    },
    {
      "epoch": 15.813477737665464,
      "grad_norm": 749.6900024414062,
      "learning_rate": 0.00018869193985542032,
      "loss": 7.3316,
      "step": 131410
    },
    {
      "epoch": 15.81468110709988,
      "grad_norm": 702.511962890625,
      "learning_rate": 0.00018869018250688714,
      "loss": 7.31,
      "step": 131420
    },
    {
      "epoch": 15.815884476534297,
      "grad_norm": 1237.361572265625,
      "learning_rate": 0.00018868842502999734,
      "loss": 7.3249,
      "step": 131430
    },
    {
      "epoch": 15.817087845968713,
      "grad_norm": 646.3370361328125,
      "learning_rate": 0.00018868666742475346,
      "loss": 7.4005,
      "step": 131440
    },
    {
      "epoch": 15.818291215403129,
      "grad_norm": 962.4224853515625,
      "learning_rate": 0.00018868490969115812,
      "loss": 7.2904,
      "step": 131450
    },
    {
      "epoch": 15.819494584837544,
      "grad_norm": 473.9652099609375,
      "learning_rate": 0.00018868315182921382,
      "loss": 7.2729,
      "step": 131460
    },
    {
      "epoch": 15.820697954271962,
      "grad_norm": 574.8343505859375,
      "learning_rate": 0.0001886813938389231,
      "loss": 7.2934,
      "step": 131470
    },
    {
      "epoch": 15.821901323706378,
      "grad_norm": 1261.2105712890625,
      "learning_rate": 0.00018867963572028848,
      "loss": 7.3167,
      "step": 131480
    },
    {
      "epoch": 15.823104693140793,
      "grad_norm": 284.2709655761719,
      "learning_rate": 0.00018867787747331257,
      "loss": 7.3167,
      "step": 131490
    },
    {
      "epoch": 15.824308062575211,
      "grad_norm": 543.37841796875,
      "learning_rate": 0.00018867611909799786,
      "loss": 7.3328,
      "step": 131500
    },
    {
      "epoch": 15.825511432009627,
      "grad_norm": 600.607666015625,
      "learning_rate": 0.0001886743605943469,
      "loss": 7.2612,
      "step": 131510
    },
    {
      "epoch": 15.826714801444043,
      "grad_norm": 246.95547485351562,
      "learning_rate": 0.00018867260196236228,
      "loss": 7.3524,
      "step": 131520
    },
    {
      "epoch": 15.82791817087846,
      "grad_norm": 356.9399719238281,
      "learning_rate": 0.00018867084320204646,
      "loss": 7.296,
      "step": 131530
    },
    {
      "epoch": 15.829121540312876,
      "grad_norm": 476.8534240722656,
      "learning_rate": 0.00018866908431340208,
      "loss": 7.3902,
      "step": 131540
    },
    {
      "epoch": 15.830324909747292,
      "grad_norm": 718.119140625,
      "learning_rate": 0.00018866732529643164,
      "loss": 7.3336,
      "step": 131550
    },
    {
      "epoch": 15.83152827918171,
      "grad_norm": 462.7386474609375,
      "learning_rate": 0.00018866556615113766,
      "loss": 7.3429,
      "step": 131560
    },
    {
      "epoch": 15.832731648616125,
      "grad_norm": 726.5955200195312,
      "learning_rate": 0.00018866380687752274,
      "loss": 7.2506,
      "step": 131570
    },
    {
      "epoch": 15.833935018050541,
      "grad_norm": 323.6458435058594,
      "learning_rate": 0.0001886620474755894,
      "loss": 7.2381,
      "step": 131580
    },
    {
      "epoch": 15.835138387484958,
      "grad_norm": 387.4329833984375,
      "learning_rate": 0.00018866028794534018,
      "loss": 7.3101,
      "step": 131590
    },
    {
      "epoch": 15.836341756919374,
      "grad_norm": 296.24725341796875,
      "learning_rate": 0.00018865852828677765,
      "loss": 7.2715,
      "step": 131600
    },
    {
      "epoch": 15.83754512635379,
      "grad_norm": 454.244140625,
      "learning_rate": 0.00018865676849990433,
      "loss": 7.268,
      "step": 131610
    },
    {
      "epoch": 15.838748495788208,
      "grad_norm": 597.741943359375,
      "learning_rate": 0.00018865500858472278,
      "loss": 7.3178,
      "step": 131620
    },
    {
      "epoch": 15.839951865222623,
      "grad_norm": 1367.4615478515625,
      "learning_rate": 0.00018865324854123552,
      "loss": 7.3464,
      "step": 131630
    },
    {
      "epoch": 15.84115523465704,
      "grad_norm": 437.6692199707031,
      "learning_rate": 0.00018865148836944514,
      "loss": 7.367,
      "step": 131640
    },
    {
      "epoch": 15.842358604091457,
      "grad_norm": 985.8402099609375,
      "learning_rate": 0.00018864972806935417,
      "loss": 7.3601,
      "step": 131650
    },
    {
      "epoch": 15.843561973525873,
      "grad_norm": 2324.87744140625,
      "learning_rate": 0.00018864796764096517,
      "loss": 7.4486,
      "step": 131660
    },
    {
      "epoch": 15.844765342960288,
      "grad_norm": 389.7665100097656,
      "learning_rate": 0.00018864620708428068,
      "loss": 7.3407,
      "step": 131670
    },
    {
      "epoch": 15.845968712394706,
      "grad_norm": 615.8221435546875,
      "learning_rate": 0.0001886444463993032,
      "loss": 7.3344,
      "step": 131680
    },
    {
      "epoch": 15.847172081829122,
      "grad_norm": 5037.34521484375,
      "learning_rate": 0.00018864268558603536,
      "loss": 7.3313,
      "step": 131690
    },
    {
      "epoch": 15.848375451263538,
      "grad_norm": 1342.1685791015625,
      "learning_rate": 0.00018864092464447967,
      "loss": 7.3743,
      "step": 131700
    },
    {
      "epoch": 15.849578820697955,
      "grad_norm": 1030.5870361328125,
      "learning_rate": 0.00018863916357463864,
      "loss": 7.4617,
      "step": 131710
    },
    {
      "epoch": 15.85078219013237,
      "grad_norm": 529.4223022460938,
      "learning_rate": 0.00018863740237651488,
      "loss": 7.3907,
      "step": 131720
    },
    {
      "epoch": 15.851985559566787,
      "grad_norm": 1138.220947265625,
      "learning_rate": 0.0001886356410501109,
      "loss": 7.494,
      "step": 131730
    },
    {
      "epoch": 15.853188929001202,
      "grad_norm": 851.51904296875,
      "learning_rate": 0.0001886338795954293,
      "loss": 7.4549,
      "step": 131740
    },
    {
      "epoch": 15.85439229843562,
      "grad_norm": 2602.716796875,
      "learning_rate": 0.00018863211801247256,
      "loss": 7.4268,
      "step": 131750
    },
    {
      "epoch": 15.855595667870036,
      "grad_norm": 1488.7198486328125,
      "learning_rate": 0.00018863035630124327,
      "loss": 7.432,
      "step": 131760
    },
    {
      "epoch": 15.856799037304452,
      "grad_norm": 1121.366943359375,
      "learning_rate": 0.00018862859446174397,
      "loss": 7.4475,
      "step": 131770
    },
    {
      "epoch": 15.85800240673887,
      "grad_norm": 3571.143798828125,
      "learning_rate": 0.00018862683249397722,
      "loss": 7.5597,
      "step": 131780
    },
    {
      "epoch": 15.859205776173285,
      "grad_norm": 676.1463012695312,
      "learning_rate": 0.00018862507039794557,
      "loss": 7.4271,
      "step": 131790
    },
    {
      "epoch": 15.8604091456077,
      "grad_norm": 1601.666748046875,
      "learning_rate": 0.00018862330817365151,
      "loss": 7.4623,
      "step": 131800
    },
    {
      "epoch": 15.861612515042118,
      "grad_norm": 1726.900634765625,
      "learning_rate": 0.00018862154582109772,
      "loss": 7.4037,
      "step": 131810
    },
    {
      "epoch": 15.862815884476534,
      "grad_norm": 1261.962890625,
      "learning_rate": 0.00018861978334028663,
      "loss": 7.4488,
      "step": 131820
    },
    {
      "epoch": 15.86401925391095,
      "grad_norm": 1210.049072265625,
      "learning_rate": 0.00018861802073122085,
      "loss": 7.4747,
      "step": 131830
    },
    {
      "epoch": 15.865222623345367,
      "grad_norm": 1083.919189453125,
      "learning_rate": 0.0001886162579939029,
      "loss": 7.4403,
      "step": 131840
    },
    {
      "epoch": 15.866425992779783,
      "grad_norm": 1613.10986328125,
      "learning_rate": 0.00018861449512833534,
      "loss": 7.5355,
      "step": 131850
    },
    {
      "epoch": 15.867629362214199,
      "grad_norm": 987.13037109375,
      "learning_rate": 0.00018861273213452076,
      "loss": 7.4347,
      "step": 131860
    },
    {
      "epoch": 15.868832731648617,
      "grad_norm": 601.0751342773438,
      "learning_rate": 0.00018861096901246166,
      "loss": 7.5164,
      "step": 131870
    },
    {
      "epoch": 15.870036101083032,
      "grad_norm": 685.5590209960938,
      "learning_rate": 0.0001886092057621606,
      "loss": 7.4458,
      "step": 131880
    },
    {
      "epoch": 15.871239470517448,
      "grad_norm": 419.60888671875,
      "learning_rate": 0.00018860744238362014,
      "loss": 7.3614,
      "step": 131890
    },
    {
      "epoch": 15.872442839951866,
      "grad_norm": 2338.96533203125,
      "learning_rate": 0.00018860567887684284,
      "loss": 7.3878,
      "step": 131900
    },
    {
      "epoch": 15.873646209386282,
      "grad_norm": 510.9063415527344,
      "learning_rate": 0.00018860391524183124,
      "loss": 7.3111,
      "step": 131910
    },
    {
      "epoch": 15.874849578820697,
      "grad_norm": 800.651611328125,
      "learning_rate": 0.00018860215147858793,
      "loss": 7.4052,
      "step": 131920
    },
    {
      "epoch": 15.876052948255115,
      "grad_norm": 814.5459594726562,
      "learning_rate": 0.0001886003875871154,
      "loss": 7.3182,
      "step": 131930
    },
    {
      "epoch": 15.87725631768953,
      "grad_norm": 552.2872314453125,
      "learning_rate": 0.00018859862356741623,
      "loss": 7.416,
      "step": 131940
    },
    {
      "epoch": 15.878459687123947,
      "grad_norm": 1093.0543212890625,
      "learning_rate": 0.00018859685941949303,
      "loss": 7.3964,
      "step": 131950
    },
    {
      "epoch": 15.879663056558364,
      "grad_norm": 928.7797241210938,
      "learning_rate": 0.00018859509514334826,
      "loss": 7.3422,
      "step": 131960
    },
    {
      "epoch": 15.88086642599278,
      "grad_norm": 1768.70849609375,
      "learning_rate": 0.00018859333073898451,
      "loss": 7.342,
      "step": 131970
    },
    {
      "epoch": 15.882069795427196,
      "grad_norm": 736.9295043945312,
      "learning_rate": 0.00018859156620640435,
      "loss": 7.3474,
      "step": 131980
    },
    {
      "epoch": 15.883273164861613,
      "grad_norm": 869.2301025390625,
      "learning_rate": 0.00018858980154561031,
      "loss": 7.4479,
      "step": 131990
    },
    {
      "epoch": 15.884476534296029,
      "grad_norm": 2636.7568359375,
      "learning_rate": 0.00018858803675660498,
      "loss": 7.3574,
      "step": 132000
    },
    {
      "epoch": 15.885679903730445,
      "grad_norm": 2663.193359375,
      "learning_rate": 0.00018858627183939087,
      "loss": 7.4161,
      "step": 132010
    },
    {
      "epoch": 15.886883273164862,
      "grad_norm": 742.95166015625,
      "learning_rate": 0.00018858450679397053,
      "loss": 7.3578,
      "step": 132020
    },
    {
      "epoch": 15.888086642599278,
      "grad_norm": 1010.35009765625,
      "learning_rate": 0.00018858274162034656,
      "loss": 7.3829,
      "step": 132030
    },
    {
      "epoch": 15.889290012033694,
      "grad_norm": 1248.4332275390625,
      "learning_rate": 0.0001885809763185215,
      "loss": 7.3209,
      "step": 132040
    },
    {
      "epoch": 15.890493381468112,
      "grad_norm": 1663.683837890625,
      "learning_rate": 0.0001885792108884979,
      "loss": 7.3999,
      "step": 132050
    },
    {
      "epoch": 15.891696750902527,
      "grad_norm": 973.1458740234375,
      "learning_rate": 0.0001885774453302783,
      "loss": 7.4251,
      "step": 132060
    },
    {
      "epoch": 15.892900120336943,
      "grad_norm": 896.4718627929688,
      "learning_rate": 0.00018857567964386526,
      "loss": 7.4159,
      "step": 132070
    },
    {
      "epoch": 15.89410348977136,
      "grad_norm": 1221.7650146484375,
      "learning_rate": 0.00018857391382926138,
      "loss": 7.2595,
      "step": 132080
    },
    {
      "epoch": 15.895306859205776,
      "grad_norm": 449.065185546875,
      "learning_rate": 0.00018857214788646915,
      "loss": 7.2464,
      "step": 132090
    },
    {
      "epoch": 15.896510228640192,
      "grad_norm": 703.9978637695312,
      "learning_rate": 0.00018857038181549116,
      "loss": 7.3952,
      "step": 132100
    },
    {
      "epoch": 15.89771359807461,
      "grad_norm": 903.7862548828125,
      "learning_rate": 0.00018856861561632995,
      "loss": 7.3301,
      "step": 132110
    },
    {
      "epoch": 15.898916967509026,
      "grad_norm": 1145.3795166015625,
      "learning_rate": 0.00018856684928898808,
      "loss": 7.2957,
      "step": 132120
    },
    {
      "epoch": 15.900120336943441,
      "grad_norm": 827.79736328125,
      "learning_rate": 0.00018856508283346813,
      "loss": 7.3222,
      "step": 132130
    },
    {
      "epoch": 15.901323706377857,
      "grad_norm": 1203.126220703125,
      "learning_rate": 0.00018856331624977264,
      "loss": 7.3951,
      "step": 132140
    },
    {
      "epoch": 15.902527075812275,
      "grad_norm": 1375.4981689453125,
      "learning_rate": 0.00018856154953790418,
      "loss": 7.4578,
      "step": 132150
    },
    {
      "epoch": 15.90373044524669,
      "grad_norm": 860.630126953125,
      "learning_rate": 0.00018855978269786527,
      "loss": 7.1981,
      "step": 132160
    },
    {
      "epoch": 15.904933814681106,
      "grad_norm": 597.31298828125,
      "learning_rate": 0.0001885580157296585,
      "loss": 7.2839,
      "step": 132170
    },
    {
      "epoch": 15.906137184115524,
      "grad_norm": 977.681396484375,
      "learning_rate": 0.0001885562486332864,
      "loss": 7.3767,
      "step": 132180
    },
    {
      "epoch": 15.90734055354994,
      "grad_norm": 1293.9266357421875,
      "learning_rate": 0.00018855448140875155,
      "loss": 7.2727,
      "step": 132190
    },
    {
      "epoch": 15.908543922984355,
      "grad_norm": 1996.4329833984375,
      "learning_rate": 0.00018855271405605653,
      "loss": 7.3439,
      "step": 132200
    },
    {
      "epoch": 15.909747292418773,
      "grad_norm": 1326.6815185546875,
      "learning_rate": 0.00018855094657520383,
      "loss": 7.3878,
      "step": 132210
    },
    {
      "epoch": 15.910950661853189,
      "grad_norm": 683.3563842773438,
      "learning_rate": 0.00018854917896619607,
      "loss": 7.4467,
      "step": 132220
    },
    {
      "epoch": 15.912154031287605,
      "grad_norm": 926.5361328125,
      "learning_rate": 0.00018854741122903578,
      "loss": 7.43,
      "step": 132230
    },
    {
      "epoch": 15.913357400722022,
      "grad_norm": 657.5689086914062,
      "learning_rate": 0.00018854564336372553,
      "loss": 7.2433,
      "step": 132240
    },
    {
      "epoch": 15.914560770156438,
      "grad_norm": 1846.4720458984375,
      "learning_rate": 0.00018854387537026786,
      "loss": 7.3929,
      "step": 132250
    },
    {
      "epoch": 15.915764139590854,
      "grad_norm": 900.9047241210938,
      "learning_rate": 0.00018854210724866536,
      "loss": 7.4495,
      "step": 132260
    },
    {
      "epoch": 15.916967509025271,
      "grad_norm": 1205.8076171875,
      "learning_rate": 0.00018854033899892055,
      "loss": 7.4043,
      "step": 132270
    },
    {
      "epoch": 15.918170878459687,
      "grad_norm": 572.4059448242188,
      "learning_rate": 0.00018853857062103605,
      "loss": 7.3304,
      "step": 132280
    },
    {
      "epoch": 15.919374247894103,
      "grad_norm": 760.0726318359375,
      "learning_rate": 0.00018853680211501433,
      "loss": 7.4405,
      "step": 132290
    },
    {
      "epoch": 15.92057761732852,
      "grad_norm": 1020.7083740234375,
      "learning_rate": 0.00018853503348085802,
      "loss": 7.3819,
      "step": 132300
    },
    {
      "epoch": 15.921780986762936,
      "grad_norm": 627.8305053710938,
      "learning_rate": 0.00018853326471856963,
      "loss": 7.3409,
      "step": 132310
    },
    {
      "epoch": 15.922984356197352,
      "grad_norm": 643.7003173828125,
      "learning_rate": 0.00018853149582815179,
      "loss": 7.3013,
      "step": 132320
    },
    {
      "epoch": 15.92418772563177,
      "grad_norm": 790.8099365234375,
      "learning_rate": 0.00018852972680960702,
      "loss": 7.3324,
      "step": 132330
    },
    {
      "epoch": 15.925391095066185,
      "grad_norm": 565.606201171875,
      "learning_rate": 0.00018852795766293785,
      "loss": 7.31,
      "step": 132340
    },
    {
      "epoch": 15.926594464500601,
      "grad_norm": 553.124267578125,
      "learning_rate": 0.00018852618838814688,
      "loss": 7.3832,
      "step": 132350
    },
    {
      "epoch": 15.927797833935019,
      "grad_norm": 779.17041015625,
      "learning_rate": 0.00018852441898523664,
      "loss": 7.3317,
      "step": 132360
    },
    {
      "epoch": 15.929001203369435,
      "grad_norm": 894.0552368164062,
      "learning_rate": 0.00018852264945420975,
      "loss": 7.2975,
      "step": 132370
    },
    {
      "epoch": 15.93020457280385,
      "grad_norm": 375.84625244140625,
      "learning_rate": 0.00018852087979506868,
      "loss": 7.2296,
      "step": 132380
    },
    {
      "epoch": 15.931407942238268,
      "grad_norm": 917.9942016601562,
      "learning_rate": 0.0001885191100078161,
      "loss": 7.2822,
      "step": 132390
    },
    {
      "epoch": 15.932611311672684,
      "grad_norm": 2386.097412109375,
      "learning_rate": 0.00018851734009245447,
      "loss": 7.2722,
      "step": 132400
    },
    {
      "epoch": 15.9338146811071,
      "grad_norm": 698.42333984375,
      "learning_rate": 0.00018851557004898642,
      "loss": 7.3083,
      "step": 132410
    },
    {
      "epoch": 15.935018050541515,
      "grad_norm": 1420.1732177734375,
      "learning_rate": 0.00018851379987741446,
      "loss": 7.3846,
      "step": 132420
    },
    {
      "epoch": 15.936221419975933,
      "grad_norm": 1176.1473388671875,
      "learning_rate": 0.00018851202957774116,
      "loss": 7.2335,
      "step": 132430
    },
    {
      "epoch": 15.937424789410349,
      "grad_norm": 1198.45166015625,
      "learning_rate": 0.00018851025914996914,
      "loss": 7.3123,
      "step": 132440
    },
    {
      "epoch": 15.938628158844764,
      "grad_norm": 788.9474487304688,
      "learning_rate": 0.00018850848859410093,
      "loss": 7.2493,
      "step": 132450
    },
    {
      "epoch": 15.939831528279182,
      "grad_norm": 759.3018188476562,
      "learning_rate": 0.00018850671791013907,
      "loss": 7.3103,
      "step": 132460
    },
    {
      "epoch": 15.941034897713598,
      "grad_norm": 1486.5074462890625,
      "learning_rate": 0.00018850494709808614,
      "loss": 7.2904,
      "step": 132470
    },
    {
      "epoch": 15.942238267148014,
      "grad_norm": 1557.888916015625,
      "learning_rate": 0.00018850317615794467,
      "loss": 7.2453,
      "step": 132480
    },
    {
      "epoch": 15.943441636582431,
      "grad_norm": 383.0955810546875,
      "learning_rate": 0.0001885014050897173,
      "loss": 7.3304,
      "step": 132490
    },
    {
      "epoch": 15.944645006016847,
      "grad_norm": 1168.932373046875,
      "learning_rate": 0.00018849963389340653,
      "loss": 7.4245,
      "step": 132500
    },
    {
      "epoch": 15.945848375451263,
      "grad_norm": 783.9414672851562,
      "learning_rate": 0.00018849786256901495,
      "loss": 7.324,
      "step": 132510
    },
    {
      "epoch": 15.94705174488568,
      "grad_norm": 1027.5152587890625,
      "learning_rate": 0.00018849609111654507,
      "loss": 7.3045,
      "step": 132520
    },
    {
      "epoch": 15.948255114320096,
      "grad_norm": 1178.9649658203125,
      "learning_rate": 0.00018849431953599954,
      "loss": 7.4402,
      "step": 132530
    },
    {
      "epoch": 15.949458483754512,
      "grad_norm": 1348.3472900390625,
      "learning_rate": 0.00018849254782738085,
      "loss": 7.3948,
      "step": 132540
    },
    {
      "epoch": 15.95066185318893,
      "grad_norm": 3098.52587890625,
      "learning_rate": 0.00018849077599069163,
      "loss": 7.2963,
      "step": 132550
    },
    {
      "epoch": 15.951865222623345,
      "grad_norm": 605.2205810546875,
      "learning_rate": 0.0001884890040259344,
      "loss": 7.211,
      "step": 132560
    },
    {
      "epoch": 15.953068592057761,
      "grad_norm": 1400.8870849609375,
      "learning_rate": 0.00018848723193311173,
      "loss": 7.233,
      "step": 132570
    },
    {
      "epoch": 15.954271961492179,
      "grad_norm": 2605.5439453125,
      "learning_rate": 0.00018848545971222617,
      "loss": 7.3758,
      "step": 132580
    },
    {
      "epoch": 15.955475330926594,
      "grad_norm": 981.1021728515625,
      "learning_rate": 0.00018848368736328034,
      "loss": 7.372,
      "step": 132590
    },
    {
      "epoch": 15.95667870036101,
      "grad_norm": 1613.8927001953125,
      "learning_rate": 0.00018848191488627674,
      "loss": 7.3922,
      "step": 132600
    },
    {
      "epoch": 15.957882069795428,
      "grad_norm": 537.8240356445312,
      "learning_rate": 0.00018848014228121796,
      "loss": 7.3488,
      "step": 132610
    },
    {
      "epoch": 15.959085439229844,
      "grad_norm": 1165.123046875,
      "learning_rate": 0.0001884783695481066,
      "loss": 7.2414,
      "step": 132620
    },
    {
      "epoch": 15.96028880866426,
      "grad_norm": 914.7545166015625,
      "learning_rate": 0.00018847659668694516,
      "loss": 7.2546,
      "step": 132630
    },
    {
      "epoch": 15.961492178098677,
      "grad_norm": 686.739501953125,
      "learning_rate": 0.00018847482369773626,
      "loss": 7.3288,
      "step": 132640
    },
    {
      "epoch": 15.962695547533093,
      "grad_norm": 449.0668640136719,
      "learning_rate": 0.00018847305058048244,
      "loss": 7.4283,
      "step": 132650
    },
    {
      "epoch": 15.963898916967509,
      "grad_norm": 449.6669921875,
      "learning_rate": 0.00018847127733518626,
      "loss": 7.2542,
      "step": 132660
    },
    {
      "epoch": 15.965102286401926,
      "grad_norm": 1582.1160888671875,
      "learning_rate": 0.00018846950396185033,
      "loss": 7.3047,
      "step": 132670
    },
    {
      "epoch": 15.966305655836342,
      "grad_norm": 729.3178100585938,
      "learning_rate": 0.00018846773046047718,
      "loss": 7.4101,
      "step": 132680
    },
    {
      "epoch": 15.967509025270758,
      "grad_norm": 610.6448974609375,
      "learning_rate": 0.00018846595683106937,
      "loss": 7.3058,
      "step": 132690
    },
    {
      "epoch": 15.968712394705175,
      "grad_norm": 852.5591430664062,
      "learning_rate": 0.00018846418307362947,
      "loss": 7.2494,
      "step": 132700
    },
    {
      "epoch": 15.969915764139591,
      "grad_norm": 894.6786499023438,
      "learning_rate": 0.00018846240918816008,
      "loss": 7.3286,
      "step": 132710
    },
    {
      "epoch": 15.971119133574007,
      "grad_norm": 933.6132202148438,
      "learning_rate": 0.0001884606351746637,
      "loss": 7.3031,
      "step": 132720
    },
    {
      "epoch": 15.972322503008424,
      "grad_norm": 1170.6556396484375,
      "learning_rate": 0.000188458861033143,
      "loss": 7.3357,
      "step": 132730
    },
    {
      "epoch": 15.97352587244284,
      "grad_norm": 695.614013671875,
      "learning_rate": 0.00018845708676360043,
      "loss": 7.353,
      "step": 132740
    },
    {
      "epoch": 15.974729241877256,
      "grad_norm": 1078.06298828125,
      "learning_rate": 0.00018845531236603863,
      "loss": 7.2228,
      "step": 132750
    },
    {
      "epoch": 15.975932611311674,
      "grad_norm": 1467.5294189453125,
      "learning_rate": 0.00018845353784046018,
      "loss": 7.2914,
      "step": 132760
    },
    {
      "epoch": 15.97713598074609,
      "grad_norm": 650.1083984375,
      "learning_rate": 0.00018845176318686761,
      "loss": 7.339,
      "step": 132770
    },
    {
      "epoch": 15.978339350180505,
      "grad_norm": 566.7533569335938,
      "learning_rate": 0.00018844998840526352,
      "loss": 7.2865,
      "step": 132780
    },
    {
      "epoch": 15.979542719614923,
      "grad_norm": 524.9175415039062,
      "learning_rate": 0.00018844821349565043,
      "loss": 7.2933,
      "step": 132790
    },
    {
      "epoch": 15.980746089049338,
      "grad_norm": 1949.8116455078125,
      "learning_rate": 0.00018844643845803094,
      "loss": 7.2193,
      "step": 132800
    },
    {
      "epoch": 15.981949458483754,
      "grad_norm": 491.0684509277344,
      "learning_rate": 0.00018844466329240764,
      "loss": 7.3184,
      "step": 132810
    },
    {
      "epoch": 15.98315282791817,
      "grad_norm": 1322.3477783203125,
      "learning_rate": 0.00018844288799878305,
      "loss": 7.2984,
      "step": 132820
    },
    {
      "epoch": 15.984356197352588,
      "grad_norm": 1918.1973876953125,
      "learning_rate": 0.00018844111257715977,
      "loss": 7.341,
      "step": 132830
    },
    {
      "epoch": 15.985559566787003,
      "grad_norm": 690.3482666015625,
      "learning_rate": 0.0001884393370275404,
      "loss": 7.3013,
      "step": 132840
    },
    {
      "epoch": 15.98676293622142,
      "grad_norm": 2622.135986328125,
      "learning_rate": 0.0001884375613499274,
      "loss": 7.2816,
      "step": 132850
    },
    {
      "epoch": 15.987966305655837,
      "grad_norm": 849.4652099609375,
      "learning_rate": 0.00018843578554432347,
      "loss": 7.3266,
      "step": 132860
    },
    {
      "epoch": 15.989169675090253,
      "grad_norm": 673.3524780273438,
      "learning_rate": 0.00018843400961073112,
      "loss": 7.2121,
      "step": 132870
    },
    {
      "epoch": 15.990373044524668,
      "grad_norm": 1238.9793701171875,
      "learning_rate": 0.00018843223354915293,
      "loss": 7.3043,
      "step": 132880
    },
    {
      "epoch": 15.991576413959086,
      "grad_norm": 964.6995239257812,
      "learning_rate": 0.00018843045735959143,
      "loss": 7.3414,
      "step": 132890
    },
    {
      "epoch": 15.992779783393502,
      "grad_norm": 1358.2208251953125,
      "learning_rate": 0.00018842868104204926,
      "loss": 7.2966,
      "step": 132900
    },
    {
      "epoch": 15.993983152827917,
      "grad_norm": 1238.5880126953125,
      "learning_rate": 0.00018842690459652895,
      "loss": 7.3556,
      "step": 132910
    },
    {
      "epoch": 15.995186522262335,
      "grad_norm": 753.2740478515625,
      "learning_rate": 0.00018842512802303308,
      "loss": 7.3793,
      "step": 132920
    },
    {
      "epoch": 15.99638989169675,
      "grad_norm": 3753.060546875,
      "learning_rate": 0.00018842335132156422,
      "loss": 7.3637,
      "step": 132930
    },
    {
      "epoch": 15.997593261131167,
      "grad_norm": 1965.354248046875,
      "learning_rate": 0.0001884215744921249,
      "loss": 7.4331,
      "step": 132940
    },
    {
      "epoch": 15.998796630565584,
      "grad_norm": 4114.978515625,
      "learning_rate": 0.00018841979753471776,
      "loss": 7.4812,
      "step": 132950
    },
    {
      "epoch": 16.0,
      "grad_norm": 2308.511962890625,
      "learning_rate": 0.00018841802044934535,
      "loss": 7.5405,
      "step": 132960
    },
    {
      "epoch": 16.0,
      "eval_loss": 7.461447238922119,
      "eval_runtime": 121.0916,
      "eval_samples_per_second": 61.003,
      "eval_steps_per_second": 7.631,
      "step": 132960
    },
    {
      "epoch": 16.001203369434418,
      "grad_norm": 2153.32568359375,
      "learning_rate": 0.00018841624323601022,
      "loss": 7.4752,
      "step": 132970
    },
    {
      "epoch": 16.00240673886883,
      "grad_norm": 889.717041015625,
      "learning_rate": 0.00018841446589471498,
      "loss": 7.4672,
      "step": 132980
    },
    {
      "epoch": 16.00361010830325,
      "grad_norm": 962.61572265625,
      "learning_rate": 0.00018841268842546217,
      "loss": 7.399,
      "step": 132990
    },
    {
      "epoch": 16.004813477737667,
      "grad_norm": 1718.97802734375,
      "learning_rate": 0.00018841091082825434,
      "loss": 7.3364,
      "step": 133000
    },
    {
      "epoch": 16.00601684717208,
      "grad_norm": 1896.02978515625,
      "learning_rate": 0.00018840913310309412,
      "loss": 7.3895,
      "step": 133010
    },
    {
      "epoch": 16.0072202166065,
      "grad_norm": 1826.3760986328125,
      "learning_rate": 0.00018840735524998405,
      "loss": 7.4432,
      "step": 133020
    },
    {
      "epoch": 16.008423586040916,
      "grad_norm": 1622.966552734375,
      "learning_rate": 0.00018840557726892675,
      "loss": 7.4268,
      "step": 133030
    },
    {
      "epoch": 16.00962695547533,
      "grad_norm": 3895.788330078125,
      "learning_rate": 0.0001884037991599247,
      "loss": 7.4765,
      "step": 133040
    },
    {
      "epoch": 16.010830324909747,
      "grad_norm": 4022.718994140625,
      "learning_rate": 0.00018840202092298055,
      "loss": 7.4086,
      "step": 133050
    },
    {
      "epoch": 16.012033694344165,
      "grad_norm": 2445.8984375,
      "learning_rate": 0.00018840024255809684,
      "loss": 7.4304,
      "step": 133060
    },
    {
      "epoch": 16.01323706377858,
      "grad_norm": 1398.728271484375,
      "learning_rate": 0.00018839846406527614,
      "loss": 7.3882,
      "step": 133070
    },
    {
      "epoch": 16.014440433212997,
      "grad_norm": 1955.9288330078125,
      "learning_rate": 0.00018839668544452107,
      "loss": 7.5115,
      "step": 133080
    },
    {
      "epoch": 16.015643802647414,
      "grad_norm": 1860.2630615234375,
      "learning_rate": 0.00018839490669583418,
      "loss": 7.383,
      "step": 133090
    },
    {
      "epoch": 16.016847172081828,
      "grad_norm": 4438.74755859375,
      "learning_rate": 0.00018839312781921802,
      "loss": 7.4305,
      "step": 133100
    },
    {
      "epoch": 16.018050541516246,
      "grad_norm": 2396.3515625,
      "learning_rate": 0.00018839134881467516,
      "loss": 7.5489,
      "step": 133110
    },
    {
      "epoch": 16.019253910950663,
      "grad_norm": 1607.8743896484375,
      "learning_rate": 0.0001883895696822082,
      "loss": 7.43,
      "step": 133120
    },
    {
      "epoch": 16.020457280385077,
      "grad_norm": 1343.34521484375,
      "learning_rate": 0.00018838779042181972,
      "loss": 7.4046,
      "step": 133130
    },
    {
      "epoch": 16.021660649819495,
      "grad_norm": 1568.716552734375,
      "learning_rate": 0.0001883860110335123,
      "loss": 7.4269,
      "step": 133140
    },
    {
      "epoch": 16.022864019253912,
      "grad_norm": 1527.429443359375,
      "learning_rate": 0.00018838423151728849,
      "loss": 7.4202,
      "step": 133150
    },
    {
      "epoch": 16.024067388688326,
      "grad_norm": 2880.157958984375,
      "learning_rate": 0.00018838245187315086,
      "loss": 7.5203,
      "step": 133160
    },
    {
      "epoch": 16.025270758122744,
      "grad_norm": 1841.1431884765625,
      "learning_rate": 0.00018838067210110202,
      "loss": 7.3747,
      "step": 133170
    },
    {
      "epoch": 16.02647412755716,
      "grad_norm": 2396.6181640625,
      "learning_rate": 0.0001883788922011445,
      "loss": 7.4847,
      "step": 133180
    },
    {
      "epoch": 16.027677496991576,
      "grad_norm": 1560.1546630859375,
      "learning_rate": 0.00018837711217328093,
      "loss": 7.5355,
      "step": 133190
    },
    {
      "epoch": 16.028880866425993,
      "grad_norm": 2516.204345703125,
      "learning_rate": 0.00018837533201751386,
      "loss": 7.4528,
      "step": 133200
    },
    {
      "epoch": 16.03008423586041,
      "grad_norm": 2364.503662109375,
      "learning_rate": 0.00018837355173384585,
      "loss": 7.382,
      "step": 133210
    },
    {
      "epoch": 16.031287605294825,
      "grad_norm": 2570.381103515625,
      "learning_rate": 0.0001883717713222795,
      "loss": 7.4052,
      "step": 133220
    },
    {
      "epoch": 16.032490974729242,
      "grad_norm": 1330.7001953125,
      "learning_rate": 0.00018836999078281736,
      "loss": 7.5175,
      "step": 133230
    },
    {
      "epoch": 16.03369434416366,
      "grad_norm": 1902.0858154296875,
      "learning_rate": 0.00018836821011546207,
      "loss": 7.381,
      "step": 133240
    },
    {
      "epoch": 16.034897713598074,
      "grad_norm": 3191.408447265625,
      "learning_rate": 0.00018836642932021613,
      "loss": 7.4846,
      "step": 133250
    },
    {
      "epoch": 16.03610108303249,
      "grad_norm": 1880.0172119140625,
      "learning_rate": 0.00018836464839708215,
      "loss": 7.329,
      "step": 133260
    },
    {
      "epoch": 16.03730445246691,
      "grad_norm": 2109.16162109375,
      "learning_rate": 0.00018836286734606273,
      "loss": 7.344,
      "step": 133270
    },
    {
      "epoch": 16.038507821901323,
      "grad_norm": 1009.3477172851562,
      "learning_rate": 0.00018836108616716038,
      "loss": 7.4741,
      "step": 133280
    },
    {
      "epoch": 16.03971119133574,
      "grad_norm": 1296.6412353515625,
      "learning_rate": 0.00018835930486037775,
      "loss": 7.4144,
      "step": 133290
    },
    {
      "epoch": 16.040914560770158,
      "grad_norm": 2100.59033203125,
      "learning_rate": 0.0001883575234257174,
      "loss": 7.4383,
      "step": 133300
    },
    {
      "epoch": 16.042117930204572,
      "grad_norm": 1190.213623046875,
      "learning_rate": 0.00018835574186318188,
      "loss": 7.4075,
      "step": 133310
    },
    {
      "epoch": 16.04332129963899,
      "grad_norm": 687.1910400390625,
      "learning_rate": 0.0001883539601727738,
      "loss": 7.4325,
      "step": 133320
    },
    {
      "epoch": 16.044524669073404,
      "grad_norm": 1521.695068359375,
      "learning_rate": 0.00018835217835449572,
      "loss": 7.3603,
      "step": 133330
    },
    {
      "epoch": 16.04572803850782,
      "grad_norm": 1121.0152587890625,
      "learning_rate": 0.00018835039640835023,
      "loss": 7.3836,
      "step": 133340
    },
    {
      "epoch": 16.04693140794224,
      "grad_norm": 558.4403686523438,
      "learning_rate": 0.0001883486143343399,
      "loss": 7.3389,
      "step": 133350
    },
    {
      "epoch": 16.048134777376653,
      "grad_norm": 1114.6622314453125,
      "learning_rate": 0.00018834683213246733,
      "loss": 7.3147,
      "step": 133360
    },
    {
      "epoch": 16.04933814681107,
      "grad_norm": 1681.1033935546875,
      "learning_rate": 0.00018834504980273504,
      "loss": 7.345,
      "step": 133370
    },
    {
      "epoch": 16.050541516245488,
      "grad_norm": 1570.2076416015625,
      "learning_rate": 0.0001883432673451457,
      "loss": 7.3738,
      "step": 133380
    },
    {
      "epoch": 16.051744885679902,
      "grad_norm": 2327.41357421875,
      "learning_rate": 0.00018834148475970183,
      "loss": 7.4091,
      "step": 133390
    },
    {
      "epoch": 16.05294825511432,
      "grad_norm": 3199.50390625,
      "learning_rate": 0.000188339702046406,
      "loss": 7.4601,
      "step": 133400
    },
    {
      "epoch": 16.054151624548737,
      "grad_norm": 1199.08984375,
      "learning_rate": 0.0001883379192052608,
      "loss": 7.5126,
      "step": 133410
    },
    {
      "epoch": 16.05535499398315,
      "grad_norm": 1048.759765625,
      "learning_rate": 0.00018833613623626884,
      "loss": 7.4348,
      "step": 133420
    },
    {
      "epoch": 16.05655836341757,
      "grad_norm": 1395.2281494140625,
      "learning_rate": 0.0001883343531394327,
      "loss": 7.3279,
      "step": 133430
    },
    {
      "epoch": 16.057761732851986,
      "grad_norm": 1758.9266357421875,
      "learning_rate": 0.0001883325699147549,
      "loss": 7.3411,
      "step": 133440
    },
    {
      "epoch": 16.0589651022864,
      "grad_norm": 2204.992431640625,
      "learning_rate": 0.0001883307865622381,
      "loss": 7.4047,
      "step": 133450
    },
    {
      "epoch": 16.060168471720818,
      "grad_norm": 1060.8265380859375,
      "learning_rate": 0.0001883290030818848,
      "loss": 7.3041,
      "step": 133460
    },
    {
      "epoch": 16.061371841155236,
      "grad_norm": 597.0863037109375,
      "learning_rate": 0.00018832721947369767,
      "loss": 7.3442,
      "step": 133470
    },
    {
      "epoch": 16.06257521058965,
      "grad_norm": 988.7749633789062,
      "learning_rate": 0.0001883254357376792,
      "loss": 7.3666,
      "step": 133480
    },
    {
      "epoch": 16.063778580024067,
      "grad_norm": 953.1688232421875,
      "learning_rate": 0.00018832365187383207,
      "loss": 7.309,
      "step": 133490
    },
    {
      "epoch": 16.064981949458485,
      "grad_norm": 483.6090087890625,
      "learning_rate": 0.0001883218678821588,
      "loss": 7.3684,
      "step": 133500
    },
    {
      "epoch": 16.0661853188929,
      "grad_norm": 692.1166381835938,
      "learning_rate": 0.00018832008376266197,
      "loss": 7.3841,
      "step": 133510
    },
    {
      "epoch": 16.067388688327316,
      "grad_norm": 654.5657958984375,
      "learning_rate": 0.00018831829951534418,
      "loss": 7.2664,
      "step": 133520
    },
    {
      "epoch": 16.068592057761734,
      "grad_norm": 796.6243896484375,
      "learning_rate": 0.00018831651514020797,
      "loss": 7.295,
      "step": 133530
    },
    {
      "epoch": 16.069795427196148,
      "grad_norm": 1008.3521118164062,
      "learning_rate": 0.000188314730637256,
      "loss": 7.3321,
      "step": 133540
    },
    {
      "epoch": 16.070998796630565,
      "grad_norm": 1003.3748168945312,
      "learning_rate": 0.00018831294600649078,
      "loss": 7.231,
      "step": 133550
    },
    {
      "epoch": 16.072202166064983,
      "grad_norm": 1938.42724609375,
      "learning_rate": 0.00018831116124791496,
      "loss": 7.3993,
      "step": 133560
    },
    {
      "epoch": 16.073405535499397,
      "grad_norm": 797.2125854492188,
      "learning_rate": 0.00018830937636153105,
      "loss": 7.3484,
      "step": 133570
    },
    {
      "epoch": 16.074608904933815,
      "grad_norm": 1339.31689453125,
      "learning_rate": 0.00018830759134734167,
      "loss": 7.4835,
      "step": 133580
    },
    {
      "epoch": 16.075812274368232,
      "grad_norm": 989.5545654296875,
      "learning_rate": 0.00018830580620534946,
      "loss": 7.2828,
      "step": 133590
    },
    {
      "epoch": 16.077015643802646,
      "grad_norm": 4392.4541015625,
      "learning_rate": 0.00018830402093555687,
      "loss": 7.352,
      "step": 133600
    },
    {
      "epoch": 16.078219013237064,
      "grad_norm": 3030.227294921875,
      "learning_rate": 0.0001883022355379666,
      "loss": 7.2869,
      "step": 133610
    },
    {
      "epoch": 16.07942238267148,
      "grad_norm": 3117.56884765625,
      "learning_rate": 0.0001883004500125812,
      "loss": 7.3385,
      "step": 133620
    },
    {
      "epoch": 16.080625752105895,
      "grad_norm": 2173.417236328125,
      "learning_rate": 0.00018829866435940323,
      "loss": 7.3315,
      "step": 133630
    },
    {
      "epoch": 16.081829121540313,
      "grad_norm": 3442.108642578125,
      "learning_rate": 0.0001882968785784353,
      "loss": 7.2368,
      "step": 133640
    },
    {
      "epoch": 16.08303249097473,
      "grad_norm": 1160.810546875,
      "learning_rate": 0.00018829509266968,
      "loss": 7.3203,
      "step": 133650
    },
    {
      "epoch": 16.084235860409144,
      "grad_norm": 1445.635009765625,
      "learning_rate": 0.00018829330663313985,
      "loss": 7.1463,
      "step": 133660
    },
    {
      "epoch": 16.085439229843562,
      "grad_norm": 814.6618041992188,
      "learning_rate": 0.00018829152046881756,
      "loss": 7.3323,
      "step": 133670
    },
    {
      "epoch": 16.08664259927798,
      "grad_norm": 1322.2098388671875,
      "learning_rate": 0.00018828973417671558,
      "loss": 7.2867,
      "step": 133680
    },
    {
      "epoch": 16.087845968712394,
      "grad_norm": 1727.97314453125,
      "learning_rate": 0.0001882879477568366,
      "loss": 7.2514,
      "step": 133690
    },
    {
      "epoch": 16.08904933814681,
      "grad_norm": 1164.9935302734375,
      "learning_rate": 0.00018828616120918312,
      "loss": 7.3356,
      "step": 133700
    },
    {
      "epoch": 16.09025270758123,
      "grad_norm": 1758.5335693359375,
      "learning_rate": 0.0001882843745337578,
      "loss": 7.1983,
      "step": 133710
    },
    {
      "epoch": 16.091456077015643,
      "grad_norm": 915.6506958007812,
      "learning_rate": 0.00018828258773056318,
      "loss": 7.3296,
      "step": 133720
    },
    {
      "epoch": 16.09265944645006,
      "grad_norm": 1868.530517578125,
      "learning_rate": 0.00018828080079960188,
      "loss": 7.3555,
      "step": 133730
    },
    {
      "epoch": 16.093862815884478,
      "grad_norm": 1778.7662353515625,
      "learning_rate": 0.00018827901374087643,
      "loss": 7.2858,
      "step": 133740
    },
    {
      "epoch": 16.095066185318892,
      "grad_norm": 1313.4185791015625,
      "learning_rate": 0.0001882772265543895,
      "loss": 7.2891,
      "step": 133750
    },
    {
      "epoch": 16.09626955475331,
      "grad_norm": 1100.2462158203125,
      "learning_rate": 0.00018827543924014356,
      "loss": 7.2885,
      "step": 133760
    },
    {
      "epoch": 16.097472924187727,
      "grad_norm": 3137.552001953125,
      "learning_rate": 0.00018827365179814134,
      "loss": 7.4447,
      "step": 133770
    },
    {
      "epoch": 16.09867629362214,
      "grad_norm": 2594.281982421875,
      "learning_rate": 0.0001882718642283853,
      "loss": 7.3153,
      "step": 133780
    },
    {
      "epoch": 16.09987966305656,
      "grad_norm": 1250.010009765625,
      "learning_rate": 0.0001882700765308781,
      "loss": 7.3383,
      "step": 133790
    },
    {
      "epoch": 16.101083032490976,
      "grad_norm": 747.5228881835938,
      "learning_rate": 0.0001882682887056223,
      "loss": 7.3911,
      "step": 133800
    },
    {
      "epoch": 16.10228640192539,
      "grad_norm": 913.8907470703125,
      "learning_rate": 0.0001882665007526205,
      "loss": 7.2911,
      "step": 133810
    },
    {
      "epoch": 16.103489771359808,
      "grad_norm": 2426.7666015625,
      "learning_rate": 0.00018826471267187526,
      "loss": 7.4006,
      "step": 133820
    },
    {
      "epoch": 16.104693140794225,
      "grad_norm": 924.7741088867188,
      "learning_rate": 0.0001882629244633892,
      "loss": 7.375,
      "step": 133830
    },
    {
      "epoch": 16.10589651022864,
      "grad_norm": 1016.3461303710938,
      "learning_rate": 0.0001882611361271649,
      "loss": 7.3458,
      "step": 133840
    },
    {
      "epoch": 16.107099879663057,
      "grad_norm": 1101.467041015625,
      "learning_rate": 0.00018825934766320498,
      "loss": 7.2704,
      "step": 133850
    },
    {
      "epoch": 16.108303249097474,
      "grad_norm": 2128.217041015625,
      "learning_rate": 0.00018825755907151193,
      "loss": 7.3006,
      "step": 133860
    },
    {
      "epoch": 16.10950661853189,
      "grad_norm": 3141.464111328125,
      "learning_rate": 0.00018825577035208845,
      "loss": 7.4474,
      "step": 133870
    },
    {
      "epoch": 16.110709987966306,
      "grad_norm": 1540.0906982421875,
      "learning_rate": 0.00018825398150493705,
      "loss": 7.459,
      "step": 133880
    },
    {
      "epoch": 16.111913357400724,
      "grad_norm": 4305.9921875,
      "learning_rate": 0.00018825219253006036,
      "loss": 7.3037,
      "step": 133890
    },
    {
      "epoch": 16.113116726835138,
      "grad_norm": 3027.341552734375,
      "learning_rate": 0.00018825040342746097,
      "loss": 7.4396,
      "step": 133900
    },
    {
      "epoch": 16.114320096269555,
      "grad_norm": 8410.6162109375,
      "learning_rate": 0.00018824861419714143,
      "loss": 7.4259,
      "step": 133910
    },
    {
      "epoch": 16.115523465703973,
      "grad_norm": 6526.33056640625,
      "learning_rate": 0.0001882468248391044,
      "loss": 7.3705,
      "step": 133920
    },
    {
      "epoch": 16.116726835138387,
      "grad_norm": 2232.2724609375,
      "learning_rate": 0.00018824503535335238,
      "loss": 7.4946,
      "step": 133930
    },
    {
      "epoch": 16.117930204572804,
      "grad_norm": 3278.3681640625,
      "learning_rate": 0.00018824324573988804,
      "loss": 7.382,
      "step": 133940
    },
    {
      "epoch": 16.119133574007222,
      "grad_norm": 2583.0341796875,
      "learning_rate": 0.00018824145599871393,
      "loss": 7.4288,
      "step": 133950
    },
    {
      "epoch": 16.120336943441636,
      "grad_norm": 2351.92724609375,
      "learning_rate": 0.00018823966612983265,
      "loss": 7.3954,
      "step": 133960
    },
    {
      "epoch": 16.121540312876053,
      "grad_norm": 2978.618896484375,
      "learning_rate": 0.00018823787613324677,
      "loss": 7.3464,
      "step": 133970
    },
    {
      "epoch": 16.12274368231047,
      "grad_norm": 2466.6240234375,
      "learning_rate": 0.00018823608600895893,
      "loss": 7.3458,
      "step": 133980
    },
    {
      "epoch": 16.123947051744885,
      "grad_norm": 2047.6640625,
      "learning_rate": 0.00018823429575697163,
      "loss": 7.3782,
      "step": 133990
    },
    {
      "epoch": 16.125150421179303,
      "grad_norm": 2854.529052734375,
      "learning_rate": 0.00018823250537728758,
      "loss": 7.3129,
      "step": 134000
    },
    {
      "epoch": 16.126353790613717,
      "grad_norm": 4302.376953125,
      "learning_rate": 0.00018823071486990929,
      "loss": 7.4665,
      "step": 134010
    },
    {
      "epoch": 16.127557160048134,
      "grad_norm": 4303.015625,
      "learning_rate": 0.00018822892423483935,
      "loss": 7.3908,
      "step": 134020
    },
    {
      "epoch": 16.128760529482552,
      "grad_norm": 4873.20361328125,
      "learning_rate": 0.0001882271334720804,
      "loss": 7.4954,
      "step": 134030
    },
    {
      "epoch": 16.129963898916966,
      "grad_norm": 2545.898681640625,
      "learning_rate": 0.00018822534258163502,
      "loss": 7.4327,
      "step": 134040
    },
    {
      "epoch": 16.131167268351383,
      "grad_norm": 3953.999755859375,
      "learning_rate": 0.00018822355156350576,
      "loss": 7.4307,
      "step": 134050
    },
    {
      "epoch": 16.1323706377858,
      "grad_norm": 4592.9296875,
      "learning_rate": 0.00018822176041769523,
      "loss": 7.3481,
      "step": 134060
    },
    {
      "epoch": 16.133574007220215,
      "grad_norm": 3657.686767578125,
      "learning_rate": 0.00018821996914420604,
      "loss": 7.4564,
      "step": 134070
    },
    {
      "epoch": 16.134777376654633,
      "grad_norm": 4818.3388671875,
      "learning_rate": 0.0001882181777430408,
      "loss": 7.5084,
      "step": 134080
    },
    {
      "epoch": 16.13598074608905,
      "grad_norm": 5988.2705078125,
      "learning_rate": 0.00018821638621420206,
      "loss": 7.5656,
      "step": 134090
    },
    {
      "epoch": 16.137184115523464,
      "grad_norm": 3293.47265625,
      "learning_rate": 0.00018821459455769244,
      "loss": 7.501,
      "step": 134100
    },
    {
      "epoch": 16.13838748495788,
      "grad_norm": 3640.26416015625,
      "learning_rate": 0.00018821280277351449,
      "loss": 7.5136,
      "step": 134110
    },
    {
      "epoch": 16.1395908543923,
      "grad_norm": 4979.64794921875,
      "learning_rate": 0.00018821101086167086,
      "loss": 7.484,
      "step": 134120
    },
    {
      "epoch": 16.140794223826713,
      "grad_norm": 4171.54931640625,
      "learning_rate": 0.0001882092188221641,
      "loss": 7.4495,
      "step": 134130
    },
    {
      "epoch": 16.14199759326113,
      "grad_norm": 9866.3525390625,
      "learning_rate": 0.00018820742665499684,
      "loss": 7.3779,
      "step": 134140
    },
    {
      "epoch": 16.14320096269555,
      "grad_norm": 4566.8125,
      "learning_rate": 0.00018820563436017164,
      "loss": 7.5587,
      "step": 134150
    },
    {
      "epoch": 16.144404332129962,
      "grad_norm": 7476.89599609375,
      "learning_rate": 0.00018820384193769113,
      "loss": 7.5376,
      "step": 134160
    },
    {
      "epoch": 16.14560770156438,
      "grad_norm": 4836.7578125,
      "learning_rate": 0.00018820204938755787,
      "loss": 7.4673,
      "step": 134170
    },
    {
      "epoch": 16.146811070998798,
      "grad_norm": 3187.313720703125,
      "learning_rate": 0.00018820025670977447,
      "loss": 7.5112,
      "step": 134180
    },
    {
      "epoch": 16.14801444043321,
      "grad_norm": 3993.05615234375,
      "learning_rate": 0.00018819846390434353,
      "loss": 7.5605,
      "step": 134190
    },
    {
      "epoch": 16.14921780986763,
      "grad_norm": 3113.9814453125,
      "learning_rate": 0.00018819667097126765,
      "loss": 7.4392,
      "step": 134200
    },
    {
      "epoch": 16.150421179302047,
      "grad_norm": 2835.6923828125,
      "learning_rate": 0.0001881948779105494,
      "loss": 7.4144,
      "step": 134210
    },
    {
      "epoch": 16.15162454873646,
      "grad_norm": 5039.08984375,
      "learning_rate": 0.00018819308472219136,
      "loss": 7.4577,
      "step": 134220
    },
    {
      "epoch": 16.15282791817088,
      "grad_norm": 2328.653076171875,
      "learning_rate": 0.0001881912914061962,
      "loss": 7.3902,
      "step": 134230
    },
    {
      "epoch": 16.154031287605296,
      "grad_norm": 3229.591064453125,
      "learning_rate": 0.00018818949796256643,
      "loss": 7.4442,
      "step": 134240
    },
    {
      "epoch": 16.15523465703971,
      "grad_norm": 2118.358154296875,
      "learning_rate": 0.00018818770439130468,
      "loss": 7.4533,
      "step": 134250
    },
    {
      "epoch": 16.156438026474127,
      "grad_norm": 2109.6953125,
      "learning_rate": 0.00018818591069241357,
      "loss": 7.418,
      "step": 134260
    },
    {
      "epoch": 16.157641395908545,
      "grad_norm": 1547.218505859375,
      "learning_rate": 0.0001881841168658957,
      "loss": 7.3236,
      "step": 134270
    },
    {
      "epoch": 16.15884476534296,
      "grad_norm": 1125.2373046875,
      "learning_rate": 0.0001881823229117536,
      "loss": 7.4228,
      "step": 134280
    },
    {
      "epoch": 16.160048134777377,
      "grad_norm": 1350.6263427734375,
      "learning_rate": 0.00018818052882998992,
      "loss": 7.315,
      "step": 134290
    },
    {
      "epoch": 16.161251504211794,
      "grad_norm": 2960.20849609375,
      "learning_rate": 0.0001881787346206072,
      "loss": 7.355,
      "step": 134300
    },
    {
      "epoch": 16.162454873646208,
      "grad_norm": 2167.13720703125,
      "learning_rate": 0.00018817694028360813,
      "loss": 7.364,
      "step": 134310
    },
    {
      "epoch": 16.163658243080626,
      "grad_norm": 3908.79052734375,
      "learning_rate": 0.00018817514581899526,
      "loss": 7.3271,
      "step": 134320
    },
    {
      "epoch": 16.164861612515043,
      "grad_norm": 5377.28076171875,
      "learning_rate": 0.00018817335122677117,
      "loss": 7.4126,
      "step": 134330
    },
    {
      "epoch": 16.166064981949457,
      "grad_norm": 1815.341796875,
      "learning_rate": 0.00018817155650693848,
      "loss": 7.3638,
      "step": 134340
    },
    {
      "epoch": 16.167268351383875,
      "grad_norm": 1928.3946533203125,
      "learning_rate": 0.0001881697616594998,
      "loss": 7.3946,
      "step": 134350
    },
    {
      "epoch": 16.168471720818292,
      "grad_norm": 1802.7713623046875,
      "learning_rate": 0.00018816796668445765,
      "loss": 7.3332,
      "step": 134360
    },
    {
      "epoch": 16.169675090252706,
      "grad_norm": 1804.3046875,
      "learning_rate": 0.0001881661715818147,
      "loss": 7.2565,
      "step": 134370
    },
    {
      "epoch": 16.170878459687124,
      "grad_norm": 635.908203125,
      "learning_rate": 0.00018816437635157355,
      "loss": 7.3992,
      "step": 134380
    },
    {
      "epoch": 16.17208182912154,
      "grad_norm": 653.279052734375,
      "learning_rate": 0.00018816258099373678,
      "loss": 7.361,
      "step": 134390
    },
    {
      "epoch": 16.173285198555956,
      "grad_norm": 1067.9378662109375,
      "learning_rate": 0.00018816078550830701,
      "loss": 7.3135,
      "step": 134400
    },
    {
      "epoch": 16.174488567990373,
      "grad_norm": 1257.174560546875,
      "learning_rate": 0.00018815898989528676,
      "loss": 7.3672,
      "step": 134410
    },
    {
      "epoch": 16.17569193742479,
      "grad_norm": 4849.39453125,
      "learning_rate": 0.00018815719415467874,
      "loss": 7.3568,
      "step": 134420
    },
    {
      "epoch": 16.176895306859205,
      "grad_norm": 9125.1962890625,
      "learning_rate": 0.00018815539828648546,
      "loss": 7.3637,
      "step": 134430
    },
    {
      "epoch": 16.178098676293622,
      "grad_norm": 7759.28271484375,
      "learning_rate": 0.00018815360229070956,
      "loss": 7.4134,
      "step": 134440
    },
    {
      "epoch": 16.17930204572804,
      "grad_norm": 3989.203125,
      "learning_rate": 0.00018815180616735364,
      "loss": 7.5372,
      "step": 134450
    },
    {
      "epoch": 16.180505415162454,
      "grad_norm": 8492.6474609375,
      "learning_rate": 0.0001881500099164203,
      "loss": 7.501,
      "step": 134460
    },
    {
      "epoch": 16.18170878459687,
      "grad_norm": 1717.7882080078125,
      "learning_rate": 0.0001881482135379121,
      "loss": 7.4058,
      "step": 134470
    },
    {
      "epoch": 16.18291215403129,
      "grad_norm": 2014.2508544921875,
      "learning_rate": 0.0001881464170318317,
      "loss": 7.4289,
      "step": 134480
    },
    {
      "epoch": 16.184115523465703,
      "grad_norm": 2031.71826171875,
      "learning_rate": 0.00018814462039818167,
      "loss": 7.2289,
      "step": 134490
    },
    {
      "epoch": 16.18531889290012,
      "grad_norm": 2361.267333984375,
      "learning_rate": 0.00018814282363696462,
      "loss": 7.3201,
      "step": 134500
    },
    {
      "epoch": 16.186522262334538,
      "grad_norm": 4718.13818359375,
      "learning_rate": 0.00018814102674818314,
      "loss": 7.1876,
      "step": 134510
    },
    {
      "epoch": 16.187725631768952,
      "grad_norm": 1611.455810546875,
      "learning_rate": 0.00018813922973183982,
      "loss": 7.4207,
      "step": 134520
    },
    {
      "epoch": 16.18892900120337,
      "grad_norm": 1906.565673828125,
      "learning_rate": 0.0001881374325879373,
      "loss": 7.3867,
      "step": 134530
    },
    {
      "epoch": 16.190132370637787,
      "grad_norm": 2264.2587890625,
      "learning_rate": 0.0001881356353164781,
      "loss": 7.2454,
      "step": 134540
    },
    {
      "epoch": 16.1913357400722,
      "grad_norm": 816.5081176757812,
      "learning_rate": 0.00018813383791746493,
      "loss": 7.3905,
      "step": 134550
    },
    {
      "epoch": 16.19253910950662,
      "grad_norm": 1701.131591796875,
      "learning_rate": 0.00018813204039090032,
      "loss": 7.3097,
      "step": 134560
    },
    {
      "epoch": 16.193742478941036,
      "grad_norm": 1724.3487548828125,
      "learning_rate": 0.00018813024273678686,
      "loss": 7.2824,
      "step": 134570
    },
    {
      "epoch": 16.19494584837545,
      "grad_norm": 2758.356689453125,
      "learning_rate": 0.0001881284449551272,
      "loss": 7.3388,
      "step": 134580
    },
    {
      "epoch": 16.196149217809868,
      "grad_norm": 982.5479125976562,
      "learning_rate": 0.00018812664704592394,
      "loss": 7.3162,
      "step": 134590
    },
    {
      "epoch": 16.197352587244286,
      "grad_norm": 1717.4306640625,
      "learning_rate": 0.00018812484900917965,
      "loss": 7.3075,
      "step": 134600
    },
    {
      "epoch": 16.1985559566787,
      "grad_norm": 2894.6103515625,
      "learning_rate": 0.00018812305084489695,
      "loss": 7.4182,
      "step": 134610
    },
    {
      "epoch": 16.199759326113117,
      "grad_norm": 1259.333251953125,
      "learning_rate": 0.00018812125255307844,
      "loss": 7.347,
      "step": 134620
    },
    {
      "epoch": 16.200962695547535,
      "grad_norm": 1081.9444580078125,
      "learning_rate": 0.0001881194541337267,
      "loss": 7.383,
      "step": 134630
    },
    {
      "epoch": 16.20216606498195,
      "grad_norm": 1318.0164794921875,
      "learning_rate": 0.00018811765558684438,
      "loss": 7.352,
      "step": 134640
    },
    {
      "epoch": 16.203369434416366,
      "grad_norm": 553.3134765625,
      "learning_rate": 0.00018811585691243406,
      "loss": 7.2723,
      "step": 134650
    },
    {
      "epoch": 16.204572803850784,
      "grad_norm": 959.5357666015625,
      "learning_rate": 0.00018811405811049831,
      "loss": 7.285,
      "step": 134660
    },
    {
      "epoch": 16.205776173285198,
      "grad_norm": 404.1699523925781,
      "learning_rate": 0.0001881122591810398,
      "loss": 7.2958,
      "step": 134670
    },
    {
      "epoch": 16.206979542719615,
      "grad_norm": 1347.3458251953125,
      "learning_rate": 0.00018811046012406107,
      "loss": 7.308,
      "step": 134680
    },
    {
      "epoch": 16.20818291215403,
      "grad_norm": 1214.8868408203125,
      "learning_rate": 0.00018810866093956475,
      "loss": 7.3352,
      "step": 134690
    },
    {
      "epoch": 16.209386281588447,
      "grad_norm": 871.9038696289062,
      "learning_rate": 0.00018810686162755346,
      "loss": 7.2537,
      "step": 134700
    },
    {
      "epoch": 16.210589651022865,
      "grad_norm": 528.3934326171875,
      "learning_rate": 0.00018810506218802978,
      "loss": 7.283,
      "step": 134710
    },
    {
      "epoch": 16.21179302045728,
      "grad_norm": 710.7511596679688,
      "learning_rate": 0.00018810326262099633,
      "loss": 7.3367,
      "step": 134720
    },
    {
      "epoch": 16.212996389891696,
      "grad_norm": 1877.6536865234375,
      "learning_rate": 0.0001881014629264557,
      "loss": 7.329,
      "step": 134730
    },
    {
      "epoch": 16.214199759326114,
      "grad_norm": 813.703857421875,
      "learning_rate": 0.00018809966310441047,
      "loss": 7.2091,
      "step": 134740
    },
    {
      "epoch": 16.215403128760528,
      "grad_norm": 707.8218383789062,
      "learning_rate": 0.00018809786315486334,
      "loss": 7.2575,
      "step": 134750
    },
    {
      "epoch": 16.216606498194945,
      "grad_norm": 3227.122314453125,
      "learning_rate": 0.0001880960630778168,
      "loss": 7.2496,
      "step": 134760
    },
    {
      "epoch": 16.217809867629363,
      "grad_norm": 268.9632263183594,
      "learning_rate": 0.00018809426287327353,
      "loss": 7.3143,
      "step": 134770
    },
    {
      "epoch": 16.219013237063777,
      "grad_norm": 1045.4754638671875,
      "learning_rate": 0.0001880924625412361,
      "loss": 7.4153,
      "step": 134780
    },
    {
      "epoch": 16.220216606498195,
      "grad_norm": 1444.0511474609375,
      "learning_rate": 0.00018809066208170716,
      "loss": 7.2603,
      "step": 134790
    },
    {
      "epoch": 16.221419975932612,
      "grad_norm": 556.0645141601562,
      "learning_rate": 0.00018808886149468924,
      "loss": 7.3158,
      "step": 134800
    },
    {
      "epoch": 16.222623345367026,
      "grad_norm": 485.6466979980469,
      "learning_rate": 0.000188087060780185,
      "loss": 7.2585,
      "step": 134810
    },
    {
      "epoch": 16.223826714801444,
      "grad_norm": 863.2359619140625,
      "learning_rate": 0.00018808525993819705,
      "loss": 7.4184,
      "step": 134820
    },
    {
      "epoch": 16.22503008423586,
      "grad_norm": 10132.2978515625,
      "learning_rate": 0.00018808345896872795,
      "loss": 7.3077,
      "step": 134830
    },
    {
      "epoch": 16.226233453670275,
      "grad_norm": 814.1869506835938,
      "learning_rate": 0.00018808165787178037,
      "loss": 7.2195,
      "step": 134840
    },
    {
      "epoch": 16.227436823104693,
      "grad_norm": 854.6044311523438,
      "learning_rate": 0.0001880798566473569,
      "loss": 7.3092,
      "step": 134850
    },
    {
      "epoch": 16.22864019253911,
      "grad_norm": 3079.143798828125,
      "learning_rate": 0.0001880780552954601,
      "loss": 7.4242,
      "step": 134860
    },
    {
      "epoch": 16.229843561973524,
      "grad_norm": 98335.5859375,
      "learning_rate": 0.00018807625381609262,
      "loss": 7.5575,
      "step": 134870
    },
    {
      "epoch": 16.231046931407942,
      "grad_norm": 68654.34375,
      "learning_rate": 0.00018807445220925706,
      "loss": 7.468,
      "step": 134880
    },
    {
      "epoch": 16.23225030084236,
      "grad_norm": 19752.994140625,
      "learning_rate": 0.000188072650474956,
      "loss": 7.5652,
      "step": 134890
    },
    {
      "epoch": 16.233453670276774,
      "grad_norm": 28384.818359375,
      "learning_rate": 0.0001880708486131921,
      "loss": 7.831,
      "step": 134900
    },
    {
      "epoch": 16.23465703971119,
      "grad_norm": 5448.560546875,
      "learning_rate": 0.0001880690466239679,
      "loss": 7.8397,
      "step": 134910
    },
    {
      "epoch": 16.23586040914561,
      "grad_norm": 3974.45751953125,
      "learning_rate": 0.00018806724450728607,
      "loss": 7.71,
      "step": 134920
    },
    {
      "epoch": 16.237063778580023,
      "grad_norm": 2246.431884765625,
      "learning_rate": 0.00018806544226314918,
      "loss": 7.5296,
      "step": 134930
    },
    {
      "epoch": 16.23826714801444,
      "grad_norm": 833.052490234375,
      "learning_rate": 0.00018806363989155988,
      "loss": 7.4504,
      "step": 134940
    },
    {
      "epoch": 16.239470517448858,
      "grad_norm": 1213.1649169921875,
      "learning_rate": 0.00018806183739252075,
      "loss": 7.5472,
      "step": 134950
    },
    {
      "epoch": 16.240673886883272,
      "grad_norm": 1169.3134765625,
      "learning_rate": 0.00018806003476603436,
      "loss": 7.4958,
      "step": 134960
    },
    {
      "epoch": 16.24187725631769,
      "grad_norm": 322.09368896484375,
      "learning_rate": 0.00018805823201210338,
      "loss": 7.3395,
      "step": 134970
    },
    {
      "epoch": 16.243080625752107,
      "grad_norm": 1242.49169921875,
      "learning_rate": 0.00018805642913073044,
      "loss": 7.5151,
      "step": 134980
    },
    {
      "epoch": 16.24428399518652,
      "grad_norm": 598.2134399414062,
      "learning_rate": 0.00018805462612191804,
      "loss": 7.3753,
      "step": 134990
    },
    {
      "epoch": 16.24548736462094,
      "grad_norm": 1171.3634033203125,
      "learning_rate": 0.0001880528229856689,
      "loss": 7.3291,
      "step": 135000
    },
    {
      "epoch": 16.246690734055356,
      "grad_norm": 987.525146484375,
      "learning_rate": 0.00018805101972198555,
      "loss": 7.4213,
      "step": 135010
    },
    {
      "epoch": 16.24789410348977,
      "grad_norm": 1815.230712890625,
      "learning_rate": 0.00018804921633087064,
      "loss": 7.3551,
      "step": 135020
    },
    {
      "epoch": 16.249097472924188,
      "grad_norm": 3428.69921875,
      "learning_rate": 0.0001880474128123268,
      "loss": 7.4784,
      "step": 135030
    },
    {
      "epoch": 16.250300842358605,
      "grad_norm": 26357.70703125,
      "learning_rate": 0.00018804560916635662,
      "loss": 7.4469,
      "step": 135040
    },
    {
      "epoch": 16.25150421179302,
      "grad_norm": 8461.775390625,
      "learning_rate": 0.00018804380539296268,
      "loss": 7.5533,
      "step": 135050
    },
    {
      "epoch": 16.252707581227437,
      "grad_norm": 38504.7265625,
      "learning_rate": 0.0001880420014921476,
      "loss": 7.4803,
      "step": 135060
    },
    {
      "epoch": 16.253910950661854,
      "grad_norm": 2633.317626953125,
      "learning_rate": 0.00018804019746391406,
      "loss": 7.4323,
      "step": 135070
    },
    {
      "epoch": 16.25511432009627,
      "grad_norm": 3646.68212890625,
      "learning_rate": 0.0001880383933082646,
      "loss": 7.3077,
      "step": 135080
    },
    {
      "epoch": 16.256317689530686,
      "grad_norm": 3053.300537109375,
      "learning_rate": 0.00018803658902520184,
      "loss": 7.4784,
      "step": 135090
    },
    {
      "epoch": 16.257521058965104,
      "grad_norm": 8867.451171875,
      "learning_rate": 0.0001880347846147284,
      "loss": 7.4092,
      "step": 135100
    },
    {
      "epoch": 16.258724428399518,
      "grad_norm": 3264.137939453125,
      "learning_rate": 0.0001880329800768469,
      "loss": 7.4702,
      "step": 135110
    },
    {
      "epoch": 16.259927797833935,
      "grad_norm": 3088.422119140625,
      "learning_rate": 0.00018803117541155991,
      "loss": 7.3454,
      "step": 135120
    },
    {
      "epoch": 16.261131167268353,
      "grad_norm": 5243.46923828125,
      "learning_rate": 0.0001880293706188701,
      "loss": 7.3988,
      "step": 135130
    },
    {
      "epoch": 16.262334536702767,
      "grad_norm": 430.4847412109375,
      "learning_rate": 0.00018802756569878006,
      "loss": 7.4231,
      "step": 135140
    },
    {
      "epoch": 16.263537906137184,
      "grad_norm": 1054.1220703125,
      "learning_rate": 0.00018802576065129242,
      "loss": 7.3821,
      "step": 135150
    },
    {
      "epoch": 16.264741275571602,
      "grad_norm": 359.6756896972656,
      "learning_rate": 0.00018802395547640974,
      "loss": 7.4069,
      "step": 135160
    },
    {
      "epoch": 16.265944645006016,
      "grad_norm": 357.1447448730469,
      "learning_rate": 0.00018802215017413464,
      "loss": 7.3935,
      "step": 135170
    },
    {
      "epoch": 16.267148014440433,
      "grad_norm": 6698.04736328125,
      "learning_rate": 0.00018802034474446982,
      "loss": 7.468,
      "step": 135180
    },
    {
      "epoch": 16.26835138387485,
      "grad_norm": 1396.1961669921875,
      "learning_rate": 0.00018801853918741778,
      "loss": 7.3965,
      "step": 135190
    },
    {
      "epoch": 16.269554753309265,
      "grad_norm": 710.7750854492188,
      "learning_rate": 0.00018801673350298122,
      "loss": 7.5052,
      "step": 135200
    },
    {
      "epoch": 16.270758122743683,
      "grad_norm": 2006.60986328125,
      "learning_rate": 0.00018801492769116268,
      "loss": 7.4554,
      "step": 135210
    },
    {
      "epoch": 16.2719614921781,
      "grad_norm": 487.1640930175781,
      "learning_rate": 0.00018801312175196482,
      "loss": 7.4907,
      "step": 135220
    },
    {
      "epoch": 16.273164861612514,
      "grad_norm": 12558.9658203125,
      "learning_rate": 0.00018801131568539026,
      "loss": 7.3869,
      "step": 135230
    },
    {
      "epoch": 16.27436823104693,
      "grad_norm": 1464.945068359375,
      "learning_rate": 0.00018800950949144157,
      "loss": 7.3742,
      "step": 135240
    },
    {
      "epoch": 16.27557160048135,
      "grad_norm": 1422.491943359375,
      "learning_rate": 0.00018800770317012143,
      "loss": 7.4689,
      "step": 135250
    },
    {
      "epoch": 16.276774969915763,
      "grad_norm": 3618.24072265625,
      "learning_rate": 0.0001880058967214324,
      "loss": 7.3944,
      "step": 135260
    },
    {
      "epoch": 16.27797833935018,
      "grad_norm": 1279.1890869140625,
      "learning_rate": 0.0001880040901453771,
      "loss": 7.3848,
      "step": 135270
    },
    {
      "epoch": 16.2791817087846,
      "grad_norm": 6718681.5,
      "learning_rate": 0.00018800228344195813,
      "loss": 7.3688,
      "step": 135280
    },
    {
      "epoch": 16.280385078219012,
      "grad_norm": 3917.640380859375,
      "learning_rate": 0.00018800047661117817,
      "loss": 7.4604,
      "step": 135290
    },
    {
      "epoch": 16.28158844765343,
      "grad_norm": 7963.70556640625,
      "learning_rate": 0.0001879986696530398,
      "loss": 7.4116,
      "step": 135300
    },
    {
      "epoch": 16.282791817087848,
      "grad_norm": 1286.34375,
      "learning_rate": 0.00018799686256754558,
      "loss": 7.3811,
      "step": 135310
    },
    {
      "epoch": 16.28399518652226,
      "grad_norm": 3045.016845703125,
      "learning_rate": 0.00018799505535469822,
      "loss": 7.3988,
      "step": 135320
    },
    {
      "epoch": 16.28519855595668,
      "grad_norm": 6178.04296875,
      "learning_rate": 0.00018799324801450025,
      "loss": 7.4496,
      "step": 135330
    },
    {
      "epoch": 16.286401925391097,
      "grad_norm": 1294.6044921875,
      "learning_rate": 0.00018799144054695437,
      "loss": 7.3696,
      "step": 135340
    },
    {
      "epoch": 16.28760529482551,
      "grad_norm": 5328.47265625,
      "learning_rate": 0.0001879896329520631,
      "loss": 7.354,
      "step": 135350
    },
    {
      "epoch": 16.28880866425993,
      "grad_norm": 2091.859375,
      "learning_rate": 0.00018798782522982913,
      "loss": 7.3048,
      "step": 135360
    },
    {
      "epoch": 16.290012033694346,
      "grad_norm": 3573.5048828125,
      "learning_rate": 0.00018798601738025507,
      "loss": 7.4095,
      "step": 135370
    },
    {
      "epoch": 16.29121540312876,
      "grad_norm": 5019.990234375,
      "learning_rate": 0.00018798420940334352,
      "loss": 7.4838,
      "step": 135380
    },
    {
      "epoch": 16.292418772563177,
      "grad_norm": 4849.22509765625,
      "learning_rate": 0.00018798240129909706,
      "loss": 7.337,
      "step": 135390
    },
    {
      "epoch": 16.29362214199759,
      "grad_norm": 1907.13330078125,
      "learning_rate": 0.00018798059306751838,
      "loss": 7.4253,
      "step": 135400
    },
    {
      "epoch": 16.29482551143201,
      "grad_norm": 6181.5908203125,
      "learning_rate": 0.00018797878470861004,
      "loss": 7.3597,
      "step": 135410
    },
    {
      "epoch": 16.296028880866427,
      "grad_norm": 837.5048217773438,
      "learning_rate": 0.00018797697622237468,
      "loss": 7.3881,
      "step": 135420
    },
    {
      "epoch": 16.29723225030084,
      "grad_norm": 4801.70703125,
      "learning_rate": 0.00018797516760881493,
      "loss": 7.4965,
      "step": 135430
    },
    {
      "epoch": 16.29843561973526,
      "grad_norm": 814.237548828125,
      "learning_rate": 0.0001879733588679334,
      "loss": 7.3403,
      "step": 135440
    },
    {
      "epoch": 16.299638989169676,
      "grad_norm": 4937.7548828125,
      "learning_rate": 0.00018797154999973268,
      "loss": 7.4315,
      "step": 135450
    },
    {
      "epoch": 16.30084235860409,
      "grad_norm": 1484.304931640625,
      "learning_rate": 0.00018796974100421541,
      "loss": 7.3327,
      "step": 135460
    },
    {
      "epoch": 16.302045728038507,
      "grad_norm": 16406.900390625,
      "learning_rate": 0.0001879679318813842,
      "loss": 7.3749,
      "step": 135470
    },
    {
      "epoch": 16.303249097472925,
      "grad_norm": 4597.98828125,
      "learning_rate": 0.00018796612263124166,
      "loss": 7.4448,
      "step": 135480
    },
    {
      "epoch": 16.30445246690734,
      "grad_norm": 1915.1119384765625,
      "learning_rate": 0.00018796431325379044,
      "loss": 7.3674,
      "step": 135490
    },
    {
      "epoch": 16.305655836341757,
      "grad_norm": 4350.32421875,
      "learning_rate": 0.00018796250374903317,
      "loss": 7.3967,
      "step": 135500
    },
    {
      "epoch": 16.306859205776174,
      "grad_norm": 35623.3515625,
      "learning_rate": 0.00018796069411697238,
      "loss": 7.3241,
      "step": 135510
    },
    {
      "epoch": 16.308062575210588,
      "grad_norm": 1764.968017578125,
      "learning_rate": 0.00018795888435761083,
      "loss": 7.3164,
      "step": 135520
    },
    {
      "epoch": 16.309265944645006,
      "grad_norm": 2249.67822265625,
      "learning_rate": 0.000187957074470951,
      "loss": 7.5236,
      "step": 135530
    },
    {
      "epoch": 16.310469314079423,
      "grad_norm": 3995.468017578125,
      "learning_rate": 0.00018795526445699557,
      "loss": 7.3412,
      "step": 135540
    },
    {
      "epoch": 16.311672683513837,
      "grad_norm": 1048.600830078125,
      "learning_rate": 0.00018795345431574717,
      "loss": 7.42,
      "step": 135550
    },
    {
      "epoch": 16.312876052948255,
      "grad_norm": 8585.6337890625,
      "learning_rate": 0.0001879516440472084,
      "loss": 7.4167,
      "step": 135560
    },
    {
      "epoch": 16.314079422382672,
      "grad_norm": 8284.8955078125,
      "learning_rate": 0.00018794983365138187,
      "loss": 7.469,
      "step": 135570
    },
    {
      "epoch": 16.315282791817086,
      "grad_norm": 12807.7412109375,
      "learning_rate": 0.00018794802312827025,
      "loss": 7.352,
      "step": 135580
    },
    {
      "epoch": 16.316486161251504,
      "grad_norm": 7363.87109375,
      "learning_rate": 0.0001879462124778761,
      "loss": 7.4129,
      "step": 135590
    },
    {
      "epoch": 16.31768953068592,
      "grad_norm": 3202.677001953125,
      "learning_rate": 0.00018794440170020207,
      "loss": 7.3481,
      "step": 135600
    },
    {
      "epoch": 16.318892900120336,
      "grad_norm": 6564.0791015625,
      "learning_rate": 0.00018794259079525077,
      "loss": 7.4588,
      "step": 135610
    },
    {
      "epoch": 16.320096269554753,
      "grad_norm": 1761.1639404296875,
      "learning_rate": 0.00018794077976302487,
      "loss": 7.4876,
      "step": 135620
    },
    {
      "epoch": 16.32129963898917,
      "grad_norm": 3432.889892578125,
      "learning_rate": 0.0001879389686035269,
      "loss": 7.3906,
      "step": 135630
    },
    {
      "epoch": 16.322503008423585,
      "grad_norm": 7886.70703125,
      "learning_rate": 0.00018793715731675955,
      "loss": 7.3919,
      "step": 135640
    },
    {
      "epoch": 16.323706377858002,
      "grad_norm": 4607.04638671875,
      "learning_rate": 0.00018793534590272543,
      "loss": 7.3867,
      "step": 135650
    },
    {
      "epoch": 16.32490974729242,
      "grad_norm": 4718.65380859375,
      "learning_rate": 0.00018793353436142717,
      "loss": 7.496,
      "step": 135660
    },
    {
      "epoch": 16.326113116726834,
      "grad_norm": 3534.19384765625,
      "learning_rate": 0.00018793172269286737,
      "loss": 7.3163,
      "step": 135670
    },
    {
      "epoch": 16.32731648616125,
      "grad_norm": 4739.6005859375,
      "learning_rate": 0.0001879299108970486,
      "loss": 7.3221,
      "step": 135680
    },
    {
      "epoch": 16.32851985559567,
      "grad_norm": 13892.4052734375,
      "learning_rate": 0.00018792809897397362,
      "loss": 7.3012,
      "step": 135690
    },
    {
      "epoch": 16.329723225030083,
      "grad_norm": 1643.28515625,
      "learning_rate": 0.00018792628692364493,
      "loss": 7.3831,
      "step": 135700
    },
    {
      "epoch": 16.3309265944645,
      "grad_norm": 2262.72021484375,
      "learning_rate": 0.0001879244747460652,
      "loss": 7.4484,
      "step": 135710
    },
    {
      "epoch": 16.332129963898918,
      "grad_norm": 1647.235595703125,
      "learning_rate": 0.00018792266244123705,
      "loss": 7.317,
      "step": 135720
    },
    {
      "epoch": 16.333333333333332,
      "grad_norm": 4474.38720703125,
      "learning_rate": 0.00018792085000916308,
      "loss": 7.3673,
      "step": 135730
    },
    {
      "epoch": 16.33453670276775,
      "grad_norm": 13284.1533203125,
      "learning_rate": 0.00018791903744984595,
      "loss": 7.4562,
      "step": 135740
    },
    {
      "epoch": 16.335740072202167,
      "grad_norm": 1539.72802734375,
      "learning_rate": 0.0001879172247632883,
      "loss": 7.2861,
      "step": 135750
    },
    {
      "epoch": 16.33694344163658,
      "grad_norm": 1045.3782958984375,
      "learning_rate": 0.00018791541194949267,
      "loss": 7.2342,
      "step": 135760
    },
    {
      "epoch": 16.338146811071,
      "grad_norm": 3884.170166015625,
      "learning_rate": 0.00018791359900846176,
      "loss": 7.4265,
      "step": 135770
    },
    {
      "epoch": 16.339350180505416,
      "grad_norm": 7942.37353515625,
      "learning_rate": 0.00018791178594019815,
      "loss": 7.3579,
      "step": 135780
    },
    {
      "epoch": 16.34055354993983,
      "grad_norm": 1561.6995849609375,
      "learning_rate": 0.0001879099727447045,
      "loss": 7.3824,
      "step": 135790
    },
    {
      "epoch": 16.341756919374248,
      "grad_norm": 5925.3896484375,
      "learning_rate": 0.00018790815942198338,
      "loss": 7.4067,
      "step": 135800
    },
    {
      "epoch": 16.342960288808666,
      "grad_norm": 1170.9866943359375,
      "learning_rate": 0.00018790634597203748,
      "loss": 7.3359,
      "step": 135810
    },
    {
      "epoch": 16.34416365824308,
      "grad_norm": 1370.93994140625,
      "learning_rate": 0.00018790453239486937,
      "loss": 7.3038,
      "step": 135820
    },
    {
      "epoch": 16.345367027677497,
      "grad_norm": 7989.8427734375,
      "learning_rate": 0.00018790271869048174,
      "loss": 7.3582,
      "step": 135830
    },
    {
      "epoch": 16.346570397111915,
      "grad_norm": 1201.0194091796875,
      "learning_rate": 0.00018790090485887713,
      "loss": 7.4095,
      "step": 135840
    },
    {
      "epoch": 16.34777376654633,
      "grad_norm": 5560.896484375,
      "learning_rate": 0.00018789909090005823,
      "loss": 7.4376,
      "step": 135850
    },
    {
      "epoch": 16.348977135980746,
      "grad_norm": 2773.318115234375,
      "learning_rate": 0.00018789727681402764,
      "loss": 7.3307,
      "step": 135860
    },
    {
      "epoch": 16.350180505415164,
      "grad_norm": 15386.08203125,
      "learning_rate": 0.00018789546260078801,
      "loss": 7.3922,
      "step": 135870
    },
    {
      "epoch": 16.351383874849578,
      "grad_norm": 2575.066650390625,
      "learning_rate": 0.0001878936482603419,
      "loss": 7.437,
      "step": 135880
    },
    {
      "epoch": 16.352587244283995,
      "grad_norm": 2834.9453125,
      "learning_rate": 0.00018789183379269204,
      "loss": 7.4451,
      "step": 135890
    },
    {
      "epoch": 16.353790613718413,
      "grad_norm": 12504.13671875,
      "learning_rate": 0.00018789001919784095,
      "loss": 7.3121,
      "step": 135900
    },
    {
      "epoch": 16.354993983152827,
      "grad_norm": 5930.501953125,
      "learning_rate": 0.00018788820447579132,
      "loss": 7.4703,
      "step": 135910
    },
    {
      "epoch": 16.356197352587245,
      "grad_norm": 4534.30615234375,
      "learning_rate": 0.00018788638962654575,
      "loss": 7.3633,
      "step": 135920
    },
    {
      "epoch": 16.357400722021662,
      "grad_norm": 3038.640869140625,
      "learning_rate": 0.00018788457465010688,
      "loss": 7.375,
      "step": 135930
    },
    {
      "epoch": 16.358604091456076,
      "grad_norm": 5623.09423828125,
      "learning_rate": 0.00018788275954647734,
      "loss": 7.3456,
      "step": 135940
    },
    {
      "epoch": 16.359807460890494,
      "grad_norm": 29531.099609375,
      "learning_rate": 0.00018788094431565974,
      "loss": 7.443,
      "step": 135950
    },
    {
      "epoch": 16.36101083032491,
      "grad_norm": 22576.966796875,
      "learning_rate": 0.0001878791289576567,
      "loss": 7.4488,
      "step": 135960
    },
    {
      "epoch": 16.362214199759325,
      "grad_norm": 378.20758056640625,
      "learning_rate": 0.0001878773134724709,
      "loss": 7.3549,
      "step": 135970
    },
    {
      "epoch": 16.363417569193743,
      "grad_norm": 3222.92822265625,
      "learning_rate": 0.0001878754978601049,
      "loss": 7.2736,
      "step": 135980
    },
    {
      "epoch": 16.36462093862816,
      "grad_norm": 422.771240234375,
      "learning_rate": 0.0001878736821205614,
      "loss": 7.4024,
      "step": 135990
    },
    {
      "epoch": 16.365824308062574,
      "grad_norm": 940.5438232421875,
      "learning_rate": 0.00018787186625384295,
      "loss": 7.4342,
      "step": 136000
    },
    {
      "epoch": 16.367027677496992,
      "grad_norm": 835.6951904296875,
      "learning_rate": 0.00018787005025995224,
      "loss": 7.3894,
      "step": 136010
    },
    {
      "epoch": 16.36823104693141,
      "grad_norm": 552.2232055664062,
      "learning_rate": 0.00018786823413889184,
      "loss": 7.3654,
      "step": 136020
    },
    {
      "epoch": 16.369434416365824,
      "grad_norm": 827.7374877929688,
      "learning_rate": 0.00018786641789066445,
      "loss": 7.423,
      "step": 136030
    },
    {
      "epoch": 16.37063778580024,
      "grad_norm": 880.3196411132812,
      "learning_rate": 0.0001878646015152726,
      "loss": 7.5086,
      "step": 136040
    },
    {
      "epoch": 16.37184115523466,
      "grad_norm": 1384.047607421875,
      "learning_rate": 0.00018786278501271904,
      "loss": 7.4775,
      "step": 136050
    },
    {
      "epoch": 16.373044524669073,
      "grad_norm": 575.7612915039062,
      "learning_rate": 0.0001878609683830063,
      "loss": 7.2752,
      "step": 136060
    },
    {
      "epoch": 16.37424789410349,
      "grad_norm": 518.8995971679688,
      "learning_rate": 0.00018785915162613707,
      "loss": 7.488,
      "step": 136070
    },
    {
      "epoch": 16.375451263537904,
      "grad_norm": 15413.3115234375,
      "learning_rate": 0.00018785733474211392,
      "loss": 7.5608,
      "step": 136080
    },
    {
      "epoch": 16.376654632972322,
      "grad_norm": 653.1942138671875,
      "learning_rate": 0.00018785551773093956,
      "loss": 7.3276,
      "step": 136090
    },
    {
      "epoch": 16.37785800240674,
      "grad_norm": 589.8455200195312,
      "learning_rate": 0.00018785370059261656,
      "loss": 7.4827,
      "step": 136100
    },
    {
      "epoch": 16.379061371841154,
      "grad_norm": 1137.1182861328125,
      "learning_rate": 0.00018785188332714757,
      "loss": 7.3475,
      "step": 136110
    },
    {
      "epoch": 16.38026474127557,
      "grad_norm": 5282.48681640625,
      "learning_rate": 0.00018785006593453521,
      "loss": 7.4058,
      "step": 136120
    },
    {
      "epoch": 16.38146811070999,
      "grad_norm": 6450.40625,
      "learning_rate": 0.0001878482484147821,
      "loss": 7.2965,
      "step": 136130
    },
    {
      "epoch": 16.382671480144403,
      "grad_norm": 30703.484375,
      "learning_rate": 0.0001878464307678909,
      "loss": 7.253,
      "step": 136140
    },
    {
      "epoch": 16.38387484957882,
      "grad_norm": 12553.6572265625,
      "learning_rate": 0.00018784461299386425,
      "loss": 7.5897,
      "step": 136150
    },
    {
      "epoch": 16.385078219013238,
      "grad_norm": 43169.89453125,
      "learning_rate": 0.00018784279509270472,
      "loss": 7.594,
      "step": 136160
    },
    {
      "epoch": 16.386281588447652,
      "grad_norm": 11119.8896484375,
      "learning_rate": 0.000187840977064415,
      "loss": 7.6198,
      "step": 136170
    },
    {
      "epoch": 16.38748495788207,
      "grad_norm": 8511.7587890625,
      "learning_rate": 0.00018783915890899767,
      "loss": 7.4384,
      "step": 136180
    },
    {
      "epoch": 16.388688327316487,
      "grad_norm": 5969.4306640625,
      "learning_rate": 0.0001878373406264554,
      "loss": 7.4835,
      "step": 136190
    },
    {
      "epoch": 16.3898916967509,
      "grad_norm": 6323.908203125,
      "learning_rate": 0.00018783552221679083,
      "loss": 7.4741,
      "step": 136200
    },
    {
      "epoch": 16.39109506618532,
      "grad_norm": 70771.2734375,
      "learning_rate": 0.00018783370368000653,
      "loss": 7.4753,
      "step": 136210
    },
    {
      "epoch": 16.392298435619736,
      "grad_norm": 696879.0625,
      "learning_rate": 0.00018783188501610525,
      "loss": 7.4855,
      "step": 136220
    },
    {
      "epoch": 16.39350180505415,
      "grad_norm": 14489.6826171875,
      "learning_rate": 0.0001878300662250895,
      "loss": 7.3957,
      "step": 136230
    },
    {
      "epoch": 16.394705174488568,
      "grad_norm": 14740.2939453125,
      "learning_rate": 0.00018782824730696196,
      "loss": 7.4846,
      "step": 136240
    },
    {
      "epoch": 16.395908543922985,
      "grad_norm": 39206.94140625,
      "learning_rate": 0.00018782642826172528,
      "loss": 7.3847,
      "step": 136250
    },
    {
      "epoch": 16.3971119133574,
      "grad_norm": 25458.478515625,
      "learning_rate": 0.00018782460908938205,
      "loss": 7.3715,
      "step": 136260
    },
    {
      "epoch": 16.398315282791817,
      "grad_norm": 1424.4176025390625,
      "learning_rate": 0.00018782278978993495,
      "loss": 7.3292,
      "step": 136270
    },
    {
      "epoch": 16.399518652226234,
      "grad_norm": 91076.3984375,
      "learning_rate": 0.00018782097036338657,
      "loss": 7.3388,
      "step": 136280
    },
    {
      "epoch": 16.40072202166065,
      "grad_norm": 25471.916015625,
      "learning_rate": 0.00018781915080973957,
      "loss": 7.3743,
      "step": 136290
    },
    {
      "epoch": 16.401925391095066,
      "grad_norm": 5020.595703125,
      "learning_rate": 0.00018781733112899657,
      "loss": 7.4635,
      "step": 136300
    },
    {
      "epoch": 16.403128760529484,
      "grad_norm": 44687.38671875,
      "learning_rate": 0.00018781551132116025,
      "loss": 7.531,
      "step": 136310
    },
    {
      "epoch": 16.404332129963898,
      "grad_norm": 107524.359375,
      "learning_rate": 0.00018781369138623318,
      "loss": 7.4699,
      "step": 136320
    },
    {
      "epoch": 16.405535499398315,
      "grad_norm": 5184.20849609375,
      "learning_rate": 0.00018781187132421799,
      "loss": 7.4869,
      "step": 136330
    },
    {
      "epoch": 16.406738868832733,
      "grad_norm": 18225.78515625,
      "learning_rate": 0.0001878100511351174,
      "loss": 7.5691,
      "step": 136340
    },
    {
      "epoch": 16.407942238267147,
      "grad_norm": 15367.7236328125,
      "learning_rate": 0.0001878082308189339,
      "loss": 7.4852,
      "step": 136350
    },
    {
      "epoch": 16.409145607701564,
      "grad_norm": 47433.625,
      "learning_rate": 0.0001878064103756703,
      "loss": 7.4459,
      "step": 136360
    },
    {
      "epoch": 16.410348977135982,
      "grad_norm": 16086.1826171875,
      "learning_rate": 0.00018780458980532908,
      "loss": 7.3971,
      "step": 136370
    },
    {
      "epoch": 16.411552346570396,
      "grad_norm": 10373.1318359375,
      "learning_rate": 0.00018780276910791298,
      "loss": 7.4956,
      "step": 136380
    },
    {
      "epoch": 16.412755716004813,
      "grad_norm": 40921.32421875,
      "learning_rate": 0.0001878009482834246,
      "loss": 7.4732,
      "step": 136390
    },
    {
      "epoch": 16.41395908543923,
      "grad_norm": 44686.60546875,
      "learning_rate": 0.00018779912733186654,
      "loss": 7.4912,
      "step": 136400
    },
    {
      "epoch": 16.415162454873645,
      "grad_norm": 68040.78125,
      "learning_rate": 0.00018779730625324148,
      "loss": 7.485,
      "step": 136410
    },
    {
      "epoch": 16.416365824308063,
      "grad_norm": 25117.5,
      "learning_rate": 0.00018779548504755205,
      "loss": 7.523,
      "step": 136420
    },
    {
      "epoch": 16.41756919374248,
      "grad_norm": 459473.1875,
      "learning_rate": 0.00018779366371480087,
      "loss": 7.5368,
      "step": 136430
    },
    {
      "epoch": 16.418772563176894,
      "grad_norm": 48309.23046875,
      "learning_rate": 0.00018779184225499056,
      "loss": 7.508,
      "step": 136440
    },
    {
      "epoch": 16.41997593261131,
      "grad_norm": 100739.703125,
      "learning_rate": 0.0001877900206681238,
      "loss": 7.411,
      "step": 136450
    },
    {
      "epoch": 16.42117930204573,
      "grad_norm": 13916.9072265625,
      "learning_rate": 0.00018778819895420322,
      "loss": 7.53,
      "step": 136460
    },
    {
      "epoch": 16.422382671480143,
      "grad_norm": 13331.6533203125,
      "learning_rate": 0.0001877863771132314,
      "loss": 7.4388,
      "step": 136470
    },
    {
      "epoch": 16.42358604091456,
      "grad_norm": 9611.49609375,
      "learning_rate": 0.00018778455514521108,
      "loss": 7.4856,
      "step": 136480
    },
    {
      "epoch": 16.42478941034898,
      "grad_norm": 114058.515625,
      "learning_rate": 0.0001877827330501448,
      "loss": 7.3812,
      "step": 136490
    },
    {
      "epoch": 16.425992779783392,
      "grad_norm": 6379.4580078125,
      "learning_rate": 0.00018778091082803525,
      "loss": 7.5142,
      "step": 136500
    },
    {
      "epoch": 16.42719614921781,
      "grad_norm": 48772.484375,
      "learning_rate": 0.000187779088478885,
      "loss": 7.5297,
      "step": 136510
    },
    {
      "epoch": 16.428399518652228,
      "grad_norm": 38506.64453125,
      "learning_rate": 0.00018777726600269679,
      "loss": 7.4822,
      "step": 136520
    },
    {
      "epoch": 16.42960288808664,
      "grad_norm": 18935.1640625,
      "learning_rate": 0.00018777544339947317,
      "loss": 7.41,
      "step": 136530
    },
    {
      "epoch": 16.43080625752106,
      "grad_norm": 23718.6171875,
      "learning_rate": 0.00018777362066921683,
      "loss": 7.4754,
      "step": 136540
    },
    {
      "epoch": 16.432009626955477,
      "grad_norm": 47956.45703125,
      "learning_rate": 0.00018777179781193039,
      "loss": 7.4338,
      "step": 136550
    },
    {
      "epoch": 16.43321299638989,
      "grad_norm": 38971.03515625,
      "learning_rate": 0.00018776997482761648,
      "loss": 7.5148,
      "step": 136560
    },
    {
      "epoch": 16.43441636582431,
      "grad_norm": 8915.9169921875,
      "learning_rate": 0.00018776815171627777,
      "loss": 7.4471,
      "step": 136570
    },
    {
      "epoch": 16.435619735258726,
      "grad_norm": 4852.09716796875,
      "learning_rate": 0.00018776632847791687,
      "loss": 7.471,
      "step": 136580
    },
    {
      "epoch": 16.43682310469314,
      "grad_norm": 19770.119140625,
      "learning_rate": 0.0001877645051125364,
      "loss": 7.4327,
      "step": 136590
    },
    {
      "epoch": 16.438026474127557,
      "grad_norm": 12467.087890625,
      "learning_rate": 0.00018776268162013903,
      "loss": 7.4982,
      "step": 136600
    },
    {
      "epoch": 16.439229843561975,
      "grad_norm": 12908.8193359375,
      "learning_rate": 0.0001877608580007274,
      "loss": 7.3856,
      "step": 136610
    },
    {
      "epoch": 16.44043321299639,
      "grad_norm": 10286.375,
      "learning_rate": 0.00018775903425430413,
      "loss": 7.3419,
      "step": 136620
    },
    {
      "epoch": 16.441636582430807,
      "grad_norm": 31747.52734375,
      "learning_rate": 0.0001877572103808719,
      "loss": 7.4967,
      "step": 136630
    },
    {
      "epoch": 16.442839951865224,
      "grad_norm": 6905.46337890625,
      "learning_rate": 0.0001877553863804333,
      "loss": 7.4347,
      "step": 136640
    },
    {
      "epoch": 16.444043321299638,
      "grad_norm": 466.9425354003906,
      "learning_rate": 0.000187753562252991,
      "loss": 7.4828,
      "step": 136650
    },
    {
      "epoch": 16.445246690734056,
      "grad_norm": 738.4029541015625,
      "learning_rate": 0.0001877517379985476,
      "loss": 7.4964,
      "step": 136660
    },
    {
      "epoch": 16.446450060168473,
      "grad_norm": 1243.8538818359375,
      "learning_rate": 0.0001877499136171058,
      "loss": 7.4728,
      "step": 136670
    },
    {
      "epoch": 16.447653429602887,
      "grad_norm": 1072.4976806640625,
      "learning_rate": 0.00018774808910866818,
      "loss": 7.538,
      "step": 136680
    },
    {
      "epoch": 16.448856799037305,
      "grad_norm": 1277.53857421875,
      "learning_rate": 0.00018774626447323742,
      "loss": 7.447,
      "step": 136690
    },
    {
      "epoch": 16.450060168471722,
      "grad_norm": 585.4208374023438,
      "learning_rate": 0.00018774443971081618,
      "loss": 7.5286,
      "step": 136700
    },
    {
      "epoch": 16.451263537906136,
      "grad_norm": 598.9535522460938,
      "learning_rate": 0.00018774261482140705,
      "loss": 7.5233,
      "step": 136710
    },
    {
      "epoch": 16.452466907340554,
      "grad_norm": 517.5478515625,
      "learning_rate": 0.00018774078980501272,
      "loss": 7.4624,
      "step": 136720
    },
    {
      "epoch": 16.45367027677497,
      "grad_norm": 509.94024658203125,
      "learning_rate": 0.00018773896466163574,
      "loss": 7.4874,
      "step": 136730
    },
    {
      "epoch": 16.454873646209386,
      "grad_norm": 923.160888671875,
      "learning_rate": 0.00018773713939127887,
      "loss": 7.5426,
      "step": 136740
    },
    {
      "epoch": 16.456077015643803,
      "grad_norm": 312.4687194824219,
      "learning_rate": 0.0001877353139939447,
      "loss": 7.4742,
      "step": 136750
    },
    {
      "epoch": 16.45728038507822,
      "grad_norm": 355.81109619140625,
      "learning_rate": 0.00018773348846963582,
      "loss": 7.4346,
      "step": 136760
    },
    {
      "epoch": 16.458483754512635,
      "grad_norm": 2573.20849609375,
      "learning_rate": 0.00018773166281835498,
      "loss": 7.4636,
      "step": 136770
    },
    {
      "epoch": 16.459687123947052,
      "grad_norm": 1106.6849365234375,
      "learning_rate": 0.00018772983704010474,
      "loss": 7.3676,
      "step": 136780
    },
    {
      "epoch": 16.460890493381466,
      "grad_norm": 329.6548156738281,
      "learning_rate": 0.00018772801113488776,
      "loss": 7.4621,
      "step": 136790
    },
    {
      "epoch": 16.462093862815884,
      "grad_norm": 7716.13525390625,
      "learning_rate": 0.0001877261851027067,
      "loss": 7.5916,
      "step": 136800
    },
    {
      "epoch": 16.4632972322503,
      "grad_norm": 1859.53857421875,
      "learning_rate": 0.00018772435894356417,
      "loss": 7.5168,
      "step": 136810
    },
    {
      "epoch": 16.464500601684716,
      "grad_norm": 501.2537841796875,
      "learning_rate": 0.00018772253265746284,
      "loss": 7.3892,
      "step": 136820
    },
    {
      "epoch": 16.465703971119133,
      "grad_norm": 3446.930419921875,
      "learning_rate": 0.00018772070624440538,
      "loss": 7.3543,
      "step": 136830
    },
    {
      "epoch": 16.46690734055355,
      "grad_norm": 831.8310546875,
      "learning_rate": 0.00018771887970439437,
      "loss": 7.4879,
      "step": 136840
    },
    {
      "epoch": 16.468110709987965,
      "grad_norm": 4313.0439453125,
      "learning_rate": 0.0001877170530374325,
      "loss": 7.394,
      "step": 136850
    },
    {
      "epoch": 16.469314079422382,
      "grad_norm": 992.0599975585938,
      "learning_rate": 0.00018771522624352237,
      "loss": 7.4155,
      "step": 136860
    },
    {
      "epoch": 16.4705174488568,
      "grad_norm": 458.1968078613281,
      "learning_rate": 0.0001877133993226667,
      "loss": 7.2855,
      "step": 136870
    },
    {
      "epoch": 16.471720818291214,
      "grad_norm": 1216.3865966796875,
      "learning_rate": 0.00018771157227486803,
      "loss": 7.3575,
      "step": 136880
    },
    {
      "epoch": 16.47292418772563,
      "grad_norm": 378.1925354003906,
      "learning_rate": 0.0001877097451001291,
      "loss": 7.3525,
      "step": 136890
    },
    {
      "epoch": 16.47412755716005,
      "grad_norm": 624.5488891601562,
      "learning_rate": 0.00018770791779845248,
      "loss": 7.4068,
      "step": 136900
    },
    {
      "epoch": 16.475330926594463,
      "grad_norm": 556.9076538085938,
      "learning_rate": 0.0001877060903698409,
      "loss": 7.3626,
      "step": 136910
    },
    {
      "epoch": 16.47653429602888,
      "grad_norm": 518.5603637695312,
      "learning_rate": 0.00018770426281429693,
      "loss": 7.3611,
      "step": 136920
    },
    {
      "epoch": 16.477737665463298,
      "grad_norm": 622.8832397460938,
      "learning_rate": 0.0001877024351318232,
      "loss": 7.3894,
      "step": 136930
    },
    {
      "epoch": 16.478941034897712,
      "grad_norm": 326.3026123046875,
      "learning_rate": 0.00018770060732242242,
      "loss": 7.3806,
      "step": 136940
    },
    {
      "epoch": 16.48014440433213,
      "grad_norm": 9292.4755859375,
      "learning_rate": 0.00018769877938609722,
      "loss": 7.2729,
      "step": 136950
    },
    {
      "epoch": 16.481347773766547,
      "grad_norm": 344.7122497558594,
      "learning_rate": 0.00018769695132285022,
      "loss": 7.3035,
      "step": 136960
    },
    {
      "epoch": 16.48255114320096,
      "grad_norm": 554.872802734375,
      "learning_rate": 0.0001876951231326841,
      "loss": 7.3936,
      "step": 136970
    },
    {
      "epoch": 16.48375451263538,
      "grad_norm": 516.7557983398438,
      "learning_rate": 0.00018769329481560146,
      "loss": 7.3623,
      "step": 136980
    },
    {
      "epoch": 16.484957882069796,
      "grad_norm": 229.0831298828125,
      "learning_rate": 0.000187691466371605,
      "loss": 7.3547,
      "step": 136990
    },
    {
      "epoch": 16.48616125150421,
      "grad_norm": 198.1752471923828,
      "learning_rate": 0.0001876896378006973,
      "loss": 7.3364,
      "step": 137000
    },
    {
      "epoch": 16.487364620938628,
      "grad_norm": 532.3231201171875,
      "learning_rate": 0.0001876878091028811,
      "loss": 7.3512,
      "step": 137010
    },
    {
      "epoch": 16.488567990373046,
      "grad_norm": 366.5466003417969,
      "learning_rate": 0.00018768598027815893,
      "loss": 7.3665,
      "step": 137020
    },
    {
      "epoch": 16.48977135980746,
      "grad_norm": 9869.470703125,
      "learning_rate": 0.00018768415132653353,
      "loss": 7.3486,
      "step": 137030
    },
    {
      "epoch": 16.490974729241877,
      "grad_norm": 518.7772827148438,
      "learning_rate": 0.0001876823222480075,
      "loss": 7.3746,
      "step": 137040
    },
    {
      "epoch": 16.492178098676295,
      "grad_norm": 16230.76953125,
      "learning_rate": 0.0001876804930425835,
      "loss": 7.407,
      "step": 137050
    },
    {
      "epoch": 16.49338146811071,
      "grad_norm": 141.1302032470703,
      "learning_rate": 0.0001876786637102642,
      "loss": 7.6212,
      "step": 137060
    },
    {
      "epoch": 16.494584837545126,
      "grad_norm": 199.4281463623047,
      "learning_rate": 0.0001876768342510522,
      "loss": 7.5952,
      "step": 137070
    },
    {
      "epoch": 16.495788206979544,
      "grad_norm": 13421.2236328125,
      "learning_rate": 0.0001876750046649502,
      "loss": 7.6444,
      "step": 137080
    },
    {
      "epoch": 16.496991576413958,
      "grad_norm": 194.61570739746094,
      "learning_rate": 0.00018767317495196079,
      "loss": 7.7355,
      "step": 137090
    },
    {
      "epoch": 16.498194945848375,
      "grad_norm": 3275.883544921875,
      "learning_rate": 0.00018767134511208668,
      "loss": 7.5391,
      "step": 137100
    },
    {
      "epoch": 16.499398315282793,
      "grad_norm": 1565.52685546875,
      "learning_rate": 0.00018766951514533048,
      "loss": 7.3521,
      "step": 137110
    },
    {
      "epoch": 16.500601684717207,
      "grad_norm": 528.7994384765625,
      "learning_rate": 0.00018766768505169482,
      "loss": 7.4239,
      "step": 137120
    },
    {
      "epoch": 16.501805054151625,
      "grad_norm": 413.8816833496094,
      "learning_rate": 0.00018766585483118243,
      "loss": 7.421,
      "step": 137130
    },
    {
      "epoch": 16.503008423586042,
      "grad_norm": 172.19161987304688,
      "learning_rate": 0.0001876640244837959,
      "loss": 7.4868,
      "step": 137140
    },
    {
      "epoch": 16.504211793020456,
      "grad_norm": 133.32363891601562,
      "learning_rate": 0.00018766219400953783,
      "loss": 7.3561,
      "step": 137150
    },
    {
      "epoch": 16.505415162454874,
      "grad_norm": 171.27767944335938,
      "learning_rate": 0.00018766036340841095,
      "loss": 7.3943,
      "step": 137160
    },
    {
      "epoch": 16.50661853188929,
      "grad_norm": 94.91278076171875,
      "learning_rate": 0.0001876585326804179,
      "loss": 7.4905,
      "step": 137170
    },
    {
      "epoch": 16.507821901323705,
      "grad_norm": 366.7793884277344,
      "learning_rate": 0.00018765670182556128,
      "loss": 7.4743,
      "step": 137180
    },
    {
      "epoch": 16.509025270758123,
      "grad_norm": 145.08108520507812,
      "learning_rate": 0.00018765487084384378,
      "loss": 7.3883,
      "step": 137190
    },
    {
      "epoch": 16.51022864019254,
      "grad_norm": 180.25515747070312,
      "learning_rate": 0.00018765303973526805,
      "loss": 7.4795,
      "step": 137200
    },
    {
      "epoch": 16.511432009626954,
      "grad_norm": 110.56512451171875,
      "learning_rate": 0.0001876512084998367,
      "loss": 7.3867,
      "step": 137210
    },
    {
      "epoch": 16.512635379061372,
      "grad_norm": 98.25045013427734,
      "learning_rate": 0.00018764937713755243,
      "loss": 7.348,
      "step": 137220
    },
    {
      "epoch": 16.51383874849579,
      "grad_norm": 147.30294799804688,
      "learning_rate": 0.0001876475456484179,
      "loss": 7.3438,
      "step": 137230
    },
    {
      "epoch": 16.515042117930204,
      "grad_norm": 64.69267272949219,
      "learning_rate": 0.00018764571403243573,
      "loss": 7.4821,
      "step": 137240
    },
    {
      "epoch": 16.51624548736462,
      "grad_norm": 418.8008728027344,
      "learning_rate": 0.00018764388228960854,
      "loss": 7.3345,
      "step": 137250
    },
    {
      "epoch": 16.51744885679904,
      "grad_norm": 94.74871826171875,
      "learning_rate": 0.00018764205041993902,
      "loss": 7.296,
      "step": 137260
    },
    {
      "epoch": 16.518652226233453,
      "grad_norm": 85.00372314453125,
      "learning_rate": 0.00018764021842342983,
      "loss": 7.36,
      "step": 137270
    },
    {
      "epoch": 16.51985559566787,
      "grad_norm": 345.0772705078125,
      "learning_rate": 0.0001876383863000836,
      "loss": 7.3627,
      "step": 137280
    },
    {
      "epoch": 16.521058965102288,
      "grad_norm": 95.23140716552734,
      "learning_rate": 0.00018763655404990297,
      "loss": 7.3586,
      "step": 137290
    },
    {
      "epoch": 16.522262334536702,
      "grad_norm": 275.6534118652344,
      "learning_rate": 0.00018763472167289064,
      "loss": 7.3728,
      "step": 137300
    },
    {
      "epoch": 16.52346570397112,
      "grad_norm": 44.374481201171875,
      "learning_rate": 0.0001876328891690492,
      "loss": 7.3683,
      "step": 137310
    },
    {
      "epoch": 16.524669073405537,
      "grad_norm": 73.00506591796875,
      "learning_rate": 0.00018763105653838132,
      "loss": 7.4326,
      "step": 137320
    },
    {
      "epoch": 16.52587244283995,
      "grad_norm": 234.020263671875,
      "learning_rate": 0.0001876292237808897,
      "loss": 7.3152,
      "step": 137330
    },
    {
      "epoch": 16.52707581227437,
      "grad_norm": 45.43013000488281,
      "learning_rate": 0.00018762739089657696,
      "loss": 7.3923,
      "step": 137340
    },
    {
      "epoch": 16.528279181708786,
      "grad_norm": 186.88645935058594,
      "learning_rate": 0.00018762555788544576,
      "loss": 7.4001,
      "step": 137350
    },
    {
      "epoch": 16.5294825511432,
      "grad_norm": 40.556644439697266,
      "learning_rate": 0.00018762372474749874,
      "loss": 7.3879,
      "step": 137360
    },
    {
      "epoch": 16.530685920577618,
      "grad_norm": 38.50154113769531,
      "learning_rate": 0.00018762189148273853,
      "loss": 7.367,
      "step": 137370
    },
    {
      "epoch": 16.531889290012035,
      "grad_norm": 128.13572692871094,
      "learning_rate": 0.00018762005809116783,
      "loss": 7.3556,
      "step": 137380
    },
    {
      "epoch": 16.53309265944645,
      "grad_norm": 50.316162109375,
      "learning_rate": 0.00018761822457278926,
      "loss": 7.4299,
      "step": 137390
    },
    {
      "epoch": 16.534296028880867,
      "grad_norm": 26.99710464477539,
      "learning_rate": 0.00018761639092760551,
      "loss": 7.3604,
      "step": 137400
    },
    {
      "epoch": 16.535499398315284,
      "grad_norm": 51.31673812866211,
      "learning_rate": 0.0001876145571556192,
      "loss": 7.303,
      "step": 137410
    },
    {
      "epoch": 16.5367027677497,
      "grad_norm": 218.17044067382812,
      "learning_rate": 0.00018761272325683296,
      "loss": 7.4029,
      "step": 137420
    },
    {
      "epoch": 16.537906137184116,
      "grad_norm": 143.8904266357422,
      "learning_rate": 0.0001876108892312495,
      "loss": 7.4381,
      "step": 137430
    },
    {
      "epoch": 16.53910950661853,
      "grad_norm": 90.31390380859375,
      "learning_rate": 0.0001876090550788715,
      "loss": 7.3692,
      "step": 137440
    },
    {
      "epoch": 16.540312876052948,
      "grad_norm": 120.62361907958984,
      "learning_rate": 0.0001876072207997015,
      "loss": 7.4033,
      "step": 137450
    },
    {
      "epoch": 16.541516245487365,
      "grad_norm": 54.00306701660156,
      "learning_rate": 0.00018760538639374226,
      "loss": 7.3523,
      "step": 137460
    },
    {
      "epoch": 16.54271961492178,
      "grad_norm": 84.41741180419922,
      "learning_rate": 0.00018760355186099638,
      "loss": 7.3048,
      "step": 137470
    },
    {
      "epoch": 16.543922984356197,
      "grad_norm": 120.92346954345703,
      "learning_rate": 0.00018760171720146653,
      "loss": 7.2617,
      "step": 137480
    },
    {
      "epoch": 16.545126353790614,
      "grad_norm": 116.1357421875,
      "learning_rate": 0.0001875998824151554,
      "loss": 7.3082,
      "step": 137490
    },
    {
      "epoch": 16.54632972322503,
      "grad_norm": 92.02224731445312,
      "learning_rate": 0.00018759804750206557,
      "loss": 7.232,
      "step": 137500
    },
    {
      "epoch": 16.547533092659446,
      "grad_norm": 75.33924102783203,
      "learning_rate": 0.00018759621246219977,
      "loss": 7.3646,
      "step": 137510
    },
    {
      "epoch": 16.548736462093864,
      "grad_norm": 232.46670532226562,
      "learning_rate": 0.0001875943772955606,
      "loss": 7.3562,
      "step": 137520
    },
    {
      "epoch": 16.549939831528278,
      "grad_norm": 98.42095184326172,
      "learning_rate": 0.00018759254200215075,
      "loss": 7.2942,
      "step": 137530
    },
    {
      "epoch": 16.551143200962695,
      "grad_norm": 141.63809204101562,
      "learning_rate": 0.00018759070658197287,
      "loss": 7.3499,
      "step": 137540
    },
    {
      "epoch": 16.552346570397113,
      "grad_norm": 92.96308135986328,
      "learning_rate": 0.0001875888710350296,
      "loss": 7.279,
      "step": 137550
    },
    {
      "epoch": 16.553549939831527,
      "grad_norm": 153.2523651123047,
      "learning_rate": 0.00018758703536132362,
      "loss": 7.3607,
      "step": 137560
    },
    {
      "epoch": 16.554753309265944,
      "grad_norm": 101.3573989868164,
      "learning_rate": 0.00018758519956085757,
      "loss": 7.2989,
      "step": 137570
    },
    {
      "epoch": 16.555956678700362,
      "grad_norm": 138.0144500732422,
      "learning_rate": 0.00018758336363363412,
      "loss": 7.2876,
      "step": 137580
    },
    {
      "epoch": 16.557160048134776,
      "grad_norm": 431.8025207519531,
      "learning_rate": 0.0001875815275796559,
      "loss": 7.2583,
      "step": 137590
    },
    {
      "epoch": 16.558363417569193,
      "grad_norm": 174.82595825195312,
      "learning_rate": 0.0001875796913989256,
      "loss": 7.3036,
      "step": 137600
    },
    {
      "epoch": 16.55956678700361,
      "grad_norm": 116.15555572509766,
      "learning_rate": 0.00018757785509144586,
      "loss": 7.3912,
      "step": 137610
    },
    {
      "epoch": 16.560770156438025,
      "grad_norm": 151.6611785888672,
      "learning_rate": 0.00018757601865721933,
      "loss": 7.2205,
      "step": 137620
    },
    {
      "epoch": 16.561973525872443,
      "grad_norm": 97.33600616455078,
      "learning_rate": 0.0001875741820962487,
      "loss": 7.3213,
      "step": 137630
    },
    {
      "epoch": 16.56317689530686,
      "grad_norm": 180.7718048095703,
      "learning_rate": 0.0001875723454085366,
      "loss": 7.4124,
      "step": 137640
    },
    {
      "epoch": 16.564380264741274,
      "grad_norm": 145.56834411621094,
      "learning_rate": 0.0001875705085940857,
      "loss": 7.335,
      "step": 137650
    },
    {
      "epoch": 16.56558363417569,
      "grad_norm": 205.83348083496094,
      "learning_rate": 0.00018756867165289865,
      "loss": 7.2019,
      "step": 137660
    },
    {
      "epoch": 16.56678700361011,
      "grad_norm": 195.31544494628906,
      "learning_rate": 0.00018756683458497813,
      "loss": 7.2055,
      "step": 137670
    },
    {
      "epoch": 16.567990373044523,
      "grad_norm": 244.1363983154297,
      "learning_rate": 0.00018756499739032675,
      "loss": 7.2788,
      "step": 137680
    },
    {
      "epoch": 16.56919374247894,
      "grad_norm": 231.9351348876953,
      "learning_rate": 0.00018756316006894724,
      "loss": 7.233,
      "step": 137690
    },
    {
      "epoch": 16.57039711191336,
      "grad_norm": 168.2899932861328,
      "learning_rate": 0.0001875613226208422,
      "loss": 7.2498,
      "step": 137700
    },
    {
      "epoch": 16.571600481347772,
      "grad_norm": 181.37579345703125,
      "learning_rate": 0.00018755948504601427,
      "loss": 7.2848,
      "step": 137710
    },
    {
      "epoch": 16.57280385078219,
      "grad_norm": 73.8011474609375,
      "learning_rate": 0.00018755764734446618,
      "loss": 7.4079,
      "step": 137720
    },
    {
      "epoch": 16.574007220216608,
      "grad_norm": 167.99313354492188,
      "learning_rate": 0.00018755580951620052,
      "loss": 7.2984,
      "step": 137730
    },
    {
      "epoch": 16.57521058965102,
      "grad_norm": 744.4867553710938,
      "learning_rate": 0.00018755397156122006,
      "loss": 7.2441,
      "step": 137740
    },
    {
      "epoch": 16.57641395908544,
      "grad_norm": 165.22177124023438,
      "learning_rate": 0.00018755213347952733,
      "loss": 7.3445,
      "step": 137750
    },
    {
      "epoch": 16.577617328519857,
      "grad_norm": 99.83592987060547,
      "learning_rate": 0.00018755029527112507,
      "loss": 7.272,
      "step": 137760
    },
    {
      "epoch": 16.57882069795427,
      "grad_norm": 94.53739166259766,
      "learning_rate": 0.0001875484569360159,
      "loss": 7.3411,
      "step": 137770
    },
    {
      "epoch": 16.58002406738869,
      "grad_norm": 119.02742767333984,
      "learning_rate": 0.0001875466184742025,
      "loss": 7.3016,
      "step": 137780
    },
    {
      "epoch": 16.581227436823106,
      "grad_norm": 99.3770751953125,
      "learning_rate": 0.00018754477988568754,
      "loss": 7.2913,
      "step": 137790
    },
    {
      "epoch": 16.58243080625752,
      "grad_norm": 314.3997802734375,
      "learning_rate": 0.00018754294117047366,
      "loss": 7.3973,
      "step": 137800
    },
    {
      "epoch": 16.583634175691937,
      "grad_norm": 180.70556640625,
      "learning_rate": 0.00018754110232856354,
      "loss": 7.3071,
      "step": 137810
    },
    {
      "epoch": 16.584837545126355,
      "grad_norm": 74.38983917236328,
      "learning_rate": 0.00018753926335995984,
      "loss": 7.3638,
      "step": 137820
    },
    {
      "epoch": 16.58604091456077,
      "grad_norm": 249.79318237304688,
      "learning_rate": 0.00018753742426466522,
      "loss": 7.3613,
      "step": 137830
    },
    {
      "epoch": 16.587244283995187,
      "grad_norm": 109.54365539550781,
      "learning_rate": 0.0001875355850426823,
      "loss": 7.3336,
      "step": 137840
    },
    {
      "epoch": 16.588447653429604,
      "grad_norm": 135.13758850097656,
      "learning_rate": 0.0001875337456940138,
      "loss": 7.2774,
      "step": 137850
    },
    {
      "epoch": 16.589651022864018,
      "grad_norm": 89.30496215820312,
      "learning_rate": 0.00018753190621866234,
      "loss": 7.3111,
      "step": 137860
    },
    {
      "epoch": 16.590854392298436,
      "grad_norm": 212.9723663330078,
      "learning_rate": 0.00018753006661663063,
      "loss": 7.2669,
      "step": 137870
    },
    {
      "epoch": 16.592057761732853,
      "grad_norm": 73.3120346069336,
      "learning_rate": 0.00018752822688792128,
      "loss": 7.347,
      "step": 137880
    },
    {
      "epoch": 16.593261131167267,
      "grad_norm": 48.44794845581055,
      "learning_rate": 0.000187526387032537,
      "loss": 7.3232,
      "step": 137890
    },
    {
      "epoch": 16.594464500601685,
      "grad_norm": 62.431514739990234,
      "learning_rate": 0.00018752454705048038,
      "loss": 7.2734,
      "step": 137900
    },
    {
      "epoch": 16.595667870036102,
      "grad_norm": 89.42121887207031,
      "learning_rate": 0.00018752270694175418,
      "loss": 7.3049,
      "step": 137910
    },
    {
      "epoch": 16.596871239470516,
      "grad_norm": 113.3212890625,
      "learning_rate": 0.000187520866706361,
      "loss": 7.3334,
      "step": 137920
    },
    {
      "epoch": 16.598074608904934,
      "grad_norm": 205.4388427734375,
      "learning_rate": 0.0001875190263443035,
      "loss": 7.3752,
      "step": 137930
    },
    {
      "epoch": 16.59927797833935,
      "grad_norm": 171.0076904296875,
      "learning_rate": 0.00018751718585558437,
      "loss": 7.2799,
      "step": 137940
    },
    {
      "epoch": 16.600481347773766,
      "grad_norm": 128.2551727294922,
      "learning_rate": 0.00018751534524020627,
      "loss": 7.3735,
      "step": 137950
    },
    {
      "epoch": 16.601684717208183,
      "grad_norm": 56.19858932495117,
      "learning_rate": 0.00018751350449817185,
      "loss": 7.3556,
      "step": 137960
    },
    {
      "epoch": 16.6028880866426,
      "grad_norm": 50.323246002197266,
      "learning_rate": 0.0001875116636294838,
      "loss": 7.2606,
      "step": 137970
    },
    {
      "epoch": 16.604091456077015,
      "grad_norm": 72.1199722290039,
      "learning_rate": 0.00018750982263414476,
      "loss": 7.2897,
      "step": 137980
    },
    {
      "epoch": 16.605294825511432,
      "grad_norm": 134.43150329589844,
      "learning_rate": 0.00018750798151215742,
      "loss": 7.3223,
      "step": 137990
    },
    {
      "epoch": 16.60649819494585,
      "grad_norm": 86.16520690917969,
      "learning_rate": 0.00018750614026352436,
      "loss": 7.4149,
      "step": 138000
    },
    {
      "epoch": 16.607701564380264,
      "grad_norm": 73.39897918701172,
      "learning_rate": 0.00018750429888824837,
      "loss": 7.4453,
      "step": 138010
    },
    {
      "epoch": 16.60890493381468,
      "grad_norm": 53.49974060058594,
      "learning_rate": 0.00018750245738633205,
      "loss": 7.3445,
      "step": 138020
    },
    {
      "epoch": 16.6101083032491,
      "grad_norm": 108.8686752319336,
      "learning_rate": 0.00018750061575777808,
      "loss": 7.2752,
      "step": 138030
    },
    {
      "epoch": 16.611311672683513,
      "grad_norm": 109.46723937988281,
      "learning_rate": 0.00018749877400258907,
      "loss": 7.3848,
      "step": 138040
    },
    {
      "epoch": 16.61251504211793,
      "grad_norm": 159.08604431152344,
      "learning_rate": 0.00018749693212076775,
      "loss": 7.3612,
      "step": 138050
    },
    {
      "epoch": 16.613718411552348,
      "grad_norm": 75.06536102294922,
      "learning_rate": 0.00018749509011231678,
      "loss": 7.2738,
      "step": 138060
    },
    {
      "epoch": 16.614921780986762,
      "grad_norm": 118.68558502197266,
      "learning_rate": 0.00018749324797723882,
      "loss": 7.3183,
      "step": 138070
    },
    {
      "epoch": 16.61612515042118,
      "grad_norm": 103.16372680664062,
      "learning_rate": 0.0001874914057155365,
      "loss": 7.3244,
      "step": 138080
    },
    {
      "epoch": 16.617328519855597,
      "grad_norm": 106.84732055664062,
      "learning_rate": 0.00018748956332721253,
      "loss": 7.2596,
      "step": 138090
    },
    {
      "epoch": 16.61853188929001,
      "grad_norm": 110.65084075927734,
      "learning_rate": 0.00018748772081226956,
      "loss": 7.2644,
      "step": 138100
    },
    {
      "epoch": 16.61973525872443,
      "grad_norm": 78.4281234741211,
      "learning_rate": 0.0001874858781707103,
      "loss": 7.3037,
      "step": 138110
    },
    {
      "epoch": 16.620938628158846,
      "grad_norm": 70.9840087890625,
      "learning_rate": 0.00018748403540253732,
      "loss": 7.199,
      "step": 138120
    },
    {
      "epoch": 16.62214199759326,
      "grad_norm": 96.12006378173828,
      "learning_rate": 0.00018748219250775335,
      "loss": 7.3796,
      "step": 138130
    },
    {
      "epoch": 16.623345367027678,
      "grad_norm": 102.58830261230469,
      "learning_rate": 0.00018748034948636106,
      "loss": 7.2708,
      "step": 138140
    },
    {
      "epoch": 16.624548736462096,
      "grad_norm": 164.28958129882812,
      "learning_rate": 0.0001874785063383631,
      "loss": 7.308,
      "step": 138150
    },
    {
      "epoch": 16.62575210589651,
      "grad_norm": 135.41746520996094,
      "learning_rate": 0.00018747666306376216,
      "loss": 7.2117,
      "step": 138160
    },
    {
      "epoch": 16.626955475330927,
      "grad_norm": 127.80256652832031,
      "learning_rate": 0.00018747481966256086,
      "loss": 7.2532,
      "step": 138170
    },
    {
      "epoch": 16.628158844765345,
      "grad_norm": 120.26117706298828,
      "learning_rate": 0.0001874729761347619,
      "loss": 7.3886,
      "step": 138180
    },
    {
      "epoch": 16.62936221419976,
      "grad_norm": 111.0550765991211,
      "learning_rate": 0.00018747113248036797,
      "loss": 7.4187,
      "step": 138190
    },
    {
      "epoch": 16.630565583634176,
      "grad_norm": 134.6328582763672,
      "learning_rate": 0.0001874692886993817,
      "loss": 7.3399,
      "step": 138200
    },
    {
      "epoch": 16.63176895306859,
      "grad_norm": 209.6514892578125,
      "learning_rate": 0.0001874674447918058,
      "loss": 7.2014,
      "step": 138210
    },
    {
      "epoch": 16.632972322503008,
      "grad_norm": 130.1690673828125,
      "learning_rate": 0.00018746560075764287,
      "loss": 7.195,
      "step": 138220
    },
    {
      "epoch": 16.634175691937426,
      "grad_norm": 85.39344787597656,
      "learning_rate": 0.00018746375659689567,
      "loss": 7.1786,
      "step": 138230
    },
    {
      "epoch": 16.63537906137184,
      "grad_norm": 199.42642211914062,
      "learning_rate": 0.0001874619123095668,
      "loss": 7.322,
      "step": 138240
    },
    {
      "epoch": 16.636582430806257,
      "grad_norm": 114.37007904052734,
      "learning_rate": 0.00018746006789565893,
      "loss": 7.3211,
      "step": 138250
    },
    {
      "epoch": 16.637785800240675,
      "grad_norm": 54.989017486572266,
      "learning_rate": 0.00018745822335517476,
      "loss": 7.31,
      "step": 138260
    },
    {
      "epoch": 16.63898916967509,
      "grad_norm": 56.570369720458984,
      "learning_rate": 0.00018745637868811696,
      "loss": 7.3263,
      "step": 138270
    },
    {
      "epoch": 16.640192539109506,
      "grad_norm": 115.92820739746094,
      "learning_rate": 0.0001874545338944882,
      "loss": 7.3327,
      "step": 138280
    },
    {
      "epoch": 16.641395908543924,
      "grad_norm": 224.2471160888672,
      "learning_rate": 0.0001874526889742911,
      "loss": 7.2967,
      "step": 138290
    },
    {
      "epoch": 16.642599277978338,
      "grad_norm": 79.9986801147461,
      "learning_rate": 0.0001874508439275284,
      "loss": 7.2535,
      "step": 138300
    },
    {
      "epoch": 16.643802647412755,
      "grad_norm": 53.839508056640625,
      "learning_rate": 0.00018744899875420268,
      "loss": 7.3326,
      "step": 138310
    },
    {
      "epoch": 16.645006016847173,
      "grad_norm": 139.7207794189453,
      "learning_rate": 0.00018744715345431673,
      "loss": 7.2236,
      "step": 138320
    },
    {
      "epoch": 16.646209386281587,
      "grad_norm": 141.90843200683594,
      "learning_rate": 0.00018744530802787315,
      "loss": 7.2426,
      "step": 138330
    },
    {
      "epoch": 16.647412755716005,
      "grad_norm": 133.6788787841797,
      "learning_rate": 0.0001874434624748746,
      "loss": 7.287,
      "step": 138340
    },
    {
      "epoch": 16.648616125150422,
      "grad_norm": 81.70609283447266,
      "learning_rate": 0.0001874416167953238,
      "loss": 7.2922,
      "step": 138350
    },
    {
      "epoch": 16.649819494584836,
      "grad_norm": 92.37747192382812,
      "learning_rate": 0.00018743977098922336,
      "loss": 7.1355,
      "step": 138360
    },
    {
      "epoch": 16.651022864019254,
      "grad_norm": 94.87637329101562,
      "learning_rate": 0.000187437925056576,
      "loss": 7.2832,
      "step": 138370
    },
    {
      "epoch": 16.65222623345367,
      "grad_norm": 256.49774169921875,
      "learning_rate": 0.00018743607899738434,
      "loss": 7.3185,
      "step": 138380
    },
    {
      "epoch": 16.653429602888085,
      "grad_norm": 99.84320068359375,
      "learning_rate": 0.00018743423281165114,
      "loss": 7.2878,
      "step": 138390
    },
    {
      "epoch": 16.654632972322503,
      "grad_norm": 172.68467712402344,
      "learning_rate": 0.000187432386499379,
      "loss": 7.3679,
      "step": 138400
    },
    {
      "epoch": 16.65583634175692,
      "grad_norm": 136.68402099609375,
      "learning_rate": 0.0001874305400605706,
      "loss": 7.3502,
      "step": 138410
    },
    {
      "epoch": 16.657039711191334,
      "grad_norm": 198.687744140625,
      "learning_rate": 0.00018742869349522862,
      "loss": 7.3141,
      "step": 138420
    },
    {
      "epoch": 16.658243080625752,
      "grad_norm": 106.51041412353516,
      "learning_rate": 0.00018742684680335577,
      "loss": 7.3389,
      "step": 138430
    },
    {
      "epoch": 16.65944645006017,
      "grad_norm": 94.6019058227539,
      "learning_rate": 0.00018742499998495465,
      "loss": 7.3798,
      "step": 138440
    },
    {
      "epoch": 16.660649819494584,
      "grad_norm": 108.6557846069336,
      "learning_rate": 0.00018742315304002798,
      "loss": 7.4042,
      "step": 138450
    },
    {
      "epoch": 16.661853188929,
      "grad_norm": 111.88874816894531,
      "learning_rate": 0.00018742130596857842,
      "loss": 7.2755,
      "step": 138460
    },
    {
      "epoch": 16.66305655836342,
      "grad_norm": 73.07344055175781,
      "learning_rate": 0.00018741945877060864,
      "loss": 7.2987,
      "step": 138470
    },
    {
      "epoch": 16.664259927797833,
      "grad_norm": 107.49629211425781,
      "learning_rate": 0.00018741761144612133,
      "loss": 7.2433,
      "step": 138480
    },
    {
      "epoch": 16.66546329723225,
      "grad_norm": 87.2340316772461,
      "learning_rate": 0.00018741576399511914,
      "loss": 7.2603,
      "step": 138490
    },
    {
      "epoch": 16.666666666666668,
      "grad_norm": 47.034210205078125,
      "learning_rate": 0.0001874139164176048,
      "loss": 7.3143,
      "step": 138500
    },
    {
      "epoch": 16.667870036101082,
      "grad_norm": 458.09979248046875,
      "learning_rate": 0.00018741206871358091,
      "loss": 7.2992,
      "step": 138510
    },
    {
      "epoch": 16.6690734055355,
      "grad_norm": 197.75437927246094,
      "learning_rate": 0.00018741022088305018,
      "loss": 7.277,
      "step": 138520
    },
    {
      "epoch": 16.670276774969917,
      "grad_norm": 106.57793426513672,
      "learning_rate": 0.0001874083729260153,
      "loss": 7.2749,
      "step": 138530
    },
    {
      "epoch": 16.67148014440433,
      "grad_norm": 248.29006958007812,
      "learning_rate": 0.00018740652484247887,
      "loss": 7.2091,
      "step": 138540
    },
    {
      "epoch": 16.67268351383875,
      "grad_norm": 44.29582595825195,
      "learning_rate": 0.0001874046766324437,
      "loss": 7.194,
      "step": 138550
    },
    {
      "epoch": 16.673886883273166,
      "grad_norm": 127.11905670166016,
      "learning_rate": 0.00018740282829591228,
      "loss": 7.2697,
      "step": 138560
    },
    {
      "epoch": 16.67509025270758,
      "grad_norm": 105.27174377441406,
      "learning_rate": 0.00018740097983288746,
      "loss": 7.3125,
      "step": 138570
    },
    {
      "epoch": 16.676293622141998,
      "grad_norm": 94.6681137084961,
      "learning_rate": 0.00018739913124337181,
      "loss": 7.1667,
      "step": 138580
    },
    {
      "epoch": 16.677496991576415,
      "grad_norm": 113.34056854248047,
      "learning_rate": 0.00018739728252736807,
      "loss": 7.2745,
      "step": 138590
    },
    {
      "epoch": 16.67870036101083,
      "grad_norm": 109.09080505371094,
      "learning_rate": 0.00018739543368487887,
      "loss": 7.2274,
      "step": 138600
    },
    {
      "epoch": 16.679903730445247,
      "grad_norm": 135.49853515625,
      "learning_rate": 0.00018739358471590688,
      "loss": 7.2697,
      "step": 138610
    },
    {
      "epoch": 16.681107099879664,
      "grad_norm": 73.32743835449219,
      "learning_rate": 0.00018739173562045482,
      "loss": 7.2995,
      "step": 138620
    },
    {
      "epoch": 16.68231046931408,
      "grad_norm": 89.45295715332031,
      "learning_rate": 0.00018738988639852534,
      "loss": 7.4163,
      "step": 138630
    },
    {
      "epoch": 16.683513838748496,
      "grad_norm": 218.6761016845703,
      "learning_rate": 0.00018738803705012114,
      "loss": 7.2111,
      "step": 138640
    },
    {
      "epoch": 16.684717208182914,
      "grad_norm": 124.73004150390625,
      "learning_rate": 0.00018738618757524483,
      "loss": 7.2773,
      "step": 138650
    },
    {
      "epoch": 16.685920577617328,
      "grad_norm": 302.06170654296875,
      "learning_rate": 0.00018738433797389913,
      "loss": 7.3111,
      "step": 138660
    },
    {
      "epoch": 16.687123947051745,
      "grad_norm": 119.40280151367188,
      "learning_rate": 0.00018738248824608675,
      "loss": 7.3417,
      "step": 138670
    },
    {
      "epoch": 16.688327316486163,
      "grad_norm": 96.8246078491211,
      "learning_rate": 0.00018738063839181035,
      "loss": 7.2684,
      "step": 138680
    },
    {
      "epoch": 16.689530685920577,
      "grad_norm": 236.0146026611328,
      "learning_rate": 0.00018737878841107257,
      "loss": 7.2409,
      "step": 138690
    },
    {
      "epoch": 16.690734055354994,
      "grad_norm": 137.79690551757812,
      "learning_rate": 0.00018737693830387611,
      "loss": 7.4255,
      "step": 138700
    },
    {
      "epoch": 16.691937424789412,
      "grad_norm": 174.46075439453125,
      "learning_rate": 0.00018737508807022363,
      "loss": 7.2236,
      "step": 138710
    },
    {
      "epoch": 16.693140794223826,
      "grad_norm": 180.52059936523438,
      "learning_rate": 0.00018737323771011785,
      "loss": 7.2406,
      "step": 138720
    },
    {
      "epoch": 16.694344163658243,
      "grad_norm": 130.89598083496094,
      "learning_rate": 0.00018737138722356144,
      "loss": 7.2732,
      "step": 138730
    },
    {
      "epoch": 16.69554753309266,
      "grad_norm": 271.3072509765625,
      "learning_rate": 0.00018736953661055702,
      "loss": 7.2451,
      "step": 138740
    },
    {
      "epoch": 16.696750902527075,
      "grad_norm": 176.72035217285156,
      "learning_rate": 0.0001873676858711073,
      "loss": 7.2727,
      "step": 138750
    },
    {
      "epoch": 16.697954271961493,
      "grad_norm": 143.26882934570312,
      "learning_rate": 0.00018736583500521503,
      "loss": 7.328,
      "step": 138760
    },
    {
      "epoch": 16.69915764139591,
      "grad_norm": 245.9073028564453,
      "learning_rate": 0.00018736398401288277,
      "loss": 7.246,
      "step": 138770
    },
    {
      "epoch": 16.700361010830324,
      "grad_norm": 103.39151763916016,
      "learning_rate": 0.0001873621328941133,
      "loss": 7.134,
      "step": 138780
    },
    {
      "epoch": 16.70156438026474,
      "grad_norm": 170.048828125,
      "learning_rate": 0.00018736028164890921,
      "loss": 7.2998,
      "step": 138790
    },
    {
      "epoch": 16.70276774969916,
      "grad_norm": 202.63087463378906,
      "learning_rate": 0.00018735843027727325,
      "loss": 7.2928,
      "step": 138800
    },
    {
      "epoch": 16.703971119133573,
      "grad_norm": 163.97991943359375,
      "learning_rate": 0.00018735657877920806,
      "loss": 7.3331,
      "step": 138810
    },
    {
      "epoch": 16.70517448856799,
      "grad_norm": 186.88934326171875,
      "learning_rate": 0.00018735472715471636,
      "loss": 7.2235,
      "step": 138820
    },
    {
      "epoch": 16.706377858002405,
      "grad_norm": 173.6659698486328,
      "learning_rate": 0.0001873528754038008,
      "loss": 7.3578,
      "step": 138830
    },
    {
      "epoch": 16.707581227436823,
      "grad_norm": 143.7120361328125,
      "learning_rate": 0.00018735102352646402,
      "loss": 7.4287,
      "step": 138840
    },
    {
      "epoch": 16.70878459687124,
      "grad_norm": 141.8917999267578,
      "learning_rate": 0.00018734917152270878,
      "loss": 7.302,
      "step": 138850
    },
    {
      "epoch": 16.709987966305654,
      "grad_norm": 252.0840301513672,
      "learning_rate": 0.0001873473193925377,
      "loss": 7.3326,
      "step": 138860
    },
    {
      "epoch": 16.71119133574007,
      "grad_norm": 306.8090515136719,
      "learning_rate": 0.0001873454671359535,
      "loss": 7.2765,
      "step": 138870
    },
    {
      "epoch": 16.71239470517449,
      "grad_norm": 259.6101379394531,
      "learning_rate": 0.00018734361475295885,
      "loss": 7.2738,
      "step": 138880
    },
    {
      "epoch": 16.713598074608903,
      "grad_norm": 221.7580108642578,
      "learning_rate": 0.0001873417622435564,
      "loss": 7.2437,
      "step": 138890
    },
    {
      "epoch": 16.71480144404332,
      "grad_norm": 161.8363494873047,
      "learning_rate": 0.00018733990960774887,
      "loss": 7.4207,
      "step": 138900
    },
    {
      "epoch": 16.71600481347774,
      "grad_norm": 275.39947509765625,
      "learning_rate": 0.00018733805684553896,
      "loss": 7.3731,
      "step": 138910
    },
    {
      "epoch": 16.717208182912152,
      "grad_norm": 107.85295867919922,
      "learning_rate": 0.00018733620395692928,
      "loss": 7.2704,
      "step": 138920
    },
    {
      "epoch": 16.71841155234657,
      "grad_norm": 67.6468734741211,
      "learning_rate": 0.00018733435094192254,
      "loss": 7.1798,
      "step": 138930
    },
    {
      "epoch": 16.719614921780988,
      "grad_norm": 93.60075378417969,
      "learning_rate": 0.00018733249780052144,
      "loss": 7.2046,
      "step": 138940
    },
    {
      "epoch": 16.7208182912154,
      "grad_norm": 109.11235809326172,
      "learning_rate": 0.00018733064453272869,
      "loss": 7.2176,
      "step": 138950
    },
    {
      "epoch": 16.72202166064982,
      "grad_norm": 86.03272247314453,
      "learning_rate": 0.0001873287911385469,
      "loss": 7.3452,
      "step": 138960
    },
    {
      "epoch": 16.723225030084237,
      "grad_norm": 100.9232177734375,
      "learning_rate": 0.00018732693761797878,
      "loss": 7.2866,
      "step": 138970
    },
    {
      "epoch": 16.72442839951865,
      "grad_norm": 48.10356521606445,
      "learning_rate": 0.00018732508397102709,
      "loss": 7.2352,
      "step": 138980
    },
    {
      "epoch": 16.72563176895307,
      "grad_norm": 30.741086959838867,
      "learning_rate": 0.00018732323019769438,
      "loss": 7.3659,
      "step": 138990
    },
    {
      "epoch": 16.726835138387486,
      "grad_norm": 194.4628448486328,
      "learning_rate": 0.0001873213762979834,
      "loss": 7.28,
      "step": 139000
    },
    {
      "epoch": 16.7280385078219,
      "grad_norm": 110.37733459472656,
      "learning_rate": 0.00018731952227189685,
      "loss": 7.3293,
      "step": 139010
    },
    {
      "epoch": 16.729241877256317,
      "grad_norm": 104.7942886352539,
      "learning_rate": 0.00018731766811943737,
      "loss": 7.2848,
      "step": 139020
    },
    {
      "epoch": 16.730445246690735,
      "grad_norm": 48.08375549316406,
      "learning_rate": 0.0001873158138406077,
      "loss": 7.3379,
      "step": 139030
    },
    {
      "epoch": 16.73164861612515,
      "grad_norm": 53.7038459777832,
      "learning_rate": 0.00018731395943541045,
      "loss": 7.2115,
      "step": 139040
    },
    {
      "epoch": 16.732851985559567,
      "grad_norm": 54.888431549072266,
      "learning_rate": 0.00018731210490384833,
      "loss": 7.2915,
      "step": 139050
    },
    {
      "epoch": 16.734055354993984,
      "grad_norm": 54.29003143310547,
      "learning_rate": 0.00018731025024592407,
      "loss": 7.279,
      "step": 139060
    },
    {
      "epoch": 16.735258724428398,
      "grad_norm": 130.731201171875,
      "learning_rate": 0.00018730839546164034,
      "loss": 7.3312,
      "step": 139070
    },
    {
      "epoch": 16.736462093862816,
      "grad_norm": 126.02831268310547,
      "learning_rate": 0.00018730654055099978,
      "loss": 7.2774,
      "step": 139080
    },
    {
      "epoch": 16.737665463297233,
      "grad_norm": 126.5333480834961,
      "learning_rate": 0.00018730468551400513,
      "loss": 7.2487,
      "step": 139090
    },
    {
      "epoch": 16.738868832731647,
      "grad_norm": 106.4551773071289,
      "learning_rate": 0.000187302830350659,
      "loss": 7.2209,
      "step": 139100
    },
    {
      "epoch": 16.740072202166065,
      "grad_norm": 185.82272338867188,
      "learning_rate": 0.00018730097506096415,
      "loss": 7.3364,
      "step": 139110
    },
    {
      "epoch": 16.741275571600482,
      "grad_norm": 519.2950439453125,
      "learning_rate": 0.00018729911964492324,
      "loss": 7.2096,
      "step": 139120
    },
    {
      "epoch": 16.742478941034896,
      "grad_norm": 224.7523956298828,
      "learning_rate": 0.00018729726410253892,
      "loss": 7.2224,
      "step": 139130
    },
    {
      "epoch": 16.743682310469314,
      "grad_norm": 362.90472412109375,
      "learning_rate": 0.00018729540843381393,
      "loss": 7.2313,
      "step": 139140
    },
    {
      "epoch": 16.74488567990373,
      "grad_norm": 104.93167877197266,
      "learning_rate": 0.00018729355263875092,
      "loss": 7.2703,
      "step": 139150
    },
    {
      "epoch": 16.746089049338146,
      "grad_norm": 91.54993438720703,
      "learning_rate": 0.0001872916967173526,
      "loss": 7.1763,
      "step": 139160
    },
    {
      "epoch": 16.747292418772563,
      "grad_norm": 162.8675537109375,
      "learning_rate": 0.00018728984066962166,
      "loss": 7.2305,
      "step": 139170
    },
    {
      "epoch": 16.74849578820698,
      "grad_norm": 52.1177978515625,
      "learning_rate": 0.00018728798449556074,
      "loss": 7.199,
      "step": 139180
    },
    {
      "epoch": 16.749699157641395,
      "grad_norm": 346.7183837890625,
      "learning_rate": 0.00018728612819517254,
      "loss": 7.2614,
      "step": 139190
    },
    {
      "epoch": 16.750902527075812,
      "grad_norm": 133.7086944580078,
      "learning_rate": 0.0001872842717684598,
      "loss": 7.2071,
      "step": 139200
    },
    {
      "epoch": 16.75210589651023,
      "grad_norm": 108.26351928710938,
      "learning_rate": 0.00018728241521542515,
      "loss": 7.3074,
      "step": 139210
    },
    {
      "epoch": 16.753309265944644,
      "grad_norm": 87.5799789428711,
      "learning_rate": 0.0001872805585360713,
      "loss": 7.2241,
      "step": 139220
    },
    {
      "epoch": 16.75451263537906,
      "grad_norm": 65.84819793701172,
      "learning_rate": 0.00018727870173040092,
      "loss": 7.183,
      "step": 139230
    },
    {
      "epoch": 16.75571600481348,
      "grad_norm": 164.9320831298828,
      "learning_rate": 0.00018727684479841674,
      "loss": 7.141,
      "step": 139240
    },
    {
      "epoch": 16.756919374247893,
      "grad_norm": 196.724853515625,
      "learning_rate": 0.00018727498774012138,
      "loss": 7.234,
      "step": 139250
    },
    {
      "epoch": 16.75812274368231,
      "grad_norm": 358.0588684082031,
      "learning_rate": 0.00018727313055551758,
      "loss": 7.1495,
      "step": 139260
    },
    {
      "epoch": 16.759326113116728,
      "grad_norm": 203.6941680908203,
      "learning_rate": 0.00018727127324460801,
      "loss": 7.272,
      "step": 139270
    },
    {
      "epoch": 16.760529482551142,
      "grad_norm": 203.03640747070312,
      "learning_rate": 0.00018726941580739538,
      "loss": 7.1698,
      "step": 139280
    },
    {
      "epoch": 16.76173285198556,
      "grad_norm": 260.54595947265625,
      "learning_rate": 0.00018726755824388235,
      "loss": 7.2256,
      "step": 139290
    },
    {
      "epoch": 16.762936221419977,
      "grad_norm": 148.79818725585938,
      "learning_rate": 0.0001872657005540716,
      "loss": 7.3323,
      "step": 139300
    },
    {
      "epoch": 16.76413959085439,
      "grad_norm": 848.3822021484375,
      "learning_rate": 0.00018726384273796588,
      "loss": 7.2515,
      "step": 139310
    },
    {
      "epoch": 16.76534296028881,
      "grad_norm": 136.0645751953125,
      "learning_rate": 0.0001872619847955678,
      "loss": 7.2265,
      "step": 139320
    },
    {
      "epoch": 16.766546329723226,
      "grad_norm": 145.14022827148438,
      "learning_rate": 0.0001872601267268801,
      "loss": 7.2299,
      "step": 139330
    },
    {
      "epoch": 16.76774969915764,
      "grad_norm": 69.01983642578125,
      "learning_rate": 0.0001872582685319054,
      "loss": 7.2612,
      "step": 139340
    },
    {
      "epoch": 16.768953068592058,
      "grad_norm": 110.63954162597656,
      "learning_rate": 0.0001872564102106465,
      "loss": 7.3256,
      "step": 139350
    },
    {
      "epoch": 16.770156438026476,
      "grad_norm": 931.7457885742188,
      "learning_rate": 0.00018725455176310602,
      "loss": 7.2788,
      "step": 139360
    },
    {
      "epoch": 16.77135980746089,
      "grad_norm": 197.388916015625,
      "learning_rate": 0.00018725269318928665,
      "loss": 7.1983,
      "step": 139370
    },
    {
      "epoch": 16.772563176895307,
      "grad_norm": 70.93016052246094,
      "learning_rate": 0.0001872508344891911,
      "loss": 7.2098,
      "step": 139380
    },
    {
      "epoch": 16.773766546329725,
      "grad_norm": 75.49091339111328,
      "learning_rate": 0.00018724897566282207,
      "loss": 7.1266,
      "step": 139390
    },
    {
      "epoch": 16.77496991576414,
      "grad_norm": 201.10296630859375,
      "learning_rate": 0.0001872471167101822,
      "loss": 7.2378,
      "step": 139400
    },
    {
      "epoch": 16.776173285198556,
      "grad_norm": 341.234375,
      "learning_rate": 0.00018724525763127422,
      "loss": 7.3339,
      "step": 139410
    },
    {
      "epoch": 16.777376654632974,
      "grad_norm": 684.5357055664062,
      "learning_rate": 0.0001872433984261008,
      "loss": 7.276,
      "step": 139420
    },
    {
      "epoch": 16.778580024067388,
      "grad_norm": 444.3345642089844,
      "learning_rate": 0.00018724153909466468,
      "loss": 7.3303,
      "step": 139430
    },
    {
      "epoch": 16.779783393501805,
      "grad_norm": 502.9717712402344,
      "learning_rate": 0.0001872396796369685,
      "loss": 7.2057,
      "step": 139440
    },
    {
      "epoch": 16.780986762936223,
      "grad_norm": 772.0226440429688,
      "learning_rate": 0.00018723782005301492,
      "loss": 7.1651,
      "step": 139450
    },
    {
      "epoch": 16.782190132370637,
      "grad_norm": 416.9214782714844,
      "learning_rate": 0.00018723596034280675,
      "loss": 7.2508,
      "step": 139460
    },
    {
      "epoch": 16.783393501805055,
      "grad_norm": 205.43426513671875,
      "learning_rate": 0.00018723410050634653,
      "loss": 7.23,
      "step": 139470
    },
    {
      "epoch": 16.784596871239472,
      "grad_norm": 383.4138488769531,
      "learning_rate": 0.00018723224054363708,
      "loss": 7.145,
      "step": 139480
    },
    {
      "epoch": 16.785800240673886,
      "grad_norm": 317.0182800292969,
      "learning_rate": 0.00018723038045468103,
      "loss": 7.2024,
      "step": 139490
    },
    {
      "epoch": 16.787003610108304,
      "grad_norm": 689.5606689453125,
      "learning_rate": 0.0001872285202394811,
      "loss": 7.2854,
      "step": 139500
    },
    {
      "epoch": 16.78820697954272,
      "grad_norm": 777.2088012695312,
      "learning_rate": 0.00018722665989803994,
      "loss": 7.2215,
      "step": 139510
    },
    {
      "epoch": 16.789410348977135,
      "grad_norm": 290.6332702636719,
      "learning_rate": 0.00018722479943036026,
      "loss": 7.27,
      "step": 139520
    },
    {
      "epoch": 16.790613718411553,
      "grad_norm": 259.62860107421875,
      "learning_rate": 0.00018722293883644477,
      "loss": 7.2449,
      "step": 139530
    },
    {
      "epoch": 16.79181708784597,
      "grad_norm": 732.4400024414062,
      "learning_rate": 0.00018722107811629618,
      "loss": 7.2064,
      "step": 139540
    },
    {
      "epoch": 16.793020457280385,
      "grad_norm": 387.3924865722656,
      "learning_rate": 0.00018721921726991712,
      "loss": 7.415,
      "step": 139550
    },
    {
      "epoch": 16.794223826714802,
      "grad_norm": 299.3282165527344,
      "learning_rate": 0.00018721735629731033,
      "loss": 7.2283,
      "step": 139560
    },
    {
      "epoch": 16.79542719614922,
      "grad_norm": 1383.3133544921875,
      "learning_rate": 0.00018721549519847848,
      "loss": 7.2772,
      "step": 139570
    },
    {
      "epoch": 16.796630565583634,
      "grad_norm": 468.2200012207031,
      "learning_rate": 0.0001872136339734243,
      "loss": 7.2919,
      "step": 139580
    },
    {
      "epoch": 16.79783393501805,
      "grad_norm": 412.17950439453125,
      "learning_rate": 0.00018721177262215046,
      "loss": 7.1286,
      "step": 139590
    },
    {
      "epoch": 16.799037304452465,
      "grad_norm": 303.6209716796875,
      "learning_rate": 0.00018720991114465964,
      "loss": 7.3256,
      "step": 139600
    },
    {
      "epoch": 16.800240673886883,
      "grad_norm": 245.2219696044922,
      "learning_rate": 0.00018720804954095455,
      "loss": 7.214,
      "step": 139610
    },
    {
      "epoch": 16.8014440433213,
      "grad_norm": 377.19915771484375,
      "learning_rate": 0.00018720618781103786,
      "loss": 7.196,
      "step": 139620
    },
    {
      "epoch": 16.802647412755714,
      "grad_norm": 278.16094970703125,
      "learning_rate": 0.00018720432595491232,
      "loss": 7.2349,
      "step": 139630
    },
    {
      "epoch": 16.803850782190132,
      "grad_norm": 664.1884765625,
      "learning_rate": 0.0001872024639725806,
      "loss": 7.1976,
      "step": 139640
    },
    {
      "epoch": 16.80505415162455,
      "grad_norm": 215.2874298095703,
      "learning_rate": 0.00018720060186404536,
      "loss": 7.103,
      "step": 139650
    },
    {
      "epoch": 16.806257521058964,
      "grad_norm": 231.60414123535156,
      "learning_rate": 0.00018719873962930932,
      "loss": 7.2328,
      "step": 139660
    },
    {
      "epoch": 16.80746089049338,
      "grad_norm": 236.93356323242188,
      "learning_rate": 0.00018719687726837515,
      "loss": 7.15,
      "step": 139670
    },
    {
      "epoch": 16.8086642599278,
      "grad_norm": 231.94137573242188,
      "learning_rate": 0.00018719501478124562,
      "loss": 7.3687,
      "step": 139680
    },
    {
      "epoch": 16.809867629362213,
      "grad_norm": 216.42724609375,
      "learning_rate": 0.00018719315216792336,
      "loss": 7.2421,
      "step": 139690
    },
    {
      "epoch": 16.81107099879663,
      "grad_norm": 218.34716796875,
      "learning_rate": 0.00018719128942841108,
      "loss": 7.1991,
      "step": 139700
    },
    {
      "epoch": 16.812274368231048,
      "grad_norm": 251.12322998046875,
      "learning_rate": 0.00018718942656271146,
      "loss": 7.2053,
      "step": 139710
    },
    {
      "epoch": 16.813477737665462,
      "grad_norm": 118.98465728759766,
      "learning_rate": 0.0001871875635708272,
      "loss": 7.2229,
      "step": 139720
    },
    {
      "epoch": 16.81468110709988,
      "grad_norm": 229.80421447753906,
      "learning_rate": 0.00018718570045276104,
      "loss": 7.3189,
      "step": 139730
    },
    {
      "epoch": 16.815884476534297,
      "grad_norm": 333.42333984375,
      "learning_rate": 0.00018718383720851565,
      "loss": 7.3007,
      "step": 139740
    },
    {
      "epoch": 16.81708784596871,
      "grad_norm": 175.28524780273438,
      "learning_rate": 0.00018718197383809374,
      "loss": 7.2219,
      "step": 139750
    },
    {
      "epoch": 16.81829121540313,
      "grad_norm": 857.9053955078125,
      "learning_rate": 0.00018718011034149794,
      "loss": 7.2849,
      "step": 139760
    },
    {
      "epoch": 16.819494584837546,
      "grad_norm": 595.9356689453125,
      "learning_rate": 0.00018717824671873103,
      "loss": 7.2844,
      "step": 139770
    },
    {
      "epoch": 16.82069795427196,
      "grad_norm": 616.8970947265625,
      "learning_rate": 0.00018717638296979566,
      "loss": 7.2565,
      "step": 139780
    },
    {
      "epoch": 16.821901323706378,
      "grad_norm": 323.5877990722656,
      "learning_rate": 0.00018717451909469455,
      "loss": 7.2226,
      "step": 139790
    },
    {
      "epoch": 16.823104693140795,
      "grad_norm": 301.0235290527344,
      "learning_rate": 0.0001871726550934304,
      "loss": 7.2297,
      "step": 139800
    },
    {
      "epoch": 16.82430806257521,
      "grad_norm": 476.3625793457031,
      "learning_rate": 0.00018717079096600584,
      "loss": 7.253,
      "step": 139810
    },
    {
      "epoch": 16.825511432009627,
      "grad_norm": 681.467041015625,
      "learning_rate": 0.00018716892671242368,
      "loss": 7.2507,
      "step": 139820
    },
    {
      "epoch": 16.826714801444044,
      "grad_norm": 567.54150390625,
      "learning_rate": 0.00018716706233268653,
      "loss": 7.2234,
      "step": 139830
    },
    {
      "epoch": 16.82791817087846,
      "grad_norm": 454.80859375,
      "learning_rate": 0.00018716519782679714,
      "loss": 7.1869,
      "step": 139840
    },
    {
      "epoch": 16.829121540312876,
      "grad_norm": 407.3795471191406,
      "learning_rate": 0.00018716333319475817,
      "loss": 7.188,
      "step": 139850
    },
    {
      "epoch": 16.830324909747294,
      "grad_norm": 660.4546508789062,
      "learning_rate": 0.00018716146843657233,
      "loss": 7.3038,
      "step": 139860
    },
    {
      "epoch": 16.831528279181708,
      "grad_norm": 380.2582092285156,
      "learning_rate": 0.00018715960355224237,
      "loss": 7.2803,
      "step": 139870
    },
    {
      "epoch": 16.832731648616125,
      "grad_norm": 585.8861694335938,
      "learning_rate": 0.00018715773854177092,
      "loss": 7.2756,
      "step": 139880
    },
    {
      "epoch": 16.833935018050543,
      "grad_norm": 1130.9735107421875,
      "learning_rate": 0.0001871558734051607,
      "loss": 7.2597,
      "step": 139890
    },
    {
      "epoch": 16.835138387484957,
      "grad_norm": 790.6238403320312,
      "learning_rate": 0.00018715400814241443,
      "loss": 7.2254,
      "step": 139900
    },
    {
      "epoch": 16.836341756919374,
      "grad_norm": 508.5173645019531,
      "learning_rate": 0.00018715214275353477,
      "loss": 7.3063,
      "step": 139910
    },
    {
      "epoch": 16.837545126353792,
      "grad_norm": 1146.9901123046875,
      "learning_rate": 0.00018715027723852442,
      "loss": 7.257,
      "step": 139920
    },
    {
      "epoch": 16.838748495788206,
      "grad_norm": 397.6241455078125,
      "learning_rate": 0.00018714841159738613,
      "loss": 7.1613,
      "step": 139930
    },
    {
      "epoch": 16.839951865222623,
      "grad_norm": 611.0847778320312,
      "learning_rate": 0.00018714654583012262,
      "loss": 7.1651,
      "step": 139940
    },
    {
      "epoch": 16.84115523465704,
      "grad_norm": 371.8877868652344,
      "learning_rate": 0.00018714467993673645,
      "loss": 7.2738,
      "step": 139950
    },
    {
      "epoch": 16.842358604091455,
      "grad_norm": 366.5208435058594,
      "learning_rate": 0.00018714281391723047,
      "loss": 7.3467,
      "step": 139960
    },
    {
      "epoch": 16.843561973525873,
      "grad_norm": 253.11341857910156,
      "learning_rate": 0.00018714094777160734,
      "loss": 7.2335,
      "step": 139970
    },
    {
      "epoch": 16.84476534296029,
      "grad_norm": 254.44203186035156,
      "learning_rate": 0.0001871390814998697,
      "loss": 7.2224,
      "step": 139980
    },
    {
      "epoch": 16.845968712394704,
      "grad_norm": 408.25006103515625,
      "learning_rate": 0.00018713721510202032,
      "loss": 7.2601,
      "step": 139990
    },
    {
      "epoch": 16.84717208182912,
      "grad_norm": 105.78641510009766,
      "learning_rate": 0.00018713534857806184,
      "loss": 7.2055,
      "step": 140000
    },
    {
      "epoch": 16.84837545126354,
      "grad_norm": 169.25010681152344,
      "learning_rate": 0.00018713348192799707,
      "loss": 7.2599,
      "step": 140010
    },
    {
      "epoch": 16.849578820697953,
      "grad_norm": 163.738525390625,
      "learning_rate": 0.00018713161515182859,
      "loss": 7.1699,
      "step": 140020
    },
    {
      "epoch": 16.85078219013237,
      "grad_norm": 118.00897216796875,
      "learning_rate": 0.00018712974824955915,
      "loss": 7.2125,
      "step": 140030
    },
    {
      "epoch": 16.85198555956679,
      "grad_norm": 179.82626342773438,
      "learning_rate": 0.00018712788122119144,
      "loss": 7.2152,
      "step": 140040
    },
    {
      "epoch": 16.853188929001202,
      "grad_norm": 230.5179901123047,
      "learning_rate": 0.0001871260140667282,
      "loss": 7.1367,
      "step": 140050
    },
    {
      "epoch": 16.85439229843562,
      "grad_norm": 117.68388366699219,
      "learning_rate": 0.0001871241467861721,
      "loss": 7.2819,
      "step": 140060
    },
    {
      "epoch": 16.855595667870038,
      "grad_norm": 102.51017761230469,
      "learning_rate": 0.00018712227937952585,
      "loss": 7.1079,
      "step": 140070
    },
    {
      "epoch": 16.85679903730445,
      "grad_norm": 91.59984588623047,
      "learning_rate": 0.00018712041184679214,
      "loss": 7.3432,
      "step": 140080
    },
    {
      "epoch": 16.85800240673887,
      "grad_norm": 132.9410400390625,
      "learning_rate": 0.00018711854418797372,
      "loss": 7.2687,
      "step": 140090
    },
    {
      "epoch": 16.859205776173287,
      "grad_norm": 122.43665313720703,
      "learning_rate": 0.00018711667640307323,
      "loss": 7.203,
      "step": 140100
    },
    {
      "epoch": 16.8604091456077,
      "grad_norm": 102.4945297241211,
      "learning_rate": 0.00018711480849209342,
      "loss": 7.2202,
      "step": 140110
    },
    {
      "epoch": 16.86161251504212,
      "grad_norm": 74.01268768310547,
      "learning_rate": 0.00018711294045503695,
      "loss": 7.2334,
      "step": 140120
    },
    {
      "epoch": 16.862815884476536,
      "grad_norm": 93.73290252685547,
      "learning_rate": 0.00018711107229190654,
      "loss": 7.3383,
      "step": 140130
    },
    {
      "epoch": 16.86401925391095,
      "grad_norm": 130.47085571289062,
      "learning_rate": 0.00018710920400270492,
      "loss": 7.298,
      "step": 140140
    },
    {
      "epoch": 16.865222623345367,
      "grad_norm": 130.94932556152344,
      "learning_rate": 0.0001871073355874348,
      "loss": 7.2934,
      "step": 140150
    },
    {
      "epoch": 16.866425992779785,
      "grad_norm": 430.466796875,
      "learning_rate": 0.0001871054670460988,
      "loss": 7.2353,
      "step": 140160
    },
    {
      "epoch": 16.8676293622142,
      "grad_norm": 130.47991943359375,
      "learning_rate": 0.00018710359837869974,
      "loss": 7.298,
      "step": 140170
    },
    {
      "epoch": 16.868832731648617,
      "grad_norm": 232.81826782226562,
      "learning_rate": 0.00018710172958524024,
      "loss": 7.2429,
      "step": 140180
    },
    {
      "epoch": 16.870036101083034,
      "grad_norm": 210.23997497558594,
      "learning_rate": 0.00018709986066572302,
      "loss": 7.196,
      "step": 140190
    },
    {
      "epoch": 16.871239470517448,
      "grad_norm": 78.10020446777344,
      "learning_rate": 0.00018709799162015082,
      "loss": 7.2458,
      "step": 140200
    },
    {
      "epoch": 16.872442839951866,
      "grad_norm": 191.275146484375,
      "learning_rate": 0.00018709612244852632,
      "loss": 7.1331,
      "step": 140210
    },
    {
      "epoch": 16.87364620938628,
      "grad_norm": 203.09738159179688,
      "learning_rate": 0.00018709425315085223,
      "loss": 7.2239,
      "step": 140220
    },
    {
      "epoch": 16.874849578820697,
      "grad_norm": 193.85629272460938,
      "learning_rate": 0.00018709238372713124,
      "loss": 7.3018,
      "step": 140230
    },
    {
      "epoch": 16.876052948255115,
      "grad_norm": 244.79351806640625,
      "learning_rate": 0.00018709051417736608,
      "loss": 7.1613,
      "step": 140240
    },
    {
      "epoch": 16.87725631768953,
      "grad_norm": 158.73377990722656,
      "learning_rate": 0.00018708864450155943,
      "loss": 7.3027,
      "step": 140250
    },
    {
      "epoch": 16.878459687123947,
      "grad_norm": 129.6427459716797,
      "learning_rate": 0.00018708677469971404,
      "loss": 7.2607,
      "step": 140260
    },
    {
      "epoch": 16.879663056558364,
      "grad_norm": 462.04388427734375,
      "learning_rate": 0.00018708490477183256,
      "loss": 7.2748,
      "step": 140270
    },
    {
      "epoch": 16.880866425992778,
      "grad_norm": 512.9693603515625,
      "learning_rate": 0.00018708303471791772,
      "loss": 7.2684,
      "step": 140280
    },
    {
      "epoch": 16.882069795427196,
      "grad_norm": 284.9821472167969,
      "learning_rate": 0.00018708116453797224,
      "loss": 7.2547,
      "step": 140290
    },
    {
      "epoch": 16.883273164861613,
      "grad_norm": 129.4783477783203,
      "learning_rate": 0.0001870792942319988,
      "loss": 7.2807,
      "step": 140300
    },
    {
      "epoch": 16.884476534296027,
      "grad_norm": 135.47842407226562,
      "learning_rate": 0.00018707742380000015,
      "loss": 7.3678,
      "step": 140310
    },
    {
      "epoch": 16.885679903730445,
      "grad_norm": 238.98765563964844,
      "learning_rate": 0.00018707555324197893,
      "loss": 7.2005,
      "step": 140320
    },
    {
      "epoch": 16.886883273164862,
      "grad_norm": 338.9595031738281,
      "learning_rate": 0.0001870736825579379,
      "loss": 7.2527,
      "step": 140330
    },
    {
      "epoch": 16.888086642599276,
      "grad_norm": 392.41009521484375,
      "learning_rate": 0.00018707181174787977,
      "loss": 7.2183,
      "step": 140340
    },
    {
      "epoch": 16.889290012033694,
      "grad_norm": 114.52117156982422,
      "learning_rate": 0.00018706994081180725,
      "loss": 7.3447,
      "step": 140350
    },
    {
      "epoch": 16.89049338146811,
      "grad_norm": 264.9649963378906,
      "learning_rate": 0.00018706806974972298,
      "loss": 7.2551,
      "step": 140360
    },
    {
      "epoch": 16.891696750902526,
      "grad_norm": 103.57659912109375,
      "learning_rate": 0.00018706619856162976,
      "loss": 7.2541,
      "step": 140370
    },
    {
      "epoch": 16.892900120336943,
      "grad_norm": 78.29728698730469,
      "learning_rate": 0.00018706432724753023,
      "loss": 7.3447,
      "step": 140380
    },
    {
      "epoch": 16.89410348977136,
      "grad_norm": 44.53462600708008,
      "learning_rate": 0.0001870624558074271,
      "loss": 7.3042,
      "step": 140390
    },
    {
      "epoch": 16.895306859205775,
      "grad_norm": 114.2386245727539,
      "learning_rate": 0.00018706058424132317,
      "loss": 7.2138,
      "step": 140400
    },
    {
      "epoch": 16.896510228640192,
      "grad_norm": 271.8332824707031,
      "learning_rate": 0.000187058712549221,
      "loss": 7.2662,
      "step": 140410
    },
    {
      "epoch": 16.89771359807461,
      "grad_norm": 101.64602661132812,
      "learning_rate": 0.00018705684073112342,
      "loss": 7.2041,
      "step": 140420
    },
    {
      "epoch": 16.898916967509024,
      "grad_norm": 197.1746063232422,
      "learning_rate": 0.0001870549687870331,
      "loss": 7.1597,
      "step": 140430
    },
    {
      "epoch": 16.90012033694344,
      "grad_norm": 119.66224670410156,
      "learning_rate": 0.00018705309671695274,
      "loss": 7.3107,
      "step": 140440
    },
    {
      "epoch": 16.90132370637786,
      "grad_norm": 107.16404724121094,
      "learning_rate": 0.00018705122452088505,
      "loss": 7.3012,
      "step": 140450
    },
    {
      "epoch": 16.902527075812273,
      "grad_norm": 112.4430160522461,
      "learning_rate": 0.00018704935219883274,
      "loss": 7.3344,
      "step": 140460
    },
    {
      "epoch": 16.90373044524669,
      "grad_norm": 147.1563262939453,
      "learning_rate": 0.00018704747975079858,
      "loss": 7.306,
      "step": 140470
    },
    {
      "epoch": 16.904933814681108,
      "grad_norm": 190.45547485351562,
      "learning_rate": 0.00018704560717678516,
      "loss": 7.2292,
      "step": 140480
    },
    {
      "epoch": 16.906137184115522,
      "grad_norm": 197.283935546875,
      "learning_rate": 0.0001870437344767953,
      "loss": 7.1409,
      "step": 140490
    },
    {
      "epoch": 16.90734055354994,
      "grad_norm": 162.4819793701172,
      "learning_rate": 0.00018704186165083165,
      "loss": 7.2511,
      "step": 140500
    },
    {
      "epoch": 16.908543922984357,
      "grad_norm": 196.75277709960938,
      "learning_rate": 0.0001870399886988969,
      "loss": 7.357,
      "step": 140510
    },
    {
      "epoch": 16.90974729241877,
      "grad_norm": 253.63694763183594,
      "learning_rate": 0.00018703811562099384,
      "loss": 7.2279,
      "step": 140520
    },
    {
      "epoch": 16.91095066185319,
      "grad_norm": 140.5008087158203,
      "learning_rate": 0.00018703624241712513,
      "loss": 7.2603,
      "step": 140530
    },
    {
      "epoch": 16.912154031287606,
      "grad_norm": 159.239990234375,
      "learning_rate": 0.00018703436908729348,
      "loss": 7.3026,
      "step": 140540
    },
    {
      "epoch": 16.91335740072202,
      "grad_norm": 249.13462829589844,
      "learning_rate": 0.00018703249563150161,
      "loss": 7.3498,
      "step": 140550
    },
    {
      "epoch": 16.914560770156438,
      "grad_norm": 155.85171508789062,
      "learning_rate": 0.00018703062204975223,
      "loss": 7.2365,
      "step": 140560
    },
    {
      "epoch": 16.915764139590856,
      "grad_norm": 494.0457763671875,
      "learning_rate": 0.00018702874834204808,
      "loss": 7.3239,
      "step": 140570
    },
    {
      "epoch": 16.91696750902527,
      "grad_norm": 658.1851196289062,
      "learning_rate": 0.0001870268745083918,
      "loss": 7.3253,
      "step": 140580
    },
    {
      "epoch": 16.918170878459687,
      "grad_norm": 362.6427307128906,
      "learning_rate": 0.00018702500054878617,
      "loss": 7.3368,
      "step": 140590
    },
    {
      "epoch": 16.919374247894105,
      "grad_norm": 192.82269287109375,
      "learning_rate": 0.0001870231264632339,
      "loss": 7.2243,
      "step": 140600
    },
    {
      "epoch": 16.92057761732852,
      "grad_norm": 413.574951171875,
      "learning_rate": 0.00018702125225173764,
      "loss": 7.2197,
      "step": 140610
    },
    {
      "epoch": 16.921780986762936,
      "grad_norm": 347.7251892089844,
      "learning_rate": 0.0001870193779143002,
      "loss": 7.193,
      "step": 140620
    },
    {
      "epoch": 16.922984356197354,
      "grad_norm": 585.346923828125,
      "learning_rate": 0.0001870175034509242,
      "loss": 7.3094,
      "step": 140630
    },
    {
      "epoch": 16.924187725631768,
      "grad_norm": 1707.060302734375,
      "learning_rate": 0.00018701562886161237,
      "loss": 7.2115,
      "step": 140640
    },
    {
      "epoch": 16.925391095066185,
      "grad_norm": 464.30328369140625,
      "learning_rate": 0.00018701375414636746,
      "loss": 7.1449,
      "step": 140650
    },
    {
      "epoch": 16.926594464500603,
      "grad_norm": 334.6609191894531,
      "learning_rate": 0.00018701187930519214,
      "loss": 7.244,
      "step": 140660
    },
    {
      "epoch": 16.927797833935017,
      "grad_norm": 379.38336181640625,
      "learning_rate": 0.00018701000433808917,
      "loss": 7.2555,
      "step": 140670
    },
    {
      "epoch": 16.929001203369435,
      "grad_norm": 750.4374389648438,
      "learning_rate": 0.00018700812924506125,
      "loss": 7.249,
      "step": 140680
    },
    {
      "epoch": 16.930204572803852,
      "grad_norm": 1935.0679931640625,
      "learning_rate": 0.0001870062540261111,
      "loss": 7.2245,
      "step": 140690
    },
    {
      "epoch": 16.931407942238266,
      "grad_norm": 1020.1409912109375,
      "learning_rate": 0.0001870043786812414,
      "loss": 7.2246,
      "step": 140700
    },
    {
      "epoch": 16.932611311672684,
      "grad_norm": 2117.8369140625,
      "learning_rate": 0.00018700250321045486,
      "loss": 7.2156,
      "step": 140710
    },
    {
      "epoch": 16.9338146811071,
      "grad_norm": 507.71435546875,
      "learning_rate": 0.00018700062761375423,
      "loss": 7.1958,
      "step": 140720
    },
    {
      "epoch": 16.935018050541515,
      "grad_norm": 448.57574462890625,
      "learning_rate": 0.00018699875189114224,
      "loss": 7.2327,
      "step": 140730
    },
    {
      "epoch": 16.936221419975933,
      "grad_norm": 584.8023681640625,
      "learning_rate": 0.00018699687604262154,
      "loss": 7.2211,
      "step": 140740
    },
    {
      "epoch": 16.93742478941035,
      "grad_norm": 502.99395751953125,
      "learning_rate": 0.0001869950000681949,
      "loss": 7.2682,
      "step": 140750
    },
    {
      "epoch": 16.938628158844764,
      "grad_norm": 662.310302734375,
      "learning_rate": 0.00018699312396786502,
      "loss": 7.2251,
      "step": 140760
    },
    {
      "epoch": 16.939831528279182,
      "grad_norm": 1193.97314453125,
      "learning_rate": 0.00018699124774163462,
      "loss": 7.2697,
      "step": 140770
    },
    {
      "epoch": 16.9410348977136,
      "grad_norm": 692.0593872070312,
      "learning_rate": 0.0001869893713895064,
      "loss": 7.2434,
      "step": 140780
    },
    {
      "epoch": 16.942238267148014,
      "grad_norm": 2674.326171875,
      "learning_rate": 0.00018698749491148306,
      "loss": 7.2773,
      "step": 140790
    },
    {
      "epoch": 16.94344163658243,
      "grad_norm": 1171.132568359375,
      "learning_rate": 0.00018698561830756736,
      "loss": 7.2644,
      "step": 140800
    },
    {
      "epoch": 16.94464500601685,
      "grad_norm": 1005.2243041992188,
      "learning_rate": 0.000186983741577762,
      "loss": 7.3121,
      "step": 140810
    },
    {
      "epoch": 16.945848375451263,
      "grad_norm": 702.234375,
      "learning_rate": 0.00018698186472206966,
      "loss": 7.1652,
      "step": 140820
    },
    {
      "epoch": 16.94705174488568,
      "grad_norm": 1431.6015625,
      "learning_rate": 0.0001869799877404931,
      "loss": 7.3035,
      "step": 140830
    },
    {
      "epoch": 16.948255114320098,
      "grad_norm": 1350.1988525390625,
      "learning_rate": 0.00018697811063303504,
      "loss": 7.1727,
      "step": 140840
    },
    {
      "epoch": 16.949458483754512,
      "grad_norm": 988.340576171875,
      "learning_rate": 0.00018697623339969816,
      "loss": 7.314,
      "step": 140850
    },
    {
      "epoch": 16.95066185318893,
      "grad_norm": 1350.4931640625,
      "learning_rate": 0.00018697435604048522,
      "loss": 7.2395,
      "step": 140860
    },
    {
      "epoch": 16.951865222623347,
      "grad_norm": 612.1477661132812,
      "learning_rate": 0.00018697247855539888,
      "loss": 7.2238,
      "step": 140870
    },
    {
      "epoch": 16.95306859205776,
      "grad_norm": 1204.4171142578125,
      "learning_rate": 0.00018697060094444192,
      "loss": 7.2783,
      "step": 140880
    },
    {
      "epoch": 16.95427196149218,
      "grad_norm": 551.55859375,
      "learning_rate": 0.00018696872320761703,
      "loss": 7.2574,
      "step": 140890
    },
    {
      "epoch": 16.955475330926596,
      "grad_norm": 885.8955688476562,
      "learning_rate": 0.0001869668453449269,
      "loss": 7.2001,
      "step": 140900
    },
    {
      "epoch": 16.95667870036101,
      "grad_norm": 1194.371337890625,
      "learning_rate": 0.0001869649673563743,
      "loss": 7.2824,
      "step": 140910
    },
    {
      "epoch": 16.957882069795428,
      "grad_norm": 689.4669799804688,
      "learning_rate": 0.00018696308924196192,
      "loss": 7.3292,
      "step": 140920
    },
    {
      "epoch": 16.959085439229845,
      "grad_norm": 2157.862060546875,
      "learning_rate": 0.00018696121100169246,
      "loss": 7.3187,
      "step": 140930
    },
    {
      "epoch": 16.96028880866426,
      "grad_norm": 1226.73388671875,
      "learning_rate": 0.00018695933263556866,
      "loss": 7.3299,
      "step": 140940
    },
    {
      "epoch": 16.961492178098677,
      "grad_norm": 5642.2080078125,
      "learning_rate": 0.00018695745414359322,
      "loss": 7.3384,
      "step": 140950
    },
    {
      "epoch": 16.96269554753309,
      "grad_norm": 1071.24853515625,
      "learning_rate": 0.0001869555755257689,
      "loss": 7.3305,
      "step": 140960
    },
    {
      "epoch": 16.96389891696751,
      "grad_norm": 1440.128173828125,
      "learning_rate": 0.0001869536967820984,
      "loss": 7.2943,
      "step": 140970
    },
    {
      "epoch": 16.965102286401926,
      "grad_norm": 5736.08544921875,
      "learning_rate": 0.00018695181791258443,
      "loss": 7.1581,
      "step": 140980
    },
    {
      "epoch": 16.96630565583634,
      "grad_norm": 2490.807861328125,
      "learning_rate": 0.0001869499389172297,
      "loss": 7.226,
      "step": 140990
    },
    {
      "epoch": 16.967509025270758,
      "grad_norm": 3486.12890625,
      "learning_rate": 0.00018694805979603695,
      "loss": 7.3359,
      "step": 141000
    },
    {
      "epoch": 16.968712394705175,
      "grad_norm": 395.9966735839844,
      "learning_rate": 0.00018694618054900889,
      "loss": 7.3345,
      "step": 141010
    },
    {
      "epoch": 16.96991576413959,
      "grad_norm": 557.5656127929688,
      "learning_rate": 0.0001869443011761482,
      "loss": 7.3539,
      "step": 141020
    },
    {
      "epoch": 16.971119133574007,
      "grad_norm": 155.63076782226562,
      "learning_rate": 0.0001869424216774577,
      "loss": 7.2907,
      "step": 141030
    },
    {
      "epoch": 16.972322503008424,
      "grad_norm": 176.70492553710938,
      "learning_rate": 0.00018694054205294002,
      "loss": 7.2284,
      "step": 141040
    },
    {
      "epoch": 16.97352587244284,
      "grad_norm": 194.2889862060547,
      "learning_rate": 0.00018693866230259793,
      "loss": 7.2408,
      "step": 141050
    },
    {
      "epoch": 16.974729241877256,
      "grad_norm": 191.15797424316406,
      "learning_rate": 0.00018693678242643412,
      "loss": 7.3036,
      "step": 141060
    },
    {
      "epoch": 16.975932611311674,
      "grad_norm": 389.35052490234375,
      "learning_rate": 0.0001869349024244513,
      "loss": 7.2751,
      "step": 141070
    },
    {
      "epoch": 16.977135980746088,
      "grad_norm": 237.237060546875,
      "learning_rate": 0.00018693302229665223,
      "loss": 7.3091,
      "step": 141080
    },
    {
      "epoch": 16.978339350180505,
      "grad_norm": 371.0087890625,
      "learning_rate": 0.00018693114204303962,
      "loss": 7.3141,
      "step": 141090
    },
    {
      "epoch": 16.979542719614923,
      "grad_norm": 159.05126953125,
      "learning_rate": 0.0001869292616636162,
      "loss": 7.2645,
      "step": 141100
    },
    {
      "epoch": 16.980746089049337,
      "grad_norm": 359.671875,
      "learning_rate": 0.00018692738115838467,
      "loss": 7.2667,
      "step": 141110
    },
    {
      "epoch": 16.981949458483754,
      "grad_norm": 911.1619262695312,
      "learning_rate": 0.0001869255005273477,
      "loss": 7.3295,
      "step": 141120
    },
    {
      "epoch": 16.983152827918172,
      "grad_norm": 941.1460571289062,
      "learning_rate": 0.00018692361977050813,
      "loss": 7.2638,
      "step": 141130
    },
    {
      "epoch": 16.984356197352586,
      "grad_norm": 852.6581420898438,
      "learning_rate": 0.0001869217388878686,
      "loss": 7.2092,
      "step": 141140
    },
    {
      "epoch": 16.985559566787003,
      "grad_norm": 1084.5054931640625,
      "learning_rate": 0.00018691985787943187,
      "loss": 7.2554,
      "step": 141150
    },
    {
      "epoch": 16.98676293622142,
      "grad_norm": 3219.71240234375,
      "learning_rate": 0.00018691797674520063,
      "loss": 7.2092,
      "step": 141160
    },
    {
      "epoch": 16.987966305655835,
      "grad_norm": 15177.5029296875,
      "learning_rate": 0.00018691609548517762,
      "loss": 7.235,
      "step": 141170
    },
    {
      "epoch": 16.989169675090253,
      "grad_norm": 2970.762451171875,
      "learning_rate": 0.00018691421409936562,
      "loss": 7.1923,
      "step": 141180
    },
    {
      "epoch": 16.99037304452467,
      "grad_norm": 3648.067626953125,
      "learning_rate": 0.0001869123325877672,
      "loss": 7.2056,
      "step": 141190
    },
    {
      "epoch": 16.991576413959084,
      "grad_norm": 2022.618408203125,
      "learning_rate": 0.00018691045095038523,
      "loss": 7.3552,
      "step": 141200
    },
    {
      "epoch": 16.9927797833935,
      "grad_norm": 3119.40771484375,
      "learning_rate": 0.00018690856918722237,
      "loss": 7.2329,
      "step": 141210
    },
    {
      "epoch": 16.99398315282792,
      "grad_norm": 2291.0751953125,
      "learning_rate": 0.00018690668729828133,
      "loss": 7.2169,
      "step": 141220
    },
    {
      "epoch": 16.995186522262333,
      "grad_norm": 9763.142578125,
      "learning_rate": 0.00018690480528356487,
      "loss": 7.3825,
      "step": 141230
    },
    {
      "epoch": 16.99638989169675,
      "grad_norm": 1863.9749755859375,
      "learning_rate": 0.0001869029231430757,
      "loss": 7.2771,
      "step": 141240
    },
    {
      "epoch": 16.99759326113117,
      "grad_norm": 5123.5166015625,
      "learning_rate": 0.00018690104087681656,
      "loss": 7.3613,
      "step": 141250
    },
    {
      "epoch": 16.998796630565582,
      "grad_norm": 1767.199951171875,
      "learning_rate": 0.00018689915848479015,
      "loss": 7.3014,
      "step": 141260
    },
    {
      "epoch": 17.0,
      "grad_norm": 1562.28857421875,
      "learning_rate": 0.00018689727596699922,
      "loss": 7.3865,
      "step": 141270
    },
    {
      "epoch": 17.0,
      "eval_loss": 7.285952091217041,
      "eval_runtime": 119.1122,
      "eval_samples_per_second": 62.017,
      "eval_steps_per_second": 7.757,
      "step": 141270
    },
    {
      "epoch": 17.001203369434418,
      "grad_norm": 924.876708984375,
      "learning_rate": 0.00018689539332344645,
      "loss": 7.3318,
      "step": 141280
    },
    {
      "epoch": 17.00240673886883,
      "grad_norm": 1089.5457763671875,
      "learning_rate": 0.0001868935105541346,
      "loss": 7.3168,
      "step": 141290
    },
    {
      "epoch": 17.00361010830325,
      "grad_norm": 1816.0389404296875,
      "learning_rate": 0.0001868916276590664,
      "loss": 7.2017,
      "step": 141300
    },
    {
      "epoch": 17.004813477737667,
      "grad_norm": 1234.37158203125,
      "learning_rate": 0.00018688974463824454,
      "loss": 7.3102,
      "step": 141310
    },
    {
      "epoch": 17.00601684717208,
      "grad_norm": 1734.231201171875,
      "learning_rate": 0.00018688786149167178,
      "loss": 7.3916,
      "step": 141320
    },
    {
      "epoch": 17.0072202166065,
      "grad_norm": 1366.3143310546875,
      "learning_rate": 0.00018688597821935084,
      "loss": 7.326,
      "step": 141330
    },
    {
      "epoch": 17.008423586040916,
      "grad_norm": 4636.26123046875,
      "learning_rate": 0.00018688409482128443,
      "loss": 7.3423,
      "step": 141340
    },
    {
      "epoch": 17.00962695547533,
      "grad_norm": 10725.2548828125,
      "learning_rate": 0.0001868822112974753,
      "loss": 7.2779,
      "step": 141350
    },
    {
      "epoch": 17.010830324909747,
      "grad_norm": 24845.275390625,
      "learning_rate": 0.00018688032764792612,
      "loss": 7.4422,
      "step": 141360
    },
    {
      "epoch": 17.012033694344165,
      "grad_norm": 13669.8125,
      "learning_rate": 0.00018687844387263968,
      "loss": 7.4754,
      "step": 141370
    },
    {
      "epoch": 17.01323706377858,
      "grad_norm": 29731.974609375,
      "learning_rate": 0.00018687655997161868,
      "loss": 7.4477,
      "step": 141380
    },
    {
      "epoch": 17.014440433212997,
      "grad_norm": 9919.509765625,
      "learning_rate": 0.00018687467594486585,
      "loss": 7.4055,
      "step": 141390
    },
    {
      "epoch": 17.015643802647414,
      "grad_norm": 17093.2421875,
      "learning_rate": 0.0001868727917923839,
      "loss": 7.3542,
      "step": 141400
    },
    {
      "epoch": 17.016847172081828,
      "grad_norm": 2228155.75,
      "learning_rate": 0.0001868709075141756,
      "loss": 7.7627,
      "step": 141410
    },
    {
      "epoch": 17.018050541516246,
      "grad_norm": 207971.96875,
      "learning_rate": 0.00018686902311024363,
      "loss": 10.0114,
      "step": 141420
    },
    {
      "epoch": 17.019253910950663,
      "grad_norm": 2111354.75,
      "learning_rate": 0.00018686713858059075,
      "loss": 10.3712,
      "step": 141430
    },
    {
      "epoch": 17.020457280385077,
      "grad_norm": 1330060.75,
      "learning_rate": 0.00018686525392521967,
      "loss": 9.8709,
      "step": 141440
    },
    {
      "epoch": 17.021660649819495,
      "grad_norm": 29423556.0,
      "learning_rate": 0.00018686336914413312,
      "loss": 9.4304,
      "step": 141450
    },
    {
      "epoch": 17.022864019253912,
      "grad_norm": 67532224.0,
      "learning_rate": 0.00018686148423733382,
      "loss": 10.2345,
      "step": 141460
    },
    {
      "epoch": 17.024067388688326,
      "grad_norm": 309209.0,
      "learning_rate": 0.0001868595992048245,
      "loss": 10.6619,
      "step": 141470
    },
    {
      "epoch": 17.025270758122744,
      "grad_norm": 476131.34375,
      "learning_rate": 0.00018685771404660794,
      "loss": 10.8618,
      "step": 141480
    },
    {
      "epoch": 17.02647412755716,
      "grad_norm": 21831126.0,
      "learning_rate": 0.00018685582876268677,
      "loss": 11.2351,
      "step": 141490
    },
    {
      "epoch": 17.027677496991576,
      "grad_norm": 10083105.0,
      "learning_rate": 0.00018685394335306378,
      "loss": 10.2909,
      "step": 141500
    },
    {
      "epoch": 17.028880866425993,
      "grad_norm": 153905440.0,
      "learning_rate": 0.00018685205781774174,
      "loss": 11.0587,
      "step": 141510
    },
    {
      "epoch": 17.03008423586041,
      "grad_norm": 630525248.0,
      "learning_rate": 0.00018685017215672327,
      "loss": 11.0391,
      "step": 141520
    },
    {
      "epoch": 17.031287605294825,
      "grad_norm": 63822776.0,
      "learning_rate": 0.00018684828637001118,
      "loss": 10.3231,
      "step": 141530
    },
    {
      "epoch": 17.032490974729242,
      "grad_norm": 60172244.0,
      "learning_rate": 0.00018684640045760816,
      "loss": 9.7273,
      "step": 141540
    },
    {
      "epoch": 17.03369434416366,
      "grad_norm": 428727744.0,
      "learning_rate": 0.000186844514419517,
      "loss": 9.6713,
      "step": 141550
    },
    {
      "epoch": 17.034897713598074,
      "grad_norm": 103376992.0,
      "learning_rate": 0.00018684262825574037,
      "loss": 10.0265,
      "step": 141560
    },
    {
      "epoch": 17.03610108303249,
      "grad_norm": 149720096.0,
      "learning_rate": 0.000186840741966281,
      "loss": 11.8091,
      "step": 141570
    },
    {
      "epoch": 17.03730445246691,
      "grad_norm": 12849386.0,
      "learning_rate": 0.00018683885555114164,
      "loss": 10.4154,
      "step": 141580
    },
    {
      "epoch": 17.038507821901323,
      "grad_norm": 3311020.25,
      "learning_rate": 0.000186836969010325,
      "loss": 9.7652,
      "step": 141590
    },
    {
      "epoch": 17.03971119133574,
      "grad_norm": 18709520.0,
      "learning_rate": 0.00018683508234383387,
      "loss": 8.5957,
      "step": 141600
    },
    {
      "epoch": 17.040914560770158,
      "grad_norm": 3221425.25,
      "learning_rate": 0.0001868331955516709,
      "loss": 8.11,
      "step": 141610
    },
    {
      "epoch": 17.042117930204572,
      "grad_norm": 503154880.0,
      "learning_rate": 0.00018683130863383885,
      "loss": 8.398,
      "step": 141620
    },
    {
      "epoch": 17.04332129963899,
      "grad_norm": 5451205.5,
      "learning_rate": 0.0001868294215903405,
      "loss": 9.0374,
      "step": 141630
    },
    {
      "epoch": 17.044524669073404,
      "grad_norm": 12103708.0,
      "learning_rate": 0.00018682753442117854,
      "loss": 8.1284,
      "step": 141640
    },
    {
      "epoch": 17.04572803850782,
      "grad_norm": 2498935.5,
      "learning_rate": 0.00018682564712635565,
      "loss": 7.9074,
      "step": 141650
    },
    {
      "epoch": 17.04693140794224,
      "grad_norm": 30658082.0,
      "learning_rate": 0.00018682375970587468,
      "loss": 7.8347,
      "step": 141660
    },
    {
      "epoch": 17.048134777376653,
      "grad_norm": 4080987.5,
      "learning_rate": 0.00018682187215973822,
      "loss": 7.7735,
      "step": 141670
    },
    {
      "epoch": 17.04933814681107,
      "grad_norm": 3283819.25,
      "learning_rate": 0.0001868199844879491,
      "loss": 8.0627,
      "step": 141680
    },
    {
      "epoch": 17.050541516245488,
      "grad_norm": 38352371712.0,
      "learning_rate": 0.00018681809669051006,
      "loss": 9.5241,
      "step": 141690
    },
    {
      "epoch": 17.051744885679902,
      "grad_norm": 6710531072.0,
      "learning_rate": 0.00018681620876742375,
      "loss": 10.4074,
      "step": 141700
    },
    {
      "epoch": 17.05294825511432,
      "grad_norm": 490279072.0,
      "learning_rate": 0.000186814320718693,
      "loss": 10.0882,
      "step": 141710
    },
    {
      "epoch": 17.054151624548737,
      "grad_norm": 109023453184.0,
      "learning_rate": 0.00018681243254432047,
      "loss": 9.7834,
      "step": 141720
    },
    {
      "epoch": 17.05535499398315,
      "grad_norm": 496933824.0,
      "learning_rate": 0.00018681054424430892,
      "loss": 9.1749,
      "step": 141730
    },
    {
      "epoch": 17.05655836341757,
      "grad_norm": 3079289088.0,
      "learning_rate": 0.0001868086558186611,
      "loss": 10.0425,
      "step": 141740
    },
    {
      "epoch": 17.057761732851986,
      "grad_norm": 635064896.0,
      "learning_rate": 0.00018680676726737968,
      "loss": 10.9633,
      "step": 141750
    },
    {
      "epoch": 17.0589651022864,
      "grad_norm": 52752132.0,
      "learning_rate": 0.00018680487859046746,
      "loss": 11.445,
      "step": 141760
    },
    {
      "epoch": 17.060168471720818,
      "grad_norm": 86162898944.0,
      "learning_rate": 0.00018680298978792717,
      "loss": 10.9618,
      "step": 141770
    },
    {
      "epoch": 17.061371841155236,
      "grad_norm": 2763698176.0,
      "learning_rate": 0.0001868011008597615,
      "loss": 10.1048,
      "step": 141780
    },
    {
      "epoch": 17.06257521058965,
      "grad_norm": 16464600064.0,
      "learning_rate": 0.0001867992118059732,
      "loss": 9.5028,
      "step": 141790
    },
    {
      "epoch": 17.063778580024067,
      "grad_norm": 3073200128.0,
      "learning_rate": 0.00018679732262656502,
      "loss": 7.5215,
      "step": 141800
    },
    {
      "epoch": 17.064981949458485,
      "grad_norm": 306863.46875,
      "learning_rate": 0.0001867954333215397,
      "loss": 7.384,
      "step": 141810
    },
    {
      "epoch": 17.0661853188929,
      "grad_norm": 4970920.5,
      "learning_rate": 0.00018679354389089991,
      "loss": 7.6664,
      "step": 141820
    },
    {
      "epoch": 17.067388688327316,
      "grad_norm": 18504586.0,
      "learning_rate": 0.00018679165433464846,
      "loss": 7.7047,
      "step": 141830
    },
    {
      "epoch": 17.068592057761734,
      "grad_norm": 47235032.0,
      "learning_rate": 0.00018678976465278809,
      "loss": 7.7192,
      "step": 141840
    },
    {
      "epoch": 17.069795427196148,
      "grad_norm": 323516224.0,
      "learning_rate": 0.00018678787484532146,
      "loss": 7.7239,
      "step": 141850
    },
    {
      "epoch": 17.070998796630565,
      "grad_norm": 220470.0625,
      "learning_rate": 0.00018678598491225136,
      "loss": 7.6575,
      "step": 141860
    },
    {
      "epoch": 17.072202166064983,
      "grad_norm": 353980.90625,
      "learning_rate": 0.00018678409485358054,
      "loss": 7.7066,
      "step": 141870
    },
    {
      "epoch": 17.073405535499397,
      "grad_norm": 28216.509765625,
      "learning_rate": 0.00018678220466931167,
      "loss": 7.6509,
      "step": 141880
    },
    {
      "epoch": 17.074608904933815,
      "grad_norm": 64755.54296875,
      "learning_rate": 0.00018678031435944754,
      "loss": 7.6948,
      "step": 141890
    },
    {
      "epoch": 17.075812274368232,
      "grad_norm": 29359730.0,
      "learning_rate": 0.00018677842392399087,
      "loss": 7.5768,
      "step": 141900
    },
    {
      "epoch": 17.077015643802646,
      "grad_norm": 568256.25,
      "learning_rate": 0.00018677653336294438,
      "loss": 7.6167,
      "step": 141910
    },
    {
      "epoch": 17.078219013237064,
      "grad_norm": 175611.65625,
      "learning_rate": 0.00018677464267631083,
      "loss": 7.6359,
      "step": 141920
    },
    {
      "epoch": 17.07942238267148,
      "grad_norm": 165359.421875,
      "learning_rate": 0.00018677275186409296,
      "loss": 7.6493,
      "step": 141930
    },
    {
      "epoch": 17.080625752105895,
      "grad_norm": 101344.546875,
      "learning_rate": 0.00018677086092629347,
      "loss": 7.6918,
      "step": 141940
    },
    {
      "epoch": 17.081829121540313,
      "grad_norm": 145652.84375,
      "learning_rate": 0.00018676896986291516,
      "loss": 7.6174,
      "step": 141950
    },
    {
      "epoch": 17.08303249097473,
      "grad_norm": 77464.8125,
      "learning_rate": 0.0001867670786739607,
      "loss": 7.6242,
      "step": 141960
    },
    {
      "epoch": 17.084235860409144,
      "grad_norm": 316192.28125,
      "learning_rate": 0.00018676518735943288,
      "loss": 7.6159,
      "step": 141970
    },
    {
      "epoch": 17.085439229843562,
      "grad_norm": 723720.1875,
      "learning_rate": 0.0001867632959193344,
      "loss": 7.6689,
      "step": 141980
    },
    {
      "epoch": 17.08664259927798,
      "grad_norm": 1426618.0,
      "learning_rate": 0.00018676140435366798,
      "loss": 7.6359,
      "step": 141990
    },
    {
      "epoch": 17.087845968712394,
      "grad_norm": 327477.0,
      "learning_rate": 0.00018675951266243642,
      "loss": 7.5401,
      "step": 142000
    },
    {
      "epoch": 17.08904933814681,
      "grad_norm": 24021530.0,
      "learning_rate": 0.00018675762084564243,
      "loss": 7.5873,
      "step": 142010
    },
    {
      "epoch": 17.09025270758123,
      "grad_norm": 307043.25,
      "learning_rate": 0.00018675572890328873,
      "loss": 7.5659,
      "step": 142020
    },
    {
      "epoch": 17.091456077015643,
      "grad_norm": 885694.4375,
      "learning_rate": 0.00018675383683537807,
      "loss": 7.6518,
      "step": 142030
    },
    {
      "epoch": 17.09265944645006,
      "grad_norm": 524654.9375,
      "learning_rate": 0.00018675194464191317,
      "loss": 7.6359,
      "step": 142040
    },
    {
      "epoch": 17.093862815884478,
      "grad_norm": 72734.875,
      "learning_rate": 0.00018675005232289681,
      "loss": 7.6203,
      "step": 142050
    },
    {
      "epoch": 17.095066185318892,
      "grad_norm": 29165812.0,
      "learning_rate": 0.0001867481598783317,
      "loss": 7.5633,
      "step": 142060
    },
    {
      "epoch": 17.09626955475331,
      "grad_norm": 89524.546875,
      "learning_rate": 0.0001867462673082206,
      "loss": 7.5212,
      "step": 142070
    },
    {
      "epoch": 17.097472924187727,
      "grad_norm": 500923.21875,
      "learning_rate": 0.00018674437461256623,
      "loss": 7.7076,
      "step": 142080
    },
    {
      "epoch": 17.09867629362214,
      "grad_norm": 5271867.0,
      "learning_rate": 0.00018674248179137133,
      "loss": 7.6236,
      "step": 142090
    },
    {
      "epoch": 17.09987966305656,
      "grad_norm": 652417.75,
      "learning_rate": 0.00018674058884463862,
      "loss": 7.6297,
      "step": 142100
    },
    {
      "epoch": 17.101083032490976,
      "grad_norm": 22941.08203125,
      "learning_rate": 0.0001867386957723709,
      "loss": 7.6299,
      "step": 142110
    },
    {
      "epoch": 17.10228640192539,
      "grad_norm": 28353.453125,
      "learning_rate": 0.00018673680257457086,
      "loss": 7.5819,
      "step": 142120
    },
    {
      "epoch": 17.103489771359808,
      "grad_norm": 55700.875,
      "learning_rate": 0.00018673490925124125,
      "loss": 7.6439,
      "step": 142130
    },
    {
      "epoch": 17.104693140794225,
      "grad_norm": 615666.875,
      "learning_rate": 0.00018673301580238478,
      "loss": 7.5065,
      "step": 142140
    },
    {
      "epoch": 17.10589651022864,
      "grad_norm": 217125.90625,
      "learning_rate": 0.00018673112222800423,
      "loss": 7.5731,
      "step": 142150
    },
    {
      "epoch": 17.107099879663057,
      "grad_norm": 196665.125,
      "learning_rate": 0.00018672922852810237,
      "loss": 7.5202,
      "step": 142160
    },
    {
      "epoch": 17.108303249097474,
      "grad_norm": 166179.25,
      "learning_rate": 0.0001867273347026819,
      "loss": 7.6613,
      "step": 142170
    },
    {
      "epoch": 17.10950661853189,
      "grad_norm": 96824.5,
      "learning_rate": 0.00018672544075174554,
      "loss": 7.5422,
      "step": 142180
    },
    {
      "epoch": 17.110709987966306,
      "grad_norm": 441479.03125,
      "learning_rate": 0.0001867235466752961,
      "loss": 7.4653,
      "step": 142190
    },
    {
      "epoch": 17.111913357400724,
      "grad_norm": 50175.7890625,
      "learning_rate": 0.00018672165247333623,
      "loss": 7.5489,
      "step": 142200
    },
    {
      "epoch": 17.113116726835138,
      "grad_norm": 69269.8046875,
      "learning_rate": 0.00018671975814586872,
      "loss": 7.6343,
      "step": 142210
    },
    {
      "epoch": 17.114320096269555,
      "grad_norm": 192029.21875,
      "learning_rate": 0.00018671786369289632,
      "loss": 7.5305,
      "step": 142220
    },
    {
      "epoch": 17.115523465703973,
      "grad_norm": 54431.62890625,
      "learning_rate": 0.00018671596911442174,
      "loss": 7.5186,
      "step": 142230
    },
    {
      "epoch": 17.116726835138387,
      "grad_norm": 219534.40625,
      "learning_rate": 0.0001867140744104478,
      "loss": 7.4882,
      "step": 142240
    },
    {
      "epoch": 17.117930204572804,
      "grad_norm": 1309003.25,
      "learning_rate": 0.00018671217958097716,
      "loss": 7.5362,
      "step": 142250
    },
    {
      "epoch": 17.119133574007222,
      "grad_norm": 535729.8125,
      "learning_rate": 0.00018671028462601257,
      "loss": 7.6497,
      "step": 142260
    },
    {
      "epoch": 17.120336943441636,
      "grad_norm": 17054.576171875,
      "learning_rate": 0.0001867083895455568,
      "loss": 7.4007,
      "step": 142270
    },
    {
      "epoch": 17.121540312876053,
      "grad_norm": 18348.041015625,
      "learning_rate": 0.0001867064943396126,
      "loss": 7.4202,
      "step": 142280
    },
    {
      "epoch": 17.12274368231047,
      "grad_norm": 134746.65625,
      "learning_rate": 0.00018670459900818265,
      "loss": 7.5017,
      "step": 142290
    },
    {
      "epoch": 17.123947051744885,
      "grad_norm": 8078.41015625,
      "learning_rate": 0.00018670270355126978,
      "loss": 7.3104,
      "step": 142300
    },
    {
      "epoch": 17.125150421179303,
      "grad_norm": 767163.0,
      "learning_rate": 0.00018670080796887666,
      "loss": 7.4925,
      "step": 142310
    },
    {
      "epoch": 17.126353790613717,
      "grad_norm": 138308.53125,
      "learning_rate": 0.00018669891226100608,
      "loss": 7.4317,
      "step": 142320
    },
    {
      "epoch": 17.127557160048134,
      "grad_norm": 12479.5205078125,
      "learning_rate": 0.00018669701642766076,
      "loss": 7.3732,
      "step": 142330
    },
    {
      "epoch": 17.128760529482552,
      "grad_norm": 157744.296875,
      "learning_rate": 0.00018669512046884347,
      "loss": 7.3109,
      "step": 142340
    },
    {
      "epoch": 17.129963898916966,
      "grad_norm": 12371.865234375,
      "learning_rate": 0.00018669322438455695,
      "loss": 7.423,
      "step": 142350
    },
    {
      "epoch": 17.131167268351383,
      "grad_norm": 12458.7294921875,
      "learning_rate": 0.0001866913281748039,
      "loss": 7.3762,
      "step": 142360
    },
    {
      "epoch": 17.1323706377858,
      "grad_norm": 11302.03515625,
      "learning_rate": 0.0001866894318395871,
      "loss": 7.4035,
      "step": 142370
    },
    {
      "epoch": 17.133574007220215,
      "grad_norm": 16933.833984375,
      "learning_rate": 0.0001866875353789093,
      "loss": 7.4945,
      "step": 142380
    },
    {
      "epoch": 17.134777376654633,
      "grad_norm": 423122.34375,
      "learning_rate": 0.00018668563879277323,
      "loss": 7.34,
      "step": 142390
    },
    {
      "epoch": 17.13598074608905,
      "grad_norm": 14923.0068359375,
      "learning_rate": 0.00018668374208118165,
      "loss": 7.3102,
      "step": 142400
    },
    {
      "epoch": 17.137184115523464,
      "grad_norm": 11712.00390625,
      "learning_rate": 0.00018668184524413728,
      "loss": 7.376,
      "step": 142410
    },
    {
      "epoch": 17.13838748495788,
      "grad_norm": 62720.671875,
      "learning_rate": 0.00018667994828164286,
      "loss": 7.317,
      "step": 142420
    },
    {
      "epoch": 17.1395908543923,
      "grad_norm": 41738.1796875,
      "learning_rate": 0.00018667805119370118,
      "loss": 7.4232,
      "step": 142430
    },
    {
      "epoch": 17.140794223826713,
      "grad_norm": 38861.62890625,
      "learning_rate": 0.00018667615398031495,
      "loss": 7.3007,
      "step": 142440
    },
    {
      "epoch": 17.14199759326113,
      "grad_norm": 153266.109375,
      "learning_rate": 0.00018667425664148693,
      "loss": 7.4199,
      "step": 142450
    },
    {
      "epoch": 17.14320096269555,
      "grad_norm": 211282.78125,
      "learning_rate": 0.00018667235917721984,
      "loss": 7.2611,
      "step": 142460
    },
    {
      "epoch": 17.144404332129962,
      "grad_norm": 13568.48828125,
      "learning_rate": 0.0001866704615875165,
      "loss": 7.3037,
      "step": 142470
    },
    {
      "epoch": 17.14560770156438,
      "grad_norm": 10499.71875,
      "learning_rate": 0.00018666856387237956,
      "loss": 7.3668,
      "step": 142480
    },
    {
      "epoch": 17.146811070998798,
      "grad_norm": 11154.3515625,
      "learning_rate": 0.00018666666603181183,
      "loss": 7.3346,
      "step": 142490
    },
    {
      "epoch": 17.14801444043321,
      "grad_norm": 160310.6875,
      "learning_rate": 0.00018666476806581604,
      "loss": 7.3174,
      "step": 142500
    },
    {
      "epoch": 17.14921780986763,
      "grad_norm": 34515.3125,
      "learning_rate": 0.00018666286997439491,
      "loss": 7.4247,
      "step": 142510
    },
    {
      "epoch": 17.150421179302047,
      "grad_norm": 36640.3125,
      "learning_rate": 0.00018666097175755122,
      "loss": 7.3735,
      "step": 142520
    },
    {
      "epoch": 17.15162454873646,
      "grad_norm": 41794.875,
      "learning_rate": 0.00018665907341528773,
      "loss": 7.4383,
      "step": 142530
    },
    {
      "epoch": 17.15282791817088,
      "grad_norm": 21124.376953125,
      "learning_rate": 0.00018665717494760714,
      "loss": 7.4022,
      "step": 142540
    },
    {
      "epoch": 17.154031287605296,
      "grad_norm": 60762.08203125,
      "learning_rate": 0.00018665527635451223,
      "loss": 7.2942,
      "step": 142550
    },
    {
      "epoch": 17.15523465703971,
      "grad_norm": 42075.8984375,
      "learning_rate": 0.00018665337763600572,
      "loss": 7.3258,
      "step": 142560
    },
    {
      "epoch": 17.156438026474127,
      "grad_norm": 36340.90234375,
      "learning_rate": 0.00018665147879209042,
      "loss": 7.3362,
      "step": 142570
    },
    {
      "epoch": 17.157641395908545,
      "grad_norm": 15715.5380859375,
      "learning_rate": 0.00018664957982276898,
      "loss": 7.468,
      "step": 142580
    },
    {
      "epoch": 17.15884476534296,
      "grad_norm": 60864.4765625,
      "learning_rate": 0.00018664768072804426,
      "loss": 7.4349,
      "step": 142590
    },
    {
      "epoch": 17.160048134777377,
      "grad_norm": 104410.78125,
      "learning_rate": 0.00018664578150791894,
      "loss": 7.3642,
      "step": 142600
    },
    {
      "epoch": 17.161251504211794,
      "grad_norm": 72621.6328125,
      "learning_rate": 0.00018664388216239577,
      "loss": 7.297,
      "step": 142610
    },
    {
      "epoch": 17.162454873646208,
      "grad_norm": 35826.2421875,
      "learning_rate": 0.0001866419826914775,
      "loss": 7.3096,
      "step": 142620
    },
    {
      "epoch": 17.163658243080626,
      "grad_norm": 50612.3984375,
      "learning_rate": 0.0001866400830951669,
      "loss": 7.4833,
      "step": 142630
    },
    {
      "epoch": 17.164861612515043,
      "grad_norm": 36865.8203125,
      "learning_rate": 0.00018663818337346676,
      "loss": 7.5272,
      "step": 142640
    },
    {
      "epoch": 17.166064981949457,
      "grad_norm": 39109.46484375,
      "learning_rate": 0.0001866362835263797,
      "loss": 7.4837,
      "step": 142650
    },
    {
      "epoch": 17.167268351383875,
      "grad_norm": 137890.71875,
      "learning_rate": 0.0001866343835539086,
      "loss": 7.4907,
      "step": 142660
    },
    {
      "epoch": 17.168471720818292,
      "grad_norm": 41040.09375,
      "learning_rate": 0.00018663248345605614,
      "loss": 7.4056,
      "step": 142670
    },
    {
      "epoch": 17.169675090252706,
      "grad_norm": 16532.912109375,
      "learning_rate": 0.0001866305832328251,
      "loss": 7.5154,
      "step": 142680
    },
    {
      "epoch": 17.170878459687124,
      "grad_norm": 22060.61328125,
      "learning_rate": 0.00018662868288421818,
      "loss": 7.5773,
      "step": 142690
    },
    {
      "epoch": 17.17208182912154,
      "grad_norm": 59600.97265625,
      "learning_rate": 0.0001866267824102382,
      "loss": 7.5662,
      "step": 142700
    },
    {
      "epoch": 17.173285198555956,
      "grad_norm": 10623.1748046875,
      "learning_rate": 0.00018662488181088785,
      "loss": 7.5215,
      "step": 142710
    },
    {
      "epoch": 17.174488567990373,
      "grad_norm": 18329.470703125,
      "learning_rate": 0.00018662298108616996,
      "loss": 7.4264,
      "step": 142720
    },
    {
      "epoch": 17.17569193742479,
      "grad_norm": 40400.17578125,
      "learning_rate": 0.00018662108023608718,
      "loss": 7.3876,
      "step": 142730
    },
    {
      "epoch": 17.176895306859205,
      "grad_norm": 13407.1435546875,
      "learning_rate": 0.00018661917926064234,
      "loss": 7.3398,
      "step": 142740
    },
    {
      "epoch": 17.178098676293622,
      "grad_norm": 42986.66796875,
      "learning_rate": 0.00018661727815983816,
      "loss": 7.4152,
      "step": 142750
    },
    {
      "epoch": 17.17930204572804,
      "grad_norm": 10154.193359375,
      "learning_rate": 0.0001866153769336774,
      "loss": 7.3588,
      "step": 142760
    },
    {
      "epoch": 17.180505415162454,
      "grad_norm": 27231.228515625,
      "learning_rate": 0.0001866134755821628,
      "loss": 7.3424,
      "step": 142770
    },
    {
      "epoch": 17.18170878459687,
      "grad_norm": 70247.015625,
      "learning_rate": 0.0001866115741052971,
      "loss": 7.3623,
      "step": 142780
    },
    {
      "epoch": 17.18291215403129,
      "grad_norm": 25750.89453125,
      "learning_rate": 0.00018660967250308305,
      "loss": 7.3052,
      "step": 142790
    },
    {
      "epoch": 17.184115523465703,
      "grad_norm": 403852.15625,
      "learning_rate": 0.00018660777077552347,
      "loss": 7.2851,
      "step": 142800
    },
    {
      "epoch": 17.18531889290012,
      "grad_norm": 427347.15625,
      "learning_rate": 0.000186605868922621,
      "loss": 7.3909,
      "step": 142810
    },
    {
      "epoch": 17.186522262334538,
      "grad_norm": 11774.193359375,
      "learning_rate": 0.00018660396694437852,
      "loss": 7.3808,
      "step": 142820
    },
    {
      "epoch": 17.187725631768952,
      "grad_norm": 35684.640625,
      "learning_rate": 0.0001866020648407987,
      "loss": 7.3358,
      "step": 142830
    },
    {
      "epoch": 17.18892900120337,
      "grad_norm": 92285.578125,
      "learning_rate": 0.00018660016261188428,
      "loss": 7.3237,
      "step": 142840
    },
    {
      "epoch": 17.190132370637787,
      "grad_norm": 10759.0458984375,
      "learning_rate": 0.0001865982602576381,
      "loss": 7.3341,
      "step": 142850
    },
    {
      "epoch": 17.1913357400722,
      "grad_norm": 18348.951171875,
      "learning_rate": 0.00018659635777806282,
      "loss": 7.2642,
      "step": 142860
    },
    {
      "epoch": 17.19253910950662,
      "grad_norm": 44980.66015625,
      "learning_rate": 0.00018659445517316123,
      "loss": 7.2788,
      "step": 142870
    },
    {
      "epoch": 17.193742478941036,
      "grad_norm": 294009.71875,
      "learning_rate": 0.0001865925524429361,
      "loss": 7.2674,
      "step": 142880
    },
    {
      "epoch": 17.19494584837545,
      "grad_norm": 22624.0703125,
      "learning_rate": 0.00018659064958739013,
      "loss": 7.2889,
      "step": 142890
    },
    {
      "epoch": 17.196149217809868,
      "grad_norm": 360896.65625,
      "learning_rate": 0.00018658874660652615,
      "loss": 7.3536,
      "step": 142900
    },
    {
      "epoch": 17.197352587244286,
      "grad_norm": 190300.53125,
      "learning_rate": 0.0001865868435003469,
      "loss": 7.3024,
      "step": 142910
    },
    {
      "epoch": 17.1985559566787,
      "grad_norm": 261780.9375,
      "learning_rate": 0.00018658494026885505,
      "loss": 7.3493,
      "step": 142920
    },
    {
      "epoch": 17.199759326113117,
      "grad_norm": 65406.859375,
      "learning_rate": 0.00018658303691205342,
      "loss": 7.2996,
      "step": 142930
    },
    {
      "epoch": 17.200962695547535,
      "grad_norm": 572977.0,
      "learning_rate": 0.00018658113342994478,
      "loss": 7.411,
      "step": 142940
    },
    {
      "epoch": 17.20216606498195,
      "grad_norm": 128455.8671875,
      "learning_rate": 0.00018657922982253186,
      "loss": 7.4283,
      "step": 142950
    },
    {
      "epoch": 17.203369434416366,
      "grad_norm": 11471.7587890625,
      "learning_rate": 0.00018657732608981742,
      "loss": 7.3861,
      "step": 142960
    },
    {
      "epoch": 17.204572803850784,
      "grad_norm": 11657.0908203125,
      "learning_rate": 0.00018657542223180423,
      "loss": 7.4415,
      "step": 142970
    },
    {
      "epoch": 17.205776173285198,
      "grad_norm": 38225.0859375,
      "learning_rate": 0.00018657351824849497,
      "loss": 7.3639,
      "step": 142980
    },
    {
      "epoch": 17.206979542719615,
      "grad_norm": 15534.462890625,
      "learning_rate": 0.00018657161413989252,
      "loss": 7.2942,
      "step": 142990
    },
    {
      "epoch": 17.20818291215403,
      "grad_norm": 14562.5517578125,
      "learning_rate": 0.00018656970990599955,
      "loss": 7.3218,
      "step": 143000
    },
    {
      "epoch": 17.209386281588447,
      "grad_norm": 10495.0380859375,
      "learning_rate": 0.00018656780554681884,
      "loss": 7.3931,
      "step": 143010
    },
    {
      "epoch": 17.210589651022865,
      "grad_norm": 16439.080078125,
      "learning_rate": 0.00018656590106235315,
      "loss": 7.3797,
      "step": 143020
    },
    {
      "epoch": 17.21179302045728,
      "grad_norm": 14018.064453125,
      "learning_rate": 0.0001865639964526052,
      "loss": 7.5922,
      "step": 143030
    },
    {
      "epoch": 17.212996389891696,
      "grad_norm": 5792.22119140625,
      "learning_rate": 0.0001865620917175778,
      "loss": 7.6399,
      "step": 143040
    },
    {
      "epoch": 17.214199759326114,
      "grad_norm": 12662.169921875,
      "learning_rate": 0.0001865601868572737,
      "loss": 7.3904,
      "step": 143050
    },
    {
      "epoch": 17.215403128760528,
      "grad_norm": 6172.56201171875,
      "learning_rate": 0.00018655828187169558,
      "loss": 7.3667,
      "step": 143060
    },
    {
      "epoch": 17.216606498194945,
      "grad_norm": 7061.3076171875,
      "learning_rate": 0.00018655637676084628,
      "loss": 7.3954,
      "step": 143070
    },
    {
      "epoch": 17.217809867629363,
      "grad_norm": 32888.35546875,
      "learning_rate": 0.00018655447152472854,
      "loss": 7.3878,
      "step": 143080
    },
    {
      "epoch": 17.219013237063777,
      "grad_norm": 13680.1015625,
      "learning_rate": 0.00018655256616334511,
      "loss": 7.3361,
      "step": 143090
    },
    {
      "epoch": 17.220216606498195,
      "grad_norm": 13983.4951171875,
      "learning_rate": 0.00018655066067669874,
      "loss": 7.3062,
      "step": 143100
    },
    {
      "epoch": 17.221419975932612,
      "grad_norm": 29049.599609375,
      "learning_rate": 0.00018654875506479222,
      "loss": 7.2493,
      "step": 143110
    },
    {
      "epoch": 17.222623345367026,
      "grad_norm": 21413.96484375,
      "learning_rate": 0.00018654684932762825,
      "loss": 7.4469,
      "step": 143120
    },
    {
      "epoch": 17.223826714801444,
      "grad_norm": 28884.884765625,
      "learning_rate": 0.00018654494346520965,
      "loss": 7.4074,
      "step": 143130
    },
    {
      "epoch": 17.22503008423586,
      "grad_norm": 39515.76953125,
      "learning_rate": 0.00018654303747753912,
      "loss": 7.455,
      "step": 143140
    },
    {
      "epoch": 17.226233453670275,
      "grad_norm": 30641.70703125,
      "learning_rate": 0.0001865411313646195,
      "loss": 7.3797,
      "step": 143150
    },
    {
      "epoch": 17.227436823104693,
      "grad_norm": 23947.599609375,
      "learning_rate": 0.00018653922512645345,
      "loss": 7.4171,
      "step": 143160
    },
    {
      "epoch": 17.22864019253911,
      "grad_norm": 161175.59375,
      "learning_rate": 0.00018653731876304379,
      "loss": 7.3796,
      "step": 143170
    },
    {
      "epoch": 17.229843561973524,
      "grad_norm": 31299.63671875,
      "learning_rate": 0.00018653541227439324,
      "loss": 7.3659,
      "step": 143180
    },
    {
      "epoch": 17.231046931407942,
      "grad_norm": 8280.3173828125,
      "learning_rate": 0.0001865335056605046,
      "loss": 7.3586,
      "step": 143190
    },
    {
      "epoch": 17.23225030084236,
      "grad_norm": 14721.0703125,
      "learning_rate": 0.00018653159892138062,
      "loss": 7.2953,
      "step": 143200
    },
    {
      "epoch": 17.233453670276774,
      "grad_norm": 12474.8017578125,
      "learning_rate": 0.00018652969205702405,
      "loss": 7.357,
      "step": 143210
    },
    {
      "epoch": 17.23465703971119,
      "grad_norm": 26803.2578125,
      "learning_rate": 0.00018652778506743768,
      "loss": 7.4322,
      "step": 143220
    },
    {
      "epoch": 17.23586040914561,
      "grad_norm": 4982.1591796875,
      "learning_rate": 0.0001865258779526242,
      "loss": 7.2978,
      "step": 143230
    },
    {
      "epoch": 17.237063778580023,
      "grad_norm": 11828.568359375,
      "learning_rate": 0.0001865239707125864,
      "loss": 7.3405,
      "step": 143240
    },
    {
      "epoch": 17.23826714801444,
      "grad_norm": 3795.12255859375,
      "learning_rate": 0.00018652206334732709,
      "loss": 7.3372,
      "step": 143250
    },
    {
      "epoch": 17.239470517448858,
      "grad_norm": 6387.31787109375,
      "learning_rate": 0.000186520155856849,
      "loss": 7.2956,
      "step": 143260
    },
    {
      "epoch": 17.240673886883272,
      "grad_norm": 8069.97216796875,
      "learning_rate": 0.00018651824824115484,
      "loss": 7.414,
      "step": 143270
    },
    {
      "epoch": 17.24187725631769,
      "grad_norm": 5350.2783203125,
      "learning_rate": 0.00018651634050024745,
      "loss": 7.2934,
      "step": 143280
    },
    {
      "epoch": 17.243080625752107,
      "grad_norm": 5435.71875,
      "learning_rate": 0.00018651443263412954,
      "loss": 7.3264,
      "step": 143290
    },
    {
      "epoch": 17.24428399518652,
      "grad_norm": 10408.421875,
      "learning_rate": 0.0001865125246428039,
      "loss": 7.2753,
      "step": 143300
    },
    {
      "epoch": 17.24548736462094,
      "grad_norm": 4312.27587890625,
      "learning_rate": 0.00018651061652627326,
      "loss": 7.406,
      "step": 143310
    },
    {
      "epoch": 17.246690734055356,
      "grad_norm": 13353.0390625,
      "learning_rate": 0.0001865087082845404,
      "loss": 7.1902,
      "step": 143320
    },
    {
      "epoch": 17.24789410348977,
      "grad_norm": 84529.4609375,
      "learning_rate": 0.00018650679991760808,
      "loss": 7.2451,
      "step": 143330
    },
    {
      "epoch": 17.249097472924188,
      "grad_norm": 5085.16650390625,
      "learning_rate": 0.0001865048914254791,
      "loss": 7.396,
      "step": 143340
    },
    {
      "epoch": 17.250300842358605,
      "grad_norm": 4344.40771484375,
      "learning_rate": 0.00018650298280815616,
      "loss": 7.267,
      "step": 143350
    },
    {
      "epoch": 17.25150421179302,
      "grad_norm": 19860.373046875,
      "learning_rate": 0.00018650107406564202,
      "loss": 7.3526,
      "step": 143360
    },
    {
      "epoch": 17.252707581227437,
      "grad_norm": 3494.416015625,
      "learning_rate": 0.0001864991651979395,
      "loss": 7.3546,
      "step": 143370
    },
    {
      "epoch": 17.253910950661854,
      "grad_norm": 7269.1943359375,
      "learning_rate": 0.00018649725620505133,
      "loss": 7.388,
      "step": 143380
    },
    {
      "epoch": 17.25511432009627,
      "grad_norm": 3985.4990234375,
      "learning_rate": 0.00018649534708698026,
      "loss": 7.4202,
      "step": 143390
    },
    {
      "epoch": 17.256317689530686,
      "grad_norm": 6875.625,
      "learning_rate": 0.00018649343784372908,
      "loss": 7.3016,
      "step": 143400
    },
    {
      "epoch": 17.257521058965104,
      "grad_norm": 47567.93359375,
      "learning_rate": 0.00018649152847530052,
      "loss": 7.5067,
      "step": 143410
    },
    {
      "epoch": 17.258724428399518,
      "grad_norm": 34000.109375,
      "learning_rate": 0.00018648961898169738,
      "loss": 7.4025,
      "step": 143420
    },
    {
      "epoch": 17.259927797833935,
      "grad_norm": 9943.953125,
      "learning_rate": 0.00018648770936292242,
      "loss": 7.3339,
      "step": 143430
    },
    {
      "epoch": 17.261131167268353,
      "grad_norm": 3363.312744140625,
      "learning_rate": 0.00018648579961897839,
      "loss": 7.3725,
      "step": 143440
    },
    {
      "epoch": 17.262334536702767,
      "grad_norm": 9100.0615234375,
      "learning_rate": 0.000186483889749868,
      "loss": 7.3353,
      "step": 143450
    },
    {
      "epoch": 17.263537906137184,
      "grad_norm": 6549.49658203125,
      "learning_rate": 0.00018648197975559415,
      "loss": 7.4322,
      "step": 143460
    },
    {
      "epoch": 17.264741275571602,
      "grad_norm": 15612.9072265625,
      "learning_rate": 0.00018648006963615946,
      "loss": 7.4149,
      "step": 143470
    },
    {
      "epoch": 17.265944645006016,
      "grad_norm": 5912.58935546875,
      "learning_rate": 0.0001864781593915668,
      "loss": 7.3577,
      "step": 143480
    },
    {
      "epoch": 17.267148014440433,
      "grad_norm": 16263.78515625,
      "learning_rate": 0.00018647624902181888,
      "loss": 7.3115,
      "step": 143490
    },
    {
      "epoch": 17.26835138387485,
      "grad_norm": 8480.08203125,
      "learning_rate": 0.0001864743385269185,
      "loss": 7.3549,
      "step": 143500
    },
    {
      "epoch": 17.269554753309265,
      "grad_norm": 4502.17822265625,
      "learning_rate": 0.00018647242790686836,
      "loss": 7.3199,
      "step": 143510
    },
    {
      "epoch": 17.270758122743683,
      "grad_norm": 5547.29052734375,
      "learning_rate": 0.00018647051716167128,
      "loss": 7.2781,
      "step": 143520
    },
    {
      "epoch": 17.2719614921781,
      "grad_norm": 6042.51806640625,
      "learning_rate": 0.00018646860629133004,
      "loss": 7.3389,
      "step": 143530
    },
    {
      "epoch": 17.273164861612514,
      "grad_norm": 8290.95703125,
      "learning_rate": 0.00018646669529584733,
      "loss": 7.3993,
      "step": 143540
    },
    {
      "epoch": 17.27436823104693,
      "grad_norm": 13297.8583984375,
      "learning_rate": 0.000186464784175226,
      "loss": 7.4346,
      "step": 143550
    },
    {
      "epoch": 17.27557160048135,
      "grad_norm": 2856.671875,
      "learning_rate": 0.00018646287292946876,
      "loss": 7.3942,
      "step": 143560
    },
    {
      "epoch": 17.276774969915763,
      "grad_norm": 17264.79296875,
      "learning_rate": 0.00018646096155857842,
      "loss": 7.2047,
      "step": 143570
    },
    {
      "epoch": 17.27797833935018,
      "grad_norm": 9158.6005859375,
      "learning_rate": 0.0001864590500625577,
      "loss": 7.4385,
      "step": 143580
    },
    {
      "epoch": 17.2791817087846,
      "grad_norm": 8827.74609375,
      "learning_rate": 0.0001864571384414094,
      "loss": 7.444,
      "step": 143590
    },
    {
      "epoch": 17.280385078219012,
      "grad_norm": 8712.2607421875,
      "learning_rate": 0.00018645522669513627,
      "loss": 7.3713,
      "step": 143600
    },
    {
      "epoch": 17.28158844765343,
      "grad_norm": 29853.666015625,
      "learning_rate": 0.0001864533148237411,
      "loss": 7.3908,
      "step": 143610
    },
    {
      "epoch": 17.282791817087848,
      "grad_norm": 4064.296630859375,
      "learning_rate": 0.00018645140282722662,
      "loss": 7.5226,
      "step": 143620
    },
    {
      "epoch": 17.28399518652226,
      "grad_norm": 10876.7958984375,
      "learning_rate": 0.00018644949070559564,
      "loss": 7.3226,
      "step": 143630
    },
    {
      "epoch": 17.28519855595668,
      "grad_norm": 6016.1044921875,
      "learning_rate": 0.00018644757845885085,
      "loss": 7.2681,
      "step": 143640
    },
    {
      "epoch": 17.286401925391097,
      "grad_norm": 7666.17919921875,
      "learning_rate": 0.00018644566608699508,
      "loss": 7.3294,
      "step": 143650
    },
    {
      "epoch": 17.28760529482551,
      "grad_norm": 3795.66455078125,
      "learning_rate": 0.00018644375359003115,
      "loss": 7.3469,
      "step": 143660
    },
    {
      "epoch": 17.28880866425993,
      "grad_norm": 12368.240234375,
      "learning_rate": 0.00018644184096796172,
      "loss": 7.4265,
      "step": 143670
    },
    {
      "epoch": 17.290012033694346,
      "grad_norm": 188565.40625,
      "learning_rate": 0.0001864399282207896,
      "loss": 7.3465,
      "step": 143680
    },
    {
      "epoch": 17.29121540312876,
      "grad_norm": 6490.89013671875,
      "learning_rate": 0.0001864380153485176,
      "loss": 7.2713,
      "step": 143690
    },
    {
      "epoch": 17.292418772563177,
      "grad_norm": 9034.33984375,
      "learning_rate": 0.00018643610235114842,
      "loss": 7.2496,
      "step": 143700
    },
    {
      "epoch": 17.29362214199759,
      "grad_norm": 6057.49267578125,
      "learning_rate": 0.00018643418922868486,
      "loss": 7.3932,
      "step": 143710
    },
    {
      "epoch": 17.29482551143201,
      "grad_norm": 2169.48876953125,
      "learning_rate": 0.00018643227598112968,
      "loss": 7.4326,
      "step": 143720
    },
    {
      "epoch": 17.296028880866427,
      "grad_norm": 2845.04541015625,
      "learning_rate": 0.00018643036260848568,
      "loss": 7.3755,
      "step": 143730
    },
    {
      "epoch": 17.29723225030084,
      "grad_norm": 4065.597412109375,
      "learning_rate": 0.00018642844911075562,
      "loss": 7.3622,
      "step": 143740
    },
    {
      "epoch": 17.29843561973526,
      "grad_norm": 2476.890380859375,
      "learning_rate": 0.0001864265354879422,
      "loss": 7.3002,
      "step": 143750
    },
    {
      "epoch": 17.299638989169676,
      "grad_norm": 2537.3583984375,
      "learning_rate": 0.00018642462174004828,
      "loss": 7.4122,
      "step": 143760
    },
    {
      "epoch": 17.30084235860409,
      "grad_norm": 16140.46484375,
      "learning_rate": 0.0001864227078670766,
      "loss": 7.3327,
      "step": 143770
    },
    {
      "epoch": 17.302045728038507,
      "grad_norm": 4514.66748046875,
      "learning_rate": 0.00018642079386902994,
      "loss": 7.377,
      "step": 143780
    },
    {
      "epoch": 17.303249097472925,
      "grad_norm": 4736.56689453125,
      "learning_rate": 0.000186418879745911,
      "loss": 7.3443,
      "step": 143790
    },
    {
      "epoch": 17.30445246690734,
      "grad_norm": 1771.5638427734375,
      "learning_rate": 0.00018641696549772264,
      "loss": 7.308,
      "step": 143800
    },
    {
      "epoch": 17.305655836341757,
      "grad_norm": 2567.355712890625,
      "learning_rate": 0.0001864150511244676,
      "loss": 7.3158,
      "step": 143810
    },
    {
      "epoch": 17.306859205776174,
      "grad_norm": 3582.007080078125,
      "learning_rate": 0.00018641313662614862,
      "loss": 7.3372,
      "step": 143820
    },
    {
      "epoch": 17.308062575210588,
      "grad_norm": 2451.7158203125,
      "learning_rate": 0.00018641122200276853,
      "loss": 7.4579,
      "step": 143830
    },
    {
      "epoch": 17.309265944645006,
      "grad_norm": 3274.827392578125,
      "learning_rate": 0.00018640930725433006,
      "loss": 7.3478,
      "step": 143840
    },
    {
      "epoch": 17.310469314079423,
      "grad_norm": 1747.725830078125,
      "learning_rate": 0.00018640739238083595,
      "loss": 7.318,
      "step": 143850
    },
    {
      "epoch": 17.311672683513837,
      "grad_norm": 1676.380126953125,
      "learning_rate": 0.00018640547738228906,
      "loss": 7.3239,
      "step": 143860
    },
    {
      "epoch": 17.312876052948255,
      "grad_norm": 4575.64453125,
      "learning_rate": 0.00018640356225869208,
      "loss": 7.2099,
      "step": 143870
    },
    {
      "epoch": 17.314079422382672,
      "grad_norm": 3088.161376953125,
      "learning_rate": 0.00018640164701004783,
      "loss": 7.301,
      "step": 143880
    },
    {
      "epoch": 17.315282791817086,
      "grad_norm": 2872.845947265625,
      "learning_rate": 0.00018639973163635907,
      "loss": 7.3168,
      "step": 143890
    },
    {
      "epoch": 17.316486161251504,
      "grad_norm": 2950.113525390625,
      "learning_rate": 0.00018639781613762855,
      "loss": 7.3247,
      "step": 143900
    },
    {
      "epoch": 17.31768953068592,
      "grad_norm": 1746.21435546875,
      "learning_rate": 0.00018639590051385907,
      "loss": 7.4187,
      "step": 143910
    },
    {
      "epoch": 17.318892900120336,
      "grad_norm": 4907.98779296875,
      "learning_rate": 0.00018639398476505336,
      "loss": 7.3506,
      "step": 143920
    },
    {
      "epoch": 17.320096269554753,
      "grad_norm": 1487.94384765625,
      "learning_rate": 0.00018639206889121428,
      "loss": 7.3712,
      "step": 143930
    },
    {
      "epoch": 17.32129963898917,
      "grad_norm": 4368.30908203125,
      "learning_rate": 0.00018639015289234452,
      "loss": 7.2957,
      "step": 143940
    },
    {
      "epoch": 17.322503008423585,
      "grad_norm": 2257.970703125,
      "learning_rate": 0.00018638823676844687,
      "loss": 7.2696,
      "step": 143950
    },
    {
      "epoch": 17.323706377858002,
      "grad_norm": 3080.18603515625,
      "learning_rate": 0.00018638632051952412,
      "loss": 7.3628,
      "step": 143960
    },
    {
      "epoch": 17.32490974729242,
      "grad_norm": 1503.36767578125,
      "learning_rate": 0.00018638440414557903,
      "loss": 7.4014,
      "step": 143970
    },
    {
      "epoch": 17.326113116726834,
      "grad_norm": 1938.1185302734375,
      "learning_rate": 0.00018638248764661438,
      "loss": 7.2477,
      "step": 143980
    },
    {
      "epoch": 17.32731648616125,
      "grad_norm": 1879.9444580078125,
      "learning_rate": 0.00018638057102263295,
      "loss": 7.2318,
      "step": 143990
    },
    {
      "epoch": 17.32851985559567,
      "grad_norm": 2374.004150390625,
      "learning_rate": 0.00018637865427363753,
      "loss": 7.4004,
      "step": 144000
    },
    {
      "epoch": 17.329723225030083,
      "grad_norm": 819.7803955078125,
      "learning_rate": 0.00018637673739963084,
      "loss": 7.3483,
      "step": 144010
    },
    {
      "epoch": 17.3309265944645,
      "grad_norm": 1696.6414794921875,
      "learning_rate": 0.00018637482040061572,
      "loss": 7.2798,
      "step": 144020
    },
    {
      "epoch": 17.332129963898918,
      "grad_norm": 1228.54443359375,
      "learning_rate": 0.00018637290327659485,
      "loss": 7.2843,
      "step": 144030
    },
    {
      "epoch": 17.333333333333332,
      "grad_norm": 1032.5482177734375,
      "learning_rate": 0.0001863709860275711,
      "loss": 7.3488,
      "step": 144040
    },
    {
      "epoch": 17.33453670276775,
      "grad_norm": 1954.9874267578125,
      "learning_rate": 0.0001863690686535472,
      "loss": 7.2892,
      "step": 144050
    },
    {
      "epoch": 17.335740072202167,
      "grad_norm": 1351.595947265625,
      "learning_rate": 0.00018636715115452596,
      "loss": 7.3964,
      "step": 144060
    },
    {
      "epoch": 17.33694344163658,
      "grad_norm": 5936.81787109375,
      "learning_rate": 0.0001863652335305101,
      "loss": 7.2856,
      "step": 144070
    },
    {
      "epoch": 17.338146811071,
      "grad_norm": 2107.088623046875,
      "learning_rate": 0.00018636331578150245,
      "loss": 7.3662,
      "step": 144080
    },
    {
      "epoch": 17.339350180505416,
      "grad_norm": 644.792724609375,
      "learning_rate": 0.0001863613979075057,
      "loss": 7.4152,
      "step": 144090
    },
    {
      "epoch": 17.34055354993983,
      "grad_norm": 1492.9910888671875,
      "learning_rate": 0.00018635947990852273,
      "loss": 7.2137,
      "step": 144100
    },
    {
      "epoch": 17.341756919374248,
      "grad_norm": 769.9312744140625,
      "learning_rate": 0.00018635756178455627,
      "loss": 7.3288,
      "step": 144110
    },
    {
      "epoch": 17.342960288808666,
      "grad_norm": 1779.5491943359375,
      "learning_rate": 0.00018635564353560913,
      "loss": 7.3602,
      "step": 144120
    },
    {
      "epoch": 17.34416365824308,
      "grad_norm": 8701.2197265625,
      "learning_rate": 0.00018635372516168398,
      "loss": 7.3381,
      "step": 144130
    },
    {
      "epoch": 17.345367027677497,
      "grad_norm": 2692.812255859375,
      "learning_rate": 0.00018635180666278374,
      "loss": 7.3142,
      "step": 144140
    },
    {
      "epoch": 17.346570397111915,
      "grad_norm": 810.0578002929688,
      "learning_rate": 0.00018634988803891106,
      "loss": 7.2351,
      "step": 144150
    },
    {
      "epoch": 17.34777376654633,
      "grad_norm": 1202.1480712890625,
      "learning_rate": 0.00018634796929006882,
      "loss": 7.2064,
      "step": 144160
    },
    {
      "epoch": 17.348977135980746,
      "grad_norm": 1173.7559814453125,
      "learning_rate": 0.00018634605041625973,
      "loss": 7.224,
      "step": 144170
    },
    {
      "epoch": 17.350180505415164,
      "grad_norm": 1695.92919921875,
      "learning_rate": 0.00018634413141748656,
      "loss": 7.2897,
      "step": 144180
    },
    {
      "epoch": 17.351383874849578,
      "grad_norm": 982.7464599609375,
      "learning_rate": 0.00018634221229375214,
      "loss": 7.303,
      "step": 144190
    },
    {
      "epoch": 17.352587244283995,
      "grad_norm": 1036.66796875,
      "learning_rate": 0.00018634029304505926,
      "loss": 7.3062,
      "step": 144200
    },
    {
      "epoch": 17.353790613718413,
      "grad_norm": 973.714111328125,
      "learning_rate": 0.0001863383736714106,
      "loss": 7.4222,
      "step": 144210
    },
    {
      "epoch": 17.354993983152827,
      "grad_norm": 1261.516845703125,
      "learning_rate": 0.00018633645417280903,
      "loss": 7.383,
      "step": 144220
    },
    {
      "epoch": 17.356197352587245,
      "grad_norm": 1612.299560546875,
      "learning_rate": 0.00018633453454925726,
      "loss": 7.2298,
      "step": 144230
    },
    {
      "epoch": 17.357400722021662,
      "grad_norm": 2724.57666015625,
      "learning_rate": 0.00018633261480075812,
      "loss": 7.3926,
      "step": 144240
    },
    {
      "epoch": 17.358604091456076,
      "grad_norm": 1550.954345703125,
      "learning_rate": 0.00018633069492731438,
      "loss": 7.2332,
      "step": 144250
    },
    {
      "epoch": 17.359807460890494,
      "grad_norm": 1035.0948486328125,
      "learning_rate": 0.00018632877492892882,
      "loss": 7.265,
      "step": 144260
    },
    {
      "epoch": 17.36101083032491,
      "grad_norm": 2157.24462890625,
      "learning_rate": 0.00018632685480560418,
      "loss": 7.2548,
      "step": 144270
    },
    {
      "epoch": 17.362214199759325,
      "grad_norm": 1274.3280029296875,
      "learning_rate": 0.00018632493455734327,
      "loss": 7.3577,
      "step": 144280
    },
    {
      "epoch": 17.363417569193743,
      "grad_norm": 874.005126953125,
      "learning_rate": 0.0001863230141841489,
      "loss": 7.1592,
      "step": 144290
    },
    {
      "epoch": 17.36462093862816,
      "grad_norm": 2269.83203125,
      "learning_rate": 0.0001863210936860238,
      "loss": 7.3021,
      "step": 144300
    },
    {
      "epoch": 17.365824308062574,
      "grad_norm": 527.2459106445312,
      "learning_rate": 0.00018631917306297075,
      "loss": 7.403,
      "step": 144310
    },
    {
      "epoch": 17.367027677496992,
      "grad_norm": 9902.6416015625,
      "learning_rate": 0.00018631725231499257,
      "loss": 7.3602,
      "step": 144320
    },
    {
      "epoch": 17.36823104693141,
      "grad_norm": 1581.8699951171875,
      "learning_rate": 0.000186315331442092,
      "loss": 7.3606,
      "step": 144330
    },
    {
      "epoch": 17.369434416365824,
      "grad_norm": 542.272216796875,
      "learning_rate": 0.00018631341044427182,
      "loss": 7.3011,
      "step": 144340
    },
    {
      "epoch": 17.37063778580024,
      "grad_norm": 1121.8876953125,
      "learning_rate": 0.00018631148932153483,
      "loss": 7.3544,
      "step": 144350
    },
    {
      "epoch": 17.37184115523466,
      "grad_norm": 341.2218322753906,
      "learning_rate": 0.0001863095680738838,
      "loss": 7.3572,
      "step": 144360
    },
    {
      "epoch": 17.373044524669073,
      "grad_norm": 1032.7603759765625,
      "learning_rate": 0.00018630764670132157,
      "loss": 7.2174,
      "step": 144370
    },
    {
      "epoch": 17.37424789410349,
      "grad_norm": 552.34228515625,
      "learning_rate": 0.00018630572520385083,
      "loss": 7.3507,
      "step": 144380
    },
    {
      "epoch": 17.375451263537904,
      "grad_norm": 969.8330688476562,
      "learning_rate": 0.0001863038035814744,
      "loss": 7.3362,
      "step": 144390
    },
    {
      "epoch": 17.376654632972322,
      "grad_norm": 651.5182495117188,
      "learning_rate": 0.00018630188183419507,
      "loss": 7.2821,
      "step": 144400
    },
    {
      "epoch": 17.37785800240674,
      "grad_norm": 1053.3939208984375,
      "learning_rate": 0.00018629995996201557,
      "loss": 7.3523,
      "step": 144410
    },
    {
      "epoch": 17.379061371841154,
      "grad_norm": 3322.80517578125,
      "learning_rate": 0.00018629803796493875,
      "loss": 7.2868,
      "step": 144420
    },
    {
      "epoch": 17.38026474127557,
      "grad_norm": 924.3411865234375,
      "learning_rate": 0.00018629611584296735,
      "loss": 7.2399,
      "step": 144430
    },
    {
      "epoch": 17.38146811070999,
      "grad_norm": 1255.8531494140625,
      "learning_rate": 0.00018629419359610417,
      "loss": 7.3268,
      "step": 144440
    },
    {
      "epoch": 17.382671480144403,
      "grad_norm": 435.2577209472656,
      "learning_rate": 0.000186292271224352,
      "loss": 7.2386,
      "step": 144450
    },
    {
      "epoch": 17.38387484957882,
      "grad_norm": 446.9579162597656,
      "learning_rate": 0.00018629034872771357,
      "loss": 7.31,
      "step": 144460
    },
    {
      "epoch": 17.385078219013238,
      "grad_norm": 1047.120849609375,
      "learning_rate": 0.00018628842610619178,
      "loss": 7.2474,
      "step": 144470
    },
    {
      "epoch": 17.386281588447652,
      "grad_norm": 427.711181640625,
      "learning_rate": 0.00018628650335978926,
      "loss": 7.3688,
      "step": 144480
    },
    {
      "epoch": 17.38748495788207,
      "grad_norm": 523.4109497070312,
      "learning_rate": 0.0001862845804885089,
      "loss": 7.3032,
      "step": 144490
    },
    {
      "epoch": 17.388688327316487,
      "grad_norm": 1037.880859375,
      "learning_rate": 0.00018628265749235344,
      "loss": 7.2783,
      "step": 144500
    },
    {
      "epoch": 17.3898916967509,
      "grad_norm": 817.0455932617188,
      "learning_rate": 0.00018628073437132567,
      "loss": 7.3336,
      "step": 144510
    },
    {
      "epoch": 17.39109506618532,
      "grad_norm": 970.6292114257812,
      "learning_rate": 0.00018627881112542836,
      "loss": 7.2893,
      "step": 144520
    },
    {
      "epoch": 17.392298435619736,
      "grad_norm": 600.50537109375,
      "learning_rate": 0.00018627688775466433,
      "loss": 7.3085,
      "step": 144530
    },
    {
      "epoch": 17.39350180505415,
      "grad_norm": 567.3610229492188,
      "learning_rate": 0.00018627496425903633,
      "loss": 7.3421,
      "step": 144540
    },
    {
      "epoch": 17.394705174488568,
      "grad_norm": 446.5660095214844,
      "learning_rate": 0.00018627304063854717,
      "loss": 7.3091,
      "step": 144550
    },
    {
      "epoch": 17.395908543922985,
      "grad_norm": 757.7503051757812,
      "learning_rate": 0.0001862711168931996,
      "loss": 7.321,
      "step": 144560
    },
    {
      "epoch": 17.3971119133574,
      "grad_norm": 1131.69384765625,
      "learning_rate": 0.00018626919302299646,
      "loss": 7.2824,
      "step": 144570
    },
    {
      "epoch": 17.398315282791817,
      "grad_norm": 236.73899841308594,
      "learning_rate": 0.00018626726902794044,
      "loss": 7.3372,
      "step": 144580
    },
    {
      "epoch": 17.399518652226234,
      "grad_norm": 689.5936889648438,
      "learning_rate": 0.00018626534490803444,
      "loss": 7.3674,
      "step": 144590
    },
    {
      "epoch": 17.40072202166065,
      "grad_norm": 1243.590087890625,
      "learning_rate": 0.00018626342066328116,
      "loss": 7.2967,
      "step": 144600
    },
    {
      "epoch": 17.401925391095066,
      "grad_norm": 508.0067443847656,
      "learning_rate": 0.0001862614962936834,
      "loss": 7.4296,
      "step": 144610
    },
    {
      "epoch": 17.403128760529484,
      "grad_norm": 74773.5234375,
      "learning_rate": 0.000186259571799244,
      "loss": 7.2877,
      "step": 144620
    },
    {
      "epoch": 17.404332129963898,
      "grad_norm": 352.3360290527344,
      "learning_rate": 0.00018625764717996568,
      "loss": 7.3145,
      "step": 144630
    },
    {
      "epoch": 17.405535499398315,
      "grad_norm": 371.5287170410156,
      "learning_rate": 0.00018625572243585127,
      "loss": 7.391,
      "step": 144640
    },
    {
      "epoch": 17.406738868832733,
      "grad_norm": 424.0631408691406,
      "learning_rate": 0.0001862537975669035,
      "loss": 7.298,
      "step": 144650
    },
    {
      "epoch": 17.407942238267147,
      "grad_norm": 328.6737365722656,
      "learning_rate": 0.00018625187257312518,
      "loss": 7.3303,
      "step": 144660
    },
    {
      "epoch": 17.409145607701564,
      "grad_norm": 603.6110229492188,
      "learning_rate": 0.00018624994745451915,
      "loss": 7.3787,
      "step": 144670
    },
    {
      "epoch": 17.410348977135982,
      "grad_norm": 829.3782958984375,
      "learning_rate": 0.0001862480222110881,
      "loss": 7.3476,
      "step": 144680
    },
    {
      "epoch": 17.411552346570396,
      "grad_norm": 814.5963134765625,
      "learning_rate": 0.0001862460968428349,
      "loss": 7.3706,
      "step": 144690
    },
    {
      "epoch": 17.412755716004813,
      "grad_norm": 514.9297485351562,
      "learning_rate": 0.0001862441713497623,
      "loss": 7.3761,
      "step": 144700
    },
    {
      "epoch": 17.41395908543923,
      "grad_norm": 416.16796875,
      "learning_rate": 0.0001862422457318731,
      "loss": 7.3654,
      "step": 144710
    },
    {
      "epoch": 17.415162454873645,
      "grad_norm": 1018.8207397460938,
      "learning_rate": 0.00018624031998917007,
      "loss": 7.3595,
      "step": 144720
    },
    {
      "epoch": 17.416365824308063,
      "grad_norm": 398.9307861328125,
      "learning_rate": 0.00018623839412165597,
      "loss": 7.2842,
      "step": 144730
    },
    {
      "epoch": 17.41756919374248,
      "grad_norm": 243.2138671875,
      "learning_rate": 0.0001862364681293337,
      "loss": 7.24,
      "step": 144740
    },
    {
      "epoch": 17.418772563176894,
      "grad_norm": 424.7333679199219,
      "learning_rate": 0.0001862345420122059,
      "loss": 7.413,
      "step": 144750
    },
    {
      "epoch": 17.41997593261131,
      "grad_norm": 668.6207275390625,
      "learning_rate": 0.00018623261577027547,
      "loss": 7.2815,
      "step": 144760
    },
    {
      "epoch": 17.42117930204573,
      "grad_norm": 602.9308471679688,
      "learning_rate": 0.00018623068940354514,
      "loss": 7.3633,
      "step": 144770
    },
    {
      "epoch": 17.422382671480143,
      "grad_norm": 511.3483581542969,
      "learning_rate": 0.0001862287629120177,
      "loss": 7.268,
      "step": 144780
    },
    {
      "epoch": 17.42358604091456,
      "grad_norm": 649.0414428710938,
      "learning_rate": 0.00018622683629569594,
      "loss": 7.3628,
      "step": 144790
    },
    {
      "epoch": 17.42478941034898,
      "grad_norm": 1002.4896850585938,
      "learning_rate": 0.0001862249095545827,
      "loss": 7.3551,
      "step": 144800
    },
    {
      "epoch": 17.425992779783392,
      "grad_norm": 337.2454528808594,
      "learning_rate": 0.0001862229826886807,
      "loss": 7.4289,
      "step": 144810
    },
    {
      "epoch": 17.42719614921781,
      "grad_norm": 919.8434448242188,
      "learning_rate": 0.00018622105569799277,
      "loss": 7.3102,
      "step": 144820
    },
    {
      "epoch": 17.428399518652228,
      "grad_norm": 1118.0283203125,
      "learning_rate": 0.0001862191285825217,
      "loss": 7.2126,
      "step": 144830
    },
    {
      "epoch": 17.42960288808664,
      "grad_norm": 469.995849609375,
      "learning_rate": 0.00018621720134227022,
      "loss": 7.3796,
      "step": 144840
    },
    {
      "epoch": 17.43080625752106,
      "grad_norm": 744.5451049804688,
      "learning_rate": 0.0001862152739772412,
      "loss": 7.3315,
      "step": 144850
    },
    {
      "epoch": 17.432009626955477,
      "grad_norm": 914.6758422851562,
      "learning_rate": 0.0001862133464874374,
      "loss": 7.3838,
      "step": 144860
    },
    {
      "epoch": 17.43321299638989,
      "grad_norm": 541.7366333007812,
      "learning_rate": 0.00018621141887286156,
      "loss": 7.3604,
      "step": 144870
    },
    {
      "epoch": 17.43441636582431,
      "grad_norm": 920.416015625,
      "learning_rate": 0.0001862094911335165,
      "loss": 7.2882,
      "step": 144880
    },
    {
      "epoch": 17.435619735258726,
      "grad_norm": 725.6030883789062,
      "learning_rate": 0.0001862075632694051,
      "loss": 7.2507,
      "step": 144890
    },
    {
      "epoch": 17.43682310469314,
      "grad_norm": 808.6946411132812,
      "learning_rate": 0.00018620563528053,
      "loss": 7.3138,
      "step": 144900
    },
    {
      "epoch": 17.438026474127557,
      "grad_norm": 1240.8717041015625,
      "learning_rate": 0.0001862037071668941,
      "loss": 7.3813,
      "step": 144910
    },
    {
      "epoch": 17.439229843561975,
      "grad_norm": 2129.42333984375,
      "learning_rate": 0.00018620177892850014,
      "loss": 7.4193,
      "step": 144920
    },
    {
      "epoch": 17.44043321299639,
      "grad_norm": 1639.1021728515625,
      "learning_rate": 0.0001861998505653509,
      "loss": 7.3415,
      "step": 144930
    },
    {
      "epoch": 17.441636582430807,
      "grad_norm": 975.15283203125,
      "learning_rate": 0.00018619792207744923,
      "loss": 7.3347,
      "step": 144940
    },
    {
      "epoch": 17.442839951865224,
      "grad_norm": 417.9345703125,
      "learning_rate": 0.00018619599346479787,
      "loss": 7.3501,
      "step": 144950
    },
    {
      "epoch": 17.444043321299638,
      "grad_norm": 636.4627685546875,
      "learning_rate": 0.00018619406472739964,
      "loss": 7.2403,
      "step": 144960
    },
    {
      "epoch": 17.445246690734056,
      "grad_norm": 475.5944519042969,
      "learning_rate": 0.0001861921358652573,
      "loss": 7.3205,
      "step": 144970
    },
    {
      "epoch": 17.446450060168473,
      "grad_norm": 935.9007568359375,
      "learning_rate": 0.00018619020687837368,
      "loss": 7.2482,
      "step": 144980
    },
    {
      "epoch": 17.447653429602887,
      "grad_norm": 513.8063354492188,
      "learning_rate": 0.00018618827776675155,
      "loss": 7.3805,
      "step": 144990
    },
    {
      "epoch": 17.448856799037305,
      "grad_norm": 279.8896789550781,
      "learning_rate": 0.00018618634853039367,
      "loss": 7.366,
      "step": 145000
    },
    {
      "epoch": 17.450060168471722,
      "grad_norm": 1896.1181640625,
      "learning_rate": 0.0001861844191693029,
      "loss": 7.3362,
      "step": 145010
    },
    {
      "epoch": 17.451263537906136,
      "grad_norm": 624.3501586914062,
      "learning_rate": 0.00018618248968348198,
      "loss": 7.2464,
      "step": 145020
    },
    {
      "epoch": 17.452466907340554,
      "grad_norm": 486.2643127441406,
      "learning_rate": 0.00018618056007293373,
      "loss": 7.2967,
      "step": 145030
    },
    {
      "epoch": 17.45367027677497,
      "grad_norm": 560.977783203125,
      "learning_rate": 0.00018617863033766092,
      "loss": 7.2483,
      "step": 145040
    },
    {
      "epoch": 17.454873646209386,
      "grad_norm": 274.99261474609375,
      "learning_rate": 0.00018617670047766638,
      "loss": 7.3979,
      "step": 145050
    },
    {
      "epoch": 17.456077015643803,
      "grad_norm": 266.1700439453125,
      "learning_rate": 0.00018617477049295286,
      "loss": 7.2552,
      "step": 145060
    },
    {
      "epoch": 17.45728038507822,
      "grad_norm": 430.4427185058594,
      "learning_rate": 0.00018617284038352315,
      "loss": 7.2915,
      "step": 145070
    },
    {
      "epoch": 17.458483754512635,
      "grad_norm": 258.2168273925781,
      "learning_rate": 0.0001861709101493801,
      "loss": 7.26,
      "step": 145080
    },
    {
      "epoch": 17.459687123947052,
      "grad_norm": 1040.547119140625,
      "learning_rate": 0.00018616897979052645,
      "loss": 7.3294,
      "step": 145090
    },
    {
      "epoch": 17.460890493381466,
      "grad_norm": 893.6184692382812,
      "learning_rate": 0.00018616704930696503,
      "loss": 7.4598,
      "step": 145100
    },
    {
      "epoch": 17.462093862815884,
      "grad_norm": 657.4937744140625,
      "learning_rate": 0.00018616511869869857,
      "loss": 7.4161,
      "step": 145110
    },
    {
      "epoch": 17.4632972322503,
      "grad_norm": 1353.1165771484375,
      "learning_rate": 0.00018616318796572996,
      "loss": 7.3446,
      "step": 145120
    },
    {
      "epoch": 17.464500601684716,
      "grad_norm": 274.4250793457031,
      "learning_rate": 0.00018616125710806193,
      "loss": 7.2106,
      "step": 145130
    },
    {
      "epoch": 17.465703971119133,
      "grad_norm": 441.3121032714844,
      "learning_rate": 0.00018615932612569725,
      "loss": 7.2857,
      "step": 145140
    },
    {
      "epoch": 17.46690734055355,
      "grad_norm": 950.4795532226562,
      "learning_rate": 0.0001861573950186388,
      "loss": 7.3201,
      "step": 145150
    },
    {
      "epoch": 17.468110709987965,
      "grad_norm": 439.0350036621094,
      "learning_rate": 0.00018615546378688934,
      "loss": 7.2845,
      "step": 145160
    },
    {
      "epoch": 17.469314079422382,
      "grad_norm": 509.9036560058594,
      "learning_rate": 0.0001861535324304516,
      "loss": 7.4241,
      "step": 145170
    },
    {
      "epoch": 17.4705174488568,
      "grad_norm": 536.127685546875,
      "learning_rate": 0.00018615160094932845,
      "loss": 7.3482,
      "step": 145180
    },
    {
      "epoch": 17.471720818291214,
      "grad_norm": 871.3175048828125,
      "learning_rate": 0.0001861496693435227,
      "loss": 7.3592,
      "step": 145190
    },
    {
      "epoch": 17.47292418772563,
      "grad_norm": 510.48870849609375,
      "learning_rate": 0.0001861477376130371,
      "loss": 7.3678,
      "step": 145200
    },
    {
      "epoch": 17.47412755716005,
      "grad_norm": 561.6532592773438,
      "learning_rate": 0.0001861458057578744,
      "loss": 7.2998,
      "step": 145210
    },
    {
      "epoch": 17.475330926594463,
      "grad_norm": 1668.0245361328125,
      "learning_rate": 0.00018614387377803747,
      "loss": 7.4022,
      "step": 145220
    },
    {
      "epoch": 17.47653429602888,
      "grad_norm": 418.8941345214844,
      "learning_rate": 0.0001861419416735291,
      "loss": 7.3786,
      "step": 145230
    },
    {
      "epoch": 17.477737665463298,
      "grad_norm": 1009.04248046875,
      "learning_rate": 0.00018614000944435207,
      "loss": 7.3459,
      "step": 145240
    },
    {
      "epoch": 17.478941034897712,
      "grad_norm": 1637.6072998046875,
      "learning_rate": 0.0001861380770905092,
      "loss": 7.3276,
      "step": 145250
    },
    {
      "epoch": 17.48014440433213,
      "grad_norm": 470.1253967285156,
      "learning_rate": 0.00018613614461200325,
      "loss": 7.2656,
      "step": 145260
    },
    {
      "epoch": 17.481347773766547,
      "grad_norm": 372.2898254394531,
      "learning_rate": 0.00018613421200883702,
      "loss": 7.2537,
      "step": 145270
    },
    {
      "epoch": 17.48255114320096,
      "grad_norm": 331.7706298828125,
      "learning_rate": 0.0001861322792810133,
      "loss": 7.3372,
      "step": 145280
    },
    {
      "epoch": 17.48375451263538,
      "grad_norm": 998.35546875,
      "learning_rate": 0.00018613034642853496,
      "loss": 7.353,
      "step": 145290
    },
    {
      "epoch": 17.484957882069796,
      "grad_norm": 663.9705810546875,
      "learning_rate": 0.0001861284134514047,
      "loss": 7.3066,
      "step": 145300
    },
    {
      "epoch": 17.48616125150421,
      "grad_norm": 432.5865173339844,
      "learning_rate": 0.0001861264803496254,
      "loss": 7.3771,
      "step": 145310
    },
    {
      "epoch": 17.487364620938628,
      "grad_norm": 1278.626953125,
      "learning_rate": 0.00018612454712319977,
      "loss": 7.3352,
      "step": 145320
    },
    {
      "epoch": 17.488567990373046,
      "grad_norm": 1087.6787109375,
      "learning_rate": 0.0001861226137721307,
      "loss": 7.2944,
      "step": 145330
    },
    {
      "epoch": 17.48977135980746,
      "grad_norm": 949.75,
      "learning_rate": 0.00018612068029642094,
      "loss": 7.2802,
      "step": 145340
    },
    {
      "epoch": 17.490974729241877,
      "grad_norm": 493.3676452636719,
      "learning_rate": 0.00018611874669607328,
      "loss": 7.302,
      "step": 145350
    },
    {
      "epoch": 17.492178098676295,
      "grad_norm": 830.3001098632812,
      "learning_rate": 0.00018611681297109054,
      "loss": 7.3606,
      "step": 145360
    },
    {
      "epoch": 17.49338146811071,
      "grad_norm": 596.345458984375,
      "learning_rate": 0.0001861148791214755,
      "loss": 7.3008,
      "step": 145370
    },
    {
      "epoch": 17.494584837545126,
      "grad_norm": 275.3224792480469,
      "learning_rate": 0.00018611294514723096,
      "loss": 7.2747,
      "step": 145380
    },
    {
      "epoch": 17.495788206979544,
      "grad_norm": 686.6500854492188,
      "learning_rate": 0.00018611101104835974,
      "loss": 7.2889,
      "step": 145390
    },
    {
      "epoch": 17.496991576413958,
      "grad_norm": 705.9725341796875,
      "learning_rate": 0.00018610907682486464,
      "loss": 7.3085,
      "step": 145400
    },
    {
      "epoch": 17.498194945848375,
      "grad_norm": 543.0433349609375,
      "learning_rate": 0.00018610714247674844,
      "loss": 7.3319,
      "step": 145410
    },
    {
      "epoch": 17.499398315282793,
      "grad_norm": 2264.55517578125,
      "learning_rate": 0.0001861052080040139,
      "loss": 7.3781,
      "step": 145420
    },
    {
      "epoch": 17.500601684717207,
      "grad_norm": 712.26416015625,
      "learning_rate": 0.00018610327340666394,
      "loss": 7.3167,
      "step": 145430
    },
    {
      "epoch": 17.501805054151625,
      "grad_norm": 924.663818359375,
      "learning_rate": 0.00018610133868470126,
      "loss": 7.36,
      "step": 145440
    },
    {
      "epoch": 17.503008423586042,
      "grad_norm": 425.78948974609375,
      "learning_rate": 0.00018609940383812866,
      "loss": 7.3095,
      "step": 145450
    },
    {
      "epoch": 17.504211793020456,
      "grad_norm": 501.5254821777344,
      "learning_rate": 0.00018609746886694896,
      "loss": 7.3466,
      "step": 145460
    },
    {
      "epoch": 17.505415162454874,
      "grad_norm": 677.3394165039062,
      "learning_rate": 0.000186095533771165,
      "loss": 7.3262,
      "step": 145470
    },
    {
      "epoch": 17.50661853188929,
      "grad_norm": 571.8798828125,
      "learning_rate": 0.00018609359855077955,
      "loss": 7.2988,
      "step": 145480
    },
    {
      "epoch": 17.507821901323705,
      "grad_norm": 2618.664794921875,
      "learning_rate": 0.00018609166320579537,
      "loss": 7.3308,
      "step": 145490
    },
    {
      "epoch": 17.509025270758123,
      "grad_norm": 889.0499267578125,
      "learning_rate": 0.00018608972773621532,
      "loss": 7.3257,
      "step": 145500
    },
    {
      "epoch": 17.51022864019254,
      "grad_norm": 574.9404907226562,
      "learning_rate": 0.00018608779214204218,
      "loss": 7.3327,
      "step": 145510
    },
    {
      "epoch": 17.511432009626954,
      "grad_norm": 1075.8699951171875,
      "learning_rate": 0.00018608585642327876,
      "loss": 7.3292,
      "step": 145520
    },
    {
      "epoch": 17.512635379061372,
      "grad_norm": 484.9434509277344,
      "learning_rate": 0.00018608392057992783,
      "loss": 7.3285,
      "step": 145530
    },
    {
      "epoch": 17.51383874849579,
      "grad_norm": 1054.6104736328125,
      "learning_rate": 0.00018608198461199222,
      "loss": 7.4039,
      "step": 145540
    },
    {
      "epoch": 17.515042117930204,
      "grad_norm": 809.6917114257812,
      "learning_rate": 0.00018608004851947474,
      "loss": 7.2362,
      "step": 145550
    },
    {
      "epoch": 17.51624548736462,
      "grad_norm": 2787.838623046875,
      "learning_rate": 0.00018607811230237817,
      "loss": 7.3033,
      "step": 145560
    },
    {
      "epoch": 17.51744885679904,
      "grad_norm": 928.9354248046875,
      "learning_rate": 0.0001860761759607053,
      "loss": 7.26,
      "step": 145570
    },
    {
      "epoch": 17.518652226233453,
      "grad_norm": 2730.526611328125,
      "learning_rate": 0.00018607423949445898,
      "loss": 7.3684,
      "step": 145580
    },
    {
      "epoch": 17.51985559566787,
      "grad_norm": 927.2328491210938,
      "learning_rate": 0.00018607230290364196,
      "loss": 7.2313,
      "step": 145590
    },
    {
      "epoch": 17.521058965102288,
      "grad_norm": 1870.129150390625,
      "learning_rate": 0.00018607036618825708,
      "loss": 7.2646,
      "step": 145600
    },
    {
      "epoch": 17.522262334536702,
      "grad_norm": 768.214111328125,
      "learning_rate": 0.00018606842934830713,
      "loss": 7.3782,
      "step": 145610
    },
    {
      "epoch": 17.52346570397112,
      "grad_norm": 1572.7806396484375,
      "learning_rate": 0.00018606649238379493,
      "loss": 7.3783,
      "step": 145620
    },
    {
      "epoch": 17.524669073405537,
      "grad_norm": 774.7906494140625,
      "learning_rate": 0.0001860645552947232,
      "loss": 7.3898,
      "step": 145630
    },
    {
      "epoch": 17.52587244283995,
      "grad_norm": 1090.316162109375,
      "learning_rate": 0.0001860626180810949,
      "loss": 7.4188,
      "step": 145640
    },
    {
      "epoch": 17.52707581227437,
      "grad_norm": 552.47119140625,
      "learning_rate": 0.0001860606807429127,
      "loss": 7.3743,
      "step": 145650
    },
    {
      "epoch": 17.528279181708786,
      "grad_norm": 1903.089599609375,
      "learning_rate": 0.00018605874328017942,
      "loss": 7.304,
      "step": 145660
    },
    {
      "epoch": 17.5294825511432,
      "grad_norm": 820.9869995117188,
      "learning_rate": 0.0001860568056928979,
      "loss": 7.3379,
      "step": 145670
    },
    {
      "epoch": 17.530685920577618,
      "grad_norm": 1438.8450927734375,
      "learning_rate": 0.000186054867981071,
      "loss": 7.2836,
      "step": 145680
    },
    {
      "epoch": 17.531889290012035,
      "grad_norm": 903.9391479492188,
      "learning_rate": 0.00018605293014470141,
      "loss": 7.342,
      "step": 145690
    },
    {
      "epoch": 17.53309265944645,
      "grad_norm": 1641.6419677734375,
      "learning_rate": 0.000186050992183792,
      "loss": 7.3934,
      "step": 145700
    },
    {
      "epoch": 17.534296028880867,
      "grad_norm": 1341.2841796875,
      "learning_rate": 0.00018604905409834555,
      "loss": 7.3568,
      "step": 145710
    },
    {
      "epoch": 17.535499398315284,
      "grad_norm": 1363.6627197265625,
      "learning_rate": 0.00018604711588836485,
      "loss": 7.3389,
      "step": 145720
    },
    {
      "epoch": 17.5367027677497,
      "grad_norm": 738.7965087890625,
      "learning_rate": 0.00018604517755385275,
      "loss": 7.2598,
      "step": 145730
    },
    {
      "epoch": 17.537906137184116,
      "grad_norm": 1507.6448974609375,
      "learning_rate": 0.00018604323909481206,
      "loss": 7.3141,
      "step": 145740
    },
    {
      "epoch": 17.53910950661853,
      "grad_norm": 618.2645874023438,
      "learning_rate": 0.00018604130051124554,
      "loss": 7.4364,
      "step": 145750
    },
    {
      "epoch": 17.540312876052948,
      "grad_norm": 572.6903076171875,
      "learning_rate": 0.000186039361803156,
      "loss": 7.3705,
      "step": 145760
    },
    {
      "epoch": 17.541516245487365,
      "grad_norm": 771.5133056640625,
      "learning_rate": 0.00018603742297054624,
      "loss": 7.426,
      "step": 145770
    },
    {
      "epoch": 17.54271961492178,
      "grad_norm": 390.690673828125,
      "learning_rate": 0.00018603548401341914,
      "loss": 7.3433,
      "step": 145780
    },
    {
      "epoch": 17.543922984356197,
      "grad_norm": 443.82257080078125,
      "learning_rate": 0.00018603354493177744,
      "loss": 7.3345,
      "step": 145790
    },
    {
      "epoch": 17.545126353790614,
      "grad_norm": 869.7037353515625,
      "learning_rate": 0.00018603160572562395,
      "loss": 7.2968,
      "step": 145800
    },
    {
      "epoch": 17.54632972322503,
      "grad_norm": 541.100341796875,
      "learning_rate": 0.0001860296663949615,
      "loss": 7.2972,
      "step": 145810
    },
    {
      "epoch": 17.547533092659446,
      "grad_norm": 598.3197021484375,
      "learning_rate": 0.00018602772693979285,
      "loss": 7.3979,
      "step": 145820
    },
    {
      "epoch": 17.548736462093864,
      "grad_norm": 358.25927734375,
      "learning_rate": 0.00018602578736012085,
      "loss": 7.386,
      "step": 145830
    },
    {
      "epoch": 17.549939831528278,
      "grad_norm": 543.0917358398438,
      "learning_rate": 0.0001860238476559483,
      "loss": 7.2739,
      "step": 145840
    },
    {
      "epoch": 17.551143200962695,
      "grad_norm": 549.5787353515625,
      "learning_rate": 0.00018602190782727803,
      "loss": 7.315,
      "step": 145850
    },
    {
      "epoch": 17.552346570397113,
      "grad_norm": 262.8681640625,
      "learning_rate": 0.0001860199678741128,
      "loss": 7.2264,
      "step": 145860
    },
    {
      "epoch": 17.553549939831527,
      "grad_norm": 539.4453125,
      "learning_rate": 0.00018601802779645544,
      "loss": 7.3807,
      "step": 145870
    },
    {
      "epoch": 17.554753309265944,
      "grad_norm": 877.7745361328125,
      "learning_rate": 0.00018601608759430876,
      "loss": 7.314,
      "step": 145880
    },
    {
      "epoch": 17.555956678700362,
      "grad_norm": 343.7127990722656,
      "learning_rate": 0.00018601414726767554,
      "loss": 7.3161,
      "step": 145890
    },
    {
      "epoch": 17.557160048134776,
      "grad_norm": 285.62847900390625,
      "learning_rate": 0.0001860122068165586,
      "loss": 7.3357,
      "step": 145900
    },
    {
      "epoch": 17.558363417569193,
      "grad_norm": 420.1763916015625,
      "learning_rate": 0.00018601026624096085,
      "loss": 7.3248,
      "step": 145910
    },
    {
      "epoch": 17.55956678700361,
      "grad_norm": 544.7677001953125,
      "learning_rate": 0.00018600832554088492,
      "loss": 7.3562,
      "step": 145920
    },
    {
      "epoch": 17.560770156438025,
      "grad_norm": 451.8050842285156,
      "learning_rate": 0.00018600638471633375,
      "loss": 7.267,
      "step": 145930
    },
    {
      "epoch": 17.561973525872443,
      "grad_norm": 773.5860595703125,
      "learning_rate": 0.00018600444376731012,
      "loss": 7.4023,
      "step": 145940
    },
    {
      "epoch": 17.56317689530686,
      "grad_norm": 448.1594543457031,
      "learning_rate": 0.0001860025026938168,
      "loss": 7.3144,
      "step": 145950
    },
    {
      "epoch": 17.564380264741274,
      "grad_norm": 313.5295104980469,
      "learning_rate": 0.00018600056149585664,
      "loss": 7.3236,
      "step": 145960
    },
    {
      "epoch": 17.56558363417569,
      "grad_norm": 538.1412353515625,
      "learning_rate": 0.0001859986201734324,
      "loss": 7.3641,
      "step": 145970
    },
    {
      "epoch": 17.56678700361011,
      "grad_norm": 349.06317138671875,
      "learning_rate": 0.00018599667872654696,
      "loss": 7.3648,
      "step": 145980
    },
    {
      "epoch": 17.567990373044523,
      "grad_norm": 716.6544799804688,
      "learning_rate": 0.0001859947371552031,
      "loss": 7.356,
      "step": 145990
    },
    {
      "epoch": 17.56919374247894,
      "grad_norm": 228.37562561035156,
      "learning_rate": 0.0001859927954594036,
      "loss": 7.3501,
      "step": 146000
    },
    {
      "epoch": 17.57039711191336,
      "grad_norm": 1217.5029296875,
      "learning_rate": 0.0001859908536391513,
      "loss": 7.3097,
      "step": 146010
    },
    {
      "epoch": 17.571600481347772,
      "grad_norm": 986.5830078125,
      "learning_rate": 0.00018598891169444902,
      "loss": 7.307,
      "step": 146020
    },
    {
      "epoch": 17.57280385078219,
      "grad_norm": 187.19424438476562,
      "learning_rate": 0.0001859869696252995,
      "loss": 7.3093,
      "step": 146030
    },
    {
      "epoch": 17.574007220216608,
      "grad_norm": 1073.8994140625,
      "learning_rate": 0.00018598502743170567,
      "loss": 7.2891,
      "step": 146040
    },
    {
      "epoch": 17.57521058965102,
      "grad_norm": 403.73626708984375,
      "learning_rate": 0.00018598308511367026,
      "loss": 7.2462,
      "step": 146050
    },
    {
      "epoch": 17.57641395908544,
      "grad_norm": 4176.06298828125,
      "learning_rate": 0.00018598114267119609,
      "loss": 7.3187,
      "step": 146060
    },
    {
      "epoch": 17.577617328519857,
      "grad_norm": 377.9479064941406,
      "learning_rate": 0.00018597920010428597,
      "loss": 7.2604,
      "step": 146070
    },
    {
      "epoch": 17.57882069795427,
      "grad_norm": 1296.099365234375,
      "learning_rate": 0.00018597725741294273,
      "loss": 7.2657,
      "step": 146080
    },
    {
      "epoch": 17.58002406738869,
      "grad_norm": 1034.092529296875,
      "learning_rate": 0.00018597531459716918,
      "loss": 7.3045,
      "step": 146090
    },
    {
      "epoch": 17.581227436823106,
      "grad_norm": 636.8875122070312,
      "learning_rate": 0.0001859733716569681,
      "loss": 7.3015,
      "step": 146100
    },
    {
      "epoch": 17.58243080625752,
      "grad_norm": 305.8636474609375,
      "learning_rate": 0.00018597142859234233,
      "loss": 7.2248,
      "step": 146110
    },
    {
      "epoch": 17.583634175691937,
      "grad_norm": 666.8740234375,
      "learning_rate": 0.00018596948540329467,
      "loss": 7.3088,
      "step": 146120
    },
    {
      "epoch": 17.584837545126355,
      "grad_norm": 248.62120056152344,
      "learning_rate": 0.00018596754208982796,
      "loss": 7.2785,
      "step": 146130
    },
    {
      "epoch": 17.58604091456077,
      "grad_norm": 5054.96044921875,
      "learning_rate": 0.000185965598651945,
      "loss": 7.1947,
      "step": 146140
    },
    {
      "epoch": 17.587244283995187,
      "grad_norm": 366.7696228027344,
      "learning_rate": 0.0001859636550896486,
      "loss": 7.2768,
      "step": 146150
    },
    {
      "epoch": 17.588447653429604,
      "grad_norm": 322.9616394042969,
      "learning_rate": 0.00018596171140294155,
      "loss": 7.1709,
      "step": 146160
    },
    {
      "epoch": 17.589651022864018,
      "grad_norm": 1128.99169921875,
      "learning_rate": 0.00018595976759182666,
      "loss": 7.2795,
      "step": 146170
    },
    {
      "epoch": 17.590854392298436,
      "grad_norm": 435.0795593261719,
      "learning_rate": 0.00018595782365630678,
      "loss": 7.246,
      "step": 146180
    },
    {
      "epoch": 17.592057761732853,
      "grad_norm": 2431923.25,
      "learning_rate": 0.0001859558795963847,
      "loss": 7.2314,
      "step": 146190
    },
    {
      "epoch": 17.593261131167267,
      "grad_norm": 4149.2431640625,
      "learning_rate": 0.00018595393541206325,
      "loss": 7.3353,
      "step": 146200
    },
    {
      "epoch": 17.594464500601685,
      "grad_norm": 705.9428100585938,
      "learning_rate": 0.00018595199110334524,
      "loss": 7.2372,
      "step": 146210
    },
    {
      "epoch": 17.595667870036102,
      "grad_norm": 697.0997314453125,
      "learning_rate": 0.00018595004667023348,
      "loss": 7.4221,
      "step": 146220
    },
    {
      "epoch": 17.596871239470516,
      "grad_norm": 11594.4716796875,
      "learning_rate": 0.00018594810211273078,
      "loss": 7.2416,
      "step": 146230
    },
    {
      "epoch": 17.598074608904934,
      "grad_norm": 5237.2587890625,
      "learning_rate": 0.00018594615743083994,
      "loss": 7.2553,
      "step": 146240
    },
    {
      "epoch": 17.59927797833935,
      "grad_norm": 3798.73388671875,
      "learning_rate": 0.0001859442126245638,
      "loss": 7.1613,
      "step": 146250
    },
    {
      "epoch": 17.600481347773766,
      "grad_norm": 3670.008056640625,
      "learning_rate": 0.00018594226769390516,
      "loss": 7.3628,
      "step": 146260
    },
    {
      "epoch": 17.601684717208183,
      "grad_norm": 537.2471313476562,
      "learning_rate": 0.00018594032263886685,
      "loss": 7.3301,
      "step": 146270
    },
    {
      "epoch": 17.6028880866426,
      "grad_norm": 23133.5234375,
      "learning_rate": 0.00018593837745945168,
      "loss": 7.2543,
      "step": 146280
    },
    {
      "epoch": 17.604091456077015,
      "grad_norm": 907.0701904296875,
      "learning_rate": 0.00018593643215566243,
      "loss": 7.3081,
      "step": 146290
    },
    {
      "epoch": 17.605294825511432,
      "grad_norm": 13418.8427734375,
      "learning_rate": 0.00018593448672750198,
      "loss": 7.2941,
      "step": 146300
    },
    {
      "epoch": 17.60649819494585,
      "grad_norm": 16172.681640625,
      "learning_rate": 0.0001859325411749731,
      "loss": 7.2694,
      "step": 146310
    },
    {
      "epoch": 17.607701564380264,
      "grad_norm": 874.1754760742188,
      "learning_rate": 0.0001859305954980786,
      "loss": 7.3033,
      "step": 146320
    },
    {
      "epoch": 17.60890493381468,
      "grad_norm": 3372.183349609375,
      "learning_rate": 0.00018592864969682133,
      "loss": 7.2456,
      "step": 146330
    },
    {
      "epoch": 17.6101083032491,
      "grad_norm": 3202.759521484375,
      "learning_rate": 0.0001859267037712041,
      "loss": 7.2307,
      "step": 146340
    },
    {
      "epoch": 17.611311672683513,
      "grad_norm": 1959.9033203125,
      "learning_rate": 0.0001859247577212297,
      "loss": 7.3286,
      "step": 146350
    },
    {
      "epoch": 17.61251504211793,
      "grad_norm": 6251.66015625,
      "learning_rate": 0.00018592281154690096,
      "loss": 7.2637,
      "step": 146360
    },
    {
      "epoch": 17.613718411552348,
      "grad_norm": 2906.01513671875,
      "learning_rate": 0.0001859208652482207,
      "loss": 7.2898,
      "step": 146370
    },
    {
      "epoch": 17.614921780986762,
      "grad_norm": 6713.6220703125,
      "learning_rate": 0.00018591891882519174,
      "loss": 7.2699,
      "step": 146380
    },
    {
      "epoch": 17.61612515042118,
      "grad_norm": 2991.421142578125,
      "learning_rate": 0.0001859169722778169,
      "loss": 7.2957,
      "step": 146390
    },
    {
      "epoch": 17.617328519855597,
      "grad_norm": 13018.111328125,
      "learning_rate": 0.00018591502560609896,
      "loss": 7.3497,
      "step": 146400
    },
    {
      "epoch": 17.61853188929001,
      "grad_norm": 184216.515625,
      "learning_rate": 0.00018591307881004078,
      "loss": 7.4097,
      "step": 146410
    },
    {
      "epoch": 17.61973525872443,
      "grad_norm": 265990.875,
      "learning_rate": 0.00018591113188964518,
      "loss": 7.2602,
      "step": 146420
    },
    {
      "epoch": 17.620938628158846,
      "grad_norm": 110093.6328125,
      "learning_rate": 0.00018590918484491495,
      "loss": 7.3023,
      "step": 146430
    },
    {
      "epoch": 17.62214199759326,
      "grad_norm": 35025.1484375,
      "learning_rate": 0.00018590723767585292,
      "loss": 7.336,
      "step": 146440
    },
    {
      "epoch": 17.623345367027678,
      "grad_norm": 3899.88623046875,
      "learning_rate": 0.00018590529038246188,
      "loss": 7.3167,
      "step": 146450
    },
    {
      "epoch": 17.624548736462096,
      "grad_norm": 10622.6640625,
      "learning_rate": 0.00018590334296474472,
      "loss": 7.3151,
      "step": 146460
    },
    {
      "epoch": 17.62575210589651,
      "grad_norm": 16361.875,
      "learning_rate": 0.0001859013954227042,
      "loss": 7.345,
      "step": 146470
    },
    {
      "epoch": 17.626955475330927,
      "grad_norm": 7798.11865234375,
      "learning_rate": 0.00018589944775634313,
      "loss": 7.3355,
      "step": 146480
    },
    {
      "epoch": 17.628158844765345,
      "grad_norm": 11577.7197265625,
      "learning_rate": 0.00018589749996566438,
      "loss": 7.2456,
      "step": 146490
    },
    {
      "epoch": 17.62936221419976,
      "grad_norm": 4457.5791015625,
      "learning_rate": 0.00018589555205067072,
      "loss": 7.3605,
      "step": 146500
    },
    {
      "epoch": 17.630565583634176,
      "grad_norm": 879.6420288085938,
      "learning_rate": 0.000185893604011365,
      "loss": 7.2893,
      "step": 146510
    },
    {
      "epoch": 17.63176895306859,
      "grad_norm": 4148.8310546875,
      "learning_rate": 0.00018589165584775,
      "loss": 7.2695,
      "step": 146520
    },
    {
      "epoch": 17.632972322503008,
      "grad_norm": 3638.1328125,
      "learning_rate": 0.0001858897075598286,
      "loss": 7.3228,
      "step": 146530
    },
    {
      "epoch": 17.634175691937426,
      "grad_norm": 48776.03125,
      "learning_rate": 0.0001858877591476036,
      "loss": 7.3248,
      "step": 146540
    },
    {
      "epoch": 17.63537906137184,
      "grad_norm": 19751.966796875,
      "learning_rate": 0.00018588581061107777,
      "loss": 7.2489,
      "step": 146550
    },
    {
      "epoch": 17.636582430806257,
      "grad_norm": 19402.21484375,
      "learning_rate": 0.000185883861950254,
      "loss": 7.2607,
      "step": 146560
    },
    {
      "epoch": 17.637785800240675,
      "grad_norm": 1876.0487060546875,
      "learning_rate": 0.00018588191316513508,
      "loss": 7.2645,
      "step": 146570
    },
    {
      "epoch": 17.63898916967509,
      "grad_norm": 2103.383544921875,
      "learning_rate": 0.0001858799642557238,
      "loss": 7.2497,
      "step": 146580
    },
    {
      "epoch": 17.640192539109506,
      "grad_norm": 1838.4342041015625,
      "learning_rate": 0.00018587801522202302,
      "loss": 7.3714,
      "step": 146590
    },
    {
      "epoch": 17.641395908543924,
      "grad_norm": 969.3784790039062,
      "learning_rate": 0.00018587606606403556,
      "loss": 7.3377,
      "step": 146600
    },
    {
      "epoch": 17.642599277978338,
      "grad_norm": 2345024.0,
      "learning_rate": 0.0001858741167817642,
      "loss": 7.2874,
      "step": 146610
    },
    {
      "epoch": 17.643802647412755,
      "grad_norm": 1383.1986083984375,
      "learning_rate": 0.00018587216737521182,
      "loss": 7.2152,
      "step": 146620
    },
    {
      "epoch": 17.645006016847173,
      "grad_norm": 104735.765625,
      "learning_rate": 0.00018587021784438119,
      "loss": 7.294,
      "step": 146630
    },
    {
      "epoch": 17.646209386281587,
      "grad_norm": 8677.5244140625,
      "learning_rate": 0.00018586826818927518,
      "loss": 7.3194,
      "step": 146640
    },
    {
      "epoch": 17.647412755716005,
      "grad_norm": 2759.0703125,
      "learning_rate": 0.0001858663184098966,
      "loss": 7.2547,
      "step": 146650
    },
    {
      "epoch": 17.648616125150422,
      "grad_norm": 6347.619140625,
      "learning_rate": 0.00018586436850624822,
      "loss": 7.2436,
      "step": 146660
    },
    {
      "epoch": 17.649819494584836,
      "grad_norm": 958.448974609375,
      "learning_rate": 0.0001858624184783329,
      "loss": 7.2641,
      "step": 146670
    },
    {
      "epoch": 17.651022864019254,
      "grad_norm": 4275.88330078125,
      "learning_rate": 0.00018586046832615348,
      "loss": 7.3091,
      "step": 146680
    },
    {
      "epoch": 17.65222623345367,
      "grad_norm": 76840.453125,
      "learning_rate": 0.00018585851804971278,
      "loss": 7.3178,
      "step": 146690
    },
    {
      "epoch": 17.653429602888085,
      "grad_norm": 19024.9375,
      "learning_rate": 0.00018585656764901355,
      "loss": 7.3103,
      "step": 146700
    },
    {
      "epoch": 17.654632972322503,
      "grad_norm": 2096.9912109375,
      "learning_rate": 0.00018585461712405872,
      "loss": 7.2463,
      "step": 146710
    },
    {
      "epoch": 17.65583634175692,
      "grad_norm": 2235.46044921875,
      "learning_rate": 0.00018585266647485105,
      "loss": 7.3411,
      "step": 146720
    },
    {
      "epoch": 17.657039711191334,
      "grad_norm": 4100.783203125,
      "learning_rate": 0.00018585071570139335,
      "loss": 7.3523,
      "step": 146730
    },
    {
      "epoch": 17.658243080625752,
      "grad_norm": 570.4086303710938,
      "learning_rate": 0.0001858487648036885,
      "loss": 7.2843,
      "step": 146740
    },
    {
      "epoch": 17.65944645006017,
      "grad_norm": 755.4594116210938,
      "learning_rate": 0.0001858468137817393,
      "loss": 7.2183,
      "step": 146750
    },
    {
      "epoch": 17.660649819494584,
      "grad_norm": 2245.36328125,
      "learning_rate": 0.00018584486263554853,
      "loss": 7.3899,
      "step": 146760
    },
    {
      "epoch": 17.661853188929,
      "grad_norm": 46659.6796875,
      "learning_rate": 0.00018584291136511903,
      "loss": 7.3418,
      "step": 146770
    },
    {
      "epoch": 17.66305655836342,
      "grad_norm": 1615.165283203125,
      "learning_rate": 0.0001858409599704537,
      "loss": 7.143,
      "step": 146780
    },
    {
      "epoch": 17.664259927797833,
      "grad_norm": 1254.905517578125,
      "learning_rate": 0.00018583900845155527,
      "loss": 7.2075,
      "step": 146790
    },
    {
      "epoch": 17.66546329723225,
      "grad_norm": 849.8490600585938,
      "learning_rate": 0.00018583705680842665,
      "loss": 7.3228,
      "step": 146800
    },
    {
      "epoch": 17.666666666666668,
      "grad_norm": 367898.90625,
      "learning_rate": 0.00018583510504107057,
      "loss": 7.2608,
      "step": 146810
    },
    {
      "epoch": 17.667870036101082,
      "grad_norm": 1528.6759033203125,
      "learning_rate": 0.00018583315314948994,
      "loss": 7.3309,
      "step": 146820
    },
    {
      "epoch": 17.6690734055355,
      "grad_norm": 645.6251831054688,
      "learning_rate": 0.0001858312011336875,
      "loss": 7.2909,
      "step": 146830
    },
    {
      "epoch": 17.670276774969917,
      "grad_norm": 1087.1163330078125,
      "learning_rate": 0.00018582924899366617,
      "loss": 7.2677,
      "step": 146840
    },
    {
      "epoch": 17.67148014440433,
      "grad_norm": 606.63818359375,
      "learning_rate": 0.0001858272967294287,
      "loss": 7.3886,
      "step": 146850
    },
    {
      "epoch": 17.67268351383875,
      "grad_norm": 806.1345825195312,
      "learning_rate": 0.0001858253443409779,
      "loss": 7.2328,
      "step": 146860
    },
    {
      "epoch": 17.673886883273166,
      "grad_norm": 312233.5625,
      "learning_rate": 0.0001858233918283167,
      "loss": 7.2273,
      "step": 146870
    },
    {
      "epoch": 17.67509025270758,
      "grad_norm": 2824082.75,
      "learning_rate": 0.00018582143919144783,
      "loss": 7.3743,
      "step": 146880
    },
    {
      "epoch": 17.676293622141998,
      "grad_norm": 92454.453125,
      "learning_rate": 0.00018581948643037416,
      "loss": 7.3626,
      "step": 146890
    },
    {
      "epoch": 17.677496991576415,
      "grad_norm": 1404.9002685546875,
      "learning_rate": 0.0001858175335450985,
      "loss": 7.2826,
      "step": 146900
    },
    {
      "epoch": 17.67870036101083,
      "grad_norm": 2343.916259765625,
      "learning_rate": 0.00018581558053562368,
      "loss": 7.2826,
      "step": 146910
    },
    {
      "epoch": 17.679903730445247,
      "grad_norm": 828.0574340820312,
      "learning_rate": 0.00018581362740195253,
      "loss": 7.2656,
      "step": 146920
    },
    {
      "epoch": 17.681107099879664,
      "grad_norm": 1012.4451293945312,
      "learning_rate": 0.00018581167414408788,
      "loss": 7.3018,
      "step": 146930
    },
    {
      "epoch": 17.68231046931408,
      "grad_norm": 3416.289306640625,
      "learning_rate": 0.00018580972076203257,
      "loss": 7.3325,
      "step": 146940
    },
    {
      "epoch": 17.683513838748496,
      "grad_norm": 762.9553833007812,
      "learning_rate": 0.00018580776725578938,
      "loss": 7.2268,
      "step": 146950
    },
    {
      "epoch": 17.684717208182914,
      "grad_norm": 700855.4375,
      "learning_rate": 0.0001858058136253612,
      "loss": 7.1951,
      "step": 146960
    },
    {
      "epoch": 17.685920577617328,
      "grad_norm": 890.2206420898438,
      "learning_rate": 0.00018580385987075078,
      "loss": 7.3135,
      "step": 146970
    },
    {
      "epoch": 17.687123947051745,
      "grad_norm": 894.7973022460938,
      "learning_rate": 0.00018580190599196098,
      "loss": 7.306,
      "step": 146980
    },
    {
      "epoch": 17.688327316486163,
      "grad_norm": 2258.790283203125,
      "learning_rate": 0.00018579995198899466,
      "loss": 7.2905,
      "step": 146990
    },
    {
      "epoch": 17.689530685920577,
      "grad_norm": 2671.80126953125,
      "learning_rate": 0.00018579799786185467,
      "loss": 7.211,
      "step": 147000
    },
    {
      "epoch": 17.690734055354994,
      "grad_norm": 574.3573608398438,
      "learning_rate": 0.00018579604361054376,
      "loss": 7.2214,
      "step": 147010
    },
    {
      "epoch": 17.691937424789412,
      "grad_norm": 2397.772705078125,
      "learning_rate": 0.0001857940892350648,
      "loss": 7.2681,
      "step": 147020
    },
    {
      "epoch": 17.693140794223826,
      "grad_norm": 44892.40234375,
      "learning_rate": 0.00018579213473542057,
      "loss": 7.216,
      "step": 147030
    },
    {
      "epoch": 17.694344163658243,
      "grad_norm": 3331.72216796875,
      "learning_rate": 0.00018579018011161399,
      "loss": 7.3013,
      "step": 147040
    },
    {
      "epoch": 17.69554753309266,
      "grad_norm": 6074.6884765625,
      "learning_rate": 0.0001857882253636478,
      "loss": 7.3451,
      "step": 147050
    },
    {
      "epoch": 17.696750902527075,
      "grad_norm": 404.1683349609375,
      "learning_rate": 0.0001857862704915249,
      "loss": 7.3689,
      "step": 147060
    },
    {
      "epoch": 17.697954271961493,
      "grad_norm": 705.8678588867188,
      "learning_rate": 0.00018578431549524807,
      "loss": 7.282,
      "step": 147070
    },
    {
      "epoch": 17.69915764139591,
      "grad_norm": 1980.671142578125,
      "learning_rate": 0.00018578236037482016,
      "loss": 7.3181,
      "step": 147080
    },
    {
      "epoch": 17.700361010830324,
      "grad_norm": 6872.78466796875,
      "learning_rate": 0.00018578040513024394,
      "loss": 7.2432,
      "step": 147090
    },
    {
      "epoch": 17.70156438026474,
      "grad_norm": 927.7977294921875,
      "learning_rate": 0.0001857784497615224,
      "loss": 7.2748,
      "step": 147100
    },
    {
      "epoch": 17.70276774969916,
      "grad_norm": 654.0418090820312,
      "learning_rate": 0.00018577649426865817,
      "loss": 7.1842,
      "step": 147110
    },
    {
      "epoch": 17.703971119133573,
      "grad_norm": 718.8326416015625,
      "learning_rate": 0.00018577453865165425,
      "loss": 7.3581,
      "step": 147120
    },
    {
      "epoch": 17.70517448856799,
      "grad_norm": 2901.184326171875,
      "learning_rate": 0.00018577258291051332,
      "loss": 7.2567,
      "step": 147130
    },
    {
      "epoch": 17.706377858002405,
      "grad_norm": 4967.12158203125,
      "learning_rate": 0.00018577062704523836,
      "loss": 7.27,
      "step": 147140
    },
    {
      "epoch": 17.707581227436823,
      "grad_norm": 547.86865234375,
      "learning_rate": 0.00018576867105583208,
      "loss": 7.1951,
      "step": 147150
    },
    {
      "epoch": 17.70878459687124,
      "grad_norm": 3806.975830078125,
      "learning_rate": 0.0001857667149422974,
      "loss": 7.2073,
      "step": 147160
    },
    {
      "epoch": 17.709987966305654,
      "grad_norm": 254.1750030517578,
      "learning_rate": 0.00018576475870463709,
      "loss": 7.1362,
      "step": 147170
    },
    {
      "epoch": 17.71119133574007,
      "grad_norm": 352.9841613769531,
      "learning_rate": 0.00018576280234285398,
      "loss": 7.3163,
      "step": 147180
    },
    {
      "epoch": 17.71239470517449,
      "grad_norm": 1729.2803955078125,
      "learning_rate": 0.00018576084585695093,
      "loss": 7.2933,
      "step": 147190
    },
    {
      "epoch": 17.713598074608903,
      "grad_norm": 986.765625,
      "learning_rate": 0.00018575888924693076,
      "loss": 7.2552,
      "step": 147200
    },
    {
      "epoch": 17.71480144404332,
      "grad_norm": 13549.70703125,
      "learning_rate": 0.0001857569325127963,
      "loss": 7.2425,
      "step": 147210
    },
    {
      "epoch": 17.71600481347774,
      "grad_norm": 488.2624816894531,
      "learning_rate": 0.0001857549756545504,
      "loss": 7.2478,
      "step": 147220
    },
    {
      "epoch": 17.717208182912152,
      "grad_norm": 1665.0059814453125,
      "learning_rate": 0.00018575301867219588,
      "loss": 7.2689,
      "step": 147230
    },
    {
      "epoch": 17.71841155234657,
      "grad_norm": 35308.27734375,
      "learning_rate": 0.00018575106156573556,
      "loss": 7.3196,
      "step": 147240
    },
    {
      "epoch": 17.719614921780988,
      "grad_norm": 230.5156707763672,
      "learning_rate": 0.00018574910433517228,
      "loss": 7.2919,
      "step": 147250
    },
    {
      "epoch": 17.7208182912154,
      "grad_norm": 156.07965087890625,
      "learning_rate": 0.00018574714698050887,
      "loss": 7.4263,
      "step": 147260
    },
    {
      "epoch": 17.72202166064982,
      "grad_norm": 11104.1474609375,
      "learning_rate": 0.0001857451895017482,
      "loss": 7.294,
      "step": 147270
    },
    {
      "epoch": 17.723225030084237,
      "grad_norm": 493.3023986816406,
      "learning_rate": 0.00018574323189889303,
      "loss": 7.2137,
      "step": 147280
    },
    {
      "epoch": 17.72442839951865,
      "grad_norm": 722.85986328125,
      "learning_rate": 0.00018574127417194628,
      "loss": 7.2701,
      "step": 147290
    },
    {
      "epoch": 17.72563176895307,
      "grad_norm": 236.43310546875,
      "learning_rate": 0.0001857393163209107,
      "loss": 7.2207,
      "step": 147300
    },
    {
      "epoch": 17.726835138387486,
      "grad_norm": 3539.54931640625,
      "learning_rate": 0.00018573735834578919,
      "loss": 7.2915,
      "step": 147310
    },
    {
      "epoch": 17.7280385078219,
      "grad_norm": 3622.7412109375,
      "learning_rate": 0.00018573540024658454,
      "loss": 7.2717,
      "step": 147320
    },
    {
      "epoch": 17.729241877256317,
      "grad_norm": 3338.32080078125,
      "learning_rate": 0.00018573344202329957,
      "loss": 7.2582,
      "step": 147330
    },
    {
      "epoch": 17.730445246690735,
      "grad_norm": 68588.2421875,
      "learning_rate": 0.00018573148367593718,
      "loss": 7.2161,
      "step": 147340
    },
    {
      "epoch": 17.73164861612515,
      "grad_norm": 18673.337890625,
      "learning_rate": 0.00018572952520450015,
      "loss": 7.2857,
      "step": 147350
    },
    {
      "epoch": 17.732851985559567,
      "grad_norm": 18418.140625,
      "learning_rate": 0.00018572756660899137,
      "loss": 7.2765,
      "step": 147360
    },
    {
      "epoch": 17.734055354993984,
      "grad_norm": 47369.05859375,
      "learning_rate": 0.00018572560788941359,
      "loss": 7.2487,
      "step": 147370
    },
    {
      "epoch": 17.735258724428398,
      "grad_norm": 7611.26123046875,
      "learning_rate": 0.00018572364904576965,
      "loss": 7.2968,
      "step": 147380
    },
    {
      "epoch": 17.736462093862816,
      "grad_norm": 544.062255859375,
      "learning_rate": 0.0001857216900780625,
      "loss": 7.3254,
      "step": 147390
    },
    {
      "epoch": 17.737665463297233,
      "grad_norm": 365.38739013671875,
      "learning_rate": 0.0001857197309862949,
      "loss": 7.1947,
      "step": 147400
    },
    {
      "epoch": 17.738868832731647,
      "grad_norm": 1484.4580078125,
      "learning_rate": 0.00018571777177046965,
      "loss": 7.2881,
      "step": 147410
    },
    {
      "epoch": 17.740072202166065,
      "grad_norm": 45483.5859375,
      "learning_rate": 0.00018571581243058963,
      "loss": 7.423,
      "step": 147420
    },
    {
      "epoch": 17.741275571600482,
      "grad_norm": 836.2890014648438,
      "learning_rate": 0.00018571385296665768,
      "loss": 7.2216,
      "step": 147430
    },
    {
      "epoch": 17.742478941034896,
      "grad_norm": 1123.537841796875,
      "learning_rate": 0.00018571189337867662,
      "loss": 7.2905,
      "step": 147440
    },
    {
      "epoch": 17.743682310469314,
      "grad_norm": 200.18394470214844,
      "learning_rate": 0.00018570993366664928,
      "loss": 7.2276,
      "step": 147450
    },
    {
      "epoch": 17.74488567990373,
      "grad_norm": 69.5604019165039,
      "learning_rate": 0.0001857079738305785,
      "loss": 7.551,
      "step": 147460
    },
    {
      "epoch": 17.746089049338146,
      "grad_norm": 1939.722900390625,
      "learning_rate": 0.0001857060138704671,
      "loss": 7.4713,
      "step": 147470
    },
    {
      "epoch": 17.747292418772563,
      "grad_norm": 462.474853515625,
      "learning_rate": 0.00018570405378631798,
      "loss": 7.4653,
      "step": 147480
    },
    {
      "epoch": 17.74849578820698,
      "grad_norm": 766.5291748046875,
      "learning_rate": 0.00018570209357813392,
      "loss": 7.3912,
      "step": 147490
    },
    {
      "epoch": 17.749699157641395,
      "grad_norm": 414.2393798828125,
      "learning_rate": 0.00018570013324591779,
      "loss": 7.2489,
      "step": 147500
    },
    {
      "epoch": 17.750902527075812,
      "grad_norm": 356.94097900390625,
      "learning_rate": 0.0001856981727896724,
      "loss": 7.3119,
      "step": 147510
    },
    {
      "epoch": 17.75210589651023,
      "grad_norm": 314.382080078125,
      "learning_rate": 0.00018569621220940058,
      "loss": 7.2416,
      "step": 147520
    },
    {
      "epoch": 17.753309265944644,
      "grad_norm": 644.947021484375,
      "learning_rate": 0.00018569425150510518,
      "loss": 7.3448,
      "step": 147530
    },
    {
      "epoch": 17.75451263537906,
      "grad_norm": 244.98236083984375,
      "learning_rate": 0.00018569229067678906,
      "loss": 7.3027,
      "step": 147540
    },
    {
      "epoch": 17.75571600481348,
      "grad_norm": 2374.938720703125,
      "learning_rate": 0.00018569032972445503,
      "loss": 7.2604,
      "step": 147550
    },
    {
      "epoch": 17.756919374247893,
      "grad_norm": 39953.62890625,
      "learning_rate": 0.00018568836864810595,
      "loss": 7.2572,
      "step": 147560
    },
    {
      "epoch": 17.75812274368231,
      "grad_norm": 963.0919799804688,
      "learning_rate": 0.0001856864074477446,
      "loss": 7.3877,
      "step": 147570
    },
    {
      "epoch": 17.759326113116728,
      "grad_norm": 1925.2265625,
      "learning_rate": 0.0001856844461233739,
      "loss": 7.1947,
      "step": 147580
    },
    {
      "epoch": 17.760529482551142,
      "grad_norm": 3750.310546875,
      "learning_rate": 0.00018568248467499667,
      "loss": 7.2778,
      "step": 147590
    },
    {
      "epoch": 17.76173285198556,
      "grad_norm": 1256.3724365234375,
      "learning_rate": 0.0001856805231026157,
      "loss": 7.2844,
      "step": 147600
    },
    {
      "epoch": 17.762936221419977,
      "grad_norm": 2453.166015625,
      "learning_rate": 0.00018567856140623386,
      "loss": 7.3343,
      "step": 147610
    },
    {
      "epoch": 17.76413959085439,
      "grad_norm": 961.9985961914062,
      "learning_rate": 0.000185676599585854,
      "loss": 7.1797,
      "step": 147620
    },
    {
      "epoch": 17.76534296028881,
      "grad_norm": 810.9354858398438,
      "learning_rate": 0.00018567463764147894,
      "loss": 7.271,
      "step": 147630
    },
    {
      "epoch": 17.766546329723226,
      "grad_norm": 3024.932861328125,
      "learning_rate": 0.0001856726755731115,
      "loss": 7.2779,
      "step": 147640
    },
    {
      "epoch": 17.76774969915764,
      "grad_norm": 2884.401611328125,
      "learning_rate": 0.0001856707133807546,
      "loss": 7.2561,
      "step": 147650
    },
    {
      "epoch": 17.768953068592058,
      "grad_norm": 213.5396270751953,
      "learning_rate": 0.000185668751064411,
      "loss": 7.2765,
      "step": 147660
    },
    {
      "epoch": 17.770156438026476,
      "grad_norm": 26260.826171875,
      "learning_rate": 0.00018566678862408357,
      "loss": 7.3276,
      "step": 147670
    },
    {
      "epoch": 17.77135980746089,
      "grad_norm": 800.679931640625,
      "learning_rate": 0.00018566482605977512,
      "loss": 7.1884,
      "step": 147680
    },
    {
      "epoch": 17.772563176895307,
      "grad_norm": 784.98583984375,
      "learning_rate": 0.00018566286337148857,
      "loss": 7.2952,
      "step": 147690
    },
    {
      "epoch": 17.773766546329725,
      "grad_norm": 631.4320068359375,
      "learning_rate": 0.00018566090055922667,
      "loss": 7.2861,
      "step": 147700
    },
    {
      "epoch": 17.77496991576414,
      "grad_norm": 4246.47021484375,
      "learning_rate": 0.0001856589376229923,
      "loss": 7.2888,
      "step": 147710
    },
    {
      "epoch": 17.776173285198556,
      "grad_norm": 1353.4666748046875,
      "learning_rate": 0.0001856569745627883,
      "loss": 7.3849,
      "step": 147720
    },
    {
      "epoch": 17.777376654632974,
      "grad_norm": 599.0125732421875,
      "learning_rate": 0.00018565501137861751,
      "loss": 7.2951,
      "step": 147730
    },
    {
      "epoch": 17.778580024067388,
      "grad_norm": 856.3153076171875,
      "learning_rate": 0.0001856530480704828,
      "loss": 7.4197,
      "step": 147740
    },
    {
      "epoch": 17.779783393501805,
      "grad_norm": 337.82281494140625,
      "learning_rate": 0.00018565108463838693,
      "loss": 7.3237,
      "step": 147750
    },
    {
      "epoch": 17.780986762936223,
      "grad_norm": 1585.96533203125,
      "learning_rate": 0.00018564912108233282,
      "loss": 7.3863,
      "step": 147760
    },
    {
      "epoch": 17.782190132370637,
      "grad_norm": 6684.890625,
      "learning_rate": 0.0001856471574023233,
      "loss": 7.3317,
      "step": 147770
    },
    {
      "epoch": 17.783393501805055,
      "grad_norm": 795.8919067382812,
      "learning_rate": 0.00018564519359836118,
      "loss": 7.2081,
      "step": 147780
    },
    {
      "epoch": 17.784596871239472,
      "grad_norm": 411.29083251953125,
      "learning_rate": 0.00018564322967044934,
      "loss": 7.2962,
      "step": 147790
    },
    {
      "epoch": 17.785800240673886,
      "grad_norm": 1413.302978515625,
      "learning_rate": 0.00018564126561859058,
      "loss": 7.3072,
      "step": 147800
    },
    {
      "epoch": 17.787003610108304,
      "grad_norm": 2164.65185546875,
      "learning_rate": 0.00018563930144278778,
      "loss": 7.3282,
      "step": 147810
    },
    {
      "epoch": 17.78820697954272,
      "grad_norm": 408.6491394042969,
      "learning_rate": 0.00018563733714304376,
      "loss": 7.3427,
      "step": 147820
    },
    {
      "epoch": 17.789410348977135,
      "grad_norm": 632.2469482421875,
      "learning_rate": 0.00018563537271936136,
      "loss": 7.2408,
      "step": 147830
    },
    {
      "epoch": 17.790613718411553,
      "grad_norm": 320.8076477050781,
      "learning_rate": 0.00018563340817174344,
      "loss": 7.3619,
      "step": 147840
    },
    {
      "epoch": 17.79181708784597,
      "grad_norm": 1550.425537109375,
      "learning_rate": 0.00018563144350019283,
      "loss": 7.3577,
      "step": 147850
    },
    {
      "epoch": 17.793020457280385,
      "grad_norm": 145.00375366210938,
      "learning_rate": 0.0001856294787047124,
      "loss": 7.283,
      "step": 147860
    },
    {
      "epoch": 17.794223826714802,
      "grad_norm": 266.2540588378906,
      "learning_rate": 0.00018562751378530497,
      "loss": 7.4418,
      "step": 147870
    },
    {
      "epoch": 17.79542719614922,
      "grad_norm": 363.5646667480469,
      "learning_rate": 0.00018562554874197335,
      "loss": 7.3938,
      "step": 147880
    },
    {
      "epoch": 17.796630565583634,
      "grad_norm": 182.42591857910156,
      "learning_rate": 0.00018562358357472047,
      "loss": 7.4441,
      "step": 147890
    },
    {
      "epoch": 17.79783393501805,
      "grad_norm": 307.95489501953125,
      "learning_rate": 0.0001856216182835491,
      "loss": 7.458,
      "step": 147900
    },
    {
      "epoch": 17.799037304452465,
      "grad_norm": 797.8807983398438,
      "learning_rate": 0.0001856196528684621,
      "loss": 7.3688,
      "step": 147910
    },
    {
      "epoch": 17.800240673886883,
      "grad_norm": 314.898681640625,
      "learning_rate": 0.00018561768732946234,
      "loss": 7.4573,
      "step": 147920
    },
    {
      "epoch": 17.8014440433213,
      "grad_norm": 267.087158203125,
      "learning_rate": 0.00018561572166655267,
      "loss": 7.433,
      "step": 147930
    },
    {
      "epoch": 17.802647412755714,
      "grad_norm": 1739.5926513671875,
      "learning_rate": 0.00018561375587973586,
      "loss": 7.407,
      "step": 147940
    },
    {
      "epoch": 17.803850782190132,
      "grad_norm": 3493.8916015625,
      "learning_rate": 0.00018561178996901484,
      "loss": 7.3938,
      "step": 147950
    },
    {
      "epoch": 17.80505415162455,
      "grad_norm": 384.19293212890625,
      "learning_rate": 0.0001856098239343924,
      "loss": 7.382,
      "step": 147960
    },
    {
      "epoch": 17.806257521058964,
      "grad_norm": 3504.621337890625,
      "learning_rate": 0.00018560785777587145,
      "loss": 7.455,
      "step": 147970
    },
    {
      "epoch": 17.80746089049338,
      "grad_norm": 631.4364624023438,
      "learning_rate": 0.00018560589149345475,
      "loss": 7.4463,
      "step": 147980
    },
    {
      "epoch": 17.8086642599278,
      "grad_norm": 334.4178771972656,
      "learning_rate": 0.0001856039250871452,
      "loss": 7.3855,
      "step": 147990
    },
    {
      "epoch": 17.809867629362213,
      "grad_norm": 1590.579345703125,
      "learning_rate": 0.00018560195855694565,
      "loss": 7.4281,
      "step": 148000
    },
    {
      "epoch": 17.81107099879663,
      "grad_norm": 219.71746826171875,
      "learning_rate": 0.00018559999190285894,
      "loss": 7.4253,
      "step": 148010
    },
    {
      "epoch": 17.812274368231048,
      "grad_norm": 156.32762145996094,
      "learning_rate": 0.00018559802512488788,
      "loss": 7.4864,
      "step": 148020
    },
    {
      "epoch": 17.813477737665462,
      "grad_norm": 744.1489868164062,
      "learning_rate": 0.0001855960582230354,
      "loss": 7.3859,
      "step": 148030
    },
    {
      "epoch": 17.81468110709988,
      "grad_norm": 266.0203552246094,
      "learning_rate": 0.0001855940911973042,
      "loss": 7.4715,
      "step": 148040
    },
    {
      "epoch": 17.815884476534297,
      "grad_norm": 357.6445617675781,
      "learning_rate": 0.0001855921240476973,
      "loss": 7.4881,
      "step": 148050
    },
    {
      "epoch": 17.81708784596871,
      "grad_norm": 134.54136657714844,
      "learning_rate": 0.0001855901567742174,
      "loss": 7.5326,
      "step": 148060
    },
    {
      "epoch": 17.81829121540313,
      "grad_norm": 476.8880920410156,
      "learning_rate": 0.00018558818937686744,
      "loss": 7.6585,
      "step": 148070
    },
    {
      "epoch": 17.819494584837546,
      "grad_norm": 976.9490356445312,
      "learning_rate": 0.00018558622185565023,
      "loss": 7.5011,
      "step": 148080
    },
    {
      "epoch": 17.82069795427196,
      "grad_norm": 250.97000122070312,
      "learning_rate": 0.00018558425421056867,
      "loss": 7.4828,
      "step": 148090
    },
    {
      "epoch": 17.821901323706378,
      "grad_norm": 202.69407653808594,
      "learning_rate": 0.00018558228644162552,
      "loss": 7.5045,
      "step": 148100
    },
    {
      "epoch": 17.823104693140795,
      "grad_norm": 265.7468566894531,
      "learning_rate": 0.00018558031854882368,
      "loss": 7.47,
      "step": 148110
    },
    {
      "epoch": 17.82430806257521,
      "grad_norm": 211.81085205078125,
      "learning_rate": 0.000185578350532166,
      "loss": 7.561,
      "step": 148120
    },
    {
      "epoch": 17.825511432009627,
      "grad_norm": 53.499542236328125,
      "learning_rate": 0.0001855763823916553,
      "loss": 7.5194,
      "step": 148130
    },
    {
      "epoch": 17.826714801444044,
      "grad_norm": 73.63157653808594,
      "learning_rate": 0.00018557441412729445,
      "loss": 7.5213,
      "step": 148140
    },
    {
      "epoch": 17.82791817087846,
      "grad_norm": 136.89990234375,
      "learning_rate": 0.00018557244573908633,
      "loss": 7.4758,
      "step": 148150
    },
    {
      "epoch": 17.829121540312876,
      "grad_norm": 66.81103515625,
      "learning_rate": 0.00018557047722703374,
      "loss": 7.4555,
      "step": 148160
    },
    {
      "epoch": 17.830324909747294,
      "grad_norm": 237.77450561523438,
      "learning_rate": 0.00018556850859113952,
      "loss": 7.5492,
      "step": 148170
    },
    {
      "epoch": 17.831528279181708,
      "grad_norm": 530.3276977539062,
      "learning_rate": 0.00018556653983140654,
      "loss": 7.4046,
      "step": 148180
    },
    {
      "epoch": 17.832731648616125,
      "grad_norm": 108.76467895507812,
      "learning_rate": 0.00018556457094783766,
      "loss": 7.4829,
      "step": 148190
    },
    {
      "epoch": 17.833935018050543,
      "grad_norm": 234.10568237304688,
      "learning_rate": 0.00018556260194043574,
      "loss": 7.4189,
      "step": 148200
    },
    {
      "epoch": 17.835138387484957,
      "grad_norm": 62.14372253417969,
      "learning_rate": 0.0001855606328092036,
      "loss": 7.5113,
      "step": 148210
    },
    {
      "epoch": 17.836341756919374,
      "grad_norm": 113.21151733398438,
      "learning_rate": 0.0001855586635541441,
      "loss": 7.4875,
      "step": 148220
    },
    {
      "epoch": 17.837545126353792,
      "grad_norm": 221.73915100097656,
      "learning_rate": 0.00018555669417526008,
      "loss": 7.3501,
      "step": 148230
    },
    {
      "epoch": 17.838748495788206,
      "grad_norm": 52.65148162841797,
      "learning_rate": 0.0001855547246725544,
      "loss": 7.4315,
      "step": 148240
    },
    {
      "epoch": 17.839951865222623,
      "grad_norm": 52.15606689453125,
      "learning_rate": 0.00018555275504602992,
      "loss": 7.3908,
      "step": 148250
    },
    {
      "epoch": 17.84115523465704,
      "grad_norm": 476.1154479980469,
      "learning_rate": 0.00018555078529568946,
      "loss": 7.4344,
      "step": 148260
    },
    {
      "epoch": 17.842358604091455,
      "grad_norm": 309.08978271484375,
      "learning_rate": 0.0001855488154215359,
      "loss": 7.3377,
      "step": 148270
    },
    {
      "epoch": 17.843561973525873,
      "grad_norm": 74.35446166992188,
      "learning_rate": 0.0001855468454235721,
      "loss": 7.4406,
      "step": 148280
    },
    {
      "epoch": 17.84476534296029,
      "grad_norm": 119.46814727783203,
      "learning_rate": 0.0001855448753018009,
      "loss": 7.4396,
      "step": 148290
    },
    {
      "epoch": 17.845968712394704,
      "grad_norm": 101.33686828613281,
      "learning_rate": 0.0001855429050562251,
      "loss": 7.4416,
      "step": 148300
    },
    {
      "epoch": 17.84717208182912,
      "grad_norm": 126.79830169677734,
      "learning_rate": 0.00018554093468684766,
      "loss": 7.5306,
      "step": 148310
    },
    {
      "epoch": 17.84837545126354,
      "grad_norm": 118.94538879394531,
      "learning_rate": 0.00018553896419367131,
      "loss": 7.4806,
      "step": 148320
    },
    {
      "epoch": 17.849578820697953,
      "grad_norm": 185.35955810546875,
      "learning_rate": 0.00018553699357669898,
      "loss": 7.3903,
      "step": 148330
    },
    {
      "epoch": 17.85078219013237,
      "grad_norm": 46.131656646728516,
      "learning_rate": 0.00018553502283593352,
      "loss": 7.467,
      "step": 148340
    },
    {
      "epoch": 17.85198555956679,
      "grad_norm": 51.31923294067383,
      "learning_rate": 0.00018553305197137774,
      "loss": 7.4876,
      "step": 148350
    },
    {
      "epoch": 17.853188929001202,
      "grad_norm": 47.45521926879883,
      "learning_rate": 0.00018553108098303452,
      "loss": 7.5148,
      "step": 148360
    },
    {
      "epoch": 17.85439229843562,
      "grad_norm": 86.61430358886719,
      "learning_rate": 0.00018552910987090676,
      "loss": 7.4264,
      "step": 148370
    },
    {
      "epoch": 17.855595667870038,
      "grad_norm": 339.5720520019531,
      "learning_rate": 0.00018552713863499717,
      "loss": 7.4492,
      "step": 148380
    },
    {
      "epoch": 17.85679903730445,
      "grad_norm": 159.4058837890625,
      "learning_rate": 0.00018552516727530875,
      "loss": 7.4231,
      "step": 148390
    },
    {
      "epoch": 17.85800240673887,
      "grad_norm": 103.97511291503906,
      "learning_rate": 0.0001855231957918443,
      "loss": 7.3985,
      "step": 148400
    },
    {
      "epoch": 17.859205776173287,
      "grad_norm": 77.51788330078125,
      "learning_rate": 0.00018552122418460666,
      "loss": 7.402,
      "step": 148410
    },
    {
      "epoch": 17.8604091456077,
      "grad_norm": 75.72032165527344,
      "learning_rate": 0.0001855192524535987,
      "loss": 7.3733,
      "step": 148420
    },
    {
      "epoch": 17.86161251504212,
      "grad_norm": 45.09842300415039,
      "learning_rate": 0.00018551728059882326,
      "loss": 7.42,
      "step": 148430
    },
    {
      "epoch": 17.862815884476536,
      "grad_norm": 49.23476791381836,
      "learning_rate": 0.00018551530862028322,
      "loss": 7.4076,
      "step": 148440
    },
    {
      "epoch": 17.86401925391095,
      "grad_norm": 76.97235870361328,
      "learning_rate": 0.00018551333651798137,
      "loss": 7.4252,
      "step": 148450
    },
    {
      "epoch": 17.865222623345367,
      "grad_norm": 53.90160369873047,
      "learning_rate": 0.00018551136429192065,
      "loss": 7.4592,
      "step": 148460
    },
    {
      "epoch": 17.866425992779785,
      "grad_norm": 233.33580017089844,
      "learning_rate": 0.00018550939194210387,
      "loss": 7.4445,
      "step": 148470
    },
    {
      "epoch": 17.8676293622142,
      "grad_norm": 161.20196533203125,
      "learning_rate": 0.00018550741946853387,
      "loss": 7.4449,
      "step": 148480
    },
    {
      "epoch": 17.868832731648617,
      "grad_norm": 53.023536682128906,
      "learning_rate": 0.00018550544687121354,
      "loss": 7.4052,
      "step": 148490
    },
    {
      "epoch": 17.870036101083034,
      "grad_norm": 174.26846313476562,
      "learning_rate": 0.00018550347415014572,
      "loss": 7.4383,
      "step": 148500
    },
    {
      "epoch": 17.871239470517448,
      "grad_norm": 148.2725372314453,
      "learning_rate": 0.0001855015013053332,
      "loss": 7.4779,
      "step": 148510
    },
    {
      "epoch": 17.872442839951866,
      "grad_norm": 237.47178649902344,
      "learning_rate": 0.000185499528336779,
      "loss": 7.4529,
      "step": 148520
    },
    {
      "epoch": 17.87364620938628,
      "grad_norm": 80.06126403808594,
      "learning_rate": 0.0001854975552444858,
      "loss": 7.3896,
      "step": 148530
    },
    {
      "epoch": 17.874849578820697,
      "grad_norm": 115.9737548828125,
      "learning_rate": 0.00018549558202845656,
      "loss": 7.4874,
      "step": 148540
    },
    {
      "epoch": 17.876052948255115,
      "grad_norm": 50.52712631225586,
      "learning_rate": 0.0001854936086886941,
      "loss": 7.4149,
      "step": 148550
    },
    {
      "epoch": 17.87725631768953,
      "grad_norm": 176.19664001464844,
      "learning_rate": 0.00018549163522520127,
      "loss": 7.4091,
      "step": 148560
    },
    {
      "epoch": 17.878459687123947,
      "grad_norm": 108.81830596923828,
      "learning_rate": 0.00018548966163798095,
      "loss": 7.4882,
      "step": 148570
    },
    {
      "epoch": 17.879663056558364,
      "grad_norm": 27.57801628112793,
      "learning_rate": 0.00018548768792703597,
      "loss": 7.3848,
      "step": 148580
    },
    {
      "epoch": 17.880866425992778,
      "grad_norm": 170.5851593017578,
      "learning_rate": 0.00018548571409236924,
      "loss": 7.3669,
      "step": 148590
    },
    {
      "epoch": 17.882069795427196,
      "grad_norm": 79.25787353515625,
      "learning_rate": 0.00018548374013398352,
      "loss": 7.5217,
      "step": 148600
    },
    {
      "epoch": 17.883273164861613,
      "grad_norm": 579.5789184570312,
      "learning_rate": 0.00018548176605188177,
      "loss": 7.4688,
      "step": 148610
    },
    {
      "epoch": 17.884476534296027,
      "grad_norm": 213.94871520996094,
      "learning_rate": 0.00018547979184606678,
      "loss": 7.4313,
      "step": 148620
    },
    {
      "epoch": 17.885679903730445,
      "grad_norm": 105.76802062988281,
      "learning_rate": 0.0001854778175165414,
      "loss": 7.4935,
      "step": 148630
    },
    {
      "epoch": 17.886883273164862,
      "grad_norm": 140.32843017578125,
      "learning_rate": 0.00018547584306330853,
      "loss": 7.326,
      "step": 148640
    },
    {
      "epoch": 17.888086642599276,
      "grad_norm": 113.10185241699219,
      "learning_rate": 0.00018547386848637103,
      "loss": 7.3528,
      "step": 148650
    },
    {
      "epoch": 17.889290012033694,
      "grad_norm": 75.72783660888672,
      "learning_rate": 0.00018547189378573172,
      "loss": 7.4721,
      "step": 148660
    },
    {
      "epoch": 17.89049338146811,
      "grad_norm": 96.25650024414062,
      "learning_rate": 0.0001854699189613935,
      "loss": 7.4121,
      "step": 148670
    },
    {
      "epoch": 17.891696750902526,
      "grad_norm": 40.19290542602539,
      "learning_rate": 0.0001854679440133592,
      "loss": 7.4876,
      "step": 148680
    },
    {
      "epoch": 17.892900120336943,
      "grad_norm": 160.92198181152344,
      "learning_rate": 0.00018546596894163166,
      "loss": 7.4104,
      "step": 148690
    },
    {
      "epoch": 17.89410348977136,
      "grad_norm": 163.02830505371094,
      "learning_rate": 0.00018546399374621376,
      "loss": 7.4879,
      "step": 148700
    },
    {
      "epoch": 17.895306859205775,
      "grad_norm": 95.01387786865234,
      "learning_rate": 0.0001854620184271084,
      "loss": 7.434,
      "step": 148710
    },
    {
      "epoch": 17.896510228640192,
      "grad_norm": 135.92922973632812,
      "learning_rate": 0.0001854600429843184,
      "loss": 7.5638,
      "step": 148720
    },
    {
      "epoch": 17.89771359807461,
      "grad_norm": 50.186397552490234,
      "learning_rate": 0.00018545806741784656,
      "loss": 7.4918,
      "step": 148730
    },
    {
      "epoch": 17.898916967509024,
      "grad_norm": 60.19594955444336,
      "learning_rate": 0.00018545609172769583,
      "loss": 7.4068,
      "step": 148740
    },
    {
      "epoch": 17.90012033694344,
      "grad_norm": 111.28684997558594,
      "learning_rate": 0.00018545411591386904,
      "loss": 7.4894,
      "step": 148750
    },
    {
      "epoch": 17.90132370637786,
      "grad_norm": 409.53533935546875,
      "learning_rate": 0.00018545213997636905,
      "loss": 7.4667,
      "step": 148760
    },
    {
      "epoch": 17.902527075812273,
      "grad_norm": 522.3593139648438,
      "learning_rate": 0.00018545016391519872,
      "loss": 7.3409,
      "step": 148770
    },
    {
      "epoch": 17.90373044524669,
      "grad_norm": 115.39041900634766,
      "learning_rate": 0.00018544818773036088,
      "loss": 7.4288,
      "step": 148780
    },
    {
      "epoch": 17.904933814681108,
      "grad_norm": 156.78968811035156,
      "learning_rate": 0.00018544621142185845,
      "loss": 7.3581,
      "step": 148790
    },
    {
      "epoch": 17.906137184115522,
      "grad_norm": 84.30831909179688,
      "learning_rate": 0.00018544423498969423,
      "loss": 7.3875,
      "step": 148800
    },
    {
      "epoch": 17.90734055354994,
      "grad_norm": 66.62500762939453,
      "learning_rate": 0.00018544225843387114,
      "loss": 7.3363,
      "step": 148810
    },
    {
      "epoch": 17.908543922984357,
      "grad_norm": 93.46416473388672,
      "learning_rate": 0.00018544028175439197,
      "loss": 7.3188,
      "step": 148820
    },
    {
      "epoch": 17.90974729241877,
      "grad_norm": 115.00357055664062,
      "learning_rate": 0.00018543830495125962,
      "loss": 7.2761,
      "step": 148830
    },
    {
      "epoch": 17.91095066185319,
      "grad_norm": 396.2091979980469,
      "learning_rate": 0.00018543632802447697,
      "loss": 7.358,
      "step": 148840
    },
    {
      "epoch": 17.912154031287606,
      "grad_norm": 521.9418334960938,
      "learning_rate": 0.00018543435097404686,
      "loss": 7.3077,
      "step": 148850
    },
    {
      "epoch": 17.91335740072202,
      "grad_norm": 278.07623291015625,
      "learning_rate": 0.00018543237379997215,
      "loss": 7.3496,
      "step": 148860
    },
    {
      "epoch": 17.914560770156438,
      "grad_norm": 134.51907348632812,
      "learning_rate": 0.0001854303965022557,
      "loss": 7.2756,
      "step": 148870
    },
    {
      "epoch": 17.915764139590856,
      "grad_norm": 229.17855834960938,
      "learning_rate": 0.00018542841908090034,
      "loss": 7.4308,
      "step": 148880
    },
    {
      "epoch": 17.91696750902527,
      "grad_norm": 352.9532165527344,
      "learning_rate": 0.000185426441535909,
      "loss": 7.3495,
      "step": 148890
    },
    {
      "epoch": 17.918170878459687,
      "grad_norm": 75.5967788696289,
      "learning_rate": 0.0001854244638672845,
      "loss": 7.413,
      "step": 148900
    },
    {
      "epoch": 17.919374247894105,
      "grad_norm": 97.91011047363281,
      "learning_rate": 0.00018542248607502973,
      "loss": 7.3611,
      "step": 148910
    },
    {
      "epoch": 17.92057761732852,
      "grad_norm": 69.5083236694336,
      "learning_rate": 0.0001854205081591475,
      "loss": 7.4491,
      "step": 148920
    },
    {
      "epoch": 17.921780986762936,
      "grad_norm": 92.9021224975586,
      "learning_rate": 0.00018541853011964072,
      "loss": 7.3431,
      "step": 148930
    },
    {
      "epoch": 17.922984356197354,
      "grad_norm": 851.4580078125,
      "learning_rate": 0.00018541655195651222,
      "loss": 7.4176,
      "step": 148940
    },
    {
      "epoch": 17.924187725631768,
      "grad_norm": 592.9284057617188,
      "learning_rate": 0.00018541457366976491,
      "loss": 7.3375,
      "step": 148950
    },
    {
      "epoch": 17.925391095066185,
      "grad_norm": 570.1676025390625,
      "learning_rate": 0.0001854125952594016,
      "loss": 7.3668,
      "step": 148960
    },
    {
      "epoch": 17.926594464500603,
      "grad_norm": 923.6080322265625,
      "learning_rate": 0.0001854106167254252,
      "loss": 7.4887,
      "step": 148970
    },
    {
      "epoch": 17.927797833935017,
      "grad_norm": 110.47050476074219,
      "learning_rate": 0.00018540863806783855,
      "loss": 7.4161,
      "step": 148980
    },
    {
      "epoch": 17.929001203369435,
      "grad_norm": 4797.53125,
      "learning_rate": 0.0001854066592866445,
      "loss": 7.3715,
      "step": 148990
    },
    {
      "epoch": 17.930204572803852,
      "grad_norm": 92.46206665039062,
      "learning_rate": 0.00018540468038184593,
      "loss": 7.4468,
      "step": 149000
    },
    {
      "epoch": 17.931407942238266,
      "grad_norm": 92.92508697509766,
      "learning_rate": 0.0001854027013534457,
      "loss": 7.4184,
      "step": 149010
    },
    {
      "epoch": 17.932611311672684,
      "grad_norm": 831.9789428710938,
      "learning_rate": 0.00018540072220144666,
      "loss": 7.5265,
      "step": 149020
    },
    {
      "epoch": 17.9338146811071,
      "grad_norm": 65246.26953125,
      "learning_rate": 0.00018539874292585173,
      "loss": 7.3995,
      "step": 149030
    },
    {
      "epoch": 17.935018050541515,
      "grad_norm": 169.46783447265625,
      "learning_rate": 0.0001853967635266637,
      "loss": 7.4446,
      "step": 149040
    },
    {
      "epoch": 17.936221419975933,
      "grad_norm": 31.190034866333008,
      "learning_rate": 0.0001853947840038855,
      "loss": 7.4576,
      "step": 149050
    },
    {
      "epoch": 17.93742478941035,
      "grad_norm": 708.36279296875,
      "learning_rate": 0.00018539280435751992,
      "loss": 7.5075,
      "step": 149060
    },
    {
      "epoch": 17.938628158844764,
      "grad_norm": 759.9669189453125,
      "learning_rate": 0.00018539082458756992,
      "loss": 7.6613,
      "step": 149070
    },
    {
      "epoch": 17.939831528279182,
      "grad_norm": 32.73354721069336,
      "learning_rate": 0.0001853888446940383,
      "loss": 7.4739,
      "step": 149080
    },
    {
      "epoch": 17.9410348977136,
      "grad_norm": 138.23373413085938,
      "learning_rate": 0.0001853868646769279,
      "loss": 7.5144,
      "step": 149090
    },
    {
      "epoch": 17.942238267148014,
      "grad_norm": 60.550025939941406,
      "learning_rate": 0.00018538488453624167,
      "loss": 7.481,
      "step": 149100
    },
    {
      "epoch": 17.94344163658243,
      "grad_norm": 77.27680206298828,
      "learning_rate": 0.00018538290427198243,
      "loss": 7.3966,
      "step": 149110
    },
    {
      "epoch": 17.94464500601685,
      "grad_norm": 349.0970153808594,
      "learning_rate": 0.00018538092388415304,
      "loss": 7.4954,
      "step": 149120
    },
    {
      "epoch": 17.945848375451263,
      "grad_norm": 326.36029052734375,
      "learning_rate": 0.00018537894337275636,
      "loss": 7.457,
      "step": 149130
    },
    {
      "epoch": 17.94705174488568,
      "grad_norm": 529.8981323242188,
      "learning_rate": 0.00018537696273779528,
      "loss": 7.4533,
      "step": 149140
    },
    {
      "epoch": 17.948255114320098,
      "grad_norm": 54.149330139160156,
      "learning_rate": 0.00018537498197927264,
      "loss": 7.4147,
      "step": 149150
    },
    {
      "epoch": 17.949458483754512,
      "grad_norm": 63.90738296508789,
      "learning_rate": 0.00018537300109719134,
      "loss": 7.423,
      "step": 149160
    },
    {
      "epoch": 17.95066185318893,
      "grad_norm": 33.161537170410156,
      "learning_rate": 0.00018537102009155422,
      "loss": 7.4976,
      "step": 149170
    },
    {
      "epoch": 17.951865222623347,
      "grad_norm": 109.87821960449219,
      "learning_rate": 0.00018536903896236418,
      "loss": 7.4773,
      "step": 149180
    },
    {
      "epoch": 17.95306859205776,
      "grad_norm": 50.217193603515625,
      "learning_rate": 0.00018536705770962403,
      "loss": 7.4163,
      "step": 149190
    },
    {
      "epoch": 17.95427196149218,
      "grad_norm": 282.6112976074219,
      "learning_rate": 0.00018536507633333668,
      "loss": 7.3643,
      "step": 149200
    },
    {
      "epoch": 17.955475330926596,
      "grad_norm": 242.79551696777344,
      "learning_rate": 0.000185363094833505,
      "loss": 7.424,
      "step": 149210
    },
    {
      "epoch": 17.95667870036101,
      "grad_norm": 258.24517822265625,
      "learning_rate": 0.00018536111321013185,
      "loss": 7.339,
      "step": 149220
    },
    {
      "epoch": 17.957882069795428,
      "grad_norm": 73.15386199951172,
      "learning_rate": 0.0001853591314632201,
      "loss": 7.3714,
      "step": 149230
    },
    {
      "epoch": 17.959085439229845,
      "grad_norm": 90.18331909179688,
      "learning_rate": 0.00018535714959277258,
      "loss": 7.4321,
      "step": 149240
    },
    {
      "epoch": 17.96028880866426,
      "grad_norm": 178.28907775878906,
      "learning_rate": 0.00018535516759879222,
      "loss": 7.4123,
      "step": 149250
    },
    {
      "epoch": 17.961492178098677,
      "grad_norm": 41.904869079589844,
      "learning_rate": 0.00018535318548128184,
      "loss": 7.3803,
      "step": 149260
    },
    {
      "epoch": 17.96269554753309,
      "grad_norm": 189.78482055664062,
      "learning_rate": 0.00018535120324024434,
      "loss": 7.4245,
      "step": 149270
    },
    {
      "epoch": 17.96389891696751,
      "grad_norm": 117.08611297607422,
      "learning_rate": 0.0001853492208756826,
      "loss": 7.4566,
      "step": 149280
    },
    {
      "epoch": 17.965102286401926,
      "grad_norm": 96.96859741210938,
      "learning_rate": 0.00018534723838759941,
      "loss": 7.4392,
      "step": 149290
    },
    {
      "epoch": 17.96630565583634,
      "grad_norm": 653.5653076171875,
      "learning_rate": 0.00018534525577599772,
      "loss": 7.3919,
      "step": 149300
    },
    {
      "epoch": 17.967509025270758,
      "grad_norm": 4584.6845703125,
      "learning_rate": 0.00018534327304088038,
      "loss": 7.3663,
      "step": 149310
    },
    {
      "epoch": 17.968712394705175,
      "grad_norm": 607.0430297851562,
      "learning_rate": 0.00018534129018225027,
      "loss": 7.4837,
      "step": 149320
    },
    {
      "epoch": 17.96991576413959,
      "grad_norm": 380.6038818359375,
      "learning_rate": 0.0001853393072001102,
      "loss": 7.3697,
      "step": 149330
    },
    {
      "epoch": 17.971119133574007,
      "grad_norm": 394.68377685546875,
      "learning_rate": 0.0001853373240944631,
      "loss": 7.3179,
      "step": 149340
    },
    {
      "epoch": 17.972322503008424,
      "grad_norm": 542.9088134765625,
      "learning_rate": 0.00018533534086531185,
      "loss": 7.4092,
      "step": 149350
    },
    {
      "epoch": 17.97352587244284,
      "grad_norm": 1883.626220703125,
      "learning_rate": 0.00018533335751265924,
      "loss": 7.2972,
      "step": 149360
    },
    {
      "epoch": 17.974729241877256,
      "grad_norm": 1593.1341552734375,
      "learning_rate": 0.00018533137403650823,
      "loss": 7.3951,
      "step": 149370
    },
    {
      "epoch": 17.975932611311674,
      "grad_norm": 1005.785888671875,
      "learning_rate": 0.00018532939043686163,
      "loss": 7.3128,
      "step": 149380
    },
    {
      "epoch": 17.977135980746088,
      "grad_norm": 2285.49072265625,
      "learning_rate": 0.00018532740671372238,
      "loss": 7.4595,
      "step": 149390
    },
    {
      "epoch": 17.978339350180505,
      "grad_norm": 361.1526184082031,
      "learning_rate": 0.00018532542286709328,
      "loss": 7.4691,
      "step": 149400
    },
    {
      "epoch": 17.979542719614923,
      "grad_norm": 7603.2060546875,
      "learning_rate": 0.0001853234388969772,
      "loss": 7.3966,
      "step": 149410
    },
    {
      "epoch": 17.980746089049337,
      "grad_norm": 26465.263671875,
      "learning_rate": 0.00018532145480337707,
      "loss": 7.3971,
      "step": 149420
    },
    {
      "epoch": 17.981949458483754,
      "grad_norm": 1138.2674560546875,
      "learning_rate": 0.00018531947058629574,
      "loss": 7.2141,
      "step": 149430
    },
    {
      "epoch": 17.983152827918172,
      "grad_norm": 430.54022216796875,
      "learning_rate": 0.00018531748624573603,
      "loss": 7.4032,
      "step": 149440
    },
    {
      "epoch": 17.984356197352586,
      "grad_norm": 340.5308837890625,
      "learning_rate": 0.0001853155017817009,
      "loss": 7.4153,
      "step": 149450
    },
    {
      "epoch": 17.985559566787003,
      "grad_norm": 539.905029296875,
      "learning_rate": 0.00018531351719419317,
      "loss": 7.3884,
      "step": 149460
    },
    {
      "epoch": 17.98676293622142,
      "grad_norm": 515.404541015625,
      "learning_rate": 0.00018531153248321568,
      "loss": 7.3655,
      "step": 149470
    },
    {
      "epoch": 17.987966305655835,
      "grad_norm": 1004.9083862304688,
      "learning_rate": 0.00018530954764877137,
      "loss": 7.4721,
      "step": 149480
    },
    {
      "epoch": 17.989169675090253,
      "grad_norm": 64222.359375,
      "learning_rate": 0.00018530756269086307,
      "loss": 7.3745,
      "step": 149490
    },
    {
      "epoch": 17.99037304452467,
      "grad_norm": 117.87918090820312,
      "learning_rate": 0.00018530557760949368,
      "loss": 7.3749,
      "step": 149500
    },
    {
      "epoch": 17.991576413959084,
      "grad_norm": 6933.921875,
      "learning_rate": 0.00018530359240466604,
      "loss": 7.3617,
      "step": 149510
    },
    {
      "epoch": 17.9927797833935,
      "grad_norm": 84.59420013427734,
      "learning_rate": 0.00018530160707638305,
      "loss": 7.3928,
      "step": 149520
    },
    {
      "epoch": 17.99398315282792,
      "grad_norm": 408.6295471191406,
      "learning_rate": 0.00018529962162464755,
      "loss": 7.3956,
      "step": 149530
    },
    {
      "epoch": 17.995186522262333,
      "grad_norm": 19631.50390625,
      "learning_rate": 0.0001852976360494625,
      "loss": 7.4022,
      "step": 149540
    },
    {
      "epoch": 17.99638989169675,
      "grad_norm": 111.64810180664062,
      "learning_rate": 0.00018529565035083065,
      "loss": 7.4755,
      "step": 149550
    },
    {
      "epoch": 17.99759326113117,
      "grad_norm": 568.51025390625,
      "learning_rate": 0.00018529366452875494,
      "loss": 7.4421,
      "step": 149560
    },
    {
      "epoch": 17.998796630565582,
      "grad_norm": 84.73117065429688,
      "learning_rate": 0.0001852916785832383,
      "loss": 7.4819,
      "step": 149570
    },
    {
      "epoch": 18.0,
      "grad_norm": 111.88549041748047,
      "learning_rate": 0.0001852896925142835,
      "loss": 7.362,
      "step": 149580
    },
    {
      "epoch": 18.0,
      "eval_loss": 7.366647720336914,
      "eval_runtime": 118.9567,
      "eval_samples_per_second": 62.098,
      "eval_steps_per_second": 7.768,
      "step": 149580
    },
    {
      "epoch": 18.001203369434418,
      "grad_norm": 2082.1474609375,
      "learning_rate": 0.0001852877063218934,
      "loss": 7.3862,
      "step": 149590
    },
    {
      "epoch": 18.00240673886883,
      "grad_norm": 151.06271362304688,
      "learning_rate": 0.000185285720006071,
      "loss": 7.3733,
      "step": 149600
    },
    {
      "epoch": 18.00361010830325,
      "grad_norm": 338.38525390625,
      "learning_rate": 0.0001852837335668191,
      "loss": 7.3697,
      "step": 149610
    },
    {
      "epoch": 18.004813477737667,
      "grad_norm": 305.02740478515625,
      "learning_rate": 0.0001852817470041406,
      "loss": 7.3517,
      "step": 149620
    },
    {
      "epoch": 18.00601684717208,
      "grad_norm": 64.78070068359375,
      "learning_rate": 0.00018527976031803832,
      "loss": 7.3814,
      "step": 149630
    },
    {
      "epoch": 18.0072202166065,
      "grad_norm": 525.9762573242188,
      "learning_rate": 0.00018527777350851516,
      "loss": 7.3542,
      "step": 149640
    },
    {
      "epoch": 18.008423586040916,
      "grad_norm": 1789.5118408203125,
      "learning_rate": 0.00018527578657557406,
      "loss": 7.3802,
      "step": 149650
    },
    {
      "epoch": 18.00962695547533,
      "grad_norm": 172.57984924316406,
      "learning_rate": 0.0001852737995192178,
      "loss": 7.4317,
      "step": 149660
    },
    {
      "epoch": 18.010830324909747,
      "grad_norm": 3822.205078125,
      "learning_rate": 0.00018527181233944932,
      "loss": 7.2932,
      "step": 149670
    },
    {
      "epoch": 18.012033694344165,
      "grad_norm": 304.72161865234375,
      "learning_rate": 0.00018526982503627148,
      "loss": 7.334,
      "step": 149680
    },
    {
      "epoch": 18.01323706377858,
      "grad_norm": 103.07905578613281,
      "learning_rate": 0.00018526783760968712,
      "loss": 7.3166,
      "step": 149690
    },
    {
      "epoch": 18.014440433212997,
      "grad_norm": 111.76778411865234,
      "learning_rate": 0.0001852658500596992,
      "loss": 7.3012,
      "step": 149700
    },
    {
      "epoch": 18.015643802647414,
      "grad_norm": 173.53530883789062,
      "learning_rate": 0.0001852638623863105,
      "loss": 7.2736,
      "step": 149710
    },
    {
      "epoch": 18.016847172081828,
      "grad_norm": 1576.4041748046875,
      "learning_rate": 0.00018526187458952394,
      "loss": 7.398,
      "step": 149720
    },
    {
      "epoch": 18.018050541516246,
      "grad_norm": 1148.1187744140625,
      "learning_rate": 0.0001852598866693424,
      "loss": 7.3654,
      "step": 149730
    },
    {
      "epoch": 18.019253910950663,
      "grad_norm": 123.3941879272461,
      "learning_rate": 0.00018525789862576875,
      "loss": 7.4426,
      "step": 149740
    },
    {
      "epoch": 18.020457280385077,
      "grad_norm": 771.9910278320312,
      "learning_rate": 0.0001852559104588059,
      "loss": 7.3355,
      "step": 149750
    },
    {
      "epoch": 18.021660649819495,
      "grad_norm": 1503.333740234375,
      "learning_rate": 0.0001852539221684567,
      "loss": 7.3849,
      "step": 149760
    },
    {
      "epoch": 18.022864019253912,
      "grad_norm": 1722.6297607421875,
      "learning_rate": 0.00018525193375472398,
      "loss": 7.371,
      "step": 149770
    },
    {
      "epoch": 18.024067388688326,
      "grad_norm": 2357.4091796875,
      "learning_rate": 0.00018524994521761069,
      "loss": 7.3561,
      "step": 149780
    },
    {
      "epoch": 18.025270758122744,
      "grad_norm": 460.76104736328125,
      "learning_rate": 0.00018524795655711968,
      "loss": 7.3813,
      "step": 149790
    },
    {
      "epoch": 18.02647412755716,
      "grad_norm": 1920.828125,
      "learning_rate": 0.00018524596777325382,
      "loss": 7.4556,
      "step": 149800
    },
    {
      "epoch": 18.027677496991576,
      "grad_norm": 9213.4033203125,
      "learning_rate": 0.00018524397886601602,
      "loss": 7.3546,
      "step": 149810
    },
    {
      "epoch": 18.028880866425993,
      "grad_norm": 9927.3623046875,
      "learning_rate": 0.00018524198983540914,
      "loss": 7.3267,
      "step": 149820
    },
    {
      "epoch": 18.03008423586041,
      "grad_norm": 18445.11328125,
      "learning_rate": 0.00018524000068143604,
      "loss": 7.4944,
      "step": 149830
    },
    {
      "epoch": 18.031287605294825,
      "grad_norm": 8243.80078125,
      "learning_rate": 0.0001852380114040996,
      "loss": 7.5059,
      "step": 149840
    },
    {
      "epoch": 18.032490974729242,
      "grad_norm": 1434.2047119140625,
      "learning_rate": 0.00018523602200340276,
      "loss": 7.4293,
      "step": 149850
    },
    {
      "epoch": 18.03369434416366,
      "grad_norm": 983.7843017578125,
      "learning_rate": 0.0001852340324793483,
      "loss": 7.4002,
      "step": 149860
    },
    {
      "epoch": 18.034897713598074,
      "grad_norm": 3585.263427734375,
      "learning_rate": 0.00018523204283193919,
      "loss": 7.3998,
      "step": 149870
    },
    {
      "epoch": 18.03610108303249,
      "grad_norm": 3442.3017578125,
      "learning_rate": 0.0001852300530611782,
      "loss": 7.4419,
      "step": 149880
    },
    {
      "epoch": 18.03730445246691,
      "grad_norm": 743.7374267578125,
      "learning_rate": 0.00018522806316706837,
      "loss": 7.4102,
      "step": 149890
    },
    {
      "epoch": 18.038507821901323,
      "grad_norm": 1866.08251953125,
      "learning_rate": 0.00018522607314961246,
      "loss": 7.501,
      "step": 149900
    },
    {
      "epoch": 18.03971119133574,
      "grad_norm": 405.5949401855469,
      "learning_rate": 0.00018522408300881337,
      "loss": 7.3552,
      "step": 149910
    },
    {
      "epoch": 18.040914560770158,
      "grad_norm": 4595.52783203125,
      "learning_rate": 0.000185222092744674,
      "loss": 7.4292,
      "step": 149920
    },
    {
      "epoch": 18.042117930204572,
      "grad_norm": 388.3169250488281,
      "learning_rate": 0.0001852201023571972,
      "loss": 7.3558,
      "step": 149930
    },
    {
      "epoch": 18.04332129963899,
      "grad_norm": 1000.1346435546875,
      "learning_rate": 0.0001852181118463859,
      "loss": 7.3738,
      "step": 149940
    },
    {
      "epoch": 18.044524669073404,
      "grad_norm": 7603.63720703125,
      "learning_rate": 0.00018521612121224294,
      "loss": 7.4176,
      "step": 149950
    },
    {
      "epoch": 18.04572803850782,
      "grad_norm": 334.28271484375,
      "learning_rate": 0.0001852141304547712,
      "loss": 7.4004,
      "step": 149960
    },
    {
      "epoch": 18.04693140794224,
      "grad_norm": 1391.88037109375,
      "learning_rate": 0.0001852121395739736,
      "loss": 7.3678,
      "step": 149970
    },
    {
      "epoch": 18.048134777376653,
      "grad_norm": 27133.986328125,
      "learning_rate": 0.000185210148569853,
      "loss": 7.4173,
      "step": 149980
    },
    {
      "epoch": 18.04933814681107,
      "grad_norm": 654.1278076171875,
      "learning_rate": 0.00018520815744241226,
      "loss": 7.4461,
      "step": 149990
    },
    {
      "epoch": 18.050541516245488,
      "grad_norm": 653.0599365234375,
      "learning_rate": 0.00018520616619165429,
      "loss": 7.4052,
      "step": 150000
    },
    {
      "epoch": 18.051744885679902,
      "grad_norm": 17197.240234375,
      "learning_rate": 0.00018520417481758194,
      "loss": 7.4436,
      "step": 150010
    },
    {
      "epoch": 18.05294825511432,
      "grad_norm": 991.9915771484375,
      "learning_rate": 0.0001852021833201981,
      "loss": 7.4564,
      "step": 150020
    },
    {
      "epoch": 18.054151624548737,
      "grad_norm": 2691.764892578125,
      "learning_rate": 0.00018520019169950572,
      "loss": 7.4842,
      "step": 150030
    },
    {
      "epoch": 18.05535499398315,
      "grad_norm": 6338.83544921875,
      "learning_rate": 0.0001851981999555076,
      "loss": 7.3698,
      "step": 150040
    },
    {
      "epoch": 18.05655836341757,
      "grad_norm": 185.8012237548828,
      "learning_rate": 0.00018519620808820663,
      "loss": 7.2721,
      "step": 150050
    },
    {
      "epoch": 18.057761732851986,
      "grad_norm": 339.1905212402344,
      "learning_rate": 0.00018519421609760573,
      "loss": 7.3596,
      "step": 150060
    },
    {
      "epoch": 18.0589651022864,
      "grad_norm": 10055.3291015625,
      "learning_rate": 0.00018519222398370776,
      "loss": 7.4733,
      "step": 150070
    },
    {
      "epoch": 18.060168471720818,
      "grad_norm": 384.69122314453125,
      "learning_rate": 0.00018519023174651562,
      "loss": 7.4054,
      "step": 150080
    },
    {
      "epoch": 18.061371841155236,
      "grad_norm": 1484.3526611328125,
      "learning_rate": 0.00018518823938603219,
      "loss": 7.4336,
      "step": 150090
    },
    {
      "epoch": 18.06257521058965,
      "grad_norm": 655.484130859375,
      "learning_rate": 0.00018518624690226033,
      "loss": 7.409,
      "step": 150100
    },
    {
      "epoch": 18.063778580024067,
      "grad_norm": 399.81085205078125,
      "learning_rate": 0.00018518425429520292,
      "loss": 7.4643,
      "step": 150110
    },
    {
      "epoch": 18.064981949458485,
      "grad_norm": 777.1300659179688,
      "learning_rate": 0.00018518226156486288,
      "loss": 7.4887,
      "step": 150120
    },
    {
      "epoch": 18.0661853188929,
      "grad_norm": 1444.622314453125,
      "learning_rate": 0.00018518026871124307,
      "loss": 7.357,
      "step": 150130
    },
    {
      "epoch": 18.067388688327316,
      "grad_norm": 2199.203125,
      "learning_rate": 0.00018517827573434642,
      "loss": 7.3952,
      "step": 150140
    },
    {
      "epoch": 18.068592057761734,
      "grad_norm": 921.2860107421875,
      "learning_rate": 0.00018517628263417572,
      "loss": 7.4424,
      "step": 150150
    },
    {
      "epoch": 18.069795427196148,
      "grad_norm": 1011.6319580078125,
      "learning_rate": 0.00018517428941073396,
      "loss": 7.4433,
      "step": 150160
    },
    {
      "epoch": 18.070998796630565,
      "grad_norm": 355.4660339355469,
      "learning_rate": 0.00018517229606402393,
      "loss": 7.4623,
      "step": 150170
    },
    {
      "epoch": 18.072202166064983,
      "grad_norm": 336.6348876953125,
      "learning_rate": 0.0001851703025940486,
      "loss": 7.4286,
      "step": 150180
    },
    {
      "epoch": 18.073405535499397,
      "grad_norm": 1935.4490966796875,
      "learning_rate": 0.00018516830900081076,
      "loss": 7.3342,
      "step": 150190
    },
    {
      "epoch": 18.074608904933815,
      "grad_norm": 1970.3548583984375,
      "learning_rate": 0.0001851663152843134,
      "loss": 7.4098,
      "step": 150200
    },
    {
      "epoch": 18.075812274368232,
      "grad_norm": 10640.7919921875,
      "learning_rate": 0.00018516432144455934,
      "loss": 7.4296,
      "step": 150210
    },
    {
      "epoch": 18.077015643802646,
      "grad_norm": 5139.22802734375,
      "learning_rate": 0.00018516232748155148,
      "loss": 7.4374,
      "step": 150220
    },
    {
      "epoch": 18.078219013237064,
      "grad_norm": 6197.271484375,
      "learning_rate": 0.0001851603333952927,
      "loss": 7.4404,
      "step": 150230
    },
    {
      "epoch": 18.07942238267148,
      "grad_norm": 814.4199829101562,
      "learning_rate": 0.0001851583391857859,
      "loss": 7.4299,
      "step": 150240
    },
    {
      "epoch": 18.080625752105895,
      "grad_norm": 7303.916015625,
      "learning_rate": 0.00018515634485303395,
      "loss": 7.3515,
      "step": 150250
    },
    {
      "epoch": 18.081829121540313,
      "grad_norm": 37418.73046875,
      "learning_rate": 0.00018515435039703974,
      "loss": 7.4337,
      "step": 150260
    },
    {
      "epoch": 18.08303249097473,
      "grad_norm": 2385.81640625,
      "learning_rate": 0.0001851523558178062,
      "loss": 7.393,
      "step": 150270
    },
    {
      "epoch": 18.084235860409144,
      "grad_norm": 365.7031555175781,
      "learning_rate": 0.00018515036111533612,
      "loss": 7.4232,
      "step": 150280
    },
    {
      "epoch": 18.085439229843562,
      "grad_norm": 4888.36572265625,
      "learning_rate": 0.0001851483662896325,
      "loss": 7.3289,
      "step": 150290
    },
    {
      "epoch": 18.08664259927798,
      "grad_norm": 719.6381225585938,
      "learning_rate": 0.00018514637134069812,
      "loss": 7.4972,
      "step": 150300
    },
    {
      "epoch": 18.087845968712394,
      "grad_norm": 4957.48974609375,
      "learning_rate": 0.00018514437626853596,
      "loss": 7.557,
      "step": 150310
    },
    {
      "epoch": 18.08904933814681,
      "grad_norm": 5745.91552734375,
      "learning_rate": 0.00018514238107314882,
      "loss": 7.4551,
      "step": 150320
    },
    {
      "epoch": 18.09025270758123,
      "grad_norm": 3058.720703125,
      "learning_rate": 0.0001851403857545397,
      "loss": 7.4735,
      "step": 150330
    },
    {
      "epoch": 18.091456077015643,
      "grad_norm": 983.1807250976562,
      "learning_rate": 0.00018513839031271135,
      "loss": 7.4255,
      "step": 150340
    },
    {
      "epoch": 18.09265944645006,
      "grad_norm": 190.457763671875,
      "learning_rate": 0.00018513639474766676,
      "loss": 7.4275,
      "step": 150350
    },
    {
      "epoch": 18.093862815884478,
      "grad_norm": 136.7671661376953,
      "learning_rate": 0.0001851343990594088,
      "loss": 7.518,
      "step": 150360
    },
    {
      "epoch": 18.095066185318892,
      "grad_norm": 62.439186096191406,
      "learning_rate": 0.00018513240324794034,
      "loss": 7.5152,
      "step": 150370
    },
    {
      "epoch": 18.09626955475331,
      "grad_norm": 53.048553466796875,
      "learning_rate": 0.00018513040731326426,
      "loss": 7.5674,
      "step": 150380
    },
    {
      "epoch": 18.097472924187727,
      "grad_norm": 30.955257415771484,
      "learning_rate": 0.0001851284112553835,
      "loss": 7.5671,
      "step": 150390
    },
    {
      "epoch": 18.09867629362214,
      "grad_norm": 35.14524459838867,
      "learning_rate": 0.00018512641507430086,
      "loss": 7.5236,
      "step": 150400
    },
    {
      "epoch": 18.09987966305656,
      "grad_norm": 105.49443817138672,
      "learning_rate": 0.0001851244187700193,
      "loss": 7.53,
      "step": 150410
    },
    {
      "epoch": 18.101083032490976,
      "grad_norm": 26.84381103515625,
      "learning_rate": 0.0001851224223425417,
      "loss": 7.5037,
      "step": 150420
    },
    {
      "epoch": 18.10228640192539,
      "grad_norm": 55.635597229003906,
      "learning_rate": 0.00018512042579187094,
      "loss": 7.4922,
      "step": 150430
    },
    {
      "epoch": 18.103489771359808,
      "grad_norm": 66.30591583251953,
      "learning_rate": 0.00018511842911800988,
      "loss": 7.5733,
      "step": 150440
    },
    {
      "epoch": 18.104693140794225,
      "grad_norm": 111.55799102783203,
      "learning_rate": 0.00018511643232096148,
      "loss": 7.5171,
      "step": 150450
    },
    {
      "epoch": 18.10589651022864,
      "grad_norm": 203.03775024414062,
      "learning_rate": 0.00018511443540072855,
      "loss": 7.5406,
      "step": 150460
    },
    {
      "epoch": 18.107099879663057,
      "grad_norm": 65.03297424316406,
      "learning_rate": 0.00018511243835731404,
      "loss": 7.3924,
      "step": 150470
    },
    {
      "epoch": 18.108303249097474,
      "grad_norm": 247.0765380859375,
      "learning_rate": 0.0001851104411907208,
      "loss": 7.5306,
      "step": 150480
    },
    {
      "epoch": 18.10950661853189,
      "grad_norm": 105.78290557861328,
      "learning_rate": 0.00018510844390095177,
      "loss": 7.4942,
      "step": 150490
    },
    {
      "epoch": 18.110709987966306,
      "grad_norm": 82.8394775390625,
      "learning_rate": 0.00018510644648800977,
      "loss": 7.5812,
      "step": 150500
    },
    {
      "epoch": 18.111913357400724,
      "grad_norm": 185.19444274902344,
      "learning_rate": 0.00018510444895189777,
      "loss": 7.3928,
      "step": 150510
    },
    {
      "epoch": 18.113116726835138,
      "grad_norm": 318.8232421875,
      "learning_rate": 0.0001851024512926186,
      "loss": 7.3806,
      "step": 150520
    },
    {
      "epoch": 18.114320096269555,
      "grad_norm": 199.4155731201172,
      "learning_rate": 0.00018510045351017516,
      "loss": 7.4369,
      "step": 150530
    },
    {
      "epoch": 18.115523465703973,
      "grad_norm": 200.80133056640625,
      "learning_rate": 0.00018509845560457037,
      "loss": 7.4884,
      "step": 150540
    },
    {
      "epoch": 18.116726835138387,
      "grad_norm": 68.57654571533203,
      "learning_rate": 0.0001850964575758071,
      "loss": 7.3612,
      "step": 150550
    },
    {
      "epoch": 18.117930204572804,
      "grad_norm": 155.61082458496094,
      "learning_rate": 0.00018509445942388824,
      "loss": 7.3387,
      "step": 150560
    },
    {
      "epoch": 18.119133574007222,
      "grad_norm": 89.49139404296875,
      "learning_rate": 0.0001850924611488167,
      "loss": 7.355,
      "step": 150570
    },
    {
      "epoch": 18.120336943441636,
      "grad_norm": 200.8621826171875,
      "learning_rate": 0.00018509046275059536,
      "loss": 7.3875,
      "step": 150580
    },
    {
      "epoch": 18.121540312876053,
      "grad_norm": 156.1785888671875,
      "learning_rate": 0.0001850884642292271,
      "loss": 7.2877,
      "step": 150590
    },
    {
      "epoch": 18.12274368231047,
      "grad_norm": 99.13287353515625,
      "learning_rate": 0.00018508646558471482,
      "loss": 7.4628,
      "step": 150600
    },
    {
      "epoch": 18.123947051744885,
      "grad_norm": 241.97088623046875,
      "learning_rate": 0.00018508446681706145,
      "loss": 7.3007,
      "step": 150610
    },
    {
      "epoch": 18.125150421179303,
      "grad_norm": 103.4607162475586,
      "learning_rate": 0.00018508246792626985,
      "loss": 7.4595,
      "step": 150620
    },
    {
      "epoch": 18.126353790613717,
      "grad_norm": 279.5429992675781,
      "learning_rate": 0.00018508046891234288,
      "loss": 7.3618,
      "step": 150630
    },
    {
      "epoch": 18.127557160048134,
      "grad_norm": 137.11724853515625,
      "learning_rate": 0.00018507846977528346,
      "loss": 7.3999,
      "step": 150640
    },
    {
      "epoch": 18.128760529482552,
      "grad_norm": 111.09395599365234,
      "learning_rate": 0.00018507647051509453,
      "loss": 7.4563,
      "step": 150650
    },
    {
      "epoch": 18.129963898916966,
      "grad_norm": 143.97569274902344,
      "learning_rate": 0.0001850744711317789,
      "loss": 7.3328,
      "step": 150660
    },
    {
      "epoch": 18.131167268351383,
      "grad_norm": 85.04871368408203,
      "learning_rate": 0.00018507247162533955,
      "loss": 7.4279,
      "step": 150670
    },
    {
      "epoch": 18.1323706377858,
      "grad_norm": 270.6744689941406,
      "learning_rate": 0.00018507047199577928,
      "loss": 7.3959,
      "step": 150680
    },
    {
      "epoch": 18.133574007220215,
      "grad_norm": 67.41458129882812,
      "learning_rate": 0.00018506847224310105,
      "loss": 7.3664,
      "step": 150690
    },
    {
      "epoch": 18.134777376654633,
      "grad_norm": 134.37625122070312,
      "learning_rate": 0.00018506647236730774,
      "loss": 7.3415,
      "step": 150700
    },
    {
      "epoch": 18.13598074608905,
      "grad_norm": 253.5409393310547,
      "learning_rate": 0.00018506447236840226,
      "loss": 7.3253,
      "step": 150710
    },
    {
      "epoch": 18.137184115523464,
      "grad_norm": 107.16642761230469,
      "learning_rate": 0.00018506247224638747,
      "loss": 7.3727,
      "step": 150720
    },
    {
      "epoch": 18.13838748495788,
      "grad_norm": 203.11355590820312,
      "learning_rate": 0.0001850604720012663,
      "loss": 7.4762,
      "step": 150730
    },
    {
      "epoch": 18.1395908543923,
      "grad_norm": 211.16729736328125,
      "learning_rate": 0.00018505847163304158,
      "loss": 7.4892,
      "step": 150740
    },
    {
      "epoch": 18.140794223826713,
      "grad_norm": 92.10486602783203,
      "learning_rate": 0.0001850564711417163,
      "loss": 7.3412,
      "step": 150750
    },
    {
      "epoch": 18.14199759326113,
      "grad_norm": 149.5958709716797,
      "learning_rate": 0.0001850544705272933,
      "loss": 7.3792,
      "step": 150760
    },
    {
      "epoch": 18.14320096269555,
      "grad_norm": 145.82102966308594,
      "learning_rate": 0.0001850524697897755,
      "loss": 7.406,
      "step": 150770
    },
    {
      "epoch": 18.144404332129962,
      "grad_norm": 198.8045654296875,
      "learning_rate": 0.0001850504689291657,
      "loss": 7.3768,
      "step": 150780
    },
    {
      "epoch": 18.14560770156438,
      "grad_norm": 115.61459350585938,
      "learning_rate": 0.00018504846794546693,
      "loss": 7.3904,
      "step": 150790
    },
    {
      "epoch": 18.146811070998798,
      "grad_norm": 98.6722183227539,
      "learning_rate": 0.00018504646683868204,
      "loss": 7.3789,
      "step": 150800
    },
    {
      "epoch": 18.14801444043321,
      "grad_norm": 156.54734802246094,
      "learning_rate": 0.0001850444656088139,
      "loss": 7.2916,
      "step": 150810
    },
    {
      "epoch": 18.14921780986763,
      "grad_norm": 105.44517517089844,
      "learning_rate": 0.00018504246425586542,
      "loss": 7.3621,
      "step": 150820
    },
    {
      "epoch": 18.150421179302047,
      "grad_norm": 177.7098388671875,
      "learning_rate": 0.0001850404627798395,
      "loss": 7.4214,
      "step": 150830
    },
    {
      "epoch": 18.15162454873646,
      "grad_norm": 88.13648223876953,
      "learning_rate": 0.00018503846118073902,
      "loss": 7.4682,
      "step": 150840
    },
    {
      "epoch": 18.15282791817088,
      "grad_norm": 93.81919860839844,
      "learning_rate": 0.00018503645945856692,
      "loss": 7.3496,
      "step": 150850
    },
    {
      "epoch": 18.154031287605296,
      "grad_norm": 137.61151123046875,
      "learning_rate": 0.00018503445761332605,
      "loss": 7.4063,
      "step": 150860
    },
    {
      "epoch": 18.15523465703971,
      "grad_norm": 57.4403076171875,
      "learning_rate": 0.00018503245564501931,
      "loss": 7.4349,
      "step": 150870
    },
    {
      "epoch": 18.156438026474127,
      "grad_norm": 138.23391723632812,
      "learning_rate": 0.00018503045355364964,
      "loss": 7.3763,
      "step": 150880
    },
    {
      "epoch": 18.157641395908545,
      "grad_norm": 151.458740234375,
      "learning_rate": 0.00018502845133921988,
      "loss": 7.35,
      "step": 150890
    },
    {
      "epoch": 18.15884476534296,
      "grad_norm": 106.25843811035156,
      "learning_rate": 0.000185026449001733,
      "loss": 7.3819,
      "step": 150900
    },
    {
      "epoch": 18.160048134777377,
      "grad_norm": 126.08541107177734,
      "learning_rate": 0.00018502444654119182,
      "loss": 7.467,
      "step": 150910
    },
    {
      "epoch": 18.161251504211794,
      "grad_norm": 122.78411102294922,
      "learning_rate": 0.0001850224439575993,
      "loss": 7.3777,
      "step": 150920
    },
    {
      "epoch": 18.162454873646208,
      "grad_norm": 255.74432373046875,
      "learning_rate": 0.00018502044125095828,
      "loss": 7.3988,
      "step": 150930
    },
    {
      "epoch": 18.163658243080626,
      "grad_norm": 167.97303771972656,
      "learning_rate": 0.00018501843842127174,
      "loss": 7.523,
      "step": 150940
    },
    {
      "epoch": 18.164861612515043,
      "grad_norm": 78.77965545654297,
      "learning_rate": 0.00018501643546854247,
      "loss": 7.369,
      "step": 150950
    },
    {
      "epoch": 18.166064981949457,
      "grad_norm": 172.5324249267578,
      "learning_rate": 0.00018501443239277347,
      "loss": 7.457,
      "step": 150960
    },
    {
      "epoch": 18.167268351383875,
      "grad_norm": 69.45236206054688,
      "learning_rate": 0.00018501242919396758,
      "loss": 7.3798,
      "step": 150970
    },
    {
      "epoch": 18.168471720818292,
      "grad_norm": 66.4688491821289,
      "learning_rate": 0.00018501042587212773,
      "loss": 7.5096,
      "step": 150980
    },
    {
      "epoch": 18.169675090252706,
      "grad_norm": 102.49665832519531,
      "learning_rate": 0.0001850084224272568,
      "loss": 7.4028,
      "step": 150990
    },
    {
      "epoch": 18.170878459687124,
      "grad_norm": 169.3874053955078,
      "learning_rate": 0.0001850064188593577,
      "loss": 7.3443,
      "step": 151000
    },
    {
      "epoch": 18.17208182912154,
      "grad_norm": 119.93611145019531,
      "learning_rate": 0.0001850044151684333,
      "loss": 7.3586,
      "step": 151010
    },
    {
      "epoch": 18.173285198555956,
      "grad_norm": 111.78881072998047,
      "learning_rate": 0.00018500241135448655,
      "loss": 7.3727,
      "step": 151020
    },
    {
      "epoch": 18.174488567990373,
      "grad_norm": 209.1993408203125,
      "learning_rate": 0.0001850004074175203,
      "loss": 7.3423,
      "step": 151030
    },
    {
      "epoch": 18.17569193742479,
      "grad_norm": 129.2776641845703,
      "learning_rate": 0.0001849984033575375,
      "loss": 7.3181,
      "step": 151040
    },
    {
      "epoch": 18.176895306859205,
      "grad_norm": 51.54475784301758,
      "learning_rate": 0.000184996399174541,
      "loss": 7.5765,
      "step": 151050
    },
    {
      "epoch": 18.178098676293622,
      "grad_norm": 85.18266296386719,
      "learning_rate": 0.00018499439486853374,
      "loss": 7.4474,
      "step": 151060
    },
    {
      "epoch": 18.17930204572804,
      "grad_norm": 49.51419448852539,
      "learning_rate": 0.00018499239043951861,
      "loss": 7.5092,
      "step": 151070
    },
    {
      "epoch": 18.180505415162454,
      "grad_norm": 89.0058364868164,
      "learning_rate": 0.0001849903858874985,
      "loss": 7.5768,
      "step": 151080
    },
    {
      "epoch": 18.18170878459687,
      "grad_norm": 558.0090942382812,
      "learning_rate": 0.00018498838121247632,
      "loss": 7.5288,
      "step": 151090
    },
    {
      "epoch": 18.18291215403129,
      "grad_norm": 78.90757751464844,
      "learning_rate": 0.00018498637641445497,
      "loss": 7.4631,
      "step": 151100
    },
    {
      "epoch": 18.184115523465703,
      "grad_norm": 510.8673095703125,
      "learning_rate": 0.00018498437149343735,
      "loss": 7.5264,
      "step": 151110
    },
    {
      "epoch": 18.18531889290012,
      "grad_norm": 73.83006286621094,
      "learning_rate": 0.00018498236644942634,
      "loss": 7.4912,
      "step": 151120
    },
    {
      "epoch": 18.186522262334538,
      "grad_norm": 104.17164611816406,
      "learning_rate": 0.00018498036128242494,
      "loss": 7.5573,
      "step": 151130
    },
    {
      "epoch": 18.187725631768952,
      "grad_norm": 66.8059310913086,
      "learning_rate": 0.0001849783559924359,
      "loss": 7.4829,
      "step": 151140
    },
    {
      "epoch": 18.18892900120337,
      "grad_norm": 45.53932571411133,
      "learning_rate": 0.00018497635057946222,
      "loss": 7.5827,
      "step": 151150
    },
    {
      "epoch": 18.190132370637787,
      "grad_norm": 52.43602752685547,
      "learning_rate": 0.00018497434504350678,
      "loss": 7.5176,
      "step": 151160
    },
    {
      "epoch": 18.1913357400722,
      "grad_norm": 1033.6990966796875,
      "learning_rate": 0.0001849723393845725,
      "loss": 7.492,
      "step": 151170
    },
    {
      "epoch": 18.19253910950662,
      "grad_norm": 188.4258270263672,
      "learning_rate": 0.00018497033360266224,
      "loss": 7.5684,
      "step": 151180
    },
    {
      "epoch": 18.193742478941036,
      "grad_norm": 136.79278564453125,
      "learning_rate": 0.00018496832769777894,
      "loss": 7.5707,
      "step": 151190
    },
    {
      "epoch": 18.19494584837545,
      "grad_norm": 56.89272689819336,
      "learning_rate": 0.00018496632166992548,
      "loss": 7.5106,
      "step": 151200
    },
    {
      "epoch": 18.196149217809868,
      "grad_norm": 208.19151306152344,
      "learning_rate": 0.0001849643155191048,
      "loss": 7.5228,
      "step": 151210
    },
    {
      "epoch": 18.197352587244286,
      "grad_norm": 1972.562255859375,
      "learning_rate": 0.00018496230924531978,
      "loss": 7.537,
      "step": 151220
    },
    {
      "epoch": 18.1985559566787,
      "grad_norm": 236.2059783935547,
      "learning_rate": 0.0001849603028485733,
      "loss": 7.5807,
      "step": 151230
    },
    {
      "epoch": 18.199759326113117,
      "grad_norm": 155.88121032714844,
      "learning_rate": 0.0001849582963288683,
      "loss": 7.4712,
      "step": 151240
    },
    {
      "epoch": 18.200962695547535,
      "grad_norm": 190.3043975830078,
      "learning_rate": 0.00018495628968620768,
      "loss": 7.4649,
      "step": 151250
    },
    {
      "epoch": 18.20216606498195,
      "grad_norm": 122.1622543334961,
      "learning_rate": 0.0001849542829205943,
      "loss": 7.5207,
      "step": 151260
    },
    {
      "epoch": 18.203369434416366,
      "grad_norm": 83.27071380615234,
      "learning_rate": 0.0001849522760320311,
      "loss": 7.5285,
      "step": 151270
    },
    {
      "epoch": 18.204572803850784,
      "grad_norm": 162.8627166748047,
      "learning_rate": 0.00018495026902052103,
      "loss": 7.4729,
      "step": 151280
    },
    {
      "epoch": 18.205776173285198,
      "grad_norm": 149.04122924804688,
      "learning_rate": 0.00018494826188606691,
      "loss": 7.4415,
      "step": 151290
    },
    {
      "epoch": 18.206979542719615,
      "grad_norm": 49.33675765991211,
      "learning_rate": 0.00018494625462867168,
      "loss": 7.4724,
      "step": 151300
    },
    {
      "epoch": 18.20818291215403,
      "grad_norm": 220.3988800048828,
      "learning_rate": 0.00018494424724833824,
      "loss": 7.412,
      "step": 151310
    },
    {
      "epoch": 18.209386281588447,
      "grad_norm": 114.99543762207031,
      "learning_rate": 0.00018494223974506953,
      "loss": 7.3662,
      "step": 151320
    },
    {
      "epoch": 18.210589651022865,
      "grad_norm": 143.4963836669922,
      "learning_rate": 0.00018494023211886843,
      "loss": 7.499,
      "step": 151330
    },
    {
      "epoch": 18.21179302045728,
      "grad_norm": 90.40047454833984,
      "learning_rate": 0.0001849382243697378,
      "loss": 7.5558,
      "step": 151340
    },
    {
      "epoch": 18.212996389891696,
      "grad_norm": 107.75560760498047,
      "learning_rate": 0.00018493621649768062,
      "loss": 7.4405,
      "step": 151350
    },
    {
      "epoch": 18.214199759326114,
      "grad_norm": 184.58242797851562,
      "learning_rate": 0.00018493420850269977,
      "loss": 7.4225,
      "step": 151360
    },
    {
      "epoch": 18.215403128760528,
      "grad_norm": 112.13935852050781,
      "learning_rate": 0.00018493220038479812,
      "loss": 7.348,
      "step": 151370
    },
    {
      "epoch": 18.216606498194945,
      "grad_norm": 130.02120971679688,
      "learning_rate": 0.00018493019214397863,
      "loss": 7.5202,
      "step": 151380
    },
    {
      "epoch": 18.217809867629363,
      "grad_norm": 308.7380676269531,
      "learning_rate": 0.0001849281837802442,
      "loss": 7.3169,
      "step": 151390
    },
    {
      "epoch": 18.219013237063777,
      "grad_norm": 235.38868713378906,
      "learning_rate": 0.0001849261752935977,
      "loss": 7.3293,
      "step": 151400
    },
    {
      "epoch": 18.220216606498195,
      "grad_norm": 723.6695556640625,
      "learning_rate": 0.00018492416668404205,
      "loss": 7.3796,
      "step": 151410
    },
    {
      "epoch": 18.221419975932612,
      "grad_norm": 285.28717041015625,
      "learning_rate": 0.00018492215795158016,
      "loss": 7.3958,
      "step": 151420
    },
    {
      "epoch": 18.222623345367026,
      "grad_norm": 628.6517944335938,
      "learning_rate": 0.00018492014909621496,
      "loss": 7.4207,
      "step": 151430
    },
    {
      "epoch": 18.223826714801444,
      "grad_norm": 779.4058227539062,
      "learning_rate": 0.00018491814011794934,
      "loss": 7.3709,
      "step": 151440
    },
    {
      "epoch": 18.22503008423586,
      "grad_norm": 401.85968017578125,
      "learning_rate": 0.0001849161310167862,
      "loss": 7.3898,
      "step": 151450
    },
    {
      "epoch": 18.226233453670275,
      "grad_norm": 797.9706420898438,
      "learning_rate": 0.00018491412179272843,
      "loss": 7.3811,
      "step": 151460
    },
    {
      "epoch": 18.227436823104693,
      "grad_norm": 615.8762817382812,
      "learning_rate": 0.000184912112445779,
      "loss": 7.455,
      "step": 151470
    },
    {
      "epoch": 18.22864019253911,
      "grad_norm": 1646.37353515625,
      "learning_rate": 0.00018491010297594075,
      "loss": 7.389,
      "step": 151480
    },
    {
      "epoch": 18.229843561973524,
      "grad_norm": 1196.743408203125,
      "learning_rate": 0.00018490809338321662,
      "loss": 7.4918,
      "step": 151490
    },
    {
      "epoch": 18.231046931407942,
      "grad_norm": 1348.6378173828125,
      "learning_rate": 0.00018490608366760952,
      "loss": 7.4224,
      "step": 151500
    },
    {
      "epoch": 18.23225030084236,
      "grad_norm": 1069.326904296875,
      "learning_rate": 0.00018490407382912237,
      "loss": 7.4181,
      "step": 151510
    },
    {
      "epoch": 18.233453670276774,
      "grad_norm": 920.8336181640625,
      "learning_rate": 0.00018490206386775806,
      "loss": 7.4824,
      "step": 151520
    },
    {
      "epoch": 18.23465703971119,
      "grad_norm": 1177.33984375,
      "learning_rate": 0.0001849000537835195,
      "loss": 7.4369,
      "step": 151530
    },
    {
      "epoch": 18.23586040914561,
      "grad_norm": 1107.93359375,
      "learning_rate": 0.0001848980435764096,
      "loss": 7.3992,
      "step": 151540
    },
    {
      "epoch": 18.237063778580023,
      "grad_norm": 671.354736328125,
      "learning_rate": 0.00018489603324643126,
      "loss": 7.4526,
      "step": 151550
    },
    {
      "epoch": 18.23826714801444,
      "grad_norm": 713.6513061523438,
      "learning_rate": 0.0001848940227935874,
      "loss": 7.4403,
      "step": 151560
    },
    {
      "epoch": 18.239470517448858,
      "grad_norm": 826.3712768554688,
      "learning_rate": 0.0001848920122178809,
      "loss": 7.3983,
      "step": 151570
    },
    {
      "epoch": 18.240673886883272,
      "grad_norm": 1284.66650390625,
      "learning_rate": 0.00018489000151931476,
      "loss": 7.4668,
      "step": 151580
    },
    {
      "epoch": 18.24187725631769,
      "grad_norm": 699.0711059570312,
      "learning_rate": 0.0001848879906978918,
      "loss": 7.4623,
      "step": 151590
    },
    {
      "epoch": 18.243080625752107,
      "grad_norm": 404.0067443847656,
      "learning_rate": 0.00018488597975361496,
      "loss": 7.4273,
      "step": 151600
    },
    {
      "epoch": 18.24428399518652,
      "grad_norm": 987.5184326171875,
      "learning_rate": 0.00018488396868648717,
      "loss": 7.4349,
      "step": 151610
    },
    {
      "epoch": 18.24548736462094,
      "grad_norm": 851.533447265625,
      "learning_rate": 0.00018488195749651127,
      "loss": 7.329,
      "step": 151620
    },
    {
      "epoch": 18.246690734055356,
      "grad_norm": 971.6378784179688,
      "learning_rate": 0.00018487994618369028,
      "loss": 7.3766,
      "step": 151630
    },
    {
      "epoch": 18.24789410348977,
      "grad_norm": 675.4795532226562,
      "learning_rate": 0.000184877934748027,
      "loss": 7.3764,
      "step": 151640
    },
    {
      "epoch": 18.249097472924188,
      "grad_norm": 861.827392578125,
      "learning_rate": 0.00018487592318952443,
      "loss": 7.3667,
      "step": 151650
    },
    {
      "epoch": 18.250300842358605,
      "grad_norm": 1099.5947265625,
      "learning_rate": 0.0001848739115081854,
      "loss": 7.3711,
      "step": 151660
    },
    {
      "epoch": 18.25150421179302,
      "grad_norm": 358.1723937988281,
      "learning_rate": 0.0001848718997040129,
      "loss": 7.4148,
      "step": 151670
    },
    {
      "epoch": 18.252707581227437,
      "grad_norm": 2679.742431640625,
      "learning_rate": 0.00018486988777700983,
      "loss": 7.3917,
      "step": 151680
    },
    {
      "epoch": 18.253910950661854,
      "grad_norm": 1006.3916015625,
      "learning_rate": 0.00018486787572717902,
      "loss": 7.444,
      "step": 151690
    },
    {
      "epoch": 18.25511432009627,
      "grad_norm": 767.1204223632812,
      "learning_rate": 0.00018486586355452348,
      "loss": 7.3879,
      "step": 151700
    },
    {
      "epoch": 18.256317689530686,
      "grad_norm": 916.1561889648438,
      "learning_rate": 0.0001848638512590461,
      "loss": 7.3784,
      "step": 151710
    },
    {
      "epoch": 18.257521058965104,
      "grad_norm": 365.0423889160156,
      "learning_rate": 0.00018486183884074971,
      "loss": 7.3472,
      "step": 151720
    },
    {
      "epoch": 18.258724428399518,
      "grad_norm": 466.4209289550781,
      "learning_rate": 0.00018485982629963736,
      "loss": 7.3599,
      "step": 151730
    },
    {
      "epoch": 18.259927797833935,
      "grad_norm": 301.1725769042969,
      "learning_rate": 0.00018485781363571183,
      "loss": 7.407,
      "step": 151740
    },
    {
      "epoch": 18.261131167268353,
      "grad_norm": 360.5256652832031,
      "learning_rate": 0.00018485580084897613,
      "loss": 7.321,
      "step": 151750
    },
    {
      "epoch": 18.262334536702767,
      "grad_norm": 867.6375732421875,
      "learning_rate": 0.00018485378793943311,
      "loss": 7.3021,
      "step": 151760
    },
    {
      "epoch": 18.263537906137184,
      "grad_norm": 422.22113037109375,
      "learning_rate": 0.00018485177490708575,
      "loss": 7.3276,
      "step": 151770
    },
    {
      "epoch": 18.264741275571602,
      "grad_norm": 544.160400390625,
      "learning_rate": 0.0001848497617519369,
      "loss": 7.3177,
      "step": 151780
    },
    {
      "epoch": 18.265944645006016,
      "grad_norm": 496.56494140625,
      "learning_rate": 0.0001848477484739895,
      "loss": 7.3398,
      "step": 151790
    },
    {
      "epoch": 18.267148014440433,
      "grad_norm": 417.65289306640625,
      "learning_rate": 0.00018484573507324644,
      "loss": 7.3748,
      "step": 151800
    },
    {
      "epoch": 18.26835138387485,
      "grad_norm": 883.1669921875,
      "learning_rate": 0.00018484372154971065,
      "loss": 7.3655,
      "step": 151810
    },
    {
      "epoch": 18.269554753309265,
      "grad_norm": 639.7716674804688,
      "learning_rate": 0.00018484170790338507,
      "loss": 7.4151,
      "step": 151820
    },
    {
      "epoch": 18.270758122743683,
      "grad_norm": 910.5934448242188,
      "learning_rate": 0.00018483969413427259,
      "loss": 7.3394,
      "step": 151830
    },
    {
      "epoch": 18.2719614921781,
      "grad_norm": 489.7258605957031,
      "learning_rate": 0.00018483768024237613,
      "loss": 7.3665,
      "step": 151840
    },
    {
      "epoch": 18.273164861612514,
      "grad_norm": 431.5764465332031,
      "learning_rate": 0.00018483566622769858,
      "loss": 7.344,
      "step": 151850
    },
    {
      "epoch": 18.27436823104693,
      "grad_norm": 542.9868774414062,
      "learning_rate": 0.0001848336520902429,
      "loss": 7.3436,
      "step": 151860
    },
    {
      "epoch": 18.27557160048135,
      "grad_norm": 371.8831481933594,
      "learning_rate": 0.00018483163783001197,
      "loss": 7.3787,
      "step": 151870
    },
    {
      "epoch": 18.276774969915763,
      "grad_norm": 570.6926879882812,
      "learning_rate": 0.00018482962344700871,
      "loss": 7.4664,
      "step": 151880
    },
    {
      "epoch": 18.27797833935018,
      "grad_norm": 425.23931884765625,
      "learning_rate": 0.00018482760894123604,
      "loss": 7.2684,
      "step": 151890
    },
    {
      "epoch": 18.2791817087846,
      "grad_norm": 390.46575927734375,
      "learning_rate": 0.0001848255943126969,
      "loss": 7.349,
      "step": 151900
    },
    {
      "epoch": 18.280385078219012,
      "grad_norm": 474.7912902832031,
      "learning_rate": 0.00018482357956139416,
      "loss": 7.4784,
      "step": 151910
    },
    {
      "epoch": 18.28158844765343,
      "grad_norm": 497.7284240722656,
      "learning_rate": 0.00018482156468733077,
      "loss": 7.42,
      "step": 151920
    },
    {
      "epoch": 18.282791817087848,
      "grad_norm": 377.16796875,
      "learning_rate": 0.00018481954969050963,
      "loss": 7.3588,
      "step": 151930
    },
    {
      "epoch": 18.28399518652226,
      "grad_norm": 509.6619873046875,
      "learning_rate": 0.00018481753457093365,
      "loss": 7.4345,
      "step": 151940
    },
    {
      "epoch": 18.28519855595668,
      "grad_norm": 505.9047546386719,
      "learning_rate": 0.00018481551932860577,
      "loss": 7.3872,
      "step": 151950
    },
    {
      "epoch": 18.286401925391097,
      "grad_norm": 504.4636535644531,
      "learning_rate": 0.0001848135039635289,
      "loss": 7.3197,
      "step": 151960
    },
    {
      "epoch": 18.28760529482551,
      "grad_norm": 331.6162109375,
      "learning_rate": 0.0001848114884757059,
      "loss": 7.4205,
      "step": 151970
    },
    {
      "epoch": 18.28880866425993,
      "grad_norm": 499.6607971191406,
      "learning_rate": 0.0001848094728651398,
      "loss": 7.4117,
      "step": 151980
    },
    {
      "epoch": 18.290012033694346,
      "grad_norm": 602.674072265625,
      "learning_rate": 0.00018480745713183344,
      "loss": 7.3608,
      "step": 151990
    },
    {
      "epoch": 18.29121540312876,
      "grad_norm": 75.54895782470703,
      "learning_rate": 0.00018480544127578972,
      "loss": 7.4137,
      "step": 152000
    },
    {
      "epoch": 18.292418772563177,
      "grad_norm": 104.66126251220703,
      "learning_rate": 0.00018480342529701162,
      "loss": 7.3288,
      "step": 152010
    },
    {
      "epoch": 18.29362214199759,
      "grad_norm": 343.62664794921875,
      "learning_rate": 0.000184801409195502,
      "loss": 7.3711,
      "step": 152020
    },
    {
      "epoch": 18.29482551143201,
      "grad_norm": 523.1060791015625,
      "learning_rate": 0.00018479939297126382,
      "loss": 7.4172,
      "step": 152030
    },
    {
      "epoch": 18.296028880866427,
      "grad_norm": 326.9176025390625,
      "learning_rate": 0.0001847973766243,
      "loss": 7.4349,
      "step": 152040
    },
    {
      "epoch": 18.29723225030084,
      "grad_norm": 264.3764343261719,
      "learning_rate": 0.0001847953601546134,
      "loss": 7.3919,
      "step": 152050
    },
    {
      "epoch": 18.29843561973526,
      "grad_norm": 459.1297607421875,
      "learning_rate": 0.00018479334356220702,
      "loss": 7.3893,
      "step": 152060
    },
    {
      "epoch": 18.299638989169676,
      "grad_norm": 290.79315185546875,
      "learning_rate": 0.0001847913268470837,
      "loss": 7.415,
      "step": 152070
    },
    {
      "epoch": 18.30084235860409,
      "grad_norm": 597.1858520507812,
      "learning_rate": 0.00018478931000924642,
      "loss": 7.3745,
      "step": 152080
    },
    {
      "epoch": 18.302045728038507,
      "grad_norm": 850.1022338867188,
      "learning_rate": 0.00018478729304869805,
      "loss": 7.3653,
      "step": 152090
    },
    {
      "epoch": 18.303249097472925,
      "grad_norm": 237.01907348632812,
      "learning_rate": 0.00018478527596544158,
      "loss": 7.3974,
      "step": 152100
    },
    {
      "epoch": 18.30445246690734,
      "grad_norm": 964.3274536132812,
      "learning_rate": 0.00018478325875947985,
      "loss": 7.3103,
      "step": 152110
    },
    {
      "epoch": 18.305655836341757,
      "grad_norm": 365.0823974609375,
      "learning_rate": 0.00018478124143081581,
      "loss": 7.3759,
      "step": 152120
    },
    {
      "epoch": 18.306859205776174,
      "grad_norm": 540.3898315429688,
      "learning_rate": 0.0001847792239794524,
      "loss": 7.3724,
      "step": 152130
    },
    {
      "epoch": 18.308062575210588,
      "grad_norm": 206.74462890625,
      "learning_rate": 0.0001847772064053925,
      "loss": 7.3672,
      "step": 152140
    },
    {
      "epoch": 18.309265944645006,
      "grad_norm": 756.8995361328125,
      "learning_rate": 0.00018477518870863905,
      "loss": 7.4172,
      "step": 152150
    },
    {
      "epoch": 18.310469314079423,
      "grad_norm": 619.6227416992188,
      "learning_rate": 0.00018477317088919502,
      "loss": 7.3557,
      "step": 152160
    },
    {
      "epoch": 18.311672683513837,
      "grad_norm": 206.07252502441406,
      "learning_rate": 0.00018477115294706321,
      "loss": 7.3845,
      "step": 152170
    },
    {
      "epoch": 18.312876052948255,
      "grad_norm": 206.29991149902344,
      "learning_rate": 0.00018476913488224667,
      "loss": 7.408,
      "step": 152180
    },
    {
      "epoch": 18.314079422382672,
      "grad_norm": 166.0495147705078,
      "learning_rate": 0.00018476711669474823,
      "loss": 7.3012,
      "step": 152190
    },
    {
      "epoch": 18.315282791817086,
      "grad_norm": 161.13961791992188,
      "learning_rate": 0.00018476509838457085,
      "loss": 7.3856,
      "step": 152200
    },
    {
      "epoch": 18.316486161251504,
      "grad_norm": 127.53560638427734,
      "learning_rate": 0.00018476307995171745,
      "loss": 7.2802,
      "step": 152210
    },
    {
      "epoch": 18.31768953068592,
      "grad_norm": 232.245361328125,
      "learning_rate": 0.00018476106139619095,
      "loss": 7.3252,
      "step": 152220
    },
    {
      "epoch": 18.318892900120336,
      "grad_norm": 189.64749145507812,
      "learning_rate": 0.00018475904271799423,
      "loss": 7.2922,
      "step": 152230
    },
    {
      "epoch": 18.320096269554753,
      "grad_norm": 135.0006103515625,
      "learning_rate": 0.00018475702391713031,
      "loss": 7.332,
      "step": 152240
    },
    {
      "epoch": 18.32129963898917,
      "grad_norm": 372.1553649902344,
      "learning_rate": 0.00018475500499360204,
      "loss": 7.35,
      "step": 152250
    },
    {
      "epoch": 18.322503008423585,
      "grad_norm": 514.6927490234375,
      "learning_rate": 0.00018475298594741233,
      "loss": 7.4239,
      "step": 152260
    },
    {
      "epoch": 18.323706377858002,
      "grad_norm": 163.652587890625,
      "learning_rate": 0.00018475096677856412,
      "loss": 7.4147,
      "step": 152270
    },
    {
      "epoch": 18.32490974729242,
      "grad_norm": 248.01438903808594,
      "learning_rate": 0.00018474894748706034,
      "loss": 7.3285,
      "step": 152280
    },
    {
      "epoch": 18.326113116726834,
      "grad_norm": 275.4211120605469,
      "learning_rate": 0.0001847469280729039,
      "loss": 7.3398,
      "step": 152290
    },
    {
      "epoch": 18.32731648616125,
      "grad_norm": 271.49530029296875,
      "learning_rate": 0.00018474490853609776,
      "loss": 7.4121,
      "step": 152300
    },
    {
      "epoch": 18.32851985559567,
      "grad_norm": 275.52764892578125,
      "learning_rate": 0.0001847428888766448,
      "loss": 7.3025,
      "step": 152310
    },
    {
      "epoch": 18.329723225030083,
      "grad_norm": 815.0440673828125,
      "learning_rate": 0.00018474086909454795,
      "loss": 7.3067,
      "step": 152320
    },
    {
      "epoch": 18.3309265944645,
      "grad_norm": 557.1624755859375,
      "learning_rate": 0.00018473884918981015,
      "loss": 7.4209,
      "step": 152330
    },
    {
      "epoch": 18.332129963898918,
      "grad_norm": 283.4029541015625,
      "learning_rate": 0.00018473682916243434,
      "loss": 7.3618,
      "step": 152340
    },
    {
      "epoch": 18.333333333333332,
      "grad_norm": 284.2005615234375,
      "learning_rate": 0.00018473480901242336,
      "loss": 7.4374,
      "step": 152350
    },
    {
      "epoch": 18.33453670276775,
      "grad_norm": 181.1649169921875,
      "learning_rate": 0.00018473278873978023,
      "loss": 7.4721,
      "step": 152360
    },
    {
      "epoch": 18.335740072202167,
      "grad_norm": 380.21990966796875,
      "learning_rate": 0.0001847307683445078,
      "loss": 7.3412,
      "step": 152370
    },
    {
      "epoch": 18.33694344163658,
      "grad_norm": 96.97758483886719,
      "learning_rate": 0.0001847287478266091,
      "loss": 7.3578,
      "step": 152380
    },
    {
      "epoch": 18.338146811071,
      "grad_norm": 197.025390625,
      "learning_rate": 0.00018472672718608692,
      "loss": 7.3366,
      "step": 152390
    },
    {
      "epoch": 18.339350180505416,
      "grad_norm": 153.88284301757812,
      "learning_rate": 0.00018472470642294427,
      "loss": 7.4359,
      "step": 152400
    },
    {
      "epoch": 18.34055354993983,
      "grad_norm": 74.3664321899414,
      "learning_rate": 0.00018472268553718405,
      "loss": 7.3939,
      "step": 152410
    },
    {
      "epoch": 18.341756919374248,
      "grad_norm": 118.60506439208984,
      "learning_rate": 0.00018472066452880918,
      "loss": 7.3732,
      "step": 152420
    },
    {
      "epoch": 18.342960288808666,
      "grad_norm": 351.7347412109375,
      "learning_rate": 0.0001847186433978226,
      "loss": 7.3547,
      "step": 152430
    },
    {
      "epoch": 18.34416365824308,
      "grad_norm": 98.69527435302734,
      "learning_rate": 0.00018471662214422722,
      "loss": 7.4385,
      "step": 152440
    },
    {
      "epoch": 18.345367027677497,
      "grad_norm": 116.80021667480469,
      "learning_rate": 0.00018471460076802595,
      "loss": 7.3886,
      "step": 152450
    },
    {
      "epoch": 18.346570397111915,
      "grad_norm": 311.6566162109375,
      "learning_rate": 0.0001847125792692218,
      "loss": 7.3342,
      "step": 152460
    },
    {
      "epoch": 18.34777376654633,
      "grad_norm": 245.01870727539062,
      "learning_rate": 0.00018471055764781758,
      "loss": 7.446,
      "step": 152470
    },
    {
      "epoch": 18.348977135980746,
      "grad_norm": 699.9287719726562,
      "learning_rate": 0.00018470853590381626,
      "loss": 7.3242,
      "step": 152480
    },
    {
      "epoch": 18.350180505415164,
      "grad_norm": 325.8674621582031,
      "learning_rate": 0.0001847065140372208,
      "loss": 7.262,
      "step": 152490
    },
    {
      "epoch": 18.351383874849578,
      "grad_norm": 309.1668701171875,
      "learning_rate": 0.0001847044920480341,
      "loss": 7.4257,
      "step": 152500
    },
    {
      "epoch": 18.352587244283995,
      "grad_norm": 352.16363525390625,
      "learning_rate": 0.0001847024699362591,
      "loss": 7.3365,
      "step": 152510
    },
    {
      "epoch": 18.353790613718413,
      "grad_norm": 669.510498046875,
      "learning_rate": 0.00018470044770189868,
      "loss": 7.2026,
      "step": 152520
    },
    {
      "epoch": 18.354993983152827,
      "grad_norm": 6747.38037109375,
      "learning_rate": 0.00018469842534495583,
      "loss": 7.2472,
      "step": 152530
    },
    {
      "epoch": 18.356197352587245,
      "grad_norm": 752.1255493164062,
      "learning_rate": 0.00018469640286543343,
      "loss": 7.3093,
      "step": 152540
    },
    {
      "epoch": 18.357400722021662,
      "grad_norm": 637.0225219726562,
      "learning_rate": 0.00018469438026333444,
      "loss": 7.3485,
      "step": 152550
    },
    {
      "epoch": 18.358604091456076,
      "grad_norm": 1597.137939453125,
      "learning_rate": 0.00018469235753866176,
      "loss": 7.3601,
      "step": 152560
    },
    {
      "epoch": 18.359807460890494,
      "grad_norm": 805.3880004882812,
      "learning_rate": 0.00018469033469141835,
      "loss": 7.4051,
      "step": 152570
    },
    {
      "epoch": 18.36101083032491,
      "grad_norm": 409.3124694824219,
      "learning_rate": 0.00018468831172160707,
      "loss": 7.2234,
      "step": 152580
    },
    {
      "epoch": 18.362214199759325,
      "grad_norm": 406.9848327636719,
      "learning_rate": 0.00018468628862923094,
      "loss": 7.3235,
      "step": 152590
    },
    {
      "epoch": 18.363417569193743,
      "grad_norm": 1156.8260498046875,
      "learning_rate": 0.00018468426541429282,
      "loss": 7.2933,
      "step": 152600
    },
    {
      "epoch": 18.36462093862816,
      "grad_norm": 831.1348266601562,
      "learning_rate": 0.00018468224207679565,
      "loss": 7.3392,
      "step": 152610
    },
    {
      "epoch": 18.365824308062574,
      "grad_norm": 961.0281372070312,
      "learning_rate": 0.0001846802186167424,
      "loss": 7.3274,
      "step": 152620
    },
    {
      "epoch": 18.367027677496992,
      "grad_norm": 15979.4677734375,
      "learning_rate": 0.00018467819503413595,
      "loss": 7.423,
      "step": 152630
    },
    {
      "epoch": 18.36823104693141,
      "grad_norm": 4876.44580078125,
      "learning_rate": 0.00018467617132897926,
      "loss": 7.399,
      "step": 152640
    },
    {
      "epoch": 18.369434416365824,
      "grad_norm": 1651.61083984375,
      "learning_rate": 0.0001846741475012752,
      "loss": 7.4468,
      "step": 152650
    },
    {
      "epoch": 18.37063778580024,
      "grad_norm": 1993.109375,
      "learning_rate": 0.0001846721235510268,
      "loss": 7.415,
      "step": 152660
    },
    {
      "epoch": 18.37184115523466,
      "grad_norm": 1994.6533203125,
      "learning_rate": 0.00018467009947823688,
      "loss": 7.4098,
      "step": 152670
    },
    {
      "epoch": 18.373044524669073,
      "grad_norm": 1205.272216796875,
      "learning_rate": 0.00018466807528290846,
      "loss": 7.3834,
      "step": 152680
    },
    {
      "epoch": 18.37424789410349,
      "grad_norm": 813.9103393554688,
      "learning_rate": 0.0001846660509650444,
      "loss": 7.3982,
      "step": 152690
    },
    {
      "epoch": 18.375451263537904,
      "grad_norm": 276.84539794921875,
      "learning_rate": 0.00018466402652464768,
      "loss": 7.2824,
      "step": 152700
    },
    {
      "epoch": 18.376654632972322,
      "grad_norm": 239.27444458007812,
      "learning_rate": 0.00018466200196172122,
      "loss": 7.3645,
      "step": 152710
    },
    {
      "epoch": 18.37785800240674,
      "grad_norm": 449.86865234375,
      "learning_rate": 0.0001846599772762679,
      "loss": 7.2669,
      "step": 152720
    },
    {
      "epoch": 18.379061371841154,
      "grad_norm": 473.189453125,
      "learning_rate": 0.00018465795246829075,
      "loss": 7.3682,
      "step": 152730
    },
    {
      "epoch": 18.38026474127557,
      "grad_norm": 981.4517211914062,
      "learning_rate": 0.0001846559275377926,
      "loss": 7.4094,
      "step": 152740
    },
    {
      "epoch": 18.38146811070999,
      "grad_norm": 671.7831420898438,
      "learning_rate": 0.00018465390248477643,
      "loss": 7.275,
      "step": 152750
    },
    {
      "epoch": 18.382671480144403,
      "grad_norm": 440.99591064453125,
      "learning_rate": 0.00018465187730924517,
      "loss": 7.3697,
      "step": 152760
    },
    {
      "epoch": 18.38387484957882,
      "grad_norm": 352.1147155761719,
      "learning_rate": 0.00018464985201120174,
      "loss": 7.2749,
      "step": 152770
    },
    {
      "epoch": 18.385078219013238,
      "grad_norm": 406.75775146484375,
      "learning_rate": 0.0001846478265906491,
      "loss": 7.4258,
      "step": 152780
    },
    {
      "epoch": 18.386281588447652,
      "grad_norm": 633.501220703125,
      "learning_rate": 0.0001846458010475901,
      "loss": 7.4049,
      "step": 152790
    },
    {
      "epoch": 18.38748495788207,
      "grad_norm": 458.1518249511719,
      "learning_rate": 0.00018464377538202778,
      "loss": 7.415,
      "step": 152800
    },
    {
      "epoch": 18.388688327316487,
      "grad_norm": 672.3319091796875,
      "learning_rate": 0.00018464174959396498,
      "loss": 7.4147,
      "step": 152810
    },
    {
      "epoch": 18.3898916967509,
      "grad_norm": 273.4700927734375,
      "learning_rate": 0.00018463972368340468,
      "loss": 7.273,
      "step": 152820
    },
    {
      "epoch": 18.39109506618532,
      "grad_norm": 239.07583618164062,
      "learning_rate": 0.00018463769765034982,
      "loss": 7.3782,
      "step": 152830
    },
    {
      "epoch": 18.392298435619736,
      "grad_norm": 638.8527221679688,
      "learning_rate": 0.0001846356714948033,
      "loss": 7.2381,
      "step": 152840
    },
    {
      "epoch": 18.39350180505415,
      "grad_norm": 776.33447265625,
      "learning_rate": 0.00018463364521676808,
      "loss": 7.4281,
      "step": 152850
    },
    {
      "epoch": 18.394705174488568,
      "grad_norm": 343.4711608886719,
      "learning_rate": 0.0001846316188162471,
      "loss": 7.3826,
      "step": 152860
    },
    {
      "epoch": 18.395908543922985,
      "grad_norm": 725.744873046875,
      "learning_rate": 0.00018462959229324324,
      "loss": 7.4091,
      "step": 152870
    },
    {
      "epoch": 18.3971119133574,
      "grad_norm": 1224.29443359375,
      "learning_rate": 0.00018462756564775946,
      "loss": 7.4163,
      "step": 152880
    },
    {
      "epoch": 18.398315282791817,
      "grad_norm": 503.4244384765625,
      "learning_rate": 0.0001846255388797987,
      "loss": 7.2828,
      "step": 152890
    },
    {
      "epoch": 18.399518652226234,
      "grad_norm": 343.1127014160156,
      "learning_rate": 0.0001846235119893639,
      "loss": 7.2975,
      "step": 152900
    },
    {
      "epoch": 18.40072202166065,
      "grad_norm": 399.8078308105469,
      "learning_rate": 0.000184621484976458,
      "loss": 7.336,
      "step": 152910
    },
    {
      "epoch": 18.401925391095066,
      "grad_norm": 283.1817321777344,
      "learning_rate": 0.0001846194578410839,
      "loss": 7.2593,
      "step": 152920
    },
    {
      "epoch": 18.403128760529484,
      "grad_norm": 149.43238830566406,
      "learning_rate": 0.00018461743058324455,
      "loss": 7.2923,
      "step": 152930
    },
    {
      "epoch": 18.404332129963898,
      "grad_norm": 882.1889038085938,
      "learning_rate": 0.0001846154032029429,
      "loss": 7.3169,
      "step": 152940
    },
    {
      "epoch": 18.405535499398315,
      "grad_norm": 232.15188598632812,
      "learning_rate": 0.00018461337570018188,
      "loss": 7.4182,
      "step": 152950
    },
    {
      "epoch": 18.406738868832733,
      "grad_norm": 810.0722045898438,
      "learning_rate": 0.0001846113480749644,
      "loss": 7.43,
      "step": 152960
    },
    {
      "epoch": 18.407942238267147,
      "grad_norm": 460.2532043457031,
      "learning_rate": 0.0001846093203272934,
      "loss": 7.4534,
      "step": 152970
    },
    {
      "epoch": 18.409145607701564,
      "grad_norm": 652.58154296875,
      "learning_rate": 0.00018460729245717183,
      "loss": 7.3855,
      "step": 152980
    },
    {
      "epoch": 18.410348977135982,
      "grad_norm": 340.3279113769531,
      "learning_rate": 0.00018460526446460263,
      "loss": 7.4736,
      "step": 152990
    },
    {
      "epoch": 18.411552346570396,
      "grad_norm": 878.8419189453125,
      "learning_rate": 0.00018460323634958873,
      "loss": 7.4742,
      "step": 153000
    },
    {
      "epoch": 18.412755716004813,
      "grad_norm": 810.9647216796875,
      "learning_rate": 0.00018460120811213305,
      "loss": 7.3896,
      "step": 153010
    },
    {
      "epoch": 18.41395908543923,
      "grad_norm": 1688.00244140625,
      "learning_rate": 0.00018459917975223854,
      "loss": 7.3993,
      "step": 153020
    },
    {
      "epoch": 18.415162454873645,
      "grad_norm": 430.4827575683594,
      "learning_rate": 0.0001845971512699081,
      "loss": 7.4924,
      "step": 153030
    },
    {
      "epoch": 18.416365824308063,
      "grad_norm": 85.9831771850586,
      "learning_rate": 0.00018459512266514474,
      "loss": 7.5139,
      "step": 153040
    },
    {
      "epoch": 18.41756919374248,
      "grad_norm": 51.568939208984375,
      "learning_rate": 0.00018459309393795132,
      "loss": 7.481,
      "step": 153050
    },
    {
      "epoch": 18.418772563176894,
      "grad_norm": 54.81188201904297,
      "learning_rate": 0.0001845910650883308,
      "loss": 7.5253,
      "step": 153060
    },
    {
      "epoch": 18.41997593261131,
      "grad_norm": 74.08306884765625,
      "learning_rate": 0.00018458903611628615,
      "loss": 7.5278,
      "step": 153070
    },
    {
      "epoch": 18.42117930204573,
      "grad_norm": 147.89413452148438,
      "learning_rate": 0.00018458700702182027,
      "loss": 7.5596,
      "step": 153080
    },
    {
      "epoch": 18.422382671480143,
      "grad_norm": 692.1793823242188,
      "learning_rate": 0.00018458497780493612,
      "loss": 7.5143,
      "step": 153090
    },
    {
      "epoch": 18.42358604091456,
      "grad_norm": 40.994956970214844,
      "learning_rate": 0.0001845829484656366,
      "loss": 7.5969,
      "step": 153100
    },
    {
      "epoch": 18.42478941034898,
      "grad_norm": 9.625137329101562,
      "learning_rate": 0.00018458091900392466,
      "loss": 7.7019,
      "step": 153110
    },
    {
      "epoch": 18.425992779783392,
      "grad_norm": 12.490511894226074,
      "learning_rate": 0.0001845788894198033,
      "loss": 7.6813,
      "step": 153120
    },
    {
      "epoch": 18.42719614921781,
      "grad_norm": 8.232507705688477,
      "learning_rate": 0.00018457685971327536,
      "loss": 7.5214,
      "step": 153130
    },
    {
      "epoch": 18.428399518652228,
      "grad_norm": 34.846214294433594,
      "learning_rate": 0.00018457482988434385,
      "loss": 7.4904,
      "step": 153140
    },
    {
      "epoch": 18.42960288808664,
      "grad_norm": 413.77508544921875,
      "learning_rate": 0.00018457279993301166,
      "loss": 7.5764,
      "step": 153150
    },
    {
      "epoch": 18.43080625752106,
      "grad_norm": 290.9382629394531,
      "learning_rate": 0.00018457076985928177,
      "loss": 7.7764,
      "step": 153160
    },
    {
      "epoch": 18.432009626955477,
      "grad_norm": 1067.5765380859375,
      "learning_rate": 0.00018456873966315708,
      "loss": 7.7062,
      "step": 153170
    },
    {
      "epoch": 18.43321299638989,
      "grad_norm": 1371.8919677734375,
      "learning_rate": 0.00018456670934464056,
      "loss": 7.6137,
      "step": 153180
    },
    {
      "epoch": 18.43441636582431,
      "grad_norm": 526.41357421875,
      "learning_rate": 0.0001845646789037351,
      "loss": 7.5463,
      "step": 153190
    },
    {
      "epoch": 18.435619735258726,
      "grad_norm": 35714.6171875,
      "learning_rate": 0.0001845626483404437,
      "loss": 7.6021,
      "step": 153200
    },
    {
      "epoch": 18.43682310469314,
      "grad_norm": 107209.7421875,
      "learning_rate": 0.00018456061765476927,
      "loss": 7.9262,
      "step": 153210
    },
    {
      "epoch": 18.438026474127557,
      "grad_norm": 290236160.0,
      "learning_rate": 0.00018455858684671475,
      "loss": 8.4073,
      "step": 153220
    },
    {
      "epoch": 18.439229843561975,
      "grad_norm": 49486820.0,
      "learning_rate": 0.00018455655591628306,
      "loss": 8.7149,
      "step": 153230
    },
    {
      "epoch": 18.44043321299639,
      "grad_norm": 354431008.0,
      "learning_rate": 0.00018455452486347714,
      "loss": 9.1406,
      "step": 153240
    },
    {
      "epoch": 18.441636582430807,
      "grad_norm": 952564928.0,
      "learning_rate": 0.00018455249368829998,
      "loss": 9.4997,
      "step": 153250
    },
    {
      "epoch": 18.442839951865224,
      "grad_norm": 343560960.0,
      "learning_rate": 0.0001845504623907545,
      "loss": 9.6089,
      "step": 153260
    },
    {
      "epoch": 18.444043321299638,
      "grad_norm": 907001152.0,
      "learning_rate": 0.0001845484309708436,
      "loss": 9.5405,
      "step": 153270
    },
    {
      "epoch": 18.445246690734056,
      "grad_norm": 1099533184.0,
      "learning_rate": 0.00018454639942857025,
      "loss": 9.7289,
      "step": 153280
    },
    {
      "epoch": 18.446450060168473,
      "grad_norm": 401878048.0,
      "learning_rate": 0.00018454436776393739,
      "loss": 9.5961,
      "step": 153290
    },
    {
      "epoch": 18.447653429602887,
      "grad_norm": 393140160.0,
      "learning_rate": 0.00018454233597694795,
      "loss": 9.6738,
      "step": 153300
    },
    {
      "epoch": 18.448856799037305,
      "grad_norm": 110440016.0,
      "learning_rate": 0.00018454030406760488,
      "loss": 9.7425,
      "step": 153310
    },
    {
      "epoch": 18.450060168471722,
      "grad_norm": 227057136.0,
      "learning_rate": 0.0001845382720359111,
      "loss": 9.9498,
      "step": 153320
    },
    {
      "epoch": 18.451263537906136,
      "grad_norm": 189533648.0,
      "learning_rate": 0.0001845362398818696,
      "loss": 10.2202,
      "step": 153330
    },
    {
      "epoch": 18.452466907340554,
      "grad_norm": 144564608.0,
      "learning_rate": 0.00018453420760548327,
      "loss": 10.5441,
      "step": 153340
    },
    {
      "epoch": 18.45367027677497,
      "grad_norm": 158687680.0,
      "learning_rate": 0.00018453217520675508,
      "loss": 10.4903,
      "step": 153350
    },
    {
      "epoch": 18.454873646209386,
      "grad_norm": 84126192.0,
      "learning_rate": 0.00018453014268568797,
      "loss": 10.4288,
      "step": 153360
    },
    {
      "epoch": 18.456077015643803,
      "grad_norm": 347420544.0,
      "learning_rate": 0.00018452811004228485,
      "loss": 10.4237,
      "step": 153370
    },
    {
      "epoch": 18.45728038507822,
      "grad_norm": 705729536.0,
      "learning_rate": 0.0001845260772765487,
      "loss": 10.4202,
      "step": 153380
    },
    {
      "epoch": 18.458483754512635,
      "grad_norm": 447884192.0,
      "learning_rate": 0.00018452404438848246,
      "loss": 10.3371,
      "step": 153390
    },
    {
      "epoch": 18.459687123947052,
      "grad_norm": 319394624.0,
      "learning_rate": 0.00018452201137808904,
      "loss": 10.5163,
      "step": 153400
    },
    {
      "epoch": 18.460890493381466,
      "grad_norm": 307767616.0,
      "learning_rate": 0.00018451997824537139,
      "loss": 10.3893,
      "step": 153410
    },
    {
      "epoch": 18.462093862815884,
      "grad_norm": 18004549632.0,
      "learning_rate": 0.00018451794499033247,
      "loss": 10.6187,
      "step": 153420
    },
    {
      "epoch": 18.4632972322503,
      "grad_norm": 2387283456.0,
      "learning_rate": 0.00018451591161297524,
      "loss": 10.5182,
      "step": 153430
    },
    {
      "epoch": 18.464500601684716,
      "grad_norm": 256399568.0,
      "learning_rate": 0.00018451387811330264,
      "loss": 10.6427,
      "step": 153440
    },
    {
      "epoch": 18.465703971119133,
      "grad_norm": 226842784.0,
      "learning_rate": 0.00018451184449131756,
      "loss": 10.4203,
      "step": 153450
    },
    {
      "epoch": 18.46690734055355,
      "grad_norm": 15965370.0,
      "learning_rate": 0.00018450981074702295,
      "loss": 10.6075,
      "step": 153460
    },
    {
      "epoch": 18.468110709987965,
      "grad_norm": 38252532.0,
      "learning_rate": 0.0001845077768804218,
      "loss": 10.7387,
      "step": 153470
    },
    {
      "epoch": 18.469314079422382,
      "grad_norm": 34838764.0,
      "learning_rate": 0.00018450574289151706,
      "loss": 10.9605,
      "step": 153480
    },
    {
      "epoch": 18.4705174488568,
      "grad_norm": 62735660.0,
      "learning_rate": 0.0001845037087803116,
      "loss": 10.692,
      "step": 153490
    },
    {
      "epoch": 18.471720818291214,
      "grad_norm": 30423376.0,
      "learning_rate": 0.00018450167454680844,
      "loss": 10.3002,
      "step": 153500
    },
    {
      "epoch": 18.47292418772563,
      "grad_norm": 1632257.75,
      "learning_rate": 0.0001844996401910105,
      "loss": 10.3527,
      "step": 153510
    },
    {
      "epoch": 18.47412755716005,
      "grad_norm": 277744.25,
      "learning_rate": 0.00018449760571292073,
      "loss": 10.359,
      "step": 153520
    },
    {
      "epoch": 18.475330926594463,
      "grad_norm": 1733779.25,
      "learning_rate": 0.000184495571112542,
      "loss": 10.1695,
      "step": 153530
    },
    {
      "epoch": 18.47653429602888,
      "grad_norm": 903284.5,
      "learning_rate": 0.00018449353638987738,
      "loss": 9.9837,
      "step": 153540
    },
    {
      "epoch": 18.477737665463298,
      "grad_norm": 1033326.5,
      "learning_rate": 0.00018449150154492973,
      "loss": 10.4487,
      "step": 153550
    },
    {
      "epoch": 18.478941034897712,
      "grad_norm": 686108.9375,
      "learning_rate": 0.00018448946657770199,
      "loss": 10.8624,
      "step": 153560
    },
    {
      "epoch": 18.48014440433213,
      "grad_norm": 272341.71875,
      "learning_rate": 0.00018448743148819717,
      "loss": 10.9326,
      "step": 153570
    },
    {
      "epoch": 18.481347773766547,
      "grad_norm": 7521502232576.0,
      "learning_rate": 0.00018448539627641817,
      "loss": 10.9476,
      "step": 153580
    },
    {
      "epoch": 18.48255114320096,
      "grad_norm": 5581615661056.0,
      "learning_rate": 0.00018448336094236792,
      "loss": 10.5312,
      "step": 153590
    },
    {
      "epoch": 18.48375451263538,
      "grad_norm": 35045565792256.0,
      "learning_rate": 0.00018448132548604942,
      "loss": 10.2854,
      "step": 153600
    },
    {
      "epoch": 18.484957882069796,
      "grad_norm": 4497267490816.0,
      "learning_rate": 0.00018447928990746558,
      "loss": 10.3303,
      "step": 153610
    },
    {
      "epoch": 18.48616125150421,
      "grad_norm": 19079746289664.0,
      "learning_rate": 0.00018447725420661933,
      "loss": 10.5231,
      "step": 153620
    },
    {
      "epoch": 18.487364620938628,
      "grad_norm": 16816830152704.0,
      "learning_rate": 0.00018447521838351364,
      "loss": 10.6547,
      "step": 153630
    },
    {
      "epoch": 18.488567990373046,
      "grad_norm": 34880473792512.0,
      "learning_rate": 0.00018447318243815147,
      "loss": 10.7887,
      "step": 153640
    },
    {
      "epoch": 18.48977135980746,
      "grad_norm": 25699289088.0,
      "learning_rate": 0.00018447114637053572,
      "loss": 10.8947,
      "step": 153650
    },
    {
      "epoch": 18.490974729241877,
      "grad_norm": 49615843328.0,
      "learning_rate": 0.0001844691101806694,
      "loss": 10.9466,
      "step": 153660
    },
    {
      "epoch": 18.492178098676295,
      "grad_norm": 63240937472.0,
      "learning_rate": 0.0001844670738685554,
      "loss": 11.0833,
      "step": 153670
    },
    {
      "epoch": 18.49338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.0001844650374341967,
      "loss": 11.1395,
      "step": 153680
    },
    {
      "epoch": 18.494584837545126,
      "grad_norm": Infinity,
      "learning_rate": 0.00018446300087759623,
      "loss": 11.1156,
      "step": 153690
    },
    {
      "epoch": 18.495788206979544,
      "grad_norm": Infinity,
      "learning_rate": 0.00018446096419875695,
      "loss": 11.1737,
      "step": 153700
    },
    {
      "epoch": 18.496991576413958,
      "grad_norm": Infinity,
      "learning_rate": 0.00018445892739768182,
      "loss": 11.1832,
      "step": 153710
    },
    {
      "epoch": 18.498194945848375,
      "grad_norm": Infinity,
      "learning_rate": 0.00018445689047437376,
      "loss": 11.184,
      "step": 153720
    },
    {
      "epoch": 18.499398315282793,
      "grad_norm": Infinity,
      "learning_rate": 0.0001844548534288357,
      "loss": 11.2043,
      "step": 153730
    },
    {
      "epoch": 18.500601684717207,
      "grad_norm": Infinity,
      "learning_rate": 0.00018445281626107063,
      "loss": 11.0292,
      "step": 153740
    },
    {
      "epoch": 18.501805054151625,
      "grad_norm": Infinity,
      "learning_rate": 0.0001844507789710815,
      "loss": 11.0998,
      "step": 153750
    },
    {
      "epoch": 18.503008423586042,
      "grad_norm": Infinity,
      "learning_rate": 0.00018444874155887123,
      "loss": 11.1909,
      "step": 153760
    },
    {
      "epoch": 18.504211793020456,
      "grad_norm": Infinity,
      "learning_rate": 0.00018444670402444278,
      "loss": 11.1862,
      "step": 153770
    },
    {
      "epoch": 18.505415162454874,
      "grad_norm": Infinity,
      "learning_rate": 0.0001844446663677991,
      "loss": 11.2138,
      "step": 153780
    },
    {
      "epoch": 18.50661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00018444262858894312,
      "loss": 11.1951,
      "step": 153790
    },
    {
      "epoch": 18.507821901323705,
      "grad_norm": Infinity,
      "learning_rate": 0.00018444059068787786,
      "loss": 11.0419,
      "step": 153800
    },
    {
      "epoch": 18.509025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.0001844385526646062,
      "loss": 11.0918,
      "step": 153810
    },
    {
      "epoch": 18.51022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.00018443651451913108,
      "loss": 11.1833,
      "step": 153820
    },
    {
      "epoch": 18.511432009626954,
      "grad_norm": Infinity,
      "learning_rate": 0.00018443447625145548,
      "loss": 11.2276,
      "step": 153830
    },
    {
      "epoch": 18.512635379061372,
      "grad_norm": Infinity,
      "learning_rate": 0.00018443243786158236,
      "loss": 11.2327,
      "step": 153840
    },
    {
      "epoch": 18.51383874849579,
      "grad_norm": Infinity,
      "learning_rate": 0.00018443039934951466,
      "loss": 11.0014,
      "step": 153850
    },
    {
      "epoch": 18.515042117930204,
      "grad_norm": Infinity,
      "learning_rate": 0.00018442836071525532,
      "loss": 11.2101,
      "step": 153860
    },
    {
      "epoch": 18.51624548736462,
      "grad_norm": Infinity,
      "learning_rate": 0.00018442632195880728,
      "loss": 11.1144,
      "step": 153870
    },
    {
      "epoch": 18.51744885679904,
      "grad_norm": Infinity,
      "learning_rate": 0.00018442428308017354,
      "loss": 11.1024,
      "step": 153880
    },
    {
      "epoch": 18.518652226233453,
      "grad_norm": Infinity,
      "learning_rate": 0.000184422244079357,
      "loss": 11.1752,
      "step": 153890
    },
    {
      "epoch": 18.51985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00018442020495636067,
      "loss": 11.108,
      "step": 153900
    },
    {
      "epoch": 18.521058965102288,
      "grad_norm": Infinity,
      "learning_rate": 0.00018441816571118742,
      "loss": 11.1124,
      "step": 153910
    },
    {
      "epoch": 18.522262334536702,
      "grad_norm": Infinity,
      "learning_rate": 0.00018441612634384025,
      "loss": 11.1661,
      "step": 153920
    },
    {
      "epoch": 18.52346570397112,
      "grad_norm": Infinity,
      "learning_rate": 0.0001844140868543221,
      "loss": 11.1174,
      "step": 153930
    },
    {
      "epoch": 18.524669073405537,
      "grad_norm": Infinity,
      "learning_rate": 0.00018441204724263592,
      "loss": 11.0932,
      "step": 153940
    },
    {
      "epoch": 18.52587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00018441000750878467,
      "loss": 11.1473,
      "step": 153950
    },
    {
      "epoch": 18.52707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.0001844079676527713,
      "loss": 11.1022,
      "step": 153960
    },
    {
      "epoch": 18.528279181708786,
      "grad_norm": Infinity,
      "learning_rate": 0.00018440592767459877,
      "loss": 11.0514,
      "step": 153970
    },
    {
      "epoch": 18.5294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00018440388757427,
      "loss": 11.1552,
      "step": 153980
    },
    {
      "epoch": 18.530685920577618,
      "grad_norm": Infinity,
      "learning_rate": 0.000184401847351788,
      "loss": 11.0548,
      "step": 153990
    },
    {
      "epoch": 18.531889290012035,
      "grad_norm": Infinity,
      "learning_rate": 0.00018439980700715568,
      "loss": 11.1378,
      "step": 154000
    },
    {
      "epoch": 18.53309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.000184397766540376,
      "loss": 11.1906,
      "step": 154010
    },
    {
      "epoch": 18.534296028880867,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843957259514519,
      "loss": 11.0453,
      "step": 154020
    },
    {
      "epoch": 18.535499398315284,
      "grad_norm": Infinity,
      "learning_rate": 0.00018439368524038633,
      "loss": 11.2196,
      "step": 154030
    },
    {
      "epoch": 18.5367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843916444071823,
      "loss": 11.2259,
      "step": 154040
    },
    {
      "epoch": 18.537906137184116,
      "grad_norm": Infinity,
      "learning_rate": 0.00018438960345184272,
      "loss": 11.239,
      "step": 154050
    },
    {
      "epoch": 18.53910950661853,
      "grad_norm": Infinity,
      "learning_rate": 0.00018438756237437054,
      "loss": 11.0352,
      "step": 154060
    },
    {
      "epoch": 18.540312876052948,
      "grad_norm": Infinity,
      "learning_rate": 0.00018438552117476872,
      "loss": 11.2338,
      "step": 154070
    },
    {
      "epoch": 18.541516245487365,
      "grad_norm": Infinity,
      "learning_rate": 0.00018438347985304024,
      "loss": 11.1626,
      "step": 154080
    },
    {
      "epoch": 18.54271961492178,
      "grad_norm": Infinity,
      "learning_rate": 0.00018438143840918798,
      "loss": 11.1056,
      "step": 154090
    },
    {
      "epoch": 18.543922984356197,
      "grad_norm": Infinity,
      "learning_rate": 0.00018437939684321498,
      "loss": 11.297,
      "step": 154100
    },
    {
      "epoch": 18.545126353790614,
      "grad_norm": Infinity,
      "learning_rate": 0.00018437735515512417,
      "loss": 11.2027,
      "step": 154110
    },
    {
      "epoch": 18.54632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00018437531334491848,
      "loss": 11.1862,
      "step": 154120
    },
    {
      "epoch": 18.547533092659446,
      "grad_norm": Infinity,
      "learning_rate": 0.00018437327141260088,
      "loss": 11.2691,
      "step": 154130
    },
    {
      "epoch": 18.548736462093864,
      "grad_norm": Infinity,
      "learning_rate": 0.00018437122935817434,
      "loss": 11.0923,
      "step": 154140
    },
    {
      "epoch": 18.549939831528278,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843691871816418,
      "loss": 11.1205,
      "step": 154150
    },
    {
      "epoch": 18.551143200962695,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843671448830062,
      "loss": 11.164,
      "step": 154160
    },
    {
      "epoch": 18.552346570397113,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843651024622705,
      "loss": 11.1202,
      "step": 154170
    },
    {
      "epoch": 18.553549939831527,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843630599194377,
      "loss": 11.1886,
      "step": 154180
    },
    {
      "epoch": 18.554753309265944,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843610172545107,
      "loss": 11.1563,
      "step": 154190
    },
    {
      "epoch": 18.555956678700362,
      "grad_norm": Infinity,
      "learning_rate": 0.00018435897446749248,
      "loss": 11.1706,
      "step": 154200
    },
    {
      "epoch": 18.557160048134776,
      "grad_norm": Infinity,
      "learning_rate": 0.000184356931558386,
      "loss": 11.0745,
      "step": 154210
    },
    {
      "epoch": 18.558363417569193,
      "grad_norm": Infinity,
      "learning_rate": 0.00018435488852719422,
      "loss": 11.1546,
      "step": 154220
    },
    {
      "epoch": 18.55956678700361,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843528453739201,
      "loss": 11.2963,
      "step": 154230
    },
    {
      "epoch": 18.560770156438025,
      "grad_norm": Infinity,
      "learning_rate": 0.00018435080209856656,
      "loss": 11.1517,
      "step": 154240
    },
    {
      "epoch": 18.561973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843487587011366,
      "loss": 11.1994,
      "step": 154250
    },
    {
      "epoch": 18.56317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00018434671518163315,
      "loss": 11.0373,
      "step": 154260
    },
    {
      "epoch": 18.564380264741274,
      "grad_norm": Infinity,
      "learning_rate": 0.00018434467154005918,
      "loss": 11.1424,
      "step": 154270
    },
    {
      "epoch": 18.56558363417569,
      "grad_norm": Infinity,
      "learning_rate": 0.00018434262777641767,
      "loss": 11.2854,
      "step": 154280
    },
    {
      "epoch": 18.56678700361011,
      "grad_norm": Infinity,
      "learning_rate": 0.00018434058389071154,
      "loss": 11.1538,
      "step": 154290
    },
    {
      "epoch": 18.567990373044523,
      "grad_norm": Infinity,
      "learning_rate": 0.00018433853988294375,
      "loss": 11.1253,
      "step": 154300
    },
    {
      "epoch": 18.56919374247894,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843364957531173,
      "loss": 11.1735,
      "step": 154310
    },
    {
      "epoch": 18.57039711191336,
      "grad_norm": Infinity,
      "learning_rate": 0.00018433445150123508,
      "loss": 11.0829,
      "step": 154320
    },
    {
      "epoch": 18.571600481347772,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843324071273001,
      "loss": 11.2586,
      "step": 154330
    },
    {
      "epoch": 18.57280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.00018433036263131532,
      "loss": 11.1248,
      "step": 154340
    },
    {
      "epoch": 18.574007220216608,
      "grad_norm": Infinity,
      "learning_rate": 0.00018432831801328366,
      "loss": 11.111,
      "step": 154350
    },
    {
      "epoch": 18.57521058965102,
      "grad_norm": Infinity,
      "learning_rate": 0.00018432627327320816,
      "loss": 11.0498,
      "step": 154360
    },
    {
      "epoch": 18.57641395908544,
      "grad_norm": Infinity,
      "learning_rate": 0.00018432422841109167,
      "loss": 11.2345,
      "step": 154370
    },
    {
      "epoch": 18.577617328519857,
      "grad_norm": Infinity,
      "learning_rate": 0.00018432218342693721,
      "loss": 11.1254,
      "step": 154380
    },
    {
      "epoch": 18.57882069795427,
      "grad_norm": Infinity,
      "learning_rate": 0.00018432013832074776,
      "loss": 11.2464,
      "step": 154390
    },
    {
      "epoch": 18.58002406738869,
      "grad_norm": Infinity,
      "learning_rate": 0.00018431809309252622,
      "loss": 11.1221,
      "step": 154400
    },
    {
      "epoch": 18.581227436823106,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843160477422756,
      "loss": 11.194,
      "step": 154410
    },
    {
      "epoch": 18.58243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.00018431400226999882,
      "loss": 11.2833,
      "step": 154420
    },
    {
      "epoch": 18.583634175691937,
      "grad_norm": Infinity,
      "learning_rate": 0.00018431195667569888,
      "loss": 11.1216,
      "step": 154430
    },
    {
      "epoch": 18.584837545126355,
      "grad_norm": Infinity,
      "learning_rate": 0.00018430991095937872,
      "loss": 11.0644,
      "step": 154440
    },
    {
      "epoch": 18.58604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00018430786512104127,
      "loss": 11.1347,
      "step": 154450
    },
    {
      "epoch": 18.587244283995187,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843058191606896,
      "loss": 11.1396,
      "step": 154460
    },
    {
      "epoch": 18.588447653429604,
      "grad_norm": Infinity,
      "learning_rate": 0.00018430377307832653,
      "loss": 11.1106,
      "step": 154470
    },
    {
      "epoch": 18.589651022864018,
      "grad_norm": Infinity,
      "learning_rate": 0.0001843017268739551,
      "loss": 11.1095,
      "step": 154480
    },
    {
      "epoch": 18.590854392298436,
      "grad_norm": Infinity,
      "learning_rate": 0.00018429968054757827,
      "loss": 11.1176,
      "step": 154490
    },
    {
      "epoch": 18.592057761732853,
      "grad_norm": Infinity,
      "learning_rate": 0.00018429763409919897,
      "loss": 11.0243,
      "step": 154500
    },
    {
      "epoch": 18.593261131167267,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842955875288202,
      "loss": 11.0968,
      "step": 154510
    },
    {
      "epoch": 18.594464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842935408364449,
      "loss": 11.139,
      "step": 154520
    },
    {
      "epoch": 18.595667870036102,
      "grad_norm": Infinity,
      "learning_rate": 0.00018429149402207604,
      "loss": 11.2446,
      "step": 154530
    },
    {
      "epoch": 18.596871239470516,
      "grad_norm": Infinity,
      "learning_rate": 0.00018428944708571656,
      "loss": 11.1045,
      "step": 154540
    },
    {
      "epoch": 18.598074608904934,
      "grad_norm": Infinity,
      "learning_rate": 0.00018428740002736946,
      "loss": 11.2959,
      "step": 154550
    },
    {
      "epoch": 18.59927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00018428535284703768,
      "loss": 11.1471,
      "step": 154560
    },
    {
      "epoch": 18.600481347773766,
      "grad_norm": Infinity,
      "learning_rate": 0.00018428330554472416,
      "loss": 11.1321,
      "step": 154570
    },
    {
      "epoch": 18.601684717208183,
      "grad_norm": Infinity,
      "learning_rate": 0.00018428125812043194,
      "loss": 11.1653,
      "step": 154580
    },
    {
      "epoch": 18.6028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.00018427921057416388,
      "loss": 11.0967,
      "step": 154590
    },
    {
      "epoch": 18.604091456077015,
      "grad_norm": Infinity,
      "learning_rate": 0.00018427716290592302,
      "loss": 11.0111,
      "step": 154600
    },
    {
      "epoch": 18.605294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00018427511511571228,
      "loss": 11.1689,
      "step": 154610
    },
    {
      "epoch": 18.60649819494585,
      "grad_norm": Infinity,
      "learning_rate": 0.00018427306720353466,
      "loss": 11.1187,
      "step": 154620
    },
    {
      "epoch": 18.607701564380264,
      "grad_norm": Infinity,
      "learning_rate": 0.00018427101916939308,
      "loss": 11.2486,
      "step": 154630
    },
    {
      "epoch": 18.60890493381468,
      "grad_norm": Infinity,
      "learning_rate": 0.00018426897101329054,
      "loss": 11.0974,
      "step": 154640
    },
    {
      "epoch": 18.6101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.00018426692273523,
      "loss": 11.228,
      "step": 154650
    },
    {
      "epoch": 18.611311672683513,
      "grad_norm": Infinity,
      "learning_rate": 0.00018426487433521445,
      "loss": 11.1379,
      "step": 154660
    },
    {
      "epoch": 18.61251504211793,
      "grad_norm": Infinity,
      "learning_rate": 0.00018426282581324678,
      "loss": 11.1701,
      "step": 154670
    },
    {
      "epoch": 18.613718411552348,
      "grad_norm": Infinity,
      "learning_rate": 0.00018426077716932998,
      "loss": 11.1545,
      "step": 154680
    },
    {
      "epoch": 18.614921780986762,
      "grad_norm": Infinity,
      "learning_rate": 0.00018425872840346707,
      "loss": 11.1638,
      "step": 154690
    },
    {
      "epoch": 18.61612515042118,
      "grad_norm": Infinity,
      "learning_rate": 0.00018425667951566096,
      "loss": 11.1699,
      "step": 154700
    },
    {
      "epoch": 18.617328519855597,
      "grad_norm": Infinity,
      "learning_rate": 0.00018425463050591464,
      "loss": 11.2327,
      "step": 154710
    },
    {
      "epoch": 18.61853188929001,
      "grad_norm": Infinity,
      "learning_rate": 0.00018425258137423108,
      "loss": 11.1129,
      "step": 154720
    },
    {
      "epoch": 18.61973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842505321206132,
      "loss": 11.1449,
      "step": 154730
    },
    {
      "epoch": 18.620938628158846,
      "grad_norm": Infinity,
      "learning_rate": 0.00018424848274506403,
      "loss": 11.019,
      "step": 154740
    },
    {
      "epoch": 18.62214199759326,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842464332475865,
      "loss": 11.1304,
      "step": 154750
    },
    {
      "epoch": 18.623345367027678,
      "grad_norm": Infinity,
      "learning_rate": 0.00018424438362818358,
      "loss": 11.1931,
      "step": 154760
    },
    {
      "epoch": 18.624548736462096,
      "grad_norm": Infinity,
      "learning_rate": 0.00018424233388685823,
      "loss": 11.0813,
      "step": 154770
    },
    {
      "epoch": 18.62575210589651,
      "grad_norm": Infinity,
      "learning_rate": 0.00018424028402361345,
      "loss": 11.1179,
      "step": 154780
    },
    {
      "epoch": 18.626955475330927,
      "grad_norm": Infinity,
      "learning_rate": 0.00018423823403845212,
      "loss": 11.2252,
      "step": 154790
    },
    {
      "epoch": 18.628158844765345,
      "grad_norm": Infinity,
      "learning_rate": 0.00018423618393137732,
      "loss": 11.2128,
      "step": 154800
    },
    {
      "epoch": 18.62936221419976,
      "grad_norm": Infinity,
      "learning_rate": 0.00018423413370239198,
      "loss": 11.1383,
      "step": 154810
    },
    {
      "epoch": 18.630565583634176,
      "grad_norm": Infinity,
      "learning_rate": 0.000184232083351499,
      "loss": 11.1558,
      "step": 154820
    },
    {
      "epoch": 18.63176895306859,
      "grad_norm": Infinity,
      "learning_rate": 0.00018423003287870143,
      "loss": 11.1106,
      "step": 154830
    },
    {
      "epoch": 18.632972322503008,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842279822840022,
      "loss": 11.2044,
      "step": 154840
    },
    {
      "epoch": 18.634175691937426,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842259315674043,
      "loss": 11.0732,
      "step": 154850
    },
    {
      "epoch": 18.63537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.00018422388072891063,
      "loss": 11.2463,
      "step": 154860
    },
    {
      "epoch": 18.636582430806257,
      "grad_norm": Infinity,
      "learning_rate": 0.00018422182976852427,
      "loss": 11.208,
      "step": 154870
    },
    {
      "epoch": 18.637785800240675,
      "grad_norm": Infinity,
      "learning_rate": 0.00018421977868624808,
      "loss": 11.0892,
      "step": 154880
    },
    {
      "epoch": 18.63898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842177274820851,
      "loss": 11.0983,
      "step": 154890
    },
    {
      "epoch": 18.640192539109506,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842156761560383,
      "loss": 11.0362,
      "step": 154900
    },
    {
      "epoch": 18.641395908543924,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842136247081106,
      "loss": 11.2536,
      "step": 154910
    },
    {
      "epoch": 18.642599277978338,
      "grad_norm": Infinity,
      "learning_rate": 0.000184211573138305,
      "loss": 11.1086,
      "step": 154920
    },
    {
      "epoch": 18.643802647412755,
      "grad_norm": Infinity,
      "learning_rate": 0.00018420952144662444,
      "loss": 11.1801,
      "step": 154930
    },
    {
      "epoch": 18.645006016847173,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842074696330719,
      "loss": 11.1866,
      "step": 154940
    },
    {
      "epoch": 18.646209386281587,
      "grad_norm": Infinity,
      "learning_rate": 0.00018420541769765042,
      "loss": 11.1017,
      "step": 154950
    },
    {
      "epoch": 18.647412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.00018420336564036287,
      "loss": 11.2785,
      "step": 154960
    },
    {
      "epoch": 18.648616125150422,
      "grad_norm": Infinity,
      "learning_rate": 0.0001842013134612123,
      "loss": 11.0446,
      "step": 154970
    },
    {
      "epoch": 18.649819494584836,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841992611602016,
      "loss": 11.1732,
      "step": 154980
    },
    {
      "epoch": 18.651022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841972087373338,
      "loss": 11.136,
      "step": 154990
    },
    {
      "epoch": 18.65222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.00018419515619261186,
      "loss": 11.0919,
      "step": 155000
    },
    {
      "epoch": 18.653429602888085,
      "grad_norm": Infinity,
      "learning_rate": 0.00018419310352603868,
      "loss": 11.1219,
      "step": 155010
    },
    {
      "epoch": 18.654632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00018419105073761737,
      "loss": 11.1603,
      "step": 155020
    },
    {
      "epoch": 18.65583634175692,
      "grad_norm": Infinity,
      "learning_rate": 0.00018418899782735078,
      "loss": 11.1119,
      "step": 155030
    },
    {
      "epoch": 18.657039711191334,
      "grad_norm": Infinity,
      "learning_rate": 0.00018418694479524193,
      "loss": 11.2415,
      "step": 155040
    },
    {
      "epoch": 18.658243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841848916412938,
      "loss": 11.1529,
      "step": 155050
    },
    {
      "epoch": 18.65944645006017,
      "grad_norm": Infinity,
      "learning_rate": 0.00018418283836550936,
      "loss": 11.0478,
      "step": 155060
    },
    {
      "epoch": 18.660649819494584,
      "grad_norm": Infinity,
      "learning_rate": 0.00018418078496789154,
      "loss": 11.1712,
      "step": 155070
    },
    {
      "epoch": 18.661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00018417873144844333,
      "loss": 11.0886,
      "step": 155080
    },
    {
      "epoch": 18.66305655836342,
      "grad_norm": Infinity,
      "learning_rate": 0.00018417667780716777,
      "loss": 11.0901,
      "step": 155090
    },
    {
      "epoch": 18.664259927797833,
      "grad_norm": Infinity,
      "learning_rate": 0.00018417462404406773,
      "loss": 11.1937,
      "step": 155100
    },
    {
      "epoch": 18.66546329723225,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841725701591462,
      "loss": 11.0931,
      "step": 155110
    },
    {
      "epoch": 18.666666666666668,
      "grad_norm": Infinity,
      "learning_rate": 0.00018417051615240624,
      "loss": 11.0515,
      "step": 155120
    },
    {
      "epoch": 18.667870036101082,
      "grad_norm": Infinity,
      "learning_rate": 0.00018416846202385073,
      "loss": 11.2758,
      "step": 155130
    },
    {
      "epoch": 18.6690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841664077734827,
      "loss": 11.1772,
      "step": 155140
    },
    {
      "epoch": 18.670276774969917,
      "grad_norm": Infinity,
      "learning_rate": 0.00018416435340130508,
      "loss": 11.0956,
      "step": 155150
    },
    {
      "epoch": 18.67148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00018416229890732083,
      "loss": 11.0273,
      "step": 155160
    },
    {
      "epoch": 18.67268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.000184160244291533,
      "loss": 11.207,
      "step": 155170
    },
    {
      "epoch": 18.673886883273166,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841581895539445,
      "loss": 11.1246,
      "step": 155180
    },
    {
      "epoch": 18.67509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841561346945583,
      "loss": 11.2164,
      "step": 155190
    },
    {
      "epoch": 18.676293622141998,
      "grad_norm": Infinity,
      "learning_rate": 0.00018415407971337743,
      "loss": 11.1303,
      "step": 155200
    },
    {
      "epoch": 18.677496991576415,
      "grad_norm": Infinity,
      "learning_rate": 0.00018415202461040484,
      "loss": 11.0508,
      "step": 155210
    },
    {
      "epoch": 18.67870036101083,
      "grad_norm": Infinity,
      "learning_rate": 0.00018414996938564347,
      "loss": 11.1942,
      "step": 155220
    },
    {
      "epoch": 18.679903730445247,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841479140390963,
      "loss": 11.1507,
      "step": 155230
    },
    {
      "epoch": 18.681107099879664,
      "grad_norm": Infinity,
      "learning_rate": 0.00018414585857076635,
      "loss": 11.094,
      "step": 155240
    },
    {
      "epoch": 18.68231046931408,
      "grad_norm": Infinity,
      "learning_rate": 0.00018414380298065658,
      "loss": 11.3475,
      "step": 155250
    },
    {
      "epoch": 18.683513838748496,
      "grad_norm": Infinity,
      "learning_rate": 0.00018414174726876992,
      "loss": 11.2109,
      "step": 155260
    },
    {
      "epoch": 18.684717208182914,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841396914351094,
      "loss": 11.328,
      "step": 155270
    },
    {
      "epoch": 18.685920577617328,
      "grad_norm": Infinity,
      "learning_rate": 0.00018413763547967796,
      "loss": 11.2317,
      "step": 155280
    },
    {
      "epoch": 18.687123947051745,
      "grad_norm": Infinity,
      "learning_rate": 0.00018413557940247862,
      "loss": 11.0034,
      "step": 155290
    },
    {
      "epoch": 18.688327316486163,
      "grad_norm": Infinity,
      "learning_rate": 0.00018413352320351427,
      "loss": 11.2417,
      "step": 155300
    },
    {
      "epoch": 18.689530685920577,
      "grad_norm": Infinity,
      "learning_rate": 0.00018413146688278797,
      "loss": 10.9798,
      "step": 155310
    },
    {
      "epoch": 18.690734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.00018412941044030267,
      "loss": 11.2268,
      "step": 155320
    },
    {
      "epoch": 18.691937424789412,
      "grad_norm": Infinity,
      "learning_rate": 0.00018412735387606131,
      "loss": 11.1689,
      "step": 155330
    },
    {
      "epoch": 18.693140794223826,
      "grad_norm": Infinity,
      "learning_rate": 0.00018412529719006695,
      "loss": 11.1137,
      "step": 155340
    },
    {
      "epoch": 18.694344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00018412324038232249,
      "loss": 11.1095,
      "step": 155350
    },
    {
      "epoch": 18.69554753309266,
      "grad_norm": Infinity,
      "learning_rate": 0.00018412118345283092,
      "loss": 10.9851,
      "step": 155360
    },
    {
      "epoch": 18.696750902527075,
      "grad_norm": Infinity,
      "learning_rate": 0.00018411912640159525,
      "loss": 11.0877,
      "step": 155370
    },
    {
      "epoch": 18.697954271961493,
      "grad_norm": Infinity,
      "learning_rate": 0.00018411706922861844,
      "loss": 11.1739,
      "step": 155380
    },
    {
      "epoch": 18.69915764139591,
      "grad_norm": Infinity,
      "learning_rate": 0.00018411501193390346,
      "loss": 11.1279,
      "step": 155390
    },
    {
      "epoch": 18.700361010830324,
      "grad_norm": Infinity,
      "learning_rate": 0.00018411295451745328,
      "loss": 11.1035,
      "step": 155400
    },
    {
      "epoch": 18.70156438026474,
      "grad_norm": Infinity,
      "learning_rate": 0.00018411089697927086,
      "loss": 11.1095,
      "step": 155410
    },
    {
      "epoch": 18.70276774969916,
      "grad_norm": Infinity,
      "learning_rate": 0.00018410883931935926,
      "loss": 11.1272,
      "step": 155420
    },
    {
      "epoch": 18.703971119133573,
      "grad_norm": Infinity,
      "learning_rate": 0.00018410678153772136,
      "loss": 11.1291,
      "step": 155430
    },
    {
      "epoch": 18.70517448856799,
      "grad_norm": Infinity,
      "learning_rate": 0.0001841047236343602,
      "loss": 11.1143,
      "step": 155440
    },
    {
      "epoch": 18.706377858002405,
      "grad_norm": Infinity,
      "learning_rate": 0.00018410266560927875,
      "loss": 11.081,
      "step": 155450
    },
    {
      "epoch": 18.707581227436823,
      "grad_norm": Infinity,
      "learning_rate": 0.00018410060746247995,
      "loss": 11.1918,
      "step": 155460
    },
    {
      "epoch": 18.70878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00018409854919396686,
      "loss": 11.1112,
      "step": 155470
    },
    {
      "epoch": 18.709987966305654,
      "grad_norm": Infinity,
      "learning_rate": 0.00018409649080374237,
      "loss": 11.1154,
      "step": 155480
    },
    {
      "epoch": 18.71119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.00018409443229180946,
      "loss": 11.1244,
      "step": 155490
    },
    {
      "epoch": 18.71239470517449,
      "grad_norm": Infinity,
      "learning_rate": 0.00018409237365817118,
      "loss": 11.1785,
      "step": 155500
    },
    {
      "epoch": 18.713598074608903,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840903149028305,
      "loss": 11.0254,
      "step": 155510
    },
    {
      "epoch": 18.71480144404332,
      "grad_norm": Infinity,
      "learning_rate": 0.00018408825602579033,
      "loss": 11.096,
      "step": 155520
    },
    {
      "epoch": 18.71600481347774,
      "grad_norm": Infinity,
      "learning_rate": 0.00018408619702705372,
      "loss": 11.1088,
      "step": 155530
    },
    {
      "epoch": 18.717208182912152,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840841379066236,
      "loss": 11.1197,
      "step": 155540
    },
    {
      "epoch": 18.71841155234657,
      "grad_norm": Infinity,
      "learning_rate": 0.000184082078664503,
      "loss": 11.043,
      "step": 155550
    },
    {
      "epoch": 18.719614921780988,
      "grad_norm": Infinity,
      "learning_rate": 0.00018408001930069484,
      "loss": 11.0468,
      "step": 155560
    },
    {
      "epoch": 18.7208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00018407795981520216,
      "loss": 11.0612,
      "step": 155570
    },
    {
      "epoch": 18.72202166064982,
      "grad_norm": Infinity,
      "learning_rate": 0.00018407590020802793,
      "loss": 11.2637,
      "step": 155580
    },
    {
      "epoch": 18.723225030084237,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840738404791751,
      "loss": 11.1286,
      "step": 155590
    },
    {
      "epoch": 18.72442839951865,
      "grad_norm": Infinity,
      "learning_rate": 0.00018407178062864664,
      "loss": 10.9937,
      "step": 155600
    },
    {
      "epoch": 18.72563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00018406972065644557,
      "loss": 11.0324,
      "step": 155610
    },
    {
      "epoch": 18.726835138387486,
      "grad_norm": Infinity,
      "learning_rate": 0.00018406766056257485,
      "loss": 11.0811,
      "step": 155620
    },
    {
      "epoch": 18.7280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.00018406560034703748,
      "loss": 11.1048,
      "step": 155630
    },
    {
      "epoch": 18.729241877256317,
      "grad_norm": Infinity,
      "learning_rate": 0.00018406354000983642,
      "loss": 10.9592,
      "step": 155640
    },
    {
      "epoch": 18.730445246690735,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840614795509747,
      "loss": 11.0839,
      "step": 155650
    },
    {
      "epoch": 18.73164861612515,
      "grad_norm": Infinity,
      "learning_rate": 0.00018405941897045523,
      "loss": 11.1915,
      "step": 155660
    },
    {
      "epoch": 18.732851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.00018405735826828107,
      "loss": 11.0202,
      "step": 155670
    },
    {
      "epoch": 18.734055354993984,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840552974444551,
      "loss": 11.1132,
      "step": 155680
    },
    {
      "epoch": 18.735258724428398,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840532364989804,
      "loss": 11.2307,
      "step": 155690
    },
    {
      "epoch": 18.736462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840511754318599,
      "loss": 11.1407,
      "step": 155700
    },
    {
      "epoch": 18.737665463297233,
      "grad_norm": Infinity,
      "learning_rate": 0.00018404911424309658,
      "loss": 11.1428,
      "step": 155710
    },
    {
      "epoch": 18.738868832731647,
      "grad_norm": Infinity,
      "learning_rate": 0.00018404705293269347,
      "loss": 11.1013,
      "step": 155720
    },
    {
      "epoch": 18.740072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00018404499150065351,
      "loss": 11.1795,
      "step": 155730
    },
    {
      "epoch": 18.741275571600482,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840429299469797,
      "loss": 11.1118,
      "step": 155740
    },
    {
      "epoch": 18.742478941034896,
      "grad_norm": Infinity,
      "learning_rate": 0.00018404086827167502,
      "loss": 11.0965,
      "step": 155750
    },
    {
      "epoch": 18.743682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00018403880647474249,
      "loss": 11.1054,
      "step": 155760
    },
    {
      "epoch": 18.74488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.00018403674455618501,
      "loss": 11.0522,
      "step": 155770
    },
    {
      "epoch": 18.746089049338146,
      "grad_norm": Infinity,
      "learning_rate": 0.00018403468251600562,
      "loss": 10.9737,
      "step": 155780
    },
    {
      "epoch": 18.747292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840326203542073,
      "loss": 11.2099,
      "step": 155790
    },
    {
      "epoch": 18.74849578820698,
      "grad_norm": Infinity,
      "learning_rate": 0.00018403055807079307,
      "loss": 11.174,
      "step": 155800
    },
    {
      "epoch": 18.749699157641395,
      "grad_norm": Infinity,
      "learning_rate": 0.00018402849566576582,
      "loss": 11.0626,
      "step": 155810
    },
    {
      "epoch": 18.750902527075812,
      "grad_norm": Infinity,
      "learning_rate": 0.00018402643313912862,
      "loss": 11.1572,
      "step": 155820
    },
    {
      "epoch": 18.75210589651023,
      "grad_norm": Infinity,
      "learning_rate": 0.00018402437049088444,
      "loss": 11.0243,
      "step": 155830
    },
    {
      "epoch": 18.753309265944644,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840223077210362,
      "loss": 11.0665,
      "step": 155840
    },
    {
      "epoch": 18.75451263537906,
      "grad_norm": Infinity,
      "learning_rate": 0.000184020244829587,
      "loss": 11.1575,
      "step": 155850
    },
    {
      "epoch": 18.75571600481348,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840181818165397,
      "loss": 11.2205,
      "step": 155860
    },
    {
      "epoch": 18.756919374247893,
      "grad_norm": Infinity,
      "learning_rate": 0.00018401611868189738,
      "loss": 11.2051,
      "step": 155870
    },
    {
      "epoch": 18.75812274368231,
      "grad_norm": Infinity,
      "learning_rate": 0.00018401405542566298,
      "loss": 11.1057,
      "step": 155880
    },
    {
      "epoch": 18.759326113116728,
      "grad_norm": Infinity,
      "learning_rate": 0.00018401199204783953,
      "loss": 11.094,
      "step": 155890
    },
    {
      "epoch": 18.760529482551142,
      "grad_norm": Infinity,
      "learning_rate": 0.00018400992854842997,
      "loss": 11.0583,
      "step": 155900
    },
    {
      "epoch": 18.76173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840078649274373,
      "loss": 11.1038,
      "step": 155910
    },
    {
      "epoch": 18.762936221419977,
      "grad_norm": Infinity,
      "learning_rate": 0.0001840058011848645,
      "loss": 11.3338,
      "step": 155920
    },
    {
      "epoch": 18.76413959085439,
      "grad_norm": Infinity,
      "learning_rate": 0.00018400373732071457,
      "loss": 11.0459,
      "step": 155930
    },
    {
      "epoch": 18.76534296028881,
      "grad_norm": Infinity,
      "learning_rate": 0.00018400167333499048,
      "loss": 11.2301,
      "step": 155940
    },
    {
      "epoch": 18.766546329723226,
      "grad_norm": Infinity,
      "learning_rate": 0.00018399960922769527,
      "loss": 11.1998,
      "step": 155950
    },
    {
      "epoch": 18.76774969915764,
      "grad_norm": Infinity,
      "learning_rate": 0.00018399754499883185,
      "loss": 11.096,
      "step": 155960
    },
    {
      "epoch": 18.768953068592058,
      "grad_norm": Infinity,
      "learning_rate": 0.00018399548064840322,
      "loss": 11.1265,
      "step": 155970
    },
    {
      "epoch": 18.770156438026476,
      "grad_norm": Infinity,
      "learning_rate": 0.00018399341617641245,
      "loss": 11.1074,
      "step": 155980
    },
    {
      "epoch": 18.77135980746089,
      "grad_norm": Infinity,
      "learning_rate": 0.00018399135158286245,
      "loss": 11.1348,
      "step": 155990
    },
    {
      "epoch": 18.772563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839892868677562,
      "loss": 11.0085,
      "step": 156000
    },
    {
      "epoch": 18.773766546329725,
      "grad_norm": Infinity,
      "learning_rate": 0.00018398722203109675,
      "loss": 11.0944,
      "step": 156010
    },
    {
      "epoch": 18.77496991576414,
      "grad_norm": Infinity,
      "learning_rate": 0.00018398515707288704,
      "loss": 11.1064,
      "step": 156020
    },
    {
      "epoch": 18.776173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00018398309199313008,
      "loss": 11.202,
      "step": 156030
    },
    {
      "epoch": 18.777376654632974,
      "grad_norm": Infinity,
      "learning_rate": 0.00018398102679182885,
      "loss": 11.0866,
      "step": 156040
    },
    {
      "epoch": 18.778580024067388,
      "grad_norm": Infinity,
      "learning_rate": 0.00018397896146898633,
      "loss": 11.1692,
      "step": 156050
    },
    {
      "epoch": 18.779783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839768960246055,
      "loss": 11.1189,
      "step": 156060
    },
    {
      "epoch": 18.780986762936223,
      "grad_norm": Infinity,
      "learning_rate": 0.00018397483045868942,
      "loss": 11.1767,
      "step": 156070
    },
    {
      "epoch": 18.782190132370637,
      "grad_norm": Infinity,
      "learning_rate": 0.00018397276477124098,
      "loss": 11.1687,
      "step": 156080
    },
    {
      "epoch": 18.783393501805055,
      "grad_norm": Infinity,
      "learning_rate": 0.00018397069896226325,
      "loss": 11.1573,
      "step": 156090
    },
    {
      "epoch": 18.784596871239472,
      "grad_norm": Infinity,
      "learning_rate": 0.00018396863303175918,
      "loss": 11.0969,
      "step": 156100
    },
    {
      "epoch": 18.785800240673886,
      "grad_norm": Infinity,
      "learning_rate": 0.00018396656697973175,
      "loss": 11.0836,
      "step": 156110
    },
    {
      "epoch": 18.787003610108304,
      "grad_norm": Infinity,
      "learning_rate": 0.00018396450080618396,
      "loss": 11.1231,
      "step": 156120
    },
    {
      "epoch": 18.78820697954272,
      "grad_norm": Infinity,
      "learning_rate": 0.00018396243451111884,
      "loss": 11.0268,
      "step": 156130
    },
    {
      "epoch": 18.789410348977135,
      "grad_norm": Infinity,
      "learning_rate": 0.00018396036809453933,
      "loss": 11.2009,
      "step": 156140
    },
    {
      "epoch": 18.790613718411553,
      "grad_norm": Infinity,
      "learning_rate": 0.00018395830155644845,
      "loss": 11.106,
      "step": 156150
    },
    {
      "epoch": 18.79181708784597,
      "grad_norm": Infinity,
      "learning_rate": 0.00018395623489684916,
      "loss": 11.1531,
      "step": 156160
    },
    {
      "epoch": 18.793020457280385,
      "grad_norm": Infinity,
      "learning_rate": 0.00018395416811574448,
      "loss": 11.1239,
      "step": 156170
    },
    {
      "epoch": 18.794223826714802,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839521012131374,
      "loss": 11.0848,
      "step": 156180
    },
    {
      "epoch": 18.79542719614922,
      "grad_norm": Infinity,
      "learning_rate": 0.00018395003418903086,
      "loss": 11.0738,
      "step": 156190
    },
    {
      "epoch": 18.796630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00018394796704342795,
      "loss": 11.1085,
      "step": 156200
    },
    {
      "epoch": 18.79783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839458997763316,
      "loss": 11.1598,
      "step": 156210
    },
    {
      "epoch": 18.799037304452465,
      "grad_norm": Infinity,
      "learning_rate": 0.00018394383238774477,
      "loss": 11.1524,
      "step": 156220
    },
    {
      "epoch": 18.800240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.00018394176487767052,
      "loss": 11.0619,
      "step": 156230
    },
    {
      "epoch": 18.8014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839396972461118,
      "loss": 11.0887,
      "step": 156240
    },
    {
      "epoch": 18.802647412755714,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839376294930716,
      "loss": 11.1339,
      "step": 156250
    },
    {
      "epoch": 18.803850782190132,
      "grad_norm": Infinity,
      "learning_rate": 0.00018393556161855297,
      "loss": 11.2568,
      "step": 156260
    },
    {
      "epoch": 18.80505415162455,
      "grad_norm": Infinity,
      "learning_rate": 0.00018393349362255882,
      "loss": 11.2225,
      "step": 156270
    },
    {
      "epoch": 18.806257521058964,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839314255050922,
      "loss": 11.1804,
      "step": 156280
    },
    {
      "epoch": 18.80746089049338,
      "grad_norm": Infinity,
      "learning_rate": 0.00018392935726615608,
      "loss": 11.0418,
      "step": 156290
    },
    {
      "epoch": 18.8086642599278,
      "grad_norm": Infinity,
      "learning_rate": 0.00018392728890575345,
      "loss": 11.169,
      "step": 156300
    },
    {
      "epoch": 18.809867629362213,
      "grad_norm": Infinity,
      "learning_rate": 0.00018392522042388733,
      "loss": 11.1228,
      "step": 156310
    },
    {
      "epoch": 18.81107099879663,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839231518205607,
      "loss": 11.2595,
      "step": 156320
    },
    {
      "epoch": 18.812274368231048,
      "grad_norm": Infinity,
      "learning_rate": 0.00018392108309577652,
      "loss": 11.2344,
      "step": 156330
    },
    {
      "epoch": 18.813477737665462,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839190142495378,
      "loss": 11.0632,
      "step": 156340
    },
    {
      "epoch": 18.81468110709988,
      "grad_norm": Infinity,
      "learning_rate": 0.00018391694528184757,
      "loss": 11.1457,
      "step": 156350
    },
    {
      "epoch": 18.815884476534297,
      "grad_norm": Infinity,
      "learning_rate": 0.00018391487619270883,
      "loss": 11.2568,
      "step": 156360
    },
    {
      "epoch": 18.81708784596871,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839128069821245,
      "loss": 11.1566,
      "step": 156370
    },
    {
      "epoch": 18.81829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.00018391073765009765,
      "loss": 11.3208,
      "step": 156380
    },
    {
      "epoch": 18.819494584837546,
      "grad_norm": Infinity,
      "learning_rate": 0.00018390866819663123,
      "loss": 11.101,
      "step": 156390
    },
    {
      "epoch": 18.82069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.0001839065986217283,
      "loss": 11.1122,
      "step": 156400
    },
    {
      "epoch": 18.821901323706378,
      "grad_norm": Infinity,
      "learning_rate": 0.00018390452892539175,
      "loss": 11.1221,
      "step": 156410
    },
    {
      "epoch": 18.823104693140795,
      "grad_norm": Infinity,
      "learning_rate": 0.00018390245910762464,
      "loss": 11.084,
      "step": 156420
    },
    {
      "epoch": 18.82430806257521,
      "grad_norm": Infinity,
      "learning_rate": 0.00018390038916842998,
      "loss": 11.1787,
      "step": 156430
    },
    {
      "epoch": 18.825511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00018389831910781073,
      "loss": 11.1843,
      "step": 156440
    },
    {
      "epoch": 18.826714801444044,
      "grad_norm": Infinity,
      "learning_rate": 0.00018389624892576988,
      "loss": 11.1098,
      "step": 156450
    },
    {
      "epoch": 18.82791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00018389417862231048,
      "loss": 11.1299,
      "step": 156460
    },
    {
      "epoch": 18.829121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00018389210819743545,
      "loss": 11.0784,
      "step": 156470
    },
    {
      "epoch": 18.830324909747294,
      "grad_norm": Infinity,
      "learning_rate": 0.00018389003765114785,
      "loss": 11.1469,
      "step": 156480
    },
    {
      "epoch": 18.831528279181708,
      "grad_norm": Infinity,
      "learning_rate": 0.00018388796698345065,
      "loss": 11.1581,
      "step": 156490
    },
    {
      "epoch": 18.832731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.00018388589619434687,
      "loss": 11.0997,
      "step": 156500
    },
    {
      "epoch": 18.833935018050543,
      "grad_norm": Infinity,
      "learning_rate": 0.00018388382528383948,
      "loss": 11.1977,
      "step": 156510
    },
    {
      "epoch": 18.835138387484957,
      "grad_norm": Infinity,
      "learning_rate": 0.00018388175425193147,
      "loss": 11.1397,
      "step": 156520
    },
    {
      "epoch": 18.836341756919374,
      "grad_norm": Infinity,
      "learning_rate": 0.00018387968309862583,
      "loss": 11.3134,
      "step": 156530
    },
    {
      "epoch": 18.837545126353792,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838776118239256,
      "loss": 11.0607,
      "step": 156540
    },
    {
      "epoch": 18.838748495788206,
      "grad_norm": Infinity,
      "learning_rate": 0.00018387554042783376,
      "loss": 11.0607,
      "step": 156550
    },
    {
      "epoch": 18.839951865222623,
      "grad_norm": Infinity,
      "learning_rate": 0.00018387346891035332,
      "loss": 11.1197,
      "step": 156560
    },
    {
      "epoch": 18.84115523465704,
      "grad_norm": Infinity,
      "learning_rate": 0.00018387139727148725,
      "loss": 11.1981,
      "step": 156570
    },
    {
      "epoch": 18.842358604091455,
      "grad_norm": Infinity,
      "learning_rate": 0.00018386932551123857,
      "loss": 11.165,
      "step": 156580
    },
    {
      "epoch": 18.843561973525873,
      "grad_norm": Infinity,
      "learning_rate": 0.00018386725362961025,
      "loss": 11.1773,
      "step": 156590
    },
    {
      "epoch": 18.84476534296029,
      "grad_norm": Infinity,
      "learning_rate": 0.00018386518162660532,
      "loss": 11.1637,
      "step": 156600
    },
    {
      "epoch": 18.845968712394704,
      "grad_norm": Infinity,
      "learning_rate": 0.00018386310950222677,
      "loss": 11.1841,
      "step": 156610
    },
    {
      "epoch": 18.84717208182912,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838610372564776,
      "loss": 11.2402,
      "step": 156620
    },
    {
      "epoch": 18.84837545126354,
      "grad_norm": Infinity,
      "learning_rate": 0.00018385896488936075,
      "loss": 11.1267,
      "step": 156630
    },
    {
      "epoch": 18.849578820697953,
      "grad_norm": Infinity,
      "learning_rate": 0.00018385689240087936,
      "loss": 11.2748,
      "step": 156640
    },
    {
      "epoch": 18.85078219013237,
      "grad_norm": Infinity,
      "learning_rate": 0.00018385481979103628,
      "loss": 11.2116,
      "step": 156650
    },
    {
      "epoch": 18.85198555956679,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838527470598346,
      "loss": 11.1882,
      "step": 156660
    },
    {
      "epoch": 18.853188929001202,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838506742072773,
      "loss": 11.1638,
      "step": 156670
    },
    {
      "epoch": 18.85439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00018384860123336734,
      "loss": 11.208,
      "step": 156680
    },
    {
      "epoch": 18.855595667870038,
      "grad_norm": Infinity,
      "learning_rate": 0.00018384652813810778,
      "loss": 11.1221,
      "step": 156690
    },
    {
      "epoch": 18.85679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00018384445492150158,
      "loss": 11.0525,
      "step": 156700
    },
    {
      "epoch": 18.85800240673887,
      "grad_norm": Infinity,
      "learning_rate": 0.00018384238158355176,
      "loss": 11.1971,
      "step": 156710
    },
    {
      "epoch": 18.859205776173287,
      "grad_norm": Infinity,
      "learning_rate": 0.00018384030812426132,
      "loss": 11.1792,
      "step": 156720
    },
    {
      "epoch": 18.8604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00018383823454363324,
      "loss": 11.1946,
      "step": 156730
    },
    {
      "epoch": 18.86161251504212,
      "grad_norm": Infinity,
      "learning_rate": 0.00018383616084167054,
      "loss": 11.064,
      "step": 156740
    },
    {
      "epoch": 18.862815884476536,
      "grad_norm": Infinity,
      "learning_rate": 0.00018383408701837622,
      "loss": 11.0675,
      "step": 156750
    },
    {
      "epoch": 18.86401925391095,
      "grad_norm": Infinity,
      "learning_rate": 0.00018383201307375327,
      "loss": 11.1382,
      "step": 156760
    },
    {
      "epoch": 18.865222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838299390078047,
      "loss": 11.054,
      "step": 156770
    },
    {
      "epoch": 18.866425992779785,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838278648205335,
      "loss": 11.105,
      "step": 156780
    },
    {
      "epoch": 18.8676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00018382579051194273,
      "loss": 11.1738,
      "step": 156790
    },
    {
      "epoch": 18.868832731648617,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838237160820353,
      "loss": 11.0556,
      "step": 156800
    },
    {
      "epoch": 18.870036101083034,
      "grad_norm": Infinity,
      "learning_rate": 0.00018382164153081426,
      "loss": 11.1819,
      "step": 156810
    },
    {
      "epoch": 18.871239470517448,
      "grad_norm": Infinity,
      "learning_rate": 0.00018381956685828264,
      "loss": 11.1954,
      "step": 156820
    },
    {
      "epoch": 18.872442839951866,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838174920644434,
      "loss": 11.17,
      "step": 156830
    },
    {
      "epoch": 18.87364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.00018381541714929955,
      "loss": 11.2456,
      "step": 156840
    },
    {
      "epoch": 18.874849578820697,
      "grad_norm": Infinity,
      "learning_rate": 0.00018381334211285412,
      "loss": 11.1573,
      "step": 156850
    },
    {
      "epoch": 18.876052948255115,
      "grad_norm": Infinity,
      "learning_rate": 0.00018381126695511004,
      "loss": 11.1414,
      "step": 156860
    },
    {
      "epoch": 18.87725631768953,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838091916760704,
      "loss": 11.2085,
      "step": 156870
    },
    {
      "epoch": 18.878459687123947,
      "grad_norm": Infinity,
      "learning_rate": 0.00018380711627573817,
      "loss": 11.1137,
      "step": 156880
    },
    {
      "epoch": 18.879663056558364,
      "grad_norm": Infinity,
      "learning_rate": 0.00018380504075411633,
      "loss": 11.1565,
      "step": 156890
    },
    {
      "epoch": 18.880866425992778,
      "grad_norm": Infinity,
      "learning_rate": 0.0001838029651112079,
      "loss": 11.1592,
      "step": 156900
    },
    {
      "epoch": 18.882069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00018380088934701592,
      "loss": 11.0634,
      "step": 156910
    },
    {
      "epoch": 18.883273164861613,
      "grad_norm": Infinity,
      "learning_rate": 0.00018379881346154337,
      "loss": 11.1236,
      "step": 156920
    },
    {
      "epoch": 18.884476534296027,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837967374547932,
      "loss": 11.0946,
      "step": 156930
    },
    {
      "epoch": 18.885679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00018379466132676853,
      "loss": 11.1062,
      "step": 156940
    },
    {
      "epoch": 18.886883273164862,
      "grad_norm": Infinity,
      "learning_rate": 0.00018379258507747223,
      "loss": 11.0293,
      "step": 156950
    },
    {
      "epoch": 18.888086642599276,
      "grad_norm": Infinity,
      "learning_rate": 0.00018379050870690738,
      "loss": 11.0878,
      "step": 156960
    },
    {
      "epoch": 18.889290012033694,
      "grad_norm": Infinity,
      "learning_rate": 0.000183788432215077,
      "loss": 10.9764,
      "step": 156970
    },
    {
      "epoch": 18.89049338146811,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837863556019841,
      "loss": 11.0772,
      "step": 156980
    },
    {
      "epoch": 18.891696750902526,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837842788676316,
      "loss": 11.1573,
      "step": 156990
    },
    {
      "epoch": 18.892900120336943,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837822020120226,
      "loss": 11.0387,
      "step": 157000
    },
    {
      "epoch": 18.89410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00018378012503516007,
      "loss": 11.1678,
      "step": 157010
    },
    {
      "epoch": 18.895306859205775,
      "grad_norm": Infinity,
      "learning_rate": 0.000183778047937047,
      "loss": 11.1658,
      "step": 157020
    },
    {
      "epoch": 18.896510228640192,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837759707176864,
      "loss": 11.1512,
      "step": 157030
    },
    {
      "epoch": 18.89771359807461,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837738933770813,
      "loss": 11.0356,
      "step": 157040
    },
    {
      "epoch": 18.898916967509024,
      "grad_norm": Infinity,
      "learning_rate": 0.00018377181591523468,
      "loss": 11.2211,
      "step": 157050
    },
    {
      "epoch": 18.90012033694344,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837697383321496,
      "loss": 11.0747,
      "step": 157060
    },
    {
      "epoch": 18.90132370637786,
      "grad_norm": Infinity,
      "learning_rate": 0.00018376766062782896,
      "loss": 11.206,
      "step": 157070
    },
    {
      "epoch": 18.902527075812273,
      "grad_norm": Infinity,
      "learning_rate": 0.00018376558280227588,
      "loss": 11.1095,
      "step": 157080
    },
    {
      "epoch": 18.90373044524669,
      "grad_norm": Infinity,
      "learning_rate": 0.00018376350485549333,
      "loss": 11.1608,
      "step": 157090
    },
    {
      "epoch": 18.904933814681108,
      "grad_norm": Infinity,
      "learning_rate": 0.00018376142678748426,
      "loss": 11.0815,
      "step": 157100
    },
    {
      "epoch": 18.906137184115522,
      "grad_norm": Infinity,
      "learning_rate": 0.00018375934859825177,
      "loss": 11.0471,
      "step": 157110
    },
    {
      "epoch": 18.90734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837572702877988,
      "loss": 11.2041,
      "step": 157120
    },
    {
      "epoch": 18.908543922984357,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837551918561284,
      "loss": 11.1298,
      "step": 157130
    },
    {
      "epoch": 18.90974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00018375311330324356,
      "loss": 11.1328,
      "step": 157140
    },
    {
      "epoch": 18.91095066185319,
      "grad_norm": Infinity,
      "learning_rate": 0.00018375103462914727,
      "loss": 11.069,
      "step": 157150
    },
    {
      "epoch": 18.912154031287606,
      "grad_norm": Infinity,
      "learning_rate": 0.00018374895583384254,
      "loss": 11.1397,
      "step": 157160
    },
    {
      "epoch": 18.91335740072202,
      "grad_norm": Infinity,
      "learning_rate": 0.00018374687691733244,
      "loss": 11.1763,
      "step": 157170
    },
    {
      "epoch": 18.914560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00018374479787961989,
      "loss": 11.2508,
      "step": 157180
    },
    {
      "epoch": 18.915764139590856,
      "grad_norm": Infinity,
      "learning_rate": 0.00018374271872070795,
      "loss": 11.2083,
      "step": 157190
    },
    {
      "epoch": 18.91696750902527,
      "grad_norm": Infinity,
      "learning_rate": 0.00018374063944059964,
      "loss": 11.1163,
      "step": 157200
    },
    {
      "epoch": 18.918170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00018373856003929794,
      "loss": 11.1841,
      "step": 157210
    },
    {
      "epoch": 18.919374247894105,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837364805168059,
      "loss": 11.0735,
      "step": 157220
    },
    {
      "epoch": 18.92057761732852,
      "grad_norm": Infinity,
      "learning_rate": 0.00018373440087312643,
      "loss": 11.1616,
      "step": 157230
    },
    {
      "epoch": 18.921780986762936,
      "grad_norm": Infinity,
      "learning_rate": 0.00018373232110826266,
      "loss": 11.1764,
      "step": 157240
    },
    {
      "epoch": 18.922984356197354,
      "grad_norm": Infinity,
      "learning_rate": 0.00018373024122221753,
      "loss": 11.16,
      "step": 157250
    },
    {
      "epoch": 18.924187725631768,
      "grad_norm": Infinity,
      "learning_rate": 0.00018372816121499405,
      "loss": 11.0687,
      "step": 157260
    },
    {
      "epoch": 18.925391095066185,
      "grad_norm": Infinity,
      "learning_rate": 0.00018372608108659528,
      "loss": 11.1236,
      "step": 157270
    },
    {
      "epoch": 18.926594464500603,
      "grad_norm": Infinity,
      "learning_rate": 0.00018372400083702418,
      "loss": 11.1659,
      "step": 157280
    },
    {
      "epoch": 18.927797833935017,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837219204662838,
      "loss": 11.0458,
      "step": 157290
    },
    {
      "epoch": 18.929001203369435,
      "grad_norm": Infinity,
      "learning_rate": 0.00018371983997437713,
      "loss": 11.0637,
      "step": 157300
    },
    {
      "epoch": 18.930204572803852,
      "grad_norm": Infinity,
      "learning_rate": 0.00018371775936130717,
      "loss": 11.1876,
      "step": 157310
    },
    {
      "epoch": 18.931407942238266,
      "grad_norm": Infinity,
      "learning_rate": 0.00018371567862707694,
      "loss": 11.2099,
      "step": 157320
    },
    {
      "epoch": 18.932611311672684,
      "grad_norm": Infinity,
      "learning_rate": 0.00018371359777168947,
      "loss": 11.1799,
      "step": 157330
    },
    {
      "epoch": 18.9338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00018371151679514775,
      "loss": 11.0395,
      "step": 157340
    },
    {
      "epoch": 18.935018050541515,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837094356974548,
      "loss": 11.1093,
      "step": 157350
    },
    {
      "epoch": 18.936221419975933,
      "grad_norm": Infinity,
      "learning_rate": 0.0001837073544786136,
      "loss": 11.1713,
      "step": 157360
    },
    {
      "epoch": 18.93742478941035,
      "grad_norm": Infinity,
      "learning_rate": 0.00018370527313862724,
      "loss": 11.2034,
      "step": 157370
    },
    {
      "epoch": 18.938628158844764,
      "grad_norm": Infinity,
      "learning_rate": 0.00018370319167749865,
      "loss": 11.1406,
      "step": 157380
    },
    {
      "epoch": 18.939831528279182,
      "grad_norm": Infinity,
      "learning_rate": 0.00018370111009523087,
      "loss": 11.2147,
      "step": 157390
    },
    {
      "epoch": 18.9410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00018369902839182695,
      "loss": 11.2302,
      "step": 157400
    },
    {
      "epoch": 18.942238267148014,
      "grad_norm": Infinity,
      "learning_rate": 0.00018369694656728984,
      "loss": 11.1735,
      "step": 157410
    },
    {
      "epoch": 18.94344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00018369486462162262,
      "loss": 11.1265,
      "step": 157420
    },
    {
      "epoch": 18.94464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00018369278255482826,
      "loss": 11.1389,
      "step": 157430
    },
    {
      "epoch": 18.945848375451263,
      "grad_norm": Infinity,
      "learning_rate": 0.00018369070036690976,
      "loss": 11.1313,
      "step": 157440
    },
    {
      "epoch": 18.94705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00018368861805787018,
      "loss": 11.151,
      "step": 157450
    },
    {
      "epoch": 18.948255114320098,
      "grad_norm": Infinity,
      "learning_rate": 0.00018368653562771245,
      "loss": 11.1124,
      "step": 157460
    },
    {
      "epoch": 18.949458483754512,
      "grad_norm": Infinity,
      "learning_rate": 0.00018368445307643969,
      "loss": 11.1773,
      "step": 157470
    },
    {
      "epoch": 18.95066185318893,
      "grad_norm": Infinity,
      "learning_rate": 0.00018368237040405485,
      "loss": 11.0857,
      "step": 157480
    },
    {
      "epoch": 18.951865222623347,
      "grad_norm": Infinity,
      "learning_rate": 0.00018368028761056094,
      "loss": 11.0975,
      "step": 157490
    },
    {
      "epoch": 18.95306859205776,
      "grad_norm": Infinity,
      "learning_rate": 0.00018367820469596105,
      "loss": 11.022,
      "step": 157500
    },
    {
      "epoch": 18.95427196149218,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836761216602581,
      "loss": 11.2443,
      "step": 157510
    },
    {
      "epoch": 18.955475330926596,
      "grad_norm": Infinity,
      "learning_rate": 0.00018367403850345516,
      "loss": 11.1917,
      "step": 157520
    },
    {
      "epoch": 18.95667870036101,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836719552255552,
      "loss": 11.022,
      "step": 157530
    },
    {
      "epoch": 18.957882069795428,
      "grad_norm": Infinity,
      "learning_rate": 0.00018366987182656127,
      "loss": 11.2335,
      "step": 157540
    },
    {
      "epoch": 18.959085439229845,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836677883064764,
      "loss": 11.2388,
      "step": 157550
    },
    {
      "epoch": 18.96028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.00018366570466530358,
      "loss": 11.0706,
      "step": 157560
    },
    {
      "epoch": 18.961492178098677,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836636209030458,
      "loss": 11.1347,
      "step": 157570
    },
    {
      "epoch": 18.96269554753309,
      "grad_norm": Infinity,
      "learning_rate": 0.00018366153701970613,
      "loss": 11.1475,
      "step": 157580
    },
    {
      "epoch": 18.96389891696751,
      "grad_norm": Infinity,
      "learning_rate": 0.00018365945301528758,
      "loss": 11.1984,
      "step": 157590
    },
    {
      "epoch": 18.965102286401926,
      "grad_norm": Infinity,
      "learning_rate": 0.00018365736888979314,
      "loss": 11.1857,
      "step": 157600
    },
    {
      "epoch": 18.96630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00018365528464322578,
      "loss": 11.0506,
      "step": 157610
    },
    {
      "epoch": 18.967509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836532002755886,
      "loss": 11.2283,
      "step": 157620
    },
    {
      "epoch": 18.968712394705175,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836511157868846,
      "loss": 11.2945,
      "step": 157630
    },
    {
      "epoch": 18.96991576413959,
      "grad_norm": Infinity,
      "learning_rate": 0.00018364903117711677,
      "loss": 11.0815,
      "step": 157640
    },
    {
      "epoch": 18.971119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.00018364694644628816,
      "loss": 11.174,
      "step": 157650
    },
    {
      "epoch": 18.972322503008424,
      "grad_norm": Infinity,
      "learning_rate": 0.00018364486159440173,
      "loss": 11.0937,
      "step": 157660
    },
    {
      "epoch": 18.97352587244284,
      "grad_norm": Infinity,
      "learning_rate": 0.00018364277662146056,
      "loss": 11.1016,
      "step": 157670
    },
    {
      "epoch": 18.974729241877256,
      "grad_norm": Infinity,
      "learning_rate": 0.00018364069152746764,
      "loss": 11.2318,
      "step": 157680
    },
    {
      "epoch": 18.975932611311674,
      "grad_norm": Infinity,
      "learning_rate": 0.00018363860631242599,
      "loss": 11.0833,
      "step": 157690
    },
    {
      "epoch": 18.977135980746088,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836365209763386,
      "loss": 11.0869,
      "step": 157700
    },
    {
      "epoch": 18.978339350180505,
      "grad_norm": Infinity,
      "learning_rate": 0.00018363443551920857,
      "loss": 11.23,
      "step": 157710
    },
    {
      "epoch": 18.979542719614923,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836323499410388,
      "loss": 11.1504,
      "step": 157720
    },
    {
      "epoch": 18.980746089049337,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836302642418324,
      "loss": 11.1502,
      "step": 157730
    },
    {
      "epoch": 18.981949458483754,
      "grad_norm": Infinity,
      "learning_rate": 0.00018362817842159236,
      "loss": 11.1727,
      "step": 157740
    },
    {
      "epoch": 18.983152827918172,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836260924803217,
      "loss": 11.0221,
      "step": 157750
    },
    {
      "epoch": 18.984356197352586,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836240064180234,
      "loss": 11.1395,
      "step": 157760
    },
    {
      "epoch": 18.985559566787003,
      "grad_norm": Infinity,
      "learning_rate": 0.00018362192023470058,
      "loss": 11.0751,
      "step": 157770
    },
    {
      "epoch": 18.98676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00018361983393035618,
      "loss": 11.1112,
      "step": 157780
    },
    {
      "epoch": 18.987966305655835,
      "grad_norm": Infinity,
      "learning_rate": 0.00018361774750499322,
      "loss": 11.0373,
      "step": 157790
    },
    {
      "epoch": 18.989169675090253,
      "grad_norm": Infinity,
      "learning_rate": 0.00018361566095861473,
      "loss": 11.0098,
      "step": 157800
    },
    {
      "epoch": 18.99037304452467,
      "grad_norm": Infinity,
      "learning_rate": 0.00018361357429122377,
      "loss": 11.1946,
      "step": 157810
    },
    {
      "epoch": 18.991576413959084,
      "grad_norm": Infinity,
      "learning_rate": 0.0001836114875028233,
      "loss": 11.1975,
      "step": 157820
    },
    {
      "epoch": 18.9927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00018360940059341635,
      "loss": 11.2993,
      "step": 157830
    },
    {
      "epoch": 18.99398315282792,
      "grad_norm": Infinity,
      "learning_rate": 0.00018360731356300596,
      "loss": 11.138,
      "step": 157840
    },
    {
      "epoch": 18.995186522262333,
      "grad_norm": Infinity,
      "learning_rate": 0.00018360522641159516,
      "loss": 11.2325,
      "step": 157850
    },
    {
      "epoch": 18.99638989169675,
      "grad_norm": Infinity,
      "learning_rate": 0.00018360313913918694,
      "loss": 11.1623,
      "step": 157860
    },
    {
      "epoch": 18.99759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00018360105174578437,
      "loss": 11.0993,
      "step": 157870
    },
    {
      "epoch": 18.998796630565582,
      "grad_norm": Infinity,
      "learning_rate": 0.0001835989642313904,
      "loss": 11.1284,
      "step": 157880
    },
    {
      "epoch": 19.0,
      "grad_norm": Infinity,
      "learning_rate": 0.0001835968765960081,
      "loss": 11.1654,
      "step": 157890
    },
    {
      "epoch": 19.0,
      "eval_loss": 11.139603614807129,
      "eval_runtime": 118.4351,
      "eval_samples_per_second": 62.372,
      "eval_steps_per_second": 7.802,
      "step": 157890
    },
    {
      "epoch": 19.001203369434418,
      "grad_norm": Infinity,
      "learning_rate": 0.0001835947888396405,
      "loss": 11.0563,
      "step": 157900
    },
    {
      "epoch": 19.00240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.00018359270096229057,
      "loss": 11.146,
      "step": 157910
    },
    {
      "epoch": 19.00361010830325,
      "grad_norm": Infinity,
      "learning_rate": 0.00018359061296396138,
      "loss": 11.1602,
      "step": 157920
    },
    {
      "epoch": 19.004813477737667,
      "grad_norm": Infinity,
      "learning_rate": 0.00018358852484465593,
      "loss": 11.0101,
      "step": 157930
    },
    {
      "epoch": 19.00601684717208,
      "grad_norm": Infinity,
      "learning_rate": 0.00018358643660437723,
      "loss": 11.1811,
      "step": 157940
    },
    {
      "epoch": 19.0072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00018358434824312835,
      "loss": 11.19,
      "step": 157950
    },
    {
      "epoch": 19.008423586040916,
      "grad_norm": Infinity,
      "learning_rate": 0.00018358225976091228,
      "loss": 11.0729,
      "step": 157960
    },
    {
      "epoch": 19.00962695547533,
      "grad_norm": Infinity,
      "learning_rate": 0.00018358017115773203,
      "loss": 11.1656,
      "step": 157970
    },
    {
      "epoch": 19.010830324909747,
      "grad_norm": Infinity,
      "learning_rate": 0.00018357808243359066,
      "loss": 11.1126,
      "step": 157980
    },
    {
      "epoch": 19.012033694344165,
      "grad_norm": Infinity,
      "learning_rate": 0.00018357599358849114,
      "loss": 11.1051,
      "step": 157990
    },
    {
      "epoch": 19.01323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00018357390462243653,
      "loss": 11.0959,
      "step": 158000
    },
    {
      "epoch": 19.014440433212997,
      "grad_norm": Infinity,
      "learning_rate": 0.00018357181553542985,
      "loss": 11.2075,
      "step": 158010
    },
    {
      "epoch": 19.015643802647414,
      "grad_norm": Infinity,
      "learning_rate": 0.00018356972632747412,
      "loss": 11.1011,
      "step": 158020
    },
    {
      "epoch": 19.016847172081828,
      "grad_norm": Infinity,
      "learning_rate": 0.0001835676369985724,
      "loss": 11.1666,
      "step": 158030
    },
    {
      "epoch": 19.018050541516246,
      "grad_norm": Infinity,
      "learning_rate": 0.0001835655475487276,
      "loss": 11.2899,
      "step": 158040
    },
    {
      "epoch": 19.019253910950663,
      "grad_norm": Infinity,
      "learning_rate": 0.00018356345797794287,
      "loss": 11.1406,
      "step": 158050
    },
    {
      "epoch": 19.020457280385077,
      "grad_norm": Infinity,
      "learning_rate": 0.00018356136828622118,
      "loss": 11.2048,
      "step": 158060
    },
    {
      "epoch": 19.021660649819495,
      "grad_norm": Infinity,
      "learning_rate": 0.00018355927847356557,
      "loss": 11.1747,
      "step": 158070
    },
    {
      "epoch": 19.022864019253912,
      "grad_norm": Infinity,
      "learning_rate": 0.00018355718853997903,
      "loss": 11.2221,
      "step": 158080
    },
    {
      "epoch": 19.024067388688326,
      "grad_norm": Infinity,
      "learning_rate": 0.00018355509848546464,
      "loss": 11.189,
      "step": 158090
    },
    {
      "epoch": 19.025270758122744,
      "grad_norm": Infinity,
      "learning_rate": 0.00018355300831002537,
      "loss": 11.1854,
      "step": 158100
    },
    {
      "epoch": 19.02647412755716,
      "grad_norm": Infinity,
      "learning_rate": 0.00018355091801366426,
      "loss": 11.2108,
      "step": 158110
    },
    {
      "epoch": 19.027677496991576,
      "grad_norm": Infinity,
      "learning_rate": 0.00018354882759638437,
      "loss": 11.1077,
      "step": 158120
    },
    {
      "epoch": 19.028880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.0001835467370581887,
      "loss": 11.1843,
      "step": 158130
    },
    {
      "epoch": 19.03008423586041,
      "grad_norm": Infinity,
      "learning_rate": 0.00018354464639908023,
      "loss": 11.0339,
      "step": 158140
    },
    {
      "epoch": 19.031287605294825,
      "grad_norm": Infinity,
      "learning_rate": 0.00018354255561906207,
      "loss": 11.1135,
      "step": 158150
    },
    {
      "epoch": 19.032490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00018354046471813718,
      "loss": 11.1554,
      "step": 158160
    },
    {
      "epoch": 19.03369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.00018353837369630864,
      "loss": 11.0672,
      "step": 158170
    },
    {
      "epoch": 19.034897713598074,
      "grad_norm": Infinity,
      "learning_rate": 0.00018353628255357945,
      "loss": 11.1759,
      "step": 158180
    },
    {
      "epoch": 19.03610108303249,
      "grad_norm": Infinity,
      "learning_rate": 0.00018353419128995258,
      "loss": 11.0807,
      "step": 158190
    },
    {
      "epoch": 19.03730445246691,
      "grad_norm": Infinity,
      "learning_rate": 0.00018353209990543115,
      "loss": 11.143,
      "step": 158200
    },
    {
      "epoch": 19.038507821901323,
      "grad_norm": Infinity,
      "learning_rate": 0.00018353000840001818,
      "loss": 11.082,
      "step": 158210
    },
    {
      "epoch": 19.03971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.00018352791677371663,
      "loss": 11.1706,
      "step": 158220
    },
    {
      "epoch": 19.040914560770158,
      "grad_norm": Infinity,
      "learning_rate": 0.00018352582502652957,
      "loss": 11.036,
      "step": 158230
    },
    {
      "epoch": 19.042117930204572,
      "grad_norm": Infinity,
      "learning_rate": 0.00018352373315846002,
      "loss": 11.0549,
      "step": 158240
    },
    {
      "epoch": 19.04332129963899,
      "grad_norm": Infinity,
      "learning_rate": 0.00018352164116951098,
      "loss": 11.2535,
      "step": 158250
    },
    {
      "epoch": 19.044524669073404,
      "grad_norm": Infinity,
      "learning_rate": 0.00018351954905968554,
      "loss": 11.1804,
      "step": 158260
    },
    {
      "epoch": 19.04572803850782,
      "grad_norm": Infinity,
      "learning_rate": 0.00018351745682898668,
      "loss": 11.1223,
      "step": 158270
    },
    {
      "epoch": 19.04693140794224,
      "grad_norm": Infinity,
      "learning_rate": 0.00018351536447741744,
      "loss": 11.0699,
      "step": 158280
    },
    {
      "epoch": 19.048134777376653,
      "grad_norm": Infinity,
      "learning_rate": 0.00018351327200498085,
      "loss": 11.1097,
      "step": 158290
    },
    {
      "epoch": 19.04933814681107,
      "grad_norm": Infinity,
      "learning_rate": 0.00018351117941167993,
      "loss": 11.1484,
      "step": 158300
    },
    {
      "epoch": 19.050541516245488,
      "grad_norm": Infinity,
      "learning_rate": 0.00018350908669751773,
      "loss": 11.1152,
      "step": 158310
    },
    {
      "epoch": 19.051744885679902,
      "grad_norm": Infinity,
      "learning_rate": 0.00018350699386249724,
      "loss": 11.0663,
      "step": 158320
    },
    {
      "epoch": 19.05294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00018350490090662152,
      "loss": 11.0921,
      "step": 158330
    },
    {
      "epoch": 19.054151624548737,
      "grad_norm": Infinity,
      "learning_rate": 0.0001835028078298936,
      "loss": 11.1356,
      "step": 158340
    },
    {
      "epoch": 19.05535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.0001835007146323165,
      "loss": 11.1691,
      "step": 158350
    },
    {
      "epoch": 19.05655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00018349862131389325,
      "loss": 11.2317,
      "step": 158360
    },
    {
      "epoch": 19.057761732851986,
      "grad_norm": Infinity,
      "learning_rate": 0.00018349652787462688,
      "loss": 11.0429,
      "step": 158370
    },
    {
      "epoch": 19.0589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834944343145204,
      "loss": 11.1741,
      "step": 158380
    },
    {
      "epoch": 19.060168471720818,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834923406335769,
      "loss": 11.1619,
      "step": 158390
    },
    {
      "epoch": 19.061371841155236,
      "grad_norm": Infinity,
      "learning_rate": 0.00018349024683179935,
      "loss": 11.2042,
      "step": 158400
    },
    {
      "epoch": 19.06257521058965,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834881529091908,
      "loss": 11.1238,
      "step": 158410
    },
    {
      "epoch": 19.063778580024067,
      "grad_norm": Infinity,
      "learning_rate": 0.00018348605886575427,
      "loss": 11.0707,
      "step": 158420
    },
    {
      "epoch": 19.064981949458485,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834839647014928,
      "loss": 11.2027,
      "step": 158430
    },
    {
      "epoch": 19.0661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00018348187041640943,
      "loss": 11.1844,
      "step": 158440
    },
    {
      "epoch": 19.067388688327316,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834797760105072,
      "loss": 11.1456,
      "step": 158450
    },
    {
      "epoch": 19.068592057761734,
      "grad_norm": Infinity,
      "learning_rate": 0.00018347768148378908,
      "loss": 11.0727,
      "step": 158460
    },
    {
      "epoch": 19.069795427196148,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834755868362582,
      "loss": 11.1388,
      "step": 158470
    },
    {
      "epoch": 19.070998796630565,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834734920679175,
      "loss": 11.1525,
      "step": 158480
    },
    {
      "epoch": 19.072202166064983,
      "grad_norm": Infinity,
      "learning_rate": 0.00018347139717877004,
      "loss": 11.1697,
      "step": 158490
    },
    {
      "epoch": 19.073405535499397,
      "grad_norm": Infinity,
      "learning_rate": 0.00018346930216881887,
      "loss": 11.2195,
      "step": 158500
    },
    {
      "epoch": 19.074608904933815,
      "grad_norm": Infinity,
      "learning_rate": 0.000183467207038067,
      "loss": 11.0772,
      "step": 158510
    },
    {
      "epoch": 19.075812274368232,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834651117865175,
      "loss": 11.2047,
      "step": 158520
    },
    {
      "epoch": 19.077015643802646,
      "grad_norm": Infinity,
      "learning_rate": 0.00018346301641417338,
      "loss": 11.0359,
      "step": 158530
    },
    {
      "epoch": 19.078219013237064,
      "grad_norm": Infinity,
      "learning_rate": 0.00018346092092103765,
      "loss": 11.1607,
      "step": 158540
    },
    {
      "epoch": 19.07942238267148,
      "grad_norm": Infinity,
      "learning_rate": 0.00018345882530711334,
      "loss": 11.087,
      "step": 158550
    },
    {
      "epoch": 19.080625752105895,
      "grad_norm": Infinity,
      "learning_rate": 0.00018345672957240358,
      "loss": 10.9896,
      "step": 158560
    },
    {
      "epoch": 19.081829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.00018345463371691127,
      "loss": 11.1302,
      "step": 158570
    },
    {
      "epoch": 19.08303249097473,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834525377406395,
      "loss": 11.1366,
      "step": 158580
    },
    {
      "epoch": 19.084235860409144,
      "grad_norm": Infinity,
      "learning_rate": 0.00018345044164359134,
      "loss": 11.0495,
      "step": 158590
    },
    {
      "epoch": 19.085439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00018344834542576973,
      "loss": 11.1212,
      "step": 158600
    },
    {
      "epoch": 19.08664259927798,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834462490871778,
      "loss": 11.2575,
      "step": 158610
    },
    {
      "epoch": 19.087845968712394,
      "grad_norm": Infinity,
      "learning_rate": 0.00018344415262781856,
      "loss": 11.1387,
      "step": 158620
    },
    {
      "epoch": 19.08904933814681,
      "grad_norm": Infinity,
      "learning_rate": 0.000183442056047695,
      "loss": 11.0817,
      "step": 158630
    },
    {
      "epoch": 19.09025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.00018343995934681017,
      "loss": 11.0478,
      "step": 158640
    },
    {
      "epoch": 19.091456077015643,
      "grad_norm": Infinity,
      "learning_rate": 0.00018343786252516715,
      "loss": 11.1995,
      "step": 158650
    },
    {
      "epoch": 19.09265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00018343576558276894,
      "loss": 11.1377,
      "step": 158660
    },
    {
      "epoch": 19.093862815884478,
      "grad_norm": Infinity,
      "learning_rate": 0.00018343366851961856,
      "loss": 11.1158,
      "step": 158670
    },
    {
      "epoch": 19.095066185318892,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834315713357191,
      "loss": 11.1496,
      "step": 158680
    },
    {
      "epoch": 19.09626955475331,
      "grad_norm": Infinity,
      "learning_rate": 0.00018342947403107352,
      "loss": 11.0106,
      "step": 158690
    },
    {
      "epoch": 19.097472924187727,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834273766056849,
      "loss": 11.1551,
      "step": 158700
    },
    {
      "epoch": 19.09867629362214,
      "grad_norm": Infinity,
      "learning_rate": 0.00018342527905955628,
      "loss": 11.1511,
      "step": 158710
    },
    {
      "epoch": 19.09987966305656,
      "grad_norm": Infinity,
      "learning_rate": 0.00018342318139269066,
      "loss": 11.2416,
      "step": 158720
    },
    {
      "epoch": 19.101083032490976,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834210836050911,
      "loss": 11.0608,
      "step": 158730
    },
    {
      "epoch": 19.10228640192539,
      "grad_norm": Infinity,
      "learning_rate": 0.00018341898569676065,
      "loss": 11.0706,
      "step": 158740
    },
    {
      "epoch": 19.103489771359808,
      "grad_norm": Infinity,
      "learning_rate": 0.00018341688766770236,
      "loss": 11.2039,
      "step": 158750
    },
    {
      "epoch": 19.104693140794225,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834147895179192,
      "loss": 11.127,
      "step": 158760
    },
    {
      "epoch": 19.10589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.00018341269124741427,
      "loss": 11.191,
      "step": 158770
    },
    {
      "epoch": 19.107099879663057,
      "grad_norm": Infinity,
      "learning_rate": 0.00018341059285619058,
      "loss": 11.231,
      "step": 158780
    },
    {
      "epoch": 19.108303249097474,
      "grad_norm": Infinity,
      "learning_rate": 0.00018340849434425117,
      "loss": 11.1613,
      "step": 158790
    },
    {
      "epoch": 19.10950661853189,
      "grad_norm": Infinity,
      "learning_rate": 0.00018340639571159907,
      "loss": 11.0705,
      "step": 158800
    },
    {
      "epoch": 19.110709987966306,
      "grad_norm": Infinity,
      "learning_rate": 0.0001834042969582373,
      "loss": 11.1376,
      "step": 158810
    },
    {
      "epoch": 19.111913357400724,
      "grad_norm": Infinity,
      "learning_rate": 0.00018340219808416893,
      "loss": 11.2207,
      "step": 158820
    },
    {
      "epoch": 19.113116726835138,
      "grad_norm": Infinity,
      "learning_rate": 0.000183400099089397,
      "loss": 11.2688,
      "step": 158830
    },
    {
      "epoch": 19.114320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.00018339799997392454,
      "loss": 11.2192,
      "step": 158840
    },
    {
      "epoch": 19.115523465703973,
      "grad_norm": Infinity,
      "learning_rate": 0.00018339590073775455,
      "loss": 11.1115,
      "step": 158850
    },
    {
      "epoch": 19.116726835138387,
      "grad_norm": Infinity,
      "learning_rate": 0.00018339380138089015,
      "loss": 11.1967,
      "step": 158860
    },
    {
      "epoch": 19.117930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833917019033343,
      "loss": 11.1094,
      "step": 158870
    },
    {
      "epoch": 19.119133574007222,
      "grad_norm": Infinity,
      "learning_rate": 0.00018338960230509006,
      "loss": 11.2252,
      "step": 158880
    },
    {
      "epoch": 19.120336943441636,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833875025861605,
      "loss": 11.1617,
      "step": 158890
    },
    {
      "epoch": 19.121540312876053,
      "grad_norm": Infinity,
      "learning_rate": 0.00018338540274654864,
      "loss": 11.1357,
      "step": 158900
    },
    {
      "epoch": 19.12274368231047,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833833027862575,
      "loss": 11.2059,
      "step": 158910
    },
    {
      "epoch": 19.123947051744885,
      "grad_norm": Infinity,
      "learning_rate": 0.00018338120270529015,
      "loss": 11.2334,
      "step": 158920
    },
    {
      "epoch": 19.125150421179303,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833791025036496,
      "loss": 11.1384,
      "step": 158930
    },
    {
      "epoch": 19.126353790613717,
      "grad_norm": Infinity,
      "learning_rate": 0.00018337700218133888,
      "loss": 11.2042,
      "step": 158940
    },
    {
      "epoch": 19.127557160048134,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833749017383611,
      "loss": 11.139,
      "step": 158950
    },
    {
      "epoch": 19.128760529482552,
      "grad_norm": Infinity,
      "learning_rate": 0.00018337280117471924,
      "loss": 11.1242,
      "step": 158960
    },
    {
      "epoch": 19.129963898916966,
      "grad_norm": Infinity,
      "learning_rate": 0.00018337070049041632,
      "loss": 11.1422,
      "step": 158970
    },
    {
      "epoch": 19.131167268351383,
      "grad_norm": Infinity,
      "learning_rate": 0.00018336859968545545,
      "loss": 11.1218,
      "step": 158980
    },
    {
      "epoch": 19.1323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00018336649875983963,
      "loss": 11.2146,
      "step": 158990
    },
    {
      "epoch": 19.133574007220215,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833643977135719,
      "loss": 11.1259,
      "step": 159000
    },
    {
      "epoch": 19.134777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.00018336229654665527,
      "loss": 11.1867,
      "step": 159010
    },
    {
      "epoch": 19.13598074608905,
      "grad_norm": Infinity,
      "learning_rate": 0.00018336019525909284,
      "loss": 11.1998,
      "step": 159020
    },
    {
      "epoch": 19.137184115523464,
      "grad_norm": Infinity,
      "learning_rate": 0.00018335809385088765,
      "loss": 11.0415,
      "step": 159030
    },
    {
      "epoch": 19.13838748495788,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833559923220427,
      "loss": 11.1145,
      "step": 159040
    },
    {
      "epoch": 19.1395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00018335389067256102,
      "loss": 11.0816,
      "step": 159050
    },
    {
      "epoch": 19.140794223826713,
      "grad_norm": Infinity,
      "learning_rate": 0.00018335178890244573,
      "loss": 11.0793,
      "step": 159060
    },
    {
      "epoch": 19.14199759326113,
      "grad_norm": Infinity,
      "learning_rate": 0.00018334968701169978,
      "loss": 11.3079,
      "step": 159070
    },
    {
      "epoch": 19.14320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833475850003263,
      "loss": 11.0181,
      "step": 159080
    },
    {
      "epoch": 19.144404332129962,
      "grad_norm": Infinity,
      "learning_rate": 0.00018334548286832824,
      "loss": 11.1338,
      "step": 159090
    },
    {
      "epoch": 19.14560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00018334338061570872,
      "loss": 11.1068,
      "step": 159100
    },
    {
      "epoch": 19.146811070998798,
      "grad_norm": Infinity,
      "learning_rate": 0.00018334127824247074,
      "loss": 11.1185,
      "step": 159110
    },
    {
      "epoch": 19.14801444043321,
      "grad_norm": Infinity,
      "learning_rate": 0.00018333917574861737,
      "loss": 11.2902,
      "step": 159120
    },
    {
      "epoch": 19.14921780986763,
      "grad_norm": Infinity,
      "learning_rate": 0.00018333707313415162,
      "loss": 11.1842,
      "step": 159130
    },
    {
      "epoch": 19.150421179302047,
      "grad_norm": Infinity,
      "learning_rate": 0.00018333497039907655,
      "loss": 11.1929,
      "step": 159140
    },
    {
      "epoch": 19.15162454873646,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833328675433952,
      "loss": 11.0934,
      "step": 159150
    },
    {
      "epoch": 19.15282791817088,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833307645671106,
      "loss": 11.0951,
      "step": 159160
    },
    {
      "epoch": 19.154031287605296,
      "grad_norm": Infinity,
      "learning_rate": 0.00018332866147022586,
      "loss": 11.0999,
      "step": 159170
    },
    {
      "epoch": 19.15523465703971,
      "grad_norm": Infinity,
      "learning_rate": 0.00018332655825274393,
      "loss": 11.2071,
      "step": 159180
    },
    {
      "epoch": 19.156438026474127,
      "grad_norm": Infinity,
      "learning_rate": 0.00018332445491466793,
      "loss": 11.1583,
      "step": 159190
    },
    {
      "epoch": 19.157641395908545,
      "grad_norm": Infinity,
      "learning_rate": 0.00018332235145600084,
      "loss": 11.0031,
      "step": 159200
    },
    {
      "epoch": 19.15884476534296,
      "grad_norm": Infinity,
      "learning_rate": 0.00018332024787674577,
      "loss": 11.1038,
      "step": 159210
    },
    {
      "epoch": 19.160048134777377,
      "grad_norm": Infinity,
      "learning_rate": 0.00018331814417690568,
      "loss": 11.1046,
      "step": 159220
    },
    {
      "epoch": 19.161251504211794,
      "grad_norm": Infinity,
      "learning_rate": 0.00018331604035648372,
      "loss": 11.1936,
      "step": 159230
    },
    {
      "epoch": 19.162454873646208,
      "grad_norm": Infinity,
      "learning_rate": 0.00018331393641548283,
      "loss": 11.0831,
      "step": 159240
    },
    {
      "epoch": 19.163658243080626,
      "grad_norm": Infinity,
      "learning_rate": 0.00018331183235390614,
      "loss": 11.1405,
      "step": 159250
    },
    {
      "epoch": 19.164861612515043,
      "grad_norm": Infinity,
      "learning_rate": 0.00018330972817175663,
      "loss": 11.1274,
      "step": 159260
    },
    {
      "epoch": 19.166064981949457,
      "grad_norm": Infinity,
      "learning_rate": 0.00018330762386903742,
      "loss": 11.1068,
      "step": 159270
    },
    {
      "epoch": 19.167268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.00018330551944575148,
      "loss": 11.1133,
      "step": 159280
    },
    {
      "epoch": 19.168471720818292,
      "grad_norm": Infinity,
      "learning_rate": 0.0001833034149019019,
      "loss": 11.2081,
      "step": 159290
    },
    {
      "epoch": 19.169675090252706,
      "grad_norm": Infinity,
      "learning_rate": 0.00018330131023749167,
      "loss": 11.1845,
      "step": 159300
    },
    {
      "epoch": 19.170878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00018329920545252394,
      "loss": 11.1335,
      "step": 159310
    },
    {
      "epoch": 19.17208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00018329710054700168,
      "loss": 11.0865,
      "step": 159320
    },
    {
      "epoch": 19.173285198555956,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832949955209279,
      "loss": 11.1383,
      "step": 159330
    },
    {
      "epoch": 19.174488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.00018329289037430572,
      "loss": 11.0965,
      "step": 159340
    },
    {
      "epoch": 19.17569193742479,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832907851071382,
      "loss": 11.2015,
      "step": 159350
    },
    {
      "epoch": 19.176895306859205,
      "grad_norm": Infinity,
      "learning_rate": 0.00018328867971942832,
      "loss": 11.237,
      "step": 159360
    },
    {
      "epoch": 19.178098676293622,
      "grad_norm": Infinity,
      "learning_rate": 0.00018328657421117915,
      "loss": 11.2032,
      "step": 159370
    },
    {
      "epoch": 19.17930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.00018328446858239375,
      "loss": 11.1951,
      "step": 159380
    },
    {
      "epoch": 19.180505415162454,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832823628330752,
      "loss": 11.0599,
      "step": 159390
    },
    {
      "epoch": 19.18170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00018328025696322646,
      "loss": 11.0815,
      "step": 159400
    },
    {
      "epoch": 19.18291215403129,
      "grad_norm": Infinity,
      "learning_rate": 0.00018327815097285064,
      "loss": 11.1372,
      "step": 159410
    },
    {
      "epoch": 19.184115523465703,
      "grad_norm": Infinity,
      "learning_rate": 0.00018327604486195078,
      "loss": 11.1505,
      "step": 159420
    },
    {
      "epoch": 19.18531889290012,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832739386305299,
      "loss": 11.233,
      "step": 159430
    },
    {
      "epoch": 19.186522262334538,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832718322785911,
      "loss": 11.1522,
      "step": 159440
    },
    {
      "epoch": 19.187725631768952,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832697258061374,
      "loss": 11.0415,
      "step": 159450
    },
    {
      "epoch": 19.18892900120337,
      "grad_norm": Infinity,
      "learning_rate": 0.00018326761921317185,
      "loss": 11.1992,
      "step": 159460
    },
    {
      "epoch": 19.190132370637787,
      "grad_norm": Infinity,
      "learning_rate": 0.00018326551249969747,
      "loss": 11.134,
      "step": 159470
    },
    {
      "epoch": 19.1913357400722,
      "grad_norm": Infinity,
      "learning_rate": 0.00018326340566571735,
      "loss": 11.0484,
      "step": 159480
    },
    {
      "epoch": 19.19253910950662,
      "grad_norm": Infinity,
      "learning_rate": 0.00018326129871123453,
      "loss": 11.146,
      "step": 159490
    },
    {
      "epoch": 19.193742478941036,
      "grad_norm": Infinity,
      "learning_rate": 0.00018325919163625205,
      "loss": 11.1345,
      "step": 159500
    },
    {
      "epoch": 19.19494584837545,
      "grad_norm": Infinity,
      "learning_rate": 0.00018325708444077296,
      "loss": 11.1604,
      "step": 159510
    },
    {
      "epoch": 19.196149217809868,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832549771248003,
      "loss": 11.1589,
      "step": 159520
    },
    {
      "epoch": 19.197352587244286,
      "grad_norm": Infinity,
      "learning_rate": 0.00018325286968833717,
      "loss": 11.1412,
      "step": 159530
    },
    {
      "epoch": 19.1985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00018325076213138657,
      "loss": 11.1329,
      "step": 159540
    },
    {
      "epoch": 19.199759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00018324865445395157,
      "loss": 11.174,
      "step": 159550
    },
    {
      "epoch": 19.200962695547535,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832465466560352,
      "loss": 11.0711,
      "step": 159560
    },
    {
      "epoch": 19.20216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832444387376405,
      "loss": 10.9949,
      "step": 159570
    },
    {
      "epoch": 19.203369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832423306987706,
      "loss": 11.13,
      "step": 159580
    },
    {
      "epoch": 19.204572803850784,
      "grad_norm": Infinity,
      "learning_rate": 0.00018324022253942844,
      "loss": 11.0963,
      "step": 159590
    },
    {
      "epoch": 19.205776173285198,
      "grad_norm": Infinity,
      "learning_rate": 0.00018323811425961716,
      "loss": 11.1516,
      "step": 159600
    },
    {
      "epoch": 19.206979542719615,
      "grad_norm": Infinity,
      "learning_rate": 0.00018323600585933977,
      "loss": 11.2114,
      "step": 159610
    },
    {
      "epoch": 19.20818291215403,
      "grad_norm": Infinity,
      "learning_rate": 0.00018323389733859933,
      "loss": 11.1687,
      "step": 159620
    },
    {
      "epoch": 19.209386281588447,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832317886973989,
      "loss": 11.151,
      "step": 159630
    },
    {
      "epoch": 19.210589651022865,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832296799357415,
      "loss": 11.2609,
      "step": 159640
    },
    {
      "epoch": 19.21179302045728,
      "grad_norm": Infinity,
      "learning_rate": 0.00018322757105363022,
      "loss": 11.2683,
      "step": 159650
    },
    {
      "epoch": 19.212996389891696,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832254620510681,
      "loss": 11.1292,
      "step": 159660
    },
    {
      "epoch": 19.214199759326114,
      "grad_norm": Infinity,
      "learning_rate": 0.00018322335292805816,
      "loss": 11.1581,
      "step": 159670
    },
    {
      "epoch": 19.215403128760528,
      "grad_norm": Infinity,
      "learning_rate": 0.00018322124368460351,
      "loss": 11.1436,
      "step": 159680
    },
    {
      "epoch": 19.216606498194945,
      "grad_norm": Infinity,
      "learning_rate": 0.00018321913432070715,
      "loss": 11.1155,
      "step": 159690
    },
    {
      "epoch": 19.217809867629363,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832170248363722,
      "loss": 11.2039,
      "step": 159700
    },
    {
      "epoch": 19.219013237063777,
      "grad_norm": Infinity,
      "learning_rate": 0.00018321491523160164,
      "loss": 11.1006,
      "step": 159710
    },
    {
      "epoch": 19.220216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.00018321280550639854,
      "loss": 11.2315,
      "step": 159720
    },
    {
      "epoch": 19.221419975932612,
      "grad_norm": Infinity,
      "learning_rate": 0.00018321069566076597,
      "loss": 11.1251,
      "step": 159730
    },
    {
      "epoch": 19.222623345367026,
      "grad_norm": Infinity,
      "learning_rate": 0.000183208585694707,
      "loss": 11.0394,
      "step": 159740
    },
    {
      "epoch": 19.223826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.00018320647560822467,
      "loss": 11.1559,
      "step": 159750
    },
    {
      "epoch": 19.22503008423586,
      "grad_norm": Infinity,
      "learning_rate": 0.000183204365401322,
      "loss": 11.1828,
      "step": 159760
    },
    {
      "epoch": 19.226233453670275,
      "grad_norm": Infinity,
      "learning_rate": 0.0001832022550740021,
      "loss": 11.2003,
      "step": 159770
    },
    {
      "epoch": 19.227436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00018320014462626797,
      "loss": 11.2145,
      "step": 159780
    },
    {
      "epoch": 19.22864019253911,
      "grad_norm": Infinity,
      "learning_rate": 0.0001831980340581227,
      "loss": 11.1368,
      "step": 159790
    },
    {
      "epoch": 19.229843561973524,
      "grad_norm": Infinity,
      "learning_rate": 0.00018319592336956932,
      "loss": 11.0966,
      "step": 159800
    },
    {
      "epoch": 19.231046931407942,
      "grad_norm": Infinity,
      "learning_rate": 0.00018319381256061093,
      "loss": 11.1462,
      "step": 159810
    },
    {
      "epoch": 19.23225030084236,
      "grad_norm": Infinity,
      "learning_rate": 0.0001831917016312505,
      "loss": 11.1368,
      "step": 159820
    },
    {
      "epoch": 19.233453670276774,
      "grad_norm": Infinity,
      "learning_rate": 0.00018318959058149118,
      "loss": 11.1017,
      "step": 159830
    },
    {
      "epoch": 19.23465703971119,
      "grad_norm": Infinity,
      "learning_rate": 0.00018318747941133595,
      "loss": 11.0607,
      "step": 159840
    },
    {
      "epoch": 19.23586040914561,
      "grad_norm": Infinity,
      "learning_rate": 0.00018318536812078794,
      "loss": 11.2353,
      "step": 159850
    },
    {
      "epoch": 19.237063778580023,
      "grad_norm": Infinity,
      "learning_rate": 0.00018318325670985014,
      "loss": 11.2015,
      "step": 159860
    },
    {
      "epoch": 19.23826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.00018318114517852565,
      "loss": 11.1767,
      "step": 159870
    },
    {
      "epoch": 19.239470517448858,
      "grad_norm": Infinity,
      "learning_rate": 0.0001831790335268175,
      "loss": 11.1298,
      "step": 159880
    },
    {
      "epoch": 19.240673886883272,
      "grad_norm": Infinity,
      "learning_rate": 0.00018317692175472873,
      "loss": 11.2561,
      "step": 159890
    },
    {
      "epoch": 19.24187725631769,
      "grad_norm": Infinity,
      "learning_rate": 0.00018317480986226246,
      "loss": 11.0442,
      "step": 159900
    },
    {
      "epoch": 19.243080625752107,
      "grad_norm": Infinity,
      "learning_rate": 0.00018317269784942167,
      "loss": 11.143,
      "step": 159910
    },
    {
      "epoch": 19.24428399518652,
      "grad_norm": Infinity,
      "learning_rate": 0.00018317058571620947,
      "loss": 11.0817,
      "step": 159920
    },
    {
      "epoch": 19.24548736462094,
      "grad_norm": Infinity,
      "learning_rate": 0.00018316847346262888,
      "loss": 11.2806,
      "step": 159930
    },
    {
      "epoch": 19.246690734055356,
      "grad_norm": Infinity,
      "learning_rate": 0.00018316636108868298,
      "loss": 11.1808,
      "step": 159940
    },
    {
      "epoch": 19.24789410348977,
      "grad_norm": Infinity,
      "learning_rate": 0.00018316424859437483,
      "loss": 11.1272,
      "step": 159950
    },
    {
      "epoch": 19.249097472924188,
      "grad_norm": Infinity,
      "learning_rate": 0.00018316213597970745,
      "loss": 11.094,
      "step": 159960
    },
    {
      "epoch": 19.250300842358605,
      "grad_norm": Infinity,
      "learning_rate": 0.00018316002324468396,
      "loss": 11.2003,
      "step": 159970
    },
    {
      "epoch": 19.25150421179302,
      "grad_norm": Infinity,
      "learning_rate": 0.00018315791038930738,
      "loss": 11.1593,
      "step": 159980
    },
    {
      "epoch": 19.252707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00018315579741358078,
      "loss": 11.0211,
      "step": 159990
    },
    {
      "epoch": 19.253910950661854,
      "grad_norm": Infinity,
      "learning_rate": 0.0001831536843175072,
      "loss": 11.2423,
      "step": 160000
    },
    {
      "epoch": 19.25511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.0001831515711010897,
      "loss": 11.1407,
      "step": 160010
    },
    {
      "epoch": 19.256317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00018314945776433137,
      "loss": 11.1076,
      "step": 160020
    },
    {
      "epoch": 19.257521058965104,
      "grad_norm": Infinity,
      "learning_rate": 0.00018314734430723522,
      "loss": 11.2068,
      "step": 160030
    },
    {
      "epoch": 19.258724428399518,
      "grad_norm": Infinity,
      "learning_rate": 0.00018314523072980436,
      "loss": 11.113,
      "step": 160040
    },
    {
      "epoch": 19.259927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.0001831431170320418,
      "loss": 10.97,
      "step": 160050
    },
    {
      "epoch": 19.261131167268353,
      "grad_norm": Infinity,
      "learning_rate": 0.00018314100321395062,
      "loss": 11.1879,
      "step": 160060
    },
    {
      "epoch": 19.262334536702767,
      "grad_norm": Infinity,
      "learning_rate": 0.00018313888927553391,
      "loss": 11.1473,
      "step": 160070
    },
    {
      "epoch": 19.263537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.00018313677521679468,
      "loss": 11.171,
      "step": 160080
    },
    {
      "epoch": 19.264741275571602,
      "grad_norm": Infinity,
      "learning_rate": 0.00018313466103773602,
      "loss": 11.224,
      "step": 160090
    },
    {
      "epoch": 19.265944645006016,
      "grad_norm": Infinity,
      "learning_rate": 0.00018313254673836099,
      "loss": 11.2056,
      "step": 160100
    },
    {
      "epoch": 19.267148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00018313043231867263,
      "loss": 11.0397,
      "step": 160110
    },
    {
      "epoch": 19.26835138387485,
      "grad_norm": Infinity,
      "learning_rate": 0.000183128317778674,
      "loss": 11.0851,
      "step": 160120
    },
    {
      "epoch": 19.269554753309265,
      "grad_norm": Infinity,
      "learning_rate": 0.00018312620311836816,
      "loss": 11.2034,
      "step": 160130
    },
    {
      "epoch": 19.270758122743683,
      "grad_norm": Infinity,
      "learning_rate": 0.0001831240883377582,
      "loss": 11.0784,
      "step": 160140
    },
    {
      "epoch": 19.2719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00018312197343684717,
      "loss": 11.2138,
      "step": 160150
    },
    {
      "epoch": 19.273164861612514,
      "grad_norm": Infinity,
      "learning_rate": 0.0001831198584156381,
      "loss": 11.1714,
      "step": 160160
    },
    {
      "epoch": 19.27436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00018311774327413408,
      "loss": 11.0458,
      "step": 160170
    },
    {
      "epoch": 19.27557160048135,
      "grad_norm": Infinity,
      "learning_rate": 0.00018311562801233818,
      "loss": 11.111,
      "step": 160180
    },
    {
      "epoch": 19.276774969915763,
      "grad_norm": Infinity,
      "learning_rate": 0.00018311351263025343,
      "loss": 11.0016,
      "step": 160190
    },
    {
      "epoch": 19.27797833935018,
      "grad_norm": Infinity,
      "learning_rate": 0.00018311139712788292,
      "loss": 11.117,
      "step": 160200
    },
    {
      "epoch": 19.2791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00018310928150522967,
      "loss": 11.0723,
      "step": 160210
    },
    {
      "epoch": 19.280385078219012,
      "grad_norm": Infinity,
      "learning_rate": 0.00018310716576229677,
      "loss": 11.17,
      "step": 160220
    },
    {
      "epoch": 19.28158844765343,
      "grad_norm": Infinity,
      "learning_rate": 0.00018310504989908733,
      "loss": 11.1638,
      "step": 160230
    },
    {
      "epoch": 19.282791817087848,
      "grad_norm": Infinity,
      "learning_rate": 0.00018310293391560434,
      "loss": 11.1958,
      "step": 160240
    },
    {
      "epoch": 19.28399518652226,
      "grad_norm": Infinity,
      "learning_rate": 0.00018310081781185088,
      "loss": 11.1338,
      "step": 160250
    },
    {
      "epoch": 19.28519855595668,
      "grad_norm": Infinity,
      "learning_rate": 0.00018309870158783004,
      "loss": 11.0154,
      "step": 160260
    },
    {
      "epoch": 19.286401925391097,
      "grad_norm": Infinity,
      "learning_rate": 0.00018309658524354484,
      "loss": 11.1427,
      "step": 160270
    },
    {
      "epoch": 19.28760529482551,
      "grad_norm": Infinity,
      "learning_rate": 0.00018309446877899839,
      "loss": 11.0997,
      "step": 160280
    },
    {
      "epoch": 19.28880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830923521941937,
      "loss": 11.1912,
      "step": 160290
    },
    {
      "epoch": 19.290012033694346,
      "grad_norm": Infinity,
      "learning_rate": 0.00018309023548913387,
      "loss": 11.2125,
      "step": 160300
    },
    {
      "epoch": 19.29121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00018308811866382197,
      "loss": 11.1763,
      "step": 160310
    },
    {
      "epoch": 19.292418772563177,
      "grad_norm": Infinity,
      "learning_rate": 0.00018308600171826103,
      "loss": 11.1663,
      "step": 160320
    },
    {
      "epoch": 19.29362214199759,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830838846524541,
      "loss": 11.1433,
      "step": 160330
    },
    {
      "epoch": 19.29482551143201,
      "grad_norm": Infinity,
      "learning_rate": 0.00018308176746640434,
      "loss": 11.1087,
      "step": 160340
    },
    {
      "epoch": 19.296028880866427,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830796501601147,
      "loss": 11.1544,
      "step": 160350
    },
    {
      "epoch": 19.29723225030084,
      "grad_norm": Infinity,
      "learning_rate": 0.00018307753273358833,
      "loss": 11.0843,
      "step": 160360
    },
    {
      "epoch": 19.29843561973526,
      "grad_norm": Infinity,
      "learning_rate": 0.00018307541518682827,
      "loss": 11.2509,
      "step": 160370
    },
    {
      "epoch": 19.299638989169676,
      "grad_norm": Infinity,
      "learning_rate": 0.00018307329751983756,
      "loss": 11.211,
      "step": 160380
    },
    {
      "epoch": 19.30084235860409,
      "grad_norm": Infinity,
      "learning_rate": 0.00018307117973261926,
      "loss": 11.1481,
      "step": 160390
    },
    {
      "epoch": 19.302045728038507,
      "grad_norm": Infinity,
      "learning_rate": 0.00018306906182517647,
      "loss": 11.1643,
      "step": 160400
    },
    {
      "epoch": 19.303249097472925,
      "grad_norm": Infinity,
      "learning_rate": 0.00018306694379751224,
      "loss": 11.1352,
      "step": 160410
    },
    {
      "epoch": 19.30445246690734,
      "grad_norm": Infinity,
      "learning_rate": 0.00018306482564962962,
      "loss": 11.1276,
      "step": 160420
    },
    {
      "epoch": 19.305655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830627073815317,
      "loss": 11.0039,
      "step": 160430
    },
    {
      "epoch": 19.306859205776174,
      "grad_norm": Infinity,
      "learning_rate": 0.00018306058899322152,
      "loss": 11.0804,
      "step": 160440
    },
    {
      "epoch": 19.308062575210588,
      "grad_norm": Infinity,
      "learning_rate": 0.00018305847048470216,
      "loss": 11.2549,
      "step": 160450
    },
    {
      "epoch": 19.309265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830563518559767,
      "loss": 11.2358,
      "step": 160460
    },
    {
      "epoch": 19.310469314079423,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830542331070482,
      "loss": 11.1776,
      "step": 160470
    },
    {
      "epoch": 19.311672683513837,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830521142379197,
      "loss": 11.156,
      "step": 160480
    },
    {
      "epoch": 19.312876052948255,
      "grad_norm": Infinity,
      "learning_rate": 0.00018304999524859433,
      "loss": 11.1994,
      "step": 160490
    },
    {
      "epoch": 19.314079422382672,
      "grad_norm": Infinity,
      "learning_rate": 0.00018304787613907504,
      "loss": 11.1623,
      "step": 160500
    },
    {
      "epoch": 19.315282791817086,
      "grad_norm": Infinity,
      "learning_rate": 0.000183045756909365,
      "loss": 11.2263,
      "step": 160510
    },
    {
      "epoch": 19.316486161251504,
      "grad_norm": Infinity,
      "learning_rate": 0.00018304363755946727,
      "loss": 11.1237,
      "step": 160520
    },
    {
      "epoch": 19.31768953068592,
      "grad_norm": Infinity,
      "learning_rate": 0.00018304151808938487,
      "loss": 11.0652,
      "step": 160530
    },
    {
      "epoch": 19.318892900120336,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830393984991209,
      "loss": 11.1131,
      "step": 160540
    },
    {
      "epoch": 19.320096269554753,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830372787886784,
      "loss": 11.1326,
      "step": 160550
    },
    {
      "epoch": 19.32129963898917,
      "grad_norm": Infinity,
      "learning_rate": 0.00018303515895806048,
      "loss": 11.1641,
      "step": 160560
    },
    {
      "epoch": 19.322503008423585,
      "grad_norm": Infinity,
      "learning_rate": 0.00018303303900727019,
      "loss": 11.1123,
      "step": 160570
    },
    {
      "epoch": 19.323706377858002,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830309189363106,
      "loss": 11.2257,
      "step": 160580
    },
    {
      "epoch": 19.32490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00018302879874518472,
      "loss": 11.082,
      "step": 160590
    },
    {
      "epoch": 19.326113116726834,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830266784338957,
      "loss": 11.0323,
      "step": 160600
    },
    {
      "epoch": 19.32731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830245580024466,
      "loss": 11.0977,
      "step": 160610
    },
    {
      "epoch": 19.32851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.00018302243745084041,
      "loss": 11.2239,
      "step": 160620
    },
    {
      "epoch": 19.329723225030083,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830203167790803,
      "loss": 11.1488,
      "step": 160630
    },
    {
      "epoch": 19.3309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830181959871693,
      "loss": 11.2383,
      "step": 160640
    },
    {
      "epoch": 19.332129963898918,
      "grad_norm": Infinity,
      "learning_rate": 0.00018301607507511045,
      "loss": 11.1904,
      "step": 160650
    },
    {
      "epoch": 19.333333333333332,
      "grad_norm": Infinity,
      "learning_rate": 0.00018301395404290687,
      "loss": 11.1402,
      "step": 160660
    },
    {
      "epoch": 19.33453670276775,
      "grad_norm": Infinity,
      "learning_rate": 0.00018301183289056157,
      "loss": 11.2458,
      "step": 160670
    },
    {
      "epoch": 19.335740072202167,
      "grad_norm": Infinity,
      "learning_rate": 0.00018300971161807768,
      "loss": 11.1825,
      "step": 160680
    },
    {
      "epoch": 19.33694344163658,
      "grad_norm": Infinity,
      "learning_rate": 0.00018300759022545822,
      "loss": 11.3386,
      "step": 160690
    },
    {
      "epoch": 19.338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.0001830054687127063,
      "loss": 11.2002,
      "step": 160700
    },
    {
      "epoch": 19.339350180505416,
      "grad_norm": Infinity,
      "learning_rate": 0.00018300334707982497,
      "loss": 11.0794,
      "step": 160710
    },
    {
      "epoch": 19.34055354993983,
      "grad_norm": Infinity,
      "learning_rate": 0.00018300122532681728,
      "loss": 11.0661,
      "step": 160720
    },
    {
      "epoch": 19.341756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.00018299910345368633,
      "loss": 10.9984,
      "step": 160730
    },
    {
      "epoch": 19.342960288808666,
      "grad_norm": Infinity,
      "learning_rate": 0.00018299698146043522,
      "loss": 11.0773,
      "step": 160740
    },
    {
      "epoch": 19.34416365824308,
      "grad_norm": Infinity,
      "learning_rate": 0.00018299485934706697,
      "loss": 11.129,
      "step": 160750
    },
    {
      "epoch": 19.345367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00018299273711358467,
      "loss": 11.1263,
      "step": 160760
    },
    {
      "epoch": 19.346570397111915,
      "grad_norm": Infinity,
      "learning_rate": 0.00018299061475999137,
      "loss": 11.1046,
      "step": 160770
    },
    {
      "epoch": 19.34777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829884922862902,
      "loss": 11.057,
      "step": 160780
    },
    {
      "epoch": 19.348977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.00018298636969248414,
      "loss": 11.1086,
      "step": 160790
    },
    {
      "epoch": 19.350180505415164,
      "grad_norm": Infinity,
      "learning_rate": 0.00018298424697857634,
      "loss": 11.1199,
      "step": 160800
    },
    {
      "epoch": 19.351383874849578,
      "grad_norm": Infinity,
      "learning_rate": 0.00018298212414456986,
      "loss": 11.0532,
      "step": 160810
    },
    {
      "epoch": 19.352587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00018298000119046774,
      "loss": 11.0792,
      "step": 160820
    },
    {
      "epoch": 19.353790613718413,
      "grad_norm": Infinity,
      "learning_rate": 0.00018297787811627307,
      "loss": 11.2113,
      "step": 160830
    },
    {
      "epoch": 19.354993983152827,
      "grad_norm": Infinity,
      "learning_rate": 0.00018297575492198895,
      "loss": 11.1289,
      "step": 160840
    },
    {
      "epoch": 19.356197352587245,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829736316076184,
      "loss": 11.0859,
      "step": 160850
    },
    {
      "epoch": 19.357400722021662,
      "grad_norm": Infinity,
      "learning_rate": 0.00018297150817316453,
      "loss": 11.1366,
      "step": 160860
    },
    {
      "epoch": 19.358604091456076,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829693846186304,
      "loss": 11.0457,
      "step": 160870
    },
    {
      "epoch": 19.359807460890494,
      "grad_norm": Infinity,
      "learning_rate": 0.00018296726094401908,
      "loss": 11.1258,
      "step": 160880
    },
    {
      "epoch": 19.36101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.00018296513714933369,
      "loss": 11.1406,
      "step": 160890
    },
    {
      "epoch": 19.362214199759325,
      "grad_norm": Infinity,
      "learning_rate": 0.00018296301323457721,
      "loss": 11.1511,
      "step": 160900
    },
    {
      "epoch": 19.363417569193743,
      "grad_norm": Infinity,
      "learning_rate": 0.00018296088919975278,
      "loss": 11.1093,
      "step": 160910
    },
    {
      "epoch": 19.36462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.00018295876504486347,
      "loss": 11.199,
      "step": 160920
    },
    {
      "epoch": 19.365824308062574,
      "grad_norm": Infinity,
      "learning_rate": 0.00018295664076991232,
      "loss": 11.1904,
      "step": 160930
    },
    {
      "epoch": 19.367027677496992,
      "grad_norm": Infinity,
      "learning_rate": 0.00018295451637490246,
      "loss": 11.2918,
      "step": 160940
    },
    {
      "epoch": 19.36823104693141,
      "grad_norm": Infinity,
      "learning_rate": 0.00018295239185983693,
      "loss": 11.1088,
      "step": 160950
    },
    {
      "epoch": 19.369434416365824,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829502672247188,
      "loss": 11.1114,
      "step": 160960
    },
    {
      "epoch": 19.37063778580024,
      "grad_norm": Infinity,
      "learning_rate": 0.00018294814246955115,
      "loss": 11.2617,
      "step": 160970
    },
    {
      "epoch": 19.37184115523466,
      "grad_norm": Infinity,
      "learning_rate": 0.00018294601759433706,
      "loss": 11.1418,
      "step": 160980
    },
    {
      "epoch": 19.373044524669073,
      "grad_norm": Infinity,
      "learning_rate": 0.00018294389259907964,
      "loss": 11.2187,
      "step": 160990
    },
    {
      "epoch": 19.37424789410349,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829417674837819,
      "loss": 11.0197,
      "step": 161000
    },
    {
      "epoch": 19.375451263537904,
      "grad_norm": Infinity,
      "learning_rate": 0.00018293964224844693,
      "loss": 11.1371,
      "step": 161010
    },
    {
      "epoch": 19.376654632972322,
      "grad_norm": Infinity,
      "learning_rate": 0.00018293751689307783,
      "loss": 11.1498,
      "step": 161020
    },
    {
      "epoch": 19.37785800240674,
      "grad_norm": Infinity,
      "learning_rate": 0.00018293539141767768,
      "loss": 11.0602,
      "step": 161030
    },
    {
      "epoch": 19.379061371841154,
      "grad_norm": Infinity,
      "learning_rate": 0.00018293326582224953,
      "loss": 11.1184,
      "step": 161040
    },
    {
      "epoch": 19.38026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00018293114010679647,
      "loss": 11.1488,
      "step": 161050
    },
    {
      "epoch": 19.38146811070999,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829290142713216,
      "loss": 11.1844,
      "step": 161060
    },
    {
      "epoch": 19.382671480144403,
      "grad_norm": Infinity,
      "learning_rate": 0.00018292688831582794,
      "loss": 11.0572,
      "step": 161070
    },
    {
      "epoch": 19.38387484957882,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829247622403186,
      "loss": 11.0694,
      "step": 161080
    },
    {
      "epoch": 19.385078219013238,
      "grad_norm": Infinity,
      "learning_rate": 0.00018292263604479666,
      "loss": 11.1746,
      "step": 161090
    },
    {
      "epoch": 19.386281588447652,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829205097292652,
      "loss": 11.0507,
      "step": 161100
    },
    {
      "epoch": 19.38748495788207,
      "grad_norm": Infinity,
      "learning_rate": 0.00018291838329372728,
      "loss": 11.167,
      "step": 161110
    },
    {
      "epoch": 19.388688327316487,
      "grad_norm": Infinity,
      "learning_rate": 0.000182916256738186,
      "loss": 11.107,
      "step": 161120
    },
    {
      "epoch": 19.3898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00018291413006264442,
      "loss": 11.1558,
      "step": 161130
    },
    {
      "epoch": 19.39109506618532,
      "grad_norm": Infinity,
      "learning_rate": 0.00018291200326710563,
      "loss": 11.1019,
      "step": 161140
    },
    {
      "epoch": 19.392298435619736,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829098763515727,
      "loss": 11.1393,
      "step": 161150
    },
    {
      "epoch": 19.39350180505415,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829077493160487,
      "loss": 11.1555,
      "step": 161160
    },
    {
      "epoch": 19.394705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.0001829056221605367,
      "loss": 11.0415,
      "step": 161170
    },
    {
      "epoch": 19.395908543922985,
      "grad_norm": Infinity,
      "learning_rate": 0.00018290349488503982,
      "loss": 11.0785,
      "step": 161180
    },
    {
      "epoch": 19.3971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.00018290136748956112,
      "loss": 11.1671,
      "step": 161190
    },
    {
      "epoch": 19.398315282791817,
      "grad_norm": Infinity,
      "learning_rate": 0.00018289923997410364,
      "loss": 11.1875,
      "step": 161200
    },
    {
      "epoch": 19.399518652226234,
      "grad_norm": Infinity,
      "learning_rate": 0.00018289711233867052,
      "loss": 11.1483,
      "step": 161210
    },
    {
      "epoch": 19.40072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828949845832648,
      "loss": 11.117,
      "step": 161220
    },
    {
      "epoch": 19.401925391095066,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828928567078896,
      "loss": 10.9823,
      "step": 161230
    },
    {
      "epoch": 19.403128760529484,
      "grad_norm": Infinity,
      "learning_rate": 0.00018289072871254796,
      "loss": 11.2061,
      "step": 161240
    },
    {
      "epoch": 19.404332129963898,
      "grad_norm": Infinity,
      "learning_rate": 0.00018288860059724295,
      "loss": 11.2871,
      "step": 161250
    },
    {
      "epoch": 19.405535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00018288647236197768,
      "loss": 11.2584,
      "step": 161260
    },
    {
      "epoch": 19.406738868832733,
      "grad_norm": Infinity,
      "learning_rate": 0.00018288434400675523,
      "loss": 11.1404,
      "step": 161270
    },
    {
      "epoch": 19.407942238267147,
      "grad_norm": Infinity,
      "learning_rate": 0.00018288221553157866,
      "loss": 11.0662,
      "step": 161280
    },
    {
      "epoch": 19.409145607701564,
      "grad_norm": Infinity,
      "learning_rate": 0.00018288008693645105,
      "loss": 11.1553,
      "step": 161290
    },
    {
      "epoch": 19.410348977135982,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828779582213755,
      "loss": 11.049,
      "step": 161300
    },
    {
      "epoch": 19.411552346570396,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828758293863551,
      "loss": 11.1687,
      "step": 161310
    },
    {
      "epoch": 19.412755716004813,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828737004313929,
      "loss": 11.1317,
      "step": 161320
    },
    {
      "epoch": 19.41395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.000182871571356492,
      "loss": 11.097,
      "step": 161330
    },
    {
      "epoch": 19.415162454873645,
      "grad_norm": Infinity,
      "learning_rate": 0.00018286944216165546,
      "loss": 11.2208,
      "step": 161340
    },
    {
      "epoch": 19.416365824308063,
      "grad_norm": Infinity,
      "learning_rate": 0.00018286731284688639,
      "loss": 11.1941,
      "step": 161350
    },
    {
      "epoch": 19.41756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.00018286518341218785,
      "loss": 11.1709,
      "step": 161360
    },
    {
      "epoch": 19.418772563176894,
      "grad_norm": Infinity,
      "learning_rate": 0.00018286305385756295,
      "loss": 11.0176,
      "step": 161370
    },
    {
      "epoch": 19.41997593261131,
      "grad_norm": Infinity,
      "learning_rate": 0.00018286092418301472,
      "loss": 11.1106,
      "step": 161380
    },
    {
      "epoch": 19.42117930204573,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828587943885463,
      "loss": 11.1976,
      "step": 161390
    },
    {
      "epoch": 19.422382671480143,
      "grad_norm": Infinity,
      "learning_rate": 0.00018285666447416077,
      "loss": 11.2034,
      "step": 161400
    },
    {
      "epoch": 19.42358604091456,
      "grad_norm": Infinity,
      "learning_rate": 0.00018285453443986116,
      "loss": 11.1639,
      "step": 161410
    },
    {
      "epoch": 19.42478941034898,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828524042856506,
      "loss": 11.1199,
      "step": 161420
    },
    {
      "epoch": 19.425992779783392,
      "grad_norm": Infinity,
      "learning_rate": 0.00018285027401153213,
      "loss": 11.1309,
      "step": 161430
    },
    {
      "epoch": 19.42719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00018284814361750886,
      "loss": 11.1051,
      "step": 161440
    },
    {
      "epoch": 19.428399518652228,
      "grad_norm": Infinity,
      "learning_rate": 0.00018284601310358388,
      "loss": 11.1438,
      "step": 161450
    },
    {
      "epoch": 19.42960288808664,
      "grad_norm": Infinity,
      "learning_rate": 0.00018284388246976027,
      "loss": 11.0844,
      "step": 161460
    },
    {
      "epoch": 19.43080625752106,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828417517160411,
      "loss": 11.1385,
      "step": 161470
    },
    {
      "epoch": 19.432009626955477,
      "grad_norm": Infinity,
      "learning_rate": 0.00018283962084242948,
      "loss": 11.2327,
      "step": 161480
    },
    {
      "epoch": 19.43321299638989,
      "grad_norm": Infinity,
      "learning_rate": 0.00018283748984892847,
      "loss": 11.1238,
      "step": 161490
    },
    {
      "epoch": 19.43441636582431,
      "grad_norm": Infinity,
      "learning_rate": 0.00018283535873554114,
      "loss": 11.0821,
      "step": 161500
    },
    {
      "epoch": 19.435619735258726,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828332275022706,
      "loss": 11.1408,
      "step": 161510
    },
    {
      "epoch": 19.43682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00018283109614911994,
      "loss": 11.0691,
      "step": 161520
    },
    {
      "epoch": 19.438026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00018282896467609224,
      "loss": 11.1385,
      "step": 161530
    },
    {
      "epoch": 19.439229843561975,
      "grad_norm": Infinity,
      "learning_rate": 0.00018282683308319057,
      "loss": 11.1372,
      "step": 161540
    },
    {
      "epoch": 19.44043321299639,
      "grad_norm": Infinity,
      "learning_rate": 0.00018282470137041804,
      "loss": 11.0371,
      "step": 161550
    },
    {
      "epoch": 19.441636582430807,
      "grad_norm": Infinity,
      "learning_rate": 0.0001828225695377777,
      "loss": 11.043,
      "step": 161560
    },
    {
      "epoch": 19.442839951865224,
      "grad_norm": Infinity,
      "learning_rate": 0.00018282043758527266,
      "loss": 11.1695,
      "step": 161570
    },
    {
      "epoch": 19.444043321299638,
      "grad_norm": Infinity,
      "learning_rate": 0.00018281830551290597,
      "loss": 11.2091,
      "step": 161580
    },
    {
      "epoch": 19.445246690734056,
      "grad_norm": Infinity,
      "learning_rate": 0.00018281617332068078,
      "loss": 11.0843,
      "step": 161590
    },
    {
      "epoch": 19.446450060168473,
      "grad_norm": Infinity,
      "learning_rate": 0.00018281404100860014,
      "loss": 11.1245,
      "step": 161600
    },
    {
      "epoch": 19.447653429602887,
      "grad_norm": Infinity,
      "learning_rate": 0.00018281190857666712,
      "loss": 11.2474,
      "step": 161610
    },
    {
      "epoch": 19.448856799037305,
      "grad_norm": Infinity,
      "learning_rate": 0.00018280977602488482,
      "loss": 11.132,
      "step": 161620
    },
    {
      "epoch": 19.450060168471722,
      "grad_norm": Infinity,
      "learning_rate": 0.00018280764335325636,
      "loss": 11.1568,
      "step": 161630
    },
    {
      "epoch": 19.451263537906136,
      "grad_norm": Infinity,
      "learning_rate": 0.00018280551056178474,
      "loss": 11.2903,
      "step": 161640
    },
    {
      "epoch": 19.452466907340554,
      "grad_norm": Infinity,
      "learning_rate": 0.00018280337765047314,
      "loss": 11.1696,
      "step": 161650
    },
    {
      "epoch": 19.45367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00018280124461932462,
      "loss": 11.2114,
      "step": 161660
    },
    {
      "epoch": 19.454873646209386,
      "grad_norm": Infinity,
      "learning_rate": 0.00018279911146834224,
      "loss": 11.2059,
      "step": 161670
    },
    {
      "epoch": 19.456077015643803,
      "grad_norm": Infinity,
      "learning_rate": 0.00018279697819752907,
      "loss": 11.056,
      "step": 161680
    },
    {
      "epoch": 19.45728038507822,
      "grad_norm": Infinity,
      "learning_rate": 0.00018279484480688828,
      "loss": 11.1237,
      "step": 161690
    },
    {
      "epoch": 19.458483754512635,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827927112964229,
      "loss": 11.2018,
      "step": 161700
    },
    {
      "epoch": 19.459687123947052,
      "grad_norm": Infinity,
      "learning_rate": 0.000182790577666136,
      "loss": 11.1812,
      "step": 161710
    },
    {
      "epoch": 19.460890493381466,
      "grad_norm": Infinity,
      "learning_rate": 0.00018278844391603075,
      "loss": 11.0153,
      "step": 161720
    },
    {
      "epoch": 19.462093862815884,
      "grad_norm": Infinity,
      "learning_rate": 0.00018278631004611015,
      "loss": 11.1073,
      "step": 161730
    },
    {
      "epoch": 19.4632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00018278417605637728,
      "loss": 11.1949,
      "step": 161740
    },
    {
      "epoch": 19.464500601684716,
      "grad_norm": Infinity,
      "learning_rate": 0.00018278204194683531,
      "loss": 11.144,
      "step": 161750
    },
    {
      "epoch": 19.465703971119133,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827799077174873,
      "loss": 11.1027,
      "step": 161760
    },
    {
      "epoch": 19.46690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827777733683363,
      "loss": 11.1226,
      "step": 161770
    },
    {
      "epoch": 19.468110709987965,
      "grad_norm": Infinity,
      "learning_rate": 0.00018277563889938542,
      "loss": 11.1499,
      "step": 161780
    },
    {
      "epoch": 19.469314079422382,
      "grad_norm": Infinity,
      "learning_rate": 0.00018277350431063777,
      "loss": 11.0822,
      "step": 161790
    },
    {
      "epoch": 19.4705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827713696020964,
      "loss": 11.1109,
      "step": 161800
    },
    {
      "epoch": 19.471720818291214,
      "grad_norm": Infinity,
      "learning_rate": 0.00018276923477376445,
      "loss": 11.1281,
      "step": 161810
    },
    {
      "epoch": 19.47292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00018276709982564498,
      "loss": 11.1451,
      "step": 161820
    },
    {
      "epoch": 19.47412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.00018276496475774106,
      "loss": 11.127,
      "step": 161830
    },
    {
      "epoch": 19.475330926594463,
      "grad_norm": Infinity,
      "learning_rate": 0.00018276282957005582,
      "loss": 11.0422,
      "step": 161840
    },
    {
      "epoch": 19.47653429602888,
      "grad_norm": Infinity,
      "learning_rate": 0.00018276069426259234,
      "loss": 11.0957,
      "step": 161850
    },
    {
      "epoch": 19.477737665463298,
      "grad_norm": Infinity,
      "learning_rate": 0.00018275855883535368,
      "loss": 11.1305,
      "step": 161860
    },
    {
      "epoch": 19.478941034897712,
      "grad_norm": Infinity,
      "learning_rate": 0.00018275642328834297,
      "loss": 11.1308,
      "step": 161870
    },
    {
      "epoch": 19.48014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827542876215633,
      "loss": 11.2344,
      "step": 161880
    },
    {
      "epoch": 19.481347773766547,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827521518350177,
      "loss": 11.0885,
      "step": 161890
    },
    {
      "epoch": 19.48255114320096,
      "grad_norm": Infinity,
      "learning_rate": 0.00018275001592870933,
      "loss": 11.1191,
      "step": 161900
    },
    {
      "epoch": 19.48375451263538,
      "grad_norm": Infinity,
      "learning_rate": 0.00018274787990264123,
      "loss": 11.2096,
      "step": 161910
    },
    {
      "epoch": 19.484957882069796,
      "grad_norm": Infinity,
      "learning_rate": 0.00018274574375681658,
      "loss": 11.1793,
      "step": 161920
    },
    {
      "epoch": 19.48616125150421,
      "grad_norm": Infinity,
      "learning_rate": 0.00018274360749123835,
      "loss": 11.1446,
      "step": 161930
    },
    {
      "epoch": 19.487364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827414711059097,
      "loss": 11.1112,
      "step": 161940
    },
    {
      "epoch": 19.488567990373046,
      "grad_norm": Infinity,
      "learning_rate": 0.00018273933460083373,
      "loss": 11.1483,
      "step": 161950
    },
    {
      "epoch": 19.48977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.00018273719797601352,
      "loss": 11.1544,
      "step": 161960
    },
    {
      "epoch": 19.490974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00018273506123145215,
      "loss": 11.184,
      "step": 161970
    },
    {
      "epoch": 19.492178098676295,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827329243671527,
      "loss": 11.1902,
      "step": 161980
    },
    {
      "epoch": 19.49338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00018273078738311832,
      "loss": 11.1775,
      "step": 161990
    },
    {
      "epoch": 19.494584837545126,
      "grad_norm": Infinity,
      "learning_rate": 0.00018272865027935204,
      "loss": 11.1122,
      "step": 162000
    },
    {
      "epoch": 19.495788206979544,
      "grad_norm": Infinity,
      "learning_rate": 0.000182726513055857,
      "loss": 11.2597,
      "step": 162010
    },
    {
      "epoch": 19.496991576413958,
      "grad_norm": Infinity,
      "learning_rate": 0.00018272437571263624,
      "loss": 11.1053,
      "step": 162020
    },
    {
      "epoch": 19.498194945848375,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827222382496929,
      "loss": 11.102,
      "step": 162030
    },
    {
      "epoch": 19.499398315282793,
      "grad_norm": Infinity,
      "learning_rate": 0.00018272010066703004,
      "loss": 11.1276,
      "step": 162040
    },
    {
      "epoch": 19.500601684717207,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827179629646508,
      "loss": 11.1512,
      "step": 162050
    },
    {
      "epoch": 19.501805054151625,
      "grad_norm": Infinity,
      "learning_rate": 0.00018271582514255825,
      "loss": 11.1552,
      "step": 162060
    },
    {
      "epoch": 19.503008423586042,
      "grad_norm": Infinity,
      "learning_rate": 0.00018271368720075546,
      "loss": 11.1152,
      "step": 162070
    },
    {
      "epoch": 19.504211793020456,
      "grad_norm": Infinity,
      "learning_rate": 0.00018271154913924553,
      "loss": 10.966,
      "step": 162080
    },
    {
      "epoch": 19.505415162454874,
      "grad_norm": Infinity,
      "learning_rate": 0.00018270941095803158,
      "loss": 11.1908,
      "step": 162090
    },
    {
      "epoch": 19.50661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00018270727265711672,
      "loss": 11.1515,
      "step": 162100
    },
    {
      "epoch": 19.507821901323705,
      "grad_norm": Infinity,
      "learning_rate": 0.00018270513423650396,
      "loss": 11.1622,
      "step": 162110
    },
    {
      "epoch": 19.509025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.0001827029956961965,
      "loss": 11.2566,
      "step": 162120
    },
    {
      "epoch": 19.51022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.00018270085703619738,
      "loss": 11.1811,
      "step": 162130
    },
    {
      "epoch": 19.511432009626954,
      "grad_norm": Infinity,
      "learning_rate": 0.00018269871825650968,
      "loss": 11.1443,
      "step": 162140
    },
    {
      "epoch": 19.512635379061372,
      "grad_norm": Infinity,
      "learning_rate": 0.00018269657935713653,
      "loss": 11.1258,
      "step": 162150
    },
    {
      "epoch": 19.51383874849579,
      "grad_norm": Infinity,
      "learning_rate": 0.000182694440338081,
      "loss": 11.1851,
      "step": 162160
    },
    {
      "epoch": 19.515042117930204,
      "grad_norm": Infinity,
      "learning_rate": 0.0001826923011993462,
      "loss": 11.0742,
      "step": 162170
    },
    {
      "epoch": 19.51624548736462,
      "grad_norm": Infinity,
      "learning_rate": 0.00018269016194093527,
      "loss": 11.215,
      "step": 162180
    },
    {
      "epoch": 19.51744885679904,
      "grad_norm": Infinity,
      "learning_rate": 0.00018268802256285121,
      "loss": 11.2069,
      "step": 162190
    },
    {
      "epoch": 19.518652226233453,
      "grad_norm": Infinity,
      "learning_rate": 0.00018268588306509718,
      "loss": 11.1003,
      "step": 162200
    },
    {
      "epoch": 19.51985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00018268374344767625,
      "loss": 11.0829,
      "step": 162210
    },
    {
      "epoch": 19.521058965102288,
      "grad_norm": Infinity,
      "learning_rate": 0.00018268160371059155,
      "loss": 11.1549,
      "step": 162220
    },
    {
      "epoch": 19.522262334536702,
      "grad_norm": Infinity,
      "learning_rate": 0.00018267946385384617,
      "loss": 11.0657,
      "step": 162230
    },
    {
      "epoch": 19.52346570397112,
      "grad_norm": Infinity,
      "learning_rate": 0.00018267732387744316,
      "loss": 11.2486,
      "step": 162240
    },
    {
      "epoch": 19.524669073405537,
      "grad_norm": Infinity,
      "learning_rate": 0.00018267518378138563,
      "loss": 11.1109,
      "step": 162250
    },
    {
      "epoch": 19.52587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00018267304356567676,
      "loss": 10.9885,
      "step": 162260
    },
    {
      "epoch": 19.52707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00018267090323031954,
      "loss": 11.1365,
      "step": 162270
    },
    {
      "epoch": 19.528279181708786,
      "grad_norm": Infinity,
      "learning_rate": 0.00018266876277531715,
      "loss": 11.0086,
      "step": 162280
    },
    {
      "epoch": 19.5294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00018266662220067258,
      "loss": 11.22,
      "step": 162290
    },
    {
      "epoch": 19.530685920577618,
      "grad_norm": Infinity,
      "learning_rate": 0.00018266448150638907,
      "loss": 11.1783,
      "step": 162300
    },
    {
      "epoch": 19.531889290012035,
      "grad_norm": Infinity,
      "learning_rate": 0.00018266234069246962,
      "loss": 11.1648,
      "step": 162310
    },
    {
      "epoch": 19.53309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.00018266019975891735,
      "loss": 11.1629,
      "step": 162320
    },
    {
      "epoch": 19.534296028880867,
      "grad_norm": Infinity,
      "learning_rate": 0.00018265805870573538,
      "loss": 11.1951,
      "step": 162330
    },
    {
      "epoch": 19.535499398315284,
      "grad_norm": Infinity,
      "learning_rate": 0.00018265591753292677,
      "loss": 11.1585,
      "step": 162340
    },
    {
      "epoch": 19.5367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00018265377624049464,
      "loss": 11.1936,
      "step": 162350
    },
    {
      "epoch": 19.537906137184116,
      "grad_norm": Infinity,
      "learning_rate": 0.0001826516348284421,
      "loss": 11.1752,
      "step": 162360
    },
    {
      "epoch": 19.53910950661853,
      "grad_norm": Infinity,
      "learning_rate": 0.00018264949329677222,
      "loss": 11.0277,
      "step": 162370
    },
    {
      "epoch": 19.540312876052948,
      "grad_norm": Infinity,
      "learning_rate": 0.00018264735164548817,
      "loss": 11.2411,
      "step": 162380
    },
    {
      "epoch": 19.541516245487365,
      "grad_norm": Infinity,
      "learning_rate": 0.00018264520987459296,
      "loss": 11.1114,
      "step": 162390
    },
    {
      "epoch": 19.54271961492178,
      "grad_norm": Infinity,
      "learning_rate": 0.0001826430679840897,
      "loss": 11.1428,
      "step": 162400
    },
    {
      "epoch": 19.543922984356197,
      "grad_norm": Infinity,
      "learning_rate": 0.00018264092597398155,
      "loss": 11.1593,
      "step": 162410
    },
    {
      "epoch": 19.545126353790614,
      "grad_norm": Infinity,
      "learning_rate": 0.00018263878384427158,
      "loss": 11.0661,
      "step": 162420
    },
    {
      "epoch": 19.54632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00018263664159496289,
      "loss": 11.0729,
      "step": 162430
    },
    {
      "epoch": 19.547533092659446,
      "grad_norm": Infinity,
      "learning_rate": 0.00018263449922605856,
      "loss": 11.0719,
      "step": 162440
    },
    {
      "epoch": 19.548736462093864,
      "grad_norm": Infinity,
      "learning_rate": 0.0001826323567375617,
      "loss": 11.1303,
      "step": 162450
    },
    {
      "epoch": 19.549939831528278,
      "grad_norm": Infinity,
      "learning_rate": 0.00018263021412947544,
      "loss": 11.2754,
      "step": 162460
    },
    {
      "epoch": 19.551143200962695,
      "grad_norm": Infinity,
      "learning_rate": 0.00018262807140180285,
      "loss": 11.2033,
      "step": 162470
    },
    {
      "epoch": 19.552346570397113,
      "grad_norm": Infinity,
      "learning_rate": 0.00018262592855454704,
      "loss": 11.1621,
      "step": 162480
    },
    {
      "epoch": 19.553549939831527,
      "grad_norm": Infinity,
      "learning_rate": 0.0001826237855877111,
      "loss": 11.1376,
      "step": 162490
    },
    {
      "epoch": 19.554753309265944,
      "grad_norm": Infinity,
      "learning_rate": 0.00018262164250129818,
      "loss": 11.0886,
      "step": 162500
    },
    {
      "epoch": 19.555956678700362,
      "grad_norm": Infinity,
      "learning_rate": 0.0001826194992953113,
      "loss": 11.1191,
      "step": 162510
    },
    {
      "epoch": 19.557160048134776,
      "grad_norm": Infinity,
      "learning_rate": 0.00018261735596975364,
      "loss": 11.1761,
      "step": 162520
    },
    {
      "epoch": 19.558363417569193,
      "grad_norm": Infinity,
      "learning_rate": 0.00018261521252462823,
      "loss": 11.1334,
      "step": 162530
    },
    {
      "epoch": 19.55956678700361,
      "grad_norm": Infinity,
      "learning_rate": 0.00018261306895993827,
      "loss": 11.1404,
      "step": 162540
    },
    {
      "epoch": 19.560770156438025,
      "grad_norm": Infinity,
      "learning_rate": 0.00018261092527568673,
      "loss": 11.1523,
      "step": 162550
    },
    {
      "epoch": 19.561973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.00018260878147187683,
      "loss": 11.1759,
      "step": 162560
    },
    {
      "epoch": 19.56317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00018260663754851163,
      "loss": 11.0084,
      "step": 162570
    },
    {
      "epoch": 19.564380264741274,
      "grad_norm": Infinity,
      "learning_rate": 0.00018260449350559422,
      "loss": 11.0994,
      "step": 162580
    },
    {
      "epoch": 19.56558363417569,
      "grad_norm": Infinity,
      "learning_rate": 0.00018260234934312772,
      "loss": 11.1455,
      "step": 162590
    },
    {
      "epoch": 19.56678700361011,
      "grad_norm": Infinity,
      "learning_rate": 0.00018260020506111522,
      "loss": 11.1078,
      "step": 162600
    },
    {
      "epoch": 19.567990373044523,
      "grad_norm": Infinity,
      "learning_rate": 0.00018259806065955983,
      "loss": 11.2148,
      "step": 162610
    },
    {
      "epoch": 19.56919374247894,
      "grad_norm": Infinity,
      "learning_rate": 0.00018259591613846465,
      "loss": 11.1882,
      "step": 162620
    },
    {
      "epoch": 19.57039711191336,
      "grad_norm": Infinity,
      "learning_rate": 0.0001825937714978328,
      "loss": 11.0433,
      "step": 162630
    },
    {
      "epoch": 19.571600481347772,
      "grad_norm": Infinity,
      "learning_rate": 0.00018259162673766737,
      "loss": 11.1789,
      "step": 162640
    },
    {
      "epoch": 19.57280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.00018258948185797148,
      "loss": 11.1198,
      "step": 162650
    },
    {
      "epoch": 19.574007220216608,
      "grad_norm": Infinity,
      "learning_rate": 0.00018258733685874815,
      "loss": 11.1387,
      "step": 162660
    },
    {
      "epoch": 19.57521058965102,
      "grad_norm": Infinity,
      "learning_rate": 0.00018258519174000061,
      "loss": 11.1339,
      "step": 162670
    },
    {
      "epoch": 19.57641395908544,
      "grad_norm": Infinity,
      "learning_rate": 0.00018258304650173193,
      "loss": 11.1132,
      "step": 162680
    },
    {
      "epoch": 19.577617328519857,
      "grad_norm": Infinity,
      "learning_rate": 0.00018258090114394514,
      "loss": 11.151,
      "step": 162690
    },
    {
      "epoch": 19.57882069795427,
      "grad_norm": Infinity,
      "learning_rate": 0.0001825787556666434,
      "loss": 11.1796,
      "step": 162700
    },
    {
      "epoch": 19.58002406738869,
      "grad_norm": Infinity,
      "learning_rate": 0.00018257661006982985,
      "loss": 11.1099,
      "step": 162710
    },
    {
      "epoch": 19.581227436823106,
      "grad_norm": Infinity,
      "learning_rate": 0.00018257446435350754,
      "loss": 11.1478,
      "step": 162720
    },
    {
      "epoch": 19.58243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.00018257231851767958,
      "loss": 11.1537,
      "step": 162730
    },
    {
      "epoch": 19.583634175691937,
      "grad_norm": Infinity,
      "learning_rate": 0.00018257017256234913,
      "loss": 11.2141,
      "step": 162740
    },
    {
      "epoch": 19.584837545126355,
      "grad_norm": Infinity,
      "learning_rate": 0.00018256802648751921,
      "loss": 11.0646,
      "step": 162750
    },
    {
      "epoch": 19.58604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00018256588029319298,
      "loss": 11.1072,
      "step": 162760
    },
    {
      "epoch": 19.587244283995187,
      "grad_norm": Infinity,
      "learning_rate": 0.00018256373397937354,
      "loss": 11.1924,
      "step": 162770
    },
    {
      "epoch": 19.588447653429604,
      "grad_norm": Infinity,
      "learning_rate": 0.000182561587546064,
      "loss": 11.1948,
      "step": 162780
    },
    {
      "epoch": 19.589651022864018,
      "grad_norm": Infinity,
      "learning_rate": 0.00018255944099326745,
      "loss": 11.1777,
      "step": 162790
    },
    {
      "epoch": 19.590854392298436,
      "grad_norm": Infinity,
      "learning_rate": 0.000182557294320987,
      "loss": 11.1685,
      "step": 162800
    },
    {
      "epoch": 19.592057761732853,
      "grad_norm": Infinity,
      "learning_rate": 0.0001825551475292258,
      "loss": 11.1947,
      "step": 162810
    },
    {
      "epoch": 19.593261131167267,
      "grad_norm": Infinity,
      "learning_rate": 0.00018255300061798687,
      "loss": 11.1744,
      "step": 162820
    },
    {
      "epoch": 19.594464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.0001825508535872734,
      "loss": 11.1296,
      "step": 162830
    },
    {
      "epoch": 19.595667870036102,
      "grad_norm": Infinity,
      "learning_rate": 0.00018254870643708845,
      "loss": 11.0999,
      "step": 162840
    },
    {
      "epoch": 19.596871239470516,
      "grad_norm": Infinity,
      "learning_rate": 0.00018254655916743513,
      "loss": 11.101,
      "step": 162850
    },
    {
      "epoch": 19.598074608904934,
      "grad_norm": Infinity,
      "learning_rate": 0.00018254441177831658,
      "loss": 11.1262,
      "step": 162860
    },
    {
      "epoch": 19.59927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00018254226426973587,
      "loss": 11.1276,
      "step": 162870
    },
    {
      "epoch": 19.600481347773766,
      "grad_norm": Infinity,
      "learning_rate": 0.00018254011664169614,
      "loss": 11.0346,
      "step": 162880
    },
    {
      "epoch": 19.601684717208183,
      "grad_norm": Infinity,
      "learning_rate": 0.00018253796889420047,
      "loss": 11.2714,
      "step": 162890
    },
    {
      "epoch": 19.6028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.000182535821027252,
      "loss": 11.2405,
      "step": 162900
    },
    {
      "epoch": 19.604091456077015,
      "grad_norm": Infinity,
      "learning_rate": 0.00018253367304085378,
      "loss": 11.1912,
      "step": 162910
    },
    {
      "epoch": 19.605294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.000182531524935009,
      "loss": 11.0689,
      "step": 162920
    },
    {
      "epoch": 19.60649819494585,
      "grad_norm": Infinity,
      "learning_rate": 0.0001825293767097207,
      "loss": 11.0899,
      "step": 162930
    },
    {
      "epoch": 19.607701564380264,
      "grad_norm": Infinity,
      "learning_rate": 0.00018252722836499205,
      "loss": 11.1685,
      "step": 162940
    },
    {
      "epoch": 19.60890493381468,
      "grad_norm": Infinity,
      "learning_rate": 0.0001825250799008261,
      "loss": 11.1937,
      "step": 162950
    },
    {
      "epoch": 19.6101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.000182522931317226,
      "loss": 11.0739,
      "step": 162960
    },
    {
      "epoch": 19.611311672683513,
      "grad_norm": Infinity,
      "learning_rate": 0.0001825207826141948,
      "loss": 11.1413,
      "step": 162970
    },
    {
      "epoch": 19.61251504211793,
      "grad_norm": Infinity,
      "learning_rate": 0.00018251863379173568,
      "loss": 11.0881,
      "step": 162980
    },
    {
      "epoch": 19.613718411552348,
      "grad_norm": Infinity,
      "learning_rate": 0.00018251648484985173,
      "loss": 11.2409,
      "step": 162990
    },
    {
      "epoch": 19.614921780986762,
      "grad_norm": Infinity,
      "learning_rate": 0.00018251433578854605,
      "loss": 11.0716,
      "step": 163000
    },
    {
      "epoch": 19.61612515042118,
      "grad_norm": Infinity,
      "learning_rate": 0.00018251218660782177,
      "loss": 11.1568,
      "step": 163010
    },
    {
      "epoch": 19.617328519855597,
      "grad_norm": Infinity,
      "learning_rate": 0.00018251003730768197,
      "loss": 11.1597,
      "step": 163020
    },
    {
      "epoch": 19.61853188929001,
      "grad_norm": Infinity,
      "learning_rate": 0.00018250788788812977,
      "loss": 11.1744,
      "step": 163030
    },
    {
      "epoch": 19.61973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.0001825057383491683,
      "loss": 11.1821,
      "step": 163040
    },
    {
      "epoch": 19.620938628158846,
      "grad_norm": Infinity,
      "learning_rate": 0.00018250358869080065,
      "loss": 11.1035,
      "step": 163050
    },
    {
      "epoch": 19.62214199759326,
      "grad_norm": Infinity,
      "learning_rate": 0.00018250143891302992,
      "loss": 11.1784,
      "step": 163060
    },
    {
      "epoch": 19.623345367027678,
      "grad_norm": Infinity,
      "learning_rate": 0.00018249928901585926,
      "loss": 11.1601,
      "step": 163070
    },
    {
      "epoch": 19.624548736462096,
      "grad_norm": Infinity,
      "learning_rate": 0.00018249713899929176,
      "loss": 11.1772,
      "step": 163080
    },
    {
      "epoch": 19.62575210589651,
      "grad_norm": Infinity,
      "learning_rate": 0.00018249498886333055,
      "loss": 11.1616,
      "step": 163090
    },
    {
      "epoch": 19.626955475330927,
      "grad_norm": Infinity,
      "learning_rate": 0.0001824928386079787,
      "loss": 11.2318,
      "step": 163100
    },
    {
      "epoch": 19.628158844765345,
      "grad_norm": Infinity,
      "learning_rate": 0.0001824906882332393,
      "loss": 11.1606,
      "step": 163110
    },
    {
      "epoch": 19.62936221419976,
      "grad_norm": Infinity,
      "learning_rate": 0.0001824885377391156,
      "loss": 11.118,
      "step": 163120
    },
    {
      "epoch": 19.630565583634176,
      "grad_norm": Infinity,
      "learning_rate": 0.00018248638712561055,
      "loss": 11.2344,
      "step": 163130
    },
    {
      "epoch": 19.63176895306859,
      "grad_norm": Infinity,
      "learning_rate": 0.00018248423639272737,
      "loss": 11.1924,
      "step": 163140
    },
    {
      "epoch": 19.632972322503008,
      "grad_norm": Infinity,
      "learning_rate": 0.0001824820855404691,
      "loss": 11.1809,
      "step": 163150
    },
    {
      "epoch": 19.634175691937426,
      "grad_norm": Infinity,
      "learning_rate": 0.00018247993456883892,
      "loss": 11.1535,
      "step": 163160
    },
    {
      "epoch": 19.63537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.0001824777834778399,
      "loss": 11.1904,
      "step": 163170
    },
    {
      "epoch": 19.636582430806257,
      "grad_norm": Infinity,
      "learning_rate": 0.00018247563226747517,
      "loss": 11.0601,
      "step": 163180
    },
    {
      "epoch": 19.637785800240675,
      "grad_norm": Infinity,
      "learning_rate": 0.00018247348093774782,
      "loss": 11.246,
      "step": 163190
    },
    {
      "epoch": 19.63898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.000182471329488661,
      "loss": 11.0633,
      "step": 163200
    },
    {
      "epoch": 19.640192539109506,
      "grad_norm": Infinity,
      "learning_rate": 0.0001824691779202178,
      "loss": 10.9845,
      "step": 163210
    },
    {
      "epoch": 19.641395908543924,
      "grad_norm": Infinity,
      "learning_rate": 0.00018246702623242133,
      "loss": 11.2375,
      "step": 163220
    },
    {
      "epoch": 19.642599277978338,
      "grad_norm": Infinity,
      "learning_rate": 0.00018246487442527472,
      "loss": 11.0849,
      "step": 163230
    },
    {
      "epoch": 19.643802647412755,
      "grad_norm": Infinity,
      "learning_rate": 0.00018246272249878106,
      "loss": 11.0735,
      "step": 163240
    },
    {
      "epoch": 19.645006016847173,
      "grad_norm": Infinity,
      "learning_rate": 0.00018246057045294352,
      "loss": 11.0803,
      "step": 163250
    },
    {
      "epoch": 19.646209386281587,
      "grad_norm": Infinity,
      "learning_rate": 0.00018245841828776516,
      "loss": 11.1412,
      "step": 163260
    },
    {
      "epoch": 19.647412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.0001824562660032491,
      "loss": 11.1985,
      "step": 163270
    },
    {
      "epoch": 19.648616125150422,
      "grad_norm": Infinity,
      "learning_rate": 0.00018245411359939847,
      "loss": 11.1095,
      "step": 163280
    },
    {
      "epoch": 19.649819494584836,
      "grad_norm": Infinity,
      "learning_rate": 0.00018245196107621636,
      "loss": 11.2471,
      "step": 163290
    },
    {
      "epoch": 19.651022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.00018244980843370594,
      "loss": 11.1323,
      "step": 163300
    },
    {
      "epoch": 19.65222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.00018244765567187028,
      "loss": 11.2032,
      "step": 163310
    },
    {
      "epoch": 19.653429602888085,
      "grad_norm": Infinity,
      "learning_rate": 0.0001824455027907125,
      "loss": 11.0448,
      "step": 163320
    },
    {
      "epoch": 19.654632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00018244334979023572,
      "loss": 11.1328,
      "step": 163330
    },
    {
      "epoch": 19.65583634175692,
      "grad_norm": Infinity,
      "learning_rate": 0.00018244119667044307,
      "loss": 11.1471,
      "step": 163340
    },
    {
      "epoch": 19.657039711191334,
      "grad_norm": Infinity,
      "learning_rate": 0.00018243904343133763,
      "loss": 11.0798,
      "step": 163350
    },
    {
      "epoch": 19.658243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.00018243689007292258,
      "loss": 11.1756,
      "step": 163360
    },
    {
      "epoch": 19.65944645006017,
      "grad_norm": Infinity,
      "learning_rate": 0.00018243473659520097,
      "loss": 11.1611,
      "step": 163370
    },
    {
      "epoch": 19.660649819494584,
      "grad_norm": Infinity,
      "learning_rate": 0.00018243258299817594,
      "loss": 11.0126,
      "step": 163380
    },
    {
      "epoch": 19.661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00018243042928185063,
      "loss": 11.1627,
      "step": 163390
    },
    {
      "epoch": 19.66305655836342,
      "grad_norm": Infinity,
      "learning_rate": 0.00018242827544622813,
      "loss": 11.1748,
      "step": 163400
    },
    {
      "epoch": 19.664259927797833,
      "grad_norm": Infinity,
      "learning_rate": 0.00018242612149131155,
      "loss": 11.1755,
      "step": 163410
    },
    {
      "epoch": 19.66546329723225,
      "grad_norm": Infinity,
      "learning_rate": 0.00018242396741710402,
      "loss": 11.043,
      "step": 163420
    },
    {
      "epoch": 19.666666666666668,
      "grad_norm": Infinity,
      "learning_rate": 0.00018242181322360867,
      "loss": 11.1601,
      "step": 163430
    },
    {
      "epoch": 19.667870036101082,
      "grad_norm": Infinity,
      "learning_rate": 0.00018241965891082864,
      "loss": 11.2145,
      "step": 163440
    },
    {
      "epoch": 19.6690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.000182417504478767,
      "loss": 11.0432,
      "step": 163450
    },
    {
      "epoch": 19.670276774969917,
      "grad_norm": Infinity,
      "learning_rate": 0.00018241534992742683,
      "loss": 11.1314,
      "step": 163460
    },
    {
      "epoch": 19.67148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00018241319525681135,
      "loss": 11.2229,
      "step": 163470
    },
    {
      "epoch": 19.67268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.00018241104046692362,
      "loss": 11.1409,
      "step": 163480
    },
    {
      "epoch": 19.673886883273166,
      "grad_norm": Infinity,
      "learning_rate": 0.00018240888555776677,
      "loss": 11.1248,
      "step": 163490
    },
    {
      "epoch": 19.67509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.0001824067305293439,
      "loss": 11.1087,
      "step": 163500
    },
    {
      "epoch": 19.676293622141998,
      "grad_norm": Infinity,
      "learning_rate": 0.00018240457538165817,
      "loss": 11.1314,
      "step": 163510
    },
    {
      "epoch": 19.677496991576415,
      "grad_norm": Infinity,
      "learning_rate": 0.00018240242011471268,
      "loss": 11.0941,
      "step": 163520
    },
    {
      "epoch": 19.67870036101083,
      "grad_norm": Infinity,
      "learning_rate": 0.00018240026472851048,
      "loss": 11.2069,
      "step": 163530
    },
    {
      "epoch": 19.679903730445247,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823981092230548,
      "loss": 11.1347,
      "step": 163540
    },
    {
      "epoch": 19.681107099879664,
      "grad_norm": Infinity,
      "learning_rate": 0.00018239595359834874,
      "loss": 11.0686,
      "step": 163550
    },
    {
      "epoch": 19.68231046931408,
      "grad_norm": Infinity,
      "learning_rate": 0.00018239379785439533,
      "loss": 11.029,
      "step": 163560
    },
    {
      "epoch": 19.683513838748496,
      "grad_norm": Infinity,
      "learning_rate": 0.00018239164199119782,
      "loss": 11.1197,
      "step": 163570
    },
    {
      "epoch": 19.684717208182914,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823894860087592,
      "loss": 11.1517,
      "step": 163580
    },
    {
      "epoch": 19.685920577617328,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823873299070827,
      "loss": 11.2753,
      "step": 163590
    },
    {
      "epoch": 19.687123947051745,
      "grad_norm": Infinity,
      "learning_rate": 0.00018238517368617137,
      "loss": 11.2353,
      "step": 163600
    },
    {
      "epoch": 19.688327316486163,
      "grad_norm": Infinity,
      "learning_rate": 0.00018238301734602836,
      "loss": 11.2015,
      "step": 163610
    },
    {
      "epoch": 19.689530685920577,
      "grad_norm": Infinity,
      "learning_rate": 0.00018238086088665675,
      "loss": 11.1367,
      "step": 163620
    },
    {
      "epoch": 19.690734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.00018237870430805976,
      "loss": 11.0568,
      "step": 163630
    },
    {
      "epoch": 19.691937424789412,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823765476102404,
      "loss": 11.1073,
      "step": 163640
    },
    {
      "epoch": 19.693140794223826,
      "grad_norm": Infinity,
      "learning_rate": 0.00018237439079320185,
      "loss": 11.1918,
      "step": 163650
    },
    {
      "epoch": 19.694344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00018237223385694722,
      "loss": 11.1728,
      "step": 163660
    },
    {
      "epoch": 19.69554753309266,
      "grad_norm": Infinity,
      "learning_rate": 0.00018237007680147962,
      "loss": 11.1465,
      "step": 163670
    },
    {
      "epoch": 19.696750902527075,
      "grad_norm": Infinity,
      "learning_rate": 0.00018236791962680218,
      "loss": 11.07,
      "step": 163680
    },
    {
      "epoch": 19.697954271961493,
      "grad_norm": Infinity,
      "learning_rate": 0.00018236576233291804,
      "loss": 11.1767,
      "step": 163690
    },
    {
      "epoch": 19.69915764139591,
      "grad_norm": Infinity,
      "learning_rate": 0.00018236360491983032,
      "loss": 11.0121,
      "step": 163700
    },
    {
      "epoch": 19.700361010830324,
      "grad_norm": Infinity,
      "learning_rate": 0.00018236144738754208,
      "loss": 10.9453,
      "step": 163710
    },
    {
      "epoch": 19.70156438026474,
      "grad_norm": Infinity,
      "learning_rate": 0.00018235928973605654,
      "loss": 11.1796,
      "step": 163720
    },
    {
      "epoch": 19.70276774969916,
      "grad_norm": Infinity,
      "learning_rate": 0.00018235713196537674,
      "loss": 11.1494,
      "step": 163730
    },
    {
      "epoch": 19.703971119133573,
      "grad_norm": Infinity,
      "learning_rate": 0.00018235497407550585,
      "loss": 11.1771,
      "step": 163740
    },
    {
      "epoch": 19.70517448856799,
      "grad_norm": Infinity,
      "learning_rate": 0.000182352816066447,
      "loss": 11.0732,
      "step": 163750
    },
    {
      "epoch": 19.706377858002405,
      "grad_norm": Infinity,
      "learning_rate": 0.00018235065793820328,
      "loss": 11.2246,
      "step": 163760
    },
    {
      "epoch": 19.707581227436823,
      "grad_norm": Infinity,
      "learning_rate": 0.00018234849969077783,
      "loss": 11.187,
      "step": 163770
    },
    {
      "epoch": 19.70878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00018234634132417374,
      "loss": 11.1288,
      "step": 163780
    },
    {
      "epoch": 19.709987966305654,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823441828383942,
      "loss": 11.0761,
      "step": 163790
    },
    {
      "epoch": 19.71119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823420242334423,
      "loss": 11.1201,
      "step": 163800
    },
    {
      "epoch": 19.71239470517449,
      "grad_norm": Infinity,
      "learning_rate": 0.00018233986550932112,
      "loss": 11.1259,
      "step": 163810
    },
    {
      "epoch": 19.713598074608903,
      "grad_norm": Infinity,
      "learning_rate": 0.00018233770666603385,
      "loss": 10.9689,
      "step": 163820
    },
    {
      "epoch": 19.71480144404332,
      "grad_norm": Infinity,
      "learning_rate": 0.00018233554770358357,
      "loss": 11.1964,
      "step": 163830
    },
    {
      "epoch": 19.71600481347774,
      "grad_norm": Infinity,
      "learning_rate": 0.00018233338862197343,
      "loss": 11.1857,
      "step": 163840
    },
    {
      "epoch": 19.717208182912152,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823312294212066,
      "loss": 11.0965,
      "step": 163850
    },
    {
      "epoch": 19.71841155234657,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823290701012861,
      "loss": 11.1701,
      "step": 163860
    },
    {
      "epoch": 19.719614921780988,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823269106622151,
      "loss": 11.1337,
      "step": 163870
    },
    {
      "epoch": 19.7208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00018232475110399677,
      "loss": 11.2088,
      "step": 163880
    },
    {
      "epoch": 19.72202166064982,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823225914266342,
      "loss": 11.0902,
      "step": 163890
    },
    {
      "epoch": 19.723225030084237,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823204316301305,
      "loss": 11.1607,
      "step": 163900
    },
    {
      "epoch": 19.72442839951865,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823182717144888,
      "loss": 11.1018,
      "step": 163910
    },
    {
      "epoch": 19.72563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00018231611167971227,
      "loss": 11.1827,
      "step": 163920
    },
    {
      "epoch": 19.726835138387486,
      "grad_norm": Infinity,
      "learning_rate": 0.000182313951525804,
      "loss": 11.0776,
      "step": 163930
    },
    {
      "epoch": 19.7280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823117912527671,
      "loss": 10.9289,
      "step": 163940
    },
    {
      "epoch": 19.729241877256317,
      "grad_norm": Infinity,
      "learning_rate": 0.00018230963086060471,
      "loss": 11.241,
      "step": 163950
    },
    {
      "epoch": 19.730445246690735,
      "grad_norm": Infinity,
      "learning_rate": 0.00018230747034932,
      "loss": 11.1012,
      "step": 163960
    },
    {
      "epoch": 19.73164861612515,
      "grad_norm": Infinity,
      "learning_rate": 0.00018230530971891602,
      "loss": 11.0606,
      "step": 163970
    },
    {
      "epoch": 19.732851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.00018230314896939596,
      "loss": 11.1034,
      "step": 163980
    },
    {
      "epoch": 19.734055354993984,
      "grad_norm": Infinity,
      "learning_rate": 0.0001823009881007629,
      "loss": 11.1319,
      "step": 163990
    },
    {
      "epoch": 19.735258724428398,
      "grad_norm": Infinity,
      "learning_rate": 0.00018229882711302,
      "loss": 11.1511,
      "step": 164000
    },
    {
      "epoch": 19.736462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.00018229666600617038,
      "loss": 11.0793,
      "step": 164010
    },
    {
      "epoch": 19.737665463297233,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822945047802172,
      "loss": 11.192,
      "step": 164020
    },
    {
      "epoch": 19.738868832731647,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822923434351635,
      "loss": 11.1414,
      "step": 164030
    },
    {
      "epoch": 19.740072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822901819710125,
      "loss": 11.1845,
      "step": 164040
    },
    {
      "epoch": 19.741275571600482,
      "grad_norm": Infinity,
      "learning_rate": 0.00018228802038776725,
      "loss": 11.1298,
      "step": 164050
    },
    {
      "epoch": 19.742478941034896,
      "grad_norm": Infinity,
      "learning_rate": 0.00018228585868543097,
      "loss": 11.2415,
      "step": 164060
    },
    {
      "epoch": 19.743682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822836968640067,
      "loss": 11.0748,
      "step": 164070
    },
    {
      "epoch": 19.74488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822815349234976,
      "loss": 11.0964,
      "step": 164080
    },
    {
      "epoch": 19.746089049338146,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822793728639068,
      "loss": 11.0841,
      "step": 164090
    },
    {
      "epoch": 19.747292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00018227721068523746,
      "loss": 11.0719,
      "step": 164100
    },
    {
      "epoch": 19.74849578820698,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822750483874927,
      "loss": 11.1493,
      "step": 164110
    },
    {
      "epoch": 19.749699157641395,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822728859706756,
      "loss": 11.1783,
      "step": 164120
    },
    {
      "epoch": 19.750902527075812,
      "grad_norm": Infinity,
      "learning_rate": 0.00018227072343478932,
      "loss": 11.1892,
      "step": 164130
    },
    {
      "epoch": 19.75210589651023,
      "grad_norm": Infinity,
      "learning_rate": 0.000182268560779837,
      "loss": 11.0644,
      "step": 164140
    },
    {
      "epoch": 19.753309265944644,
      "grad_norm": Infinity,
      "learning_rate": 0.00018226639800582176,
      "loss": 11.1722,
      "step": 164150
    },
    {
      "epoch": 19.75451263537906,
      "grad_norm": Infinity,
      "learning_rate": 0.00018226423511274672,
      "loss": 11.0826,
      "step": 164160
    },
    {
      "epoch": 19.75571600481348,
      "grad_norm": Infinity,
      "learning_rate": 0.00018226207210061507,
      "loss": 11.1833,
      "step": 164170
    },
    {
      "epoch": 19.756919374247893,
      "grad_norm": Infinity,
      "learning_rate": 0.00018225990896942983,
      "loss": 11.0197,
      "step": 164180
    },
    {
      "epoch": 19.75812274368231,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822577457191942,
      "loss": 11.2286,
      "step": 164190
    },
    {
      "epoch": 19.759326113116728,
      "grad_norm": Infinity,
      "learning_rate": 0.00018225558234991133,
      "loss": 11.1398,
      "step": 164200
    },
    {
      "epoch": 19.760529482551142,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822534188615843,
      "loss": 11.154,
      "step": 164210
    },
    {
      "epoch": 19.76173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00018225125525421628,
      "loss": 11.0459,
      "step": 164220
    },
    {
      "epoch": 19.762936221419977,
      "grad_norm": Infinity,
      "learning_rate": 0.00018224909152781042,
      "loss": 11.0892,
      "step": 164230
    },
    {
      "epoch": 19.76413959085439,
      "grad_norm": Infinity,
      "learning_rate": 0.00018224692768236978,
      "loss": 11.1035,
      "step": 164240
    },
    {
      "epoch": 19.76534296028881,
      "grad_norm": Infinity,
      "learning_rate": 0.00018224476371789754,
      "loss": 11.0743,
      "step": 164250
    },
    {
      "epoch": 19.766546329723226,
      "grad_norm": Infinity,
      "learning_rate": 0.00018224259963439682,
      "loss": 11.0997,
      "step": 164260
    },
    {
      "epoch": 19.76774969915764,
      "grad_norm": Infinity,
      "learning_rate": 0.00018224043543187073,
      "loss": 11.0573,
      "step": 164270
    },
    {
      "epoch": 19.768953068592058,
      "grad_norm": Infinity,
      "learning_rate": 0.00018223827111032246,
      "loss": 11.1615,
      "step": 164280
    },
    {
      "epoch": 19.770156438026476,
      "grad_norm": Infinity,
      "learning_rate": 0.00018223610666975507,
      "loss": 11.2072,
      "step": 164290
    },
    {
      "epoch": 19.77135980746089,
      "grad_norm": Infinity,
      "learning_rate": 0.00018223394211017176,
      "loss": 11.0979,
      "step": 164300
    },
    {
      "epoch": 19.772563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00018223177743157562,
      "loss": 11.1511,
      "step": 164310
    },
    {
      "epoch": 19.773766546329725,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822296126339698,
      "loss": 11.1447,
      "step": 164320
    },
    {
      "epoch": 19.77496991576414,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822274477173574,
      "loss": 11.2087,
      "step": 164330
    },
    {
      "epoch": 19.776173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00018222528268174165,
      "loss": 11.1231,
      "step": 164340
    },
    {
      "epoch": 19.777376654632974,
      "grad_norm": Infinity,
      "learning_rate": 0.00018222311752712558,
      "loss": 11.1377,
      "step": 164350
    },
    {
      "epoch": 19.778580024067388,
      "grad_norm": Infinity,
      "learning_rate": 0.00018222095225351234,
      "loss": 11.1396,
      "step": 164360
    },
    {
      "epoch": 19.779783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822187868609051,
      "loss": 11.2541,
      "step": 164370
    },
    {
      "epoch": 19.780986762936223,
      "grad_norm": Infinity,
      "learning_rate": 0.00018221662134930696,
      "loss": 11.1231,
      "step": 164380
    },
    {
      "epoch": 19.782190132370637,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822144557187211,
      "loss": 11.1657,
      "step": 164390
    },
    {
      "epoch": 19.783393501805055,
      "grad_norm": Infinity,
      "learning_rate": 0.00018221228996915062,
      "loss": 11.199,
      "step": 164400
    },
    {
      "epoch": 19.784596871239472,
      "grad_norm": Infinity,
      "learning_rate": 0.00018221012410059865,
      "loss": 11.2341,
      "step": 164410
    },
    {
      "epoch": 19.785800240673886,
      "grad_norm": Infinity,
      "learning_rate": 0.0001822079581130683,
      "loss": 11.1467,
      "step": 164420
    },
    {
      "epoch": 19.787003610108304,
      "grad_norm": Infinity,
      "learning_rate": 0.00018220579200656277,
      "loss": 11.1569,
      "step": 164430
    },
    {
      "epoch": 19.78820697954272,
      "grad_norm": Infinity,
      "learning_rate": 0.00018220362578108516,
      "loss": 11.0901,
      "step": 164440
    },
    {
      "epoch": 19.789410348977135,
      "grad_norm": Infinity,
      "learning_rate": 0.00018220145943663864,
      "loss": 11.1769,
      "step": 164450
    },
    {
      "epoch": 19.790613718411553,
      "grad_norm": Infinity,
      "learning_rate": 0.00018219929297322627,
      "loss": 11.0786,
      "step": 164460
    },
    {
      "epoch": 19.79181708784597,
      "grad_norm": Infinity,
      "learning_rate": 0.00018219712639085123,
      "loss": 11.1887,
      "step": 164470
    },
    {
      "epoch": 19.793020457280385,
      "grad_norm": Infinity,
      "learning_rate": 0.00018219495968951668,
      "loss": 11.1999,
      "step": 164480
    },
    {
      "epoch": 19.794223826714802,
      "grad_norm": Infinity,
      "learning_rate": 0.0001821927928692257,
      "loss": 11.2034,
      "step": 164490
    },
    {
      "epoch": 19.79542719614922,
      "grad_norm": Infinity,
      "learning_rate": 0.00018219062592998147,
      "loss": 11.1517,
      "step": 164500
    },
    {
      "epoch": 19.796630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00018218845887178712,
      "loss": 11.1718,
      "step": 164510
    },
    {
      "epoch": 19.79783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.00018218629169464577,
      "loss": 11.1323,
      "step": 164520
    },
    {
      "epoch": 19.799037304452465,
      "grad_norm": Infinity,
      "learning_rate": 0.00018218412439856057,
      "loss": 11.2685,
      "step": 164530
    },
    {
      "epoch": 19.800240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.00018218195698353464,
      "loss": 11.199,
      "step": 164540
    },
    {
      "epoch": 19.8014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.00018217978944957115,
      "loss": 11.2102,
      "step": 164550
    },
    {
      "epoch": 19.802647412755714,
      "grad_norm": Infinity,
      "learning_rate": 0.00018217762179667318,
      "loss": 11.0861,
      "step": 164560
    },
    {
      "epoch": 19.803850782190132,
      "grad_norm": Infinity,
      "learning_rate": 0.00018217545402484392,
      "loss": 11.1543,
      "step": 164570
    },
    {
      "epoch": 19.80505415162455,
      "grad_norm": Infinity,
      "learning_rate": 0.0001821732861340865,
      "loss": 11.3454,
      "step": 164580
    },
    {
      "epoch": 19.806257521058964,
      "grad_norm": Infinity,
      "learning_rate": 0.00018217111812440404,
      "loss": 11.17,
      "step": 164590
    },
    {
      "epoch": 19.80746089049338,
      "grad_norm": Infinity,
      "learning_rate": 0.00018216894999579969,
      "loss": 11.07,
      "step": 164600
    },
    {
      "epoch": 19.8086642599278,
      "grad_norm": Infinity,
      "learning_rate": 0.00018216678174827656,
      "loss": 11.1024,
      "step": 164610
    },
    {
      "epoch": 19.809867629362213,
      "grad_norm": Infinity,
      "learning_rate": 0.00018216461338183785,
      "loss": 11.1108,
      "step": 164620
    },
    {
      "epoch": 19.81107099879663,
      "grad_norm": Infinity,
      "learning_rate": 0.00018216244489648663,
      "loss": 11.1018,
      "step": 164630
    },
    {
      "epoch": 19.812274368231048,
      "grad_norm": Infinity,
      "learning_rate": 0.00018216027629222608,
      "loss": 11.2308,
      "step": 164640
    },
    {
      "epoch": 19.813477737665462,
      "grad_norm": Infinity,
      "learning_rate": 0.00018215810756905932,
      "loss": 10.9164,
      "step": 164650
    },
    {
      "epoch": 19.81468110709988,
      "grad_norm": Infinity,
      "learning_rate": 0.0001821559387269895,
      "loss": 11.0672,
      "step": 164660
    },
    {
      "epoch": 19.815884476534297,
      "grad_norm": Infinity,
      "learning_rate": 0.00018215376976601975,
      "loss": 11.1489,
      "step": 164670
    },
    {
      "epoch": 19.81708784596871,
      "grad_norm": Infinity,
      "learning_rate": 0.00018215160068615321,
      "loss": 11.0974,
      "step": 164680
    },
    {
      "epoch": 19.81829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.000182149431487393,
      "loss": 11.1122,
      "step": 164690
    },
    {
      "epoch": 19.819494584837546,
      "grad_norm": Infinity,
      "learning_rate": 0.00018214726216974229,
      "loss": 11.0882,
      "step": 164700
    },
    {
      "epoch": 19.82069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00018214509273320424,
      "loss": 11.1044,
      "step": 164710
    },
    {
      "epoch": 19.821901323706378,
      "grad_norm": Infinity,
      "learning_rate": 0.00018214292317778193,
      "loss": 11.2734,
      "step": 164720
    },
    {
      "epoch": 19.823104693140795,
      "grad_norm": Infinity,
      "learning_rate": 0.00018214075350347854,
      "loss": 11.1612,
      "step": 164730
    },
    {
      "epoch": 19.82430806257521,
      "grad_norm": Infinity,
      "learning_rate": 0.00018213858371029723,
      "loss": 11.087,
      "step": 164740
    },
    {
      "epoch": 19.825511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00018213641379824104,
      "loss": 11.0612,
      "step": 164750
    },
    {
      "epoch": 19.826714801444044,
      "grad_norm": Infinity,
      "learning_rate": 0.00018213424376731322,
      "loss": 11.096,
      "step": 164760
    },
    {
      "epoch": 19.82791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.0001821320736175169,
      "loss": 11.2982,
      "step": 164770
    },
    {
      "epoch": 19.829121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00018212990334885514,
      "loss": 11.2142,
      "step": 164780
    },
    {
      "epoch": 19.830324909747294,
      "grad_norm": Infinity,
      "learning_rate": 0.00018212773296133117,
      "loss": 11.2466,
      "step": 164790
    },
    {
      "epoch": 19.831528279181708,
      "grad_norm": Infinity,
      "learning_rate": 0.00018212556245494806,
      "loss": 11.0668,
      "step": 164800
    },
    {
      "epoch": 19.832731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.00018212339182970902,
      "loss": 11.0883,
      "step": 164810
    },
    {
      "epoch": 19.833935018050543,
      "grad_norm": Infinity,
      "learning_rate": 0.00018212122108561714,
      "loss": 11.1541,
      "step": 164820
    },
    {
      "epoch": 19.835138387484957,
      "grad_norm": Infinity,
      "learning_rate": 0.00018211905022267558,
      "loss": 11.1137,
      "step": 164830
    },
    {
      "epoch": 19.836341756919374,
      "grad_norm": Infinity,
      "learning_rate": 0.00018211687924088747,
      "loss": 11.0337,
      "step": 164840
    },
    {
      "epoch": 19.837545126353792,
      "grad_norm": Infinity,
      "learning_rate": 0.00018211470814025597,
      "loss": 11.083,
      "step": 164850
    },
    {
      "epoch": 19.838748495788206,
      "grad_norm": Infinity,
      "learning_rate": 0.0001821125369207842,
      "loss": 11.278,
      "step": 164860
    },
    {
      "epoch": 19.839951865222623,
      "grad_norm": Infinity,
      "learning_rate": 0.00018211036558247537,
      "loss": 11.2,
      "step": 164870
    },
    {
      "epoch": 19.84115523465704,
      "grad_norm": Infinity,
      "learning_rate": 0.0001821081941253325,
      "loss": 11.1798,
      "step": 164880
    },
    {
      "epoch": 19.842358604091455,
      "grad_norm": Infinity,
      "learning_rate": 0.00018210602254935886,
      "loss": 11.2188,
      "step": 164890
    },
    {
      "epoch": 19.843561973525873,
      "grad_norm": Infinity,
      "learning_rate": 0.00018210385085455747,
      "loss": 11.0565,
      "step": 164900
    },
    {
      "epoch": 19.84476534296029,
      "grad_norm": Infinity,
      "learning_rate": 0.00018210167904093158,
      "loss": 11.0445,
      "step": 164910
    },
    {
      "epoch": 19.845968712394704,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820995071084843,
      "loss": 11.0802,
      "step": 164920
    },
    {
      "epoch": 19.84717208182912,
      "grad_norm": Infinity,
      "learning_rate": 0.00018209733505721872,
      "loss": 11.2569,
      "step": 164930
    },
    {
      "epoch": 19.84837545126354,
      "grad_norm": Infinity,
      "learning_rate": 0.00018209516288713806,
      "loss": 11.2663,
      "step": 164940
    },
    {
      "epoch": 19.849578820697953,
      "grad_norm": Infinity,
      "learning_rate": 0.00018209299059824543,
      "loss": 11.1048,
      "step": 164950
    },
    {
      "epoch": 19.85078219013237,
      "grad_norm": Infinity,
      "learning_rate": 0.00018209081819054395,
      "loss": 11.2474,
      "step": 164960
    },
    {
      "epoch": 19.85198555956679,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820886456640368,
      "loss": 11.2436,
      "step": 164970
    },
    {
      "epoch": 19.853188929001202,
      "grad_norm": Infinity,
      "learning_rate": 0.00018208647301872714,
      "loss": 11.1079,
      "step": 164980
    },
    {
      "epoch": 19.85439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00018208430025461807,
      "loss": 11.0617,
      "step": 164990
    },
    {
      "epoch": 19.855595667870038,
      "grad_norm": Infinity,
      "learning_rate": 0.00018208212737171273,
      "loss": 11.1149,
      "step": 165000
    },
    {
      "epoch": 19.85679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820799543700143,
      "loss": 11.203,
      "step": 165010
    },
    {
      "epoch": 19.85800240673887,
      "grad_norm": Infinity,
      "learning_rate": 0.00018207778124952593,
      "loss": 11.2121,
      "step": 165020
    },
    {
      "epoch": 19.859205776173287,
      "grad_norm": Infinity,
      "learning_rate": 0.00018207560801025073,
      "loss": 11.0755,
      "step": 165030
    },
    {
      "epoch": 19.8604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00018207343465219186,
      "loss": 11.1553,
      "step": 165040
    },
    {
      "epoch": 19.86161251504212,
      "grad_norm": Infinity,
      "learning_rate": 0.00018207126117535246,
      "loss": 11.0546,
      "step": 165050
    },
    {
      "epoch": 19.862815884476536,
      "grad_norm": Infinity,
      "learning_rate": 0.00018206908757973568,
      "loss": 11.2138,
      "step": 165060
    },
    {
      "epoch": 19.86401925391095,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820669138653447,
      "loss": 11.0992,
      "step": 165070
    },
    {
      "epoch": 19.865222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.00018206474003218263,
      "loss": 11.1009,
      "step": 165080
    },
    {
      "epoch": 19.866425992779785,
      "grad_norm": Infinity,
      "learning_rate": 0.00018206256608025258,
      "loss": 11.1346,
      "step": 165090
    },
    {
      "epoch": 19.8676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00018206039200955776,
      "loss": 11.1552,
      "step": 165100
    },
    {
      "epoch": 19.868832731648617,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820582178201013,
      "loss": 11.1733,
      "step": 165110
    },
    {
      "epoch": 19.870036101083034,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820560435118863,
      "loss": 11.1219,
      "step": 165120
    },
    {
      "epoch": 19.871239470517448,
      "grad_norm": Infinity,
      "learning_rate": 0.00018205386908491599,
      "loss": 11.1515,
      "step": 165130
    },
    {
      "epoch": 19.872442839951866,
      "grad_norm": Infinity,
      "learning_rate": 0.00018205169453919346,
      "loss": 11.2341,
      "step": 165140
    },
    {
      "epoch": 19.87364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.00018204951987472188,
      "loss": 11.0963,
      "step": 165150
    },
    {
      "epoch": 19.874849578820697,
      "grad_norm": Infinity,
      "learning_rate": 0.00018204734509150438,
      "loss": 11.0325,
      "step": 165160
    },
    {
      "epoch": 19.876052948255115,
      "grad_norm": Infinity,
      "learning_rate": 0.00018204517018954413,
      "loss": 11.2082,
      "step": 165170
    },
    {
      "epoch": 19.87725631768953,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820429951688442,
      "loss": 11.1202,
      "step": 165180
    },
    {
      "epoch": 19.878459687123947,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820408200294079,
      "loss": 11.1797,
      "step": 165190
    },
    {
      "epoch": 19.879663056558364,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820386447712382,
      "loss": 11.2218,
      "step": 165200
    },
    {
      "epoch": 19.880866425992778,
      "grad_norm": Infinity,
      "learning_rate": 0.00018203646939433834,
      "loss": 11.2033,
      "step": 165210
    },
    {
      "epoch": 19.882069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00018203429389871146,
      "loss": 10.9957,
      "step": 165220
    },
    {
      "epoch": 19.883273164861613,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820321182843607,
      "loss": 11.1282,
      "step": 165230
    },
    {
      "epoch": 19.884476534296027,
      "grad_norm": Infinity,
      "learning_rate": 0.00018202994255128924,
      "loss": 11.2358,
      "step": 165240
    },
    {
      "epoch": 19.885679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00018202776669950016,
      "loss": 11.0358,
      "step": 165250
    },
    {
      "epoch": 19.886883273164862,
      "grad_norm": Infinity,
      "learning_rate": 0.00018202559072899666,
      "loss": 11.1336,
      "step": 165260
    },
    {
      "epoch": 19.888086642599276,
      "grad_norm": Infinity,
      "learning_rate": 0.0001820234146397819,
      "loss": 11.1526,
      "step": 165270
    },
    {
      "epoch": 19.889290012033694,
      "grad_norm": Infinity,
      "learning_rate": 0.000182021238431859,
      "loss": 11.2606,
      "step": 165280
    },
    {
      "epoch": 19.89049338146811,
      "grad_norm": Infinity,
      "learning_rate": 0.00018201906210523112,
      "loss": 11.1386,
      "step": 165290
    },
    {
      "epoch": 19.891696750902526,
      "grad_norm": Infinity,
      "learning_rate": 0.00018201688565990142,
      "loss": 11.1824,
      "step": 165300
    },
    {
      "epoch": 19.892900120336943,
      "grad_norm": Infinity,
      "learning_rate": 0.00018201470909587298,
      "loss": 11.1232,
      "step": 165310
    },
    {
      "epoch": 19.89410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00018201253241314905,
      "loss": 11.0581,
      "step": 165320
    },
    {
      "epoch": 19.895306859205775,
      "grad_norm": Infinity,
      "learning_rate": 0.00018201035561173274,
      "loss": 11.0982,
      "step": 165330
    },
    {
      "epoch": 19.896510228640192,
      "grad_norm": Infinity,
      "learning_rate": 0.00018200817869162722,
      "loss": 11.1737,
      "step": 165340
    },
    {
      "epoch": 19.89771359807461,
      "grad_norm": Infinity,
      "learning_rate": 0.00018200600165283558,
      "loss": 11.0467,
      "step": 165350
    },
    {
      "epoch": 19.898916967509024,
      "grad_norm": Infinity,
      "learning_rate": 0.000182003824495361,
      "loss": 11.0762,
      "step": 165360
    },
    {
      "epoch": 19.90012033694344,
      "grad_norm": Infinity,
      "learning_rate": 0.00018200164721920666,
      "loss": 11.1541,
      "step": 165370
    },
    {
      "epoch": 19.90132370637786,
      "grad_norm": Infinity,
      "learning_rate": 0.0001819994698243757,
      "loss": 11.0707,
      "step": 165380
    },
    {
      "epoch": 19.902527075812273,
      "grad_norm": Infinity,
      "learning_rate": 0.00018199729231087124,
      "loss": 11.2656,
      "step": 165390
    },
    {
      "epoch": 19.90373044524669,
      "grad_norm": Infinity,
      "learning_rate": 0.00018199511467869645,
      "loss": 11.1462,
      "step": 165400
    },
    {
      "epoch": 19.904933814681108,
      "grad_norm": Infinity,
      "learning_rate": 0.00018199293692785452,
      "loss": 11.1108,
      "step": 165410
    },
    {
      "epoch": 19.906137184115522,
      "grad_norm": Infinity,
      "learning_rate": 0.00018199075905834855,
      "loss": 11.2118,
      "step": 165420
    },
    {
      "epoch": 19.90734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.0001819885810701817,
      "loss": 11.2167,
      "step": 165430
    },
    {
      "epoch": 19.908543922984357,
      "grad_norm": Infinity,
      "learning_rate": 0.00018198640296335713,
      "loss": 11.2291,
      "step": 165440
    },
    {
      "epoch": 19.90974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.000181984224737878,
      "loss": 11.2354,
      "step": 165450
    },
    {
      "epoch": 19.91095066185319,
      "grad_norm": Infinity,
      "learning_rate": 0.00018198204639374745,
      "loss": 11.2053,
      "step": 165460
    },
    {
      "epoch": 19.912154031287606,
      "grad_norm": Infinity,
      "learning_rate": 0.00018197986793096863,
      "loss": 11.1204,
      "step": 165470
    },
    {
      "epoch": 19.91335740072202,
      "grad_norm": Infinity,
      "learning_rate": 0.00018197768934954473,
      "loss": 11.146,
      "step": 165480
    },
    {
      "epoch": 19.914560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00018197551064947886,
      "loss": 11.1402,
      "step": 165490
    },
    {
      "epoch": 19.915764139590856,
      "grad_norm": Infinity,
      "learning_rate": 0.0001819733318307742,
      "loss": 11.1543,
      "step": 165500
    },
    {
      "epoch": 19.91696750902527,
      "grad_norm": Infinity,
      "learning_rate": 0.00018197115289343384,
      "loss": 11.1014,
      "step": 165510
    },
    {
      "epoch": 19.918170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00018196897383746102,
      "loss": 11.1375,
      "step": 165520
    },
    {
      "epoch": 19.919374247894105,
      "grad_norm": Infinity,
      "learning_rate": 0.00018196679466285888,
      "loss": 11.2382,
      "step": 165530
    },
    {
      "epoch": 19.92057761732852,
      "grad_norm": Infinity,
      "learning_rate": 0.00018196461536963053,
      "loss": 11.0677,
      "step": 165540
    },
    {
      "epoch": 19.921780986762936,
      "grad_norm": Infinity,
      "learning_rate": 0.00018196243595777913,
      "loss": 11.1583,
      "step": 165550
    },
    {
      "epoch": 19.922984356197354,
      "grad_norm": Infinity,
      "learning_rate": 0.00018196025642730788,
      "loss": 11.2589,
      "step": 165560
    },
    {
      "epoch": 19.924187725631768,
      "grad_norm": Infinity,
      "learning_rate": 0.00018195807677821987,
      "loss": 11.2583,
      "step": 165570
    },
    {
      "epoch": 19.925391095066185,
      "grad_norm": Infinity,
      "learning_rate": 0.00018195589701051833,
      "loss": 11.0953,
      "step": 165580
    },
    {
      "epoch": 19.926594464500603,
      "grad_norm": Infinity,
      "learning_rate": 0.00018195371712420631,
      "loss": 11.2733,
      "step": 165590
    },
    {
      "epoch": 19.927797833935017,
      "grad_norm": Infinity,
      "learning_rate": 0.00018195153711928708,
      "loss": 11.1126,
      "step": 165600
    },
    {
      "epoch": 19.929001203369435,
      "grad_norm": Infinity,
      "learning_rate": 0.00018194935699576372,
      "loss": 11.2585,
      "step": 165610
    },
    {
      "epoch": 19.930204572803852,
      "grad_norm": Infinity,
      "learning_rate": 0.0001819471767536394,
      "loss": 11.1855,
      "step": 165620
    },
    {
      "epoch": 19.931407942238266,
      "grad_norm": Infinity,
      "learning_rate": 0.0001819449963929173,
      "loss": 11.1656,
      "step": 165630
    },
    {
      "epoch": 19.932611311672684,
      "grad_norm": Infinity,
      "learning_rate": 0.00018194281591360056,
      "loss": 11.1314,
      "step": 165640
    },
    {
      "epoch": 19.9338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00018194063531569234,
      "loss": 11.1184,
      "step": 165650
    },
    {
      "epoch": 19.935018050541515,
      "grad_norm": Infinity,
      "learning_rate": 0.00018193845459919578,
      "loss": 11.184,
      "step": 165660
    },
    {
      "epoch": 19.936221419975933,
      "grad_norm": Infinity,
      "learning_rate": 0.00018193627376411403,
      "loss": 11.1132,
      "step": 165670
    },
    {
      "epoch": 19.93742478941035,
      "grad_norm": Infinity,
      "learning_rate": 0.0001819340928104503,
      "loss": 11.0606,
      "step": 165680
    },
    {
      "epoch": 19.938628158844764,
      "grad_norm": Infinity,
      "learning_rate": 0.00018193191173820767,
      "loss": 11.1,
      "step": 165690
    },
    {
      "epoch": 19.939831528279182,
      "grad_norm": Infinity,
      "learning_rate": 0.00018192973054738935,
      "loss": 11.0695,
      "step": 165700
    },
    {
      "epoch": 19.9410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.0001819275492379985,
      "loss": 11.0693,
      "step": 165710
    },
    {
      "epoch": 19.942238267148014,
      "grad_norm": Infinity,
      "learning_rate": 0.00018192536781003825,
      "loss": 11.1839,
      "step": 165720
    },
    {
      "epoch": 19.94344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00018192318626351173,
      "loss": 11.1644,
      "step": 165730
    },
    {
      "epoch": 19.94464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00018192100459842218,
      "loss": 11.2392,
      "step": 165740
    },
    {
      "epoch": 19.945848375451263,
      "grad_norm": Infinity,
      "learning_rate": 0.0001819188228147727,
      "loss": 11.0428,
      "step": 165750
    },
    {
      "epoch": 19.94705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00018191664091256647,
      "loss": 11.0952,
      "step": 165760
    },
    {
      "epoch": 19.948255114320098,
      "grad_norm": Infinity,
      "learning_rate": 0.00018191445889180661,
      "loss": 11.254,
      "step": 165770
    },
    {
      "epoch": 19.949458483754512,
      "grad_norm": Infinity,
      "learning_rate": 0.0001819122767524963,
      "loss": 11.123,
      "step": 165780
    },
    {
      "epoch": 19.95066185318893,
      "grad_norm": Infinity,
      "learning_rate": 0.00018191009449463873,
      "loss": 11.2211,
      "step": 165790
    },
    {
      "epoch": 19.951865222623347,
      "grad_norm": Infinity,
      "learning_rate": 0.000181907912118237,
      "loss": 11.1385,
      "step": 165800
    },
    {
      "epoch": 19.95306859205776,
      "grad_norm": Infinity,
      "learning_rate": 0.00018190572962329432,
      "loss": 11.0776,
      "step": 165810
    },
    {
      "epoch": 19.95427196149218,
      "grad_norm": Infinity,
      "learning_rate": 0.00018190354700981382,
      "loss": 11.1519,
      "step": 165820
    },
    {
      "epoch": 19.955475330926596,
      "grad_norm": Infinity,
      "learning_rate": 0.00018190136427779867,
      "loss": 11.1281,
      "step": 165830
    },
    {
      "epoch": 19.95667870036101,
      "grad_norm": Infinity,
      "learning_rate": 0.000181899181427252,
      "loss": 11.1464,
      "step": 165840
    },
    {
      "epoch": 19.957882069795428,
      "grad_norm": Infinity,
      "learning_rate": 0.00018189699845817705,
      "loss": 11.1463,
      "step": 165850
    },
    {
      "epoch": 19.959085439229845,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818948153705769,
      "loss": 11.1786,
      "step": 165860
    },
    {
      "epoch": 19.96028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818926321644547,
      "loss": 11.2121,
      "step": 165870
    },
    {
      "epoch": 19.961492178098677,
      "grad_norm": Infinity,
      "learning_rate": 0.00018189044883981366,
      "loss": 11.2523,
      "step": 165880
    },
    {
      "epoch": 19.96269554753309,
      "grad_norm": Infinity,
      "learning_rate": 0.00018188826539665694,
      "loss": 11.1515,
      "step": 165890
    },
    {
      "epoch": 19.96389891696751,
      "grad_norm": Infinity,
      "learning_rate": 0.00018188608183498768,
      "loss": 11.0813,
      "step": 165900
    },
    {
      "epoch": 19.965102286401926,
      "grad_norm": Infinity,
      "learning_rate": 0.00018188389815480902,
      "loss": 11.0842,
      "step": 165910
    },
    {
      "epoch": 19.96630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00018188171435612414,
      "loss": 11.2649,
      "step": 165920
    },
    {
      "epoch": 19.967509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.00018187953043893624,
      "loss": 11.0825,
      "step": 165930
    },
    {
      "epoch": 19.968712394705175,
      "grad_norm": Infinity,
      "learning_rate": 0.00018187734640324842,
      "loss": 11.1482,
      "step": 165940
    },
    {
      "epoch": 19.96991576413959,
      "grad_norm": Infinity,
      "learning_rate": 0.00018187516224906386,
      "loss": 11.0872,
      "step": 165950
    },
    {
      "epoch": 19.971119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.00018187297797638574,
      "loss": 11.2438,
      "step": 165960
    },
    {
      "epoch": 19.972322503008424,
      "grad_norm": Infinity,
      "learning_rate": 0.00018187079358521723,
      "loss": 11.0543,
      "step": 165970
    },
    {
      "epoch": 19.97352587244284,
      "grad_norm": Infinity,
      "learning_rate": 0.00018186860907556143,
      "loss": 11.1657,
      "step": 165980
    },
    {
      "epoch": 19.974729241877256,
      "grad_norm": Infinity,
      "learning_rate": 0.00018186642444742153,
      "loss": 11.2358,
      "step": 165990
    },
    {
      "epoch": 19.975932611311674,
      "grad_norm": Infinity,
      "learning_rate": 0.00018186423970080074,
      "loss": 11.0724,
      "step": 166000
    },
    {
      "epoch": 19.977135980746088,
      "grad_norm": Infinity,
      "learning_rate": 0.00018186205483570214,
      "loss": 11.162,
      "step": 166010
    },
    {
      "epoch": 19.978339350180505,
      "grad_norm": Infinity,
      "learning_rate": 0.00018185986985212897,
      "loss": 11.1845,
      "step": 166020
    },
    {
      "epoch": 19.979542719614923,
      "grad_norm": Infinity,
      "learning_rate": 0.00018185768475008436,
      "loss": 11.2258,
      "step": 166030
    },
    {
      "epoch": 19.980746089049337,
      "grad_norm": Infinity,
      "learning_rate": 0.00018185549952957146,
      "loss": 11.2137,
      "step": 166040
    },
    {
      "epoch": 19.981949458483754,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818533141905934,
      "loss": 11.196,
      "step": 166050
    },
    {
      "epoch": 19.983152827918172,
      "grad_norm": Infinity,
      "learning_rate": 0.00018185112873315347,
      "loss": 11.1218,
      "step": 166060
    },
    {
      "epoch": 19.984356197352586,
      "grad_norm": Infinity,
      "learning_rate": 0.00018184894315725469,
      "loss": 11.2698,
      "step": 166070
    },
    {
      "epoch": 19.985559566787003,
      "grad_norm": Infinity,
      "learning_rate": 0.00018184675746290028,
      "loss": 11.117,
      "step": 166080
    },
    {
      "epoch": 19.98676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00018184457165009343,
      "loss": 11.1865,
      "step": 166090
    },
    {
      "epoch": 19.987966305655835,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818423857188373,
      "loss": 11.182,
      "step": 166100
    },
    {
      "epoch": 19.989169675090253,
      "grad_norm": Infinity,
      "learning_rate": 0.00018184019966913497,
      "loss": 11.0618,
      "step": 166110
    },
    {
      "epoch": 19.99037304452467,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818380135009897,
      "loss": 11.3382,
      "step": 166120
    },
    {
      "epoch": 19.991576413959084,
      "grad_norm": Infinity,
      "learning_rate": 0.00018183582721440462,
      "loss": 11.1709,
      "step": 166130
    },
    {
      "epoch": 19.9927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818336408093829,
      "loss": 11.1113,
      "step": 166140
    },
    {
      "epoch": 19.99398315282792,
      "grad_norm": Infinity,
      "learning_rate": 0.00018183145428592767,
      "loss": 11.0988,
      "step": 166150
    },
    {
      "epoch": 19.995186522262333,
      "grad_norm": Infinity,
      "learning_rate": 0.00018182926764404215,
      "loss": 11.1319,
      "step": 166160
    },
    {
      "epoch": 19.99638989169675,
      "grad_norm": Infinity,
      "learning_rate": 0.00018182708088372945,
      "loss": 11.1248,
      "step": 166170
    },
    {
      "epoch": 19.99759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00018182489400499278,
      "loss": 11.1114,
      "step": 166180
    },
    {
      "epoch": 19.998796630565582,
      "grad_norm": Infinity,
      "learning_rate": 0.00018182270700783527,
      "loss": 11.0689,
      "step": 166190
    },
    {
      "epoch": 20.0,
      "grad_norm": Infinity,
      "learning_rate": 0.00018182051989226015,
      "loss": 11.1714,
      "step": 166200
    },
    {
      "epoch": 20.0,
      "eval_loss": 11.139604568481445,
      "eval_runtime": 118.6697,
      "eval_samples_per_second": 62.248,
      "eval_steps_per_second": 7.786,
      "step": 166200
    },
    {
      "epoch": 20.001203369434418,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818183326582705,
      "loss": 11.069,
      "step": 166210
    },
    {
      "epoch": 20.00240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818161453058695,
      "loss": 11.1056,
      "step": 166220
    },
    {
      "epoch": 20.00361010830325,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818139578350604,
      "loss": 11.0334,
      "step": 166230
    },
    {
      "epoch": 20.004813477737667,
      "grad_norm": Infinity,
      "learning_rate": 0.00018181177024584623,
      "loss": 11.2063,
      "step": 166240
    },
    {
      "epoch": 20.00601684717208,
      "grad_norm": Infinity,
      "learning_rate": 0.00018180958253823028,
      "loss": 11.1498,
      "step": 166250
    },
    {
      "epoch": 20.0072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00018180739471221563,
      "loss": 11.19,
      "step": 166260
    },
    {
      "epoch": 20.008423586040916,
      "grad_norm": Infinity,
      "learning_rate": 0.0001818052067678055,
      "loss": 11.0813,
      "step": 166270
    },
    {
      "epoch": 20.00962695547533,
      "grad_norm": Infinity,
      "learning_rate": 0.00018180301870500304,
      "loss": 11.1756,
      "step": 166280
    },
    {
      "epoch": 20.010830324909747,
      "grad_norm": Infinity,
      "learning_rate": 0.00018180083052381143,
      "loss": 11.1816,
      "step": 166290
    },
    {
      "epoch": 20.012033694344165,
      "grad_norm": Infinity,
      "learning_rate": 0.0001817986422242338,
      "loss": 11.1397,
      "step": 166300
    },
    {
      "epoch": 20.01323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00018179645380627337,
      "loss": 11.0872,
      "step": 166310
    },
    {
      "epoch": 20.014440433212997,
      "grad_norm": Infinity,
      "learning_rate": 0.00018179426526993326,
      "loss": 11.0543,
      "step": 166320
    },
    {
      "epoch": 20.015643802647414,
      "grad_norm": Infinity,
      "learning_rate": 0.00018179207661521665,
      "loss": 11.1851,
      "step": 166330
    },
    {
      "epoch": 20.016847172081828,
      "grad_norm": Infinity,
      "learning_rate": 0.00018178988784212669,
      "loss": 11.2014,
      "step": 166340
    },
    {
      "epoch": 20.018050541516246,
      "grad_norm": Infinity,
      "learning_rate": 0.00018178769895066662,
      "loss": 11.0478,
      "step": 166350
    },
    {
      "epoch": 20.019253910950663,
      "grad_norm": Infinity,
      "learning_rate": 0.0001817855099408395,
      "loss": 11.1208,
      "step": 166360
    },
    {
      "epoch": 20.020457280385077,
      "grad_norm": Infinity,
      "learning_rate": 0.00018178332081264859,
      "loss": 10.9829,
      "step": 166370
    },
    {
      "epoch": 20.021660649819495,
      "grad_norm": Infinity,
      "learning_rate": 0.00018178113156609702,
      "loss": 11.2705,
      "step": 166380
    },
    {
      "epoch": 20.022864019253912,
      "grad_norm": Infinity,
      "learning_rate": 0.00018177894220118795,
      "loss": 11.1191,
      "step": 166390
    },
    {
      "epoch": 20.024067388688326,
      "grad_norm": Infinity,
      "learning_rate": 0.0001817767527179246,
      "loss": 11.0681,
      "step": 166400
    },
    {
      "epoch": 20.025270758122744,
      "grad_norm": Infinity,
      "learning_rate": 0.00018177456311631004,
      "loss": 11.1591,
      "step": 166410
    },
    {
      "epoch": 20.02647412755716,
      "grad_norm": Infinity,
      "learning_rate": 0.00018177237339634753,
      "loss": 11.2164,
      "step": 166420
    },
    {
      "epoch": 20.027677496991576,
      "grad_norm": Infinity,
      "learning_rate": 0.00018177018355804018,
      "loss": 11.1491,
      "step": 166430
    },
    {
      "epoch": 20.028880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.00018176799360139124,
      "loss": 11.0098,
      "step": 166440
    },
    {
      "epoch": 20.03008423586041,
      "grad_norm": Infinity,
      "learning_rate": 0.00018176580352640377,
      "loss": 11.121,
      "step": 166450
    },
    {
      "epoch": 20.031287605294825,
      "grad_norm": Infinity,
      "learning_rate": 0.00018176361333308103,
      "loss": 11.1314,
      "step": 166460
    },
    {
      "epoch": 20.032490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00018176142302142614,
      "loss": 11.0975,
      "step": 166470
    },
    {
      "epoch": 20.03369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.0001817592325914423,
      "loss": 11.2076,
      "step": 166480
    },
    {
      "epoch": 20.034897713598074,
      "grad_norm": Infinity,
      "learning_rate": 0.00018175704204313262,
      "loss": 11.1241,
      "step": 166490
    },
    {
      "epoch": 20.03610108303249,
      "grad_norm": Infinity,
      "learning_rate": 0.00018175485137650034,
      "loss": 11.1964,
      "step": 166500
    },
    {
      "epoch": 20.03730445246691,
      "grad_norm": Infinity,
      "learning_rate": 0.00018175266059154862,
      "loss": 11.1089,
      "step": 166510
    },
    {
      "epoch": 20.038507821901323,
      "grad_norm": Infinity,
      "learning_rate": 0.00018175046968828062,
      "loss": 11.2726,
      "step": 166520
    },
    {
      "epoch": 20.03971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.00018174827866669947,
      "loss": 11.1519,
      "step": 166530
    },
    {
      "epoch": 20.040914560770158,
      "grad_norm": Infinity,
      "learning_rate": 0.0001817460875268084,
      "loss": 11.2426,
      "step": 166540
    },
    {
      "epoch": 20.042117930204572,
      "grad_norm": Infinity,
      "learning_rate": 0.00018174389626861057,
      "loss": 11.2022,
      "step": 166550
    },
    {
      "epoch": 20.04332129963899,
      "grad_norm": Infinity,
      "learning_rate": 0.00018174170489210917,
      "loss": 11.0936,
      "step": 166560
    },
    {
      "epoch": 20.044524669073404,
      "grad_norm": Infinity,
      "learning_rate": 0.00018173951339730727,
      "loss": 11.1601,
      "step": 166570
    },
    {
      "epoch": 20.04572803850782,
      "grad_norm": Infinity,
      "learning_rate": 0.00018173732178420815,
      "loss": 11.1186,
      "step": 166580
    },
    {
      "epoch": 20.04693140794224,
      "grad_norm": Infinity,
      "learning_rate": 0.00018173513005281493,
      "loss": 11.1114,
      "step": 166590
    },
    {
      "epoch": 20.048134777376653,
      "grad_norm": Infinity,
      "learning_rate": 0.00018173293820313079,
      "loss": 11.1871,
      "step": 166600
    },
    {
      "epoch": 20.04933814681107,
      "grad_norm": Infinity,
      "learning_rate": 0.00018173074623515896,
      "loss": 11.0846,
      "step": 166610
    },
    {
      "epoch": 20.050541516245488,
      "grad_norm": Infinity,
      "learning_rate": 0.00018172855414890252,
      "loss": 11.1424,
      "step": 166620
    },
    {
      "epoch": 20.051744885679902,
      "grad_norm": Infinity,
      "learning_rate": 0.0001817263619443647,
      "loss": 11.1876,
      "step": 166630
    },
    {
      "epoch": 20.05294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00018172416962154863,
      "loss": 11.0225,
      "step": 166640
    },
    {
      "epoch": 20.054151624548737,
      "grad_norm": Infinity,
      "learning_rate": 0.00018172197718045754,
      "loss": 11.1876,
      "step": 166650
    },
    {
      "epoch": 20.05535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00018171978462109455,
      "loss": 11.1497,
      "step": 166660
    },
    {
      "epoch": 20.05655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00018171759194346288,
      "loss": 11.0555,
      "step": 166670
    },
    {
      "epoch": 20.057761732851986,
      "grad_norm": Infinity,
      "learning_rate": 0.00018171539914756566,
      "loss": 11.1797,
      "step": 166680
    },
    {
      "epoch": 20.0589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.0001817132062334061,
      "loss": 11.1347,
      "step": 166690
    },
    {
      "epoch": 20.060168471720818,
      "grad_norm": Infinity,
      "learning_rate": 0.00018171101320098732,
      "loss": 11.1256,
      "step": 166700
    },
    {
      "epoch": 20.061371841155236,
      "grad_norm": Infinity,
      "learning_rate": 0.00018170882005031258,
      "loss": 11.0957,
      "step": 166710
    },
    {
      "epoch": 20.06257521058965,
      "grad_norm": Infinity,
      "learning_rate": 0.00018170662678138497,
      "loss": 11.1812,
      "step": 166720
    },
    {
      "epoch": 20.063778580024067,
      "grad_norm": Infinity,
      "learning_rate": 0.0001817044333942077,
      "loss": 11.1599,
      "step": 166730
    },
    {
      "epoch": 20.064981949458485,
      "grad_norm": Infinity,
      "learning_rate": 0.00018170223988878394,
      "loss": 11.2837,
      "step": 166740
    },
    {
      "epoch": 20.0661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.0001817000462651169,
      "loss": 11.1328,
      "step": 166750
    },
    {
      "epoch": 20.067388688327316,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816978525232097,
      "loss": 11.075,
      "step": 166760
    },
    {
      "epoch": 20.068592057761734,
      "grad_norm": Infinity,
      "learning_rate": 0.00018169565866306552,
      "loss": 11.209,
      "step": 166770
    },
    {
      "epoch": 20.069795427196148,
      "grad_norm": Infinity,
      "learning_rate": 0.00018169346468468756,
      "loss": 11.2069,
      "step": 166780
    },
    {
      "epoch": 20.070998796630565,
      "grad_norm": Infinity,
      "learning_rate": 0.000181691270588079,
      "loss": 11.1312,
      "step": 166790
    },
    {
      "epoch": 20.072202166064983,
      "grad_norm": Infinity,
      "learning_rate": 0.000181689076373243,
      "loss": 11.0913,
      "step": 166800
    },
    {
      "epoch": 20.073405535499397,
      "grad_norm": Infinity,
      "learning_rate": 0.00018168688204018275,
      "loss": 11.017,
      "step": 166810
    },
    {
      "epoch": 20.074608904933815,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816846875889014,
      "loss": 11.1794,
      "step": 166820
    },
    {
      "epoch": 20.075812274368232,
      "grad_norm": Infinity,
      "learning_rate": 0.00018168249301940215,
      "loss": 11.0485,
      "step": 166830
    },
    {
      "epoch": 20.077015643802646,
      "grad_norm": Infinity,
      "learning_rate": 0.00018168029833168814,
      "loss": 11.0515,
      "step": 166840
    },
    {
      "epoch": 20.078219013237064,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816781035257626,
      "loss": 11.158,
      "step": 166850
    },
    {
      "epoch": 20.07942238267148,
      "grad_norm": Infinity,
      "learning_rate": 0.00018167590860162865,
      "loss": 11.1102,
      "step": 166860
    },
    {
      "epoch": 20.080625752105895,
      "grad_norm": Infinity,
      "learning_rate": 0.00018167371355928952,
      "loss": 11.1662,
      "step": 166870
    },
    {
      "epoch": 20.081829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.00018167151839874833,
      "loss": 11.0974,
      "step": 166880
    },
    {
      "epoch": 20.08303249097473,
      "grad_norm": Infinity,
      "learning_rate": 0.00018166932312000834,
      "loss": 11.1498,
      "step": 166890
    },
    {
      "epoch": 20.084235860409144,
      "grad_norm": Infinity,
      "learning_rate": 0.00018166712772307264,
      "loss": 11.2222,
      "step": 166900
    },
    {
      "epoch": 20.085439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00018166493220794444,
      "loss": 11.0615,
      "step": 166910
    },
    {
      "epoch": 20.08664259927798,
      "grad_norm": Infinity,
      "learning_rate": 0.00018166273657462694,
      "loss": 11.0623,
      "step": 166920
    },
    {
      "epoch": 20.087845968712394,
      "grad_norm": Infinity,
      "learning_rate": 0.00018166054082312328,
      "loss": 11.1037,
      "step": 166930
    },
    {
      "epoch": 20.08904933814681,
      "grad_norm": Infinity,
      "learning_rate": 0.00018165834495343668,
      "loss": 11.1357,
      "step": 166940
    },
    {
      "epoch": 20.09025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.00018165614896557028,
      "loss": 11.1944,
      "step": 166950
    },
    {
      "epoch": 20.091456077015643,
      "grad_norm": Infinity,
      "learning_rate": 0.00018165395285952727,
      "loss": 11.2273,
      "step": 166960
    },
    {
      "epoch": 20.09265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00018165175663531082,
      "loss": 11.1936,
      "step": 166970
    },
    {
      "epoch": 20.093862815884478,
      "grad_norm": Infinity,
      "learning_rate": 0.00018164956029292415,
      "loss": 11.1253,
      "step": 166980
    },
    {
      "epoch": 20.095066185318892,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816473638323704,
      "loss": 11.1734,
      "step": 166990
    },
    {
      "epoch": 20.09626955475331,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816451672536527,
      "loss": 11.1823,
      "step": 167000
    },
    {
      "epoch": 20.097472924187727,
      "grad_norm": Infinity,
      "learning_rate": 0.00018164297055677436,
      "loss": 11.1539,
      "step": 167010
    },
    {
      "epoch": 20.09867629362214,
      "grad_norm": Infinity,
      "learning_rate": 0.00018164077374173843,
      "loss": 11.1835,
      "step": 167020
    },
    {
      "epoch": 20.09987966305656,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816385768085482,
      "loss": 11.0116,
      "step": 167030
    },
    {
      "epoch": 20.101083032490976,
      "grad_norm": Infinity,
      "learning_rate": 0.00018163637975720677,
      "loss": 11.2357,
      "step": 167040
    },
    {
      "epoch": 20.10228640192539,
      "grad_norm": Infinity,
      "learning_rate": 0.00018163418258771734,
      "loss": 11.1763,
      "step": 167050
    },
    {
      "epoch": 20.103489771359808,
      "grad_norm": Infinity,
      "learning_rate": 0.00018163198530008308,
      "loss": 11.0937,
      "step": 167060
    },
    {
      "epoch": 20.104693140794225,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816297878943072,
      "loss": 11.1731,
      "step": 167070
    },
    {
      "epoch": 20.10589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.00018162759037039283,
      "loss": 11.0829,
      "step": 167080
    },
    {
      "epoch": 20.107099879663057,
      "grad_norm": Infinity,
      "learning_rate": 0.00018162539272834324,
      "loss": 11.0928,
      "step": 167090
    },
    {
      "epoch": 20.108303249097474,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816231949681615,
      "loss": 11.1036,
      "step": 167100
    },
    {
      "epoch": 20.10950661853189,
      "grad_norm": Infinity,
      "learning_rate": 0.00018162099708985088,
      "loss": 11.2172,
      "step": 167110
    },
    {
      "epoch": 20.110709987966306,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816187990934145,
      "loss": 11.1132,
      "step": 167120
    },
    {
      "epoch": 20.111913357400724,
      "grad_norm": Infinity,
      "learning_rate": 0.00018161660097885558,
      "loss": 11.1859,
      "step": 167130
    },
    {
      "epoch": 20.113116726835138,
      "grad_norm": Infinity,
      "learning_rate": 0.00018161440274617728,
      "loss": 11.2652,
      "step": 167140
    },
    {
      "epoch": 20.114320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.00018161220439538283,
      "loss": 11.2855,
      "step": 167150
    },
    {
      "epoch": 20.115523465703973,
      "grad_norm": Infinity,
      "learning_rate": 0.0001816100059264753,
      "loss": 11.0792,
      "step": 167160
    },
    {
      "epoch": 20.116726835138387,
      "grad_norm": Infinity,
      "learning_rate": 0.00018160780733945798,
      "loss": 11.1545,
      "step": 167170
    },
    {
      "epoch": 20.117930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.00018160560863433403,
      "loss": 11.1262,
      "step": 167180
    },
    {
      "epoch": 20.119133574007222,
      "grad_norm": Infinity,
      "learning_rate": 0.00018160340981110662,
      "loss": 11.1201,
      "step": 167190
    },
    {
      "epoch": 20.120336943441636,
      "grad_norm": Infinity,
      "learning_rate": 0.00018160121086977887,
      "loss": 11.0896,
      "step": 167200
    },
    {
      "epoch": 20.121540312876053,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815990118103541,
      "loss": 11.007,
      "step": 167210
    },
    {
      "epoch": 20.12274368231047,
      "grad_norm": Infinity,
      "learning_rate": 0.00018159681263283533,
      "loss": 11.1382,
      "step": 167220
    },
    {
      "epoch": 20.123947051744885,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815946133372259,
      "loss": 11.1126,
      "step": 167230
    },
    {
      "epoch": 20.125150421179303,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815924139235289,
      "loss": 11.1613,
      "step": 167240
    },
    {
      "epoch": 20.126353790613717,
      "grad_norm": Infinity,
      "learning_rate": 0.00018159021439174752,
      "loss": 11.1604,
      "step": 167250
    },
    {
      "epoch": 20.127557160048134,
      "grad_norm": Infinity,
      "learning_rate": 0.00018158801474188496,
      "loss": 11.2473,
      "step": 167260
    },
    {
      "epoch": 20.128760529482552,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815858149739444,
      "loss": 11.125,
      "step": 167270
    },
    {
      "epoch": 20.129963898916966,
      "grad_norm": Infinity,
      "learning_rate": 0.000181583615087929,
      "loss": 11.1334,
      "step": 167280
    },
    {
      "epoch": 20.131167268351383,
      "grad_norm": Infinity,
      "learning_rate": 0.00018158141508384202,
      "loss": 11.1533,
      "step": 167290
    },
    {
      "epoch": 20.1323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00018157921496168654,
      "loss": 11.0755,
      "step": 167300
    },
    {
      "epoch": 20.133574007220215,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815770147214658,
      "loss": 11.1248,
      "step": 167310
    },
    {
      "epoch": 20.134777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.00018157481436318303,
      "loss": 11.084,
      "step": 167320
    },
    {
      "epoch": 20.13598074608905,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815726138868413,
      "loss": 11.0828,
      "step": 167330
    },
    {
      "epoch": 20.137184115523464,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815704132924439,
      "loss": 11.1401,
      "step": 167340
    },
    {
      "epoch": 20.13838748495788,
      "grad_norm": Infinity,
      "learning_rate": 0.00018156821257999398,
      "loss": 11.1719,
      "step": 167350
    },
    {
      "epoch": 20.1395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00018156601174949466,
      "loss": 11.2306,
      "step": 167360
    },
    {
      "epoch": 20.140794223826713,
      "grad_norm": Infinity,
      "learning_rate": 0.00018156381080094923,
      "loss": 11.0992,
      "step": 167370
    },
    {
      "epoch": 20.14199759326113,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815616097343608,
      "loss": 11.1099,
      "step": 167380
    },
    {
      "epoch": 20.14320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.00018155940854973262,
      "loss": 11.085,
      "step": 167390
    },
    {
      "epoch": 20.144404332129962,
      "grad_norm": Infinity,
      "learning_rate": 0.00018155720724706778,
      "loss": 11.1913,
      "step": 167400
    },
    {
      "epoch": 20.14560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815550058263696,
      "loss": 11.1644,
      "step": 167410
    },
    {
      "epoch": 20.146811070998798,
      "grad_norm": Infinity,
      "learning_rate": 0.00018155280428764115,
      "loss": 11.1332,
      "step": 167420
    },
    {
      "epoch": 20.14801444043321,
      "grad_norm": Infinity,
      "learning_rate": 0.00018155060263088564,
      "loss": 11.1433,
      "step": 167430
    },
    {
      "epoch": 20.14921780986763,
      "grad_norm": Infinity,
      "learning_rate": 0.00018154840085610633,
      "loss": 11.2023,
      "step": 167440
    },
    {
      "epoch": 20.150421179302047,
      "grad_norm": Infinity,
      "learning_rate": 0.00018154619896330632,
      "loss": 11.0908,
      "step": 167450
    },
    {
      "epoch": 20.15162454873646,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815439969524888,
      "loss": 11.235,
      "step": 167460
    },
    {
      "epoch": 20.15282791817088,
      "grad_norm": Infinity,
      "learning_rate": 0.00018154179482365702,
      "loss": 11.1091,
      "step": 167470
    },
    {
      "epoch": 20.154031287605296,
      "grad_norm": Infinity,
      "learning_rate": 0.00018153959257681413,
      "loss": 11.1528,
      "step": 167480
    },
    {
      "epoch": 20.15523465703971,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815373902119633,
      "loss": 11.0912,
      "step": 167490
    },
    {
      "epoch": 20.156438026474127,
      "grad_norm": Infinity,
      "learning_rate": 0.00018153518772910773,
      "loss": 11.1255,
      "step": 167500
    },
    {
      "epoch": 20.157641395908545,
      "grad_norm": Infinity,
      "learning_rate": 0.00018153298512825066,
      "loss": 11.0745,
      "step": 167510
    },
    {
      "epoch": 20.15884476534296,
      "grad_norm": Infinity,
      "learning_rate": 0.00018153078240939517,
      "loss": 11.2918,
      "step": 167520
    },
    {
      "epoch": 20.160048134777377,
      "grad_norm": Infinity,
      "learning_rate": 0.00018152857957254455,
      "loss": 11.1125,
      "step": 167530
    },
    {
      "epoch": 20.161251504211794,
      "grad_norm": Infinity,
      "learning_rate": 0.00018152637661770192,
      "loss": 11.1964,
      "step": 167540
    },
    {
      "epoch": 20.162454873646208,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815241735448705,
      "loss": 11.2311,
      "step": 167550
    },
    {
      "epoch": 20.163658243080626,
      "grad_norm": Infinity,
      "learning_rate": 0.00018152197035405348,
      "loss": 11.0796,
      "step": 167560
    },
    {
      "epoch": 20.164861612515043,
      "grad_norm": Infinity,
      "learning_rate": 0.00018151976704525408,
      "loss": 11.1609,
      "step": 167570
    },
    {
      "epoch": 20.166064981949457,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815175636184754,
      "loss": 11.0817,
      "step": 167580
    },
    {
      "epoch": 20.167268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.00018151536007372066,
      "loss": 11.1746,
      "step": 167590
    },
    {
      "epoch": 20.168471720818292,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815131564109931,
      "loss": 11.1585,
      "step": 167600
    },
    {
      "epoch": 20.169675090252706,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815109526302959,
      "loss": 11.0773,
      "step": 167610
    },
    {
      "epoch": 20.170878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815087487316322,
      "loss": 11.0182,
      "step": 167620
    },
    {
      "epoch": 20.17208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.0001815065447150052,
      "loss": 11.113,
      "step": 167630
    },
    {
      "epoch": 20.173285198555956,
      "grad_norm": Infinity,
      "learning_rate": 0.00018150434058041815,
      "loss": 11.1181,
      "step": 167640
    },
    {
      "epoch": 20.174488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.00018150213632787417,
      "loss": 11.1721,
      "step": 167650
    },
    {
      "epoch": 20.17569193742479,
      "grad_norm": Infinity,
      "learning_rate": 0.00018149993195737646,
      "loss": 11.1159,
      "step": 167660
    },
    {
      "epoch": 20.176895306859205,
      "grad_norm": Infinity,
      "learning_rate": 0.00018149772746892828,
      "loss": 11.0641,
      "step": 167670
    },
    {
      "epoch": 20.178098676293622,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814955228625327,
      "loss": 11.1873,
      "step": 167680
    },
    {
      "epoch": 20.17930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.00018149331813819303,
      "loss": 11.2447,
      "step": 167690
    },
    {
      "epoch": 20.180505415162454,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814911132959124,
      "loss": 11.1306,
      "step": 167700
    },
    {
      "epoch": 20.18170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.000181488908335694,
      "loss": 11.0634,
      "step": 167710
    },
    {
      "epoch": 20.18291215403129,
      "grad_norm": Infinity,
      "learning_rate": 0.00018148670325754106,
      "loss": 11.0495,
      "step": 167720
    },
    {
      "epoch": 20.184115523465703,
      "grad_norm": Infinity,
      "learning_rate": 0.00018148449806145667,
      "loss": 11.0815,
      "step": 167730
    },
    {
      "epoch": 20.18531889290012,
      "grad_norm": Infinity,
      "learning_rate": 0.00018148229274744416,
      "loss": 11.1151,
      "step": 167740
    },
    {
      "epoch": 20.186522262334538,
      "grad_norm": Infinity,
      "learning_rate": 0.00018148008731550664,
      "loss": 11.3008,
      "step": 167750
    },
    {
      "epoch": 20.187725631768952,
      "grad_norm": Infinity,
      "learning_rate": 0.00018147788176564727,
      "loss": 11.0794,
      "step": 167760
    },
    {
      "epoch": 20.18892900120337,
      "grad_norm": Infinity,
      "learning_rate": 0.00018147567609786935,
      "loss": 11.1313,
      "step": 167770
    },
    {
      "epoch": 20.190132370637787,
      "grad_norm": Infinity,
      "learning_rate": 0.00018147347031217598,
      "loss": 11.2121,
      "step": 167780
    },
    {
      "epoch": 20.1913357400722,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814712644085704,
      "loss": 11.079,
      "step": 167790
    },
    {
      "epoch": 20.19253910950662,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814690583870558,
      "loss": 11.1556,
      "step": 167800
    },
    {
      "epoch": 20.193742478941036,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814668522476353,
      "loss": 11.2027,
      "step": 167810
    },
    {
      "epoch": 20.19494584837545,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814646459903122,
      "loss": 11.1316,
      "step": 167820
    },
    {
      "epoch": 20.196149217809868,
      "grad_norm": Infinity,
      "learning_rate": 0.00018146243961508963,
      "loss": 11.161,
      "step": 167830
    },
    {
      "epoch": 20.197352587244286,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814602331219708,
      "loss": 11.2116,
      "step": 167840
    },
    {
      "epoch": 20.1985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814580265109589,
      "loss": 11.1085,
      "step": 167850
    },
    {
      "epoch": 20.199759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00018145581978205711,
      "loss": 11.1425,
      "step": 167860
    },
    {
      "epoch": 20.200962695547535,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814536129352687,
      "loss": 11.0938,
      "step": 167870
    },
    {
      "epoch": 20.20216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.00018145140597059672,
      "loss": 11.1118,
      "step": 167880
    },
    {
      "epoch": 20.203369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814491988880445,
      "loss": 11.0124,
      "step": 167890
    },
    {
      "epoch": 20.204572803850784,
      "grad_norm": Infinity,
      "learning_rate": 0.00018144699168761515,
      "loss": 11.1443,
      "step": 167900
    },
    {
      "epoch": 20.205776173285198,
      "grad_norm": Infinity,
      "learning_rate": 0.00018144478436931189,
      "loss": 11.0882,
      "step": 167910
    },
    {
      "epoch": 20.206979542719615,
      "grad_norm": Infinity,
      "learning_rate": 0.00018144257693313796,
      "loss": 11.2014,
      "step": 167920
    },
    {
      "epoch": 20.20818291215403,
      "grad_norm": Infinity,
      "learning_rate": 0.00018144036937909648,
      "loss": 11.1425,
      "step": 167930
    },
    {
      "epoch": 20.209386281588447,
      "grad_norm": Infinity,
      "learning_rate": 0.00018143816170719068,
      "loss": 11.1595,
      "step": 167940
    },
    {
      "epoch": 20.210589651022865,
      "grad_norm": Infinity,
      "learning_rate": 0.00018143595391742376,
      "loss": 11.0945,
      "step": 167950
    },
    {
      "epoch": 20.21179302045728,
      "grad_norm": Infinity,
      "learning_rate": 0.00018143374600979893,
      "loss": 11.0637,
      "step": 167960
    },
    {
      "epoch": 20.212996389891696,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814315379843193,
      "loss": 11.2481,
      "step": 167970
    },
    {
      "epoch": 20.214199759326114,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814293298409882,
      "loss": 11.1392,
      "step": 167980
    },
    {
      "epoch": 20.215403128760528,
      "grad_norm": Infinity,
      "learning_rate": 0.00018142712157980873,
      "loss": 11.0839,
      "step": 167990
    },
    {
      "epoch": 20.216606498194945,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814249132007841,
      "loss": 11.1314,
      "step": 168000
    },
    {
      "epoch": 20.217809867629363,
      "grad_norm": Infinity,
      "learning_rate": 0.00018142270470391752,
      "loss": 11.1778,
      "step": 168010
    },
    {
      "epoch": 20.219013237063777,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814204960892122,
      "loss": 11.1252,
      "step": 168020
    },
    {
      "epoch": 20.220216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.0001814182873566713,
      "loss": 11.1722,
      "step": 168030
    },
    {
      "epoch": 20.221419975932612,
      "grad_norm": Infinity,
      "learning_rate": 0.00018141607850629806,
      "loss": 11.2076,
      "step": 168040
    },
    {
      "epoch": 20.222623345367026,
      "grad_norm": Infinity,
      "learning_rate": 0.00018141386953809564,
      "loss": 10.9646,
      "step": 168050
    },
    {
      "epoch": 20.223826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.00018141166045206722,
      "loss": 11.1962,
      "step": 168060
    },
    {
      "epoch": 20.22503008423586,
      "grad_norm": Infinity,
      "learning_rate": 0.00018140945124821606,
      "loss": 11.1501,
      "step": 168070
    },
    {
      "epoch": 20.226233453670275,
      "grad_norm": Infinity,
      "learning_rate": 0.00018140724192654536,
      "loss": 11.1453,
      "step": 168080
    },
    {
      "epoch": 20.227436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00018140503248705823,
      "loss": 11.1853,
      "step": 168090
    },
    {
      "epoch": 20.22864019253911,
      "grad_norm": Infinity,
      "learning_rate": 0.00018140282292975792,
      "loss": 11.3167,
      "step": 168100
    },
    {
      "epoch": 20.229843561973524,
      "grad_norm": Infinity,
      "learning_rate": 0.00018140061325464766,
      "loss": 11.099,
      "step": 168110
    },
    {
      "epoch": 20.231046931407942,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813984034617306,
      "loss": 11.0346,
      "step": 168120
    },
    {
      "epoch": 20.23225030084236,
      "grad_norm": Infinity,
      "learning_rate": 0.00018139619355100995,
      "loss": 11.1847,
      "step": 168130
    },
    {
      "epoch": 20.233453670276774,
      "grad_norm": Infinity,
      "learning_rate": 0.00018139398352248892,
      "loss": 11.046,
      "step": 168140
    },
    {
      "epoch": 20.23465703971119,
      "grad_norm": Infinity,
      "learning_rate": 0.00018139177337617068,
      "loss": 11.2267,
      "step": 168150
    },
    {
      "epoch": 20.23586040914561,
      "grad_norm": Infinity,
      "learning_rate": 0.00018138956311205846,
      "loss": 11.1438,
      "step": 168160
    },
    {
      "epoch": 20.237063778580023,
      "grad_norm": Infinity,
      "learning_rate": 0.00018138735273015544,
      "loss": 11.1627,
      "step": 168170
    },
    {
      "epoch": 20.23826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.00018138514223046484,
      "loss": 11.0866,
      "step": 168180
    },
    {
      "epoch": 20.239470517448858,
      "grad_norm": Infinity,
      "learning_rate": 0.00018138293161298986,
      "loss": 11.2642,
      "step": 168190
    },
    {
      "epoch": 20.240673886883272,
      "grad_norm": Infinity,
      "learning_rate": 0.00018138072087773365,
      "loss": 11.1357,
      "step": 168200
    },
    {
      "epoch": 20.24187725631769,
      "grad_norm": Infinity,
      "learning_rate": 0.00018137851002469945,
      "loss": 11.162,
      "step": 168210
    },
    {
      "epoch": 20.243080625752107,
      "grad_norm": Infinity,
      "learning_rate": 0.00018137629905389048,
      "loss": 11.0163,
      "step": 168220
    },
    {
      "epoch": 20.24428399518652,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813740879653099,
      "loss": 11.09,
      "step": 168230
    },
    {
      "epoch": 20.24548736462094,
      "grad_norm": Infinity,
      "learning_rate": 0.00018137187675896095,
      "loss": 11.1475,
      "step": 168240
    },
    {
      "epoch": 20.246690734055356,
      "grad_norm": Infinity,
      "learning_rate": 0.00018136966543484678,
      "loss": 11.1754,
      "step": 168250
    },
    {
      "epoch": 20.24789410348977,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813674539929706,
      "loss": 11.0882,
      "step": 168260
    },
    {
      "epoch": 20.249097472924188,
      "grad_norm": Infinity,
      "learning_rate": 0.00018136524243333562,
      "loss": 11.1498,
      "step": 168270
    },
    {
      "epoch": 20.250300842358605,
      "grad_norm": Infinity,
      "learning_rate": 0.00018136303075594508,
      "loss": 11.1468,
      "step": 168280
    },
    {
      "epoch": 20.25150421179302,
      "grad_norm": Infinity,
      "learning_rate": 0.00018136081896080213,
      "loss": 11.2255,
      "step": 168290
    },
    {
      "epoch": 20.252707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00018135860704791,
      "loss": 11.1239,
      "step": 168300
    },
    {
      "epoch": 20.253910950661854,
      "grad_norm": Infinity,
      "learning_rate": 0.00018135639501727185,
      "loss": 11.1368,
      "step": 168310
    },
    {
      "epoch": 20.25511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00018135418286889095,
      "loss": 11.201,
      "step": 168320
    },
    {
      "epoch": 20.256317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00018135197060277042,
      "loss": 11.1481,
      "step": 168330
    },
    {
      "epoch": 20.257521058965104,
      "grad_norm": Infinity,
      "learning_rate": 0.00018134975821891352,
      "loss": 11.1933,
      "step": 168340
    },
    {
      "epoch": 20.258724428399518,
      "grad_norm": Infinity,
      "learning_rate": 0.00018134754571732343,
      "loss": 11.1831,
      "step": 168350
    },
    {
      "epoch": 20.259927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00018134533309800336,
      "loss": 11.2204,
      "step": 168360
    },
    {
      "epoch": 20.261131167268353,
      "grad_norm": Infinity,
      "learning_rate": 0.00018134312036095654,
      "loss": 11.1646,
      "step": 168370
    },
    {
      "epoch": 20.262334536702767,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813409075061861,
      "loss": 11.1786,
      "step": 168380
    },
    {
      "epoch": 20.263537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.00018133869453369533,
      "loss": 11.088,
      "step": 168390
    },
    {
      "epoch": 20.264741275571602,
      "grad_norm": Infinity,
      "learning_rate": 0.00018133648144348733,
      "loss": 11.2518,
      "step": 168400
    },
    {
      "epoch": 20.265944645006016,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813342682355654,
      "loss": 11.1197,
      "step": 168410
    },
    {
      "epoch": 20.267148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813320549099327,
      "loss": 11.1416,
      "step": 168420
    },
    {
      "epoch": 20.26835138387485,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813298414665924,
      "loss": 11.1635,
      "step": 168430
    },
    {
      "epoch": 20.269554753309265,
      "grad_norm": Infinity,
      "learning_rate": 0.00018132762790554777,
      "loss": 11.064,
      "step": 168440
    },
    {
      "epoch": 20.270758122743683,
      "grad_norm": Infinity,
      "learning_rate": 0.000181325414226802,
      "loss": 11.0857,
      "step": 168450
    },
    {
      "epoch": 20.2719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00018132320043035826,
      "loss": 11.19,
      "step": 168460
    },
    {
      "epoch": 20.273164861612514,
      "grad_norm": Infinity,
      "learning_rate": 0.00018132098651621977,
      "loss": 11.064,
      "step": 168470
    },
    {
      "epoch": 20.27436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00018131877248438972,
      "loss": 11.0883,
      "step": 168480
    },
    {
      "epoch": 20.27557160048135,
      "grad_norm": Infinity,
      "learning_rate": 0.00018131655833487136,
      "loss": 11.275,
      "step": 168490
    },
    {
      "epoch": 20.276774969915763,
      "grad_norm": Infinity,
      "learning_rate": 0.00018131434406766782,
      "loss": 11.2177,
      "step": 168500
    },
    {
      "epoch": 20.27797833935018,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813121296827824,
      "loss": 11.2364,
      "step": 168510
    },
    {
      "epoch": 20.2791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00018130991518021822,
      "loss": 11.1248,
      "step": 168520
    },
    {
      "epoch": 20.280385078219012,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813077005599785,
      "loss": 11.1157,
      "step": 168530
    },
    {
      "epoch": 20.28158844765343,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813054858220665,
      "loss": 11.116,
      "step": 168540
    },
    {
      "epoch": 20.282791817087848,
      "grad_norm": Infinity,
      "learning_rate": 0.0001813032709664854,
      "loss": 11.0292,
      "step": 168550
    },
    {
      "epoch": 20.28399518652226,
      "grad_norm": Infinity,
      "learning_rate": 0.00018130105599323837,
      "loss": 11.1008,
      "step": 168560
    },
    {
      "epoch": 20.28519855595668,
      "grad_norm": Infinity,
      "learning_rate": 0.00018129884090232863,
      "loss": 11.0948,
      "step": 168570
    },
    {
      "epoch": 20.286401925391097,
      "grad_norm": Infinity,
      "learning_rate": 0.00018129662569375943,
      "loss": 11.1765,
      "step": 168580
    },
    {
      "epoch": 20.28760529482551,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812944103675339,
      "loss": 11.1143,
      "step": 168590
    },
    {
      "epoch": 20.28880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812921949236553,
      "loss": 11.1647,
      "step": 168600
    },
    {
      "epoch": 20.290012033694346,
      "grad_norm": Infinity,
      "learning_rate": 0.00018128997936212683,
      "loss": 11.1665,
      "step": 168610
    },
    {
      "epoch": 20.29121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00018128776368295168,
      "loss": 11.1361,
      "step": 168620
    },
    {
      "epoch": 20.292418772563177,
      "grad_norm": Infinity,
      "learning_rate": 0.00018128554788613308,
      "loss": 11.186,
      "step": 168630
    },
    {
      "epoch": 20.29362214199759,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812833319716742,
      "loss": 11.2919,
      "step": 168640
    },
    {
      "epoch": 20.29482551143201,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812811159395783,
      "loss": 11.1262,
      "step": 168650
    },
    {
      "epoch": 20.296028880866427,
      "grad_norm": Infinity,
      "learning_rate": 0.00018127889978984855,
      "loss": 11.1271,
      "step": 168660
    },
    {
      "epoch": 20.29723225030084,
      "grad_norm": Infinity,
      "learning_rate": 0.00018127668352248817,
      "loss": 11.0992,
      "step": 168670
    },
    {
      "epoch": 20.29843561973526,
      "grad_norm": Infinity,
      "learning_rate": 0.00018127446713750033,
      "loss": 11.1489,
      "step": 168680
    },
    {
      "epoch": 20.299638989169676,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812722506348883,
      "loss": 11.1718,
      "step": 168690
    },
    {
      "epoch": 20.30084235860409,
      "grad_norm": Infinity,
      "learning_rate": 0.00018127003401465525,
      "loss": 11.0129,
      "step": 168700
    },
    {
      "epoch": 20.302045728038507,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812678172768044,
      "loss": 11.1778,
      "step": 168710
    },
    {
      "epoch": 20.303249097472925,
      "grad_norm": Infinity,
      "learning_rate": 0.00018126560042133896,
      "loss": 11.1122,
      "step": 168720
    },
    {
      "epoch": 20.30445246690734,
      "grad_norm": Infinity,
      "learning_rate": 0.00018126338344826213,
      "loss": 11.1422,
      "step": 168730
    },
    {
      "epoch": 20.305655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00018126116635757707,
      "loss": 11.036,
      "step": 168740
    },
    {
      "epoch": 20.306859205776174,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812589491492871,
      "loss": 11.2096,
      "step": 168750
    },
    {
      "epoch": 20.308062575210588,
      "grad_norm": Infinity,
      "learning_rate": 0.00018125673182339534,
      "loss": 11.1884,
      "step": 168760
    },
    {
      "epoch": 20.309265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00018125451437990507,
      "loss": 11.1084,
      "step": 168770
    },
    {
      "epoch": 20.310469314079423,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812522968188194,
      "loss": 11.1523,
      "step": 168780
    },
    {
      "epoch": 20.311672683513837,
      "grad_norm": Infinity,
      "learning_rate": 0.00018125007914014164,
      "loss": 11.143,
      "step": 168790
    },
    {
      "epoch": 20.312876052948255,
      "grad_norm": Infinity,
      "learning_rate": 0.00018124786134387494,
      "loss": 11.1536,
      "step": 168800
    },
    {
      "epoch": 20.314079422382672,
      "grad_norm": Infinity,
      "learning_rate": 0.00018124564343002252,
      "loss": 11.1718,
      "step": 168810
    },
    {
      "epoch": 20.315282791817086,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812434253985876,
      "loss": 11.1878,
      "step": 168820
    },
    {
      "epoch": 20.316486161251504,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812412072495734,
      "loss": 11.1281,
      "step": 168830
    },
    {
      "epoch": 20.31768953068592,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812389889829831,
      "loss": 11.2728,
      "step": 168840
    },
    {
      "epoch": 20.318892900120336,
      "grad_norm": Infinity,
      "learning_rate": 0.00018123677059881993,
      "loss": 11.2612,
      "step": 168850
    },
    {
      "epoch": 20.320096269554753,
      "grad_norm": Infinity,
      "learning_rate": 0.00018123455209708708,
      "loss": 11.1249,
      "step": 168860
    },
    {
      "epoch": 20.32129963898917,
      "grad_norm": Infinity,
      "learning_rate": 0.00018123233347778776,
      "loss": 11.0236,
      "step": 168870
    },
    {
      "epoch": 20.322503008423585,
      "grad_norm": Infinity,
      "learning_rate": 0.00018123011474092523,
      "loss": 11.0864,
      "step": 168880
    },
    {
      "epoch": 20.323706377858002,
      "grad_norm": Infinity,
      "learning_rate": 0.00018122789588650266,
      "loss": 11.0723,
      "step": 168890
    },
    {
      "epoch": 20.32490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812256769145233,
      "loss": 11.1216,
      "step": 168900
    },
    {
      "epoch": 20.326113116726834,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812234578249903,
      "loss": 11.1565,
      "step": 168910
    },
    {
      "epoch": 20.32731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812212386179069,
      "loss": 11.2159,
      "step": 168920
    },
    {
      "epoch": 20.32851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.00018121901929327633,
      "loss": 11.0953,
      "step": 168930
    },
    {
      "epoch": 20.329723225030083,
      "grad_norm": Infinity,
      "learning_rate": 0.00018121679985110177,
      "loss": 11.0642,
      "step": 168940
    },
    {
      "epoch": 20.3309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.00018121458029138644,
      "loss": 11.2041,
      "step": 168950
    },
    {
      "epoch": 20.332129963898918,
      "grad_norm": Infinity,
      "learning_rate": 0.0001812123606141336,
      "loss": 11.1106,
      "step": 168960
    },
    {
      "epoch": 20.333333333333332,
      "grad_norm": Infinity,
      "learning_rate": 0.00018121014081934643,
      "loss": 11.197,
      "step": 168970
    },
    {
      "epoch": 20.33453670276775,
      "grad_norm": Infinity,
      "learning_rate": 0.00018120792090702809,
      "loss": 11.1328,
      "step": 168980
    },
    {
      "epoch": 20.335740072202167,
      "grad_norm": Infinity,
      "learning_rate": 0.00018120570087718188,
      "loss": 11.1587,
      "step": 168990
    },
    {
      "epoch": 20.33694344163658,
      "grad_norm": Infinity,
      "learning_rate": 0.00018120348072981092,
      "loss": 11.1417,
      "step": 169000
    },
    {
      "epoch": 20.338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00018120126046491852,
      "loss": 11.0425,
      "step": 169010
    },
    {
      "epoch": 20.339350180505416,
      "grad_norm": Infinity,
      "learning_rate": 0.00018119904008250784,
      "loss": 11.2006,
      "step": 169020
    },
    {
      "epoch": 20.34055354993983,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811968195825821,
      "loss": 11.2156,
      "step": 169030
    },
    {
      "epoch": 20.341756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811945989651445,
      "loss": 11.0456,
      "step": 169040
    },
    {
      "epoch": 20.342960288808666,
      "grad_norm": Infinity,
      "learning_rate": 0.00018119237823019827,
      "loss": 11.2371,
      "step": 169050
    },
    {
      "epoch": 20.34416365824308,
      "grad_norm": Infinity,
      "learning_rate": 0.00018119015737774662,
      "loss": 11.1413,
      "step": 169060
    },
    {
      "epoch": 20.345367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811879364077928,
      "loss": 11.0901,
      "step": 169070
    },
    {
      "epoch": 20.346570397111915,
      "grad_norm": Infinity,
      "learning_rate": 0.00018118571532033996,
      "loss": 11.1257,
      "step": 169080
    },
    {
      "epoch": 20.34777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.00018118349411539136,
      "loss": 11.1537,
      "step": 169090
    },
    {
      "epoch": 20.348977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811812727929502,
      "loss": 11.0659,
      "step": 169100
    },
    {
      "epoch": 20.350180505415164,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811790513530197,
      "loss": 11.2239,
      "step": 169110
    },
    {
      "epoch": 20.351383874849578,
      "grad_norm": Infinity,
      "learning_rate": 0.00018117682979560304,
      "loss": 11.141,
      "step": 169120
    },
    {
      "epoch": 20.352587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811746081207035,
      "loss": 11.1378,
      "step": 169130
    },
    {
      "epoch": 20.353790613718413,
      "grad_norm": Infinity,
      "learning_rate": 0.00018117238632832426,
      "loss": 11.0068,
      "step": 169140
    },
    {
      "epoch": 20.354993983152827,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811701644184685,
      "loss": 11.2291,
      "step": 169150
    },
    {
      "epoch": 20.356197352587245,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811679423911395,
      "loss": 11.1351,
      "step": 169160
    },
    {
      "epoch": 20.357400722021662,
      "grad_norm": Infinity,
      "learning_rate": 0.00018116572024634044,
      "loss": 11.1022,
      "step": 169170
    },
    {
      "epoch": 20.358604091456076,
      "grad_norm": Infinity,
      "learning_rate": 0.00018116349798407458,
      "loss": 11.0519,
      "step": 169180
    },
    {
      "epoch": 20.359807460890494,
      "grad_norm": Infinity,
      "learning_rate": 0.00018116127560434503,
      "loss": 11.1231,
      "step": 169190
    },
    {
      "epoch": 20.36101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811590531071551,
      "loss": 11.1083,
      "step": 169200
    },
    {
      "epoch": 20.362214199759325,
      "grad_norm": Infinity,
      "learning_rate": 0.00018115683049250801,
      "loss": 11.1163,
      "step": 169210
    },
    {
      "epoch": 20.363417569193743,
      "grad_norm": Infinity,
      "learning_rate": 0.00018115460776040694,
      "loss": 11.1834,
      "step": 169220
    },
    {
      "epoch": 20.36462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.00018115238491085513,
      "loss": 11.3076,
      "step": 169230
    },
    {
      "epoch": 20.365824308062574,
      "grad_norm": Infinity,
      "learning_rate": 0.00018115016194385577,
      "loss": 11.1896,
      "step": 169240
    },
    {
      "epoch": 20.367027677496992,
      "grad_norm": Infinity,
      "learning_rate": 0.00018114793885941208,
      "loss": 11.1026,
      "step": 169250
    },
    {
      "epoch": 20.36823104693141,
      "grad_norm": Infinity,
      "learning_rate": 0.00018114571565752726,
      "loss": 11.2092,
      "step": 169260
    },
    {
      "epoch": 20.369434416365824,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811434923382046,
      "loss": 11.2338,
      "step": 169270
    },
    {
      "epoch": 20.37063778580024,
      "grad_norm": Infinity,
      "learning_rate": 0.00018114126890144726,
      "loss": 11.068,
      "step": 169280
    },
    {
      "epoch": 20.37184115523466,
      "grad_norm": Infinity,
      "learning_rate": 0.00018113904534725847,
      "loss": 11.2205,
      "step": 169290
    },
    {
      "epoch": 20.373044524669073,
      "grad_norm": Infinity,
      "learning_rate": 0.00018113682167564146,
      "loss": 11.2902,
      "step": 169300
    },
    {
      "epoch": 20.37424789410349,
      "grad_norm": Infinity,
      "learning_rate": 0.00018113459788659945,
      "loss": 11.0897,
      "step": 169310
    },
    {
      "epoch": 20.375451263537904,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811323739801356,
      "loss": 11.0878,
      "step": 169320
    },
    {
      "epoch": 20.376654632972322,
      "grad_norm": Infinity,
      "learning_rate": 0.00018113014995625322,
      "loss": 11.1376,
      "step": 169330
    },
    {
      "epoch": 20.37785800240674,
      "grad_norm": Infinity,
      "learning_rate": 0.00018112792581495544,
      "loss": 11.1547,
      "step": 169340
    },
    {
      "epoch": 20.379061371841154,
      "grad_norm": Infinity,
      "learning_rate": 0.00018112570155624552,
      "loss": 11.0835,
      "step": 169350
    },
    {
      "epoch": 20.38026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00018112347718012673,
      "loss": 11.1521,
      "step": 169360
    },
    {
      "epoch": 20.38146811070999,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811212526866022,
      "loss": 11.1616,
      "step": 169370
    },
    {
      "epoch": 20.382671480144403,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811190280756752,
      "loss": 11.0925,
      "step": 169380
    },
    {
      "epoch": 20.38387484957882,
      "grad_norm": Infinity,
      "learning_rate": 0.00018111680334734892,
      "loss": 11.1141,
      "step": 169390
    },
    {
      "epoch": 20.385078219013238,
      "grad_norm": Infinity,
      "learning_rate": 0.00018111457850162666,
      "loss": 11.0996,
      "step": 169400
    },
    {
      "epoch": 20.386281588447652,
      "grad_norm": Infinity,
      "learning_rate": 0.00018111235353851152,
      "loss": 11.0842,
      "step": 169410
    },
    {
      "epoch": 20.38748495788207,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811101284580068,
      "loss": 11.1849,
      "step": 169420
    },
    {
      "epoch": 20.388688327316487,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811079032601157,
      "loss": 11.0988,
      "step": 169430
    },
    {
      "epoch": 20.3898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00018110567794484141,
      "loss": 11.2283,
      "step": 169440
    },
    {
      "epoch": 20.39109506618532,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811034525121872,
      "loss": 11.1412,
      "step": 169450
    },
    {
      "epoch": 20.392298435619736,
      "grad_norm": Infinity,
      "learning_rate": 0.0001811012269621563,
      "loss": 11.136,
      "step": 169460
    },
    {
      "epoch": 20.39350180505415,
      "grad_norm": Infinity,
      "learning_rate": 0.00018109900129475184,
      "loss": 11.143,
      "step": 169470
    },
    {
      "epoch": 20.394705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00018109677550997715,
      "loss": 11.0315,
      "step": 169480
    },
    {
      "epoch": 20.395908543922985,
      "grad_norm": Infinity,
      "learning_rate": 0.0001810945496078354,
      "loss": 11.158,
      "step": 169490
    },
    {
      "epoch": 20.3971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.0001810923235883298,
      "loss": 11.2196,
      "step": 169500
    },
    {
      "epoch": 20.398315282791817,
      "grad_norm": Infinity,
      "learning_rate": 0.00018109009745146358,
      "loss": 11.2157,
      "step": 169510
    },
    {
      "epoch": 20.399518652226234,
      "grad_norm": Infinity,
      "learning_rate": 0.00018108787119724,
      "loss": 11.1027,
      "step": 169520
    },
    {
      "epoch": 20.40072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00018108564482566223,
      "loss": 11.1171,
      "step": 169530
    },
    {
      "epoch": 20.401925391095066,
      "grad_norm": Infinity,
      "learning_rate": 0.00018108341833673351,
      "loss": 11.014,
      "step": 169540
    },
    {
      "epoch": 20.403128760529484,
      "grad_norm": Infinity,
      "learning_rate": 0.00018108119173045707,
      "loss": 11.2019,
      "step": 169550
    },
    {
      "epoch": 20.404332129963898,
      "grad_norm": Infinity,
      "learning_rate": 0.00018107896500683612,
      "loss": 11.3003,
      "step": 169560
    },
    {
      "epoch": 20.405535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.0001810767381658739,
      "loss": 11.0753,
      "step": 169570
    },
    {
      "epoch": 20.406738868832733,
      "grad_norm": Infinity,
      "learning_rate": 0.00018107451120757362,
      "loss": 11.1968,
      "step": 169580
    },
    {
      "epoch": 20.407942238267147,
      "grad_norm": Infinity,
      "learning_rate": 0.0001810722841319385,
      "loss": 11.2494,
      "step": 169590
    },
    {
      "epoch": 20.409145607701564,
      "grad_norm": Infinity,
      "learning_rate": 0.00018107005693897174,
      "loss": 11.1288,
      "step": 169600
    },
    {
      "epoch": 20.410348977135982,
      "grad_norm": Infinity,
      "learning_rate": 0.00018106782962867666,
      "loss": 11.1042,
      "step": 169610
    },
    {
      "epoch": 20.411552346570396,
      "grad_norm": Infinity,
      "learning_rate": 0.00018106560220105638,
      "loss": 11.2676,
      "step": 169620
    },
    {
      "epoch": 20.412755716004813,
      "grad_norm": Infinity,
      "learning_rate": 0.00018106337465611415,
      "loss": 11.1516,
      "step": 169630
    },
    {
      "epoch": 20.41395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00018106114699385323,
      "loss": 11.1647,
      "step": 169640
    },
    {
      "epoch": 20.415162454873645,
      "grad_norm": Infinity,
      "learning_rate": 0.00018105891921427677,
      "loss": 11.1637,
      "step": 169650
    },
    {
      "epoch": 20.416365824308063,
      "grad_norm": Infinity,
      "learning_rate": 0.00018105669131738807,
      "loss": 11.0832,
      "step": 169660
    },
    {
      "epoch": 20.41756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.00018105446330319035,
      "loss": 11.1426,
      "step": 169670
    },
    {
      "epoch": 20.418772563176894,
      "grad_norm": Infinity,
      "learning_rate": 0.00018105223517168677,
      "loss": 11.108,
      "step": 169680
    },
    {
      "epoch": 20.41997593261131,
      "grad_norm": Infinity,
      "learning_rate": 0.0001810500069228806,
      "loss": 11.1267,
      "step": 169690
    },
    {
      "epoch": 20.42117930204573,
      "grad_norm": Infinity,
      "learning_rate": 0.00018104777855677506,
      "loss": 11.1771,
      "step": 169700
    },
    {
      "epoch": 20.422382671480143,
      "grad_norm": Infinity,
      "learning_rate": 0.0001810455500733734,
      "loss": 11.0843,
      "step": 169710
    },
    {
      "epoch": 20.42358604091456,
      "grad_norm": Infinity,
      "learning_rate": 0.0001810433214726788,
      "loss": 11.2624,
      "step": 169720
    },
    {
      "epoch": 20.42478941034898,
      "grad_norm": Infinity,
      "learning_rate": 0.00018104109275469453,
      "loss": 11.1661,
      "step": 169730
    },
    {
      "epoch": 20.425992779783392,
      "grad_norm": Infinity,
      "learning_rate": 0.00018103886391942375,
      "loss": 11.178,
      "step": 169740
    },
    {
      "epoch": 20.42719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00018103663496686973,
      "loss": 11.1344,
      "step": 169750
    },
    {
      "epoch": 20.428399518652228,
      "grad_norm": Infinity,
      "learning_rate": 0.00018103440589703574,
      "loss": 11.131,
      "step": 169760
    },
    {
      "epoch": 20.42960288808664,
      "grad_norm": Infinity,
      "learning_rate": 0.00018103217670992493,
      "loss": 11.1171,
      "step": 169770
    },
    {
      "epoch": 20.43080625752106,
      "grad_norm": Infinity,
      "learning_rate": 0.00018102994740554053,
      "loss": 11.242,
      "step": 169780
    },
    {
      "epoch": 20.432009626955477,
      "grad_norm": Infinity,
      "learning_rate": 0.0001810277179838858,
      "loss": 11.1073,
      "step": 169790
    },
    {
      "epoch": 20.43321299638989,
      "grad_norm": Infinity,
      "learning_rate": 0.00018102548844496397,
      "loss": 11.1494,
      "step": 169800
    },
    {
      "epoch": 20.43441636582431,
      "grad_norm": Infinity,
      "learning_rate": 0.00018102325878877827,
      "loss": 11.097,
      "step": 169810
    },
    {
      "epoch": 20.435619735258726,
      "grad_norm": Infinity,
      "learning_rate": 0.00018102102901533187,
      "loss": 11.1085,
      "step": 169820
    },
    {
      "epoch": 20.43682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00018101879912462806,
      "loss": 11.0895,
      "step": 169830
    },
    {
      "epoch": 20.438026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00018101656911667004,
      "loss": 11.0728,
      "step": 169840
    },
    {
      "epoch": 20.439229843561975,
      "grad_norm": Infinity,
      "learning_rate": 0.00018101433899146105,
      "loss": 11.1207,
      "step": 169850
    },
    {
      "epoch": 20.44043321299639,
      "grad_norm": Infinity,
      "learning_rate": 0.0001810121087490043,
      "loss": 11.1676,
      "step": 169860
    },
    {
      "epoch": 20.441636582430807,
      "grad_norm": Infinity,
      "learning_rate": 0.00018100987838930304,
      "loss": 11.1478,
      "step": 169870
    },
    {
      "epoch": 20.442839951865224,
      "grad_norm": Infinity,
      "learning_rate": 0.00018100764791236046,
      "loss": 11.2377,
      "step": 169880
    },
    {
      "epoch": 20.444043321299638,
      "grad_norm": Infinity,
      "learning_rate": 0.00018100541731817983,
      "loss": 11.1416,
      "step": 169890
    },
    {
      "epoch": 20.445246690734056,
      "grad_norm": Infinity,
      "learning_rate": 0.00018100318660676438,
      "loss": 11.0905,
      "step": 169900
    },
    {
      "epoch": 20.446450060168473,
      "grad_norm": Infinity,
      "learning_rate": 0.00018100095577811728,
      "loss": 11.3147,
      "step": 169910
    },
    {
      "epoch": 20.447653429602887,
      "grad_norm": Infinity,
      "learning_rate": 0.00018099872483224184,
      "loss": 11.1742,
      "step": 169920
    },
    {
      "epoch": 20.448856799037305,
      "grad_norm": Infinity,
      "learning_rate": 0.00018099649376914122,
      "loss": 11.0597,
      "step": 169930
    },
    {
      "epoch": 20.450060168471722,
      "grad_norm": Infinity,
      "learning_rate": 0.00018099426258881869,
      "loss": 11.1532,
      "step": 169940
    },
    {
      "epoch": 20.451263537906136,
      "grad_norm": Infinity,
      "learning_rate": 0.00018099203129127745,
      "loss": 11.1107,
      "step": 169950
    },
    {
      "epoch": 20.452466907340554,
      "grad_norm": Infinity,
      "learning_rate": 0.00018098979987652073,
      "loss": 11.1537,
      "step": 169960
    },
    {
      "epoch": 20.45367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00018098756834455183,
      "loss": 11.0945,
      "step": 169970
    },
    {
      "epoch": 20.454873646209386,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809853366953739,
      "loss": 11.2382,
      "step": 169980
    },
    {
      "epoch": 20.456077015643803,
      "grad_norm": Infinity,
      "learning_rate": 0.00018098310492899015,
      "loss": 11.1639,
      "step": 169990
    },
    {
      "epoch": 20.45728038507822,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809808730454039,
      "loss": 11.147,
      "step": 170000
    },
    {
      "epoch": 20.458483754512635,
      "grad_norm": Infinity,
      "learning_rate": 0.00018097864104461834,
      "loss": 11.12,
      "step": 170010
    },
    {
      "epoch": 20.459687123947052,
      "grad_norm": Infinity,
      "learning_rate": 0.00018097640892663665,
      "loss": 11.1234,
      "step": 170020
    },
    {
      "epoch": 20.460890493381466,
      "grad_norm": Infinity,
      "learning_rate": 0.00018097417669146214,
      "loss": 11.184,
      "step": 170030
    },
    {
      "epoch": 20.462093862815884,
      "grad_norm": Infinity,
      "learning_rate": 0.00018097194433909798,
      "loss": 11.2513,
      "step": 170040
    },
    {
      "epoch": 20.4632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00018096971186954742,
      "loss": 11.1851,
      "step": 170050
    },
    {
      "epoch": 20.464500601684716,
      "grad_norm": Infinity,
      "learning_rate": 0.00018096747928281372,
      "loss": 11.1678,
      "step": 170060
    },
    {
      "epoch": 20.465703971119133,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809652465789001,
      "loss": 11.1197,
      "step": 170070
    },
    {
      "epoch": 20.46690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.00018096301375780975,
      "loss": 11.1274,
      "step": 170080
    },
    {
      "epoch": 20.468110709987965,
      "grad_norm": Infinity,
      "learning_rate": 0.00018096078081954593,
      "loss": 11.1493,
      "step": 170090
    },
    {
      "epoch": 20.469314079422382,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809585477641119,
      "loss": 11.0888,
      "step": 170100
    },
    {
      "epoch": 20.4705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00018095631459151082,
      "loss": 11.2197,
      "step": 170110
    },
    {
      "epoch": 20.471720818291214,
      "grad_norm": Infinity,
      "learning_rate": 0.000180954081301746,
      "loss": 11.113,
      "step": 170120
    },
    {
      "epoch": 20.47292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00018095184789482062,
      "loss": 11.2568,
      "step": 170130
    },
    {
      "epoch": 20.47412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.00018094961437073795,
      "loss": 11.1379,
      "step": 170140
    },
    {
      "epoch": 20.475330926594463,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809473807295012,
      "loss": 11.1916,
      "step": 170150
    },
    {
      "epoch": 20.47653429602888,
      "grad_norm": Infinity,
      "learning_rate": 0.00018094514697111357,
      "loss": 11.0953,
      "step": 170160
    },
    {
      "epoch": 20.477737665463298,
      "grad_norm": Infinity,
      "learning_rate": 0.00018094291309557836,
      "loss": 11.1652,
      "step": 170170
    },
    {
      "epoch": 20.478941034897712,
      "grad_norm": Infinity,
      "learning_rate": 0.00018094067910289876,
      "loss": 11.1372,
      "step": 170180
    },
    {
      "epoch": 20.48014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.000180938444993078,
      "loss": 11.1689,
      "step": 170190
    },
    {
      "epoch": 20.481347773766547,
      "grad_norm": Infinity,
      "learning_rate": 0.00018093621076611935,
      "loss": 11.2801,
      "step": 170200
    },
    {
      "epoch": 20.48255114320096,
      "grad_norm": Infinity,
      "learning_rate": 0.000180933976422026,
      "loss": 11.1455,
      "step": 170210
    },
    {
      "epoch": 20.48375451263538,
      "grad_norm": Infinity,
      "learning_rate": 0.00018093174196080124,
      "loss": 11.1376,
      "step": 170220
    },
    {
      "epoch": 20.484957882069796,
      "grad_norm": Infinity,
      "learning_rate": 0.00018092950738244823,
      "loss": 11.0514,
      "step": 170230
    },
    {
      "epoch": 20.48616125150421,
      "grad_norm": Infinity,
      "learning_rate": 0.00018092727268697028,
      "loss": 11.0518,
      "step": 170240
    },
    {
      "epoch": 20.487364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.00018092503787437054,
      "loss": 11.1309,
      "step": 170250
    },
    {
      "epoch": 20.488567990373046,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809228029446523,
      "loss": 11.0402,
      "step": 170260
    },
    {
      "epoch": 20.48977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.00018092056789781883,
      "loss": 11.1491,
      "step": 170270
    },
    {
      "epoch": 20.490974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809183327338733,
      "loss": 11.2244,
      "step": 170280
    },
    {
      "epoch": 20.492178098676295,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809160974528189,
      "loss": 11.1063,
      "step": 170290
    },
    {
      "epoch": 20.49338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00018091386205465901,
      "loss": 11.2281,
      "step": 170300
    },
    {
      "epoch": 20.494584837545126,
      "grad_norm": Infinity,
      "learning_rate": 0.00018091162653939677,
      "loss": 11.1546,
      "step": 170310
    },
    {
      "epoch": 20.495788206979544,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809093909070354,
      "loss": 11.0695,
      "step": 170320
    },
    {
      "epoch": 20.496991576413958,
      "grad_norm": Infinity,
      "learning_rate": 0.0001809071551575782,
      "loss": 10.9658,
      "step": 170330
    },
    {
      "epoch": 20.498194945848375,
      "grad_norm": Infinity,
      "learning_rate": 0.00018090491929102833,
      "loss": 11.1502,
      "step": 170340
    },
    {
      "epoch": 20.499398315282793,
      "grad_norm": Infinity,
      "learning_rate": 0.00018090268330738912,
      "loss": 11.1716,
      "step": 170350
    },
    {
      "epoch": 20.500601684717207,
      "grad_norm": Infinity,
      "learning_rate": 0.00018090044720666372,
      "loss": 11.1951,
      "step": 170360
    },
    {
      "epoch": 20.501805054151625,
      "grad_norm": Infinity,
      "learning_rate": 0.00018089821098885538,
      "loss": 11.2606,
      "step": 170370
    },
    {
      "epoch": 20.503008423586042,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808959746539674,
      "loss": 11.155,
      "step": 170380
    },
    {
      "epoch": 20.504211793020456,
      "grad_norm": Infinity,
      "learning_rate": 0.00018089373820200292,
      "loss": 11.0203,
      "step": 170390
    },
    {
      "epoch": 20.505415162454874,
      "grad_norm": Infinity,
      "learning_rate": 0.00018089150163296526,
      "loss": 11.0893,
      "step": 170400
    },
    {
      "epoch": 20.50661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00018088926494685766,
      "loss": 11.052,
      "step": 170410
    },
    {
      "epoch": 20.507821901323705,
      "grad_norm": Infinity,
      "learning_rate": 0.00018088702814368328,
      "loss": 11.1476,
      "step": 170420
    },
    {
      "epoch": 20.509025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808847912234454,
      "loss": 11.0949,
      "step": 170430
    },
    {
      "epoch": 20.51022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.00018088255418614727,
      "loss": 11.1898,
      "step": 170440
    },
    {
      "epoch": 20.511432009626954,
      "grad_norm": Infinity,
      "learning_rate": 0.00018088031703179208,
      "loss": 11.2116,
      "step": 170450
    },
    {
      "epoch": 20.512635379061372,
      "grad_norm": Infinity,
      "learning_rate": 0.00018087807976038315,
      "loss": 11.143,
      "step": 170460
    },
    {
      "epoch": 20.51383874849579,
      "grad_norm": Infinity,
      "learning_rate": 0.00018087584237192364,
      "loss": 11.2086,
      "step": 170470
    },
    {
      "epoch": 20.515042117930204,
      "grad_norm": Infinity,
      "learning_rate": 0.00018087360486641684,
      "loss": 11.0903,
      "step": 170480
    },
    {
      "epoch": 20.51624548736462,
      "grad_norm": Infinity,
      "learning_rate": 0.00018087136724386594,
      "loss": 11.0871,
      "step": 170490
    },
    {
      "epoch": 20.51744885679904,
      "grad_norm": Infinity,
      "learning_rate": 0.00018086912950427423,
      "loss": 11.1235,
      "step": 170500
    },
    {
      "epoch": 20.518652226233453,
      "grad_norm": Infinity,
      "learning_rate": 0.00018086689164764492,
      "loss": 11.124,
      "step": 170510
    },
    {
      "epoch": 20.51985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00018086465367398127,
      "loss": 11.1239,
      "step": 170520
    },
    {
      "epoch": 20.521058965102288,
      "grad_norm": Infinity,
      "learning_rate": 0.00018086241558328648,
      "loss": 11.1464,
      "step": 170530
    },
    {
      "epoch": 20.522262334536702,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808601773755638,
      "loss": 11.0619,
      "step": 170540
    },
    {
      "epoch": 20.52346570397112,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808579390508165,
      "loss": 11.1267,
      "step": 170550
    },
    {
      "epoch": 20.524669073405537,
      "grad_norm": Infinity,
      "learning_rate": 0.00018085570060904778,
      "loss": 11.1167,
      "step": 170560
    },
    {
      "epoch": 20.52587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808534620502609,
      "loss": 11.1912,
      "step": 170570
    },
    {
      "epoch": 20.52707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00018085122337445911,
      "loss": 11.0377,
      "step": 170580
    },
    {
      "epoch": 20.528279181708786,
      "grad_norm": Infinity,
      "learning_rate": 0.00018084898458164562,
      "loss": 11.1624,
      "step": 170590
    },
    {
      "epoch": 20.5294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808467456718237,
      "loss": 11.1101,
      "step": 170600
    },
    {
      "epoch": 20.530685920577618,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808445066449966,
      "loss": 11.1022,
      "step": 170610
    },
    {
      "epoch": 20.531889290012035,
      "grad_norm": Infinity,
      "learning_rate": 0.00018084226750116753,
      "loss": 11.1836,
      "step": 170620
    },
    {
      "epoch": 20.53309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808400282403397,
      "loss": 11.1366,
      "step": 170630
    },
    {
      "epoch": 20.534296028880867,
      "grad_norm": Infinity,
      "learning_rate": 0.00018083778886251642,
      "loss": 11.1776,
      "step": 170640
    },
    {
      "epoch": 20.535499398315284,
      "grad_norm": Infinity,
      "learning_rate": 0.00018083554936770092,
      "loss": 11.1768,
      "step": 170650
    },
    {
      "epoch": 20.5367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808333097558964,
      "loss": 11.202,
      "step": 170660
    },
    {
      "epoch": 20.537906137184116,
      "grad_norm": Infinity,
      "learning_rate": 0.00018083107002710613,
      "loss": 11.2465,
      "step": 170670
    },
    {
      "epoch": 20.53910950661853,
      "grad_norm": Infinity,
      "learning_rate": 0.00018082883018133333,
      "loss": 11.197,
      "step": 170680
    },
    {
      "epoch": 20.540312876052948,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808265902185813,
      "loss": 11.1777,
      "step": 170690
    },
    {
      "epoch": 20.541516245487365,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808243501388532,
      "loss": 11.2165,
      "step": 170700
    },
    {
      "epoch": 20.54271961492178,
      "grad_norm": Infinity,
      "learning_rate": 0.00018082210994215233,
      "loss": 11.1918,
      "step": 170710
    },
    {
      "epoch": 20.543922984356197,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808198696284819,
      "loss": 11.1845,
      "step": 170720
    },
    {
      "epoch": 20.545126353790614,
      "grad_norm": Infinity,
      "learning_rate": 0.00018081762919784514,
      "loss": 11.026,
      "step": 170730
    },
    {
      "epoch": 20.54632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00018081538865024534,
      "loss": 11.1739,
      "step": 170740
    },
    {
      "epoch": 20.547533092659446,
      "grad_norm": Infinity,
      "learning_rate": 0.00018081314798568574,
      "loss": 11.2232,
      "step": 170750
    },
    {
      "epoch": 20.548736462093864,
      "grad_norm": Infinity,
      "learning_rate": 0.00018081090720416953,
      "loss": 11.1036,
      "step": 170760
    },
    {
      "epoch": 20.549939831528278,
      "grad_norm": Infinity,
      "learning_rate": 0.00018080866630569998,
      "loss": 11.2451,
      "step": 170770
    },
    {
      "epoch": 20.551143200962695,
      "grad_norm": Infinity,
      "learning_rate": 0.0001808064252902804,
      "loss": 11.0609,
      "step": 170780
    },
    {
      "epoch": 20.552346570397113,
      "grad_norm": Infinity,
      "learning_rate": 0.00018080418415791392,
      "loss": 11.1309,
      "step": 170790
    },
    {
      "epoch": 20.553549939831527,
      "grad_norm": Infinity,
      "learning_rate": 0.00018080194290860386,
      "loss": 11.1908,
      "step": 170800
    },
    {
      "epoch": 20.554753309265944,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807997015423534,
      "loss": 11.1889,
      "step": 170810
    },
    {
      "epoch": 20.555956678700362,
      "grad_norm": Infinity,
      "learning_rate": 0.00018079746005916588,
      "loss": 11.1928,
      "step": 170820
    },
    {
      "epoch": 20.557160048134776,
      "grad_norm": Infinity,
      "learning_rate": 0.00018079521845904445,
      "loss": 11.2115,
      "step": 170830
    },
    {
      "epoch": 20.558363417569193,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807929767419924,
      "loss": 11.0234,
      "step": 170840
    },
    {
      "epoch": 20.55956678700361,
      "grad_norm": Infinity,
      "learning_rate": 0.00018079073490801295,
      "loss": 11.1885,
      "step": 170850
    },
    {
      "epoch": 20.560770156438025,
      "grad_norm": Infinity,
      "learning_rate": 0.00018078849295710938,
      "loss": 11.0748,
      "step": 170860
    },
    {
      "epoch": 20.561973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807862508892849,
      "loss": 11.0524,
      "step": 170870
    },
    {
      "epoch": 20.56317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807840087045428,
      "loss": 11.0896,
      "step": 170880
    },
    {
      "epoch": 20.564380264741274,
      "grad_norm": Infinity,
      "learning_rate": 0.00018078176640288625,
      "loss": 11.225,
      "step": 170890
    },
    {
      "epoch": 20.56558363417569,
      "grad_norm": Infinity,
      "learning_rate": 0.00018077952398431858,
      "loss": 11.1962,
      "step": 170900
    },
    {
      "epoch": 20.56678700361011,
      "grad_norm": Infinity,
      "learning_rate": 0.00018077728144884296,
      "loss": 11.1384,
      "step": 170910
    },
    {
      "epoch": 20.567990373044523,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807750387964627,
      "loss": 11.1526,
      "step": 170920
    },
    {
      "epoch": 20.56919374247894,
      "grad_norm": Infinity,
      "learning_rate": 0.000180772796027181,
      "loss": 11.1168,
      "step": 170930
    },
    {
      "epoch": 20.57039711191336,
      "grad_norm": Infinity,
      "learning_rate": 0.00018077055314100112,
      "loss": 11.0887,
      "step": 170940
    },
    {
      "epoch": 20.571600481347772,
      "grad_norm": Infinity,
      "learning_rate": 0.00018076831013792631,
      "loss": 11.2093,
      "step": 170950
    },
    {
      "epoch": 20.57280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.00018076606701795985,
      "loss": 11.085,
      "step": 170960
    },
    {
      "epoch": 20.574007220216608,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807638237811049,
      "loss": 11.0918,
      "step": 170970
    },
    {
      "epoch": 20.57521058965102,
      "grad_norm": Infinity,
      "learning_rate": 0.00018076158042736477,
      "loss": 11.0611,
      "step": 170980
    },
    {
      "epoch": 20.57641395908544,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807593369567427,
      "loss": 11.2597,
      "step": 170990
    },
    {
      "epoch": 20.577617328519857,
      "grad_norm": Infinity,
      "learning_rate": 0.00018075709336924195,
      "loss": 11.1704,
      "step": 171000
    },
    {
      "epoch": 20.57882069795427,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807548496648657,
      "loss": 11.0639,
      "step": 171010
    },
    {
      "epoch": 20.58002406738869,
      "grad_norm": Infinity,
      "learning_rate": 0.00018075260584361728,
      "loss": 11.143,
      "step": 171020
    },
    {
      "epoch": 20.581227436823106,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807503619054999,
      "loss": 11.1953,
      "step": 171030
    },
    {
      "epoch": 20.58243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807481178505168,
      "loss": 11.1244,
      "step": 171040
    },
    {
      "epoch": 20.583634175691937,
      "grad_norm": Infinity,
      "learning_rate": 0.00018074587367867127,
      "loss": 11.1587,
      "step": 171050
    },
    {
      "epoch": 20.584837545126355,
      "grad_norm": Infinity,
      "learning_rate": 0.00018074362938996647,
      "loss": 11.0961,
      "step": 171060
    },
    {
      "epoch": 20.58604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00018074138498440573,
      "loss": 11.0667,
      "step": 171070
    },
    {
      "epoch": 20.587244283995187,
      "grad_norm": Infinity,
      "learning_rate": 0.00018073914046199226,
      "loss": 11.1862,
      "step": 171080
    },
    {
      "epoch": 20.588447653429604,
      "grad_norm": Infinity,
      "learning_rate": 0.00018073689582272934,
      "loss": 11.1575,
      "step": 171090
    },
    {
      "epoch": 20.589651022864018,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807346510666202,
      "loss": 11.1074,
      "step": 171100
    },
    {
      "epoch": 20.590854392298436,
      "grad_norm": Infinity,
      "learning_rate": 0.00018073240619366807,
      "loss": 11.1979,
      "step": 171110
    },
    {
      "epoch": 20.592057761732853,
      "grad_norm": Infinity,
      "learning_rate": 0.00018073016120387622,
      "loss": 11.1893,
      "step": 171120
    },
    {
      "epoch": 20.593261131167267,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807279160972479,
      "loss": 11.2165,
      "step": 171130
    },
    {
      "epoch": 20.594464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00018072567087378636,
      "loss": 11.2428,
      "step": 171140
    },
    {
      "epoch": 20.595667870036102,
      "grad_norm": Infinity,
      "learning_rate": 0.00018072342553349482,
      "loss": 11.1328,
      "step": 171150
    },
    {
      "epoch": 20.596871239470516,
      "grad_norm": Infinity,
      "learning_rate": 0.00018072118007637656,
      "loss": 11.1582,
      "step": 171160
    },
    {
      "epoch": 20.598074608904934,
      "grad_norm": Infinity,
      "learning_rate": 0.00018071893450243484,
      "loss": 11.1275,
      "step": 171170
    },
    {
      "epoch": 20.59927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807166888116729,
      "loss": 11.1685,
      "step": 171180
    },
    {
      "epoch": 20.600481347773766,
      "grad_norm": Infinity,
      "learning_rate": 0.000180714443004094,
      "loss": 11.1814,
      "step": 171190
    },
    {
      "epoch": 20.601684717208183,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807121970797013,
      "loss": 11.1649,
      "step": 171200
    },
    {
      "epoch": 20.6028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807099510384982,
      "loss": 11.2236,
      "step": 171210
    },
    {
      "epoch": 20.604091456077015,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807077048804878,
      "loss": 11.2317,
      "step": 171220
    },
    {
      "epoch": 20.605294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00018070545860567348,
      "loss": 11.2011,
      "step": 171230
    },
    {
      "epoch": 20.60649819494585,
      "grad_norm": Infinity,
      "learning_rate": 0.00018070321221405844,
      "loss": 11.0179,
      "step": 171240
    },
    {
      "epoch": 20.607701564380264,
      "grad_norm": Infinity,
      "learning_rate": 0.0001807009657056459,
      "loss": 11.1776,
      "step": 171250
    },
    {
      "epoch": 20.60890493381468,
      "grad_norm": Infinity,
      "learning_rate": 0.00018069871908043915,
      "loss": 11.2294,
      "step": 171260
    },
    {
      "epoch": 20.6101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806964723384414,
      "loss": 11.1094,
      "step": 171270
    },
    {
      "epoch": 20.611311672683513,
      "grad_norm": Infinity,
      "learning_rate": 0.00018069422547965598,
      "loss": 11.057,
      "step": 171280
    },
    {
      "epoch": 20.61251504211793,
      "grad_norm": Infinity,
      "learning_rate": 0.00018069197850408605,
      "loss": 11.1786,
      "step": 171290
    },
    {
      "epoch": 20.613718411552348,
      "grad_norm": Infinity,
      "learning_rate": 0.00018068973141173493,
      "loss": 11.0535,
      "step": 171300
    },
    {
      "epoch": 20.614921780986762,
      "grad_norm": Infinity,
      "learning_rate": 0.00018068748420260584,
      "loss": 11.1982,
      "step": 171310
    },
    {
      "epoch": 20.61612515042118,
      "grad_norm": Infinity,
      "learning_rate": 0.00018068523687670205,
      "loss": 11.1507,
      "step": 171320
    },
    {
      "epoch": 20.617328519855597,
      "grad_norm": Infinity,
      "learning_rate": 0.00018068298943402678,
      "loss": 11.2291,
      "step": 171330
    },
    {
      "epoch": 20.61853188929001,
      "grad_norm": Infinity,
      "learning_rate": 0.00018068074187458332,
      "loss": 11.1811,
      "step": 171340
    },
    {
      "epoch": 20.61973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806784941983749,
      "loss": 11.1592,
      "step": 171350
    },
    {
      "epoch": 20.620938628158846,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806762464054048,
      "loss": 11.1216,
      "step": 171360
    },
    {
      "epoch": 20.62214199759326,
      "grad_norm": Infinity,
      "learning_rate": 0.00018067399849567623,
      "loss": 11.1326,
      "step": 171370
    },
    {
      "epoch": 20.623345367027678,
      "grad_norm": Infinity,
      "learning_rate": 0.00018067175046919246,
      "loss": 11.1273,
      "step": 171380
    },
    {
      "epoch": 20.624548736462096,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806695023259568,
      "loss": 11.1192,
      "step": 171390
    },
    {
      "epoch": 20.62575210589651,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806672540659724,
      "loss": 11.2,
      "step": 171400
    },
    {
      "epoch": 20.626955475330927,
      "grad_norm": Infinity,
      "learning_rate": 0.00018066500568924258,
      "loss": 11.087,
      "step": 171410
    },
    {
      "epoch": 20.628158844765345,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806627571957706,
      "loss": 11.1723,
      "step": 171420
    },
    {
      "epoch": 20.62936221419976,
      "grad_norm": Infinity,
      "learning_rate": 0.00018066050858555967,
      "loss": 11.114,
      "step": 171430
    },
    {
      "epoch": 20.630565583634176,
      "grad_norm": Infinity,
      "learning_rate": 0.00018065825985861308,
      "loss": 11.0254,
      "step": 171440
    },
    {
      "epoch": 20.63176895306859,
      "grad_norm": Infinity,
      "learning_rate": 0.00018065601101493407,
      "loss": 11.2022,
      "step": 171450
    },
    {
      "epoch": 20.632972322503008,
      "grad_norm": Infinity,
      "learning_rate": 0.00018065376205452592,
      "loss": 11.153,
      "step": 171460
    },
    {
      "epoch": 20.634175691937426,
      "grad_norm": Infinity,
      "learning_rate": 0.00018065151297739184,
      "loss": 11.0004,
      "step": 171470
    },
    {
      "epoch": 20.63537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806492637835351,
      "loss": 11.1548,
      "step": 171480
    },
    {
      "epoch": 20.636582430806257,
      "grad_norm": Infinity,
      "learning_rate": 0.00018064701447295897,
      "loss": 11.223,
      "step": 171490
    },
    {
      "epoch": 20.637785800240675,
      "grad_norm": Infinity,
      "learning_rate": 0.00018064476504566672,
      "loss": 11.1358,
      "step": 171500
    },
    {
      "epoch": 20.63898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00018064251550166158,
      "loss": 11.1763,
      "step": 171510
    },
    {
      "epoch": 20.640192539109506,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806402658409468,
      "loss": 11.2437,
      "step": 171520
    },
    {
      "epoch": 20.641395908543924,
      "grad_norm": Infinity,
      "learning_rate": 0.00018063801606352567,
      "loss": 11.0512,
      "step": 171530
    },
    {
      "epoch": 20.642599277978338,
      "grad_norm": Infinity,
      "learning_rate": 0.00018063576616940137,
      "loss": 11.046,
      "step": 171540
    },
    {
      "epoch": 20.643802647412755,
      "grad_norm": Infinity,
      "learning_rate": 0.00018063351615857725,
      "loss": 11.0892,
      "step": 171550
    },
    {
      "epoch": 20.645006016847173,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806312660310565,
      "loss": 11.1015,
      "step": 171560
    },
    {
      "epoch": 20.646209386281587,
      "grad_norm": Infinity,
      "learning_rate": 0.00018062901578684242,
      "loss": 11.0387,
      "step": 171570
    },
    {
      "epoch": 20.647412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.00018062676542593824,
      "loss": 11.2202,
      "step": 171580
    },
    {
      "epoch": 20.648616125150422,
      "grad_norm": Infinity,
      "learning_rate": 0.00018062451494834723,
      "loss": 11.2432,
      "step": 171590
    },
    {
      "epoch": 20.649819494584836,
      "grad_norm": Infinity,
      "learning_rate": 0.00018062226435407261,
      "loss": 11.0643,
      "step": 171600
    },
    {
      "epoch": 20.651022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806200136431177,
      "loss": 11.1514,
      "step": 171610
    },
    {
      "epoch": 20.65222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.00018061776281548574,
      "loss": 11.2147,
      "step": 171620
    },
    {
      "epoch": 20.653429602888085,
      "grad_norm": Infinity,
      "learning_rate": 0.00018061551187117995,
      "loss": 11.0922,
      "step": 171630
    },
    {
      "epoch": 20.654632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00018061326081020363,
      "loss": 11.2056,
      "step": 171640
    },
    {
      "epoch": 20.65583634175692,
      "grad_norm": Infinity,
      "learning_rate": 0.00018061100963256,
      "loss": 11.0293,
      "step": 171650
    },
    {
      "epoch": 20.657039711191334,
      "grad_norm": Infinity,
      "learning_rate": 0.00018060875833825232,
      "loss": 11.2296,
      "step": 171660
    },
    {
      "epoch": 20.658243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.00018060650692728388,
      "loss": 11.2358,
      "step": 171670
    },
    {
      "epoch": 20.65944645006017,
      "grad_norm": Infinity,
      "learning_rate": 0.00018060425539965795,
      "loss": 11.1237,
      "step": 171680
    },
    {
      "epoch": 20.660649819494584,
      "grad_norm": Infinity,
      "learning_rate": 0.0001806020037553777,
      "loss": 11.1111,
      "step": 171690
    },
    {
      "epoch": 20.661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00018059975199444654,
      "loss": 11.1746,
      "step": 171700
    },
    {
      "epoch": 20.66305655836342,
      "grad_norm": Infinity,
      "learning_rate": 0.00018059750011686758,
      "loss": 11.0923,
      "step": 171710
    },
    {
      "epoch": 20.664259927797833,
      "grad_norm": Infinity,
      "learning_rate": 0.00018059524812264415,
      "loss": 11.0619,
      "step": 171720
    },
    {
      "epoch": 20.66546329723225,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805929960117795,
      "loss": 11.2698,
      "step": 171730
    },
    {
      "epoch": 20.666666666666668,
      "grad_norm": Infinity,
      "learning_rate": 0.00018059074378427688,
      "loss": 11.1528,
      "step": 171740
    },
    {
      "epoch": 20.667870036101082,
      "grad_norm": Infinity,
      "learning_rate": 0.00018058849144013956,
      "loss": 11.1561,
      "step": 171750
    },
    {
      "epoch": 20.6690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805862389793708,
      "loss": 11.1256,
      "step": 171760
    },
    {
      "epoch": 20.670276774969917,
      "grad_norm": Infinity,
      "learning_rate": 0.00018058398640197386,
      "loss": 11.1578,
      "step": 171770
    },
    {
      "epoch": 20.67148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00018058173370795198,
      "loss": 11.0442,
      "step": 171780
    },
    {
      "epoch": 20.67268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.00018057948089730843,
      "loss": 11.1268,
      "step": 171790
    },
    {
      "epoch": 20.673886883273166,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805772279700465,
      "loss": 11.0373,
      "step": 171800
    },
    {
      "epoch": 20.67509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805749749261694,
      "loss": 11.1178,
      "step": 171810
    },
    {
      "epoch": 20.676293622141998,
      "grad_norm": Infinity,
      "learning_rate": 0.00018057272176568047,
      "loss": 11.2226,
      "step": 171820
    },
    {
      "epoch": 20.677496991576415,
      "grad_norm": Infinity,
      "learning_rate": 0.00018057046848858288,
      "loss": 11.1206,
      "step": 171830
    },
    {
      "epoch": 20.67870036101083,
      "grad_norm": Infinity,
      "learning_rate": 0.00018056821509487993,
      "loss": 11.0796,
      "step": 171840
    },
    {
      "epoch": 20.679903730445247,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805659615845749,
      "loss": 11.0635,
      "step": 171850
    },
    {
      "epoch": 20.681107099879664,
      "grad_norm": Infinity,
      "learning_rate": 0.00018056370795767101,
      "loss": 11.1728,
      "step": 171860
    },
    {
      "epoch": 20.68231046931408,
      "grad_norm": Infinity,
      "learning_rate": 0.00018056145421417155,
      "loss": 11.1653,
      "step": 171870
    },
    {
      "epoch": 20.683513838748496,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805592003540798,
      "loss": 11.1041,
      "step": 171880
    },
    {
      "epoch": 20.684717208182914,
      "grad_norm": Infinity,
      "learning_rate": 0.00018055694637739898,
      "loss": 11.285,
      "step": 171890
    },
    {
      "epoch": 20.685920577617328,
      "grad_norm": Infinity,
      "learning_rate": 0.00018055469228413234,
      "loss": 11.2047,
      "step": 171900
    },
    {
      "epoch": 20.687123947051745,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805524380742832,
      "loss": 11.1747,
      "step": 171910
    },
    {
      "epoch": 20.688327316486163,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805501837478548,
      "loss": 11.1744,
      "step": 171920
    },
    {
      "epoch": 20.689530685920577,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805479293048504,
      "loss": 11.276,
      "step": 171930
    },
    {
      "epoch": 20.690734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.00018054567474527325,
      "loss": 11.0996,
      "step": 171940
    },
    {
      "epoch": 20.691937424789412,
      "grad_norm": Infinity,
      "learning_rate": 0.00018054342006912664,
      "loss": 11.1789,
      "step": 171950
    },
    {
      "epoch": 20.693140794223826,
      "grad_norm": Infinity,
      "learning_rate": 0.00018054116527641378,
      "loss": 11.0768,
      "step": 171960
    },
    {
      "epoch": 20.694344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00018053891036713798,
      "loss": 11.0863,
      "step": 171970
    },
    {
      "epoch": 20.69554753309266,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805366553413025,
      "loss": 11.0095,
      "step": 171980
    },
    {
      "epoch": 20.696750902527075,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805344001989106,
      "loss": 11.1681,
      "step": 171990
    },
    {
      "epoch": 20.697954271961493,
      "grad_norm": Infinity,
      "learning_rate": 0.00018053214493996553,
      "loss": 11.1088,
      "step": 172000
    },
    {
      "epoch": 20.69915764139591,
      "grad_norm": Infinity,
      "learning_rate": 0.00018052988956447058,
      "loss": 11.1341,
      "step": 172010
    },
    {
      "epoch": 20.700361010830324,
      "grad_norm": Infinity,
      "learning_rate": 0.000180527634072429,
      "loss": 11.1545,
      "step": 172020
    },
    {
      "epoch": 20.70156438026474,
      "grad_norm": Infinity,
      "learning_rate": 0.00018052537846384401,
      "loss": 11.1379,
      "step": 172030
    },
    {
      "epoch": 20.70276774969916,
      "grad_norm": Infinity,
      "learning_rate": 0.00018052312273871893,
      "loss": 11.1307,
      "step": 172040
    },
    {
      "epoch": 20.703971119133573,
      "grad_norm": Infinity,
      "learning_rate": 0.00018052086689705705,
      "loss": 11.2065,
      "step": 172050
    },
    {
      "epoch": 20.70517448856799,
      "grad_norm": Infinity,
      "learning_rate": 0.00018051861093886155,
      "loss": 11.1411,
      "step": 172060
    },
    {
      "epoch": 20.706377858002405,
      "grad_norm": Infinity,
      "learning_rate": 0.00018051635486413578,
      "loss": 11.169,
      "step": 172070
    },
    {
      "epoch": 20.707581227436823,
      "grad_norm": Infinity,
      "learning_rate": 0.00018051409867288293,
      "loss": 11.1897,
      "step": 172080
    },
    {
      "epoch": 20.70878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00018051184236510633,
      "loss": 11.0845,
      "step": 172090
    },
    {
      "epoch": 20.709987966305654,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805095859408092,
      "loss": 11.0914,
      "step": 172100
    },
    {
      "epoch": 20.71119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.00018050732939999483,
      "loss": 11.1459,
      "step": 172110
    },
    {
      "epoch": 20.71239470517449,
      "grad_norm": Infinity,
      "learning_rate": 0.00018050507274266648,
      "loss": 11.2425,
      "step": 172120
    },
    {
      "epoch": 20.713598074608903,
      "grad_norm": Infinity,
      "learning_rate": 0.0001805028159688274,
      "loss": 11.1846,
      "step": 172130
    },
    {
      "epoch": 20.71480144404332,
      "grad_norm": Infinity,
      "learning_rate": 0.00018050055907848088,
      "loss": 11.0592,
      "step": 172140
    },
    {
      "epoch": 20.71600481347774,
      "grad_norm": Infinity,
      "learning_rate": 0.00018049830207163015,
      "loss": 11.1487,
      "step": 172150
    },
    {
      "epoch": 20.717208182912152,
      "grad_norm": Infinity,
      "learning_rate": 0.00018049604494827854,
      "loss": 11.1103,
      "step": 172160
    },
    {
      "epoch": 20.71841155234657,
      "grad_norm": Infinity,
      "learning_rate": 0.00018049378770842927,
      "loss": 11.1495,
      "step": 172170
    },
    {
      "epoch": 20.719614921780988,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804915303520856,
      "loss": 11.1195,
      "step": 172180
    },
    {
      "epoch": 20.7208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00018048927287925084,
      "loss": 11.0399,
      "step": 172190
    },
    {
      "epoch": 20.72202166064982,
      "grad_norm": Infinity,
      "learning_rate": 0.00018048701528992819,
      "loss": 11.1366,
      "step": 172200
    },
    {
      "epoch": 20.723225030084237,
      "grad_norm": Infinity,
      "learning_rate": 0.000180484757584121,
      "loss": 11.1213,
      "step": 172210
    },
    {
      "epoch": 20.72442839951865,
      "grad_norm": Infinity,
      "learning_rate": 0.00018048249976183246,
      "loss": 10.9934,
      "step": 172220
    },
    {
      "epoch": 20.72563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00018048024182306592,
      "loss": 11.1397,
      "step": 172230
    },
    {
      "epoch": 20.726835138387486,
      "grad_norm": Infinity,
      "learning_rate": 0.00018047798376782454,
      "loss": 11.0982,
      "step": 172240
    },
    {
      "epoch": 20.7280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804757255961117,
      "loss": 11.1895,
      "step": 172250
    },
    {
      "epoch": 20.729241877256317,
      "grad_norm": Infinity,
      "learning_rate": 0.00018047346730793058,
      "loss": 11.164,
      "step": 172260
    },
    {
      "epoch": 20.730445246690735,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804712089032845,
      "loss": 11.2391,
      "step": 172270
    },
    {
      "epoch": 20.73164861612515,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804689503821767,
      "loss": 11.1563,
      "step": 172280
    },
    {
      "epoch": 20.732851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.00018046669174461046,
      "loss": 11.1299,
      "step": 172290
    },
    {
      "epoch": 20.734055354993984,
      "grad_norm": Infinity,
      "learning_rate": 0.00018046443299058907,
      "loss": 11.1237,
      "step": 172300
    },
    {
      "epoch": 20.735258724428398,
      "grad_norm": Infinity,
      "learning_rate": 0.00018046217412011578,
      "loss": 11.1034,
      "step": 172310
    },
    {
      "epoch": 20.736462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.00018045991513319383,
      "loss": 11.186,
      "step": 172320
    },
    {
      "epoch": 20.737665463297233,
      "grad_norm": Infinity,
      "learning_rate": 0.00018045765602982657,
      "loss": 11.1002,
      "step": 172330
    },
    {
      "epoch": 20.738868832731647,
      "grad_norm": Infinity,
      "learning_rate": 0.00018045539681001718,
      "loss": 11.045,
      "step": 172340
    },
    {
      "epoch": 20.740072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00018045313747376898,
      "loss": 11.2122,
      "step": 172350
    },
    {
      "epoch": 20.741275571600482,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804508780210852,
      "loss": 11.0942,
      "step": 172360
    },
    {
      "epoch": 20.742478941034896,
      "grad_norm": Infinity,
      "learning_rate": 0.00018044861845196915,
      "loss": 11.0911,
      "step": 172370
    },
    {
      "epoch": 20.743682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804463587664241,
      "loss": 11.2691,
      "step": 172380
    },
    {
      "epoch": 20.74488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804440989644533,
      "loss": 11.1032,
      "step": 172390
    },
    {
      "epoch": 20.746089049338146,
      "grad_norm": Infinity,
      "learning_rate": 0.00018044183904606005,
      "loss": 10.9427,
      "step": 172400
    },
    {
      "epoch": 20.747292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00018043957901124754,
      "loss": 11.1419,
      "step": 172410
    },
    {
      "epoch": 20.74849578820698,
      "grad_norm": Infinity,
      "learning_rate": 0.00018043731886001915,
      "loss": 11.0311,
      "step": 172420
    },
    {
      "epoch": 20.749699157641395,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804350585923781,
      "loss": 11.1091,
      "step": 172430
    },
    {
      "epoch": 20.750902527075812,
      "grad_norm": Infinity,
      "learning_rate": 0.00018043279820832762,
      "loss": 11.2184,
      "step": 172440
    },
    {
      "epoch": 20.75210589651023,
      "grad_norm": Infinity,
      "learning_rate": 0.00018043053770787107,
      "loss": 11.0518,
      "step": 172450
    },
    {
      "epoch": 20.753309265944644,
      "grad_norm": Infinity,
      "learning_rate": 0.00018042827709101165,
      "loss": 11.1686,
      "step": 172460
    },
    {
      "epoch": 20.75451263537906,
      "grad_norm": Infinity,
      "learning_rate": 0.00018042601635775266,
      "loss": 11.1105,
      "step": 172470
    },
    {
      "epoch": 20.75571600481348,
      "grad_norm": Infinity,
      "learning_rate": 0.00018042375550809738,
      "loss": 11.1936,
      "step": 172480
    },
    {
      "epoch": 20.756919374247893,
      "grad_norm": Infinity,
      "learning_rate": 0.00018042149454204905,
      "loss": 11.0519,
      "step": 172490
    },
    {
      "epoch": 20.75812274368231,
      "grad_norm": Infinity,
      "learning_rate": 0.00018041923345961096,
      "loss": 11.1132,
      "step": 172500
    },
    {
      "epoch": 20.759326113116728,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804169722607864,
      "loss": 11.1417,
      "step": 172510
    },
    {
      "epoch": 20.760529482551142,
      "grad_norm": Infinity,
      "learning_rate": 0.00018041471094557864,
      "loss": 11.1195,
      "step": 172520
    },
    {
      "epoch": 20.76173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804124495139909,
      "loss": 11.1559,
      "step": 172530
    },
    {
      "epoch": 20.762936221419977,
      "grad_norm": Infinity,
      "learning_rate": 0.00018041018796602652,
      "loss": 11.1252,
      "step": 172540
    },
    {
      "epoch": 20.76413959085439,
      "grad_norm": Infinity,
      "learning_rate": 0.00018040792630168875,
      "loss": 11.1344,
      "step": 172550
    },
    {
      "epoch": 20.76534296028881,
      "grad_norm": Infinity,
      "learning_rate": 0.00018040566452098085,
      "loss": 11.1014,
      "step": 172560
    },
    {
      "epoch": 20.766546329723226,
      "grad_norm": Infinity,
      "learning_rate": 0.0001804034026239061,
      "loss": 11.2088,
      "step": 172570
    },
    {
      "epoch": 20.76774969915764,
      "grad_norm": Infinity,
      "learning_rate": 0.00018040114061046775,
      "loss": 11.0848,
      "step": 172580
    },
    {
      "epoch": 20.768953068592058,
      "grad_norm": Infinity,
      "learning_rate": 0.00018039887848066914,
      "loss": 11.1741,
      "step": 172590
    },
    {
      "epoch": 20.770156438026476,
      "grad_norm": Infinity,
      "learning_rate": 0.00018039661623451346,
      "loss": 11.0817,
      "step": 172600
    },
    {
      "epoch": 20.77135980746089,
      "grad_norm": Infinity,
      "learning_rate": 0.00018039435387200408,
      "loss": 11.1698,
      "step": 172610
    },
    {
      "epoch": 20.772563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.0001803920913931442,
      "loss": 11.2413,
      "step": 172620
    },
    {
      "epoch": 20.773766546329725,
      "grad_norm": Infinity,
      "learning_rate": 0.00018038982879793708,
      "loss": 11.2063,
      "step": 172630
    },
    {
      "epoch": 20.77496991576414,
      "grad_norm": Infinity,
      "learning_rate": 0.00018038756608638608,
      "loss": 11.2472,
      "step": 172640
    },
    {
      "epoch": 20.776173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.0001803853032584944,
      "loss": 11.2178,
      "step": 172650
    },
    {
      "epoch": 20.777376654632974,
      "grad_norm": Infinity,
      "learning_rate": 0.00018038304031426535,
      "loss": 11.1067,
      "step": 172660
    },
    {
      "epoch": 20.778580024067388,
      "grad_norm": Infinity,
      "learning_rate": 0.0001803807772537022,
      "loss": 11.0927,
      "step": 172670
    },
    {
      "epoch": 20.779783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.00018037851407680822,
      "loss": 11.0596,
      "step": 172680
    },
    {
      "epoch": 20.780986762936223,
      "grad_norm": Infinity,
      "learning_rate": 0.00018037625078358666,
      "loss": 11.2017,
      "step": 172690
    },
    {
      "epoch": 20.782190132370637,
      "grad_norm": Infinity,
      "learning_rate": 0.00018037398737404082,
      "loss": 11.2352,
      "step": 172700
    },
    {
      "epoch": 20.783393501805055,
      "grad_norm": Infinity,
      "learning_rate": 0.000180371723848174,
      "loss": 11.1597,
      "step": 172710
    },
    {
      "epoch": 20.784596871239472,
      "grad_norm": Infinity,
      "learning_rate": 0.00018036946020598947,
      "loss": 11.1804,
      "step": 172720
    },
    {
      "epoch": 20.785800240673886,
      "grad_norm": Infinity,
      "learning_rate": 0.00018036719644749045,
      "loss": 11.2442,
      "step": 172730
    },
    {
      "epoch": 20.787003610108304,
      "grad_norm": Infinity,
      "learning_rate": 0.00018036493257268026,
      "loss": 11.0907,
      "step": 172740
    },
    {
      "epoch": 20.78820697954272,
      "grad_norm": Infinity,
      "learning_rate": 0.00018036266858156218,
      "loss": 11.0912,
      "step": 172750
    },
    {
      "epoch": 20.789410348977135,
      "grad_norm": Infinity,
      "learning_rate": 0.0001803604044741395,
      "loss": 11.0766,
      "step": 172760
    },
    {
      "epoch": 20.790613718411553,
      "grad_norm": Infinity,
      "learning_rate": 0.00018035814025041543,
      "loss": 11.1466,
      "step": 172770
    },
    {
      "epoch": 20.79181708784597,
      "grad_norm": Infinity,
      "learning_rate": 0.00018035587591039333,
      "loss": 11.1713,
      "step": 172780
    },
    {
      "epoch": 20.793020457280385,
      "grad_norm": Infinity,
      "learning_rate": 0.0001803536114540764,
      "loss": 11.219,
      "step": 172790
    },
    {
      "epoch": 20.794223826714802,
      "grad_norm": Infinity,
      "learning_rate": 0.00018035134688146798,
      "loss": 11.1891,
      "step": 172800
    },
    {
      "epoch": 20.79542719614922,
      "grad_norm": Infinity,
      "learning_rate": 0.00018034908219257133,
      "loss": 11.193,
      "step": 172810
    },
    {
      "epoch": 20.796630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00018034681738738972,
      "loss": 11.1871,
      "step": 172820
    },
    {
      "epoch": 20.79783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.00018034455246592642,
      "loss": 11.232,
      "step": 172830
    },
    {
      "epoch": 20.799037304452465,
      "grad_norm": Infinity,
      "learning_rate": 0.0001803422874281847,
      "loss": 11.1497,
      "step": 172840
    },
    {
      "epoch": 20.800240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.0001803400222741679,
      "loss": 11.1892,
      "step": 172850
    },
    {
      "epoch": 20.8014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.00018033775700387918,
      "loss": 11.1423,
      "step": 172860
    },
    {
      "epoch": 20.802647412755714,
      "grad_norm": Infinity,
      "learning_rate": 0.00018033549161732194,
      "loss": 11.1968,
      "step": 172870
    },
    {
      "epoch": 20.803850782190132,
      "grad_norm": Infinity,
      "learning_rate": 0.00018033322611449938,
      "loss": 11.1266,
      "step": 172880
    },
    {
      "epoch": 20.80505415162455,
      "grad_norm": Infinity,
      "learning_rate": 0.00018033096049541484,
      "loss": 11.1276,
      "step": 172890
    },
    {
      "epoch": 20.806257521058964,
      "grad_norm": Infinity,
      "learning_rate": 0.00018032869476007158,
      "loss": 11.2131,
      "step": 172900
    },
    {
      "epoch": 20.80746089049338,
      "grad_norm": Infinity,
      "learning_rate": 0.00018032642890847283,
      "loss": 11.1898,
      "step": 172910
    },
    {
      "epoch": 20.8086642599278,
      "grad_norm": Infinity,
      "learning_rate": 0.00018032416294062192,
      "loss": 11.0419,
      "step": 172920
    },
    {
      "epoch": 20.809867629362213,
      "grad_norm": Infinity,
      "learning_rate": 0.00018032189685652212,
      "loss": 11.1756,
      "step": 172930
    },
    {
      "epoch": 20.81107099879663,
      "grad_norm": Infinity,
      "learning_rate": 0.0001803196306561767,
      "loss": 11.1386,
      "step": 172940
    },
    {
      "epoch": 20.812274368231048,
      "grad_norm": Infinity,
      "learning_rate": 0.00018031736433958892,
      "loss": 11.1541,
      "step": 172950
    },
    {
      "epoch": 20.813477737665462,
      "grad_norm": Infinity,
      "learning_rate": 0.00018031509790676214,
      "loss": 11.2423,
      "step": 172960
    },
    {
      "epoch": 20.81468110709988,
      "grad_norm": Infinity,
      "learning_rate": 0.00018031283135769952,
      "loss": 11.0973,
      "step": 172970
    },
    {
      "epoch": 20.815884476534297,
      "grad_norm": Infinity,
      "learning_rate": 0.00018031056469240445,
      "loss": 11.1409,
      "step": 172980
    },
    {
      "epoch": 20.81708784596871,
      "grad_norm": Infinity,
      "learning_rate": 0.00018030829791088016,
      "loss": 11.023,
      "step": 172990
    },
    {
      "epoch": 20.81829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.0001803060310131299,
      "loss": 11.0497,
      "step": 173000
    },
    {
      "epoch": 20.819494584837546,
      "grad_norm": Infinity,
      "learning_rate": 0.00018030376399915703,
      "loss": 11.2143,
      "step": 173010
    },
    {
      "epoch": 20.82069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00018030149686896476,
      "loss": 10.9892,
      "step": 173020
    },
    {
      "epoch": 20.821901323706378,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802992296225564,
      "loss": 11.1598,
      "step": 173030
    },
    {
      "epoch": 20.823104693140795,
      "grad_norm": Infinity,
      "learning_rate": 0.00018029696225993524,
      "loss": 11.2072,
      "step": 173040
    },
    {
      "epoch": 20.82430806257521,
      "grad_norm": Infinity,
      "learning_rate": 0.00018029469478110453,
      "loss": 11.176,
      "step": 173050
    },
    {
      "epoch": 20.825511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802924271860676,
      "loss": 10.9815,
      "step": 173060
    },
    {
      "epoch": 20.826714801444044,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802901594748277,
      "loss": 11.1104,
      "step": 173070
    },
    {
      "epoch": 20.82791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00018028789164738812,
      "loss": 11.1583,
      "step": 173080
    },
    {
      "epoch": 20.829121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802856237037521,
      "loss": 11.2288,
      "step": 173090
    },
    {
      "epoch": 20.830324909747294,
      "grad_norm": Infinity,
      "learning_rate": 0.000180283355643923,
      "loss": 11.23,
      "step": 173100
    },
    {
      "epoch": 20.831528279181708,
      "grad_norm": Infinity,
      "learning_rate": 0.00018028108746790403,
      "loss": 11.2722,
      "step": 173110
    },
    {
      "epoch": 20.832731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.00018027881917569854,
      "loss": 11.1888,
      "step": 173120
    },
    {
      "epoch": 20.833935018050543,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802765507673098,
      "loss": 11.0948,
      "step": 173130
    },
    {
      "epoch": 20.835138387484957,
      "grad_norm": Infinity,
      "learning_rate": 0.00018027428224274102,
      "loss": 11.1078,
      "step": 173140
    },
    {
      "epoch": 20.836341756919374,
      "grad_norm": Infinity,
      "learning_rate": 0.00018027201360199555,
      "loss": 11.2255,
      "step": 173150
    },
    {
      "epoch": 20.837545126353792,
      "grad_norm": Infinity,
      "learning_rate": 0.00018026974484507667,
      "loss": 11.0824,
      "step": 173160
    },
    {
      "epoch": 20.838748495788206,
      "grad_norm": Infinity,
      "learning_rate": 0.00018026747597198764,
      "loss": 11.1556,
      "step": 173170
    },
    {
      "epoch": 20.839951865222623,
      "grad_norm": Infinity,
      "learning_rate": 0.00018026520698273175,
      "loss": 11.1401,
      "step": 173180
    },
    {
      "epoch": 20.84115523465704,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802629378773123,
      "loss": 11.2589,
      "step": 173190
    },
    {
      "epoch": 20.842358604091455,
      "grad_norm": Infinity,
      "learning_rate": 0.00018026066865573255,
      "loss": 11.2058,
      "step": 173200
    },
    {
      "epoch": 20.843561973525873,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802583993179958,
      "loss": 11.1996,
      "step": 173210
    },
    {
      "epoch": 20.84476534296029,
      "grad_norm": Infinity,
      "learning_rate": 0.00018025612986410534,
      "loss": 11.0883,
      "step": 173220
    },
    {
      "epoch": 20.845968712394704,
      "grad_norm": Infinity,
      "learning_rate": 0.00018025386029406443,
      "loss": 11.1341,
      "step": 173230
    },
    {
      "epoch": 20.84717208182912,
      "grad_norm": Infinity,
      "learning_rate": 0.00018025159060787639,
      "loss": 11.1927,
      "step": 173240
    },
    {
      "epoch": 20.84837545126354,
      "grad_norm": Infinity,
      "learning_rate": 0.00018024932080554446,
      "loss": 11.1739,
      "step": 173250
    },
    {
      "epoch": 20.849578820697953,
      "grad_norm": Infinity,
      "learning_rate": 0.00018024705088707198,
      "loss": 11.0904,
      "step": 173260
    },
    {
      "epoch": 20.85078219013237,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802447808524622,
      "loss": 11.159,
      "step": 173270
    },
    {
      "epoch": 20.85198555956679,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802425107017184,
      "loss": 11.0507,
      "step": 173280
    },
    {
      "epoch": 20.853188929001202,
      "grad_norm": Infinity,
      "learning_rate": 0.00018024024043484388,
      "loss": 11.1128,
      "step": 173290
    },
    {
      "epoch": 20.85439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00018023797005184192,
      "loss": 11.2208,
      "step": 173300
    },
    {
      "epoch": 20.855595667870038,
      "grad_norm": Infinity,
      "learning_rate": 0.00018023569955271577,
      "loss": 11.1155,
      "step": 173310
    },
    {
      "epoch": 20.85679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802334289374688,
      "loss": 11.174,
      "step": 173320
    },
    {
      "epoch": 20.85800240673887,
      "grad_norm": Infinity,
      "learning_rate": 0.00018023115820610427,
      "loss": 11.0918,
      "step": 173330
    },
    {
      "epoch": 20.859205776173287,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802288873586254,
      "loss": 10.9758,
      "step": 173340
    },
    {
      "epoch": 20.8604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802266163950355,
      "loss": 11.2076,
      "step": 173350
    },
    {
      "epoch": 20.86161251504212,
      "grad_norm": Infinity,
      "learning_rate": 0.00018022434531533794,
      "loss": 11.1549,
      "step": 173360
    },
    {
      "epoch": 20.862815884476536,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802220741195359,
      "loss": 11.2134,
      "step": 173370
    },
    {
      "epoch": 20.86401925391095,
      "grad_norm": Infinity,
      "learning_rate": 0.00018021980280763275,
      "loss": 11.2158,
      "step": 173380
    },
    {
      "epoch": 20.865222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.00018021753137963172,
      "loss": 11.0854,
      "step": 173390
    },
    {
      "epoch": 20.866425992779785,
      "grad_norm": Infinity,
      "learning_rate": 0.00018021525983553613,
      "loss": 11.2,
      "step": 173400
    },
    {
      "epoch": 20.8676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00018021298817534923,
      "loss": 11.0791,
      "step": 173410
    },
    {
      "epoch": 20.868832731648617,
      "grad_norm": Infinity,
      "learning_rate": 0.00018021071639907435,
      "loss": 11.0264,
      "step": 173420
    },
    {
      "epoch": 20.870036101083034,
      "grad_norm": Infinity,
      "learning_rate": 0.00018020844450671475,
      "loss": 11.1703,
      "step": 173430
    },
    {
      "epoch": 20.871239470517448,
      "grad_norm": Infinity,
      "learning_rate": 0.00018020617249827375,
      "loss": 11.1173,
      "step": 173440
    },
    {
      "epoch": 20.872442839951866,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802039003737546,
      "loss": 11.25,
      "step": 173450
    },
    {
      "epoch": 20.87364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.0001802016281331606,
      "loss": 11.1557,
      "step": 173460
    },
    {
      "epoch": 20.874849578820697,
      "grad_norm": Infinity,
      "learning_rate": 0.00018019935577649502,
      "loss": 11.0223,
      "step": 173470
    },
    {
      "epoch": 20.876052948255115,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801970833037612,
      "loss": 11.1554,
      "step": 173480
    },
    {
      "epoch": 20.87725631768953,
      "grad_norm": Infinity,
      "learning_rate": 0.00018019481071496242,
      "loss": 11.0719,
      "step": 173490
    },
    {
      "epoch": 20.878459687123947,
      "grad_norm": Infinity,
      "learning_rate": 0.00018019253801010194,
      "loss": 11.2168,
      "step": 173500
    },
    {
      "epoch": 20.879663056558364,
      "grad_norm": Infinity,
      "learning_rate": 0.00018019026518918303,
      "loss": 11.0576,
      "step": 173510
    },
    {
      "epoch": 20.880866425992778,
      "grad_norm": Infinity,
      "learning_rate": 0.00018018799225220903,
      "loss": 11.0965,
      "step": 173520
    },
    {
      "epoch": 20.882069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801857191991832,
      "loss": 11.0693,
      "step": 173530
    },
    {
      "epoch": 20.883273164861613,
      "grad_norm": Infinity,
      "learning_rate": 0.00018018344603010884,
      "loss": 11.1587,
      "step": 173540
    },
    {
      "epoch": 20.884476534296027,
      "grad_norm": Infinity,
      "learning_rate": 0.00018018117274498923,
      "loss": 11.2679,
      "step": 173550
    },
    {
      "epoch": 20.885679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00018017889934382768,
      "loss": 11.0494,
      "step": 173560
    },
    {
      "epoch": 20.886883273164862,
      "grad_norm": Infinity,
      "learning_rate": 0.00018017662582662744,
      "loss": 11.1472,
      "step": 173570
    },
    {
      "epoch": 20.888086642599276,
      "grad_norm": Infinity,
      "learning_rate": 0.00018017435219339187,
      "loss": 11.0953,
      "step": 173580
    },
    {
      "epoch": 20.889290012033694,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801720784441242,
      "loss": 11.0539,
      "step": 173590
    },
    {
      "epoch": 20.89049338146811,
      "grad_norm": Infinity,
      "learning_rate": 0.00018016980457882773,
      "loss": 11.0687,
      "step": 173600
    },
    {
      "epoch": 20.891696750902526,
      "grad_norm": Infinity,
      "learning_rate": 0.00018016753059750578,
      "loss": 11.1559,
      "step": 173610
    },
    {
      "epoch": 20.892900120336943,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801652565001616,
      "loss": 11.1049,
      "step": 173620
    },
    {
      "epoch": 20.89410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00018016298228679852,
      "loss": 11.1504,
      "step": 173630
    },
    {
      "epoch": 20.895306859205775,
      "grad_norm": Infinity,
      "learning_rate": 0.00018016070795741978,
      "loss": 11.1768,
      "step": 173640
    },
    {
      "epoch": 20.896510228640192,
      "grad_norm": Infinity,
      "learning_rate": 0.00018015843351202872,
      "loss": 11.1212,
      "step": 173650
    },
    {
      "epoch": 20.89771359807461,
      "grad_norm": Infinity,
      "learning_rate": 0.00018015615895062862,
      "loss": 11.1894,
      "step": 173660
    },
    {
      "epoch": 20.898916967509024,
      "grad_norm": Infinity,
      "learning_rate": 0.00018015388427322277,
      "loss": 11.0539,
      "step": 173670
    },
    {
      "epoch": 20.90012033694344,
      "grad_norm": Infinity,
      "learning_rate": 0.00018015160947981447,
      "loss": 11.114,
      "step": 173680
    },
    {
      "epoch": 20.90132370637786,
      "grad_norm": Infinity,
      "learning_rate": 0.00018014933457040697,
      "loss": 11.0948,
      "step": 173690
    },
    {
      "epoch": 20.902527075812273,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801470595450036,
      "loss": 11.0909,
      "step": 173700
    },
    {
      "epoch": 20.90373044524669,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801447844036077,
      "loss": 11.1265,
      "step": 173710
    },
    {
      "epoch": 20.904933814681108,
      "grad_norm": Infinity,
      "learning_rate": 0.00018014250914622247,
      "loss": 11.0683,
      "step": 173720
    },
    {
      "epoch": 20.906137184115522,
      "grad_norm": Infinity,
      "learning_rate": 0.00018014023377285122,
      "loss": 11.0873,
      "step": 173730
    },
    {
      "epoch": 20.90734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801379582834973,
      "loss": 11.1706,
      "step": 173740
    },
    {
      "epoch": 20.908543922984357,
      "grad_norm": Infinity,
      "learning_rate": 0.00018013568267816397,
      "loss": 11.1743,
      "step": 173750
    },
    {
      "epoch": 20.90974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801334069568545,
      "loss": 11.0408,
      "step": 173760
    },
    {
      "epoch": 20.91095066185319,
      "grad_norm": Infinity,
      "learning_rate": 0.00018013113111957222,
      "loss": 11.0911,
      "step": 173770
    },
    {
      "epoch": 20.912154031287606,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801288551663204,
      "loss": 11.1863,
      "step": 173780
    },
    {
      "epoch": 20.91335740072202,
      "grad_norm": Infinity,
      "learning_rate": 0.00018012657909710237,
      "loss": 11.1792,
      "step": 173790
    },
    {
      "epoch": 20.914560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00018012430291192136,
      "loss": 11.2082,
      "step": 173800
    },
    {
      "epoch": 20.915764139590856,
      "grad_norm": Infinity,
      "learning_rate": 0.00018012202661078073,
      "loss": 11.1883,
      "step": 173810
    },
    {
      "epoch": 20.91696750902527,
      "grad_norm": Infinity,
      "learning_rate": 0.00018011975019368375,
      "loss": 11.0674,
      "step": 173820
    },
    {
      "epoch": 20.918170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801174736606337,
      "loss": 11.1176,
      "step": 173830
    },
    {
      "epoch": 20.919374247894105,
      "grad_norm": Infinity,
      "learning_rate": 0.00018011519701163387,
      "loss": 11.1708,
      "step": 173840
    },
    {
      "epoch": 20.92057761732852,
      "grad_norm": Infinity,
      "learning_rate": 0.00018011292024668757,
      "loss": 11.1329,
      "step": 173850
    },
    {
      "epoch": 20.921780986762936,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801106433657981,
      "loss": 11.0115,
      "step": 173860
    },
    {
      "epoch": 20.922984356197354,
      "grad_norm": Infinity,
      "learning_rate": 0.0001801083663689688,
      "loss": 11.0324,
      "step": 173870
    },
    {
      "epoch": 20.924187725631768,
      "grad_norm": Infinity,
      "learning_rate": 0.00018010608925620285,
      "loss": 11.27,
      "step": 173880
    },
    {
      "epoch": 20.925391095066185,
      "grad_norm": Infinity,
      "learning_rate": 0.00018010381202750365,
      "loss": 11.1616,
      "step": 173890
    },
    {
      "epoch": 20.926594464500603,
      "grad_norm": Infinity,
      "learning_rate": 0.00018010153468287442,
      "loss": 11.1951,
      "step": 173900
    },
    {
      "epoch": 20.927797833935017,
      "grad_norm": Infinity,
      "learning_rate": 0.00018009925722231853,
      "loss": 11.2645,
      "step": 173910
    },
    {
      "epoch": 20.929001203369435,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800969796458392,
      "loss": 11.157,
      "step": 173920
    },
    {
      "epoch": 20.930204572803852,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800947019534398,
      "loss": 11.1395,
      "step": 173930
    },
    {
      "epoch": 20.931407942238266,
      "grad_norm": Infinity,
      "learning_rate": 0.00018009242414512356,
      "loss": 11.1996,
      "step": 173940
    },
    {
      "epoch": 20.932611311672684,
      "grad_norm": Infinity,
      "learning_rate": 0.00018009014622089384,
      "loss": 11.1171,
      "step": 173950
    },
    {
      "epoch": 20.9338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800878681807539,
      "loss": 11.2146,
      "step": 173960
    },
    {
      "epoch": 20.935018050541515,
      "grad_norm": Infinity,
      "learning_rate": 0.00018008559002470704,
      "loss": 11.1485,
      "step": 173970
    },
    {
      "epoch": 20.936221419975933,
      "grad_norm": Infinity,
      "learning_rate": 0.00018008331175275655,
      "loss": 11.1549,
      "step": 173980
    },
    {
      "epoch": 20.93742478941035,
      "grad_norm": Infinity,
      "learning_rate": 0.00018008103336490574,
      "loss": 11.1026,
      "step": 173990
    },
    {
      "epoch": 20.938628158844764,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800787548611579,
      "loss": 11.1206,
      "step": 174000
    },
    {
      "epoch": 20.939831528279182,
      "grad_norm": Infinity,
      "learning_rate": 0.00018007647624151634,
      "loss": 11.0992,
      "step": 174010
    },
    {
      "epoch": 20.9410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00018007419750598434,
      "loss": 11.1332,
      "step": 174020
    },
    {
      "epoch": 20.942238267148014,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800719186545652,
      "loss": 11.1261,
      "step": 174030
    },
    {
      "epoch": 20.94344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800696396872622,
      "loss": 11.1842,
      "step": 174040
    },
    {
      "epoch": 20.94464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00018006736060407875,
      "loss": 11.1692,
      "step": 174050
    },
    {
      "epoch": 20.945848375451263,
      "grad_norm": Infinity,
      "learning_rate": 0.00018006508140501799,
      "loss": 11.1657,
      "step": 174060
    },
    {
      "epoch": 20.94705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00018006280209008333,
      "loss": 11.0909,
      "step": 174070
    },
    {
      "epoch": 20.948255114320098,
      "grad_norm": Infinity,
      "learning_rate": 0.00018006052265927798,
      "loss": 11.0538,
      "step": 174080
    },
    {
      "epoch": 20.949458483754512,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800582431126053,
      "loss": 11.0875,
      "step": 174090
    },
    {
      "epoch": 20.95066185318893,
      "grad_norm": Infinity,
      "learning_rate": 0.00018005596345006862,
      "loss": 11.1126,
      "step": 174100
    },
    {
      "epoch": 20.951865222623347,
      "grad_norm": Infinity,
      "learning_rate": 0.00018005368367167117,
      "loss": 11.1622,
      "step": 174110
    },
    {
      "epoch": 20.95306859205776,
      "grad_norm": Infinity,
      "learning_rate": 0.00018005140377741626,
      "loss": 11.1136,
      "step": 174120
    },
    {
      "epoch": 20.95427196149218,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800491237673072,
      "loss": 11.2881,
      "step": 174130
    },
    {
      "epoch": 20.955475330926596,
      "grad_norm": Infinity,
      "learning_rate": 0.00018004684364134732,
      "loss": 11.0797,
      "step": 174140
    },
    {
      "epoch": 20.95667870036101,
      "grad_norm": Infinity,
      "learning_rate": 0.00018004456339953988,
      "loss": 11.1497,
      "step": 174150
    },
    {
      "epoch": 20.957882069795428,
      "grad_norm": Infinity,
      "learning_rate": 0.00018004228304188818,
      "loss": 11.0825,
      "step": 174160
    },
    {
      "epoch": 20.959085439229845,
      "grad_norm": Infinity,
      "learning_rate": 0.00018004000256839557,
      "loss": 11.0937,
      "step": 174170
    },
    {
      "epoch": 20.96028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800377219790653,
      "loss": 11.0773,
      "step": 174180
    },
    {
      "epoch": 20.961492178098677,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800354412739007,
      "loss": 11.1334,
      "step": 174190
    },
    {
      "epoch": 20.96269554753309,
      "grad_norm": Infinity,
      "learning_rate": 0.00018003316045290505,
      "loss": 11.1376,
      "step": 174200
    },
    {
      "epoch": 20.96389891696751,
      "grad_norm": Infinity,
      "learning_rate": 0.00018003087951608162,
      "loss": 11.1124,
      "step": 174210
    },
    {
      "epoch": 20.965102286401926,
      "grad_norm": Infinity,
      "learning_rate": 0.00018002859846343378,
      "loss": 11.175,
      "step": 174220
    },
    {
      "epoch": 20.96630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00018002631729496478,
      "loss": 11.1262,
      "step": 174230
    },
    {
      "epoch": 20.967509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.00018002403601067796,
      "loss": 11.04,
      "step": 174240
    },
    {
      "epoch": 20.968712394705175,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800217546105766,
      "loss": 11.2401,
      "step": 174250
    },
    {
      "epoch": 20.96991576413959,
      "grad_norm": Infinity,
      "learning_rate": 0.00018001947309466402,
      "loss": 11.0645,
      "step": 174260
    },
    {
      "epoch": 20.971119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.00018001719146294346,
      "loss": 11.2181,
      "step": 174270
    },
    {
      "epoch": 20.972322503008424,
      "grad_norm": Infinity,
      "learning_rate": 0.00018001490971541833,
      "loss": 11.1317,
      "step": 174280
    },
    {
      "epoch": 20.97352587244284,
      "grad_norm": Infinity,
      "learning_rate": 0.00018001262785209183,
      "loss": 11.0471,
      "step": 174290
    },
    {
      "epoch": 20.974729241877256,
      "grad_norm": Infinity,
      "learning_rate": 0.0001800103458729673,
      "loss": 11.2109,
      "step": 174300
    },
    {
      "epoch": 20.975932611311674,
      "grad_norm": Infinity,
      "learning_rate": 0.00018000806377804806,
      "loss": 11.1007,
      "step": 174310
    },
    {
      "epoch": 20.977135980746088,
      "grad_norm": Infinity,
      "learning_rate": 0.00018000578156733741,
      "loss": 11.2041,
      "step": 174320
    },
    {
      "epoch": 20.978339350180505,
      "grad_norm": Infinity,
      "learning_rate": 0.00018000349924083862,
      "loss": 11.1238,
      "step": 174330
    },
    {
      "epoch": 20.979542719614923,
      "grad_norm": Infinity,
      "learning_rate": 0.00018000121679855502,
      "loss": 11.1583,
      "step": 174340
    },
    {
      "epoch": 20.980746089049337,
      "grad_norm": Infinity,
      "learning_rate": 0.00017999893424048994,
      "loss": 11.226,
      "step": 174350
    },
    {
      "epoch": 20.981949458483754,
      "grad_norm": Infinity,
      "learning_rate": 0.00017999665156664661,
      "loss": 11.1327,
      "step": 174360
    },
    {
      "epoch": 20.983152827918172,
      "grad_norm": Infinity,
      "learning_rate": 0.0001799943687770284,
      "loss": 11.1759,
      "step": 174370
    },
    {
      "epoch": 20.984356197352586,
      "grad_norm": Infinity,
      "learning_rate": 0.00017999208587163854,
      "loss": 11.2303,
      "step": 174380
    },
    {
      "epoch": 20.985559566787003,
      "grad_norm": Infinity,
      "learning_rate": 0.0001799898028504804,
      "loss": 11.2441,
      "step": 174390
    },
    {
      "epoch": 20.98676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.0001799875197135573,
      "loss": 11.1695,
      "step": 174400
    },
    {
      "epoch": 20.987966305655835,
      "grad_norm": Infinity,
      "learning_rate": 0.00017998523646087247,
      "loss": 10.9997,
      "step": 174410
    },
    {
      "epoch": 20.989169675090253,
      "grad_norm": Infinity,
      "learning_rate": 0.0001799829530924293,
      "loss": 11.2273,
      "step": 174420
    },
    {
      "epoch": 20.99037304452467,
      "grad_norm": Infinity,
      "learning_rate": 0.000179980669608231,
      "loss": 11.1485,
      "step": 174430
    },
    {
      "epoch": 20.991576413959084,
      "grad_norm": Infinity,
      "learning_rate": 0.00017997838600828097,
      "loss": 11.2727,
      "step": 174440
    },
    {
      "epoch": 20.9927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00017997610229258246,
      "loss": 11.1697,
      "step": 174450
    },
    {
      "epoch": 20.99398315282792,
      "grad_norm": Infinity,
      "learning_rate": 0.00017997381846113875,
      "loss": 11.0973,
      "step": 174460
    },
    {
      "epoch": 20.995186522262333,
      "grad_norm": Infinity,
      "learning_rate": 0.00017997153451395322,
      "loss": 11.1313,
      "step": 174470
    },
    {
      "epoch": 20.99638989169675,
      "grad_norm": Infinity,
      "learning_rate": 0.00017996925045102912,
      "loss": 11.1774,
      "step": 174480
    },
    {
      "epoch": 20.99759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00017996696627236978,
      "loss": 11.2222,
      "step": 174490
    },
    {
      "epoch": 20.998796630565582,
      "grad_norm": Infinity,
      "learning_rate": 0.00017996468197797847,
      "loss": 11.1186,
      "step": 174500
    },
    {
      "epoch": 21.0,
      "grad_norm": Infinity,
      "learning_rate": 0.00017996239756785855,
      "loss": 11.1338,
      "step": 174510
    },
    {
      "epoch": 21.0,
      "eval_loss": 11.139604568481445,
      "eval_runtime": 120.1616,
      "eval_samples_per_second": 61.476,
      "eval_steps_per_second": 7.69,
      "step": 174510
    },
    {
      "epoch": 21.001203369434418,
      "grad_norm": Infinity,
      "learning_rate": 0.0001799601130420133,
      "loss": 11.0037,
      "step": 174520
    },
    {
      "epoch": 21.00240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.000179957828400446,
      "loss": 11.0998,
      "step": 174530
    },
    {
      "epoch": 21.00361010830325,
      "grad_norm": Infinity,
      "learning_rate": 0.00017995554364316,
      "loss": 11.1011,
      "step": 174540
    },
    {
      "epoch": 21.004813477737667,
      "grad_norm": Infinity,
      "learning_rate": 0.00017995325877015857,
      "loss": 11.1358,
      "step": 174550
    },
    {
      "epoch": 21.00601684717208,
      "grad_norm": Infinity,
      "learning_rate": 0.00017995097378144506,
      "loss": 11.194,
      "step": 174560
    },
    {
      "epoch": 21.0072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00017994868867702275,
      "loss": 11.1631,
      "step": 174570
    },
    {
      "epoch": 21.008423586040916,
      "grad_norm": Infinity,
      "learning_rate": 0.00017994640345689493,
      "loss": 11.2274,
      "step": 174580
    },
    {
      "epoch": 21.00962695547533,
      "grad_norm": Infinity,
      "learning_rate": 0.00017994411812106493,
      "loss": 11.0714,
      "step": 174590
    },
    {
      "epoch": 21.010830324909747,
      "grad_norm": Infinity,
      "learning_rate": 0.00017994183266953605,
      "loss": 11.0898,
      "step": 174600
    },
    {
      "epoch": 21.012033694344165,
      "grad_norm": Infinity,
      "learning_rate": 0.00017993954710231161,
      "loss": 11.0924,
      "step": 174610
    },
    {
      "epoch": 21.01323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00017993726141939492,
      "loss": 11.2257,
      "step": 174620
    },
    {
      "epoch": 21.014440433212997,
      "grad_norm": Infinity,
      "learning_rate": 0.00017993497562078926,
      "loss": 11.1523,
      "step": 174630
    },
    {
      "epoch": 21.015643802647414,
      "grad_norm": Infinity,
      "learning_rate": 0.00017993268970649794,
      "loss": 11.0388,
      "step": 174640
    },
    {
      "epoch": 21.016847172081828,
      "grad_norm": Infinity,
      "learning_rate": 0.00017993040367652433,
      "loss": 11.1122,
      "step": 174650
    },
    {
      "epoch": 21.018050541516246,
      "grad_norm": Infinity,
      "learning_rate": 0.00017992811753087166,
      "loss": 11.162,
      "step": 174660
    },
    {
      "epoch": 21.019253910950663,
      "grad_norm": Infinity,
      "learning_rate": 0.00017992583126954327,
      "loss": 11.088,
      "step": 174670
    },
    {
      "epoch": 21.020457280385077,
      "grad_norm": Infinity,
      "learning_rate": 0.00017992354489254248,
      "loss": 11.2144,
      "step": 174680
    },
    {
      "epoch": 21.021660649819495,
      "grad_norm": Infinity,
      "learning_rate": 0.00017992125839987259,
      "loss": 11.1326,
      "step": 174690
    },
    {
      "epoch": 21.022864019253912,
      "grad_norm": Infinity,
      "learning_rate": 0.00017991897179153688,
      "loss": 11.204,
      "step": 174700
    },
    {
      "epoch": 21.024067388688326,
      "grad_norm": Infinity,
      "learning_rate": 0.0001799166850675387,
      "loss": 11.2448,
      "step": 174710
    },
    {
      "epoch": 21.025270758122744,
      "grad_norm": Infinity,
      "learning_rate": 0.00017991439822788137,
      "loss": 11.0121,
      "step": 174720
    },
    {
      "epoch": 21.02647412755716,
      "grad_norm": Infinity,
      "learning_rate": 0.00017991211127256818,
      "loss": 11.2258,
      "step": 174730
    },
    {
      "epoch": 21.027677496991576,
      "grad_norm": Infinity,
      "learning_rate": 0.00017990982420160238,
      "loss": 11.1731,
      "step": 174740
    },
    {
      "epoch": 21.028880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.00017990753701498737,
      "loss": 11.1214,
      "step": 174750
    },
    {
      "epoch": 21.03008423586041,
      "grad_norm": Infinity,
      "learning_rate": 0.00017990524971272644,
      "loss": 11.101,
      "step": 174760
    },
    {
      "epoch": 21.031287605294825,
      "grad_norm": Infinity,
      "learning_rate": 0.00017990296229482286,
      "loss": 11.0451,
      "step": 174770
    },
    {
      "epoch": 21.032490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00017990067476127995,
      "loss": 11.0666,
      "step": 174780
    },
    {
      "epoch": 21.03369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798983871121011,
      "loss": 11.1861,
      "step": 174790
    },
    {
      "epoch": 21.034897713598074,
      "grad_norm": Infinity,
      "learning_rate": 0.00017989609934728954,
      "loss": 11.2275,
      "step": 174800
    },
    {
      "epoch": 21.03610108303249,
      "grad_norm": Infinity,
      "learning_rate": 0.00017989381146684856,
      "loss": 11.0896,
      "step": 174810
    },
    {
      "epoch": 21.03730445246691,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798915234707815,
      "loss": 11.0762,
      "step": 174820
    },
    {
      "epoch": 21.038507821901323,
      "grad_norm": Infinity,
      "learning_rate": 0.00017988923535909172,
      "loss": 11.1872,
      "step": 174830
    },
    {
      "epoch": 21.03971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017988694713178249,
      "loss": 11.1857,
      "step": 174840
    },
    {
      "epoch": 21.040914560770158,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798846587888571,
      "loss": 11.2217,
      "step": 174850
    },
    {
      "epoch": 21.042117930204572,
      "grad_norm": Infinity,
      "learning_rate": 0.00017988237033031892,
      "loss": 11.2011,
      "step": 174860
    },
    {
      "epoch": 21.04332129963899,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798800817561712,
      "loss": 11.1494,
      "step": 174870
    },
    {
      "epoch": 21.044524669073404,
      "grad_norm": Infinity,
      "learning_rate": 0.00017987779306641728,
      "loss": 11.15,
      "step": 174880
    },
    {
      "epoch": 21.04572803850782,
      "grad_norm": Infinity,
      "learning_rate": 0.00017987550426106047,
      "loss": 11.1517,
      "step": 174890
    },
    {
      "epoch": 21.04693140794224,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798732153401041,
      "loss": 11.0835,
      "step": 174900
    },
    {
      "epoch": 21.048134777376653,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798709263035515,
      "loss": 11.1482,
      "step": 174910
    },
    {
      "epoch": 21.04933814681107,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798686371514059,
      "loss": 11.1244,
      "step": 174920
    },
    {
      "epoch": 21.050541516245488,
      "grad_norm": Infinity,
      "learning_rate": 0.00017986634788367067,
      "loss": 11.2063,
      "step": 174930
    },
    {
      "epoch": 21.051744885679902,
      "grad_norm": Infinity,
      "learning_rate": 0.00017986405850034913,
      "loss": 11.1859,
      "step": 174940
    },
    {
      "epoch": 21.05294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00017986176900144457,
      "loss": 11.1427,
      "step": 174950
    },
    {
      "epoch": 21.054151624548737,
      "grad_norm": Infinity,
      "learning_rate": 0.00017985947938696032,
      "loss": 11.1569,
      "step": 174960
    },
    {
      "epoch": 21.05535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00017985718965689968,
      "loss": 11.0558,
      "step": 174970
    },
    {
      "epoch": 21.05655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00017985489981126596,
      "loss": 11.1052,
      "step": 174980
    },
    {
      "epoch": 21.057761732851986,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798526098500625,
      "loss": 11.155,
      "step": 174990
    },
    {
      "epoch": 21.0589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798503197732926,
      "loss": 11.1384,
      "step": 175000
    },
    {
      "epoch": 21.060168471720818,
      "grad_norm": Infinity,
      "learning_rate": 0.00017984802958095957,
      "loss": 10.9837,
      "step": 175010
    },
    {
      "epoch": 21.061371841155236,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798457392730667,
      "loss": 11.1521,
      "step": 175020
    },
    {
      "epoch": 21.06257521058965,
      "grad_norm": Infinity,
      "learning_rate": 0.00017984344884961736,
      "loss": 11.0969,
      "step": 175030
    },
    {
      "epoch": 21.063778580024067,
      "grad_norm": Infinity,
      "learning_rate": 0.00017984115831061486,
      "loss": 11.0779,
      "step": 175040
    },
    {
      "epoch": 21.064981949458485,
      "grad_norm": Infinity,
      "learning_rate": 0.00017983886765606246,
      "loss": 11.042,
      "step": 175050
    },
    {
      "epoch": 21.0661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798365768859635,
      "loss": 11.2205,
      "step": 175060
    },
    {
      "epoch": 21.067388688327316,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798342860003213,
      "loss": 11.1827,
      "step": 175070
    },
    {
      "epoch": 21.068592057761734,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798319949991392,
      "loss": 11.1997,
      "step": 175080
    },
    {
      "epoch": 21.069795427196148,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798297038824205,
      "loss": 11.072,
      "step": 175090
    },
    {
      "epoch": 21.070998796630565,
      "grad_norm": Infinity,
      "learning_rate": 0.00017982741265016848,
      "loss": 11.169,
      "step": 175100
    },
    {
      "epoch": 21.072202166064983,
      "grad_norm": Infinity,
      "learning_rate": 0.00017982512130238653,
      "loss": 11.1379,
      "step": 175110
    },
    {
      "epoch": 21.073405535499397,
      "grad_norm": Infinity,
      "learning_rate": 0.00017982282983907788,
      "loss": 11.2083,
      "step": 175120
    },
    {
      "epoch": 21.074608904933815,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798205382602459,
      "loss": 11.2237,
      "step": 175130
    },
    {
      "epoch": 21.075812274368232,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798182465658939,
      "loss": 11.1737,
      "step": 175140
    },
    {
      "epoch": 21.077015643802646,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798159547560252,
      "loss": 11.1462,
      "step": 175150
    },
    {
      "epoch": 21.078219013237064,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798136628306431,
      "loss": 11.1439,
      "step": 175160
    },
    {
      "epoch": 21.07942238267148,
      "grad_norm": Infinity,
      "learning_rate": 0.0001798113707897509,
      "loss": 11.066,
      "step": 175170
    },
    {
      "epoch": 21.080625752105895,
      "grad_norm": Infinity,
      "learning_rate": 0.00017980907863335196,
      "loss": 11.1613,
      "step": 175180
    },
    {
      "epoch": 21.081829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.00017980678636144955,
      "loss": 11.0563,
      "step": 175190
    },
    {
      "epoch": 21.08303249097473,
      "grad_norm": Infinity,
      "learning_rate": 0.00017980449397404707,
      "loss": 11.1427,
      "step": 175200
    },
    {
      "epoch": 21.084235860409144,
      "grad_norm": Infinity,
      "learning_rate": 0.00017980220147114775,
      "loss": 11.1727,
      "step": 175210
    },
    {
      "epoch": 21.085439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00017979990885275495,
      "loss": 11.2068,
      "step": 175220
    },
    {
      "epoch": 21.08664259927798,
      "grad_norm": Infinity,
      "learning_rate": 0.00017979761611887197,
      "loss": 11.1031,
      "step": 175230
    },
    {
      "epoch": 21.087845968712394,
      "grad_norm": Infinity,
      "learning_rate": 0.00017979532326950216,
      "loss": 11.1011,
      "step": 175240
    },
    {
      "epoch": 21.08904933814681,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797930303046488,
      "loss": 11.0811,
      "step": 175250
    },
    {
      "epoch": 21.09025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.00017979073722431524,
      "loss": 11.1707,
      "step": 175260
    },
    {
      "epoch": 21.091456077015643,
      "grad_norm": Infinity,
      "learning_rate": 0.00017978844402850475,
      "loss": 11.164,
      "step": 175270
    },
    {
      "epoch": 21.09265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00017978615071722072,
      "loss": 11.0354,
      "step": 175280
    },
    {
      "epoch": 21.093862815884478,
      "grad_norm": Infinity,
      "learning_rate": 0.00017978385729046644,
      "loss": 11.229,
      "step": 175290
    },
    {
      "epoch": 21.095066185318892,
      "grad_norm": Infinity,
      "learning_rate": 0.00017978156374824517,
      "loss": 10.9969,
      "step": 175300
    },
    {
      "epoch": 21.09626955475331,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797792700905603,
      "loss": 11.1586,
      "step": 175310
    },
    {
      "epoch": 21.097472924187727,
      "grad_norm": Infinity,
      "learning_rate": 0.00017977697631741513,
      "loss": 11.0191,
      "step": 175320
    },
    {
      "epoch": 21.09867629362214,
      "grad_norm": Infinity,
      "learning_rate": 0.00017977468242881298,
      "loss": 11.0642,
      "step": 175330
    },
    {
      "epoch": 21.09987966305656,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797723884247572,
      "loss": 11.118,
      "step": 175340
    },
    {
      "epoch": 21.101083032490976,
      "grad_norm": Infinity,
      "learning_rate": 0.00017977009430525105,
      "loss": 11.16,
      "step": 175350
    },
    {
      "epoch": 21.10228640192539,
      "grad_norm": Infinity,
      "learning_rate": 0.00017976780007029788,
      "loss": 11.0596,
      "step": 175360
    },
    {
      "epoch": 21.103489771359808,
      "grad_norm": Infinity,
      "learning_rate": 0.000179765505719901,
      "loss": 11.1998,
      "step": 175370
    },
    {
      "epoch": 21.104693140794225,
      "grad_norm": Infinity,
      "learning_rate": 0.00017976321125406374,
      "loss": 11.1866,
      "step": 175380
    },
    {
      "epoch": 21.10589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.00017976091667278946,
      "loss": 11.1864,
      "step": 175390
    },
    {
      "epoch": 21.107099879663057,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797586219760814,
      "loss": 11.1361,
      "step": 175400
    },
    {
      "epoch": 21.108303249097474,
      "grad_norm": Infinity,
      "learning_rate": 0.00017975632716394296,
      "loss": 11.1497,
      "step": 175410
    },
    {
      "epoch": 21.10950661853189,
      "grad_norm": Infinity,
      "learning_rate": 0.00017975403223637737,
      "loss": 11.0481,
      "step": 175420
    },
    {
      "epoch": 21.110709987966306,
      "grad_norm": Infinity,
      "learning_rate": 0.00017975173719338803,
      "loss": 11.074,
      "step": 175430
    },
    {
      "epoch": 21.111913357400724,
      "grad_norm": Infinity,
      "learning_rate": 0.00017974944203497824,
      "loss": 11.1384,
      "step": 175440
    },
    {
      "epoch": 21.113116726835138,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797471467611513,
      "loss": 11.1639,
      "step": 175450
    },
    {
      "epoch": 21.114320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797448513719106,
      "loss": 11.0698,
      "step": 175460
    },
    {
      "epoch": 21.115523465703973,
      "grad_norm": Infinity,
      "learning_rate": 0.00017974255586725935,
      "loss": 11.1396,
      "step": 175470
    },
    {
      "epoch": 21.116726835138387,
      "grad_norm": Infinity,
      "learning_rate": 0.00017974026024720097,
      "loss": 11.1011,
      "step": 175480
    },
    {
      "epoch": 21.117930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.00017973796451173875,
      "loss": 11.0587,
      "step": 175490
    },
    {
      "epoch": 21.119133574007222,
      "grad_norm": Infinity,
      "learning_rate": 0.00017973566866087598,
      "loss": 11.1652,
      "step": 175500
    },
    {
      "epoch": 21.120336943441636,
      "grad_norm": Infinity,
      "learning_rate": 0.00017973337269461604,
      "loss": 11.2248,
      "step": 175510
    },
    {
      "epoch": 21.121540312876053,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797310766129622,
      "loss": 11.0861,
      "step": 175520
    },
    {
      "epoch": 21.12274368231047,
      "grad_norm": Infinity,
      "learning_rate": 0.00017972878041591783,
      "loss": 11.1933,
      "step": 175530
    },
    {
      "epoch": 21.123947051744885,
      "grad_norm": Infinity,
      "learning_rate": 0.00017972648410348623,
      "loss": 11.0916,
      "step": 175540
    },
    {
      "epoch": 21.125150421179303,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797241876756707,
      "loss": 11.1908,
      "step": 175550
    },
    {
      "epoch": 21.126353790613717,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797218911324746,
      "loss": 11.1684,
      "step": 175560
    },
    {
      "epoch": 21.127557160048134,
      "grad_norm": Infinity,
      "learning_rate": 0.00017971959447390124,
      "loss": 11.2317,
      "step": 175570
    },
    {
      "epoch": 21.128760529482552,
      "grad_norm": Infinity,
      "learning_rate": 0.00017971729769995396,
      "loss": 11.1633,
      "step": 175580
    },
    {
      "epoch": 21.129963898916966,
      "grad_norm": Infinity,
      "learning_rate": 0.00017971500081063603,
      "loss": 11.0136,
      "step": 175590
    },
    {
      "epoch": 21.131167268351383,
      "grad_norm": Infinity,
      "learning_rate": 0.00017971270380595086,
      "loss": 11.105,
      "step": 175600
    },
    {
      "epoch": 21.1323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00017971040668590172,
      "loss": 11.1389,
      "step": 175610
    },
    {
      "epoch": 21.133574007220215,
      "grad_norm": Infinity,
      "learning_rate": 0.0001797081094504919,
      "loss": 11.1323,
      "step": 175620
    },
    {
      "epoch": 21.134777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.00017970581209972482,
      "loss": 11.1936,
      "step": 175630
    },
    {
      "epoch": 21.13598074608905,
      "grad_norm": Infinity,
      "learning_rate": 0.00017970351463360371,
      "loss": 11.1362,
      "step": 175640
    },
    {
      "epoch": 21.137184115523464,
      "grad_norm": Infinity,
      "learning_rate": 0.00017970121705213193,
      "loss": 11.1278,
      "step": 175650
    },
    {
      "epoch": 21.13838748495788,
      "grad_norm": Infinity,
      "learning_rate": 0.00017969891935531285,
      "loss": 11.1496,
      "step": 175660
    },
    {
      "epoch": 21.1395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017969662154314973,
      "loss": 11.1811,
      "step": 175670
    },
    {
      "epoch": 21.140794223826713,
      "grad_norm": Infinity,
      "learning_rate": 0.00017969432361564594,
      "loss": 11.0146,
      "step": 175680
    },
    {
      "epoch": 21.14199759326113,
      "grad_norm": Infinity,
      "learning_rate": 0.00017969202557280477,
      "loss": 11.2045,
      "step": 175690
    },
    {
      "epoch": 21.14320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.00017968972741462958,
      "loss": 11.1125,
      "step": 175700
    },
    {
      "epoch": 21.144404332129962,
      "grad_norm": Infinity,
      "learning_rate": 0.00017968742914112367,
      "loss": 11.2293,
      "step": 175710
    },
    {
      "epoch": 21.14560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00017968513075229038,
      "loss": 11.1283,
      "step": 175720
    },
    {
      "epoch": 21.146811070998798,
      "grad_norm": Infinity,
      "learning_rate": 0.00017968283224813303,
      "loss": 11.0612,
      "step": 175730
    },
    {
      "epoch": 21.14801444043321,
      "grad_norm": Infinity,
      "learning_rate": 0.00017968053362865492,
      "loss": 11.074,
      "step": 175740
    },
    {
      "epoch": 21.14921780986763,
      "grad_norm": Infinity,
      "learning_rate": 0.00017967823489385945,
      "loss": 11.164,
      "step": 175750
    },
    {
      "epoch": 21.150421179302047,
      "grad_norm": Infinity,
      "learning_rate": 0.00017967593604374988,
      "loss": 11.1004,
      "step": 175760
    },
    {
      "epoch": 21.15162454873646,
      "grad_norm": Infinity,
      "learning_rate": 0.00017967363707832958,
      "loss": 11.1457,
      "step": 175770
    },
    {
      "epoch": 21.15282791817088,
      "grad_norm": Infinity,
      "learning_rate": 0.00017967133799760184,
      "loss": 11.3109,
      "step": 175780
    },
    {
      "epoch": 21.154031287605296,
      "grad_norm": Infinity,
      "learning_rate": 0.00017966903880157,
      "loss": 11.103,
      "step": 175790
    },
    {
      "epoch": 21.15523465703971,
      "grad_norm": Infinity,
      "learning_rate": 0.00017966673949023737,
      "loss": 11.1572,
      "step": 175800
    },
    {
      "epoch": 21.156438026474127,
      "grad_norm": Infinity,
      "learning_rate": 0.00017966444006360733,
      "loss": 11.1813,
      "step": 175810
    },
    {
      "epoch": 21.157641395908545,
      "grad_norm": Infinity,
      "learning_rate": 0.00017966214052168319,
      "loss": 11.1746,
      "step": 175820
    },
    {
      "epoch": 21.15884476534296,
      "grad_norm": Infinity,
      "learning_rate": 0.00017965984086446826,
      "loss": 11.2223,
      "step": 175830
    },
    {
      "epoch": 21.160048134777377,
      "grad_norm": Infinity,
      "learning_rate": 0.00017965754109196586,
      "loss": 11.1993,
      "step": 175840
    },
    {
      "epoch": 21.161251504211794,
      "grad_norm": Infinity,
      "learning_rate": 0.00017965524120417932,
      "loss": 11.1044,
      "step": 175850
    },
    {
      "epoch": 21.162454873646208,
      "grad_norm": Infinity,
      "learning_rate": 0.000179652941201112,
      "loss": 11.2017,
      "step": 175860
    },
    {
      "epoch": 21.163658243080626,
      "grad_norm": Infinity,
      "learning_rate": 0.0001796506410827672,
      "loss": 11.2222,
      "step": 175870
    },
    {
      "epoch": 21.164861612515043,
      "grad_norm": Infinity,
      "learning_rate": 0.00017964834084914822,
      "loss": 11.198,
      "step": 175880
    },
    {
      "epoch": 21.166064981949457,
      "grad_norm": Infinity,
      "learning_rate": 0.0001796460405002585,
      "loss": 11.0992,
      "step": 175890
    },
    {
      "epoch": 21.167268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.00017964374003610125,
      "loss": 11.1333,
      "step": 175900
    },
    {
      "epoch": 21.168471720818292,
      "grad_norm": Infinity,
      "learning_rate": 0.00017964143945667984,
      "loss": 11.1504,
      "step": 175910
    },
    {
      "epoch": 21.169675090252706,
      "grad_norm": Infinity,
      "learning_rate": 0.00017963913876199763,
      "loss": 11.19,
      "step": 175920
    },
    {
      "epoch": 21.170878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.0001796368379520579,
      "loss": 11.1756,
      "step": 175930
    },
    {
      "epoch": 21.17208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00017963453702686403,
      "loss": 11.1727,
      "step": 175940
    },
    {
      "epoch": 21.173285198555956,
      "grad_norm": Infinity,
      "learning_rate": 0.0001796322359864193,
      "loss": 11.113,
      "step": 175950
    },
    {
      "epoch": 21.174488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.00017962993483072704,
      "loss": 11.1965,
      "step": 175960
    },
    {
      "epoch": 21.17569193742479,
      "grad_norm": Infinity,
      "learning_rate": 0.00017962763355979067,
      "loss": 11.0987,
      "step": 175970
    },
    {
      "epoch": 21.176895306859205,
      "grad_norm": Infinity,
      "learning_rate": 0.00017962533217361342,
      "loss": 11.1025,
      "step": 175980
    },
    {
      "epoch": 21.178098676293622,
      "grad_norm": Infinity,
      "learning_rate": 0.00017962303067219865,
      "loss": 11.1558,
      "step": 175990
    },
    {
      "epoch": 21.17930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.0001796207290555497,
      "loss": 11.0474,
      "step": 176000
    },
    {
      "epoch": 21.180505415162454,
      "grad_norm": Infinity,
      "learning_rate": 0.0001796184273236699,
      "loss": 11.1678,
      "step": 176010
    },
    {
      "epoch": 21.18170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00017961612547656258,
      "loss": 11.0584,
      "step": 176020
    },
    {
      "epoch": 21.18291215403129,
      "grad_norm": Infinity,
      "learning_rate": 0.00017961382351423105,
      "loss": 11.1787,
      "step": 176030
    },
    {
      "epoch": 21.184115523465703,
      "grad_norm": Infinity,
      "learning_rate": 0.0001796115214366787,
      "loss": 11.0706,
      "step": 176040
    },
    {
      "epoch": 21.18531889290012,
      "grad_norm": Infinity,
      "learning_rate": 0.00017960921924390882,
      "loss": 11.0195,
      "step": 176050
    },
    {
      "epoch": 21.186522262334538,
      "grad_norm": Infinity,
      "learning_rate": 0.00017960691693592472,
      "loss": 11.0011,
      "step": 176060
    },
    {
      "epoch": 21.187725631768952,
      "grad_norm": Infinity,
      "learning_rate": 0.00017960461451272977,
      "loss": 11.1685,
      "step": 176070
    },
    {
      "epoch": 21.18892900120337,
      "grad_norm": Infinity,
      "learning_rate": 0.00017960231197432726,
      "loss": 11.1919,
      "step": 176080
    },
    {
      "epoch": 21.190132370637787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017960000932072058,
      "loss": 11.1221,
      "step": 176090
    },
    {
      "epoch": 21.1913357400722,
      "grad_norm": Infinity,
      "learning_rate": 0.00017959770655191304,
      "loss": 11.1551,
      "step": 176100
    },
    {
      "epoch": 21.19253910950662,
      "grad_norm": Infinity,
      "learning_rate": 0.00017959540366790794,
      "loss": 11.1241,
      "step": 176110
    },
    {
      "epoch": 21.193742478941036,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795931006687087,
      "loss": 11.1233,
      "step": 176120
    },
    {
      "epoch": 21.19494584837545,
      "grad_norm": Infinity,
      "learning_rate": 0.00017959079755431852,
      "loss": 11.1798,
      "step": 176130
    },
    {
      "epoch": 21.196149217809868,
      "grad_norm": Infinity,
      "learning_rate": 0.00017958849432474084,
      "loss": 11.1839,
      "step": 176140
    },
    {
      "epoch": 21.197352587244286,
      "grad_norm": Infinity,
      "learning_rate": 0.00017958619097997893,
      "loss": 11.225,
      "step": 176150
    },
    {
      "epoch": 21.1985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017958388752003617,
      "loss": 11.1674,
      "step": 176160
    },
    {
      "epoch": 21.199759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795815839449159,
      "loss": 11.2309,
      "step": 176170
    },
    {
      "epoch": 21.200962695547535,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795792802546214,
      "loss": 11.1631,
      "step": 176180
    },
    {
      "epoch": 21.20216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.00017957697644915605,
      "loss": 11.1528,
      "step": 176190
    },
    {
      "epoch": 21.203369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.00017957467252852319,
      "loss": 11.1066,
      "step": 176200
    },
    {
      "epoch": 21.204572803850784,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795723684927261,
      "loss": 11.1619,
      "step": 176210
    },
    {
      "epoch": 21.205776173285198,
      "grad_norm": Infinity,
      "learning_rate": 0.00017957006434176815,
      "loss": 11.0592,
      "step": 176220
    },
    {
      "epoch": 21.206979542719615,
      "grad_norm": Infinity,
      "learning_rate": 0.00017956776007565267,
      "loss": 11.1485,
      "step": 176230
    },
    {
      "epoch": 21.20818291215403,
      "grad_norm": Infinity,
      "learning_rate": 0.000179565455694383,
      "loss": 11.2799,
      "step": 176240
    },
    {
      "epoch": 21.209386281588447,
      "grad_norm": Infinity,
      "learning_rate": 0.00017956315119796248,
      "loss": 11.0851,
      "step": 176250
    },
    {
      "epoch": 21.210589651022865,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795608465863944,
      "loss": 11.171,
      "step": 176260
    },
    {
      "epoch": 21.21179302045728,
      "grad_norm": Infinity,
      "learning_rate": 0.00017955854185968217,
      "loss": 11.1642,
      "step": 176270
    },
    {
      "epoch": 21.212996389891696,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795562370178291,
      "loss": 11.2228,
      "step": 176280
    },
    {
      "epoch": 21.214199759326114,
      "grad_norm": Infinity,
      "learning_rate": 0.00017955393206083847,
      "loss": 11.1731,
      "step": 176290
    },
    {
      "epoch": 21.215403128760528,
      "grad_norm": Infinity,
      "learning_rate": 0.00017955162698871368,
      "loss": 11.1642,
      "step": 176300
    },
    {
      "epoch": 21.216606498194945,
      "grad_norm": Infinity,
      "learning_rate": 0.00017954932180145805,
      "loss": 11.1389,
      "step": 176310
    },
    {
      "epoch": 21.217809867629363,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795470164990749,
      "loss": 11.2054,
      "step": 176320
    },
    {
      "epoch": 21.219013237063777,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795447110815676,
      "loss": 11.1474,
      "step": 176330
    },
    {
      "epoch": 21.220216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.00017954240554893947,
      "loss": 11.0303,
      "step": 176340
    },
    {
      "epoch": 21.221419975932612,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795400999011938,
      "loss": 11.2145,
      "step": 176350
    },
    {
      "epoch": 21.222623345367026,
      "grad_norm": Infinity,
      "learning_rate": 0.000179537794138334,
      "loss": 11.1101,
      "step": 176360
    },
    {
      "epoch": 21.223826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.00017953548826036336,
      "loss": 11.1943,
      "step": 176370
    },
    {
      "epoch": 21.22503008423586,
      "grad_norm": Infinity,
      "learning_rate": 0.00017953318226728523,
      "loss": 11.1527,
      "step": 176380
    },
    {
      "epoch": 21.226233453670275,
      "grad_norm": Infinity,
      "learning_rate": 0.00017953087615910294,
      "loss": 11.2004,
      "step": 176390
    },
    {
      "epoch": 21.227436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00017952856993581984,
      "loss": 11.1601,
      "step": 176400
    },
    {
      "epoch": 21.22864019253911,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795262635974393,
      "loss": 11.0547,
      "step": 176410
    },
    {
      "epoch": 21.229843561973524,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795239571439646,
      "loss": 11.0258,
      "step": 176420
    },
    {
      "epoch": 21.231046931407942,
      "grad_norm": Infinity,
      "learning_rate": 0.00017952165057539906,
      "loss": 11.1601,
      "step": 176430
    },
    {
      "epoch": 21.23225030084236,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795193438917461,
      "loss": 11.1631,
      "step": 176440
    },
    {
      "epoch": 21.233453670276774,
      "grad_norm": Infinity,
      "learning_rate": 0.00017951703709300903,
      "loss": 11.0708,
      "step": 176450
    },
    {
      "epoch": 21.23465703971119,
      "grad_norm": Infinity,
      "learning_rate": 0.00017951473017919115,
      "loss": 11.194,
      "step": 176460
    },
    {
      "epoch": 21.23586040914561,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795124231502958,
      "loss": 11.1157,
      "step": 176470
    },
    {
      "epoch": 21.237063778580023,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795101160063264,
      "loss": 11.2554,
      "step": 176480
    },
    {
      "epoch": 21.23826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795078087472862,
      "loss": 11.1329,
      "step": 176490
    },
    {
      "epoch": 21.239470517448858,
      "grad_norm": Infinity,
      "learning_rate": 0.0001795055013731786,
      "loss": 11.1903,
      "step": 176500
    },
    {
      "epoch": 21.240673886883272,
      "grad_norm": Infinity,
      "learning_rate": 0.00017950319388400688,
      "loss": 11.1696,
      "step": 176510
    },
    {
      "epoch": 21.24187725631769,
      "grad_norm": Infinity,
      "learning_rate": 0.00017950088627977442,
      "loss": 11.0923,
      "step": 176520
    },
    {
      "epoch": 21.243080625752107,
      "grad_norm": Infinity,
      "learning_rate": 0.00017949857856048454,
      "loss": 11.1562,
      "step": 176530
    },
    {
      "epoch": 21.24428399518652,
      "grad_norm": Infinity,
      "learning_rate": 0.00017949627072614058,
      "loss": 11.1852,
      "step": 176540
    },
    {
      "epoch": 21.24548736462094,
      "grad_norm": Infinity,
      "learning_rate": 0.00017949396277674594,
      "loss": 11.2694,
      "step": 176550
    },
    {
      "epoch": 21.246690734055356,
      "grad_norm": Infinity,
      "learning_rate": 0.00017949165471230384,
      "loss": 11.1554,
      "step": 176560
    },
    {
      "epoch": 21.24789410348977,
      "grad_norm": Infinity,
      "learning_rate": 0.00017948934653281773,
      "loss": 11.081,
      "step": 176570
    },
    {
      "epoch": 21.249097472924188,
      "grad_norm": Infinity,
      "learning_rate": 0.00017948703823829088,
      "loss": 11.2633,
      "step": 176580
    },
    {
      "epoch": 21.250300842358605,
      "grad_norm": Infinity,
      "learning_rate": 0.00017948472982872671,
      "loss": 11.0338,
      "step": 176590
    },
    {
      "epoch": 21.25150421179302,
      "grad_norm": Infinity,
      "learning_rate": 0.00017948242130412848,
      "loss": 11.1355,
      "step": 176600
    },
    {
      "epoch": 21.252707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00017948011266449957,
      "loss": 11.1979,
      "step": 176610
    },
    {
      "epoch": 21.253910950661854,
      "grad_norm": Infinity,
      "learning_rate": 0.0001794778039098433,
      "loss": 11.0544,
      "step": 176620
    },
    {
      "epoch": 21.25511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00017947549504016306,
      "loss": 11.1873,
      "step": 176630
    },
    {
      "epoch": 21.256317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00017947318605546211,
      "loss": 11.317,
      "step": 176640
    },
    {
      "epoch": 21.257521058965104,
      "grad_norm": Infinity,
      "learning_rate": 0.00017947087695574388,
      "loss": 11.2159,
      "step": 176650
    },
    {
      "epoch": 21.258724428399518,
      "grad_norm": Infinity,
      "learning_rate": 0.00017946856774101163,
      "loss": 11.1819,
      "step": 176660
    },
    {
      "epoch": 21.259927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.0001794662584112688,
      "loss": 11.0854,
      "step": 176670
    },
    {
      "epoch": 21.261131167268353,
      "grad_norm": Infinity,
      "learning_rate": 0.00017946394896651863,
      "loss": 11.2274,
      "step": 176680
    },
    {
      "epoch": 21.262334536702767,
      "grad_norm": Infinity,
      "learning_rate": 0.00017946163940676452,
      "loss": 11.2985,
      "step": 176690
    },
    {
      "epoch": 21.263537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.0001794593297320098,
      "loss": 11.1206,
      "step": 176700
    },
    {
      "epoch": 21.264741275571602,
      "grad_norm": Infinity,
      "learning_rate": 0.00017945701994225778,
      "loss": 11.2249,
      "step": 176710
    },
    {
      "epoch": 21.265944645006016,
      "grad_norm": Infinity,
      "learning_rate": 0.0001794547100375119,
      "loss": 11.0384,
      "step": 176720
    },
    {
      "epoch": 21.267148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.0001794524000177754,
      "loss": 11.1015,
      "step": 176730
    },
    {
      "epoch": 21.26835138387485,
      "grad_norm": Infinity,
      "learning_rate": 0.00017945008988305166,
      "loss": 11.1558,
      "step": 176740
    },
    {
      "epoch": 21.269554753309265,
      "grad_norm": Infinity,
      "learning_rate": 0.00017944777963334403,
      "loss": 11.1869,
      "step": 176750
    },
    {
      "epoch": 21.270758122743683,
      "grad_norm": Infinity,
      "learning_rate": 0.00017944546926865584,
      "loss": 11.0758,
      "step": 176760
    },
    {
      "epoch": 21.2719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00017944315878899042,
      "loss": 11.134,
      "step": 176770
    },
    {
      "epoch": 21.273164861612514,
      "grad_norm": Infinity,
      "learning_rate": 0.00017944084819435118,
      "loss": 11.1136,
      "step": 176780
    },
    {
      "epoch": 21.27436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.0001794385374847414,
      "loss": 11.2003,
      "step": 176790
    },
    {
      "epoch": 21.27557160048135,
      "grad_norm": Infinity,
      "learning_rate": 0.00017943622666016445,
      "loss": 11.0783,
      "step": 176800
    },
    {
      "epoch": 21.276774969915763,
      "grad_norm": Infinity,
      "learning_rate": 0.00017943391572062365,
      "loss": 11.1038,
      "step": 176810
    },
    {
      "epoch": 21.27797833935018,
      "grad_norm": Infinity,
      "learning_rate": 0.00017943160466612237,
      "loss": 11.1468,
      "step": 176820
    },
    {
      "epoch": 21.2791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00017942929349666398,
      "loss": 10.9171,
      "step": 176830
    },
    {
      "epoch": 21.280385078219012,
      "grad_norm": Infinity,
      "learning_rate": 0.00017942698221225174,
      "loss": 11.0567,
      "step": 176840
    },
    {
      "epoch": 21.28158844765343,
      "grad_norm": Infinity,
      "learning_rate": 0.00017942467081288906,
      "loss": 11.2039,
      "step": 176850
    },
    {
      "epoch": 21.282791817087848,
      "grad_norm": Infinity,
      "learning_rate": 0.0001794223592985793,
      "loss": 11.2548,
      "step": 176860
    },
    {
      "epoch": 21.28399518652226,
      "grad_norm": Infinity,
      "learning_rate": 0.00017942004766932575,
      "loss": 11.097,
      "step": 176870
    },
    {
      "epoch": 21.28519855595668,
      "grad_norm": Infinity,
      "learning_rate": 0.0001794177359251318,
      "loss": 11.2278,
      "step": 176880
    },
    {
      "epoch": 21.286401925391097,
      "grad_norm": Infinity,
      "learning_rate": 0.00017941542406600078,
      "loss": 11.1818,
      "step": 176890
    },
    {
      "epoch": 21.28760529482551,
      "grad_norm": Infinity,
      "learning_rate": 0.000179413112091936,
      "loss": 11.2058,
      "step": 176900
    },
    {
      "epoch": 21.28880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.0001794108000029409,
      "loss": 11.1192,
      "step": 176910
    },
    {
      "epoch": 21.290012033694346,
      "grad_norm": Infinity,
      "learning_rate": 0.00017940848779901872,
      "loss": 11.1827,
      "step": 176920
    },
    {
      "epoch": 21.29121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00017940617548017286,
      "loss": 11.1723,
      "step": 176930
    },
    {
      "epoch": 21.292418772563177,
      "grad_norm": Infinity,
      "learning_rate": 0.00017940386304640666,
      "loss": 11.1733,
      "step": 176940
    },
    {
      "epoch": 21.29362214199759,
      "grad_norm": Infinity,
      "learning_rate": 0.00017940155049772348,
      "loss": 11.1362,
      "step": 176950
    },
    {
      "epoch": 21.29482551143201,
      "grad_norm": Infinity,
      "learning_rate": 0.00017939923783412661,
      "loss": 11.2565,
      "step": 176960
    },
    {
      "epoch": 21.296028880866427,
      "grad_norm": Infinity,
      "learning_rate": 0.00017939692505561952,
      "loss": 11.0689,
      "step": 176970
    },
    {
      "epoch": 21.29723225030084,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793946121622054,
      "loss": 11.171,
      "step": 176980
    },
    {
      "epoch": 21.29843561973526,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793922991538877,
      "loss": 11.1513,
      "step": 176990
    },
    {
      "epoch": 21.299638989169676,
      "grad_norm": Infinity,
      "learning_rate": 0.00017938998603066975,
      "loss": 11.0984,
      "step": 177000
    },
    {
      "epoch": 21.30084235860409,
      "grad_norm": Infinity,
      "learning_rate": 0.00017938767279255489,
      "loss": 11.1211,
      "step": 177010
    },
    {
      "epoch": 21.302045728038507,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793853594395465,
      "loss": 11.1399,
      "step": 177020
    },
    {
      "epoch": 21.303249097472925,
      "grad_norm": Infinity,
      "learning_rate": 0.00017938304597164782,
      "loss": 11.0951,
      "step": 177030
    },
    {
      "epoch": 21.30445246690734,
      "grad_norm": Infinity,
      "learning_rate": 0.00017938073238886232,
      "loss": 11.1552,
      "step": 177040
    },
    {
      "epoch": 21.305655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00017937841869119326,
      "loss": 11.0345,
      "step": 177050
    },
    {
      "epoch": 21.306859205776174,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793761048786441,
      "loss": 11.1799,
      "step": 177060
    },
    {
      "epoch": 21.308062575210588,
      "grad_norm": Infinity,
      "learning_rate": 0.00017937379095121807,
      "loss": 11.1944,
      "step": 177070
    },
    {
      "epoch": 21.309265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00017937147690891858,
      "loss": 11.043,
      "step": 177080
    },
    {
      "epoch": 21.310469314079423,
      "grad_norm": Infinity,
      "learning_rate": 0.00017936916275174895,
      "loss": 11.1842,
      "step": 177090
    },
    {
      "epoch": 21.311672683513837,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793668484797126,
      "loss": 11.134,
      "step": 177100
    },
    {
      "epoch": 21.312876052948255,
      "grad_norm": Infinity,
      "learning_rate": 0.00017936453409281278,
      "loss": 11.159,
      "step": 177110
    },
    {
      "epoch": 21.314079422382672,
      "grad_norm": Infinity,
      "learning_rate": 0.00017936221959105288,
      "loss": 11.1535,
      "step": 177120
    },
    {
      "epoch": 21.315282791817086,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793599049744363,
      "loss": 11.2275,
      "step": 177130
    },
    {
      "epoch": 21.316486161251504,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793575902429663,
      "loss": 11.1148,
      "step": 177140
    },
    {
      "epoch": 21.31768953068592,
      "grad_norm": Infinity,
      "learning_rate": 0.00017935527539664626,
      "loss": 11.1367,
      "step": 177150
    },
    {
      "epoch": 21.318892900120336,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793529604354796,
      "loss": 11.1976,
      "step": 177160
    },
    {
      "epoch": 21.320096269554753,
      "grad_norm": Infinity,
      "learning_rate": 0.00017935064535946955,
      "loss": 11.1856,
      "step": 177170
    },
    {
      "epoch": 21.32129963898917,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793483301686196,
      "loss": 11.2488,
      "step": 177180
    },
    {
      "epoch": 21.322503008423585,
      "grad_norm": Infinity,
      "learning_rate": 0.00017934601486293295,
      "loss": 11.1916,
      "step": 177190
    },
    {
      "epoch": 21.323706377858002,
      "grad_norm": Infinity,
      "learning_rate": 0.00017934369944241308,
      "loss": 11.0954,
      "step": 177200
    },
    {
      "epoch": 21.32490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00017934138390706327,
      "loss": 11.018,
      "step": 177210
    },
    {
      "epoch": 21.326113116726834,
      "grad_norm": Infinity,
      "learning_rate": 0.00017933906825688686,
      "loss": 11.0858,
      "step": 177220
    },
    {
      "epoch": 21.32731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793367524918873,
      "loss": 11.1276,
      "step": 177230
    },
    {
      "epoch": 21.32851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793344366120678,
      "loss": 11.0127,
      "step": 177240
    },
    {
      "epoch": 21.329723225030083,
      "grad_norm": Infinity,
      "learning_rate": 0.00017933212061743182,
      "loss": 11.1165,
      "step": 177250
    },
    {
      "epoch": 21.3309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.00017932980450798266,
      "loss": 11.1582,
      "step": 177260
    },
    {
      "epoch": 21.332129963898918,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793274882837237,
      "loss": 11.1298,
      "step": 177270
    },
    {
      "epoch": 21.333333333333332,
      "grad_norm": Infinity,
      "learning_rate": 0.00017932517194465828,
      "loss": 11.2097,
      "step": 177280
    },
    {
      "epoch": 21.33453670276775,
      "grad_norm": Infinity,
      "learning_rate": 0.00017932285549078974,
      "loss": 11.2558,
      "step": 177290
    },
    {
      "epoch": 21.335740072202167,
      "grad_norm": Infinity,
      "learning_rate": 0.00017932053892212143,
      "loss": 11.2204,
      "step": 177300
    },
    {
      "epoch": 21.33694344163658,
      "grad_norm": Infinity,
      "learning_rate": 0.00017931822223865676,
      "loss": 11.1964,
      "step": 177310
    },
    {
      "epoch": 21.338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.000179315905440399,
      "loss": 11.0676,
      "step": 177320
    },
    {
      "epoch": 21.339350180505416,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793135885273516,
      "loss": 11.0256,
      "step": 177330
    },
    {
      "epoch": 21.34055354993983,
      "grad_norm": Infinity,
      "learning_rate": 0.00017931127149951778,
      "loss": 11.1466,
      "step": 177340
    },
    {
      "epoch": 21.341756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.000179308954356901,
      "loss": 11.0819,
      "step": 177350
    },
    {
      "epoch": 21.342960288808666,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793066370995046,
      "loss": 11.1377,
      "step": 177360
    },
    {
      "epoch": 21.34416365824308,
      "grad_norm": Infinity,
      "learning_rate": 0.0001793043197273319,
      "loss": 11.0592,
      "step": 177370
    },
    {
      "epoch": 21.345367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017930200224038626,
      "loss": 11.0842,
      "step": 177380
    },
    {
      "epoch": 21.346570397111915,
      "grad_norm": Infinity,
      "learning_rate": 0.00017929968463867107,
      "loss": 11.1002,
      "step": 177390
    },
    {
      "epoch": 21.34777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792973669221896,
      "loss": 11.2114,
      "step": 177400
    },
    {
      "epoch": 21.348977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.00017929504909094534,
      "loss": 11.138,
      "step": 177410
    },
    {
      "epoch": 21.350180505415164,
      "grad_norm": Infinity,
      "learning_rate": 0.00017929273114494152,
      "loss": 11.0799,
      "step": 177420
    },
    {
      "epoch": 21.351383874849578,
      "grad_norm": Infinity,
      "learning_rate": 0.00017929041308418155,
      "loss": 11.2754,
      "step": 177430
    },
    {
      "epoch": 21.352587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00017928809490866878,
      "loss": 11.1381,
      "step": 177440
    },
    {
      "epoch": 21.353790613718413,
      "grad_norm": Infinity,
      "learning_rate": 0.00017928577661840658,
      "loss": 11.1783,
      "step": 177450
    },
    {
      "epoch": 21.354993983152827,
      "grad_norm": Infinity,
      "learning_rate": 0.00017928345821339824,
      "loss": 11.158,
      "step": 177460
    },
    {
      "epoch": 21.356197352587245,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792811396936472,
      "loss": 11.172,
      "step": 177470
    },
    {
      "epoch": 21.357400722021662,
      "grad_norm": Infinity,
      "learning_rate": 0.00017927882105915675,
      "loss": 11.2189,
      "step": 177480
    },
    {
      "epoch": 21.358604091456076,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792765023099303,
      "loss": 11.0759,
      "step": 177490
    },
    {
      "epoch": 21.359807460890494,
      "grad_norm": Infinity,
      "learning_rate": 0.00017927418344597116,
      "loss": 11.0952,
      "step": 177500
    },
    {
      "epoch": 21.36101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792718644672827,
      "loss": 11.2672,
      "step": 177510
    },
    {
      "epoch": 21.362214199759325,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792695453738683,
      "loss": 11.1047,
      "step": 177520
    },
    {
      "epoch": 21.363417569193743,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792672261657313,
      "loss": 11.1513,
      "step": 177530
    },
    {
      "epoch": 21.36462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.000179264906842875,
      "loss": 11.1857,
      "step": 177540
    },
    {
      "epoch": 21.365824308062574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017926258740530288,
      "loss": 11.0515,
      "step": 177550
    },
    {
      "epoch": 21.367027677496992,
      "grad_norm": Infinity,
      "learning_rate": 0.00017926026785301818,
      "loss": 11.097,
      "step": 177560
    },
    {
      "epoch": 21.36823104693141,
      "grad_norm": Infinity,
      "learning_rate": 0.00017925794818602432,
      "loss": 11.0533,
      "step": 177570
    },
    {
      "epoch": 21.369434416365824,
      "grad_norm": Infinity,
      "learning_rate": 0.00017925562840432466,
      "loss": 11.1579,
      "step": 177580
    },
    {
      "epoch": 21.37063778580024,
      "grad_norm": Infinity,
      "learning_rate": 0.00017925330850792254,
      "loss": 11.0705,
      "step": 177590
    },
    {
      "epoch": 21.37184115523466,
      "grad_norm": Infinity,
      "learning_rate": 0.00017925098849682126,
      "loss": 11.2422,
      "step": 177600
    },
    {
      "epoch": 21.373044524669073,
      "grad_norm": Infinity,
      "learning_rate": 0.00017924866837102429,
      "loss": 11.1921,
      "step": 177610
    },
    {
      "epoch": 21.37424789410349,
      "grad_norm": Infinity,
      "learning_rate": 0.00017924634813053488,
      "loss": 11.2032,
      "step": 177620
    },
    {
      "epoch": 21.375451263537904,
      "grad_norm": Infinity,
      "learning_rate": 0.00017924402777535646,
      "loss": 11.1397,
      "step": 177630
    },
    {
      "epoch": 21.376654632972322,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792417073054924,
      "loss": 11.1978,
      "step": 177640
    },
    {
      "epoch": 21.37785800240674,
      "grad_norm": Infinity,
      "learning_rate": 0.000179239386720946,
      "loss": 11.1372,
      "step": 177650
    },
    {
      "epoch": 21.379061371841154,
      "grad_norm": Infinity,
      "learning_rate": 0.00017923706602172065,
      "loss": 11.2596,
      "step": 177660
    },
    {
      "epoch": 21.38026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792347452078197,
      "loss": 11.1096,
      "step": 177670
    },
    {
      "epoch": 21.38146811070999,
      "grad_norm": Infinity,
      "learning_rate": 0.00017923242427924648,
      "loss": 11.0652,
      "step": 177680
    },
    {
      "epoch": 21.382671480144403,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792301032360044,
      "loss": 11.2321,
      "step": 177690
    },
    {
      "epoch": 21.38387484957882,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792277820780968,
      "loss": 11.1908,
      "step": 177700
    },
    {
      "epoch": 21.385078219013238,
      "grad_norm": Infinity,
      "learning_rate": 0.00017922546080552707,
      "loss": 11.1872,
      "step": 177710
    },
    {
      "epoch": 21.386281588447652,
      "grad_norm": Infinity,
      "learning_rate": 0.00017922313941829852,
      "loss": 11.2141,
      "step": 177720
    },
    {
      "epoch": 21.38748495788207,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792208179164145,
      "loss": 11.1941,
      "step": 177730
    },
    {
      "epoch": 21.388688327316487,
      "grad_norm": Infinity,
      "learning_rate": 0.00017921849629987842,
      "loss": 11.2062,
      "step": 177740
    },
    {
      "epoch": 21.3898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00017921617456869365,
      "loss": 11.1188,
      "step": 177750
    },
    {
      "epoch": 21.39109506618532,
      "grad_norm": Infinity,
      "learning_rate": 0.00017921385272286346,
      "loss": 10.9906,
      "step": 177760
    },
    {
      "epoch": 21.392298435619736,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792115307623913,
      "loss": 11.2406,
      "step": 177770
    },
    {
      "epoch": 21.39350180505415,
      "grad_norm": Infinity,
      "learning_rate": 0.00017920920868728048,
      "loss": 11.1204,
      "step": 177780
    },
    {
      "epoch": 21.394705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017920688649753438,
      "loss": 11.1523,
      "step": 177790
    },
    {
      "epoch": 21.395908543922985,
      "grad_norm": Infinity,
      "learning_rate": 0.0001792045641931564,
      "loss": 11.1775,
      "step": 177800
    },
    {
      "epoch": 21.3971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017920224177414982,
      "loss": 11.0372,
      "step": 177810
    },
    {
      "epoch": 21.398315282791817,
      "grad_norm": Infinity,
      "learning_rate": 0.00017919991924051804,
      "loss": 11.0467,
      "step": 177820
    },
    {
      "epoch": 21.399518652226234,
      "grad_norm": Infinity,
      "learning_rate": 0.00017919759659226444,
      "loss": 11.1863,
      "step": 177830
    },
    {
      "epoch": 21.40072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00017919527382939237,
      "loss": 11.1306,
      "step": 177840
    },
    {
      "epoch": 21.401925391095066,
      "grad_norm": Infinity,
      "learning_rate": 0.00017919295095190517,
      "loss": 11.2199,
      "step": 177850
    },
    {
      "epoch": 21.403128760529484,
      "grad_norm": Infinity,
      "learning_rate": 0.00017919062795980622,
      "loss": 11.0467,
      "step": 177860
    },
    {
      "epoch": 21.404332129963898,
      "grad_norm": Infinity,
      "learning_rate": 0.0001791883048530989,
      "loss": 11.1701,
      "step": 177870
    },
    {
      "epoch": 21.405535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00017918598163178652,
      "loss": 11.1886,
      "step": 177880
    },
    {
      "epoch": 21.406738868832733,
      "grad_norm": Infinity,
      "learning_rate": 0.00017918365829587248,
      "loss": 11.059,
      "step": 177890
    },
    {
      "epoch": 21.407942238267147,
      "grad_norm": Infinity,
      "learning_rate": 0.00017918133484536014,
      "loss": 11.1136,
      "step": 177900
    },
    {
      "epoch": 21.409145607701564,
      "grad_norm": Infinity,
      "learning_rate": 0.00017917901128025287,
      "loss": 11.146,
      "step": 177910
    },
    {
      "epoch": 21.410348977135982,
      "grad_norm": Infinity,
      "learning_rate": 0.00017917668760055402,
      "loss": 11.1497,
      "step": 177920
    },
    {
      "epoch": 21.411552346570396,
      "grad_norm": Infinity,
      "learning_rate": 0.00017917436380626695,
      "loss": 11.1289,
      "step": 177930
    },
    {
      "epoch": 21.412755716004813,
      "grad_norm": Infinity,
      "learning_rate": 0.00017917203989739505,
      "loss": 11.0559,
      "step": 177940
    },
    {
      "epoch": 21.41395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017916971587394163,
      "loss": 11.1901,
      "step": 177950
    },
    {
      "epoch": 21.415162454873645,
      "grad_norm": Infinity,
      "learning_rate": 0.0001791673917359101,
      "loss": 11.1796,
      "step": 177960
    },
    {
      "epoch": 21.416365824308063,
      "grad_norm": Infinity,
      "learning_rate": 0.00017916506748330379,
      "loss": 11.0309,
      "step": 177970
    },
    {
      "epoch": 21.41756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.0001791627431161261,
      "loss": 11.2446,
      "step": 177980
    },
    {
      "epoch": 21.418772563176894,
      "grad_norm": Infinity,
      "learning_rate": 0.00017916041863438037,
      "loss": 11.1847,
      "step": 177990
    },
    {
      "epoch": 21.41997593261131,
      "grad_norm": Infinity,
      "learning_rate": 0.00017915809403806998,
      "loss": 11.1351,
      "step": 178000
    },
    {
      "epoch": 21.42117930204573,
      "grad_norm": Infinity,
      "learning_rate": 0.00017915576932719826,
      "loss": 11.1855,
      "step": 178010
    },
    {
      "epoch": 21.422382671480143,
      "grad_norm": Infinity,
      "learning_rate": 0.00017915344450176863,
      "loss": 11.1336,
      "step": 178020
    },
    {
      "epoch": 21.42358604091456,
      "grad_norm": Infinity,
      "learning_rate": 0.00017915111956178442,
      "loss": 11.1641,
      "step": 178030
    },
    {
      "epoch": 21.42478941034898,
      "grad_norm": Infinity,
      "learning_rate": 0.00017914879450724899,
      "loss": 11.0796,
      "step": 178040
    },
    {
      "epoch": 21.425992779783392,
      "grad_norm": Infinity,
      "learning_rate": 0.0001791464693381657,
      "loss": 11.3461,
      "step": 178050
    },
    {
      "epoch": 21.42719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00017914414405453792,
      "loss": 11.0922,
      "step": 178060
    },
    {
      "epoch": 21.428399518652228,
      "grad_norm": Infinity,
      "learning_rate": 0.00017914181865636905,
      "loss": 11.1003,
      "step": 178070
    },
    {
      "epoch": 21.42960288808664,
      "grad_norm": Infinity,
      "learning_rate": 0.00017913949314366244,
      "loss": 11.1272,
      "step": 178080
    },
    {
      "epoch": 21.43080625752106,
      "grad_norm": Infinity,
      "learning_rate": 0.00017913716751642142,
      "loss": 11.2766,
      "step": 178090
    },
    {
      "epoch": 21.432009626955477,
      "grad_norm": Infinity,
      "learning_rate": 0.0001791348417746494,
      "loss": 11.2385,
      "step": 178100
    },
    {
      "epoch": 21.43321299638989,
      "grad_norm": Infinity,
      "learning_rate": 0.0001791325159183497,
      "loss": 11.2543,
      "step": 178110
    },
    {
      "epoch": 21.43441636582431,
      "grad_norm": Infinity,
      "learning_rate": 0.00017913018994752573,
      "loss": 11.0706,
      "step": 178120
    },
    {
      "epoch": 21.435619735258726,
      "grad_norm": Infinity,
      "learning_rate": 0.0001791278638621808,
      "loss": 11.1407,
      "step": 178130
    },
    {
      "epoch": 21.43682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00017912553766231837,
      "loss": 11.1637,
      "step": 178140
    },
    {
      "epoch": 21.438026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00017912321134794172,
      "loss": 11.1057,
      "step": 178150
    },
    {
      "epoch": 21.439229843561975,
      "grad_norm": Infinity,
      "learning_rate": 0.00017912088491905425,
      "loss": 11.1744,
      "step": 178160
    },
    {
      "epoch": 21.44043321299639,
      "grad_norm": Infinity,
      "learning_rate": 0.00017911855837565935,
      "loss": 11.0698,
      "step": 178170
    },
    {
      "epoch": 21.441636582430807,
      "grad_norm": Infinity,
      "learning_rate": 0.00017911623171776032,
      "loss": 11.0919,
      "step": 178180
    },
    {
      "epoch": 21.442839951865224,
      "grad_norm": Infinity,
      "learning_rate": 0.00017911390494536062,
      "loss": 11.1396,
      "step": 178190
    },
    {
      "epoch": 21.444043321299638,
      "grad_norm": Infinity,
      "learning_rate": 0.00017911157805846354,
      "loss": 11.069,
      "step": 178200
    },
    {
      "epoch": 21.445246690734056,
      "grad_norm": Infinity,
      "learning_rate": 0.00017910925105707248,
      "loss": 11.2572,
      "step": 178210
    },
    {
      "epoch": 21.446450060168473,
      "grad_norm": Infinity,
      "learning_rate": 0.00017910692394119078,
      "loss": 11.1759,
      "step": 178220
    },
    {
      "epoch": 21.447653429602887,
      "grad_norm": Infinity,
      "learning_rate": 0.00017910459671082184,
      "loss": 11.1208,
      "step": 178230
    },
    {
      "epoch": 21.448856799037305,
      "grad_norm": Infinity,
      "learning_rate": 0.00017910226936596906,
      "loss": 11.133,
      "step": 178240
    },
    {
      "epoch": 21.450060168471722,
      "grad_norm": Infinity,
      "learning_rate": 0.00017909994190663572,
      "loss": 11.1646,
      "step": 178250
    },
    {
      "epoch": 21.451263537906136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017909761433282525,
      "loss": 11.1254,
      "step": 178260
    },
    {
      "epoch": 21.452466907340554,
      "grad_norm": Infinity,
      "learning_rate": 0.000179095286644541,
      "loss": 11.0085,
      "step": 178270
    },
    {
      "epoch": 21.45367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017909295884178635,
      "loss": 11.1745,
      "step": 178280
    },
    {
      "epoch": 21.454873646209386,
      "grad_norm": Infinity,
      "learning_rate": 0.00017909063092456465,
      "loss": 11.0254,
      "step": 178290
    },
    {
      "epoch": 21.456077015643803,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790883028928793,
      "loss": 11.1441,
      "step": 178300
    },
    {
      "epoch": 21.45728038507822,
      "grad_norm": Infinity,
      "learning_rate": 0.00017908597474673363,
      "loss": 10.9994,
      "step": 178310
    },
    {
      "epoch": 21.458483754512635,
      "grad_norm": Infinity,
      "learning_rate": 0.00017908364648613106,
      "loss": 11.1418,
      "step": 178320
    },
    {
      "epoch": 21.459687123947052,
      "grad_norm": Infinity,
      "learning_rate": 0.00017908131811107488,
      "loss": 11.0872,
      "step": 178330
    },
    {
      "epoch": 21.460890493381466,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790789896215685,
      "loss": 10.9967,
      "step": 178340
    },
    {
      "epoch": 21.462093862815884,
      "grad_norm": Infinity,
      "learning_rate": 0.00017907666101761536,
      "loss": 11.186,
      "step": 178350
    },
    {
      "epoch": 21.4632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00017907433229921873,
      "loss": 11.1498,
      "step": 178360
    },
    {
      "epoch": 21.464500601684716,
      "grad_norm": Infinity,
      "learning_rate": 0.00017907200346638203,
      "loss": 11.098,
      "step": 178370
    },
    {
      "epoch": 21.465703971119133,
      "grad_norm": Infinity,
      "learning_rate": 0.00017906967451910858,
      "loss": 11.1497,
      "step": 178380
    },
    {
      "epoch": 21.46690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.00017906734545740185,
      "loss": 11.0304,
      "step": 178390
    },
    {
      "epoch": 21.468110709987965,
      "grad_norm": Infinity,
      "learning_rate": 0.00017906501628126513,
      "loss": 11.1167,
      "step": 178400
    },
    {
      "epoch": 21.469314079422382,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790626869907018,
      "loss": 11.2074,
      "step": 178410
    },
    {
      "epoch": 21.4705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017906035758571525,
      "loss": 11.1508,
      "step": 178420
    },
    {
      "epoch": 21.471720818291214,
      "grad_norm": Infinity,
      "learning_rate": 0.00017905802806630883,
      "loss": 11.1496,
      "step": 178430
    },
    {
      "epoch": 21.47292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00017905569843248596,
      "loss": 11.1121,
      "step": 178440
    },
    {
      "epoch": 21.47412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.00017905336868424994,
      "loss": 11.1338,
      "step": 178450
    },
    {
      "epoch": 21.475330926594463,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790510388216042,
      "loss": 11.0631,
      "step": 178460
    },
    {
      "epoch": 21.47653429602888,
      "grad_norm": Infinity,
      "learning_rate": 0.00017904870884455205,
      "loss": 11.1566,
      "step": 178470
    },
    {
      "epoch": 21.477737665463298,
      "grad_norm": Infinity,
      "learning_rate": 0.00017904637875309693,
      "loss": 11.1467,
      "step": 178480
    },
    {
      "epoch": 21.478941034897712,
      "grad_norm": Infinity,
      "learning_rate": 0.00017904404854724218,
      "loss": 11.0822,
      "step": 178490
    },
    {
      "epoch": 21.48014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017904171822699118,
      "loss": 11.1704,
      "step": 178500
    },
    {
      "epoch": 21.481347773766547,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790393877923473,
      "loss": 11.1729,
      "step": 178510
    },
    {
      "epoch": 21.48255114320096,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790370572433139,
      "loss": 11.1836,
      "step": 178520
    },
    {
      "epoch": 21.48375451263538,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790347265798944,
      "loss": 11.2747,
      "step": 178530
    },
    {
      "epoch": 21.484957882069796,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790323958020921,
      "loss": 11.1204,
      "step": 178540
    },
    {
      "epoch": 21.48616125150421,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790300649099104,
      "loss": 11.1644,
      "step": 178550
    },
    {
      "epoch": 21.487364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.00017902773390335273,
      "loss": 11.1467,
      "step": 178560
    },
    {
      "epoch": 21.488567990373046,
      "grad_norm": Infinity,
      "learning_rate": 0.00017902540278242238,
      "loss": 11.1352,
      "step": 178570
    },
    {
      "epoch": 21.48977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790230715471228,
      "loss": 11.1543,
      "step": 178580
    },
    {
      "epoch": 21.490974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00017902074019745729,
      "loss": 11.1715,
      "step": 178590
    },
    {
      "epoch": 21.492178098676295,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790184087334293,
      "loss": 11.1144,
      "step": 178600
    },
    {
      "epoch": 21.49338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00017901607715504212,
      "loss": 11.1829,
      "step": 178610
    },
    {
      "epoch": 21.494584837545126,
      "grad_norm": Infinity,
      "learning_rate": 0.00017901374546229918,
      "loss": 11.2078,
      "step": 178620
    },
    {
      "epoch": 21.495788206979544,
      "grad_norm": Infinity,
      "learning_rate": 0.0001790114136552038,
      "loss": 11.0871,
      "step": 178630
    },
    {
      "epoch": 21.496991576413958,
      "grad_norm": Infinity,
      "learning_rate": 0.00017900908173375948,
      "loss": 11.0597,
      "step": 178640
    },
    {
      "epoch": 21.498194945848375,
      "grad_norm": Infinity,
      "learning_rate": 0.00017900674969796948,
      "loss": 11.1074,
      "step": 178650
    },
    {
      "epoch": 21.499398315282793,
      "grad_norm": Infinity,
      "learning_rate": 0.00017900441754783718,
      "loss": 11.1345,
      "step": 178660
    },
    {
      "epoch": 21.500601684717207,
      "grad_norm": Infinity,
      "learning_rate": 0.00017900208528336603,
      "loss": 11.2014,
      "step": 178670
    },
    {
      "epoch": 21.501805054151625,
      "grad_norm": Infinity,
      "learning_rate": 0.00017899975290455933,
      "loss": 11.1527,
      "step": 178680
    },
    {
      "epoch": 21.503008423586042,
      "grad_norm": Infinity,
      "learning_rate": 0.00017899742041142046,
      "loss": 11.2237,
      "step": 178690
    },
    {
      "epoch": 21.504211793020456,
      "grad_norm": Infinity,
      "learning_rate": 0.00017899508780395287,
      "loss": 11.1291,
      "step": 178700
    },
    {
      "epoch": 21.505415162454874,
      "grad_norm": Infinity,
      "learning_rate": 0.00017899275508215985,
      "loss": 11.0648,
      "step": 178710
    },
    {
      "epoch": 21.50661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00017899042224604484,
      "loss": 11.0635,
      "step": 178720
    },
    {
      "epoch": 21.507821901323705,
      "grad_norm": Infinity,
      "learning_rate": 0.00017898808929561115,
      "loss": 11.1899,
      "step": 178730
    },
    {
      "epoch": 21.509025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789857562308622,
      "loss": 11.1596,
      "step": 178740
    },
    {
      "epoch": 21.51022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.00017898342305180134,
      "loss": 11.1185,
      "step": 178750
    },
    {
      "epoch": 21.511432009626954,
      "grad_norm": Infinity,
      "learning_rate": 0.000178981089758432,
      "loss": 11.0803,
      "step": 178760
    },
    {
      "epoch": 21.512635379061372,
      "grad_norm": Infinity,
      "learning_rate": 0.00017897875635075754,
      "loss": 11.0858,
      "step": 178770
    },
    {
      "epoch": 21.51383874849579,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789764228287813,
      "loss": 11.2377,
      "step": 178780
    },
    {
      "epoch": 21.515042117930204,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789740891925067,
      "loss": 11.1412,
      "step": 178790
    },
    {
      "epoch": 21.51624548736462,
      "grad_norm": Infinity,
      "learning_rate": 0.00017897175544193705,
      "loss": 11.0815,
      "step": 178800
    },
    {
      "epoch": 21.51744885679904,
      "grad_norm": Infinity,
      "learning_rate": 0.00017896942157707578,
      "loss": 11.1827,
      "step": 178810
    },
    {
      "epoch": 21.518652226233453,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789670875979263,
      "loss": 11.2028,
      "step": 178820
    },
    {
      "epoch": 21.51985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017896475350449192,
      "loss": 11.1479,
      "step": 178830
    },
    {
      "epoch": 21.521058965102288,
      "grad_norm": Infinity,
      "learning_rate": 0.00017896241929677605,
      "loss": 11.0077,
      "step": 178840
    },
    {
      "epoch": 21.522262334536702,
      "grad_norm": Infinity,
      "learning_rate": 0.00017896008497478205,
      "loss": 11.1626,
      "step": 178850
    },
    {
      "epoch": 21.52346570397112,
      "grad_norm": Infinity,
      "learning_rate": 0.00017895775053851333,
      "loss": 11.1765,
      "step": 178860
    },
    {
      "epoch": 21.524669073405537,
      "grad_norm": Infinity,
      "learning_rate": 0.00017895541598797322,
      "loss": 11.1911,
      "step": 178870
    },
    {
      "epoch": 21.52587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00017895308132316517,
      "loss": 11.2009,
      "step": 178880
    },
    {
      "epoch": 21.52707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789507465440925,
      "loss": 11.1698,
      "step": 178890
    },
    {
      "epoch": 21.528279181708786,
      "grad_norm": Infinity,
      "learning_rate": 0.00017894841165075863,
      "loss": 11.1596,
      "step": 178900
    },
    {
      "epoch": 21.5294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789460766431669,
      "loss": 11.2234,
      "step": 178910
    },
    {
      "epoch": 21.530685920577618,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789437415213207,
      "loss": 11.236,
      "step": 178920
    },
    {
      "epoch": 21.531889290012035,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789414062852234,
      "loss": 11.016,
      "step": 178930
    },
    {
      "epoch": 21.53309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.00017893907093487842,
      "loss": 11.1852,
      "step": 178940
    },
    {
      "epoch": 21.534296028880867,
      "grad_norm": Infinity,
      "learning_rate": 0.00017893673547028912,
      "loss": 11.1889,
      "step": 178950
    },
    {
      "epoch": 21.535499398315284,
      "grad_norm": Infinity,
      "learning_rate": 0.00017893439989145885,
      "loss": 11.1744,
      "step": 178960
    },
    {
      "epoch": 21.5367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017893206419839102,
      "loss": 11.0609,
      "step": 178970
    },
    {
      "epoch": 21.537906137184116,
      "grad_norm": Infinity,
      "learning_rate": 0.00017892972839108904,
      "loss": 11.164,
      "step": 178980
    },
    {
      "epoch": 21.53910950661853,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789273924695562,
      "loss": 11.1089,
      "step": 178990
    },
    {
      "epoch": 21.540312876052948,
      "grad_norm": Infinity,
      "learning_rate": 0.00017892505643379595,
      "loss": 11.0832,
      "step": 179000
    },
    {
      "epoch": 21.541516245487365,
      "grad_norm": Infinity,
      "learning_rate": 0.00017892272028381169,
      "loss": 11.1279,
      "step": 179010
    },
    {
      "epoch": 21.54271961492178,
      "grad_norm": Infinity,
      "learning_rate": 0.00017892038401960675,
      "loss": 11.1563,
      "step": 179020
    },
    {
      "epoch": 21.543922984356197,
      "grad_norm": Infinity,
      "learning_rate": 0.00017891804764118453,
      "loss": 11.0521,
      "step": 179030
    },
    {
      "epoch": 21.545126353790614,
      "grad_norm": Infinity,
      "learning_rate": 0.00017891571114854842,
      "loss": 11.22,
      "step": 179040
    },
    {
      "epoch": 21.54632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00017891337454170177,
      "loss": 11.2028,
      "step": 179050
    },
    {
      "epoch": 21.547533092659446,
      "grad_norm": Infinity,
      "learning_rate": 0.000178911037820648,
      "loss": 11.1106,
      "step": 179060
    },
    {
      "epoch": 21.548736462093864,
      "grad_norm": Infinity,
      "learning_rate": 0.00017890870098539048,
      "loss": 11.2388,
      "step": 179070
    },
    {
      "epoch": 21.549939831528278,
      "grad_norm": Infinity,
      "learning_rate": 0.00017890636403593258,
      "loss": 11.2003,
      "step": 179080
    },
    {
      "epoch": 21.551143200962695,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789040269722777,
      "loss": 11.1938,
      "step": 179090
    },
    {
      "epoch": 21.552346570397113,
      "grad_norm": Infinity,
      "learning_rate": 0.0001789016897944292,
      "loss": 11.1349,
      "step": 179100
    },
    {
      "epoch": 21.553549939831527,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788993525023905,
      "loss": 11.1204,
      "step": 179110
    },
    {
      "epoch": 21.554753309265944,
      "grad_norm": Infinity,
      "learning_rate": 0.00017889701509616492,
      "loss": 11.0467,
      "step": 179120
    },
    {
      "epoch": 21.555956678700362,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788946775757559,
      "loss": 11.1042,
      "step": 179130
    },
    {
      "epoch": 21.557160048134776,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788923399411668,
      "loss": 11.2238,
      "step": 179140
    },
    {
      "epoch": 21.558363417569193,
      "grad_norm": Infinity,
      "learning_rate": 0.000178890002192401,
      "loss": 11.1167,
      "step": 179150
    },
    {
      "epoch": 21.55956678700361,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788876643294619,
      "loss": 11.19,
      "step": 179160
    },
    {
      "epoch": 21.560770156438025,
      "grad_norm": Infinity,
      "learning_rate": 0.00017888532635235288,
      "loss": 11.1259,
      "step": 179170
    },
    {
      "epoch": 21.561973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788829882610773,
      "loss": 11.1525,
      "step": 179180
    },
    {
      "epoch": 21.56317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00017888065005563854,
      "loss": 11.1196,
      "step": 179190
    },
    {
      "epoch": 21.564380264741274,
      "grad_norm": Infinity,
      "learning_rate": 0.00017887831173604005,
      "loss": 11.2967,
      "step": 179200
    },
    {
      "epoch": 21.56558363417569,
      "grad_norm": Infinity,
      "learning_rate": 0.00017887597330228517,
      "loss": 11.0438,
      "step": 179210
    },
    {
      "epoch": 21.56678700361011,
      "grad_norm": Infinity,
      "learning_rate": 0.00017887363475437723,
      "loss": 11.1433,
      "step": 179220
    },
    {
      "epoch": 21.567990373044523,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788712960923197,
      "loss": 11.162,
      "step": 179230
    },
    {
      "epoch": 21.56919374247894,
      "grad_norm": Infinity,
      "learning_rate": 0.00017886895731611591,
      "loss": 11.1669,
      "step": 179240
    },
    {
      "epoch": 21.57039711191336,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788666184257693,
      "loss": 11.0664,
      "step": 179250
    },
    {
      "epoch": 21.571600481347772,
      "grad_norm": Infinity,
      "learning_rate": 0.00017886427942128324,
      "loss": 11.0326,
      "step": 179260
    },
    {
      "epoch": 21.57280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.00017886194030266103,
      "loss": 11.1657,
      "step": 179270
    },
    {
      "epoch": 21.574007220216608,
      "grad_norm": Infinity,
      "learning_rate": 0.00017885960106990619,
      "loss": 11.0824,
      "step": 179280
    },
    {
      "epoch": 21.57521058965102,
      "grad_norm": Infinity,
      "learning_rate": 0.00017885726172302198,
      "loss": 11.1467,
      "step": 179290
    },
    {
      "epoch": 21.57641395908544,
      "grad_norm": Infinity,
      "learning_rate": 0.00017885492226201187,
      "loss": 11.1381,
      "step": 179300
    },
    {
      "epoch": 21.577617328519857,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788525826868792,
      "loss": 11.1116,
      "step": 179310
    },
    {
      "epoch": 21.57882069795427,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788502429976274,
      "loss": 11.1916,
      "step": 179320
    },
    {
      "epoch": 21.58002406738869,
      "grad_norm": Infinity,
      "learning_rate": 0.00017884790319425982,
      "loss": 11.2377,
      "step": 179330
    },
    {
      "epoch": 21.581227436823106,
      "grad_norm": Infinity,
      "learning_rate": 0.00017884556327677986,
      "loss": 11.0818,
      "step": 179340
    },
    {
      "epoch": 21.58243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.00017884322324519092,
      "loss": 11.16,
      "step": 179350
    },
    {
      "epoch": 21.583634175691937,
      "grad_norm": Infinity,
      "learning_rate": 0.00017884088309949635,
      "loss": 11.1542,
      "step": 179360
    },
    {
      "epoch": 21.584837545126355,
      "grad_norm": Infinity,
      "learning_rate": 0.00017883854283969955,
      "loss": 11.1207,
      "step": 179370
    },
    {
      "epoch": 21.58604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00017883620246580392,
      "loss": 11.012,
      "step": 179380
    },
    {
      "epoch": 21.587244283995187,
      "grad_norm": Infinity,
      "learning_rate": 0.00017883386197781284,
      "loss": 11.0529,
      "step": 179390
    },
    {
      "epoch": 21.588447653429604,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788315213757297,
      "loss": 11.2252,
      "step": 179400
    },
    {
      "epoch": 21.589651022864018,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788291806595579,
      "loss": 11.1763,
      "step": 179410
    },
    {
      "epoch": 21.590854392298436,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788268398293008,
      "loss": 11.0922,
      "step": 179420
    },
    {
      "epoch": 21.592057761732853,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788244988849618,
      "loss": 11.1546,
      "step": 179430
    },
    {
      "epoch": 21.593261131167267,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788221578265443,
      "loss": 11.1087,
      "step": 179440
    },
    {
      "epoch": 21.594464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00017881981665405166,
      "loss": 11.189,
      "step": 179450
    },
    {
      "epoch": 21.595667870036102,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788174753674873,
      "loss": 10.9882,
      "step": 179460
    },
    {
      "epoch": 21.596871239470516,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788151339668546,
      "loss": 11.1569,
      "step": 179470
    },
    {
      "epoch": 21.598074608904934,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788127924521569,
      "loss": 11.1781,
      "step": 179480
    },
    {
      "epoch": 21.59927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788104508233977,
      "loss": 11.0443,
      "step": 179490
    },
    {
      "epoch": 21.600481347773766,
      "grad_norm": Infinity,
      "learning_rate": 0.00017880810908058026,
      "loss": 11.0961,
      "step": 179500
    },
    {
      "epoch": 21.601684717208183,
      "grad_norm": Infinity,
      "learning_rate": 0.00017880576722370805,
      "loss": 11.0358,
      "step": 179510
    },
    {
      "epoch": 21.6028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.00017880342525278443,
      "loss": 11.198,
      "step": 179520
    },
    {
      "epoch": 21.604091456077015,
      "grad_norm": Infinity,
      "learning_rate": 0.0001788010831678128,
      "loss": 11.1522,
      "step": 179530
    },
    {
      "epoch": 21.605294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00017879874096879655,
      "loss": 11.218,
      "step": 179540
    },
    {
      "epoch": 21.60649819494585,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787963986557391,
      "loss": 11.1059,
      "step": 179550
    },
    {
      "epoch": 21.607701564380264,
      "grad_norm": Infinity,
      "learning_rate": 0.00017879405622864378,
      "loss": 11.0864,
      "step": 179560
    },
    {
      "epoch": 21.60890493381468,
      "grad_norm": Infinity,
      "learning_rate": 0.00017879171368751398,
      "loss": 11.1138,
      "step": 179570
    },
    {
      "epoch": 21.6101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.00017878937103235317,
      "loss": 11.0421,
      "step": 179580
    },
    {
      "epoch": 21.611311672683513,
      "grad_norm": Infinity,
      "learning_rate": 0.00017878702826316464,
      "loss": 11.164,
      "step": 179590
    },
    {
      "epoch": 21.61251504211793,
      "grad_norm": Infinity,
      "learning_rate": 0.00017878468537995185,
      "loss": 11.1227,
      "step": 179600
    },
    {
      "epoch": 21.613718411552348,
      "grad_norm": Infinity,
      "learning_rate": 0.00017878234238271818,
      "loss": 11.12,
      "step": 179610
    },
    {
      "epoch": 21.614921780986762,
      "grad_norm": Infinity,
      "learning_rate": 0.00017877999927146698,
      "loss": 11.1591,
      "step": 179620
    },
    {
      "epoch": 21.61612515042118,
      "grad_norm": Infinity,
      "learning_rate": 0.00017877765604620168,
      "loss": 11.1408,
      "step": 179630
    },
    {
      "epoch": 21.617328519855597,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787753127069257,
      "loss": 11.1705,
      "step": 179640
    },
    {
      "epoch": 21.61853188929001,
      "grad_norm": Infinity,
      "learning_rate": 0.00017877296925364236,
      "loss": 11.032,
      "step": 179650
    },
    {
      "epoch": 21.61973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.00017877062568635507,
      "loss": 11.2358,
      "step": 179660
    },
    {
      "epoch": 21.620938628158846,
      "grad_norm": Infinity,
      "learning_rate": 0.00017876828200506727,
      "loss": 11.1123,
      "step": 179670
    },
    {
      "epoch": 21.62214199759326,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787659382097823,
      "loss": 11.2215,
      "step": 179680
    },
    {
      "epoch": 21.623345367027678,
      "grad_norm": Infinity,
      "learning_rate": 0.00017876359430050358,
      "loss": 11.1523,
      "step": 179690
    },
    {
      "epoch": 21.624548736462096,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787612502772345,
      "loss": 11.1506,
      "step": 179700
    },
    {
      "epoch": 21.62575210589651,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787589061399784,
      "loss": 11.1127,
      "step": 179710
    },
    {
      "epoch": 21.626955475330927,
      "grad_norm": Infinity,
      "learning_rate": 0.00017875656188873874,
      "loss": 11.0368,
      "step": 179720
    },
    {
      "epoch": 21.628158844765345,
      "grad_norm": Infinity,
      "learning_rate": 0.00017875421752351894,
      "loss": 11.042,
      "step": 179730
    },
    {
      "epoch": 21.62936221419976,
      "grad_norm": Infinity,
      "learning_rate": 0.00017875187304432227,
      "loss": 11.2149,
      "step": 179740
    },
    {
      "epoch": 21.630565583634176,
      "grad_norm": Infinity,
      "learning_rate": 0.00017874952845115224,
      "loss": 11.1029,
      "step": 179750
    },
    {
      "epoch": 21.63176895306859,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787471837440122,
      "loss": 11.1455,
      "step": 179760
    },
    {
      "epoch": 21.632972322503008,
      "grad_norm": Infinity,
      "learning_rate": 0.00017874483892290553,
      "loss": 11.1065,
      "step": 179770
    },
    {
      "epoch": 21.634175691937426,
      "grad_norm": Infinity,
      "learning_rate": 0.00017874249398783565,
      "loss": 11.0886,
      "step": 179780
    },
    {
      "epoch": 21.63537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.00017874014893880594,
      "loss": 11.2223,
      "step": 179790
    },
    {
      "epoch": 21.636582430806257,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787378037758198,
      "loss": 11.0646,
      "step": 179800
    },
    {
      "epoch": 21.637785800240675,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787354584988806,
      "loss": 11.1551,
      "step": 179810
    },
    {
      "epoch": 21.63898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00017873311310799173,
      "loss": 11.2092,
      "step": 179820
    },
    {
      "epoch": 21.640192539109506,
      "grad_norm": Infinity,
      "learning_rate": 0.00017873076760315665,
      "loss": 11.1924,
      "step": 179830
    },
    {
      "epoch": 21.641395908543924,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787284219843787,
      "loss": 11.1122,
      "step": 179840
    },
    {
      "epoch": 21.642599277978338,
      "grad_norm": Infinity,
      "learning_rate": 0.00017872607625166124,
      "loss": 11.2104,
      "step": 179850
    },
    {
      "epoch": 21.643802647412755,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787237304050078,
      "loss": 11.1106,
      "step": 179860
    },
    {
      "epoch": 21.645006016847173,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787213844444216,
      "loss": 11.1625,
      "step": 179870
    },
    {
      "epoch": 21.646209386281587,
      "grad_norm": Infinity,
      "learning_rate": 0.00017871903836990616,
      "loss": 11.2445,
      "step": 179880
    },
    {
      "epoch": 21.647412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.00017871669218146483,
      "loss": 11.0936,
      "step": 179890
    },
    {
      "epoch": 21.648616125150422,
      "grad_norm": Infinity,
      "learning_rate": 0.000178714345879101,
      "loss": 11.0613,
      "step": 179900
    },
    {
      "epoch": 21.649819494584836,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787119994628181,
      "loss": 11.2723,
      "step": 179910
    },
    {
      "epoch": 21.651022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787096529326195,
      "loss": 11.2057,
      "step": 179920
    },
    {
      "epoch": 21.65222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787073062885086,
      "loss": 11.0551,
      "step": 179930
    },
    {
      "epoch": 21.653429602888085,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787049595304888,
      "loss": 11.1519,
      "step": 179940
    },
    {
      "epoch": 21.654632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.0001787026126585635,
      "loss": 11.2491,
      "step": 179950
    },
    {
      "epoch": 21.65583634175692,
      "grad_norm": Infinity,
      "learning_rate": 0.00017870026567273607,
      "loss": 11.0742,
      "step": 179960
    },
    {
      "epoch": 21.657039711191334,
      "grad_norm": Infinity,
      "learning_rate": 0.0001786979185730099,
      "loss": 11.0986,
      "step": 179970
    },
    {
      "epoch": 21.658243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.00017869557135938844,
      "loss": 11.2278,
      "step": 179980
    },
    {
      "epoch": 21.65944645006017,
      "grad_norm": Infinity,
      "learning_rate": 0.00017869322403187506,
      "loss": 11.2291,
      "step": 179990
    },
    {
      "epoch": 21.660649819494584,
      "grad_norm": Infinity,
      "learning_rate": 0.00017869087659047315,
      "loss": 11.0964,
      "step": 180000
    },
    {
      "epoch": 21.661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00017868852903518614,
      "loss": 11.2089,
      "step": 180010
    },
    {
      "epoch": 21.66305655836342,
      "grad_norm": Infinity,
      "learning_rate": 0.00017868618136601737,
      "loss": 11.0344,
      "step": 180020
    },
    {
      "epoch": 21.664259927797833,
      "grad_norm": Infinity,
      "learning_rate": 0.00017868383358297027,
      "loss": 11.1704,
      "step": 180030
    },
    {
      "epoch": 21.66546329723225,
      "grad_norm": Infinity,
      "learning_rate": 0.00017868148568604824,
      "loss": 11.1504,
      "step": 180040
    },
    {
      "epoch": 21.666666666666668,
      "grad_norm": Infinity,
      "learning_rate": 0.00017867913767525472,
      "loss": 11.1858,
      "step": 180050
    },
    {
      "epoch": 21.667870036101082,
      "grad_norm": Infinity,
      "learning_rate": 0.000178676789550593,
      "loss": 11.2171,
      "step": 180060
    },
    {
      "epoch": 21.6690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.00017867444131206654,
      "loss": 11.188,
      "step": 180070
    },
    {
      "epoch": 21.670276774969917,
      "grad_norm": Infinity,
      "learning_rate": 0.00017867209295967877,
      "loss": 11.1513,
      "step": 180080
    },
    {
      "epoch": 21.67148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00017866974449343305,
      "loss": 11.108,
      "step": 180090
    },
    {
      "epoch": 21.67268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.0001786673959133328,
      "loss": 11.2216,
      "step": 180100
    },
    {
      "epoch": 21.673886883273166,
      "grad_norm": Infinity,
      "learning_rate": 0.00017866504721938137,
      "loss": 11.1627,
      "step": 180110
    },
    {
      "epoch": 21.67509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.00017866269841158223,
      "loss": 11.0329,
      "step": 180120
    },
    {
      "epoch": 21.676293622141998,
      "grad_norm": Infinity,
      "learning_rate": 0.00017866034948993873,
      "loss": 11.1457,
      "step": 180130
    },
    {
      "epoch": 21.677496991576415,
      "grad_norm": Infinity,
      "learning_rate": 0.0001786580004544543,
      "loss": 11.1479,
      "step": 180140
    },
    {
      "epoch": 21.67870036101083,
      "grad_norm": Infinity,
      "learning_rate": 0.00017865565130513228,
      "loss": 11.1805,
      "step": 180150
    },
    {
      "epoch": 21.679903730445247,
      "grad_norm": Infinity,
      "learning_rate": 0.00017865330204197615,
      "loss": 11.1952,
      "step": 180160
    },
    {
      "epoch": 21.681107099879664,
      "grad_norm": Infinity,
      "learning_rate": 0.00017865095266498925,
      "loss": 10.9904,
      "step": 180170
    },
    {
      "epoch": 21.68231046931408,
      "grad_norm": Infinity,
      "learning_rate": 0.00017864860317417502,
      "loss": 11.1384,
      "step": 180180
    },
    {
      "epoch": 21.683513838748496,
      "grad_norm": Infinity,
      "learning_rate": 0.00017864625356953684,
      "loss": 11.1221,
      "step": 180190
    },
    {
      "epoch": 21.684717208182914,
      "grad_norm": Infinity,
      "learning_rate": 0.0001786439038510781,
      "loss": 11.2118,
      "step": 180200
    },
    {
      "epoch": 21.685920577617328,
      "grad_norm": Infinity,
      "learning_rate": 0.00017864155401880224,
      "loss": 11.0066,
      "step": 180210
    },
    {
      "epoch": 21.687123947051745,
      "grad_norm": Infinity,
      "learning_rate": 0.00017863920407271264,
      "loss": 11.1237,
      "step": 180220
    },
    {
      "epoch": 21.688327316486163,
      "grad_norm": Infinity,
      "learning_rate": 0.00017863685401281267,
      "loss": 11.2164,
      "step": 180230
    },
    {
      "epoch": 21.689530685920577,
      "grad_norm": Infinity,
      "learning_rate": 0.00017863450383910577,
      "loss": 11.2231,
      "step": 180240
    },
    {
      "epoch": 21.690734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.00017863215355159531,
      "loss": 11.0839,
      "step": 180250
    },
    {
      "epoch": 21.691937424789412,
      "grad_norm": Infinity,
      "learning_rate": 0.00017862980315028475,
      "loss": 11.1832,
      "step": 180260
    },
    {
      "epoch": 21.693140794223826,
      "grad_norm": Infinity,
      "learning_rate": 0.00017862745263517744,
      "loss": 11.1609,
      "step": 180270
    },
    {
      "epoch": 21.694344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00017862510200627676,
      "loss": 11.1112,
      "step": 180280
    },
    {
      "epoch": 21.69554753309266,
      "grad_norm": Infinity,
      "learning_rate": 0.00017862275126358617,
      "loss": 11.1118,
      "step": 180290
    },
    {
      "epoch": 21.696750902527075,
      "grad_norm": Infinity,
      "learning_rate": 0.00017862040040710905,
      "loss": 11.1723,
      "step": 180300
    },
    {
      "epoch": 21.697954271961493,
      "grad_norm": Infinity,
      "learning_rate": 0.00017861804943684878,
      "loss": 11.2083,
      "step": 180310
    },
    {
      "epoch": 21.69915764139591,
      "grad_norm": Infinity,
      "learning_rate": 0.0001786156983528088,
      "loss": 11.2532,
      "step": 180320
    },
    {
      "epoch": 21.700361010830324,
      "grad_norm": Infinity,
      "learning_rate": 0.0001786133471549925,
      "loss": 11.1602,
      "step": 180330
    },
    {
      "epoch": 21.70156438026474,
      "grad_norm": Infinity,
      "learning_rate": 0.00017861099584340326,
      "loss": 11.0888,
      "step": 180340
    },
    {
      "epoch": 21.70276774969916,
      "grad_norm": Infinity,
      "learning_rate": 0.0001786086444180445,
      "loss": 11.0899,
      "step": 180350
    },
    {
      "epoch": 21.703971119133573,
      "grad_norm": Infinity,
      "learning_rate": 0.00017860629287891965,
      "loss": 11.0778,
      "step": 180360
    },
    {
      "epoch": 21.70517448856799,
      "grad_norm": Infinity,
      "learning_rate": 0.0001786039412260321,
      "loss": 11.1506,
      "step": 180370
    },
    {
      "epoch": 21.706377858002405,
      "grad_norm": Infinity,
      "learning_rate": 0.00017860158945938519,
      "loss": 11.1819,
      "step": 180380
    },
    {
      "epoch": 21.707581227436823,
      "grad_norm": Infinity,
      "learning_rate": 0.00017859923757898238,
      "loss": 11.1748,
      "step": 180390
    },
    {
      "epoch": 21.70878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785968855848271,
      "loss": 11.1569,
      "step": 180400
    },
    {
      "epoch": 21.709987966305654,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785945334769227,
      "loss": 11.1512,
      "step": 180410
    },
    {
      "epoch": 21.71119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785921812552726,
      "loss": 11.1474,
      "step": 180420
    },
    {
      "epoch": 21.71239470517449,
      "grad_norm": Infinity,
      "learning_rate": 0.00017858982891988024,
      "loss": 11.1456,
      "step": 180430
    },
    {
      "epoch": 21.713598074608903,
      "grad_norm": Infinity,
      "learning_rate": 0.00017858747647074897,
      "loss": 11.1321,
      "step": 180440
    },
    {
      "epoch": 21.71480144404332,
      "grad_norm": Infinity,
      "learning_rate": 0.00017858512390788223,
      "loss": 11.2342,
      "step": 180450
    },
    {
      "epoch": 21.71600481347774,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785827712312834,
      "loss": 11.1463,
      "step": 180460
    },
    {
      "epoch": 21.717208182912152,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785804184409559,
      "loss": 11.1173,
      "step": 180470
    },
    {
      "epoch": 21.71841155234657,
      "grad_norm": Infinity,
      "learning_rate": 0.00017857806553690314,
      "loss": 11.1319,
      "step": 180480
    },
    {
      "epoch": 21.719614921780988,
      "grad_norm": Infinity,
      "learning_rate": 0.00017857571251912851,
      "loss": 11.0869,
      "step": 180490
    },
    {
      "epoch": 21.7208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00017857335938763544,
      "loss": 11.2537,
      "step": 180500
    },
    {
      "epoch": 21.72202166064982,
      "grad_norm": Infinity,
      "learning_rate": 0.00017857100614242731,
      "loss": 11.2086,
      "step": 180510
    },
    {
      "epoch": 21.723225030084237,
      "grad_norm": Infinity,
      "learning_rate": 0.00017856865278350755,
      "loss": 11.1466,
      "step": 180520
    },
    {
      "epoch": 21.72442839951865,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785662993108795,
      "loss": 11.1545,
      "step": 180530
    },
    {
      "epoch": 21.72563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00017856394572454665,
      "loss": 11.1046,
      "step": 180540
    },
    {
      "epoch": 21.726835138387486,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785615920245124,
      "loss": 11.1605,
      "step": 180550
    },
    {
      "epoch": 21.7280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.00017855923821078013,
      "loss": 11.172,
      "step": 180560
    },
    {
      "epoch": 21.729241877256317,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785568842833532,
      "loss": 11.0667,
      "step": 180570
    },
    {
      "epoch": 21.730445246690735,
      "grad_norm": Infinity,
      "learning_rate": 0.00017855453024223511,
      "loss": 11.1737,
      "step": 180580
    },
    {
      "epoch": 21.73164861612515,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785521760874292,
      "loss": 11.0979,
      "step": 180590
    },
    {
      "epoch": 21.732851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785498218189389,
      "loss": 11.164,
      "step": 180600
    },
    {
      "epoch": 21.734055354993984,
      "grad_norm": Infinity,
      "learning_rate": 0.00017854746743676763,
      "loss": 11.1741,
      "step": 180610
    },
    {
      "epoch": 21.735258724428398,
      "grad_norm": Infinity,
      "learning_rate": 0.00017854511294091875,
      "loss": 11.1297,
      "step": 180620
    },
    {
      "epoch": 21.736462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785427583313957,
      "loss": 11.1815,
      "step": 180630
    },
    {
      "epoch": 21.737665463297233,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785404036082019,
      "loss": 11.1617,
      "step": 180640
    },
    {
      "epoch": 21.738868832731647,
      "grad_norm": Infinity,
      "learning_rate": 0.00017853804877134076,
      "loss": 11.2375,
      "step": 180650
    },
    {
      "epoch": 21.740072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00017853569382081564,
      "loss": 11.1206,
      "step": 180660
    },
    {
      "epoch": 21.741275571600482,
      "grad_norm": Infinity,
      "learning_rate": 0.00017853333875663,
      "loss": 11.2529,
      "step": 180670
    },
    {
      "epoch": 21.742478941034896,
      "grad_norm": Infinity,
      "learning_rate": 0.00017853098357878725,
      "loss": 11.0324,
      "step": 180680
    },
    {
      "epoch": 21.743682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00017852862828729075,
      "loss": 11.1769,
      "step": 180690
    },
    {
      "epoch": 21.74488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.00017852627288214397,
      "loss": 11.1054,
      "step": 180700
    },
    {
      "epoch": 21.746089049338146,
      "grad_norm": Infinity,
      "learning_rate": 0.00017852391736335025,
      "loss": 11.1726,
      "step": 180710
    },
    {
      "epoch": 21.747292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00017852156173091304,
      "loss": 11.2157,
      "step": 180720
    },
    {
      "epoch": 21.74849578820698,
      "grad_norm": Infinity,
      "learning_rate": 0.00017851920598483575,
      "loss": 11.1392,
      "step": 180730
    },
    {
      "epoch": 21.749699157641395,
      "grad_norm": Infinity,
      "learning_rate": 0.00017851685012512178,
      "loss": 11.2039,
      "step": 180740
    },
    {
      "epoch": 21.750902527075812,
      "grad_norm": Infinity,
      "learning_rate": 0.00017851449415177455,
      "loss": 11.1707,
      "step": 180750
    },
    {
      "epoch": 21.75210589651023,
      "grad_norm": Infinity,
      "learning_rate": 0.00017851213806479744,
      "loss": 11.0881,
      "step": 180760
    },
    {
      "epoch": 21.753309265944644,
      "grad_norm": Infinity,
      "learning_rate": 0.00017850978186419392,
      "loss": 11.0546,
      "step": 180770
    },
    {
      "epoch": 21.75451263537906,
      "grad_norm": Infinity,
      "learning_rate": 0.00017850742554996734,
      "loss": 11.163,
      "step": 180780
    },
    {
      "epoch": 21.75571600481348,
      "grad_norm": Infinity,
      "learning_rate": 0.0001785050691221211,
      "loss": 11.1368,
      "step": 180790
    },
    {
      "epoch": 21.756919374247893,
      "grad_norm": Infinity,
      "learning_rate": 0.00017850271258065868,
      "loss": 11.1986,
      "step": 180800
    },
    {
      "epoch": 21.75812274368231,
      "grad_norm": Infinity,
      "learning_rate": 0.00017850035592558347,
      "loss": 11.121,
      "step": 180810
    },
    {
      "epoch": 21.759326113116728,
      "grad_norm": Infinity,
      "learning_rate": 0.00017849799915689886,
      "loss": 11.0966,
      "step": 180820
    },
    {
      "epoch": 21.760529482551142,
      "grad_norm": Infinity,
      "learning_rate": 0.00017849564227460823,
      "loss": 11.1748,
      "step": 180830
    },
    {
      "epoch": 21.76173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00017849328527871504,
      "loss": 11.2282,
      "step": 180840
    },
    {
      "epoch": 21.762936221419977,
      "grad_norm": Infinity,
      "learning_rate": 0.0001784909281692227,
      "loss": 11.1932,
      "step": 180850
    },
    {
      "epoch": 21.76413959085439,
      "grad_norm": Infinity,
      "learning_rate": 0.00017848857094613458,
      "loss": 11.1459,
      "step": 180860
    },
    {
      "epoch": 21.76534296028881,
      "grad_norm": Infinity,
      "learning_rate": 0.00017848621360945415,
      "loss": 11.2191,
      "step": 180870
    },
    {
      "epoch": 21.766546329723226,
      "grad_norm": Infinity,
      "learning_rate": 0.00017848385615918479,
      "loss": 10.993,
      "step": 180880
    },
    {
      "epoch": 21.76774969915764,
      "grad_norm": Infinity,
      "learning_rate": 0.0001784814985953299,
      "loss": 11.1873,
      "step": 180890
    },
    {
      "epoch": 21.768953068592058,
      "grad_norm": Infinity,
      "learning_rate": 0.0001784791409178929,
      "loss": 11.1758,
      "step": 180900
    },
    {
      "epoch": 21.770156438026476,
      "grad_norm": Infinity,
      "learning_rate": 0.0001784767831268772,
      "loss": 11.2121,
      "step": 180910
    },
    {
      "epoch": 21.77135980746089,
      "grad_norm": Infinity,
      "learning_rate": 0.00017847442522228626,
      "loss": 11.1558,
      "step": 180920
    },
    {
      "epoch": 21.772563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00017847206720412345,
      "loss": 11.159,
      "step": 180930
    },
    {
      "epoch": 21.773766546329725,
      "grad_norm": Infinity,
      "learning_rate": 0.00017846970907239214,
      "loss": 11.1322,
      "step": 180940
    },
    {
      "epoch": 21.77496991576414,
      "grad_norm": Infinity,
      "learning_rate": 0.0001784673508270958,
      "loss": 11.047,
      "step": 180950
    },
    {
      "epoch": 21.776173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00017846499246823786,
      "loss": 11.1279,
      "step": 180960
    },
    {
      "epoch": 21.777376654632974,
      "grad_norm": Infinity,
      "learning_rate": 0.00017846263399582166,
      "loss": 11.1287,
      "step": 180970
    },
    {
      "epoch": 21.778580024067388,
      "grad_norm": Infinity,
      "learning_rate": 0.00017846027540985068,
      "loss": 11.0994,
      "step": 180980
    },
    {
      "epoch": 21.779783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.0001784579167103283,
      "loss": 11.031,
      "step": 180990
    },
    {
      "epoch": 21.780986762936223,
      "grad_norm": Infinity,
      "learning_rate": 0.00017845555789725796,
      "loss": 11.1557,
      "step": 181000
    },
    {
      "epoch": 21.782190132370637,
      "grad_norm": Infinity,
      "learning_rate": 0.00017845319897064304,
      "loss": 11.1921,
      "step": 181010
    },
    {
      "epoch": 21.783393501805055,
      "grad_norm": Infinity,
      "learning_rate": 0.00017845083993048698,
      "loss": 11.1343,
      "step": 181020
    },
    {
      "epoch": 21.784596871239472,
      "grad_norm": Infinity,
      "learning_rate": 0.0001784484807767932,
      "loss": 11.2043,
      "step": 181030
    },
    {
      "epoch": 21.785800240673886,
      "grad_norm": Infinity,
      "learning_rate": 0.0001784461215095651,
      "loss": 11.2302,
      "step": 181040
    },
    {
      "epoch": 21.787003610108304,
      "grad_norm": Infinity,
      "learning_rate": 0.00017844376212880608,
      "loss": 11.1428,
      "step": 181050
    },
    {
      "epoch": 21.78820697954272,
      "grad_norm": Infinity,
      "learning_rate": 0.00017844140263451958,
      "loss": 11.1244,
      "step": 181060
    },
    {
      "epoch": 21.789410348977135,
      "grad_norm": Infinity,
      "learning_rate": 0.00017843904302670898,
      "loss": 11.1617,
      "step": 181070
    },
    {
      "epoch": 21.790613718411553,
      "grad_norm": Infinity,
      "learning_rate": 0.00017843668330537774,
      "loss": 11.1106,
      "step": 181080
    },
    {
      "epoch": 21.79181708784597,
      "grad_norm": Infinity,
      "learning_rate": 0.00017843432347052924,
      "loss": 11.2189,
      "step": 181090
    },
    {
      "epoch": 21.793020457280385,
      "grad_norm": Infinity,
      "learning_rate": 0.00017843196352216695,
      "loss": 11.0325,
      "step": 181100
    },
    {
      "epoch": 21.794223826714802,
      "grad_norm": Infinity,
      "learning_rate": 0.00017842960346029418,
      "loss": 11.2234,
      "step": 181110
    },
    {
      "epoch": 21.79542719614922,
      "grad_norm": Infinity,
      "learning_rate": 0.00017842724328491448,
      "loss": 11.1873,
      "step": 181120
    },
    {
      "epoch": 21.796630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00017842488299603116,
      "loss": 11.0861,
      "step": 181130
    },
    {
      "epoch": 21.79783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.00017842252259364766,
      "loss": 11.0352,
      "step": 181140
    },
    {
      "epoch": 21.799037304452465,
      "grad_norm": Infinity,
      "learning_rate": 0.00017842016207776743,
      "loss": 11.0851,
      "step": 181150
    },
    {
      "epoch": 21.800240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.00017841780144839384,
      "loss": 11.0293,
      "step": 181160
    },
    {
      "epoch": 21.8014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017841544070553033,
      "loss": 11.1587,
      "step": 181170
    },
    {
      "epoch": 21.802647412755714,
      "grad_norm": Infinity,
      "learning_rate": 0.00017841307984918032,
      "loss": 11.1068,
      "step": 181180
    },
    {
      "epoch": 21.803850782190132,
      "grad_norm": Infinity,
      "learning_rate": 0.00017841071887934727,
      "loss": 11.182,
      "step": 181190
    },
    {
      "epoch": 21.80505415162455,
      "grad_norm": Infinity,
      "learning_rate": 0.0001784083577960345,
      "loss": 11.1532,
      "step": 181200
    },
    {
      "epoch": 21.806257521058964,
      "grad_norm": Infinity,
      "learning_rate": 0.00017840599659924548,
      "loss": 11.0924,
      "step": 181210
    },
    {
      "epoch": 21.80746089049338,
      "grad_norm": Infinity,
      "learning_rate": 0.00017840363528898366,
      "loss": 11.1952,
      "step": 181220
    },
    {
      "epoch": 21.8086642599278,
      "grad_norm": Infinity,
      "learning_rate": 0.00017840127386525238,
      "loss": 11.2437,
      "step": 181230
    },
    {
      "epoch": 21.809867629362213,
      "grad_norm": Infinity,
      "learning_rate": 0.0001783989123280551,
      "loss": 11.2601,
      "step": 181240
    },
    {
      "epoch": 21.81107099879663,
      "grad_norm": Infinity,
      "learning_rate": 0.0001783965506773953,
      "loss": 11.1212,
      "step": 181250
    },
    {
      "epoch": 21.812274368231048,
      "grad_norm": Infinity,
      "learning_rate": 0.00017839418891327626,
      "loss": 11.1074,
      "step": 181260
    },
    {
      "epoch": 21.813477737665462,
      "grad_norm": Infinity,
      "learning_rate": 0.00017839182703570153,
      "loss": 11.2556,
      "step": 181270
    },
    {
      "epoch": 21.81468110709988,
      "grad_norm": Infinity,
      "learning_rate": 0.00017838946504467444,
      "loss": 11.0453,
      "step": 181280
    },
    {
      "epoch": 21.815884476534297,
      "grad_norm": Infinity,
      "learning_rate": 0.00017838710294019845,
      "loss": 11.252,
      "step": 181290
    },
    {
      "epoch": 21.81708784596871,
      "grad_norm": Infinity,
      "learning_rate": 0.00017838474072227696,
      "loss": 11.1918,
      "step": 181300
    },
    {
      "epoch": 21.81829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.0001783823783909134,
      "loss": 11.1111,
      "step": 181310
    },
    {
      "epoch": 21.819494584837546,
      "grad_norm": Infinity,
      "learning_rate": 0.00017838001594611122,
      "loss": 10.9816,
      "step": 181320
    },
    {
      "epoch": 21.82069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00017837765338787377,
      "loss": 11.1914,
      "step": 181330
    },
    {
      "epoch": 21.821901323706378,
      "grad_norm": Infinity,
      "learning_rate": 0.0001783752907162045,
      "loss": 11.1404,
      "step": 181340
    },
    {
      "epoch": 21.823104693140795,
      "grad_norm": Infinity,
      "learning_rate": 0.00017837292793110686,
      "loss": 11.2338,
      "step": 181350
    },
    {
      "epoch": 21.82430806257521,
      "grad_norm": Infinity,
      "learning_rate": 0.00017837056503258424,
      "loss": 11.1739,
      "step": 181360
    },
    {
      "epoch": 21.825511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00017836820202064003,
      "loss": 11.0121,
      "step": 181370
    },
    {
      "epoch": 21.826714801444044,
      "grad_norm": Infinity,
      "learning_rate": 0.00017836583889527772,
      "loss": 11.0616,
      "step": 181380
    },
    {
      "epoch": 21.82791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.0001783634756565007,
      "loss": 11.1607,
      "step": 181390
    },
    {
      "epoch": 21.829121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00017836111230431234,
      "loss": 11.1626,
      "step": 181400
    },
    {
      "epoch": 21.830324909747294,
      "grad_norm": Infinity,
      "learning_rate": 0.00017835874883871612,
      "loss": 11.2276,
      "step": 181410
    },
    {
      "epoch": 21.831528279181708,
      "grad_norm": Infinity,
      "learning_rate": 0.00017835638525971547,
      "loss": 11.1743,
      "step": 181420
    },
    {
      "epoch": 21.832731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.00017835402156731375,
      "loss": 11.1401,
      "step": 181430
    },
    {
      "epoch": 21.833935018050543,
      "grad_norm": Infinity,
      "learning_rate": 0.00017835165776151443,
      "loss": 11.1726,
      "step": 181440
    },
    {
      "epoch": 21.835138387484957,
      "grad_norm": Infinity,
      "learning_rate": 0.00017834929384232093,
      "loss": 11.1171,
      "step": 181450
    },
    {
      "epoch": 21.836341756919374,
      "grad_norm": Infinity,
      "learning_rate": 0.00017834692980973665,
      "loss": 11.1887,
      "step": 181460
    },
    {
      "epoch": 21.837545126353792,
      "grad_norm": Infinity,
      "learning_rate": 0.000178344565663765,
      "loss": 11.1097,
      "step": 181470
    },
    {
      "epoch": 21.838748495788206,
      "grad_norm": Infinity,
      "learning_rate": 0.00017834220140440946,
      "loss": 11.1601,
      "step": 181480
    },
    {
      "epoch": 21.839951865222623,
      "grad_norm": Infinity,
      "learning_rate": 0.00017833983703167337,
      "loss": 11.1485,
      "step": 181490
    },
    {
      "epoch": 21.84115523465704,
      "grad_norm": Infinity,
      "learning_rate": 0.00017833747254556023,
      "loss": 11.1811,
      "step": 181500
    },
    {
      "epoch": 21.842358604091455,
      "grad_norm": Infinity,
      "learning_rate": 0.00017833510794607342,
      "loss": 11.1527,
      "step": 181510
    },
    {
      "epoch": 21.843561973525873,
      "grad_norm": Infinity,
      "learning_rate": 0.00017833274323321633,
      "loss": 11.104,
      "step": 181520
    },
    {
      "epoch": 21.84476534296029,
      "grad_norm": Infinity,
      "learning_rate": 0.00017833037840699246,
      "loss": 11.1038,
      "step": 181530
    },
    {
      "epoch": 21.845968712394704,
      "grad_norm": Infinity,
      "learning_rate": 0.0001783280134674052,
      "loss": 11.2436,
      "step": 181540
    },
    {
      "epoch": 21.84717208182912,
      "grad_norm": Infinity,
      "learning_rate": 0.00017832564841445793,
      "loss": 11.0612,
      "step": 181550
    },
    {
      "epoch": 21.84837545126354,
      "grad_norm": Infinity,
      "learning_rate": 0.00017832328324815416,
      "loss": 11.1391,
      "step": 181560
    },
    {
      "epoch": 21.849578820697953,
      "grad_norm": Infinity,
      "learning_rate": 0.0001783209179684972,
      "loss": 11.0787,
      "step": 181570
    },
    {
      "epoch": 21.85078219013237,
      "grad_norm": Infinity,
      "learning_rate": 0.00017831855257549058,
      "loss": 11.18,
      "step": 181580
    },
    {
      "epoch": 21.85198555956679,
      "grad_norm": Infinity,
      "learning_rate": 0.00017831618706913766,
      "loss": 11.065,
      "step": 181590
    },
    {
      "epoch": 21.853188929001202,
      "grad_norm": Infinity,
      "learning_rate": 0.0001783138214494419,
      "loss": 11.2187,
      "step": 181600
    },
    {
      "epoch": 21.85439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00017831145571640668,
      "loss": 11.1288,
      "step": 181610
    },
    {
      "epoch": 21.855595667870038,
      "grad_norm": Infinity,
      "learning_rate": 0.00017830908987003546,
      "loss": 11.0917,
      "step": 181620
    },
    {
      "epoch": 21.85679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00017830672391033165,
      "loss": 11.1151,
      "step": 181630
    },
    {
      "epoch": 21.85800240673887,
      "grad_norm": Infinity,
      "learning_rate": 0.00017830435783729867,
      "loss": 11.1632,
      "step": 181640
    },
    {
      "epoch": 21.859205776173287,
      "grad_norm": Infinity,
      "learning_rate": 0.00017830199165093998,
      "loss": 11.157,
      "step": 181650
    },
    {
      "epoch": 21.8604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00017829962535125896,
      "loss": 11.1263,
      "step": 181660
    },
    {
      "epoch": 21.86161251504212,
      "grad_norm": Infinity,
      "learning_rate": 0.00017829725893825905,
      "loss": 11.1513,
      "step": 181670
    },
    {
      "epoch": 21.862815884476536,
      "grad_norm": Infinity,
      "learning_rate": 0.00017829489241194367,
      "loss": 11.1217,
      "step": 181680
    },
    {
      "epoch": 21.86401925391095,
      "grad_norm": Infinity,
      "learning_rate": 0.00017829252577231626,
      "loss": 11.1202,
      "step": 181690
    },
    {
      "epoch": 21.865222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.0001782901590193802,
      "loss": 11.1383,
      "step": 181700
    },
    {
      "epoch": 21.866425992779785,
      "grad_norm": Infinity,
      "learning_rate": 0.000178287792153139,
      "loss": 11.2577,
      "step": 181710
    },
    {
      "epoch": 21.8676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00017828542517359602,
      "loss": 11.1333,
      "step": 181720
    },
    {
      "epoch": 21.868832731648617,
      "grad_norm": Infinity,
      "learning_rate": 0.00017828305808075467,
      "loss": 11.1896,
      "step": 181730
    },
    {
      "epoch": 21.870036101083034,
      "grad_norm": Infinity,
      "learning_rate": 0.00017828069087461845,
      "loss": 11.1563,
      "step": 181740
    },
    {
      "epoch": 21.871239470517448,
      "grad_norm": Infinity,
      "learning_rate": 0.00017827832355519073,
      "loss": 11.1597,
      "step": 181750
    },
    {
      "epoch": 21.872442839951866,
      "grad_norm": Infinity,
      "learning_rate": 0.00017827595612247493,
      "loss": 11.1044,
      "step": 181760
    },
    {
      "epoch": 21.87364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.00017827358857647454,
      "loss": 11.1056,
      "step": 181770
    },
    {
      "epoch": 21.874849578820697,
      "grad_norm": Infinity,
      "learning_rate": 0.0001782712209171929,
      "loss": 11.2507,
      "step": 181780
    },
    {
      "epoch": 21.876052948255115,
      "grad_norm": Infinity,
      "learning_rate": 0.00017826885314463348,
      "loss": 11.1047,
      "step": 181790
    },
    {
      "epoch": 21.87725631768953,
      "grad_norm": Infinity,
      "learning_rate": 0.00017826648525879973,
      "loss": 11.06,
      "step": 181800
    },
    {
      "epoch": 21.878459687123947,
      "grad_norm": Infinity,
      "learning_rate": 0.00017826411725969506,
      "loss": 11.0438,
      "step": 181810
    },
    {
      "epoch": 21.879663056558364,
      "grad_norm": Infinity,
      "learning_rate": 0.00017826174914732285,
      "loss": 11.1146,
      "step": 181820
    },
    {
      "epoch": 21.880866425992778,
      "grad_norm": Infinity,
      "learning_rate": 0.00017825938092168658,
      "loss": 11.1669,
      "step": 181830
    },
    {
      "epoch": 21.882069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00017825701258278968,
      "loss": 11.0667,
      "step": 181840
    },
    {
      "epoch": 21.883273164861613,
      "grad_norm": Infinity,
      "learning_rate": 0.00017825464413063558,
      "loss": 11.0749,
      "step": 181850
    },
    {
      "epoch": 21.884476534296027,
      "grad_norm": Infinity,
      "learning_rate": 0.00017825227556522765,
      "loss": 11.1729,
      "step": 181860
    },
    {
      "epoch": 21.885679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00017824990688656937,
      "loss": 11.0918,
      "step": 181870
    },
    {
      "epoch": 21.886883273164862,
      "grad_norm": Infinity,
      "learning_rate": 0.00017824753809466418,
      "loss": 11.0952,
      "step": 181880
    },
    {
      "epoch": 21.888086642599276,
      "grad_norm": Infinity,
      "learning_rate": 0.00017824516918951545,
      "loss": 11.2068,
      "step": 181890
    },
    {
      "epoch": 21.889290012033694,
      "grad_norm": Infinity,
      "learning_rate": 0.00017824280017112666,
      "loss": 11.0649,
      "step": 181900
    },
    {
      "epoch": 21.89049338146811,
      "grad_norm": Infinity,
      "learning_rate": 0.0001782404310395012,
      "loss": 11.036,
      "step": 181910
    },
    {
      "epoch": 21.891696750902526,
      "grad_norm": Infinity,
      "learning_rate": 0.00017823806179464253,
      "loss": 11.2186,
      "step": 181920
    },
    {
      "epoch": 21.892900120336943,
      "grad_norm": Infinity,
      "learning_rate": 0.00017823569243655406,
      "loss": 11.0482,
      "step": 181930
    },
    {
      "epoch": 21.89410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017823332296523924,
      "loss": 11.2153,
      "step": 181940
    },
    {
      "epoch": 21.895306859205775,
      "grad_norm": Infinity,
      "learning_rate": 0.00017823095338070149,
      "loss": 11.073,
      "step": 181950
    },
    {
      "epoch": 21.896510228640192,
      "grad_norm": Infinity,
      "learning_rate": 0.00017822858368294424,
      "loss": 11.1148,
      "step": 181960
    },
    {
      "epoch": 21.89771359807461,
      "grad_norm": Infinity,
      "learning_rate": 0.00017822621387197088,
      "loss": 11.1757,
      "step": 181970
    },
    {
      "epoch": 21.898916967509024,
      "grad_norm": Infinity,
      "learning_rate": 0.00017822384394778492,
      "loss": 11.1131,
      "step": 181980
    },
    {
      "epoch": 21.90012033694344,
      "grad_norm": Infinity,
      "learning_rate": 0.00017822147391038975,
      "loss": 11.0978,
      "step": 181990
    },
    {
      "epoch": 21.90132370637786,
      "grad_norm": Infinity,
      "learning_rate": 0.00017821910375978876,
      "loss": 11.1209,
      "step": 182000
    },
    {
      "epoch": 21.902527075812273,
      "grad_norm": Infinity,
      "learning_rate": 0.00017821673349598542,
      "loss": 11.1225,
      "step": 182010
    },
    {
      "epoch": 21.90373044524669,
      "grad_norm": Infinity,
      "learning_rate": 0.00017821436311898315,
      "loss": 10.9861,
      "step": 182020
    },
    {
      "epoch": 21.904933814681108,
      "grad_norm": Infinity,
      "learning_rate": 0.0001782119926287854,
      "loss": 11.2235,
      "step": 182030
    },
    {
      "epoch": 21.906137184115522,
      "grad_norm": Infinity,
      "learning_rate": 0.0001782096220253956,
      "loss": 11.1922,
      "step": 182040
    },
    {
      "epoch": 21.90734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.00017820725130881715,
      "loss": 11.2977,
      "step": 182050
    },
    {
      "epoch": 21.908543922984357,
      "grad_norm": Infinity,
      "learning_rate": 0.0001782048804790535,
      "loss": 11.1831,
      "step": 182060
    },
    {
      "epoch": 21.90974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00017820250953610808,
      "loss": 11.1861,
      "step": 182070
    },
    {
      "epoch": 21.91095066185319,
      "grad_norm": Infinity,
      "learning_rate": 0.0001782001384799843,
      "loss": 11.1354,
      "step": 182080
    },
    {
      "epoch": 21.912154031287606,
      "grad_norm": Infinity,
      "learning_rate": 0.00017819776731068564,
      "loss": 11.1224,
      "step": 182090
    },
    {
      "epoch": 21.91335740072202,
      "grad_norm": Infinity,
      "learning_rate": 0.00017819539602821548,
      "loss": 11.2226,
      "step": 182100
    },
    {
      "epoch": 21.914560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00017819302463257733,
      "loss": 11.2407,
      "step": 182110
    },
    {
      "epoch": 21.915764139590856,
      "grad_norm": Infinity,
      "learning_rate": 0.00017819065312377452,
      "loss": 11.1271,
      "step": 182120
    },
    {
      "epoch": 21.91696750902527,
      "grad_norm": Infinity,
      "learning_rate": 0.00017818828150181055,
      "loss": 11.1881,
      "step": 182130
    },
    {
      "epoch": 21.918170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00017818590976668883,
      "loss": 11.2033,
      "step": 182140
    },
    {
      "epoch": 21.919374247894105,
      "grad_norm": Infinity,
      "learning_rate": 0.00017818353791841277,
      "loss": 11.1064,
      "step": 182150
    },
    {
      "epoch": 21.92057761732852,
      "grad_norm": Infinity,
      "learning_rate": 0.00017818116595698586,
      "loss": 11.1049,
      "step": 182160
    },
    {
      "epoch": 21.921780986762936,
      "grad_norm": Infinity,
      "learning_rate": 0.0001781787938824115,
      "loss": 10.9041,
      "step": 182170
    },
    {
      "epoch": 21.922984356197354,
      "grad_norm": Infinity,
      "learning_rate": 0.0001781764216946931,
      "loss": 11.1451,
      "step": 182180
    },
    {
      "epoch": 21.924187725631768,
      "grad_norm": Infinity,
      "learning_rate": 0.00017817404939383415,
      "loss": 11.1456,
      "step": 182190
    },
    {
      "epoch": 21.925391095066185,
      "grad_norm": Infinity,
      "learning_rate": 0.00017817167697983802,
      "loss": 11.0745,
      "step": 182200
    },
    {
      "epoch": 21.926594464500603,
      "grad_norm": Infinity,
      "learning_rate": 0.00017816930445270818,
      "loss": 11.1366,
      "step": 182210
    },
    {
      "epoch": 21.927797833935017,
      "grad_norm": Infinity,
      "learning_rate": 0.00017816693181244806,
      "loss": 11.0964,
      "step": 182220
    },
    {
      "epoch": 21.929001203369435,
      "grad_norm": Infinity,
      "learning_rate": 0.00017816455905906109,
      "loss": 11.2673,
      "step": 182230
    },
    {
      "epoch": 21.930204572803852,
      "grad_norm": Infinity,
      "learning_rate": 0.00017816218619255068,
      "loss": 11.0652,
      "step": 182240
    },
    {
      "epoch": 21.931407942238266,
      "grad_norm": Infinity,
      "learning_rate": 0.00017815981321292034,
      "loss": 11.153,
      "step": 182250
    },
    {
      "epoch": 21.932611311672684,
      "grad_norm": Infinity,
      "learning_rate": 0.0001781574401201734,
      "loss": 11.1923,
      "step": 182260
    },
    {
      "epoch": 21.9338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.0001781550669143134,
      "loss": 11.096,
      "step": 182270
    },
    {
      "epoch": 21.935018050541515,
      "grad_norm": Infinity,
      "learning_rate": 0.0001781526935953437,
      "loss": 11.1599,
      "step": 182280
    },
    {
      "epoch": 21.936221419975933,
      "grad_norm": Infinity,
      "learning_rate": 0.00017815032016326777,
      "loss": 11.1602,
      "step": 182290
    },
    {
      "epoch": 21.93742478941035,
      "grad_norm": Infinity,
      "learning_rate": 0.000178147946618089,
      "loss": 11.155,
      "step": 182300
    },
    {
      "epoch": 21.938628158844764,
      "grad_norm": Infinity,
      "learning_rate": 0.00017814557295981086,
      "loss": 11.0521,
      "step": 182310
    },
    {
      "epoch": 21.939831528279182,
      "grad_norm": Infinity,
      "learning_rate": 0.00017814319918843681,
      "loss": 11.3187,
      "step": 182320
    },
    {
      "epoch": 21.9410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017814082530397024,
      "loss": 11.224,
      "step": 182330
    },
    {
      "epoch": 21.942238267148014,
      "grad_norm": Infinity,
      "learning_rate": 0.0001781384513064146,
      "loss": 11.2031,
      "step": 182340
    },
    {
      "epoch": 21.94344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00017813607719577334,
      "loss": 11.1313,
      "step": 182350
    },
    {
      "epoch": 21.94464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00017813370297204987,
      "loss": 11.1111,
      "step": 182360
    },
    {
      "epoch": 21.945848375451263,
      "grad_norm": Infinity,
      "learning_rate": 0.0001781313286352477,
      "loss": 11.0944,
      "step": 182370
    },
    {
      "epoch": 21.94705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017812895418537012,
      "loss": 11.0591,
      "step": 182380
    },
    {
      "epoch": 21.948255114320098,
      "grad_norm": Infinity,
      "learning_rate": 0.0001781265796224207,
      "loss": 11.1925,
      "step": 182390
    },
    {
      "epoch": 21.949458483754512,
      "grad_norm": Infinity,
      "learning_rate": 0.00017812420494640283,
      "loss": 11.1055,
      "step": 182400
    },
    {
      "epoch": 21.95066185318893,
      "grad_norm": Infinity,
      "learning_rate": 0.00017812183015731995,
      "loss": 11.1512,
      "step": 182410
    },
    {
      "epoch": 21.951865222623347,
      "grad_norm": Infinity,
      "learning_rate": 0.00017811945525517548,
      "loss": 11.1603,
      "step": 182420
    },
    {
      "epoch": 21.95306859205776,
      "grad_norm": Infinity,
      "learning_rate": 0.00017811708023997289,
      "loss": 11.213,
      "step": 182430
    },
    {
      "epoch": 21.95427196149218,
      "grad_norm": Infinity,
      "learning_rate": 0.00017811470511171557,
      "loss": 11.0421,
      "step": 182440
    },
    {
      "epoch": 21.955475330926596,
      "grad_norm": Infinity,
      "learning_rate": 0.000178112329870407,
      "loss": 11.1248,
      "step": 182450
    },
    {
      "epoch": 21.95667870036101,
      "grad_norm": Infinity,
      "learning_rate": 0.00017810995451605062,
      "loss": 11.1249,
      "step": 182460
    },
    {
      "epoch": 21.957882069795428,
      "grad_norm": Infinity,
      "learning_rate": 0.00017810757904864984,
      "loss": 11.1359,
      "step": 182470
    },
    {
      "epoch": 21.959085439229845,
      "grad_norm": Infinity,
      "learning_rate": 0.00017810520346820812,
      "loss": 11.1427,
      "step": 182480
    },
    {
      "epoch": 21.96028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.00017810282777472883,
      "loss": 11.1613,
      "step": 182490
    },
    {
      "epoch": 21.961492178098677,
      "grad_norm": Infinity,
      "learning_rate": 0.00017810045196821553,
      "loss": 11.0584,
      "step": 182500
    },
    {
      "epoch": 21.96269554753309,
      "grad_norm": Infinity,
      "learning_rate": 0.00017809807604867158,
      "loss": 11.183,
      "step": 182510
    },
    {
      "epoch": 21.96389891696751,
      "grad_norm": Infinity,
      "learning_rate": 0.00017809570001610042,
      "loss": 11.0937,
      "step": 182520
    },
    {
      "epoch": 21.965102286401926,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780933238705055,
      "loss": 11.1415,
      "step": 182530
    },
    {
      "epoch": 21.96630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00017809094761189024,
      "loss": 11.06,
      "step": 182540
    },
    {
      "epoch": 21.967509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.00017808857124025816,
      "loss": 11.1841,
      "step": 182550
    },
    {
      "epoch": 21.968712394705175,
      "grad_norm": Infinity,
      "learning_rate": 0.00017808619475561259,
      "loss": 11.1151,
      "step": 182560
    },
    {
      "epoch": 21.96991576413959,
      "grad_norm": Infinity,
      "learning_rate": 0.00017808381815795701,
      "loss": 11.069,
      "step": 182570
    },
    {
      "epoch": 21.971119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780814414472949,
      "loss": 11.2328,
      "step": 182580
    },
    {
      "epoch": 21.972322503008424,
      "grad_norm": Infinity,
      "learning_rate": 0.00017807906462362968,
      "loss": 11.0846,
      "step": 182590
    },
    {
      "epoch": 21.97352587244284,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780766876869647,
      "loss": 11.0604,
      "step": 182600
    },
    {
      "epoch": 21.974729241877256,
      "grad_norm": Infinity,
      "learning_rate": 0.00017807431063730355,
      "loss": 11.1187,
      "step": 182610
    },
    {
      "epoch": 21.975932611311674,
      "grad_norm": Infinity,
      "learning_rate": 0.00017807193347464957,
      "loss": 11.1575,
      "step": 182620
    },
    {
      "epoch": 21.977135980746088,
      "grad_norm": Infinity,
      "learning_rate": 0.00017806955619900622,
      "loss": 11.2411,
      "step": 182630
    },
    {
      "epoch": 21.978339350180505,
      "grad_norm": Infinity,
      "learning_rate": 0.00017806717881037695,
      "loss": 11.1222,
      "step": 182640
    },
    {
      "epoch": 21.979542719614923,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780648013087652,
      "loss": 11.1311,
      "step": 182650
    },
    {
      "epoch": 21.980746089049337,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780624236941744,
      "loss": 11.2069,
      "step": 182660
    },
    {
      "epoch": 21.981949458483754,
      "grad_norm": Infinity,
      "learning_rate": 0.00017806004596660803,
      "loss": 11.0799,
      "step": 182670
    },
    {
      "epoch": 21.983152827918172,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780576681260695,
      "loss": 11.1148,
      "step": 182680
    },
    {
      "epoch": 21.984356197352586,
      "grad_norm": Infinity,
      "learning_rate": 0.00017805529017256224,
      "loss": 11.0971,
      "step": 182690
    },
    {
      "epoch": 21.985559566787003,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780529121060897,
      "loss": 11.1542,
      "step": 182700
    },
    {
      "epoch": 21.98676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780505339266553,
      "loss": 11.1696,
      "step": 182710
    },
    {
      "epoch": 21.987966305655835,
      "grad_norm": Infinity,
      "learning_rate": 0.00017804815563426254,
      "loss": 11.2658,
      "step": 182720
    },
    {
      "epoch": 21.989169675090253,
      "grad_norm": Infinity,
      "learning_rate": 0.00017804577722891482,
      "loss": 11.1954,
      "step": 182730
    },
    {
      "epoch": 21.99037304452467,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780433987106156,
      "loss": 11.175,
      "step": 182740
    },
    {
      "epoch": 21.991576413959084,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780410200793683,
      "loss": 11.0907,
      "step": 182750
    },
    {
      "epoch": 21.9927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780386413351764,
      "loss": 11.2459,
      "step": 182760
    },
    {
      "epoch": 21.99398315282792,
      "grad_norm": Infinity,
      "learning_rate": 0.00017803626247804328,
      "loss": 11.0128,
      "step": 182770
    },
    {
      "epoch": 21.995186522262333,
      "grad_norm": Infinity,
      "learning_rate": 0.00017803388350797245,
      "loss": 11.1205,
      "step": 182780
    },
    {
      "epoch": 21.99638989169675,
      "grad_norm": Infinity,
      "learning_rate": 0.00017803150442496733,
      "loss": 11.1914,
      "step": 182790
    },
    {
      "epoch": 21.99759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780291252290313,
      "loss": 11.149,
      "step": 182800
    },
    {
      "epoch": 21.998796630565582,
      "grad_norm": Infinity,
      "learning_rate": 0.00017802674592016794,
      "loss": 11.1617,
      "step": 182810
    },
    {
      "epoch": 22.0,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780243664983806,
      "loss": 11.1725,
      "step": 182820
    },
    {
      "epoch": 22.0,
      "eval_loss": 11.139603614807129,
      "eval_runtime": 120.6846,
      "eval_samples_per_second": 61.209,
      "eval_steps_per_second": 7.656,
      "step": 182820
    },
    {
      "epoch": 22.001203369434418,
      "grad_norm": Infinity,
      "learning_rate": 0.0001780219869636727,
      "loss": 11.1769,
      "step": 182830
    },
    {
      "epoch": 22.00240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.00017801960731604773,
      "loss": 11.1089,
      "step": 182840
    },
    {
      "epoch": 22.00361010830325,
      "grad_norm": Infinity,
      "learning_rate": 0.00017801722755550917,
      "loss": 11.2128,
      "step": 182850
    },
    {
      "epoch": 22.004813477737667,
      "grad_norm": Infinity,
      "learning_rate": 0.00017801484768206036,
      "loss": 11.1146,
      "step": 182860
    },
    {
      "epoch": 22.00601684717208,
      "grad_norm": Infinity,
      "learning_rate": 0.00017801246769570485,
      "loss": 11.0829,
      "step": 182870
    },
    {
      "epoch": 22.0072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.000178010087596446,
      "loss": 11.1403,
      "step": 182880
    },
    {
      "epoch": 22.008423586040916,
      "grad_norm": Infinity,
      "learning_rate": 0.00017800770738428731,
      "loss": 11.0535,
      "step": 182890
    },
    {
      "epoch": 22.00962695547533,
      "grad_norm": Infinity,
      "learning_rate": 0.00017800532705923222,
      "loss": 11.0978,
      "step": 182900
    },
    {
      "epoch": 22.010830324909747,
      "grad_norm": Infinity,
      "learning_rate": 0.00017800294662128417,
      "loss": 11.0584,
      "step": 182910
    },
    {
      "epoch": 22.012033694344165,
      "grad_norm": Infinity,
      "learning_rate": 0.00017800056607044658,
      "loss": 11.1655,
      "step": 182920
    },
    {
      "epoch": 22.01323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00017799818540672292,
      "loss": 11.1153,
      "step": 182930
    },
    {
      "epoch": 22.014440433212997,
      "grad_norm": Infinity,
      "learning_rate": 0.00017799580463011662,
      "loss": 11.016,
      "step": 182940
    },
    {
      "epoch": 22.015643802647414,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779934237406311,
      "loss": 11.2378,
      "step": 182950
    },
    {
      "epoch": 22.016847172081828,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779910427382699,
      "loss": 11.1098,
      "step": 182960
    },
    {
      "epoch": 22.018050541516246,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779886616230364,
      "loss": 11.0959,
      "step": 182970
    },
    {
      "epoch": 22.019253910950663,
      "grad_norm": Infinity,
      "learning_rate": 0.000177986280394934,
      "loss": 11.143,
      "step": 182980
    },
    {
      "epoch": 22.020457280385077,
      "grad_norm": Infinity,
      "learning_rate": 0.00017798389905396625,
      "loss": 11.072,
      "step": 182990
    },
    {
      "epoch": 22.021660649819495,
      "grad_norm": Infinity,
      "learning_rate": 0.00017798151760013651,
      "loss": 11.2087,
      "step": 183000
    },
    {
      "epoch": 22.022864019253912,
      "grad_norm": Infinity,
      "learning_rate": 0.00017797913603344832,
      "loss": 11.1155,
      "step": 183010
    },
    {
      "epoch": 22.024067388688326,
      "grad_norm": Infinity,
      "learning_rate": 0.000177976754353905,
      "loss": 11.1776,
      "step": 183020
    },
    {
      "epoch": 22.025270758122744,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779743725615101,
      "loss": 11.1987,
      "step": 183030
    },
    {
      "epoch": 22.02647412755716,
      "grad_norm": Infinity,
      "learning_rate": 0.00017797199065626703,
      "loss": 11.0672,
      "step": 183040
    },
    {
      "epoch": 22.027677496991576,
      "grad_norm": Infinity,
      "learning_rate": 0.00017796960863817923,
      "loss": 11.1907,
      "step": 183050
    },
    {
      "epoch": 22.028880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.00017796722650725014,
      "loss": 11.1238,
      "step": 183060
    },
    {
      "epoch": 22.03008423586041,
      "grad_norm": Infinity,
      "learning_rate": 0.00017796484426348325,
      "loss": 11.1398,
      "step": 183070
    },
    {
      "epoch": 22.031287605294825,
      "grad_norm": Infinity,
      "learning_rate": 0.000177962461906882,
      "loss": 11.2278,
      "step": 183080
    },
    {
      "epoch": 22.032490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00017796007943744977,
      "loss": 11.1665,
      "step": 183090
    },
    {
      "epoch": 22.03369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779576968551901,
      "loss": 11.1412,
      "step": 183100
    },
    {
      "epoch": 22.034897713598074,
      "grad_norm": Infinity,
      "learning_rate": 0.00017795531416010636,
      "loss": 11.0917,
      "step": 183110
    },
    {
      "epoch": 22.03610108303249,
      "grad_norm": Infinity,
      "learning_rate": 0.00017795293135220206,
      "loss": 11.2043,
      "step": 183120
    },
    {
      "epoch": 22.03730445246691,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779505484314806,
      "loss": 11.1065,
      "step": 183130
    },
    {
      "epoch": 22.038507821901323,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779481653979455,
      "loss": 11.1322,
      "step": 183140
    },
    {
      "epoch": 22.03971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779457822516001,
      "loss": 11.1741,
      "step": 183150
    },
    {
      "epoch": 22.040914560770158,
      "grad_norm": Infinity,
      "learning_rate": 0.00017794339899244793,
      "loss": 11.1394,
      "step": 183160
    },
    {
      "epoch": 22.042117930204572,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779410156204924,
      "loss": 11.1185,
      "step": 183170
    },
    {
      "epoch": 22.04332129963899,
      "grad_norm": Infinity,
      "learning_rate": 0.000177938632135737,
      "loss": 11.1088,
      "step": 183180
    },
    {
      "epoch": 22.044524669073404,
      "grad_norm": Infinity,
      "learning_rate": 0.00017793624853818516,
      "loss": 11.1863,
      "step": 183190
    },
    {
      "epoch": 22.04572803850782,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779338648278403,
      "loss": 11.192,
      "step": 183200
    },
    {
      "epoch": 22.04693140794224,
      "grad_norm": Infinity,
      "learning_rate": 0.00017793148100470594,
      "loss": 11.1201,
      "step": 183210
    },
    {
      "epoch": 22.048134777376653,
      "grad_norm": Infinity,
      "learning_rate": 0.00017792909706878547,
      "loss": 11.3088,
      "step": 183220
    },
    {
      "epoch": 22.04933814681107,
      "grad_norm": Infinity,
      "learning_rate": 0.00017792671302008234,
      "loss": 11.151,
      "step": 183230
    },
    {
      "epoch": 22.050541516245488,
      "grad_norm": Infinity,
      "learning_rate": 0.00017792432885860003,
      "loss": 11.0786,
      "step": 183240
    },
    {
      "epoch": 22.051744885679902,
      "grad_norm": Infinity,
      "learning_rate": 0.00017792194458434197,
      "loss": 11.1415,
      "step": 183250
    },
    {
      "epoch": 22.05294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00017791956019731162,
      "loss": 11.1922,
      "step": 183260
    },
    {
      "epoch": 22.054151624548737,
      "grad_norm": Infinity,
      "learning_rate": 0.00017791717569751243,
      "loss": 11.1008,
      "step": 183270
    },
    {
      "epoch": 22.05535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00017791479108494787,
      "loss": 11.0372,
      "step": 183280
    },
    {
      "epoch": 22.05655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779124063596213,
      "loss": 10.9654,
      "step": 183290
    },
    {
      "epoch": 22.057761732851986,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779100215215363,
      "loss": 11.1651,
      "step": 183300
    },
    {
      "epoch": 22.0589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779076365706963,
      "loss": 11.1447,
      "step": 183310
    },
    {
      "epoch": 22.060168471720818,
      "grad_norm": Infinity,
      "learning_rate": 0.00017790525150710465,
      "loss": 11.2377,
      "step": 183320
    },
    {
      "epoch": 22.061371841155236,
      "grad_norm": Infinity,
      "learning_rate": 0.0001779028663307649,
      "loss": 11.1066,
      "step": 183330
    },
    {
      "epoch": 22.06257521058965,
      "grad_norm": Infinity,
      "learning_rate": 0.00017790048104168043,
      "loss": 11.1419,
      "step": 183340
    },
    {
      "epoch": 22.063778580024067,
      "grad_norm": Infinity,
      "learning_rate": 0.00017789809563985477,
      "loss": 11.2599,
      "step": 183350
    },
    {
      "epoch": 22.064981949458485,
      "grad_norm": Infinity,
      "learning_rate": 0.00017789571012529135,
      "loss": 11.1508,
      "step": 183360
    },
    {
      "epoch": 22.0661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00017789332449799354,
      "loss": 11.1465,
      "step": 183370
    },
    {
      "epoch": 22.067388688327316,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778909387579649,
      "loss": 11.1886,
      "step": 183380
    },
    {
      "epoch": 22.068592057761734,
      "grad_norm": Infinity,
      "learning_rate": 0.00017788855290520883,
      "loss": 11.217,
      "step": 183390
    },
    {
      "epoch": 22.069795427196148,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778861669397288,
      "loss": 11.1068,
      "step": 183400
    },
    {
      "epoch": 22.070998796630565,
      "grad_norm": Infinity,
      "learning_rate": 0.00017788378086152826,
      "loss": 11.1964,
      "step": 183410
    },
    {
      "epoch": 22.072202166064983,
      "grad_norm": Infinity,
      "learning_rate": 0.00017788139467061066,
      "loss": 11.2453,
      "step": 183420
    },
    {
      "epoch": 22.073405535499397,
      "grad_norm": Infinity,
      "learning_rate": 0.00017787900836697945,
      "loss": 11.1021,
      "step": 183430
    },
    {
      "epoch": 22.074608904933815,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778766219506381,
      "loss": 11.1761,
      "step": 183440
    },
    {
      "epoch": 22.075812274368232,
      "grad_norm": Infinity,
      "learning_rate": 0.00017787423542159003,
      "loss": 11.1881,
      "step": 183450
    },
    {
      "epoch": 22.077015643802646,
      "grad_norm": Infinity,
      "learning_rate": 0.00017787184877983872,
      "loss": 11.2353,
      "step": 183460
    },
    {
      "epoch": 22.078219013237064,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778694620253876,
      "loss": 11.1493,
      "step": 183470
    },
    {
      "epoch": 22.07942238267148,
      "grad_norm": Infinity,
      "learning_rate": 0.00017786707515824018,
      "loss": 11.1424,
      "step": 183480
    },
    {
      "epoch": 22.080625752105895,
      "grad_norm": Infinity,
      "learning_rate": 0.00017786468817839986,
      "loss": 11.0966,
      "step": 183490
    },
    {
      "epoch": 22.081829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.00017786230108587012,
      "loss": 11.0919,
      "step": 183500
    },
    {
      "epoch": 22.08303249097473,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778599138806544,
      "loss": 11.1699,
      "step": 183510
    },
    {
      "epoch": 22.084235860409144,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778575265627562,
      "loss": 11.1028,
      "step": 183520
    },
    {
      "epoch": 22.085439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00017785513913217885,
      "loss": 11.1392,
      "step": 183530
    },
    {
      "epoch": 22.08664259927798,
      "grad_norm": Infinity,
      "learning_rate": 0.00017785275158892596,
      "loss": 11.2001,
      "step": 183540
    },
    {
      "epoch": 22.087845968712394,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778503639330009,
      "loss": 11.0794,
      "step": 183550
    },
    {
      "epoch": 22.08904933814681,
      "grad_norm": Infinity,
      "learning_rate": 0.00017784797616440714,
      "loss": 11.14,
      "step": 183560
    },
    {
      "epoch": 22.09025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.00017784558828314815,
      "loss": 11.2234,
      "step": 183570
    },
    {
      "epoch": 22.091456077015643,
      "grad_norm": Infinity,
      "learning_rate": 0.00017784320028922735,
      "loss": 11.0392,
      "step": 183580
    },
    {
      "epoch": 22.09265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00017784081218264825,
      "loss": 11.1166,
      "step": 183590
    },
    {
      "epoch": 22.093862815884478,
      "grad_norm": Infinity,
      "learning_rate": 0.00017783842396341425,
      "loss": 11.0499,
      "step": 183600
    },
    {
      "epoch": 22.095066185318892,
      "grad_norm": Infinity,
      "learning_rate": 0.00017783603563152885,
      "loss": 11.1152,
      "step": 183610
    },
    {
      "epoch": 22.09626955475331,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778336471869955,
      "loss": 11.168,
      "step": 183620
    },
    {
      "epoch": 22.097472924187727,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778312586298176,
      "loss": 11.1864,
      "step": 183630
    },
    {
      "epoch": 22.09867629362214,
      "grad_norm": Infinity,
      "learning_rate": 0.00017782886995999867,
      "loss": 11.1706,
      "step": 183640
    },
    {
      "epoch": 22.09987966305656,
      "grad_norm": Infinity,
      "learning_rate": 0.00017782648117754216,
      "loss": 11.0227,
      "step": 183650
    },
    {
      "epoch": 22.101083032490976,
      "grad_norm": Infinity,
      "learning_rate": 0.00017782409228245155,
      "loss": 11.0737,
      "step": 183660
    },
    {
      "epoch": 22.10228640192539,
      "grad_norm": Infinity,
      "learning_rate": 0.00017782170327473024,
      "loss": 11.1925,
      "step": 183670
    },
    {
      "epoch": 22.103489771359808,
      "grad_norm": Infinity,
      "learning_rate": 0.00017781931415438172,
      "loss": 11.0555,
      "step": 183680
    },
    {
      "epoch": 22.104693140794225,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778169249214094,
      "loss": 11.2993,
      "step": 183690
    },
    {
      "epoch": 22.10589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778145355758168,
      "loss": 11.1392,
      "step": 183700
    },
    {
      "epoch": 22.107099879663057,
      "grad_norm": Infinity,
      "learning_rate": 0.00017781214611760734,
      "loss": 11.1894,
      "step": 183710
    },
    {
      "epoch": 22.108303249097474,
      "grad_norm": Infinity,
      "learning_rate": 0.00017780975654678452,
      "loss": 11.152,
      "step": 183720
    },
    {
      "epoch": 22.10950661853189,
      "grad_norm": Infinity,
      "learning_rate": 0.00017780736686335178,
      "loss": 11.0664,
      "step": 183730
    },
    {
      "epoch": 22.110709987966306,
      "grad_norm": Infinity,
      "learning_rate": 0.00017780497706731255,
      "loss": 11.1938,
      "step": 183740
    },
    {
      "epoch": 22.111913357400724,
      "grad_norm": Infinity,
      "learning_rate": 0.0001778025871586703,
      "loss": 11.1373,
      "step": 183750
    },
    {
      "epoch": 22.113116726835138,
      "grad_norm": Infinity,
      "learning_rate": 0.00017780019713742852,
      "loss": 11.1037,
      "step": 183760
    },
    {
      "epoch": 22.114320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.00017779780700359063,
      "loss": 11.2328,
      "step": 183770
    },
    {
      "epoch": 22.115523465703973,
      "grad_norm": Infinity,
      "learning_rate": 0.00017779541675716013,
      "loss": 11.1739,
      "step": 183780
    },
    {
      "epoch": 22.116726835138387,
      "grad_norm": Infinity,
      "learning_rate": 0.00017779302639814044,
      "loss": 11.2937,
      "step": 183790
    },
    {
      "epoch": 22.117930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.00017779063592653503,
      "loss": 11.1634,
      "step": 183800
    },
    {
      "epoch": 22.119133574007222,
      "grad_norm": Infinity,
      "learning_rate": 0.00017778824534234734,
      "loss": 11.1178,
      "step": 183810
    },
    {
      "epoch": 22.120336943441636,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777858546455809,
      "loss": 11.0466,
      "step": 183820
    },
    {
      "epoch": 22.121540312876053,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777834638362391,
      "loss": 11.1625,
      "step": 183830
    },
    {
      "epoch": 22.12274368231047,
      "grad_norm": Infinity,
      "learning_rate": 0.00017778107291432539,
      "loss": 11.1816,
      "step": 183840
    },
    {
      "epoch": 22.123947051744885,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777786818798433,
      "loss": 11.1585,
      "step": 183850
    },
    {
      "epoch": 22.125150421179303,
      "grad_norm": Infinity,
      "learning_rate": 0.00017777629073279628,
      "loss": 11.0821,
      "step": 183860
    },
    {
      "epoch": 22.126353790613717,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777738994731877,
      "loss": 11.235,
      "step": 183870
    },
    {
      "epoch": 22.127557160048134,
      "grad_norm": Infinity,
      "learning_rate": 0.00017777150810102115,
      "loss": 11.1687,
      "step": 183880
    },
    {
      "epoch": 22.128760529482552,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777691166163,
      "loss": 11.116,
      "step": 183890
    },
    {
      "epoch": 22.129963898916966,
      "grad_norm": Infinity,
      "learning_rate": 0.00017776672501902774,
      "loss": 11.144,
      "step": 183900
    },
    {
      "epoch": 22.131167268351383,
      "grad_norm": Infinity,
      "learning_rate": 0.00017776433330920783,
      "loss": 11.1895,
      "step": 183910
    },
    {
      "epoch": 22.1323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00017776194148684372,
      "loss": 11.2596,
      "step": 183920
    },
    {
      "epoch": 22.133574007220215,
      "grad_norm": Infinity,
      "learning_rate": 0.00017775954955193888,
      "loss": 11.1265,
      "step": 183930
    },
    {
      "epoch": 22.134777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777571575044968,
      "loss": 11.0285,
      "step": 183940
    },
    {
      "epoch": 22.13598074608905,
      "grad_norm": Infinity,
      "learning_rate": 0.00017775476534452088,
      "loss": 11.2185,
      "step": 183950
    },
    {
      "epoch": 22.137184115523464,
      "grad_norm": Infinity,
      "learning_rate": 0.00017775237307201468,
      "loss": 11.0788,
      "step": 183960
    },
    {
      "epoch": 22.13838748495788,
      "grad_norm": Infinity,
      "learning_rate": 0.00017774998068698151,
      "loss": 11.1457,
      "step": 183970
    },
    {
      "epoch": 22.1395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017774758818942497,
      "loss": 11.201,
      "step": 183980
    },
    {
      "epoch": 22.140794223826713,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777451955793485,
      "loss": 11.0011,
      "step": 183990
    },
    {
      "epoch": 22.14199759326113,
      "grad_norm": Infinity,
      "learning_rate": 0.00017774280285675552,
      "loss": 11.165,
      "step": 184000
    },
    {
      "epoch": 22.14320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777404100216495,
      "loss": 11.1346,
      "step": 184010
    },
    {
      "epoch": 22.144404332129962,
      "grad_norm": Infinity,
      "learning_rate": 0.00017773801707403388,
      "loss": 11.0519,
      "step": 184020
    },
    {
      "epoch": 22.14560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777356240139122,
      "loss": 11.2038,
      "step": 184030
    },
    {
      "epoch": 22.146811070998798,
      "grad_norm": Infinity,
      "learning_rate": 0.00017773323084128789,
      "loss": 11.146,
      "step": 184040
    },
    {
      "epoch": 22.14801444043321,
      "grad_norm": Infinity,
      "learning_rate": 0.00017773083755616436,
      "loss": 11.2012,
      "step": 184050
    },
    {
      "epoch": 22.14921780986763,
      "grad_norm": Infinity,
      "learning_rate": 0.00017772844415854514,
      "loss": 11.147,
      "step": 184060
    },
    {
      "epoch": 22.150421179302047,
      "grad_norm": Infinity,
      "learning_rate": 0.00017772605064843367,
      "loss": 11.2395,
      "step": 184070
    },
    {
      "epoch": 22.15162454873646,
      "grad_norm": Infinity,
      "learning_rate": 0.00017772365702583342,
      "loss": 11.1637,
      "step": 184080
    },
    {
      "epoch": 22.15282791817088,
      "grad_norm": Infinity,
      "learning_rate": 0.00017772126329074786,
      "loss": 10.9943,
      "step": 184090
    },
    {
      "epoch": 22.154031287605296,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777188694431804,
      "loss": 11.1741,
      "step": 184100
    },
    {
      "epoch": 22.15523465703971,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777164754831346,
      "loss": 11.1082,
      "step": 184110
    },
    {
      "epoch": 22.156438026474127,
      "grad_norm": Infinity,
      "learning_rate": 0.00017771408141061383,
      "loss": 11.133,
      "step": 184120
    },
    {
      "epoch": 22.157641395908545,
      "grad_norm": Infinity,
      "learning_rate": 0.00017771168722562162,
      "loss": 11.1978,
      "step": 184130
    },
    {
      "epoch": 22.15884476534296,
      "grad_norm": Infinity,
      "learning_rate": 0.0001777092929281614,
      "loss": 11.0815,
      "step": 184140
    },
    {
      "epoch": 22.160048134777377,
      "grad_norm": Infinity,
      "learning_rate": 0.00017770689851823664,
      "loss": 11.1047,
      "step": 184150
    },
    {
      "epoch": 22.161251504211794,
      "grad_norm": Infinity,
      "learning_rate": 0.00017770450399585082,
      "loss": 11.1961,
      "step": 184160
    },
    {
      "epoch": 22.162454873646208,
      "grad_norm": Infinity,
      "learning_rate": 0.00017770210936100743,
      "loss": 11.0904,
      "step": 184170
    },
    {
      "epoch": 22.163658243080626,
      "grad_norm": Infinity,
      "learning_rate": 0.00017769971461370986,
      "loss": 11.1686,
      "step": 184180
    },
    {
      "epoch": 22.164861612515043,
      "grad_norm": Infinity,
      "learning_rate": 0.00017769731975396166,
      "loss": 11.0963,
      "step": 184190
    },
    {
      "epoch": 22.166064981949457,
      "grad_norm": Infinity,
      "learning_rate": 0.00017769492478176622,
      "loss": 11.1637,
      "step": 184200
    },
    {
      "epoch": 22.167268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.00017769252969712705,
      "loss": 11.1323,
      "step": 184210
    },
    {
      "epoch": 22.168471720818292,
      "grad_norm": Infinity,
      "learning_rate": 0.00017769013450004764,
      "loss": 11.0723,
      "step": 184220
    },
    {
      "epoch": 22.169675090252706,
      "grad_norm": Infinity,
      "learning_rate": 0.0001776877391905314,
      "loss": 11.1148,
      "step": 184230
    },
    {
      "epoch": 22.170878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00017768534376858181,
      "loss": 11.1893,
      "step": 184240
    },
    {
      "epoch": 22.17208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00017768294823420236,
      "loss": 11.0502,
      "step": 184250
    },
    {
      "epoch": 22.173285198555956,
      "grad_norm": Infinity,
      "learning_rate": 0.0001776805525873965,
      "loss": 11.2387,
      "step": 184260
    },
    {
      "epoch": 22.174488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.0001776781568281677,
      "loss": 11.1015,
      "step": 184270
    },
    {
      "epoch": 22.17569193742479,
      "grad_norm": Infinity,
      "learning_rate": 0.00017767576095651944,
      "loss": 11.1129,
      "step": 184280
    },
    {
      "epoch": 22.176895306859205,
      "grad_norm": Infinity,
      "learning_rate": 0.00017767336497245514,
      "loss": 11.2017,
      "step": 184290
    },
    {
      "epoch": 22.178098676293622,
      "grad_norm": Infinity,
      "learning_rate": 0.00017767096887597833,
      "loss": 11.1187,
      "step": 184300
    },
    {
      "epoch": 22.17930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.00017766857266709245,
      "loss": 11.1473,
      "step": 184310
    },
    {
      "epoch": 22.180505415162454,
      "grad_norm": Infinity,
      "learning_rate": 0.00017766617634580102,
      "loss": 11.0872,
      "step": 184320
    },
    {
      "epoch": 22.18170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00017766377991210739,
      "loss": 11.2037,
      "step": 184330
    },
    {
      "epoch": 22.18291215403129,
      "grad_norm": Infinity,
      "learning_rate": 0.00017766138336601509,
      "loss": 11.0925,
      "step": 184340
    },
    {
      "epoch": 22.184115523465703,
      "grad_norm": Infinity,
      "learning_rate": 0.00017765898670752764,
      "loss": 11.1631,
      "step": 184350
    },
    {
      "epoch": 22.18531889290012,
      "grad_norm": Infinity,
      "learning_rate": 0.00017765658993664844,
      "loss": 11.1585,
      "step": 184360
    },
    {
      "epoch": 22.186522262334538,
      "grad_norm": Infinity,
      "learning_rate": 0.000177654193053381,
      "loss": 11.1756,
      "step": 184370
    },
    {
      "epoch": 22.187725631768952,
      "grad_norm": Infinity,
      "learning_rate": 0.00017765179605772874,
      "loss": 11.1538,
      "step": 184380
    },
    {
      "epoch": 22.18892900120337,
      "grad_norm": Infinity,
      "learning_rate": 0.00017764939894969519,
      "loss": 11.2333,
      "step": 184390
    },
    {
      "epoch": 22.190132370637787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017764700172928378,
      "loss": 11.1714,
      "step": 184400
    },
    {
      "epoch": 22.1913357400722,
      "grad_norm": Infinity,
      "learning_rate": 0.00017764460439649796,
      "loss": 11.1207,
      "step": 184410
    },
    {
      "epoch": 22.19253910950662,
      "grad_norm": Infinity,
      "learning_rate": 0.00017764220695134125,
      "loss": 11.2096,
      "step": 184420
    },
    {
      "epoch": 22.193742478941036,
      "grad_norm": Infinity,
      "learning_rate": 0.0001776398093938171,
      "loss": 11.1228,
      "step": 184430
    },
    {
      "epoch": 22.19494584837545,
      "grad_norm": Infinity,
      "learning_rate": 0.00017763741172392895,
      "loss": 11.218,
      "step": 184440
    },
    {
      "epoch": 22.196149217809868,
      "grad_norm": Infinity,
      "learning_rate": 0.00017763501394168035,
      "loss": 11.1506,
      "step": 184450
    },
    {
      "epoch": 22.197352587244286,
      "grad_norm": Infinity,
      "learning_rate": 0.00017763261604707468,
      "loss": 11.0503,
      "step": 184460
    },
    {
      "epoch": 22.1985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017763021804011547,
      "loss": 11.0846,
      "step": 184470
    },
    {
      "epoch": 22.199759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00017762781992080613,
      "loss": 11.1866,
      "step": 184480
    },
    {
      "epoch": 22.200962695547535,
      "grad_norm": Infinity,
      "learning_rate": 0.0001776254216891502,
      "loss": 11.128,
      "step": 184490
    },
    {
      "epoch": 22.20216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.0001776230233451511,
      "loss": 11.1469,
      "step": 184500
    },
    {
      "epoch": 22.203369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.00017762062488881235,
      "loss": 11.2191,
      "step": 184510
    },
    {
      "epoch": 22.204572803850784,
      "grad_norm": Infinity,
      "learning_rate": 0.00017761822632013737,
      "loss": 10.9944,
      "step": 184520
    },
    {
      "epoch": 22.205776173285198,
      "grad_norm": Infinity,
      "learning_rate": 0.00017761582763912967,
      "loss": 11.0869,
      "step": 184530
    },
    {
      "epoch": 22.206979542719615,
      "grad_norm": Infinity,
      "learning_rate": 0.0001776134288457927,
      "loss": 11.097,
      "step": 184540
    },
    {
      "epoch": 22.20818291215403,
      "grad_norm": Infinity,
      "learning_rate": 0.0001776110299401299,
      "loss": 11.2775,
      "step": 184550
    },
    {
      "epoch": 22.209386281588447,
      "grad_norm": Infinity,
      "learning_rate": 0.00017760863092214485,
      "loss": 11.1367,
      "step": 184560
    },
    {
      "epoch": 22.210589651022865,
      "grad_norm": Infinity,
      "learning_rate": 0.0001776062317918409,
      "loss": 11.0717,
      "step": 184570
    },
    {
      "epoch": 22.21179302045728,
      "grad_norm": Infinity,
      "learning_rate": 0.00017760383254922157,
      "loss": 11.1684,
      "step": 184580
    },
    {
      "epoch": 22.212996389891696,
      "grad_norm": Infinity,
      "learning_rate": 0.00017760143319429036,
      "loss": 11.1855,
      "step": 184590
    },
    {
      "epoch": 22.214199759326114,
      "grad_norm": Infinity,
      "learning_rate": 0.00017759903372705068,
      "loss": 11.0781,
      "step": 184600
    },
    {
      "epoch": 22.215403128760528,
      "grad_norm": Infinity,
      "learning_rate": 0.00017759663414750607,
      "loss": 11.1864,
      "step": 184610
    },
    {
      "epoch": 22.216606498194945,
      "grad_norm": Infinity,
      "learning_rate": 0.00017759423445565996,
      "loss": 11.1486,
      "step": 184620
    },
    {
      "epoch": 22.217809867629363,
      "grad_norm": Infinity,
      "learning_rate": 0.00017759183465151584,
      "loss": 11.0688,
      "step": 184630
    },
    {
      "epoch": 22.219013237063777,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775894347350772,
      "loss": 11.1402,
      "step": 184640
    },
    {
      "epoch": 22.220216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.00017758703470634743,
      "loss": 11.1887,
      "step": 184650
    },
    {
      "epoch": 22.221419975932612,
      "grad_norm": Infinity,
      "learning_rate": 0.00017758463456533013,
      "loss": 11.1494,
      "step": 184660
    },
    {
      "epoch": 22.222623345367026,
      "grad_norm": Infinity,
      "learning_rate": 0.00017758223431202868,
      "loss": 11.1789,
      "step": 184670
    },
    {
      "epoch": 22.223826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.00017757983394644657,
      "loss": 11.166,
      "step": 184680
    },
    {
      "epoch": 22.22503008423586,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775774334685873,
      "loss": 11.169,
      "step": 184690
    },
    {
      "epoch": 22.226233453670275,
      "grad_norm": Infinity,
      "learning_rate": 0.00017757503287845435,
      "loss": 11.1179,
      "step": 184700
    },
    {
      "epoch": 22.227436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00017757263217605115,
      "loss": 11.1811,
      "step": 184710
    },
    {
      "epoch": 22.22864019253911,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775702313613812,
      "loss": 11.283,
      "step": 184720
    },
    {
      "epoch": 22.229843561973524,
      "grad_norm": Infinity,
      "learning_rate": 0.000177567830434448,
      "loss": 11.1423,
      "step": 184730
    },
    {
      "epoch": 22.231046931407942,
      "grad_norm": Infinity,
      "learning_rate": 0.00017756542939525497,
      "loss": 11.1595,
      "step": 184740
    },
    {
      "epoch": 22.23225030084236,
      "grad_norm": Infinity,
      "learning_rate": 0.00017756302824380562,
      "loss": 11.193,
      "step": 184750
    },
    {
      "epoch": 22.233453670276774,
      "grad_norm": Infinity,
      "learning_rate": 0.00017756062698010342,
      "loss": 11.1113,
      "step": 184760
    },
    {
      "epoch": 22.23465703971119,
      "grad_norm": Infinity,
      "learning_rate": 0.00017755822560415184,
      "loss": 11.0974,
      "step": 184770
    },
    {
      "epoch": 22.23586040914561,
      "grad_norm": Infinity,
      "learning_rate": 0.00017755582411595438,
      "loss": 11.1278,
      "step": 184780
    },
    {
      "epoch": 22.237063778580023,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775534225155145,
      "loss": 11.1743,
      "step": 184790
    },
    {
      "epoch": 22.23826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775510208028356,
      "loss": 11.2146,
      "step": 184800
    },
    {
      "epoch": 22.239470517448858,
      "grad_norm": Infinity,
      "learning_rate": 0.00017754861897792127,
      "loss": 11.0847,
      "step": 184810
    },
    {
      "epoch": 22.240673886883272,
      "grad_norm": Infinity,
      "learning_rate": 0.00017754621704077497,
      "loss": 11.1111,
      "step": 184820
    },
    {
      "epoch": 22.24187725631769,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775438149914001,
      "loss": 11.1007,
      "step": 184830
    },
    {
      "epoch": 22.243080625752107,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775414128298002,
      "loss": 11.1749,
      "step": 184840
    },
    {
      "epoch": 22.24428399518652,
      "grad_norm": Infinity,
      "learning_rate": 0.00017753901055597878,
      "loss": 11.1378,
      "step": 184850
    },
    {
      "epoch": 22.24548736462094,
      "grad_norm": Infinity,
      "learning_rate": 0.00017753660816993922,
      "loss": 11.0832,
      "step": 184860
    },
    {
      "epoch": 22.246690734055356,
      "grad_norm": Infinity,
      "learning_rate": 0.00017753420567168506,
      "loss": 11.1766,
      "step": 184870
    },
    {
      "epoch": 22.24789410348977,
      "grad_norm": Infinity,
      "learning_rate": 0.00017753180306121974,
      "loss": 11.1564,
      "step": 184880
    },
    {
      "epoch": 22.249097472924188,
      "grad_norm": Infinity,
      "learning_rate": 0.00017752940033854676,
      "loss": 11.1274,
      "step": 184890
    },
    {
      "epoch": 22.250300842358605,
      "grad_norm": Infinity,
      "learning_rate": 0.00017752699750366962,
      "loss": 11.1084,
      "step": 184900
    },
    {
      "epoch": 22.25150421179302,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775245945565918,
      "loss": 11.0816,
      "step": 184910
    },
    {
      "epoch": 22.252707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775221914973167,
      "loss": 11.2123,
      "step": 184920
    },
    {
      "epoch": 22.253910950661854,
      "grad_norm": Infinity,
      "learning_rate": 0.00017751978832584786,
      "loss": 11.1126,
      "step": 184930
    },
    {
      "epoch": 22.25511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00017751738504218875,
      "loss": 11.0857,
      "step": 184940
    },
    {
      "epoch": 22.256317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00017751498164634283,
      "loss": 11.1787,
      "step": 184950
    },
    {
      "epoch": 22.257521058965104,
      "grad_norm": Infinity,
      "learning_rate": 0.00017751257813831362,
      "loss": 11.3094,
      "step": 184960
    },
    {
      "epoch": 22.258724428399518,
      "grad_norm": Infinity,
      "learning_rate": 0.00017751017451810456,
      "loss": 11.1281,
      "step": 184970
    },
    {
      "epoch": 22.259927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00017750777078571917,
      "loss": 11.07,
      "step": 184980
    },
    {
      "epoch": 22.261131167268353,
      "grad_norm": Infinity,
      "learning_rate": 0.00017750536694116086,
      "loss": 11.1521,
      "step": 184990
    },
    {
      "epoch": 22.262334536702767,
      "grad_norm": Infinity,
      "learning_rate": 0.0001775029629844332,
      "loss": 11.1335,
      "step": 185000
    },
    {
      "epoch": 22.263537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.00017750055891553953,
      "loss": 11.103,
      "step": 185010
    },
    {
      "epoch": 22.264741275571602,
      "grad_norm": Infinity,
      "learning_rate": 0.00017749815473448347,
      "loss": 11.2661,
      "step": 185020
    },
    {
      "epoch": 22.265944645006016,
      "grad_norm": Infinity,
      "learning_rate": 0.00017749575044126846,
      "loss": 11.1043,
      "step": 185030
    },
    {
      "epoch": 22.267148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00017749334603589796,
      "loss": 11.1832,
      "step": 185040
    },
    {
      "epoch": 22.26835138387485,
      "grad_norm": Infinity,
      "learning_rate": 0.00017749094151837545,
      "loss": 11.261,
      "step": 185050
    },
    {
      "epoch": 22.269554753309265,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774885368887044,
      "loss": 11.1062,
      "step": 185060
    },
    {
      "epoch": 22.270758122743683,
      "grad_norm": Infinity,
      "learning_rate": 0.00017748613214688832,
      "loss": 11.1086,
      "step": 185070
    },
    {
      "epoch": 22.2719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00017748372729293067,
      "loss": 11.1146,
      "step": 185080
    },
    {
      "epoch": 22.273164861612514,
      "grad_norm": Infinity,
      "learning_rate": 0.00017748132232683493,
      "loss": 10.9917,
      "step": 185090
    },
    {
      "epoch": 22.27436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774789172486046,
      "loss": 11.199,
      "step": 185100
    },
    {
      "epoch": 22.27557160048135,
      "grad_norm": Infinity,
      "learning_rate": 0.00017747651205824317,
      "loss": 11.1472,
      "step": 185110
    },
    {
      "epoch": 22.276774969915763,
      "grad_norm": Infinity,
      "learning_rate": 0.00017747410675575406,
      "loss": 11.224,
      "step": 185120
    },
    {
      "epoch": 22.27797833935018,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774717013411408,
      "loss": 10.9945,
      "step": 185130
    },
    {
      "epoch": 22.2791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00017746929581440686,
      "loss": 11.1843,
      "step": 185140
    },
    {
      "epoch": 22.280385078219012,
      "grad_norm": Infinity,
      "learning_rate": 0.00017746689017555573,
      "loss": 11.1287,
      "step": 185150
    },
    {
      "epoch": 22.28158844765343,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774644844245909,
      "loss": 11.0673,
      "step": 185160
    },
    {
      "epoch": 22.282791817087848,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774620785615158,
      "loss": 11.2445,
      "step": 185170
    },
    {
      "epoch": 22.28399518652226,
      "grad_norm": Infinity,
      "learning_rate": 0.00017745967258633397,
      "loss": 11.1657,
      "step": 185180
    },
    {
      "epoch": 22.28519855595668,
      "grad_norm": Infinity,
      "learning_rate": 0.00017745726649904885,
      "loss": 11.1121,
      "step": 185190
    },
    {
      "epoch": 22.286401925391097,
      "grad_norm": Infinity,
      "learning_rate": 0.00017745486029966396,
      "loss": 11.1775,
      "step": 185200
    },
    {
      "epoch": 22.28760529482551,
      "grad_norm": Infinity,
      "learning_rate": 0.00017745245398818278,
      "loss": 11.2244,
      "step": 185210
    },
    {
      "epoch": 22.28880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.00017745004756460875,
      "loss": 11.0251,
      "step": 185220
    },
    {
      "epoch": 22.290012033694346,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774476410289454,
      "loss": 11.1388,
      "step": 185230
    },
    {
      "epoch": 22.29121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00017744523438119615,
      "loss": 11.2072,
      "step": 185240
    },
    {
      "epoch": 22.292418772563177,
      "grad_norm": Infinity,
      "learning_rate": 0.00017744282762136457,
      "loss": 11.1566,
      "step": 185250
    },
    {
      "epoch": 22.29362214199759,
      "grad_norm": Infinity,
      "learning_rate": 0.00017744042074945409,
      "loss": 11.132,
      "step": 185260
    },
    {
      "epoch": 22.29482551143201,
      "grad_norm": Infinity,
      "learning_rate": 0.00017743801376546818,
      "loss": 11.2767,
      "step": 185270
    },
    {
      "epoch": 22.296028880866427,
      "grad_norm": Infinity,
      "learning_rate": 0.00017743560666941036,
      "loss": 11.1979,
      "step": 185280
    },
    {
      "epoch": 22.29723225030084,
      "grad_norm": Infinity,
      "learning_rate": 0.00017743319946128408,
      "loss": 11.1081,
      "step": 185290
    },
    {
      "epoch": 22.29843561973526,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774307921410929,
      "loss": 11.0549,
      "step": 185300
    },
    {
      "epoch": 22.299638989169676,
      "grad_norm": Infinity,
      "learning_rate": 0.00017742838470884018,
      "loss": 11.1552,
      "step": 185310
    },
    {
      "epoch": 22.30084235860409,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774259771645295,
      "loss": 11.0598,
      "step": 185320
    },
    {
      "epoch": 22.302045728038507,
      "grad_norm": Infinity,
      "learning_rate": 0.00017742356950816433,
      "loss": 11.0117,
      "step": 185330
    },
    {
      "epoch": 22.303249097472925,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774211617397481,
      "loss": 11.1422,
      "step": 185340
    },
    {
      "epoch": 22.30445246690734,
      "grad_norm": Infinity,
      "learning_rate": 0.00017741875385928437,
      "loss": 11.1606,
      "step": 185350
    },
    {
      "epoch": 22.305655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00017741634586677658,
      "loss": 11.1669,
      "step": 185360
    },
    {
      "epoch": 22.306859205776174,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774139377622282,
      "loss": 11.093,
      "step": 185370
    },
    {
      "epoch": 22.308062575210588,
      "grad_norm": Infinity,
      "learning_rate": 0.00017741152954564278,
      "loss": 11.0746,
      "step": 185380
    },
    {
      "epoch": 22.309265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00017740912121702373,
      "loss": 11.0157,
      "step": 185390
    },
    {
      "epoch": 22.310469314079423,
      "grad_norm": Infinity,
      "learning_rate": 0.0001774067127763746,
      "loss": 11.237,
      "step": 185400
    },
    {
      "epoch": 22.311672683513837,
      "grad_norm": Infinity,
      "learning_rate": 0.00017740430422369882,
      "loss": 11.1536,
      "step": 185410
    },
    {
      "epoch": 22.312876052948255,
      "grad_norm": Infinity,
      "learning_rate": 0.00017740189555899991,
      "loss": 11.2214,
      "step": 185420
    },
    {
      "epoch": 22.314079422382672,
      "grad_norm": Infinity,
      "learning_rate": 0.00017739948678228133,
      "loss": 11.1147,
      "step": 185430
    },
    {
      "epoch": 22.315282791817086,
      "grad_norm": Infinity,
      "learning_rate": 0.00017739707789354663,
      "loss": 11.1594,
      "step": 185440
    },
    {
      "epoch": 22.316486161251504,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773946688927992,
      "loss": 11.2079,
      "step": 185450
    },
    {
      "epoch": 22.31768953068592,
      "grad_norm": Infinity,
      "learning_rate": 0.00017739225978004262,
      "loss": 11.1436,
      "step": 185460
    },
    {
      "epoch": 22.318892900120336,
      "grad_norm": Infinity,
      "learning_rate": 0.00017738985055528028,
      "loss": 11.1081,
      "step": 185470
    },
    {
      "epoch": 22.320096269554753,
      "grad_norm": Infinity,
      "learning_rate": 0.00017738744121851576,
      "loss": 11.1852,
      "step": 185480
    },
    {
      "epoch": 22.32129963898917,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773850317697525,
      "loss": 11.0762,
      "step": 185490
    },
    {
      "epoch": 22.322503008423585,
      "grad_norm": Infinity,
      "learning_rate": 0.000177382622208994,
      "loss": 11.2403,
      "step": 185500
    },
    {
      "epoch": 22.323706377858002,
      "grad_norm": Infinity,
      "learning_rate": 0.00017738021253624372,
      "loss": 11.1424,
      "step": 185510
    },
    {
      "epoch": 22.32490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00017737780275150518,
      "loss": 11.1715,
      "step": 185520
    },
    {
      "epoch": 22.326113116726834,
      "grad_norm": Infinity,
      "learning_rate": 0.00017737539285478184,
      "loss": 11.1282,
      "step": 185530
    },
    {
      "epoch": 22.32731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.00017737298284607722,
      "loss": 11.1532,
      "step": 185540
    },
    {
      "epoch": 22.32851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773705727253948,
      "loss": 11.2329,
      "step": 185550
    },
    {
      "epoch": 22.329723225030083,
      "grad_norm": Infinity,
      "learning_rate": 0.00017736816249273806,
      "loss": 11.1544,
      "step": 185560
    },
    {
      "epoch": 22.3309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.00017736575214811044,
      "loss": 11.1745,
      "step": 185570
    },
    {
      "epoch": 22.332129963898918,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773633416915155,
      "loss": 11.2214,
      "step": 185580
    },
    {
      "epoch": 22.333333333333332,
      "grad_norm": Infinity,
      "learning_rate": 0.00017736093112295673,
      "loss": 11.1923,
      "step": 185590
    },
    {
      "epoch": 22.33453670276775,
      "grad_norm": Infinity,
      "learning_rate": 0.00017735852044243756,
      "loss": 11.1805,
      "step": 185600
    },
    {
      "epoch": 22.335740072202167,
      "grad_norm": Infinity,
      "learning_rate": 0.00017735610964996154,
      "loss": 11.102,
      "step": 185610
    },
    {
      "epoch": 22.33694344163658,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773536987455321,
      "loss": 11.2309,
      "step": 185620
    },
    {
      "epoch": 22.338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773512877291528,
      "loss": 11.2199,
      "step": 185630
    },
    {
      "epoch": 22.339350180505416,
      "grad_norm": Infinity,
      "learning_rate": 0.00017734887660082705,
      "loss": 11.3025,
      "step": 185640
    },
    {
      "epoch": 22.34055354993983,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773464653605584,
      "loss": 11.1069,
      "step": 185650
    },
    {
      "epoch": 22.341756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.00017734405400835032,
      "loss": 11.0437,
      "step": 185660
    },
    {
      "epoch": 22.342960288808666,
      "grad_norm": Infinity,
      "learning_rate": 0.00017734164254420629,
      "loss": 11.1679,
      "step": 185670
    },
    {
      "epoch": 22.34416365824308,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773392309681298,
      "loss": 11.2021,
      "step": 185680
    },
    {
      "epoch": 22.345367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017733681928012434,
      "loss": 11.2601,
      "step": 185690
    },
    {
      "epoch": 22.346570397111915,
      "grad_norm": Infinity,
      "learning_rate": 0.00017733440748019346,
      "loss": 11.1717,
      "step": 185700
    },
    {
      "epoch": 22.34777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.00017733199556834058,
      "loss": 11.0309,
      "step": 185710
    },
    {
      "epoch": 22.348977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773295835445692,
      "loss": 11.128,
      "step": 185720
    },
    {
      "epoch": 22.350180505415164,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773271714088828,
      "loss": 11.0878,
      "step": 185730
    },
    {
      "epoch": 22.351383874849578,
      "grad_norm": Infinity,
      "learning_rate": 0.00017732475916128491,
      "loss": 11.1033,
      "step": 185740
    },
    {
      "epoch": 22.352587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00017732234680177902,
      "loss": 11.1801,
      "step": 185750
    },
    {
      "epoch": 22.353790613718413,
      "grad_norm": Infinity,
      "learning_rate": 0.00017731993433036855,
      "loss": 11.2125,
      "step": 185760
    },
    {
      "epoch": 22.354993983152827,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773175217470571,
      "loss": 11.2979,
      "step": 185770
    },
    {
      "epoch": 22.356197352587245,
      "grad_norm": Infinity,
      "learning_rate": 0.00017731510905184808,
      "loss": 11.0352,
      "step": 185780
    },
    {
      "epoch": 22.357400722021662,
      "grad_norm": Infinity,
      "learning_rate": 0.00017731269624474502,
      "loss": 11.0381,
      "step": 185790
    },
    {
      "epoch": 22.358604091456076,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773102833257514,
      "loss": 11.1038,
      "step": 185800
    },
    {
      "epoch": 22.359807460890494,
      "grad_norm": Infinity,
      "learning_rate": 0.0001773078702948707,
      "loss": 11.1792,
      "step": 185810
    },
    {
      "epoch": 22.36101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.00017730545715210645,
      "loss": 11.1818,
      "step": 185820
    },
    {
      "epoch": 22.362214199759325,
      "grad_norm": Infinity,
      "learning_rate": 0.00017730304389746207,
      "loss": 11.0299,
      "step": 185830
    },
    {
      "epoch": 22.363417569193743,
      "grad_norm": Infinity,
      "learning_rate": 0.00017730063053094115,
      "loss": 11.2224,
      "step": 185840
    },
    {
      "epoch": 22.36462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772982170525471,
      "loss": 11.2642,
      "step": 185850
    },
    {
      "epoch": 22.365824308062574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017729580346228347,
      "loss": 11.1858,
      "step": 185860
    },
    {
      "epoch": 22.367027677496992,
      "grad_norm": Infinity,
      "learning_rate": 0.00017729338976015373,
      "loss": 11.1779,
      "step": 185870
    },
    {
      "epoch": 22.36823104693141,
      "grad_norm": Infinity,
      "learning_rate": 0.00017729097594616135,
      "loss": 11.0451,
      "step": 185880
    },
    {
      "epoch": 22.369434416365824,
      "grad_norm": Infinity,
      "learning_rate": 0.00017728856202030988,
      "loss": 11.1909,
      "step": 185890
    },
    {
      "epoch": 22.37063778580024,
      "grad_norm": Infinity,
      "learning_rate": 0.00017728614798260273,
      "loss": 11.1536,
      "step": 185900
    },
    {
      "epoch": 22.37184115523466,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772837338330435,
      "loss": 11.3102,
      "step": 185910
    },
    {
      "epoch": 22.373044524669073,
      "grad_norm": Infinity,
      "learning_rate": 0.00017728131957163556,
      "loss": 11.2002,
      "step": 185920
    },
    {
      "epoch": 22.37424789410349,
      "grad_norm": Infinity,
      "learning_rate": 0.00017727890519838254,
      "loss": 11.1803,
      "step": 185930
    },
    {
      "epoch": 22.375451263537904,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772764907132878,
      "loss": 11.2243,
      "step": 185940
    },
    {
      "epoch": 22.376654632972322,
      "grad_norm": Infinity,
      "learning_rate": 0.00017727407611635494,
      "loss": 11.1154,
      "step": 185950
    },
    {
      "epoch": 22.37785800240674,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772716614075874,
      "loss": 11.1048,
      "step": 185960
    },
    {
      "epoch": 22.379061371841154,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772692465869887,
      "loss": 11.2587,
      "step": 185970
    },
    {
      "epoch": 22.38026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772668316545623,
      "loss": 11.2595,
      "step": 185980
    },
    {
      "epoch": 22.38146811070999,
      "grad_norm": Infinity,
      "learning_rate": 0.00017726441661031173,
      "loss": 11.1677,
      "step": 185990
    },
    {
      "epoch": 22.382671480144403,
      "grad_norm": Infinity,
      "learning_rate": 0.00017726200145424045,
      "loss": 11.0761,
      "step": 186000
    },
    {
      "epoch": 22.38387484957882,
      "grad_norm": Infinity,
      "learning_rate": 0.000177259586186352,
      "loss": 11.1223,
      "step": 186010
    },
    {
      "epoch": 22.385078219013238,
      "grad_norm": Infinity,
      "learning_rate": 0.00017725717080664988,
      "loss": 11.1536,
      "step": 186020
    },
    {
      "epoch": 22.386281588447652,
      "grad_norm": Infinity,
      "learning_rate": 0.00017725475531513754,
      "loss": 11.032,
      "step": 186030
    },
    {
      "epoch": 22.38748495788207,
      "grad_norm": Infinity,
      "learning_rate": 0.00017725233971181847,
      "loss": 11.1847,
      "step": 186040
    },
    {
      "epoch": 22.388688327316487,
      "grad_norm": Infinity,
      "learning_rate": 0.00017724992399669622,
      "loss": 11.1611,
      "step": 186050
    },
    {
      "epoch": 22.3898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00017724750816977425,
      "loss": 11.1265,
      "step": 186060
    },
    {
      "epoch": 22.39109506618532,
      "grad_norm": Infinity,
      "learning_rate": 0.00017724509223105605,
      "loss": 11.1782,
      "step": 186070
    },
    {
      "epoch": 22.392298435619736,
      "grad_norm": Infinity,
      "learning_rate": 0.00017724267618054516,
      "loss": 11.1578,
      "step": 186080
    },
    {
      "epoch": 22.39350180505415,
      "grad_norm": Infinity,
      "learning_rate": 0.00017724026001824506,
      "loss": 11.0779,
      "step": 186090
    },
    {
      "epoch": 22.394705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017723784374415916,
      "loss": 11.2025,
      "step": 186100
    },
    {
      "epoch": 22.395908543922985,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772354273582911,
      "loss": 11.1966,
      "step": 186110
    },
    {
      "epoch": 22.3971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017723301086064428,
      "loss": 11.1856,
      "step": 186120
    },
    {
      "epoch": 22.398315282791817,
      "grad_norm": Infinity,
      "learning_rate": 0.00017723059425122224,
      "loss": 11.1591,
      "step": 186130
    },
    {
      "epoch": 22.399518652226234,
      "grad_norm": Infinity,
      "learning_rate": 0.00017722817753002844,
      "loss": 11.214,
      "step": 186140
    },
    {
      "epoch": 22.40072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772257606970664,
      "loss": 11.1031,
      "step": 186150
    },
    {
      "epoch": 22.401925391095066,
      "grad_norm": Infinity,
      "learning_rate": 0.00017722334375233964,
      "loss": 11.1691,
      "step": 186160
    },
    {
      "epoch": 22.403128760529484,
      "grad_norm": Infinity,
      "learning_rate": 0.00017722092669585163,
      "loss": 11.1589,
      "step": 186170
    },
    {
      "epoch": 22.404332129963898,
      "grad_norm": Infinity,
      "learning_rate": 0.00017721850952760585,
      "loss": 11.175,
      "step": 186180
    },
    {
      "epoch": 22.405535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00017721609224760588,
      "loss": 11.1472,
      "step": 186190
    },
    {
      "epoch": 22.406738868832733,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772136748558551,
      "loss": 11.1941,
      "step": 186200
    },
    {
      "epoch": 22.407942238267147,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772112573523571,
      "loss": 10.985,
      "step": 186210
    },
    {
      "epoch": 22.409145607701564,
      "grad_norm": Infinity,
      "learning_rate": 0.00017720883973711533,
      "loss": 11.2462,
      "step": 186220
    },
    {
      "epoch": 22.410348977135982,
      "grad_norm": Infinity,
      "learning_rate": 0.0001772064220101333,
      "loss": 11.1168,
      "step": 186230
    },
    {
      "epoch": 22.411552346570396,
      "grad_norm": Infinity,
      "learning_rate": 0.00017720400417141455,
      "loss": 11.1905,
      "step": 186240
    },
    {
      "epoch": 22.412755716004813,
      "grad_norm": Infinity,
      "learning_rate": 0.00017720158622096252,
      "loss": 11.0778,
      "step": 186250
    },
    {
      "epoch": 22.41395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017719916815878075,
      "loss": 11.1179,
      "step": 186260
    },
    {
      "epoch": 22.415162454873645,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771967499848727,
      "loss": 11.2511,
      "step": 186270
    },
    {
      "epoch": 22.416365824308063,
      "grad_norm": Infinity,
      "learning_rate": 0.00017719433169924192,
      "loss": 11.1963,
      "step": 186280
    },
    {
      "epoch": 22.41756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.00017719191330189187,
      "loss": 11.1283,
      "step": 186290
    },
    {
      "epoch": 22.418772563176894,
      "grad_norm": Infinity,
      "learning_rate": 0.00017718949479282606,
      "loss": 11.1111,
      "step": 186300
    },
    {
      "epoch": 22.41997593261131,
      "grad_norm": Infinity,
      "learning_rate": 0.00017718707617204798,
      "loss": 11.0909,
      "step": 186310
    },
    {
      "epoch": 22.42117930204573,
      "grad_norm": Infinity,
      "learning_rate": 0.00017718465743956117,
      "loss": 11.1416,
      "step": 186320
    },
    {
      "epoch": 22.422382671480143,
      "grad_norm": Infinity,
      "learning_rate": 0.00017718223859536911,
      "loss": 11.1364,
      "step": 186330
    },
    {
      "epoch": 22.42358604091456,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771798196394753,
      "loss": 11.1929,
      "step": 186340
    },
    {
      "epoch": 22.42478941034898,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771774005718832,
      "loss": 11.2197,
      "step": 186350
    },
    {
      "epoch": 22.425992779783392,
      "grad_norm": Infinity,
      "learning_rate": 0.00017717498139259633,
      "loss": 11.0222,
      "step": 186360
    },
    {
      "epoch": 22.42719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00017717256210161825,
      "loss": 11.1915,
      "step": 186370
    },
    {
      "epoch": 22.428399518652228,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771701426989524,
      "loss": 11.1596,
      "step": 186380
    },
    {
      "epoch": 22.42960288808664,
      "grad_norm": Infinity,
      "learning_rate": 0.00017716772318460234,
      "loss": 11.2313,
      "step": 186390
    },
    {
      "epoch": 22.43080625752106,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771653035585715,
      "loss": 11.0301,
      "step": 186400
    },
    {
      "epoch": 22.432009626955477,
      "grad_norm": Infinity,
      "learning_rate": 0.00017716288382086344,
      "loss": 11.1357,
      "step": 186410
    },
    {
      "epoch": 22.43321299638989,
      "grad_norm": Infinity,
      "learning_rate": 0.00017716046397148161,
      "loss": 11.1981,
      "step": 186420
    },
    {
      "epoch": 22.43441636582431,
      "grad_norm": Infinity,
      "learning_rate": 0.00017715804401042954,
      "loss": 11.0884,
      "step": 186430
    },
    {
      "epoch": 22.435619735258726,
      "grad_norm": Infinity,
      "learning_rate": 0.00017715562393771074,
      "loss": 11.221,
      "step": 186440
    },
    {
      "epoch": 22.43682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00017715320375332868,
      "loss": 11.1002,
      "step": 186450
    },
    {
      "epoch": 22.438026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00017715078345728692,
      "loss": 11.1157,
      "step": 186460
    },
    {
      "epoch": 22.439229843561975,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771483630495889,
      "loss": 11.2329,
      "step": 186470
    },
    {
      "epoch": 22.44043321299639,
      "grad_norm": Infinity,
      "learning_rate": 0.00017714594253023816,
      "loss": 11.1669,
      "step": 186480
    },
    {
      "epoch": 22.441636582430807,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771435218992382,
      "loss": 11.1054,
      "step": 186490
    },
    {
      "epoch": 22.442839951865224,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771411011565925,
      "loss": 11.1214,
      "step": 186500
    },
    {
      "epoch": 22.444043321299638,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771386803023046,
      "loss": 11.1538,
      "step": 186510
    },
    {
      "epoch": 22.445246690734056,
      "grad_norm": Infinity,
      "learning_rate": 0.000177136259336378,
      "loss": 11.0772,
      "step": 186520
    },
    {
      "epoch": 22.446450060168473,
      "grad_norm": Infinity,
      "learning_rate": 0.00017713383825881615,
      "loss": 11.1226,
      "step": 186530
    },
    {
      "epoch": 22.447653429602887,
      "grad_norm": Infinity,
      "learning_rate": 0.00017713141706962262,
      "loss": 11.1928,
      "step": 186540
    },
    {
      "epoch": 22.448856799037305,
      "grad_norm": Infinity,
      "learning_rate": 0.00017712899576880086,
      "loss": 11.0675,
      "step": 186550
    },
    {
      "epoch": 22.450060168471722,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771265743563544,
      "loss": 11.0459,
      "step": 186560
    },
    {
      "epoch": 22.451263537906136,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771241528322868,
      "loss": 11.237,
      "step": 186570
    },
    {
      "epoch": 22.452466907340554,
      "grad_norm": Infinity,
      "learning_rate": 0.00017712173119660145,
      "loss": 11.1534,
      "step": 186580
    },
    {
      "epoch": 22.45367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017711930944930195,
      "loss": 11.2796,
      "step": 186590
    },
    {
      "epoch": 22.454873646209386,
      "grad_norm": Infinity,
      "learning_rate": 0.00017711688759039176,
      "loss": 11.1256,
      "step": 186600
    },
    {
      "epoch": 22.456077015643803,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771144656198744,
      "loss": 10.9658,
      "step": 186610
    },
    {
      "epoch": 22.45728038507822,
      "grad_norm": Infinity,
      "learning_rate": 0.00017711204353775334,
      "loss": 11.1732,
      "step": 186620
    },
    {
      "epoch": 22.458483754512635,
      "grad_norm": Infinity,
      "learning_rate": 0.00017710962134403216,
      "loss": 11.2053,
      "step": 186630
    },
    {
      "epoch": 22.459687123947052,
      "grad_norm": Infinity,
      "learning_rate": 0.00017710719903871427,
      "loss": 11.0598,
      "step": 186640
    },
    {
      "epoch": 22.460890493381466,
      "grad_norm": Infinity,
      "learning_rate": 0.00017710477662180325,
      "loss": 11.0909,
      "step": 186650
    },
    {
      "epoch": 22.462093862815884,
      "grad_norm": Infinity,
      "learning_rate": 0.0001771023540933026,
      "loss": 11.1607,
      "step": 186660
    },
    {
      "epoch": 22.4632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.0001770999314532158,
      "loss": 11.2159,
      "step": 186670
    },
    {
      "epoch": 22.464500601684716,
      "grad_norm": Infinity,
      "learning_rate": 0.00017709750870154633,
      "loss": 11.101,
      "step": 186680
    },
    {
      "epoch": 22.465703971119133,
      "grad_norm": Infinity,
      "learning_rate": 0.00017709508583829776,
      "loss": 11.196,
      "step": 186690
    },
    {
      "epoch": 22.46690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.00017709266286347356,
      "loss": 11.1604,
      "step": 186700
    },
    {
      "epoch": 22.468110709987965,
      "grad_norm": Infinity,
      "learning_rate": 0.00017709023977707724,
      "loss": 11.0374,
      "step": 186710
    },
    {
      "epoch": 22.469314079422382,
      "grad_norm": Infinity,
      "learning_rate": 0.0001770878165791123,
      "loss": 11.1322,
      "step": 186720
    },
    {
      "epoch": 22.4705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017708539326958227,
      "loss": 11.1592,
      "step": 186730
    },
    {
      "epoch": 22.471720818291214,
      "grad_norm": Infinity,
      "learning_rate": 0.00017708296984849064,
      "loss": 11.1446,
      "step": 186740
    },
    {
      "epoch": 22.47292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00017708054631584093,
      "loss": 11.2014,
      "step": 186750
    },
    {
      "epoch": 22.47412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.00017707812267163663,
      "loss": 11.1451,
      "step": 186760
    },
    {
      "epoch": 22.475330926594463,
      "grad_norm": Infinity,
      "learning_rate": 0.00017707569891588126,
      "loss": 11.2231,
      "step": 186770
    },
    {
      "epoch": 22.47653429602888,
      "grad_norm": Infinity,
      "learning_rate": 0.00017707327504857834,
      "loss": 11.1823,
      "step": 186780
    },
    {
      "epoch": 22.477737665463298,
      "grad_norm": Infinity,
      "learning_rate": 0.00017707085106973133,
      "loss": 11.1684,
      "step": 186790
    },
    {
      "epoch": 22.478941034897712,
      "grad_norm": Infinity,
      "learning_rate": 0.00017706842697934381,
      "loss": 11.1713,
      "step": 186800
    },
    {
      "epoch": 22.48014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017706600277741922,
      "loss": 11.1839,
      "step": 186810
    },
    {
      "epoch": 22.481347773766547,
      "grad_norm": Infinity,
      "learning_rate": 0.00017706357846396112,
      "loss": 11.1706,
      "step": 186820
    },
    {
      "epoch": 22.48255114320096,
      "grad_norm": Infinity,
      "learning_rate": 0.00017706115403897296,
      "loss": 11.1758,
      "step": 186830
    },
    {
      "epoch": 22.48375451263538,
      "grad_norm": Infinity,
      "learning_rate": 0.0001770587295024583,
      "loss": 11.0375,
      "step": 186840
    },
    {
      "epoch": 22.484957882069796,
      "grad_norm": Infinity,
      "learning_rate": 0.00017705630485442066,
      "loss": 11.2197,
      "step": 186850
    },
    {
      "epoch": 22.48616125150421,
      "grad_norm": Infinity,
      "learning_rate": 0.00017705388009486348,
      "loss": 11.2333,
      "step": 186860
    },
    {
      "epoch": 22.487364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.00017705145522379034,
      "loss": 11.258,
      "step": 186870
    },
    {
      "epoch": 22.488567990373046,
      "grad_norm": Infinity,
      "learning_rate": 0.00017704903024120473,
      "loss": 11.1586,
      "step": 186880
    },
    {
      "epoch": 22.48977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.00017704660514711011,
      "loss": 11.2632,
      "step": 186890
    },
    {
      "epoch": 22.490974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00017704417994151002,
      "loss": 11.1358,
      "step": 186900
    },
    {
      "epoch": 22.492178098676295,
      "grad_norm": Infinity,
      "learning_rate": 0.00017704175462440803,
      "loss": 11.0721,
      "step": 186910
    },
    {
      "epoch": 22.49338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.0001770393291958076,
      "loss": 11.0954,
      "step": 186920
    },
    {
      "epoch": 22.494584837545126,
      "grad_norm": Infinity,
      "learning_rate": 0.0001770369036557122,
      "loss": 11.1236,
      "step": 186930
    },
    {
      "epoch": 22.495788206979544,
      "grad_norm": Infinity,
      "learning_rate": 0.00017703447800412542,
      "loss": 11.1668,
      "step": 186940
    },
    {
      "epoch": 22.496991576413958,
      "grad_norm": Infinity,
      "learning_rate": 0.00017703205224105074,
      "loss": 11.0386,
      "step": 186950
    },
    {
      "epoch": 22.498194945848375,
      "grad_norm": Infinity,
      "learning_rate": 0.0001770296263664916,
      "loss": 11.2007,
      "step": 186960
    },
    {
      "epoch": 22.499398315282793,
      "grad_norm": Infinity,
      "learning_rate": 0.0001770272003804516,
      "loss": 11.1729,
      "step": 186970
    },
    {
      "epoch": 22.500601684717207,
      "grad_norm": Infinity,
      "learning_rate": 0.00017702477428293425,
      "loss": 11.2161,
      "step": 186980
    },
    {
      "epoch": 22.501805054151625,
      "grad_norm": Infinity,
      "learning_rate": 0.00017702234807394303,
      "loss": 11.2265,
      "step": 186990
    },
    {
      "epoch": 22.503008423586042,
      "grad_norm": Infinity,
      "learning_rate": 0.00017701992175348142,
      "loss": 11.0526,
      "step": 187000
    },
    {
      "epoch": 22.504211793020456,
      "grad_norm": Infinity,
      "learning_rate": 0.000177017495321553,
      "loss": 11.1864,
      "step": 187010
    },
    {
      "epoch": 22.505415162454874,
      "grad_norm": Infinity,
      "learning_rate": 0.00017701506877816127,
      "loss": 11.1147,
      "step": 187020
    },
    {
      "epoch": 22.50661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00017701264212330968,
      "loss": 11.1837,
      "step": 187030
    },
    {
      "epoch": 22.507821901323705,
      "grad_norm": Infinity,
      "learning_rate": 0.00017701021535700178,
      "loss": 11.168,
      "step": 187040
    },
    {
      "epoch": 22.509025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.00017700778847924112,
      "loss": 11.1252,
      "step": 187050
    },
    {
      "epoch": 22.51022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.00017700536149003115,
      "loss": 11.1354,
      "step": 187060
    },
    {
      "epoch": 22.511432009626954,
      "grad_norm": Infinity,
      "learning_rate": 0.0001770029343893754,
      "loss": 11.1933,
      "step": 187070
    },
    {
      "epoch": 22.512635379061372,
      "grad_norm": Infinity,
      "learning_rate": 0.0001770005071772774,
      "loss": 11.1285,
      "step": 187080
    },
    {
      "epoch": 22.51383874849579,
      "grad_norm": Infinity,
      "learning_rate": 0.0001769980798537407,
      "loss": 11.1466,
      "step": 187090
    },
    {
      "epoch": 22.515042117930204,
      "grad_norm": Infinity,
      "learning_rate": 0.00017699565241876872,
      "loss": 11.0605,
      "step": 187100
    },
    {
      "epoch": 22.51624548736462,
      "grad_norm": Infinity,
      "learning_rate": 0.00017699322487236502,
      "loss": 11.2709,
      "step": 187110
    },
    {
      "epoch": 22.51744885679904,
      "grad_norm": Infinity,
      "learning_rate": 0.00017699079721453313,
      "loss": 11.1914,
      "step": 187120
    },
    {
      "epoch": 22.518652226233453,
      "grad_norm": Infinity,
      "learning_rate": 0.00017698836944527656,
      "loss": 11.1101,
      "step": 187130
    },
    {
      "epoch": 22.51985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017698594156459878,
      "loss": 11.221,
      "step": 187140
    },
    {
      "epoch": 22.521058965102288,
      "grad_norm": Infinity,
      "learning_rate": 0.00017698351357250337,
      "loss": 11.2434,
      "step": 187150
    },
    {
      "epoch": 22.522262334536702,
      "grad_norm": Infinity,
      "learning_rate": 0.00017698108546899377,
      "loss": 11.1105,
      "step": 187160
    },
    {
      "epoch": 22.52346570397112,
      "grad_norm": Infinity,
      "learning_rate": 0.00017697865725407357,
      "loss": 10.9253,
      "step": 187170
    },
    {
      "epoch": 22.524669073405537,
      "grad_norm": Infinity,
      "learning_rate": 0.00017697622892774625,
      "loss": 11.0859,
      "step": 187180
    },
    {
      "epoch": 22.52587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.0001769738004900153,
      "loss": 11.2238,
      "step": 187190
    },
    {
      "epoch": 22.52707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00017697137194088426,
      "loss": 11.16,
      "step": 187200
    },
    {
      "epoch": 22.528279181708786,
      "grad_norm": Infinity,
      "learning_rate": 0.00017696894328035666,
      "loss": 11.0697,
      "step": 187210
    },
    {
      "epoch": 22.5294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00017696651450843596,
      "loss": 11.181,
      "step": 187220
    },
    {
      "epoch": 22.530685920577618,
      "grad_norm": Infinity,
      "learning_rate": 0.00017696408562512575,
      "loss": 11.1725,
      "step": 187230
    },
    {
      "epoch": 22.531889290012035,
      "grad_norm": Infinity,
      "learning_rate": 0.00017696165663042948,
      "loss": 11.0334,
      "step": 187240
    },
    {
      "epoch": 22.53309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.0001769592275243507,
      "loss": 11.2224,
      "step": 187250
    },
    {
      "epoch": 22.534296028880867,
      "grad_norm": Infinity,
      "learning_rate": 0.0001769567983068929,
      "loss": 11.147,
      "step": 187260
    },
    {
      "epoch": 22.535499398315284,
      "grad_norm": Infinity,
      "learning_rate": 0.00017695436897805964,
      "loss": 11.1405,
      "step": 187270
    },
    {
      "epoch": 22.5367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.0001769519395378544,
      "loss": 11.1163,
      "step": 187280
    },
    {
      "epoch": 22.537906137184116,
      "grad_norm": Infinity,
      "learning_rate": 0.0001769495099862807,
      "loss": 11.1465,
      "step": 187290
    },
    {
      "epoch": 22.53910950661853,
      "grad_norm": Infinity,
      "learning_rate": 0.00017694708032334205,
      "loss": 11.0935,
      "step": 187300
    },
    {
      "epoch": 22.540312876052948,
      "grad_norm": Infinity,
      "learning_rate": 0.000176944650549042,
      "loss": 11.0713,
      "step": 187310
    },
    {
      "epoch": 22.541516245487365,
      "grad_norm": Infinity,
      "learning_rate": 0.00017694222066338405,
      "loss": 11.1573,
      "step": 187320
    },
    {
      "epoch": 22.54271961492178,
      "grad_norm": Infinity,
      "learning_rate": 0.00017693979066637168,
      "loss": 11.094,
      "step": 187330
    },
    {
      "epoch": 22.543922984356197,
      "grad_norm": Infinity,
      "learning_rate": 0.00017693736055800845,
      "loss": 11.0219,
      "step": 187340
    },
    {
      "epoch": 22.545126353790614,
      "grad_norm": Infinity,
      "learning_rate": 0.00017693493033829784,
      "loss": 11.187,
      "step": 187350
    },
    {
      "epoch": 22.54632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00017693250000724343,
      "loss": 11.2475,
      "step": 187360
    },
    {
      "epoch": 22.547533092659446,
      "grad_norm": Infinity,
      "learning_rate": 0.00017693006956484867,
      "loss": 11.1387,
      "step": 187370
    },
    {
      "epoch": 22.548736462093864,
      "grad_norm": Infinity,
      "learning_rate": 0.00017692763901111714,
      "loss": 11.1853,
      "step": 187380
    },
    {
      "epoch": 22.549939831528278,
      "grad_norm": Infinity,
      "learning_rate": 0.00017692520834605227,
      "loss": 11.0701,
      "step": 187390
    },
    {
      "epoch": 22.551143200962695,
      "grad_norm": Infinity,
      "learning_rate": 0.0001769227775696577,
      "loss": 11.2376,
      "step": 187400
    },
    {
      "epoch": 22.552346570397113,
      "grad_norm": Infinity,
      "learning_rate": 0.00017692034668193683,
      "loss": 11.1516,
      "step": 187410
    },
    {
      "epoch": 22.553549939831527,
      "grad_norm": Infinity,
      "learning_rate": 0.00017691791568289324,
      "loss": 11.1306,
      "step": 187420
    },
    {
      "epoch": 22.554753309265944,
      "grad_norm": Infinity,
      "learning_rate": 0.00017691548457253044,
      "loss": 11.1122,
      "step": 187430
    },
    {
      "epoch": 22.555956678700362,
      "grad_norm": Infinity,
      "learning_rate": 0.00017691305335085195,
      "loss": 11.1582,
      "step": 187440
    },
    {
      "epoch": 22.557160048134776,
      "grad_norm": Infinity,
      "learning_rate": 0.00017691062201786125,
      "loss": 11.0114,
      "step": 187450
    },
    {
      "epoch": 22.558363417569193,
      "grad_norm": Infinity,
      "learning_rate": 0.00017690819057356193,
      "loss": 11.1611,
      "step": 187460
    },
    {
      "epoch": 22.55956678700361,
      "grad_norm": Infinity,
      "learning_rate": 0.00017690575901795747,
      "loss": 11.2243,
      "step": 187470
    },
    {
      "epoch": 22.560770156438025,
      "grad_norm": Infinity,
      "learning_rate": 0.0001769033273510514,
      "loss": 11.028,
      "step": 187480
    },
    {
      "epoch": 22.561973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.00017690089557284718,
      "loss": 11.195,
      "step": 187490
    },
    {
      "epoch": 22.56317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768984636833484,
      "loss": 11.177,
      "step": 187500
    },
    {
      "epoch": 22.564380264741274,
      "grad_norm": Infinity,
      "learning_rate": 0.00017689603168255855,
      "loss": 11.0852,
      "step": 187510
    },
    {
      "epoch": 22.56558363417569,
      "grad_norm": Infinity,
      "learning_rate": 0.00017689359957048117,
      "loss": 11.1243,
      "step": 187520
    },
    {
      "epoch": 22.56678700361011,
      "grad_norm": Infinity,
      "learning_rate": 0.00017689116734711976,
      "loss": 11.1217,
      "step": 187530
    },
    {
      "epoch": 22.567990373044523,
      "grad_norm": Infinity,
      "learning_rate": 0.00017688873501247785,
      "loss": 11.1652,
      "step": 187540
    },
    {
      "epoch": 22.56919374247894,
      "grad_norm": Infinity,
      "learning_rate": 0.00017688630256655898,
      "loss": 11.1238,
      "step": 187550
    },
    {
      "epoch": 22.57039711191336,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768838700093666,
      "loss": 11.1144,
      "step": 187560
    },
    {
      "epoch": 22.571600481347772,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768814373409043,
      "loss": 11.1112,
      "step": 187570
    },
    {
      "epoch": 22.57280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768790045611756,
      "loss": 11.0239,
      "step": 187580
    },
    {
      "epoch": 22.574007220216608,
      "grad_norm": Infinity,
      "learning_rate": 0.00017687657167018397,
      "loss": 11.1385,
      "step": 187590
    },
    {
      "epoch": 22.57521058965102,
      "grad_norm": Infinity,
      "learning_rate": 0.000176874138667933,
      "loss": 11.0655,
      "step": 187600
    },
    {
      "epoch": 22.57641395908544,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768717055544261,
      "loss": 11.0885,
      "step": 187610
    },
    {
      "epoch": 22.577617328519857,
      "grad_norm": Infinity,
      "learning_rate": 0.00017686927232966692,
      "loss": 11.2164,
      "step": 187620
    },
    {
      "epoch": 22.57882069795427,
      "grad_norm": Infinity,
      "learning_rate": 0.00017686683899365891,
      "loss": 11.0837,
      "step": 187630
    },
    {
      "epoch": 22.58002406738869,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768644055464056,
      "loss": 11.1123,
      "step": 187640
    },
    {
      "epoch": 22.581227436823106,
      "grad_norm": Infinity,
      "learning_rate": 0.00017686197198791054,
      "loss": 11.1777,
      "step": 187650
    },
    {
      "epoch": 22.58243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.00017685953831817722,
      "loss": 11.0424,
      "step": 187660
    },
    {
      "epoch": 22.583634175691937,
      "grad_norm": Infinity,
      "learning_rate": 0.00017685710453720917,
      "loss": 11.173,
      "step": 187670
    },
    {
      "epoch": 22.584837545126355,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768546706450099,
      "loss": 11.1405,
      "step": 187680
    },
    {
      "epoch": 22.58604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00017685223664158298,
      "loss": 11.1449,
      "step": 187690
    },
    {
      "epoch": 22.587244283995187,
      "grad_norm": Infinity,
      "learning_rate": 0.00017684980252693187,
      "loss": 11.0346,
      "step": 187700
    },
    {
      "epoch": 22.588447653429604,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768473683010601,
      "loss": 11.1039,
      "step": 187710
    },
    {
      "epoch": 22.589651022864018,
      "grad_norm": Infinity,
      "learning_rate": 0.00017684493396397127,
      "loss": 11.1379,
      "step": 187720
    },
    {
      "epoch": 22.590854392298436,
      "grad_norm": Infinity,
      "learning_rate": 0.00017684249951566882,
      "loss": 11.1374,
      "step": 187730
    },
    {
      "epoch": 22.592057761732853,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768400649561563,
      "loss": 11.1715,
      "step": 187740
    },
    {
      "epoch": 22.593261131167267,
      "grad_norm": Infinity,
      "learning_rate": 0.00017683763028543723,
      "loss": 11.146,
      "step": 187750
    },
    {
      "epoch": 22.594464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00017683519550351515,
      "loss": 11.1752,
      "step": 187760
    },
    {
      "epoch": 22.595667870036102,
      "grad_norm": Infinity,
      "learning_rate": 0.00017683276061039357,
      "loss": 11.1637,
      "step": 187770
    },
    {
      "epoch": 22.596871239470516,
      "grad_norm": Infinity,
      "learning_rate": 0.000176830325606076,
      "loss": 11.0903,
      "step": 187780
    },
    {
      "epoch": 22.598074608904934,
      "grad_norm": Infinity,
      "learning_rate": 0.00017682789049056603,
      "loss": 11.1792,
      "step": 187790
    },
    {
      "epoch": 22.59927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00017682545526386708,
      "loss": 11.1525,
      "step": 187800
    },
    {
      "epoch": 22.600481347773766,
      "grad_norm": Infinity,
      "learning_rate": 0.00017682301992598276,
      "loss": 11.21,
      "step": 187810
    },
    {
      "epoch": 22.601684717208183,
      "grad_norm": Infinity,
      "learning_rate": 0.00017682058447691655,
      "loss": 11.1317,
      "step": 187820
    },
    {
      "epoch": 22.6028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.000176818148916672,
      "loss": 11.0641,
      "step": 187830
    },
    {
      "epoch": 22.604091456077015,
      "grad_norm": Infinity,
      "learning_rate": 0.00017681571324525258,
      "loss": 11.0839,
      "step": 187840
    },
    {
      "epoch": 22.605294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768132774626619,
      "loss": 11.2821,
      "step": 187850
    },
    {
      "epoch": 22.60649819494585,
      "grad_norm": Infinity,
      "learning_rate": 0.00017681084156890344,
      "loss": 11.1419,
      "step": 187860
    },
    {
      "epoch": 22.607701564380264,
      "grad_norm": Infinity,
      "learning_rate": 0.00017680840556398073,
      "loss": 11.1512,
      "step": 187870
    },
    {
      "epoch": 22.60890493381468,
      "grad_norm": Infinity,
      "learning_rate": 0.00017680596944789727,
      "loss": 11.158,
      "step": 187880
    },
    {
      "epoch": 22.6101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.00017680353322065663,
      "loss": 11.1676,
      "step": 187890
    },
    {
      "epoch": 22.611311672683513,
      "grad_norm": Infinity,
      "learning_rate": 0.0001768010968822623,
      "loss": 11.1941,
      "step": 187900
    },
    {
      "epoch": 22.61251504211793,
      "grad_norm": Infinity,
      "learning_rate": 0.00017679866043271786,
      "loss": 11.1175,
      "step": 187910
    },
    {
      "epoch": 22.613718411552348,
      "grad_norm": Infinity,
      "learning_rate": 0.00017679622387202674,
      "loss": 11.1983,
      "step": 187920
    },
    {
      "epoch": 22.614921780986762,
      "grad_norm": Infinity,
      "learning_rate": 0.00017679378720019258,
      "loss": 11.0447,
      "step": 187930
    },
    {
      "epoch": 22.61612515042118,
      "grad_norm": Infinity,
      "learning_rate": 0.00017679135041721882,
      "loss": 11.2145,
      "step": 187940
    },
    {
      "epoch": 22.617328519855597,
      "grad_norm": Infinity,
      "learning_rate": 0.00017678891352310904,
      "loss": 11.1853,
      "step": 187950
    },
    {
      "epoch": 22.61853188929001,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767864765178667,
      "loss": 11.1153,
      "step": 187960
    },
    {
      "epoch": 22.61973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.00017678403940149537,
      "loss": 11.1473,
      "step": 187970
    },
    {
      "epoch": 22.620938628158846,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767816021739986,
      "loss": 11.2434,
      "step": 187980
    },
    {
      "epoch": 22.62214199759326,
      "grad_norm": Infinity,
      "learning_rate": 0.00017677916483537992,
      "loss": 11.206,
      "step": 187990
    },
    {
      "epoch": 22.623345367027678,
      "grad_norm": Infinity,
      "learning_rate": 0.00017677672738564278,
      "loss": 11.1623,
      "step": 188000
    },
    {
      "epoch": 22.624548736462096,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767742898247908,
      "loss": 11.1728,
      "step": 188010
    },
    {
      "epoch": 22.62575210589651,
      "grad_norm": Infinity,
      "learning_rate": 0.00017677185215282746,
      "loss": 11.106,
      "step": 188020
    },
    {
      "epoch": 22.626955475330927,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767694143697563,
      "loss": 10.9994,
      "step": 188030
    },
    {
      "epoch": 22.628158844765345,
      "grad_norm": Infinity,
      "learning_rate": 0.00017676697647558084,
      "loss": 11.1421,
      "step": 188040
    },
    {
      "epoch": 22.62936221419976,
      "grad_norm": Infinity,
      "learning_rate": 0.00017676453847030458,
      "loss": 11.2661,
      "step": 188050
    },
    {
      "epoch": 22.630565583634176,
      "grad_norm": Infinity,
      "learning_rate": 0.00017676210035393112,
      "loss": 11.1114,
      "step": 188060
    },
    {
      "epoch": 22.63176895306859,
      "grad_norm": Infinity,
      "learning_rate": 0.00017675966212646392,
      "loss": 11.1273,
      "step": 188070
    },
    {
      "epoch": 22.632972322503008,
      "grad_norm": Infinity,
      "learning_rate": 0.00017675722378790657,
      "loss": 11.1042,
      "step": 188080
    },
    {
      "epoch": 22.634175691937426,
      "grad_norm": Infinity,
      "learning_rate": 0.00017675478533826254,
      "loss": 11.1341,
      "step": 188090
    },
    {
      "epoch": 22.63537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.00017675234677753542,
      "loss": 11.1282,
      "step": 188100
    },
    {
      "epoch": 22.636582430806257,
      "grad_norm": Infinity,
      "learning_rate": 0.00017674990810572866,
      "loss": 11.075,
      "step": 188110
    },
    {
      "epoch": 22.637785800240675,
      "grad_norm": Infinity,
      "learning_rate": 0.00017674746932284588,
      "loss": 11.1286,
      "step": 188120
    },
    {
      "epoch": 22.63898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00017674503042889052,
      "loss": 11.0815,
      "step": 188130
    },
    {
      "epoch": 22.640192539109506,
      "grad_norm": Infinity,
      "learning_rate": 0.00017674259142386618,
      "loss": 11.1622,
      "step": 188140
    },
    {
      "epoch": 22.641395908543924,
      "grad_norm": Infinity,
      "learning_rate": 0.00017674015230777635,
      "loss": 11.1906,
      "step": 188150
    },
    {
      "epoch": 22.642599277978338,
      "grad_norm": Infinity,
      "learning_rate": 0.00017673771308062458,
      "loss": 11.1892,
      "step": 188160
    },
    {
      "epoch": 22.643802647412755,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767352737424144,
      "loss": 11.1136,
      "step": 188170
    },
    {
      "epoch": 22.645006016847173,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767328342931493,
      "loss": 11.2288,
      "step": 188180
    },
    {
      "epoch": 22.646209386281587,
      "grad_norm": Infinity,
      "learning_rate": 0.00017673039473283287,
      "loss": 11.1405,
      "step": 188190
    },
    {
      "epoch": 22.647412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767279550614686,
      "loss": 11.1905,
      "step": 188200
    },
    {
      "epoch": 22.648616125150422,
      "grad_norm": Infinity,
      "learning_rate": 0.00017672551527906006,
      "loss": 11.183,
      "step": 188210
    },
    {
      "epoch": 22.649819494584836,
      "grad_norm": Infinity,
      "learning_rate": 0.00017672307538561075,
      "loss": 11.0984,
      "step": 188220
    },
    {
      "epoch": 22.651022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767206353811242,
      "loss": 11.0781,
      "step": 188230
    },
    {
      "epoch": 22.65222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.00017671819526560395,
      "loss": 11.2771,
      "step": 188240
    },
    {
      "epoch": 22.653429602888085,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767157550390535,
      "loss": 11.0902,
      "step": 188250
    },
    {
      "epoch": 22.654632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00017671331470147647,
      "loss": 11.155,
      "step": 188260
    },
    {
      "epoch": 22.65583634175692,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767108742528763,
      "loss": 11.2088,
      "step": 188270
    },
    {
      "epoch": 22.657039711191334,
      "grad_norm": Infinity,
      "learning_rate": 0.00017670843369325655,
      "loss": 11.0392,
      "step": 188280
    },
    {
      "epoch": 22.658243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767059930226208,
      "loss": 11.163,
      "step": 188290
    },
    {
      "epoch": 22.65944645006017,
      "grad_norm": Infinity,
      "learning_rate": 0.0001767035522409725,
      "loss": 11.1256,
      "step": 188300
    },
    {
      "epoch": 22.660649819494584,
      "grad_norm": Infinity,
      "learning_rate": 0.00017670111134831525,
      "loss": 11.1662,
      "step": 188310
    },
    {
      "epoch": 22.661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00017669867034465252,
      "loss": 11.0707,
      "step": 188320
    },
    {
      "epoch": 22.66305655836342,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766962292299879,
      "loss": 11.0481,
      "step": 188330
    },
    {
      "epoch": 22.664259927797833,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766937880043249,
      "loss": 11.1178,
      "step": 188340
    },
    {
      "epoch": 22.66546329723225,
      "grad_norm": Infinity,
      "learning_rate": 0.00017669134666766705,
      "loss": 11.1569,
      "step": 188350
    },
    {
      "epoch": 22.666666666666668,
      "grad_norm": Infinity,
      "learning_rate": 0.00017668890522001788,
      "loss": 11.1828,
      "step": 188360
    },
    {
      "epoch": 22.667870036101082,
      "grad_norm": Infinity,
      "learning_rate": 0.00017668646366138092,
      "loss": 11.1584,
      "step": 188370
    },
    {
      "epoch": 22.6690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.00017668402199175976,
      "loss": 11.1627,
      "step": 188380
    },
    {
      "epoch": 22.670276774969917,
      "grad_norm": Infinity,
      "learning_rate": 0.00017668158021115783,
      "loss": 11.1631,
      "step": 188390
    },
    {
      "epoch": 22.67148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00017667913831957875,
      "loss": 11.077,
      "step": 188400
    },
    {
      "epoch": 22.67268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.00017667669631702605,
      "loss": 11.0115,
      "step": 188410
    },
    {
      "epoch": 22.673886883273166,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766742542035032,
      "loss": 11.1311,
      "step": 188420
    },
    {
      "epoch": 22.67509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766718119790138,
      "loss": 11.0342,
      "step": 188430
    },
    {
      "epoch": 22.676293622141998,
      "grad_norm": Infinity,
      "learning_rate": 0.00017666936964356132,
      "loss": 11.1917,
      "step": 188440
    },
    {
      "epoch": 22.677496991576415,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766669271971494,
      "loss": 11.1531,
      "step": 188450
    },
    {
      "epoch": 22.67870036101083,
      "grad_norm": Infinity,
      "learning_rate": 0.00017666448463978145,
      "loss": 11.0852,
      "step": 188460
    },
    {
      "epoch": 22.679903730445247,
      "grad_norm": Infinity,
      "learning_rate": 0.00017666204197146107,
      "loss": 11.1602,
      "step": 188470
    },
    {
      "epoch": 22.681107099879664,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766595991921918,
      "loss": 11.0548,
      "step": 188480
    },
    {
      "epoch": 22.68231046931408,
      "grad_norm": Infinity,
      "learning_rate": 0.00017665715630197718,
      "loss": 10.9654,
      "step": 188490
    },
    {
      "epoch": 22.683513838748496,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766547133008207,
      "loss": 11.1588,
      "step": 188500
    },
    {
      "epoch": 22.684717208182914,
      "grad_norm": Infinity,
      "learning_rate": 0.00017665227018872595,
      "loss": 11.0222,
      "step": 188510
    },
    {
      "epoch": 22.685920577617328,
      "grad_norm": Infinity,
      "learning_rate": 0.00017664982696569644,
      "loss": 11.0054,
      "step": 188520
    },
    {
      "epoch": 22.687123947051745,
      "grad_norm": Infinity,
      "learning_rate": 0.00017664738363173566,
      "loss": 10.9621,
      "step": 188530
    },
    {
      "epoch": 22.688327316486163,
      "grad_norm": Infinity,
      "learning_rate": 0.00017664494018684724,
      "loss": 11.2123,
      "step": 188540
    },
    {
      "epoch": 22.689530685920577,
      "grad_norm": Infinity,
      "learning_rate": 0.00017664249663103463,
      "loss": 11.2189,
      "step": 188550
    },
    {
      "epoch": 22.690734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.00017664005296430145,
      "loss": 11.0421,
      "step": 188560
    },
    {
      "epoch": 22.691937424789412,
      "grad_norm": Infinity,
      "learning_rate": 0.00017663760918665115,
      "loss": 11.2462,
      "step": 188570
    },
    {
      "epoch": 22.693140794223826,
      "grad_norm": Infinity,
      "learning_rate": 0.00017663516529808733,
      "loss": 11.0542,
      "step": 188580
    },
    {
      "epoch": 22.694344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766327212986135,
      "loss": 11.2431,
      "step": 188590
    },
    {
      "epoch": 22.69554753309266,
      "grad_norm": Infinity,
      "learning_rate": 0.00017663027718823322,
      "loss": 11.1883,
      "step": 188600
    },
    {
      "epoch": 22.696750902527075,
      "grad_norm": Infinity,
      "learning_rate": 0.00017662783296695,
      "loss": 11.1646,
      "step": 188610
    },
    {
      "epoch": 22.697954271961493,
      "grad_norm": Infinity,
      "learning_rate": 0.00017662538863476738,
      "loss": 11.2338,
      "step": 188620
    },
    {
      "epoch": 22.69915764139591,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766229441916889,
      "loss": 11.1293,
      "step": 188630
    },
    {
      "epoch": 22.700361010830324,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766204996377181,
      "loss": 11.0257,
      "step": 188640
    },
    {
      "epoch": 22.70156438026474,
      "grad_norm": Infinity,
      "learning_rate": 0.00017661805497285854,
      "loss": 11.1802,
      "step": 188650
    },
    {
      "epoch": 22.70276774969916,
      "grad_norm": Infinity,
      "learning_rate": 0.00017661561019711373,
      "loss": 11.1375,
      "step": 188660
    },
    {
      "epoch": 22.703971119133573,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766131653104872,
      "loss": 11.1061,
      "step": 188670
    },
    {
      "epoch": 22.70517448856799,
      "grad_norm": Infinity,
      "learning_rate": 0.00017661072031298252,
      "loss": 11.0686,
      "step": 188680
    },
    {
      "epoch": 22.706377858002405,
      "grad_norm": Infinity,
      "learning_rate": 0.00017660827520460322,
      "loss": 11.078,
      "step": 188690
    },
    {
      "epoch": 22.707581227436823,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766058299853528,
      "loss": 11.1074,
      "step": 188700
    },
    {
      "epoch": 22.70878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00017660338465523486,
      "loss": 11.1721,
      "step": 188710
    },
    {
      "epoch": 22.709987966305654,
      "grad_norm": Infinity,
      "learning_rate": 0.0001766009392142529,
      "loss": 11.1626,
      "step": 188720
    },
    {
      "epoch": 22.71119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.00017659849366241046,
      "loss": 11.1575,
      "step": 188730
    },
    {
      "epoch": 22.71239470517449,
      "grad_norm": Infinity,
      "learning_rate": 0.0001765960479997111,
      "loss": 11.1587,
      "step": 188740
    },
    {
      "epoch": 22.713598074608903,
      "grad_norm": Infinity,
      "learning_rate": 0.00017659360222615835,
      "loss": 11.2074,
      "step": 188750
    },
    {
      "epoch": 22.71480144404332,
      "grad_norm": Infinity,
      "learning_rate": 0.00017659115634175576,
      "loss": 11.133,
      "step": 188760
    },
    {
      "epoch": 22.71600481347774,
      "grad_norm": Infinity,
      "learning_rate": 0.00017658871034650683,
      "loss": 11.107,
      "step": 188770
    },
    {
      "epoch": 22.717208182912152,
      "grad_norm": Infinity,
      "learning_rate": 0.00017658626424041515,
      "loss": 11.1735,
      "step": 188780
    },
    {
      "epoch": 22.71841155234657,
      "grad_norm": Infinity,
      "learning_rate": 0.00017658381802348422,
      "loss": 11.0705,
      "step": 188790
    },
    {
      "epoch": 22.719614921780988,
      "grad_norm": Infinity,
      "learning_rate": 0.0001765813716957176,
      "loss": 11.1287,
      "step": 188800
    },
    {
      "epoch": 22.7208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00017657892525711882,
      "loss": 11.1294,
      "step": 188810
    },
    {
      "epoch": 22.72202166064982,
      "grad_norm": Infinity,
      "learning_rate": 0.00017657647870769145,
      "loss": 11.0942,
      "step": 188820
    },
    {
      "epoch": 22.723225030084237,
      "grad_norm": Infinity,
      "learning_rate": 0.00017657403204743898,
      "loss": 11.2001,
      "step": 188830
    },
    {
      "epoch": 22.72442839951865,
      "grad_norm": Infinity,
      "learning_rate": 0.000176571585276365,
      "loss": 11.2361,
      "step": 188840
    },
    {
      "epoch": 22.72563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00017656913839447306,
      "loss": 11.1385,
      "step": 188850
    },
    {
      "epoch": 22.726835138387486,
      "grad_norm": Infinity,
      "learning_rate": 0.00017656669140176664,
      "loss": 11.0191,
      "step": 188860
    },
    {
      "epoch": 22.7280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.00017656424429824929,
      "loss": 11.0866,
      "step": 188870
    },
    {
      "epoch": 22.729241877256317,
      "grad_norm": Infinity,
      "learning_rate": 0.00017656179708392463,
      "loss": 11.0602,
      "step": 188880
    },
    {
      "epoch": 22.730445246690735,
      "grad_norm": Infinity,
      "learning_rate": 0.00017655934975879608,
      "loss": 11.1853,
      "step": 188890
    },
    {
      "epoch": 22.73164861612515,
      "grad_norm": Infinity,
      "learning_rate": 0.0001765569023228673,
      "loss": 11.0442,
      "step": 188900
    },
    {
      "epoch": 22.732851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.00017655445477614176,
      "loss": 11.1074,
      "step": 188910
    },
    {
      "epoch": 22.734055354993984,
      "grad_norm": Infinity,
      "learning_rate": 0.00017655200711862303,
      "loss": 11.2331,
      "step": 188920
    },
    {
      "epoch": 22.735258724428398,
      "grad_norm": Infinity,
      "learning_rate": 0.00017654955935031465,
      "loss": 11.1989,
      "step": 188930
    },
    {
      "epoch": 22.736462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.00017654711147122016,
      "loss": 11.1129,
      "step": 188940
    },
    {
      "epoch": 22.737665463297233,
      "grad_norm": Infinity,
      "learning_rate": 0.0001765446634813431,
      "loss": 11.0734,
      "step": 188950
    },
    {
      "epoch": 22.738868832731647,
      "grad_norm": Infinity,
      "learning_rate": 0.000176542215380687,
      "loss": 11.1119,
      "step": 188960
    },
    {
      "epoch": 22.740072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00017653976716925542,
      "loss": 11.1422,
      "step": 188970
    },
    {
      "epoch": 22.741275571600482,
      "grad_norm": Infinity,
      "learning_rate": 0.00017653731884705192,
      "loss": 11.1614,
      "step": 188980
    },
    {
      "epoch": 22.742478941034896,
      "grad_norm": Infinity,
      "learning_rate": 0.00017653487041408,
      "loss": 11.1652,
      "step": 188990
    },
    {
      "epoch": 22.743682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00017653242187034323,
      "loss": 11.0447,
      "step": 189000
    },
    {
      "epoch": 22.74488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.00017652997321584516,
      "loss": 11.1534,
      "step": 189010
    },
    {
      "epoch": 22.746089049338146,
      "grad_norm": Infinity,
      "learning_rate": 0.00017652752445058932,
      "loss": 11.1196,
      "step": 189020
    },
    {
      "epoch": 22.747292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00017652507557457928,
      "loss": 11.1514,
      "step": 189030
    },
    {
      "epoch": 22.74849578820698,
      "grad_norm": Infinity,
      "learning_rate": 0.00017652262658781852,
      "loss": 11.103,
      "step": 189040
    },
    {
      "epoch": 22.749699157641395,
      "grad_norm": Infinity,
      "learning_rate": 0.00017652017749031064,
      "loss": 11.2375,
      "step": 189050
    },
    {
      "epoch": 22.750902527075812,
      "grad_norm": Infinity,
      "learning_rate": 0.0001765177282820592,
      "loss": 11.2562,
      "step": 189060
    },
    {
      "epoch": 22.75210589651023,
      "grad_norm": Infinity,
      "learning_rate": 0.0001765152789630677,
      "loss": 11.1239,
      "step": 189070
    },
    {
      "epoch": 22.753309265944644,
      "grad_norm": Infinity,
      "learning_rate": 0.00017651282953333968,
      "loss": 11.0871,
      "step": 189080
    },
    {
      "epoch": 22.75451263537906,
      "grad_norm": Infinity,
      "learning_rate": 0.00017651037999287872,
      "loss": 11.1931,
      "step": 189090
    },
    {
      "epoch": 22.75571600481348,
      "grad_norm": Infinity,
      "learning_rate": 0.00017650793034168833,
      "loss": 11.2694,
      "step": 189100
    },
    {
      "epoch": 22.756919374247893,
      "grad_norm": Infinity,
      "learning_rate": 0.0001765054805797721,
      "loss": 11.1285,
      "step": 189110
    },
    {
      "epoch": 22.75812274368231,
      "grad_norm": Infinity,
      "learning_rate": 0.00017650303070713353,
      "loss": 11.1306,
      "step": 189120
    },
    {
      "epoch": 22.759326113116728,
      "grad_norm": Infinity,
      "learning_rate": 0.00017650058072377623,
      "loss": 11.1381,
      "step": 189130
    },
    {
      "epoch": 22.760529482551142,
      "grad_norm": Infinity,
      "learning_rate": 0.00017649813062970367,
      "loss": 11.164,
      "step": 189140
    },
    {
      "epoch": 22.76173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00017649568042491943,
      "loss": 11.0868,
      "step": 189150
    },
    {
      "epoch": 22.762936221419977,
      "grad_norm": Infinity,
      "learning_rate": 0.00017649323010942707,
      "loss": 11.1251,
      "step": 189160
    },
    {
      "epoch": 22.76413959085439,
      "grad_norm": Infinity,
      "learning_rate": 0.0001764907796832301,
      "loss": 11.175,
      "step": 189170
    },
    {
      "epoch": 22.76534296028881,
      "grad_norm": Infinity,
      "learning_rate": 0.00017648832914633208,
      "loss": 11.1305,
      "step": 189180
    },
    {
      "epoch": 22.766546329723226,
      "grad_norm": Infinity,
      "learning_rate": 0.0001764858784987366,
      "loss": 11.1068,
      "step": 189190
    },
    {
      "epoch": 22.76774969915764,
      "grad_norm": Infinity,
      "learning_rate": 0.00017648342774044716,
      "loss": 11.0022,
      "step": 189200
    },
    {
      "epoch": 22.768953068592058,
      "grad_norm": Infinity,
      "learning_rate": 0.0001764809768714673,
      "loss": 11.1103,
      "step": 189210
    },
    {
      "epoch": 22.770156438026476,
      "grad_norm": Infinity,
      "learning_rate": 0.00017647852589180058,
      "loss": 11.2056,
      "step": 189220
    },
    {
      "epoch": 22.77135980746089,
      "grad_norm": Infinity,
      "learning_rate": 0.00017647607480145055,
      "loss": 11.1104,
      "step": 189230
    },
    {
      "epoch": 22.772563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.0001764736236004208,
      "loss": 11.2225,
      "step": 189240
    },
    {
      "epoch": 22.773766546329725,
      "grad_norm": Infinity,
      "learning_rate": 0.00017647117228871477,
      "loss": 11.2761,
      "step": 189250
    },
    {
      "epoch": 22.77496991576414,
      "grad_norm": Infinity,
      "learning_rate": 0.00017646872086633612,
      "loss": 11.1101,
      "step": 189260
    },
    {
      "epoch": 22.776173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00017646626933328833,
      "loss": 11.0977,
      "step": 189270
    },
    {
      "epoch": 22.777376654632974,
      "grad_norm": Infinity,
      "learning_rate": 0.00017646381768957498,
      "loss": 11.1827,
      "step": 189280
    },
    {
      "epoch": 22.778580024067388,
      "grad_norm": Infinity,
      "learning_rate": 0.0001764613659351996,
      "loss": 11.0711,
      "step": 189290
    },
    {
      "epoch": 22.779783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.00017645891407016577,
      "loss": 11.1122,
      "step": 189300
    },
    {
      "epoch": 22.780986762936223,
      "grad_norm": Infinity,
      "learning_rate": 0.000176456462094477,
      "loss": 11.1671,
      "step": 189310
    },
    {
      "epoch": 22.782190132370637,
      "grad_norm": Infinity,
      "learning_rate": 0.00017645401000813683,
      "loss": 11.0999,
      "step": 189320
    },
    {
      "epoch": 22.783393501805055,
      "grad_norm": Infinity,
      "learning_rate": 0.00017645155781114886,
      "loss": 11.1402,
      "step": 189330
    },
    {
      "epoch": 22.784596871239472,
      "grad_norm": Infinity,
      "learning_rate": 0.00017644910550351659,
      "loss": 11.1618,
      "step": 189340
    },
    {
      "epoch": 22.785800240673886,
      "grad_norm": Infinity,
      "learning_rate": 0.00017644665308524362,
      "loss": 11.1845,
      "step": 189350
    },
    {
      "epoch": 22.787003610108304,
      "grad_norm": Infinity,
      "learning_rate": 0.00017644420055633344,
      "loss": 11.2174,
      "step": 189360
    },
    {
      "epoch": 22.78820697954272,
      "grad_norm": Infinity,
      "learning_rate": 0.00017644174791678964,
      "loss": 11.1245,
      "step": 189370
    },
    {
      "epoch": 22.789410348977135,
      "grad_norm": Infinity,
      "learning_rate": 0.00017643929516661575,
      "loss": 11.1825,
      "step": 189380
    },
    {
      "epoch": 22.790613718411553,
      "grad_norm": Infinity,
      "learning_rate": 0.00017643684230581535,
      "loss": 11.1393,
      "step": 189390
    },
    {
      "epoch": 22.79181708784597,
      "grad_norm": Infinity,
      "learning_rate": 0.00017643438933439195,
      "loss": 10.9904,
      "step": 189400
    },
    {
      "epoch": 22.793020457280385,
      "grad_norm": Infinity,
      "learning_rate": 0.00017643193625234912,
      "loss": 11.1743,
      "step": 189410
    },
    {
      "epoch": 22.794223826714802,
      "grad_norm": Infinity,
      "learning_rate": 0.0001764294830596904,
      "loss": 11.2213,
      "step": 189420
    },
    {
      "epoch": 22.79542719614922,
      "grad_norm": Infinity,
      "learning_rate": 0.00017642702975641936,
      "loss": 11.2268,
      "step": 189430
    },
    {
      "epoch": 22.796630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00017642457634253953,
      "loss": 11.1792,
      "step": 189440
    },
    {
      "epoch": 22.79783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.00017642212281805447,
      "loss": 11.1201,
      "step": 189450
    },
    {
      "epoch": 22.799037304452465,
      "grad_norm": Infinity,
      "learning_rate": 0.00017641966918296773,
      "loss": 11.0136,
      "step": 189460
    },
    {
      "epoch": 22.800240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.00017641721543728287,
      "loss": 11.1602,
      "step": 189470
    },
    {
      "epoch": 22.8014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017641476158100343,
      "loss": 11.0964,
      "step": 189480
    },
    {
      "epoch": 22.802647412755714,
      "grad_norm": Infinity,
      "learning_rate": 0.00017641230761413296,
      "loss": 11.1713,
      "step": 189490
    },
    {
      "epoch": 22.803850782190132,
      "grad_norm": Infinity,
      "learning_rate": 0.00017640985353667502,
      "loss": 11.0016,
      "step": 189500
    },
    {
      "epoch": 22.80505415162455,
      "grad_norm": Infinity,
      "learning_rate": 0.00017640739934863314,
      "loss": 11.2487,
      "step": 189510
    },
    {
      "epoch": 22.806257521058964,
      "grad_norm": Infinity,
      "learning_rate": 0.00017640494505001092,
      "loss": 11.2146,
      "step": 189520
    },
    {
      "epoch": 22.80746089049338,
      "grad_norm": Infinity,
      "learning_rate": 0.00017640249064081184,
      "loss": 11.1064,
      "step": 189530
    },
    {
      "epoch": 22.8086642599278,
      "grad_norm": Infinity,
      "learning_rate": 0.00017640003612103951,
      "loss": 11.117,
      "step": 189540
    },
    {
      "epoch": 22.809867629362213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017639758149069748,
      "loss": 11.0608,
      "step": 189550
    },
    {
      "epoch": 22.81107099879663,
      "grad_norm": Infinity,
      "learning_rate": 0.00017639512674978927,
      "loss": 11.1272,
      "step": 189560
    },
    {
      "epoch": 22.812274368231048,
      "grad_norm": Infinity,
      "learning_rate": 0.00017639267189831846,
      "loss": 11.0979,
      "step": 189570
    },
    {
      "epoch": 22.813477737665462,
      "grad_norm": Infinity,
      "learning_rate": 0.00017639021693628858,
      "loss": 11.0264,
      "step": 189580
    },
    {
      "epoch": 22.81468110709988,
      "grad_norm": Infinity,
      "learning_rate": 0.00017638776186370322,
      "loss": 11.1974,
      "step": 189590
    },
    {
      "epoch": 22.815884476534297,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763853066805659,
      "loss": 11.2255,
      "step": 189600
    },
    {
      "epoch": 22.81708784596871,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763828513868802,
      "loss": 11.0818,
      "step": 189610
    },
    {
      "epoch": 22.81829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.00017638039598264963,
      "loss": 11.147,
      "step": 189620
    },
    {
      "epoch": 22.819494584837546,
      "grad_norm": Infinity,
      "learning_rate": 0.00017637794046787778,
      "loss": 11.1031,
      "step": 189630
    },
    {
      "epoch": 22.82069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00017637548484256818,
      "loss": 11.2972,
      "step": 189640
    },
    {
      "epoch": 22.821901323706378,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763730291067244,
      "loss": 11.165,
      "step": 189650
    },
    {
      "epoch": 22.823104693140795,
      "grad_norm": Infinity,
      "learning_rate": 0.00017637057326035,
      "loss": 11.1457,
      "step": 189660
    },
    {
      "epoch": 22.82430806257521,
      "grad_norm": Infinity,
      "learning_rate": 0.00017636811730344852,
      "loss": 11.2185,
      "step": 189670
    },
    {
      "epoch": 22.825511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763656612360235,
      "loss": 11.1695,
      "step": 189680
    },
    {
      "epoch": 22.826714801444044,
      "grad_norm": Infinity,
      "learning_rate": 0.00017636320505807853,
      "loss": 11.1892,
      "step": 189690
    },
    {
      "epoch": 22.82791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00017636074876961716,
      "loss": 11.0703,
      "step": 189700
    },
    {
      "epoch": 22.829121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00017635829237064294,
      "loss": 11.1229,
      "step": 189710
    },
    {
      "epoch": 22.830324909747294,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763558358611594,
      "loss": 11.1212,
      "step": 189720
    },
    {
      "epoch": 22.831528279181708,
      "grad_norm": Infinity,
      "learning_rate": 0.00017635337924117012,
      "loss": 11.0298,
      "step": 189730
    },
    {
      "epoch": 22.832731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.00017635092251067866,
      "loss": 11.0389,
      "step": 189740
    },
    {
      "epoch": 22.833935018050543,
      "grad_norm": Infinity,
      "learning_rate": 0.00017634846566968855,
      "loss": 11.1126,
      "step": 189750
    },
    {
      "epoch": 22.835138387484957,
      "grad_norm": Infinity,
      "learning_rate": 0.00017634600871820335,
      "loss": 11.0647,
      "step": 189760
    },
    {
      "epoch": 22.836341756919374,
      "grad_norm": Infinity,
      "learning_rate": 0.00017634355165622662,
      "loss": 11.1533,
      "step": 189770
    },
    {
      "epoch": 22.837545126353792,
      "grad_norm": Infinity,
      "learning_rate": 0.00017634109448376196,
      "loss": 11.0943,
      "step": 189780
    },
    {
      "epoch": 22.838748495788206,
      "grad_norm": Infinity,
      "learning_rate": 0.00017633863720081284,
      "loss": 11.1534,
      "step": 189790
    },
    {
      "epoch": 22.839951865222623,
      "grad_norm": Infinity,
      "learning_rate": 0.00017633617980738293,
      "loss": 11.072,
      "step": 189800
    },
    {
      "epoch": 22.84115523465704,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763337223034757,
      "loss": 11.2568,
      "step": 189810
    },
    {
      "epoch": 22.842358604091455,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763312646890947,
      "loss": 11.2142,
      "step": 189820
    },
    {
      "epoch": 22.843561973525873,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763288069642435,
      "loss": 11.1366,
      "step": 189830
    },
    {
      "epoch": 22.84476534296029,
      "grad_norm": Infinity,
      "learning_rate": 0.00017632634912892573,
      "loss": 11.0713,
      "step": 189840
    },
    {
      "epoch": 22.845968712394704,
      "grad_norm": Infinity,
      "learning_rate": 0.00017632389118314484,
      "loss": 11.2236,
      "step": 189850
    },
    {
      "epoch": 22.84717208182912,
      "grad_norm": Infinity,
      "learning_rate": 0.00017632143312690447,
      "loss": 11.1332,
      "step": 189860
    },
    {
      "epoch": 22.84837545126354,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763189749602081,
      "loss": 11.1097,
      "step": 189870
    },
    {
      "epoch": 22.849578820697953,
      "grad_norm": Infinity,
      "learning_rate": 0.00017631651668305939,
      "loss": 10.9052,
      "step": 189880
    },
    {
      "epoch": 22.85078219013237,
      "grad_norm": Infinity,
      "learning_rate": 0.00017631405829546178,
      "loss": 11.1419,
      "step": 189890
    },
    {
      "epoch": 22.85198555956679,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763115997974189,
      "loss": 11.1203,
      "step": 189900
    },
    {
      "epoch": 22.853188929001202,
      "grad_norm": Infinity,
      "learning_rate": 0.00017630914118893433,
      "loss": 11.1216,
      "step": 189910
    },
    {
      "epoch": 22.85439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00017630668247001156,
      "loss": 11.1302,
      "step": 189920
    },
    {
      "epoch": 22.855595667870038,
      "grad_norm": Infinity,
      "learning_rate": 0.0001763042236406542,
      "loss": 11.1505,
      "step": 189930
    },
    {
      "epoch": 22.85679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00017630176470086577,
      "loss": 11.1267,
      "step": 189940
    },
    {
      "epoch": 22.85800240673887,
      "grad_norm": Infinity,
      "learning_rate": 0.00017629930565064985,
      "loss": 11.14,
      "step": 189950
    },
    {
      "epoch": 22.859205776173287,
      "grad_norm": Infinity,
      "learning_rate": 0.00017629684649001003,
      "loss": 11.0138,
      "step": 189960
    },
    {
      "epoch": 22.8604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762943872189498,
      "loss": 11.1343,
      "step": 189970
    },
    {
      "epoch": 22.86161251504212,
      "grad_norm": Infinity,
      "learning_rate": 0.00017629192783747276,
      "loss": 11.0876,
      "step": 189980
    },
    {
      "epoch": 22.862815884476536,
      "grad_norm": Infinity,
      "learning_rate": 0.00017628946834558247,
      "loss": 11.1311,
      "step": 189990
    },
    {
      "epoch": 22.86401925391095,
      "grad_norm": Infinity,
      "learning_rate": 0.00017628700874328248,
      "loss": 11.0517,
      "step": 190000
    },
    {
      "epoch": 22.865222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.00017628454903057637,
      "loss": 11.1689,
      "step": 190010
    },
    {
      "epoch": 22.866425992779785,
      "grad_norm": Infinity,
      "learning_rate": 0.00017628208920746766,
      "loss": 11.2026,
      "step": 190020
    },
    {
      "epoch": 22.8676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00017627962927395995,
      "loss": 11.1512,
      "step": 190030
    },
    {
      "epoch": 22.868832731648617,
      "grad_norm": Infinity,
      "learning_rate": 0.00017627716923005677,
      "loss": 11.2832,
      "step": 190040
    },
    {
      "epoch": 22.870036101083034,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762747090757617,
      "loss": 11.0216,
      "step": 190050
    },
    {
      "epoch": 22.871239470517448,
      "grad_norm": Infinity,
      "learning_rate": 0.00017627224881107832,
      "loss": 11.0908,
      "step": 190060
    },
    {
      "epoch": 22.872442839951866,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762697884360101,
      "loss": 11.1863,
      "step": 190070
    },
    {
      "epoch": 22.87364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.00017626732795056072,
      "loss": 11.1152,
      "step": 190080
    },
    {
      "epoch": 22.874849578820697,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762648673547337,
      "loss": 11.2341,
      "step": 190090
    },
    {
      "epoch": 22.876052948255115,
      "grad_norm": Infinity,
      "learning_rate": 0.00017626240664853256,
      "loss": 11.1993,
      "step": 190100
    },
    {
      "epoch": 22.87725631768953,
      "grad_norm": Infinity,
      "learning_rate": 0.00017625994583196088,
      "loss": 11.2688,
      "step": 190110
    },
    {
      "epoch": 22.878459687123947,
      "grad_norm": Infinity,
      "learning_rate": 0.00017625748490502223,
      "loss": 11.0799,
      "step": 190120
    },
    {
      "epoch": 22.879663056558364,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762550238677202,
      "loss": 11.1661,
      "step": 190130
    },
    {
      "epoch": 22.880866425992778,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762525627200583,
      "loss": 11.1794,
      "step": 190140
    },
    {
      "epoch": 22.882069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00017625010146204013,
      "loss": 11.2047,
      "step": 190150
    },
    {
      "epoch": 22.883273164861613,
      "grad_norm": Infinity,
      "learning_rate": 0.00017624764009366924,
      "loss": 11.1169,
      "step": 190160
    },
    {
      "epoch": 22.884476534296027,
      "grad_norm": Infinity,
      "learning_rate": 0.00017624517861494917,
      "loss": 11.0868,
      "step": 190170
    },
    {
      "epoch": 22.885679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00017624271702588348,
      "loss": 11.2496,
      "step": 190180
    },
    {
      "epoch": 22.886883273164862,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762402553264758,
      "loss": 11.1704,
      "step": 190190
    },
    {
      "epoch": 22.888086642599276,
      "grad_norm": Infinity,
      "learning_rate": 0.00017623779351672963,
      "loss": 11.1896,
      "step": 190200
    },
    {
      "epoch": 22.889290012033694,
      "grad_norm": Infinity,
      "learning_rate": 0.00017623533159664856,
      "loss": 11.2113,
      "step": 190210
    },
    {
      "epoch": 22.89049338146811,
      "grad_norm": Infinity,
      "learning_rate": 0.00017623286956623612,
      "loss": 11.2825,
      "step": 190220
    },
    {
      "epoch": 22.891696750902526,
      "grad_norm": Infinity,
      "learning_rate": 0.00017623040742549594,
      "loss": 11.1869,
      "step": 190230
    },
    {
      "epoch": 22.892900120336943,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762279451744315,
      "loss": 11.0981,
      "step": 190240
    },
    {
      "epoch": 22.89410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017622548281304642,
      "loss": 11.1534,
      "step": 190250
    },
    {
      "epoch": 22.895306859205775,
      "grad_norm": Infinity,
      "learning_rate": 0.00017622302034134423,
      "loss": 11.2106,
      "step": 190260
    },
    {
      "epoch": 22.896510228640192,
      "grad_norm": Infinity,
      "learning_rate": 0.00017622055775932854,
      "loss": 11.115,
      "step": 190270
    },
    {
      "epoch": 22.89771359807461,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762180950670029,
      "loss": 11.071,
      "step": 190280
    },
    {
      "epoch": 22.898916967509024,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762156322643708,
      "loss": 11.1501,
      "step": 190290
    },
    {
      "epoch": 22.90012033694344,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762131693514359,
      "loss": 11.2295,
      "step": 190300
    },
    {
      "epoch": 22.90132370637786,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762107063282017,
      "loss": 11.108,
      "step": 190310
    },
    {
      "epoch": 22.902527075812273,
      "grad_norm": Infinity,
      "learning_rate": 0.00017620824319467183,
      "loss": 11.2575,
      "step": 190320
    },
    {
      "epoch": 22.90373044524669,
      "grad_norm": Infinity,
      "learning_rate": 0.00017620577995084978,
      "loss": 11.1869,
      "step": 190330
    },
    {
      "epoch": 22.904933814681108,
      "grad_norm": Infinity,
      "learning_rate": 0.0001762033165967392,
      "loss": 11.1661,
      "step": 190340
    },
    {
      "epoch": 22.906137184115522,
      "grad_norm": Infinity,
      "learning_rate": 0.00017620085313234355,
      "loss": 11.1495,
      "step": 190350
    },
    {
      "epoch": 22.90734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.00017619838955766648,
      "loss": 11.1451,
      "step": 190360
    },
    {
      "epoch": 22.908543922984357,
      "grad_norm": Infinity,
      "learning_rate": 0.00017619592587271154,
      "loss": 11.1432,
      "step": 190370
    },
    {
      "epoch": 22.90974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00017619346207748226,
      "loss": 11.0582,
      "step": 190380
    },
    {
      "epoch": 22.91095066185319,
      "grad_norm": Infinity,
      "learning_rate": 0.00017619099817198225,
      "loss": 11.1593,
      "step": 190390
    },
    {
      "epoch": 22.912154031287606,
      "grad_norm": Infinity,
      "learning_rate": 0.00017618853415621503,
      "loss": 11.1102,
      "step": 190400
    },
    {
      "epoch": 22.91335740072202,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761860700301842,
      "loss": 11.0924,
      "step": 190410
    },
    {
      "epoch": 22.914560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00017618360579389332,
      "loss": 10.9973,
      "step": 190420
    },
    {
      "epoch": 22.915764139590856,
      "grad_norm": Infinity,
      "learning_rate": 0.00017618114144734594,
      "loss": 11.1166,
      "step": 190430
    },
    {
      "epoch": 22.91696750902527,
      "grad_norm": Infinity,
      "learning_rate": 0.00017617867699054566,
      "loss": 11.1088,
      "step": 190440
    },
    {
      "epoch": 22.918170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00017617621242349603,
      "loss": 11.1023,
      "step": 190450
    },
    {
      "epoch": 22.919374247894105,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761737477462006,
      "loss": 11.0637,
      "step": 190460
    },
    {
      "epoch": 22.92057761732852,
      "grad_norm": Infinity,
      "learning_rate": 0.00017617128295866293,
      "loss": 11.2578,
      "step": 190470
    },
    {
      "epoch": 22.921780986762936,
      "grad_norm": Infinity,
      "learning_rate": 0.00017616881806088665,
      "loss": 11.0875,
      "step": 190480
    },
    {
      "epoch": 22.922984356197354,
      "grad_norm": Infinity,
      "learning_rate": 0.00017616635305287527,
      "loss": 11.1683,
      "step": 190490
    },
    {
      "epoch": 22.924187725631768,
      "grad_norm": Infinity,
      "learning_rate": 0.00017616388793463238,
      "loss": 11.1869,
      "step": 190500
    },
    {
      "epoch": 22.925391095066185,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761614227061615,
      "loss": 11.1292,
      "step": 190510
    },
    {
      "epoch": 22.926594464500603,
      "grad_norm": Infinity,
      "learning_rate": 0.00017615895736746627,
      "loss": 11.251,
      "step": 190520
    },
    {
      "epoch": 22.927797833935017,
      "grad_norm": Infinity,
      "learning_rate": 0.00017615649191855023,
      "loss": 11.1146,
      "step": 190530
    },
    {
      "epoch": 22.929001203369435,
      "grad_norm": Infinity,
      "learning_rate": 0.00017615402635941694,
      "loss": 11.1738,
      "step": 190540
    },
    {
      "epoch": 22.930204572803852,
      "grad_norm": Infinity,
      "learning_rate": 0.00017615156069006996,
      "loss": 11.148,
      "step": 190550
    },
    {
      "epoch": 22.931407942238266,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761490949105129,
      "loss": 11.1293,
      "step": 190560
    },
    {
      "epoch": 22.932611311672684,
      "grad_norm": Infinity,
      "learning_rate": 0.00017614662902074927,
      "loss": 11.2317,
      "step": 190570
    },
    {
      "epoch": 22.9338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00017614416302078268,
      "loss": 11.2083,
      "step": 190580
    },
    {
      "epoch": 22.935018050541515,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761416969106167,
      "loss": 11.1864,
      "step": 190590
    },
    {
      "epoch": 22.936221419975933,
      "grad_norm": Infinity,
      "learning_rate": 0.00017613923069025484,
      "loss": 11.0079,
      "step": 190600
    },
    {
      "epoch": 22.93742478941035,
      "grad_norm": Infinity,
      "learning_rate": 0.00017613676435970075,
      "loss": 11.093,
      "step": 190610
    },
    {
      "epoch": 22.938628158844764,
      "grad_norm": Infinity,
      "learning_rate": 0.00017613429791895794,
      "loss": 11.0738,
      "step": 190620
    },
    {
      "epoch": 22.939831528279182,
      "grad_norm": Infinity,
      "learning_rate": 0.00017613183136803003,
      "loss": 11.0428,
      "step": 190630
    },
    {
      "epoch": 22.9410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017612936470692055,
      "loss": 11.0331,
      "step": 190640
    },
    {
      "epoch": 22.942238267148014,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761268979356331,
      "loss": 11.0988,
      "step": 190650
    },
    {
      "epoch": 22.94344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761244310541712,
      "loss": 11.1847,
      "step": 190660
    },
    {
      "epoch": 22.94464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761219640625385,
      "loss": 11.1531,
      "step": 190670
    },
    {
      "epoch": 22.945848375451263,
      "grad_norm": Infinity,
      "learning_rate": 0.00017611949696073848,
      "loss": 11.1233,
      "step": 190680
    },
    {
      "epoch": 22.94705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017611702974877478,
      "loss": 11.2216,
      "step": 190690
    },
    {
      "epoch": 22.948255114320098,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761145624266509,
      "loss": 11.1471,
      "step": 190700
    },
    {
      "epoch": 22.949458483754512,
      "grad_norm": Infinity,
      "learning_rate": 0.00017611209499437055,
      "loss": 11.2306,
      "step": 190710
    },
    {
      "epoch": 22.95066185318893,
      "grad_norm": Infinity,
      "learning_rate": 0.00017610962745193715,
      "loss": 11.1139,
      "step": 190720
    },
    {
      "epoch": 22.951865222623347,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761071597993543,
      "loss": 11.1324,
      "step": 190730
    },
    {
      "epoch": 22.95306859205776,
      "grad_norm": Infinity,
      "learning_rate": 0.00017610469203662566,
      "loss": 11.1031,
      "step": 190740
    },
    {
      "epoch": 22.95427196149218,
      "grad_norm": Infinity,
      "learning_rate": 0.0001761022241637547,
      "loss": 11.1528,
      "step": 190750
    },
    {
      "epoch": 22.955475330926596,
      "grad_norm": Infinity,
      "learning_rate": 0.00017609975618074506,
      "loss": 11.199,
      "step": 190760
    },
    {
      "epoch": 22.95667870036101,
      "grad_norm": Infinity,
      "learning_rate": 0.00017609728808760028,
      "loss": 11.1168,
      "step": 190770
    },
    {
      "epoch": 22.957882069795428,
      "grad_norm": Infinity,
      "learning_rate": 0.00017609481988432388,
      "loss": 11.1278,
      "step": 190780
    },
    {
      "epoch": 22.959085439229845,
      "grad_norm": Infinity,
      "learning_rate": 0.00017609235157091955,
      "loss": 11.1338,
      "step": 190790
    },
    {
      "epoch": 22.96028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.00017608988314739076,
      "loss": 11.1349,
      "step": 190800
    },
    {
      "epoch": 22.961492178098677,
      "grad_norm": Infinity,
      "learning_rate": 0.00017608741461374116,
      "loss": 11.1423,
      "step": 190810
    },
    {
      "epoch": 22.96269554753309,
      "grad_norm": Infinity,
      "learning_rate": 0.00017608494596997427,
      "loss": 11.2185,
      "step": 190820
    },
    {
      "epoch": 22.96389891696751,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760824772160937,
      "loss": 11.1235,
      "step": 190830
    },
    {
      "epoch": 22.965102286401926,
      "grad_norm": Infinity,
      "learning_rate": 0.00017608000835210298,
      "loss": 11.1686,
      "step": 190840
    },
    {
      "epoch": 22.96630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00017607753937800568,
      "loss": 11.0293,
      "step": 190850
    },
    {
      "epoch": 22.967509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.00017607507029380543,
      "loss": 11.149,
      "step": 190860
    },
    {
      "epoch": 22.968712394705175,
      "grad_norm": Infinity,
      "learning_rate": 0.00017607260109950577,
      "loss": 11.1118,
      "step": 190870
    },
    {
      "epoch": 22.96991576413959,
      "grad_norm": Infinity,
      "learning_rate": 0.00017607013179511027,
      "loss": 11.0345,
      "step": 190880
    },
    {
      "epoch": 22.971119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760676623806225,
      "loss": 11.0491,
      "step": 190890
    },
    {
      "epoch": 22.972322503008424,
      "grad_norm": Infinity,
      "learning_rate": 0.00017606519285604606,
      "loss": 11.1346,
      "step": 190900
    },
    {
      "epoch": 22.97352587244284,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760627232213845,
      "loss": 11.2246,
      "step": 190910
    },
    {
      "epoch": 22.974729241877256,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760602534766414,
      "loss": 11.2425,
      "step": 190920
    },
    {
      "epoch": 22.975932611311674,
      "grad_norm": Infinity,
      "learning_rate": 0.00017605778362182032,
      "loss": 11.1684,
      "step": 190930
    },
    {
      "epoch": 22.977135980746088,
      "grad_norm": Infinity,
      "learning_rate": 0.00017605531365692486,
      "loss": 11.2066,
      "step": 190940
    },
    {
      "epoch": 22.978339350180505,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760528435819586,
      "loss": 11.1009,
      "step": 190950
    },
    {
      "epoch": 22.979542719614923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017605037339692508,
      "loss": 11.1612,
      "step": 190960
    },
    {
      "epoch": 22.980746089049337,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760479031018279,
      "loss": 11.1445,
      "step": 190970
    },
    {
      "epoch": 22.981949458483754,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760454326966706,
      "loss": 10.9852,
      "step": 190980
    },
    {
      "epoch": 22.983152827918172,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760429621814568,
      "loss": 11.0465,
      "step": 190990
    },
    {
      "epoch": 22.984356197352586,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760404915561901,
      "loss": 11.1964,
      "step": 191000
    },
    {
      "epoch": 22.985559566787003,
      "grad_norm": Infinity,
      "learning_rate": 0.00017603802082087397,
      "loss": 11.0606,
      "step": 191010
    },
    {
      "epoch": 22.98676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760355499755121,
      "loss": 11.2024,
      "step": 191020
    },
    {
      "epoch": 22.987966305655835,
      "grad_norm": Infinity,
      "learning_rate": 0.000176033079020108,
      "loss": 11.093,
      "step": 191030
    },
    {
      "epoch": 22.989169675090253,
      "grad_norm": Infinity,
      "learning_rate": 0.00017603060795466527,
      "loss": 11.0497,
      "step": 191040
    },
    {
      "epoch": 22.99037304452467,
      "grad_norm": Infinity,
      "learning_rate": 0.00017602813677918745,
      "loss": 11.118,
      "step": 191050
    },
    {
      "epoch": 22.991576413959084,
      "grad_norm": Infinity,
      "learning_rate": 0.00017602566549367819,
      "loss": 11.156,
      "step": 191060
    },
    {
      "epoch": 22.9927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.000176023194098141,
      "loss": 11.2422,
      "step": 191070
    },
    {
      "epoch": 22.99398315282792,
      "grad_norm": Infinity,
      "learning_rate": 0.00017602072259257945,
      "loss": 11.0094,
      "step": 191080
    },
    {
      "epoch": 22.995186522262333,
      "grad_norm": Infinity,
      "learning_rate": 0.00017601825097699717,
      "loss": 11.1688,
      "step": 191090
    },
    {
      "epoch": 22.99638989169675,
      "grad_norm": Infinity,
      "learning_rate": 0.00017601577925139772,
      "loss": 10.9977,
      "step": 191100
    },
    {
      "epoch": 22.99759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00017601330741578468,
      "loss": 11.1709,
      "step": 191110
    },
    {
      "epoch": 22.998796630565582,
      "grad_norm": Infinity,
      "learning_rate": 0.0001760108354701616,
      "loss": 11.1603,
      "step": 191120
    },
    {
      "epoch": 23.0,
      "grad_norm": Infinity,
      "learning_rate": 0.00017600836341453207,
      "loss": 11.1038,
      "step": 191130
    },
    {
      "epoch": 23.0,
      "eval_loss": 11.139604568481445,
      "eval_runtime": 118.3041,
      "eval_samples_per_second": 62.441,
      "eval_steps_per_second": 7.81,
      "step": 191130
    },
    {
      "epoch": 23.001203369434418,
      "grad_norm": Infinity,
      "learning_rate": 0.00017600589124889968,
      "loss": 11.1964,
      "step": 191140
    },
    {
      "epoch": 23.00240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.000176003418973268,
      "loss": 11.0488,
      "step": 191150
    },
    {
      "epoch": 23.00361010830325,
      "grad_norm": Infinity,
      "learning_rate": 0.00017600094658764062,
      "loss": 11.2841,
      "step": 191160
    },
    {
      "epoch": 23.004813477737667,
      "grad_norm": Infinity,
      "learning_rate": 0.00017599847409202108,
      "loss": 11.2014,
      "step": 191170
    },
    {
      "epoch": 23.00601684717208,
      "grad_norm": Infinity,
      "learning_rate": 0.000175996001486413,
      "loss": 11.2959,
      "step": 191180
    },
    {
      "epoch": 23.0072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759935287708199,
      "loss": 11.2211,
      "step": 191190
    },
    {
      "epoch": 23.008423586040916,
      "grad_norm": Infinity,
      "learning_rate": 0.00017599105594524547,
      "loss": 11.22,
      "step": 191200
    },
    {
      "epoch": 23.00962695547533,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759885830096932,
      "loss": 11.158,
      "step": 191210
    },
    {
      "epoch": 23.010830324909747,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759861099641667,
      "loss": 11.1659,
      "step": 191220
    },
    {
      "epoch": 23.012033694344165,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759836368086695,
      "loss": 11.175,
      "step": 191230
    },
    {
      "epoch": 23.01323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00017598116354320523,
      "loss": 11.1375,
      "step": 191240
    },
    {
      "epoch": 23.014440433212997,
      "grad_norm": Infinity,
      "learning_rate": 0.00017597869016777748,
      "loss": 11.1379,
      "step": 191250
    },
    {
      "epoch": 23.015643802647414,
      "grad_norm": Infinity,
      "learning_rate": 0.00017597621668238977,
      "loss": 11.0362,
      "step": 191260
    },
    {
      "epoch": 23.016847172081828,
      "grad_norm": Infinity,
      "learning_rate": 0.00017597374308704576,
      "loss": 11.1441,
      "step": 191270
    },
    {
      "epoch": 23.018050541516246,
      "grad_norm": Infinity,
      "learning_rate": 0.00017597126938174898,
      "loss": 11.2475,
      "step": 191280
    },
    {
      "epoch": 23.019253910950663,
      "grad_norm": Infinity,
      "learning_rate": 0.00017596879556650298,
      "loss": 11.225,
      "step": 191290
    },
    {
      "epoch": 23.020457280385077,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759663216413114,
      "loss": 11.2133,
      "step": 191300
    },
    {
      "epoch": 23.021660649819495,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759638476061778,
      "loss": 11.1681,
      "step": 191310
    },
    {
      "epoch": 23.022864019253912,
      "grad_norm": Infinity,
      "learning_rate": 0.00017596137346110574,
      "loss": 11.0869,
      "step": 191320
    },
    {
      "epoch": 23.024067388688326,
      "grad_norm": Infinity,
      "learning_rate": 0.00017595889920609886,
      "loss": 11.1001,
      "step": 191330
    },
    {
      "epoch": 23.025270758122744,
      "grad_norm": Infinity,
      "learning_rate": 0.00017595642484116066,
      "loss": 11.107,
      "step": 191340
    },
    {
      "epoch": 23.02647412755716,
      "grad_norm": Infinity,
      "learning_rate": 0.00017595395036629476,
      "loss": 11.2042,
      "step": 191350
    },
    {
      "epoch": 23.027677496991576,
      "grad_norm": Infinity,
      "learning_rate": 0.00017595147578150475,
      "loss": 11.0908,
      "step": 191360
    },
    {
      "epoch": 23.028880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759490010867942,
      "loss": 11.1936,
      "step": 191370
    },
    {
      "epoch": 23.03008423586041,
      "grad_norm": Infinity,
      "learning_rate": 0.00017594652628216673,
      "loss": 11.1785,
      "step": 191380
    },
    {
      "epoch": 23.031287605294825,
      "grad_norm": Infinity,
      "learning_rate": 0.00017594405136762582,
      "loss": 11.1884,
      "step": 191390
    },
    {
      "epoch": 23.032490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00017594157634317516,
      "loss": 11.0629,
      "step": 191400
    },
    {
      "epoch": 23.03369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759391012088183,
      "loss": 11.0853,
      "step": 191410
    },
    {
      "epoch": 23.034897713598074,
      "grad_norm": Infinity,
      "learning_rate": 0.00017593662596455877,
      "loss": 11.2413,
      "step": 191420
    },
    {
      "epoch": 23.03610108303249,
      "grad_norm": Infinity,
      "learning_rate": 0.00017593415061040025,
      "loss": 11.2708,
      "step": 191430
    },
    {
      "epoch": 23.03730445246691,
      "grad_norm": Infinity,
      "learning_rate": 0.00017593167514634624,
      "loss": 10.9999,
      "step": 191440
    },
    {
      "epoch": 23.038507821901323,
      "grad_norm": Infinity,
      "learning_rate": 0.00017592919957240032,
      "loss": 11.0191,
      "step": 191450
    },
    {
      "epoch": 23.03971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017592672388856614,
      "loss": 11.1132,
      "step": 191460
    },
    {
      "epoch": 23.040914560770158,
      "grad_norm": Infinity,
      "learning_rate": 0.00017592424809484723,
      "loss": 11.1149,
      "step": 191470
    },
    {
      "epoch": 23.042117930204572,
      "grad_norm": Infinity,
      "learning_rate": 0.00017592177219124717,
      "loss": 11.11,
      "step": 191480
    },
    {
      "epoch": 23.04332129963899,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759192961777696,
      "loss": 11.1361,
      "step": 191490
    },
    {
      "epoch": 23.044524669073404,
      "grad_norm": Infinity,
      "learning_rate": 0.00017591682005441804,
      "loss": 11.1655,
      "step": 191500
    },
    {
      "epoch": 23.04572803850782,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759143438211961,
      "loss": 11.147,
      "step": 191510
    },
    {
      "epoch": 23.04693140794224,
      "grad_norm": Infinity,
      "learning_rate": 0.00017591186747810736,
      "loss": 11.1585,
      "step": 191520
    },
    {
      "epoch": 23.048134777376653,
      "grad_norm": Infinity,
      "learning_rate": 0.00017590939102515542,
      "loss": 11.0279,
      "step": 191530
    },
    {
      "epoch": 23.04933814681107,
      "grad_norm": Infinity,
      "learning_rate": 0.00017590691446234383,
      "loss": 11.0423,
      "step": 191540
    },
    {
      "epoch": 23.050541516245488,
      "grad_norm": Infinity,
      "learning_rate": 0.0001759044377896762,
      "loss": 11.1259,
      "step": 191550
    },
    {
      "epoch": 23.051744885679902,
      "grad_norm": Infinity,
      "learning_rate": 0.00017590196100715613,
      "loss": 11.2332,
      "step": 191560
    },
    {
      "epoch": 23.05294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00017589948411478714,
      "loss": 11.0941,
      "step": 191570
    },
    {
      "epoch": 23.054151624548737,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758970071125729,
      "loss": 11.1196,
      "step": 191580
    },
    {
      "epoch": 23.05535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00017589453000051692,
      "loss": 11.1721,
      "step": 191590
    },
    {
      "epoch": 23.05655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00017589205277862284,
      "loss": 11.0877,
      "step": 191600
    },
    {
      "epoch": 23.057761732851986,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758895754468942,
      "loss": 11.1361,
      "step": 191610
    },
    {
      "epoch": 23.0589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.00017588709800533465,
      "loss": 11.1026,
      "step": 191620
    },
    {
      "epoch": 23.060168471720818,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758846204539477,
      "loss": 11.1873,
      "step": 191630
    },
    {
      "epoch": 23.061371841155236,
      "grad_norm": Infinity,
      "learning_rate": 0.00017588214279273698,
      "loss": 10.9926,
      "step": 191640
    },
    {
      "epoch": 23.06257521058965,
      "grad_norm": Infinity,
      "learning_rate": 0.00017587966502170604,
      "loss": 11.1515,
      "step": 191650
    },
    {
      "epoch": 23.063778580024067,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758771871408585,
      "loss": 11.1253,
      "step": 191660
    },
    {
      "epoch": 23.064981949458485,
      "grad_norm": Infinity,
      "learning_rate": 0.00017587470915019795,
      "loss": 11.2217,
      "step": 191670
    },
    {
      "epoch": 23.0661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00017587223104972797,
      "loss": 11.0556,
      "step": 191680
    },
    {
      "epoch": 23.067388688327316,
      "grad_norm": Infinity,
      "learning_rate": 0.00017586975283945212,
      "loss": 11.1713,
      "step": 191690
    },
    {
      "epoch": 23.068592057761734,
      "grad_norm": Infinity,
      "learning_rate": 0.00017586727451937405,
      "loss": 11.1873,
      "step": 191700
    },
    {
      "epoch": 23.069795427196148,
      "grad_norm": Infinity,
      "learning_rate": 0.00017586479608949724,
      "loss": 11.2209,
      "step": 191710
    },
    {
      "epoch": 23.070998796630565,
      "grad_norm": Infinity,
      "learning_rate": 0.00017586231754982537,
      "loss": 11.0533,
      "step": 191720
    },
    {
      "epoch": 23.072202166064983,
      "grad_norm": Infinity,
      "learning_rate": 0.000175859838900362,
      "loss": 11.0896,
      "step": 191730
    },
    {
      "epoch": 23.073405535499397,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758573601411107,
      "loss": 11.1186,
      "step": 191740
    },
    {
      "epoch": 23.074608904933815,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758548812720751,
      "loss": 11.051,
      "step": 191750
    },
    {
      "epoch": 23.075812274368232,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758524022932587,
      "loss": 10.9249,
      "step": 191760
    },
    {
      "epoch": 23.077015643802646,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758499232046652,
      "loss": 11.1358,
      "step": 191770
    },
    {
      "epoch": 23.078219013237064,
      "grad_norm": Infinity,
      "learning_rate": 0.00017584744400629814,
      "loss": 11.2267,
      "step": 191780
    },
    {
      "epoch": 23.07942238267148,
      "grad_norm": Infinity,
      "learning_rate": 0.00017584496469816107,
      "loss": 11.1696,
      "step": 191790
    },
    {
      "epoch": 23.080625752105895,
      "grad_norm": Infinity,
      "learning_rate": 0.00017584248528025763,
      "loss": 11.1102,
      "step": 191800
    },
    {
      "epoch": 23.081829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.00017584000575259138,
      "loss": 11.1422,
      "step": 191810
    },
    {
      "epoch": 23.08303249097473,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758375261151659,
      "loss": 11.1524,
      "step": 191820
    },
    {
      "epoch": 23.084235860409144,
      "grad_norm": Infinity,
      "learning_rate": 0.00017583504636798484,
      "loss": 11.0193,
      "step": 191830
    },
    {
      "epoch": 23.085439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758325665110517,
      "loss": 11.0701,
      "step": 191840
    },
    {
      "epoch": 23.08664259927798,
      "grad_norm": Infinity,
      "learning_rate": 0.00017583008654437014,
      "loss": 11.1242,
      "step": 191850
    },
    {
      "epoch": 23.087845968712394,
      "grad_norm": Infinity,
      "learning_rate": 0.00017582760646794372,
      "loss": 11.1676,
      "step": 191860
    },
    {
      "epoch": 23.08904933814681,
      "grad_norm": Infinity,
      "learning_rate": 0.000175825126281776,
      "loss": 11.0349,
      "step": 191870
    },
    {
      "epoch": 23.09025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.00017582264598587066,
      "loss": 11.1606,
      "step": 191880
    },
    {
      "epoch": 23.091456077015643,
      "grad_norm": Infinity,
      "learning_rate": 0.00017582016558023117,
      "loss": 11.1211,
      "step": 191890
    },
    {
      "epoch": 23.09265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00017581768506486122,
      "loss": 11.0956,
      "step": 191900
    },
    {
      "epoch": 23.093862815884478,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758152044397644,
      "loss": 11.1547,
      "step": 191910
    },
    {
      "epoch": 23.095066185318892,
      "grad_norm": Infinity,
      "learning_rate": 0.0001758127237049442,
      "loss": 11.2541,
      "step": 191920
    },
    {
      "epoch": 23.09626955475331,
      "grad_norm": Infinity,
      "learning_rate": 0.00017581024286040427,
      "loss": 11.1044,
      "step": 191930
    },
    {
      "epoch": 23.097472924187727,
      "grad_norm": Infinity,
      "learning_rate": 0.00017580776190614822,
      "loss": 11.1746,
      "step": 191940
    },
    {
      "epoch": 23.09867629362214,
      "grad_norm": Infinity,
      "learning_rate": 0.00017580528084217963,
      "loss": 11.1385,
      "step": 191950
    },
    {
      "epoch": 23.09987966305656,
      "grad_norm": Infinity,
      "learning_rate": 0.00017580279966850207,
      "loss": 11.1221,
      "step": 191960
    },
    {
      "epoch": 23.101083032490976,
      "grad_norm": Infinity,
      "learning_rate": 0.00017580031838511912,
      "loss": 11.17,
      "step": 191970
    },
    {
      "epoch": 23.10228640192539,
      "grad_norm": Infinity,
      "learning_rate": 0.00017579783699203443,
      "loss": 11.1612,
      "step": 191980
    },
    {
      "epoch": 23.103489771359808,
      "grad_norm": Infinity,
      "learning_rate": 0.00017579535548925155,
      "loss": 11.0318,
      "step": 191990
    },
    {
      "epoch": 23.104693140794225,
      "grad_norm": Infinity,
      "learning_rate": 0.00017579287387677408,
      "loss": 11.1372,
      "step": 192000
    },
    {
      "epoch": 23.10589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757903921546056,
      "loss": 11.177,
      "step": 192010
    },
    {
      "epoch": 23.107099879663057,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757879103227497,
      "loss": 11.1301,
      "step": 192020
    },
    {
      "epoch": 23.108303249097474,
      "grad_norm": Infinity,
      "learning_rate": 0.00017578542838121,
      "loss": 11.2456,
      "step": 192030
    },
    {
      "epoch": 23.10950661853189,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757829463299901,
      "loss": 11.0623,
      "step": 192040
    },
    {
      "epoch": 23.110709987966306,
      "grad_norm": Infinity,
      "learning_rate": 0.00017578046416909352,
      "loss": 10.9927,
      "step": 192050
    },
    {
      "epoch": 23.111913357400724,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757779818985239,
      "loss": 11.0926,
      "step": 192060
    },
    {
      "epoch": 23.113116726835138,
      "grad_norm": Infinity,
      "learning_rate": 0.00017577549951828484,
      "loss": 11.2262,
      "step": 192070
    },
    {
      "epoch": 23.114320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.00017577301702837995,
      "loss": 11.2446,
      "step": 192080
    },
    {
      "epoch": 23.115523465703973,
      "grad_norm": Infinity,
      "learning_rate": 0.00017577053442881276,
      "loss": 11.0588,
      "step": 192090
    },
    {
      "epoch": 23.116726835138387,
      "grad_norm": Infinity,
      "learning_rate": 0.00017576805171958695,
      "loss": 11.1521,
      "step": 192100
    },
    {
      "epoch": 23.117930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.000175765568900706,
      "loss": 11.2255,
      "step": 192110
    },
    {
      "epoch": 23.119133574007222,
      "grad_norm": Infinity,
      "learning_rate": 0.00017576308597217358,
      "loss": 11.0934,
      "step": 192120
    },
    {
      "epoch": 23.120336943441636,
      "grad_norm": Infinity,
      "learning_rate": 0.00017576060293399333,
      "loss": 11.2551,
      "step": 192130
    },
    {
      "epoch": 23.121540312876053,
      "grad_norm": Infinity,
      "learning_rate": 0.00017575811978616873,
      "loss": 11.1774,
      "step": 192140
    },
    {
      "epoch": 23.12274368231047,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757556365287034,
      "loss": 11.1375,
      "step": 192150
    },
    {
      "epoch": 23.123947051744885,
      "grad_norm": Infinity,
      "learning_rate": 0.000175753153161601,
      "loss": 11.2116,
      "step": 192160
    },
    {
      "epoch": 23.125150421179303,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757506696848651,
      "loss": 11.1012,
      "step": 192170
    },
    {
      "epoch": 23.126353790613717,
      "grad_norm": Infinity,
      "learning_rate": 0.00017574818609849925,
      "loss": 11.2283,
      "step": 192180
    },
    {
      "epoch": 23.127557160048134,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757457024025071,
      "loss": 11.2024,
      "step": 192190
    },
    {
      "epoch": 23.128760529482552,
      "grad_norm": Infinity,
      "learning_rate": 0.00017574321859689218,
      "loss": 11.0866,
      "step": 192200
    },
    {
      "epoch": 23.129963898916966,
      "grad_norm": Infinity,
      "learning_rate": 0.00017574073468165814,
      "loss": 11.2123,
      "step": 192210
    },
    {
      "epoch": 23.131167268351383,
      "grad_norm": Infinity,
      "learning_rate": 0.00017573825065680856,
      "loss": 11.1511,
      "step": 192220
    },
    {
      "epoch": 23.1323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.00017573576652234703,
      "loss": 11.1686,
      "step": 192230
    },
    {
      "epoch": 23.133574007220215,
      "grad_norm": Infinity,
      "learning_rate": 0.00017573328227827713,
      "loss": 11.2081,
      "step": 192240
    },
    {
      "epoch": 23.134777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757307979246025,
      "loss": 11.1236,
      "step": 192250
    },
    {
      "epoch": 23.13598074608905,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757283134613267,
      "loss": 11.1247,
      "step": 192260
    },
    {
      "epoch": 23.137184115523464,
      "grad_norm": Infinity,
      "learning_rate": 0.00017572582888845332,
      "loss": 11.2671,
      "step": 192270
    },
    {
      "epoch": 23.13838748495788,
      "grad_norm": Infinity,
      "learning_rate": 0.000175723344205986,
      "loss": 11.1052,
      "step": 192280
    },
    {
      "epoch": 23.1395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757208594139283,
      "loss": 11.0528,
      "step": 192290
    },
    {
      "epoch": 23.140794223826713,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757183745122838,
      "loss": 11.2499,
      "step": 192300
    },
    {
      "epoch": 23.14199759326113,
      "grad_norm": Infinity,
      "learning_rate": 0.00017571588950105613,
      "loss": 11.242,
      "step": 192310
    },
    {
      "epoch": 23.14320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.00017571340438024885,
      "loss": 11.0757,
      "step": 192320
    },
    {
      "epoch": 23.144404332129962,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757109191498656,
      "loss": 11.1414,
      "step": 192330
    },
    {
      "epoch": 23.14560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00017570843380990998,
      "loss": 11.1974,
      "step": 192340
    },
    {
      "epoch": 23.146811070998798,
      "grad_norm": Infinity,
      "learning_rate": 0.00017570594836038551,
      "loss": 11.2422,
      "step": 192350
    },
    {
      "epoch": 23.14801444043321,
      "grad_norm": Infinity,
      "learning_rate": 0.0001757034628012959,
      "loss": 11.1928,
      "step": 192360
    },
    {
      "epoch": 23.14921780986763,
      "grad_norm": Infinity,
      "learning_rate": 0.00017570097713264464,
      "loss": 11.1876,
      "step": 192370
    },
    {
      "epoch": 23.150421179302047,
      "grad_norm": Infinity,
      "learning_rate": 0.00017569849135443542,
      "loss": 11.1964,
      "step": 192380
    },
    {
      "epoch": 23.15162454873646,
      "grad_norm": Infinity,
      "learning_rate": 0.00017569600546667177,
      "loss": 11.0677,
      "step": 192390
    },
    {
      "epoch": 23.15282791817088,
      "grad_norm": Infinity,
      "learning_rate": 0.00017569351946935733,
      "loss": 11.1434,
      "step": 192400
    },
    {
      "epoch": 23.154031287605296,
      "grad_norm": Infinity,
      "learning_rate": 0.00017569103336249564,
      "loss": 11.1399,
      "step": 192410
    },
    {
      "epoch": 23.15523465703971,
      "grad_norm": Infinity,
      "learning_rate": 0.00017568854714609036,
      "loss": 11.0888,
      "step": 192420
    },
    {
      "epoch": 23.156438026474127,
      "grad_norm": Infinity,
      "learning_rate": 0.00017568606082014507,
      "loss": 11.2442,
      "step": 192430
    },
    {
      "epoch": 23.157641395908545,
      "grad_norm": Infinity,
      "learning_rate": 0.00017568357438466336,
      "loss": 11.1043,
      "step": 192440
    },
    {
      "epoch": 23.15884476534296,
      "grad_norm": Infinity,
      "learning_rate": 0.00017568108783964882,
      "loss": 11.1085,
      "step": 192450
    },
    {
      "epoch": 23.160048134777377,
      "grad_norm": Infinity,
      "learning_rate": 0.00017567860118510508,
      "loss": 11.1456,
      "step": 192460
    },
    {
      "epoch": 23.161251504211794,
      "grad_norm": Infinity,
      "learning_rate": 0.0001756761144210357,
      "loss": 11.1575,
      "step": 192470
    },
    {
      "epoch": 23.162454873646208,
      "grad_norm": Infinity,
      "learning_rate": 0.00017567362754744434,
      "loss": 11.2116,
      "step": 192480
    },
    {
      "epoch": 23.163658243080626,
      "grad_norm": Infinity,
      "learning_rate": 0.00017567114056433452,
      "loss": 11.1255,
      "step": 192490
    },
    {
      "epoch": 23.164861612515043,
      "grad_norm": Infinity,
      "learning_rate": 0.00017566865347170988,
      "loss": 11.0535,
      "step": 192500
    },
    {
      "epoch": 23.166064981949457,
      "grad_norm": Infinity,
      "learning_rate": 0.00017566616626957403,
      "loss": 11.178,
      "step": 192510
    },
    {
      "epoch": 23.167268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.00017566367895793054,
      "loss": 11.0545,
      "step": 192520
    },
    {
      "epoch": 23.168471720818292,
      "grad_norm": Infinity,
      "learning_rate": 0.00017566119153678304,
      "loss": 11.163,
      "step": 192530
    },
    {
      "epoch": 23.169675090252706,
      "grad_norm": Infinity,
      "learning_rate": 0.0001756587040061351,
      "loss": 11.058,
      "step": 192540
    },
    {
      "epoch": 23.170878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00017565621636599036,
      "loss": 11.0659,
      "step": 192550
    },
    {
      "epoch": 23.17208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.0001756537286163524,
      "loss": 11.1096,
      "step": 192560
    },
    {
      "epoch": 23.173285198555956,
      "grad_norm": Infinity,
      "learning_rate": 0.00017565124075722476,
      "loss": 11.1551,
      "step": 192570
    },
    {
      "epoch": 23.174488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.00017564875278861116,
      "loss": 11.205,
      "step": 192580
    },
    {
      "epoch": 23.17569193742479,
      "grad_norm": Infinity,
      "learning_rate": 0.0001756462647105151,
      "loss": 11.3652,
      "step": 192590
    },
    {
      "epoch": 23.176895306859205,
      "grad_norm": Infinity,
      "learning_rate": 0.00017564377652294023,
      "loss": 11.2155,
      "step": 192600
    },
    {
      "epoch": 23.178098676293622,
      "grad_norm": Infinity,
      "learning_rate": 0.00017564128822589013,
      "loss": 11.1632,
      "step": 192610
    },
    {
      "epoch": 23.17930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.00017563879981936844,
      "loss": 11.1482,
      "step": 192620
    },
    {
      "epoch": 23.180505415162454,
      "grad_norm": Infinity,
      "learning_rate": 0.0001756363113033787,
      "loss": 11.1527,
      "step": 192630
    },
    {
      "epoch": 23.18170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00017563382267792455,
      "loss": 10.9933,
      "step": 192640
    },
    {
      "epoch": 23.18291215403129,
      "grad_norm": Infinity,
      "learning_rate": 0.00017563133394300962,
      "loss": 11.1944,
      "step": 192650
    },
    {
      "epoch": 23.184115523465703,
      "grad_norm": Infinity,
      "learning_rate": 0.0001756288450986374,
      "loss": 11.1868,
      "step": 192660
    },
    {
      "epoch": 23.18531889290012,
      "grad_norm": Infinity,
      "learning_rate": 0.00017562635614481166,
      "loss": 11.1149,
      "step": 192670
    },
    {
      "epoch": 23.186522262334538,
      "grad_norm": Infinity,
      "learning_rate": 0.00017562386708153585,
      "loss": 11.227,
      "step": 192680
    },
    {
      "epoch": 23.187725631768952,
      "grad_norm": Infinity,
      "learning_rate": 0.00017562137790881366,
      "loss": 11.2388,
      "step": 192690
    },
    {
      "epoch": 23.18892900120337,
      "grad_norm": Infinity,
      "learning_rate": 0.00017561888862664866,
      "loss": 11.1108,
      "step": 192700
    },
    {
      "epoch": 23.190132370637787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017561639923504446,
      "loss": 11.1981,
      "step": 192710
    },
    {
      "epoch": 23.1913357400722,
      "grad_norm": Infinity,
      "learning_rate": 0.00017561390973400465,
      "loss": 11.0734,
      "step": 192720
    },
    {
      "epoch": 23.19253910950662,
      "grad_norm": Infinity,
      "learning_rate": 0.00017561142012353286,
      "loss": 11.1146,
      "step": 192730
    },
    {
      "epoch": 23.193742478941036,
      "grad_norm": Infinity,
      "learning_rate": 0.00017560893040363267,
      "loss": 11.1375,
      "step": 192740
    },
    {
      "epoch": 23.19494584837545,
      "grad_norm": Infinity,
      "learning_rate": 0.0001756064405743077,
      "loss": 11.132,
      "step": 192750
    },
    {
      "epoch": 23.196149217809868,
      "grad_norm": Infinity,
      "learning_rate": 0.0001756039506355615,
      "loss": 11.1629,
      "step": 192760
    },
    {
      "epoch": 23.197352587244286,
      "grad_norm": Infinity,
      "learning_rate": 0.00017560146058739776,
      "loss": 11.1582,
      "step": 192770
    },
    {
      "epoch": 23.1985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017559897042982003,
      "loss": 11.1921,
      "step": 192780
    },
    {
      "epoch": 23.199759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00017559648016283194,
      "loss": 11.1661,
      "step": 192790
    },
    {
      "epoch": 23.200962695547535,
      "grad_norm": Infinity,
      "learning_rate": 0.00017559398978643707,
      "loss": 11.139,
      "step": 192800
    },
    {
      "epoch": 23.20216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.00017559149930063904,
      "loss": 11.2258,
      "step": 192810
    },
    {
      "epoch": 23.203369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.00017558900870544143,
      "loss": 11.223,
      "step": 192820
    },
    {
      "epoch": 23.204572803850784,
      "grad_norm": Infinity,
      "learning_rate": 0.00017558651800084787,
      "loss": 11.0765,
      "step": 192830
    },
    {
      "epoch": 23.205776173285198,
      "grad_norm": Infinity,
      "learning_rate": 0.00017558402718686195,
      "loss": 11.1681,
      "step": 192840
    },
    {
      "epoch": 23.206979542719615,
      "grad_norm": Infinity,
      "learning_rate": 0.0001755815362634873,
      "loss": 11.0835,
      "step": 192850
    },
    {
      "epoch": 23.20818291215403,
      "grad_norm": Infinity,
      "learning_rate": 0.00017557904523072748,
      "loss": 11.1196,
      "step": 192860
    },
    {
      "epoch": 23.209386281588447,
      "grad_norm": Infinity,
      "learning_rate": 0.00017557655408858614,
      "loss": 11.2878,
      "step": 192870
    },
    {
      "epoch": 23.210589651022865,
      "grad_norm": Infinity,
      "learning_rate": 0.00017557406283706686,
      "loss": 11.2261,
      "step": 192880
    },
    {
      "epoch": 23.21179302045728,
      "grad_norm": Infinity,
      "learning_rate": 0.00017557157147617326,
      "loss": 11.2108,
      "step": 192890
    },
    {
      "epoch": 23.212996389891696,
      "grad_norm": Infinity,
      "learning_rate": 0.00017556908000590892,
      "loss": 11.223,
      "step": 192900
    },
    {
      "epoch": 23.214199759326114,
      "grad_norm": Infinity,
      "learning_rate": 0.00017556658842627748,
      "loss": 11.1965,
      "step": 192910
    },
    {
      "epoch": 23.215403128760528,
      "grad_norm": Infinity,
      "learning_rate": 0.00017556409673728253,
      "loss": 11.1263,
      "step": 192920
    },
    {
      "epoch": 23.216606498194945,
      "grad_norm": Infinity,
      "learning_rate": 0.00017556160493892767,
      "loss": 11.158,
      "step": 192930
    },
    {
      "epoch": 23.217809867629363,
      "grad_norm": Infinity,
      "learning_rate": 0.00017555911303121654,
      "loss": 11.2259,
      "step": 192940
    },
    {
      "epoch": 23.219013237063777,
      "grad_norm": Infinity,
      "learning_rate": 0.0001755566210141527,
      "loss": 11.2653,
      "step": 192950
    },
    {
      "epoch": 23.220216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.00017555412888773975,
      "loss": 11.2626,
      "step": 192960
    },
    {
      "epoch": 23.221419975932612,
      "grad_norm": Infinity,
      "learning_rate": 0.00017555163665198132,
      "loss": 11.1391,
      "step": 192970
    },
    {
      "epoch": 23.222623345367026,
      "grad_norm": Infinity,
      "learning_rate": 0.00017554914430688106,
      "loss": 11.1943,
      "step": 192980
    },
    {
      "epoch": 23.223826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.0001755466518524425,
      "loss": 11.1316,
      "step": 192990
    },
    {
      "epoch": 23.22503008423586,
      "grad_norm": Infinity,
      "learning_rate": 0.0001755441592886693,
      "loss": 11.153,
      "step": 193000
    },
    {
      "epoch": 23.226233453670275,
      "grad_norm": Infinity,
      "learning_rate": 0.00017554166661556505,
      "loss": 11.1116,
      "step": 193010
    },
    {
      "epoch": 23.227436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00017553917383313335,
      "loss": 11.252,
      "step": 193020
    },
    {
      "epoch": 23.22864019253911,
      "grad_norm": Infinity,
      "learning_rate": 0.00017553668094137784,
      "loss": 11.1595,
      "step": 193030
    },
    {
      "epoch": 23.229843561973524,
      "grad_norm": Infinity,
      "learning_rate": 0.00017553418794030208,
      "loss": 11.0905,
      "step": 193040
    },
    {
      "epoch": 23.231046931407942,
      "grad_norm": Infinity,
      "learning_rate": 0.0001755316948299097,
      "loss": 11.1444,
      "step": 193050
    },
    {
      "epoch": 23.23225030084236,
      "grad_norm": Infinity,
      "learning_rate": 0.0001755292016102043,
      "loss": 11.064,
      "step": 193060
    },
    {
      "epoch": 23.233453670276774,
      "grad_norm": Infinity,
      "learning_rate": 0.00017552670828118952,
      "loss": 11.0443,
      "step": 193070
    },
    {
      "epoch": 23.23465703971119,
      "grad_norm": Infinity,
      "learning_rate": 0.00017552421484286893,
      "loss": 11.23,
      "step": 193080
    },
    {
      "epoch": 23.23586040914561,
      "grad_norm": Infinity,
      "learning_rate": 0.00017552172129524618,
      "loss": 11.2578,
      "step": 193090
    },
    {
      "epoch": 23.237063778580023,
      "grad_norm": Infinity,
      "learning_rate": 0.00017551922763832484,
      "loss": 11.1379,
      "step": 193100
    },
    {
      "epoch": 23.23826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.00017551673387210853,
      "loss": 11.1566,
      "step": 193110
    },
    {
      "epoch": 23.239470517448858,
      "grad_norm": Infinity,
      "learning_rate": 0.00017551423999660087,
      "loss": 11.1361,
      "step": 193120
    },
    {
      "epoch": 23.240673886883272,
      "grad_norm": Infinity,
      "learning_rate": 0.0001755117460118054,
      "loss": 11.1516,
      "step": 193130
    },
    {
      "epoch": 23.24187725631769,
      "grad_norm": Infinity,
      "learning_rate": 0.0001755092519177259,
      "loss": 11.1199,
      "step": 193140
    },
    {
      "epoch": 23.243080625752107,
      "grad_norm": Infinity,
      "learning_rate": 0.0001755067577143658,
      "loss": 11.1568,
      "step": 193150
    },
    {
      "epoch": 23.24428399518652,
      "grad_norm": Infinity,
      "learning_rate": 0.00017550426340172878,
      "loss": 11.1194,
      "step": 193160
    },
    {
      "epoch": 23.24548736462094,
      "grad_norm": Infinity,
      "learning_rate": 0.00017550176897981846,
      "loss": 11.1797,
      "step": 193170
    },
    {
      "epoch": 23.246690734055356,
      "grad_norm": Infinity,
      "learning_rate": 0.00017549927444863845,
      "loss": 11.1113,
      "step": 193180
    },
    {
      "epoch": 23.24789410348977,
      "grad_norm": Infinity,
      "learning_rate": 0.00017549677980819236,
      "loss": 11.1766,
      "step": 193190
    },
    {
      "epoch": 23.249097472924188,
      "grad_norm": Infinity,
      "learning_rate": 0.00017549428505848376,
      "loss": 11.1253,
      "step": 193200
    },
    {
      "epoch": 23.250300842358605,
      "grad_norm": Infinity,
      "learning_rate": 0.00017549179019951628,
      "loss": 11.0924,
      "step": 193210
    },
    {
      "epoch": 23.25150421179302,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754892952312936,
      "loss": 11.2372,
      "step": 193220
    },
    {
      "epoch": 23.252707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00017548680015381919,
      "loss": 11.136,
      "step": 193230
    },
    {
      "epoch": 23.253910950661854,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754843049670968,
      "loss": 11.2663,
      "step": 193240
    },
    {
      "epoch": 23.25511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00017548180967112998,
      "loss": 11.0405,
      "step": 193250
    },
    {
      "epoch": 23.256317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00017547931426592235,
      "loss": 11.0678,
      "step": 193260
    },
    {
      "epoch": 23.257521058965104,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754768187514775,
      "loss": 11.2024,
      "step": 193270
    },
    {
      "epoch": 23.258724428399518,
      "grad_norm": Infinity,
      "learning_rate": 0.00017547432312779905,
      "loss": 11.1946,
      "step": 193280
    },
    {
      "epoch": 23.259927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00017547182739489063,
      "loss": 11.1317,
      "step": 193290
    },
    {
      "epoch": 23.261131167268353,
      "grad_norm": Infinity,
      "learning_rate": 0.00017546933155275585,
      "loss": 11.0937,
      "step": 193300
    },
    {
      "epoch": 23.262334536702767,
      "grad_norm": Infinity,
      "learning_rate": 0.00017546683560139829,
      "loss": 11.0807,
      "step": 193310
    },
    {
      "epoch": 23.263537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.00017546433954082163,
      "loss": 11.2003,
      "step": 193320
    },
    {
      "epoch": 23.264741275571602,
      "grad_norm": Infinity,
      "learning_rate": 0.00017546184337102942,
      "loss": 11.076,
      "step": 193330
    },
    {
      "epoch": 23.265944645006016,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754593470920253,
      "loss": 11.0659,
      "step": 193340
    },
    {
      "epoch": 23.267148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00017545685070381284,
      "loss": 11.1645,
      "step": 193350
    },
    {
      "epoch": 23.26835138387485,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754543542063957,
      "loss": 11.1878,
      "step": 193360
    },
    {
      "epoch": 23.269554753309265,
      "grad_norm": Infinity,
      "learning_rate": 0.00017545185759977749,
      "loss": 11.2552,
      "step": 193370
    },
    {
      "epoch": 23.270758122743683,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754493608839618,
      "loss": 11.2498,
      "step": 193380
    },
    {
      "epoch": 23.2719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754468640589523,
      "loss": 11.0623,
      "step": 193390
    },
    {
      "epoch": 23.273164861612514,
      "grad_norm": Infinity,
      "learning_rate": 0.00017544436712475255,
      "loss": 11.1857,
      "step": 193400
    },
    {
      "epoch": 23.27436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00017544187008136612,
      "loss": 11.0667,
      "step": 193410
    },
    {
      "epoch": 23.27557160048135,
      "grad_norm": Infinity,
      "learning_rate": 0.00017543937292879673,
      "loss": 11.0717,
      "step": 193420
    },
    {
      "epoch": 23.276774969915763,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754368756670479,
      "loss": 11.0537,
      "step": 193430
    },
    {
      "epoch": 23.27797833935018,
      "grad_norm": Infinity,
      "learning_rate": 0.00017543437829612332,
      "loss": 11.0827,
      "step": 193440
    },
    {
      "epoch": 23.2791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754318808160266,
      "loss": 11.1383,
      "step": 193450
    },
    {
      "epoch": 23.280385078219012,
      "grad_norm": Infinity,
      "learning_rate": 0.00017542938322676128,
      "loss": 11.2067,
      "step": 193460
    },
    {
      "epoch": 23.28158844765343,
      "grad_norm": Infinity,
      "learning_rate": 0.000175426885528331,
      "loss": 11.0624,
      "step": 193470
    },
    {
      "epoch": 23.282791817087848,
      "grad_norm": Infinity,
      "learning_rate": 0.00017542438772073945,
      "loss": 11.0763,
      "step": 193480
    },
    {
      "epoch": 23.28399518652226,
      "grad_norm": Infinity,
      "learning_rate": 0.00017542188980399016,
      "loss": 11.1682,
      "step": 193490
    },
    {
      "epoch": 23.28519855595668,
      "grad_norm": Infinity,
      "learning_rate": 0.00017541939177808676,
      "loss": 11.2145,
      "step": 193500
    },
    {
      "epoch": 23.286401925391097,
      "grad_norm": Infinity,
      "learning_rate": 0.00017541689364303288,
      "loss": 11.0859,
      "step": 193510
    },
    {
      "epoch": 23.28760529482551,
      "grad_norm": Infinity,
      "learning_rate": 0.00017541439539883215,
      "loss": 11.1552,
      "step": 193520
    },
    {
      "epoch": 23.28880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754118970454882,
      "loss": 11.0117,
      "step": 193530
    },
    {
      "epoch": 23.290012033694346,
      "grad_norm": Infinity,
      "learning_rate": 0.00017540939858300456,
      "loss": 11.1963,
      "step": 193540
    },
    {
      "epoch": 23.29121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00017540690001138494,
      "loss": 11.1663,
      "step": 193550
    },
    {
      "epoch": 23.292418772563177,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754044013306329,
      "loss": 11.1226,
      "step": 193560
    },
    {
      "epoch": 23.29362214199759,
      "grad_norm": Infinity,
      "learning_rate": 0.0001754019025407521,
      "loss": 11.2278,
      "step": 193570
    },
    {
      "epoch": 23.29482551143201,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753994036417461,
      "loss": 11.129,
      "step": 193580
    },
    {
      "epoch": 23.296028880866427,
      "grad_norm": Infinity,
      "learning_rate": 0.00017539690463361856,
      "loss": 11.0296,
      "step": 193590
    },
    {
      "epoch": 23.29723225030084,
      "grad_norm": Infinity,
      "learning_rate": 0.00017539440551637308,
      "loss": 11.1274,
      "step": 193600
    },
    {
      "epoch": 23.29843561973526,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753919062900133,
      "loss": 11.1067,
      "step": 193610
    },
    {
      "epoch": 23.299638989169676,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753894069545428,
      "loss": 11.1685,
      "step": 193620
    },
    {
      "epoch": 23.30084235860409,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753869075099652,
      "loss": 11.019,
      "step": 193630
    },
    {
      "epoch": 23.302045728038507,
      "grad_norm": Infinity,
      "learning_rate": 0.00017538440795628418,
      "loss": 11.1925,
      "step": 193640
    },
    {
      "epoch": 23.303249097472925,
      "grad_norm": Infinity,
      "learning_rate": 0.00017538190829350327,
      "loss": 11.1634,
      "step": 193650
    },
    {
      "epoch": 23.30445246690734,
      "grad_norm": Infinity,
      "learning_rate": 0.00017537940852162614,
      "loss": 11.0511,
      "step": 193660
    },
    {
      "epoch": 23.305655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00017537690864065638,
      "loss": 11.1921,
      "step": 193670
    },
    {
      "epoch": 23.306859205776174,
      "grad_norm": Infinity,
      "learning_rate": 0.00017537440865059763,
      "loss": 11.1834,
      "step": 193680
    },
    {
      "epoch": 23.308062575210588,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753719085514535,
      "loss": 11.1029,
      "step": 193690
    },
    {
      "epoch": 23.309265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00017536940834322762,
      "loss": 11.1336,
      "step": 193700
    },
    {
      "epoch": 23.310469314079423,
      "grad_norm": Infinity,
      "learning_rate": 0.00017536690802592362,
      "loss": 11.1643,
      "step": 193710
    },
    {
      "epoch": 23.311672683513837,
      "grad_norm": Infinity,
      "learning_rate": 0.00017536440759954506,
      "loss": 11.0536,
      "step": 193720
    },
    {
      "epoch": 23.312876052948255,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753619070640956,
      "loss": 11.0405,
      "step": 193730
    },
    {
      "epoch": 23.314079422382672,
      "grad_norm": Infinity,
      "learning_rate": 0.00017535940641957883,
      "loss": 11.2063,
      "step": 193740
    },
    {
      "epoch": 23.315282791817086,
      "grad_norm": Infinity,
      "learning_rate": 0.00017535690566599846,
      "loss": 11.0488,
      "step": 193750
    },
    {
      "epoch": 23.316486161251504,
      "grad_norm": Infinity,
      "learning_rate": 0.000175354404803358,
      "loss": 11.0255,
      "step": 193760
    },
    {
      "epoch": 23.31768953068592,
      "grad_norm": Infinity,
      "learning_rate": 0.00017535190383166112,
      "loss": 11.1295,
      "step": 193770
    },
    {
      "epoch": 23.318892900120336,
      "grad_norm": Infinity,
      "learning_rate": 0.00017534940275091144,
      "loss": 11.1352,
      "step": 193780
    },
    {
      "epoch": 23.320096269554753,
      "grad_norm": Infinity,
      "learning_rate": 0.00017534690156111252,
      "loss": 11.0754,
      "step": 193790
    },
    {
      "epoch": 23.32129963898917,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753444002622681,
      "loss": 11.1256,
      "step": 193800
    },
    {
      "epoch": 23.322503008423585,
      "grad_norm": Infinity,
      "learning_rate": 0.00017534189885438171,
      "loss": 11.0896,
      "step": 193810
    },
    {
      "epoch": 23.323706377858002,
      "grad_norm": Infinity,
      "learning_rate": 0.000175339397337457,
      "loss": 11.0594,
      "step": 193820
    },
    {
      "epoch": 23.32490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00017533689571149754,
      "loss": 11.1942,
      "step": 193830
    },
    {
      "epoch": 23.326113116726834,
      "grad_norm": Infinity,
      "learning_rate": 0.00017533439397650702,
      "loss": 11.1883,
      "step": 193840
    },
    {
      "epoch": 23.32731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.00017533189213248904,
      "loss": 11.0362,
      "step": 193850
    },
    {
      "epoch": 23.32851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.00017532939017944718,
      "loss": 11.2057,
      "step": 193860
    },
    {
      "epoch": 23.329723225030083,
      "grad_norm": Infinity,
      "learning_rate": 0.00017532688811738514,
      "loss": 11.1193,
      "step": 193870
    },
    {
      "epoch": 23.3309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.00017532438594630647,
      "loss": 11.123,
      "step": 193880
    },
    {
      "epoch": 23.332129963898918,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753218836662148,
      "loss": 11.1341,
      "step": 193890
    },
    {
      "epoch": 23.333333333333332,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753193812771138,
      "loss": 11.0933,
      "step": 193900
    },
    {
      "epoch": 23.33453670276775,
      "grad_norm": Infinity,
      "learning_rate": 0.00017531687877900704,
      "loss": 11.1447,
      "step": 193910
    },
    {
      "epoch": 23.335740072202167,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753143761718982,
      "loss": 11.1411,
      "step": 193920
    },
    {
      "epoch": 23.33694344163658,
      "grad_norm": Infinity,
      "learning_rate": 0.0001753118734557908,
      "loss": 11.1185,
      "step": 193930
    },
    {
      "epoch": 23.338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00017530937063068859,
      "loss": 11.0155,
      "step": 193940
    },
    {
      "epoch": 23.339350180505416,
      "grad_norm": Infinity,
      "learning_rate": 0.00017530686769659507,
      "loss": 11.24,
      "step": 193950
    },
    {
      "epoch": 23.34055354993983,
      "grad_norm": Infinity,
      "learning_rate": 0.00017530436465351394,
      "loss": 11.1776,
      "step": 193960
    },
    {
      "epoch": 23.341756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.00017530186150144879,
      "loss": 11.1802,
      "step": 193970
    },
    {
      "epoch": 23.342960288808666,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752993582404033,
      "loss": 11.1322,
      "step": 193980
    },
    {
      "epoch": 23.34416365824308,
      "grad_norm": Infinity,
      "learning_rate": 0.00017529685487038102,
      "loss": 11.0815,
      "step": 193990
    },
    {
      "epoch": 23.345367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017529435139138557,
      "loss": 11.1885,
      "step": 194000
    },
    {
      "epoch": 23.346570397111915,
      "grad_norm": Infinity,
      "learning_rate": 0.00017529184780342065,
      "loss": 11.2131,
      "step": 194010
    },
    {
      "epoch": 23.34777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.00017528934410648983,
      "loss": 11.248,
      "step": 194020
    },
    {
      "epoch": 23.348977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752868403005967,
      "loss": 11.0511,
      "step": 194030
    },
    {
      "epoch": 23.350180505415164,
      "grad_norm": Infinity,
      "learning_rate": 0.00017528433638574497,
      "loss": 11.1162,
      "step": 194040
    },
    {
      "epoch": 23.351383874849578,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752818323619382,
      "loss": 11.1686,
      "step": 194050
    },
    {
      "epoch": 23.352587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00017527932822918003,
      "loss": 11.1534,
      "step": 194060
    },
    {
      "epoch": 23.353790613718413,
      "grad_norm": Infinity,
      "learning_rate": 0.00017527682398747413,
      "loss": 11.1434,
      "step": 194070
    },
    {
      "epoch": 23.354993983152827,
      "grad_norm": Infinity,
      "learning_rate": 0.000175274319636824,
      "loss": 11.0826,
      "step": 194080
    },
    {
      "epoch": 23.356197352587245,
      "grad_norm": Infinity,
      "learning_rate": 0.00017527181517723342,
      "loss": 11.1348,
      "step": 194090
    },
    {
      "epoch": 23.357400722021662,
      "grad_norm": Infinity,
      "learning_rate": 0.00017526931060870592,
      "loss": 11.1702,
      "step": 194100
    },
    {
      "epoch": 23.358604091456076,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752668059312451,
      "loss": 11.1623,
      "step": 194110
    },
    {
      "epoch": 23.359807460890494,
      "grad_norm": Infinity,
      "learning_rate": 0.00017526430114485467,
      "loss": 11.0346,
      "step": 194120
    },
    {
      "epoch": 23.36101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752617962495382,
      "loss": 11.1335,
      "step": 194130
    },
    {
      "epoch": 23.362214199759325,
      "grad_norm": Infinity,
      "learning_rate": 0.00017525929124529932,
      "loss": 11.0116,
      "step": 194140
    },
    {
      "epoch": 23.363417569193743,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752567861321417,
      "loss": 11.1241,
      "step": 194150
    },
    {
      "epoch": 23.36462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.00017525428091006887,
      "loss": 11.286,
      "step": 194160
    },
    {
      "epoch": 23.365824308062574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017525177557908456,
      "loss": 11.1952,
      "step": 194170
    },
    {
      "epoch": 23.367027677496992,
      "grad_norm": Infinity,
      "learning_rate": 0.00017524927013919234,
      "loss": 11.1059,
      "step": 194180
    },
    {
      "epoch": 23.36823104693141,
      "grad_norm": Infinity,
      "learning_rate": 0.00017524676459039587,
      "loss": 11.1561,
      "step": 194190
    },
    {
      "epoch": 23.369434416365824,
      "grad_norm": Infinity,
      "learning_rate": 0.00017524425893269873,
      "loss": 11.2982,
      "step": 194200
    },
    {
      "epoch": 23.37063778580024,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752417531661046,
      "loss": 11.168,
      "step": 194210
    },
    {
      "epoch": 23.37184115523466,
      "grad_norm": Infinity,
      "learning_rate": 0.00017523924729061701,
      "loss": 11.1924,
      "step": 194220
    },
    {
      "epoch": 23.373044524669073,
      "grad_norm": Infinity,
      "learning_rate": 0.00017523674130623973,
      "loss": 11.1715,
      "step": 194230
    },
    {
      "epoch": 23.37424789410349,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752342352129763,
      "loss": 11.1251,
      "step": 194240
    },
    {
      "epoch": 23.375451263537904,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752317290108303,
      "loss": 11.0972,
      "step": 194250
    },
    {
      "epoch": 23.376654632972322,
      "grad_norm": Infinity,
      "learning_rate": 0.00017522922269980545,
      "loss": 11.1484,
      "step": 194260
    },
    {
      "epoch": 23.37785800240674,
      "grad_norm": Infinity,
      "learning_rate": 0.00017522671627990535,
      "loss": 11.1995,
      "step": 194270
    },
    {
      "epoch": 23.379061371841154,
      "grad_norm": Infinity,
      "learning_rate": 0.00017522420975113362,
      "loss": 11.1673,
      "step": 194280
    },
    {
      "epoch": 23.38026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00017522170311349387,
      "loss": 11.124,
      "step": 194290
    },
    {
      "epoch": 23.38146811070999,
      "grad_norm": Infinity,
      "learning_rate": 0.00017521919636698973,
      "loss": 11.2001,
      "step": 194300
    },
    {
      "epoch": 23.382671480144403,
      "grad_norm": Infinity,
      "learning_rate": 0.00017521668951162486,
      "loss": 11.212,
      "step": 194310
    },
    {
      "epoch": 23.38387484957882,
      "grad_norm": Infinity,
      "learning_rate": 0.00017521418254740287,
      "loss": 11.0148,
      "step": 194320
    },
    {
      "epoch": 23.385078219013238,
      "grad_norm": Infinity,
      "learning_rate": 0.0001752116754743274,
      "loss": 11.1504,
      "step": 194330
    },
    {
      "epoch": 23.386281588447652,
      "grad_norm": Infinity,
      "learning_rate": 0.00017520916829240205,
      "loss": 11.1509,
      "step": 194340
    },
    {
      "epoch": 23.38748495788207,
      "grad_norm": Infinity,
      "learning_rate": 0.00017520666100163048,
      "loss": 11.0231,
      "step": 194350
    },
    {
      "epoch": 23.388688327316487,
      "grad_norm": Infinity,
      "learning_rate": 0.00017520415360201626,
      "loss": 11.1179,
      "step": 194360
    },
    {
      "epoch": 23.3898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00017520164609356314,
      "loss": 11.2122,
      "step": 194370
    },
    {
      "epoch": 23.39109506618532,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751991384762746,
      "loss": 11.2921,
      "step": 194380
    },
    {
      "epoch": 23.392298435619736,
      "grad_norm": Infinity,
      "learning_rate": 0.00017519663075015439,
      "loss": 11.2267,
      "step": 194390
    },
    {
      "epoch": 23.39350180505415,
      "grad_norm": Infinity,
      "learning_rate": 0.00017519412291520605,
      "loss": 11.097,
      "step": 194400
    },
    {
      "epoch": 23.394705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017519161497143326,
      "loss": 11.1614,
      "step": 194410
    },
    {
      "epoch": 23.395908543922985,
      "grad_norm": Infinity,
      "learning_rate": 0.00017518910691883966,
      "loss": 11.1221,
      "step": 194420
    },
    {
      "epoch": 23.3971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751865987574288,
      "loss": 11.1046,
      "step": 194430
    },
    {
      "epoch": 23.398315282791817,
      "grad_norm": Infinity,
      "learning_rate": 0.00017518409048720445,
      "loss": 11.0423,
      "step": 194440
    },
    {
      "epoch": 23.399518652226234,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751815821081701,
      "loss": 11.1207,
      "step": 194450
    },
    {
      "epoch": 23.40072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00017517907362032948,
      "loss": 11.1068,
      "step": 194460
    },
    {
      "epoch": 23.401925391095066,
      "grad_norm": Infinity,
      "learning_rate": 0.00017517656502368616,
      "loss": 11.0872,
      "step": 194470
    },
    {
      "epoch": 23.403128760529484,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751740563182438,
      "loss": 11.1543,
      "step": 194480
    },
    {
      "epoch": 23.404332129963898,
      "grad_norm": Infinity,
      "learning_rate": 0.000175171547504006,
      "loss": 11.2175,
      "step": 194490
    },
    {
      "epoch": 23.405535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00017516903858097642,
      "loss": 11.2075,
      "step": 194500
    },
    {
      "epoch": 23.406738868832733,
      "grad_norm": Infinity,
      "learning_rate": 0.00017516652954915868,
      "loss": 10.9674,
      "step": 194510
    },
    {
      "epoch": 23.407942238267147,
      "grad_norm": Infinity,
      "learning_rate": 0.00017516402040855641,
      "loss": 11.05,
      "step": 194520
    },
    {
      "epoch": 23.409145607701564,
      "grad_norm": Infinity,
      "learning_rate": 0.00017516151115917324,
      "loss": 11.037,
      "step": 194530
    },
    {
      "epoch": 23.410348977135982,
      "grad_norm": Infinity,
      "learning_rate": 0.00017515900180101282,
      "loss": 11.1966,
      "step": 194540
    },
    {
      "epoch": 23.411552346570396,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751564923340788,
      "loss": 11.2884,
      "step": 194550
    },
    {
      "epoch": 23.412755716004813,
      "grad_norm": Infinity,
      "learning_rate": 0.00017515398275837476,
      "loss": 11.2552,
      "step": 194560
    },
    {
      "epoch": 23.41395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017515147307390431,
      "loss": 11.1307,
      "step": 194570
    },
    {
      "epoch": 23.415162454873645,
      "grad_norm": Infinity,
      "learning_rate": 0.00017514896328067116,
      "loss": 11.0145,
      "step": 194580
    },
    {
      "epoch": 23.416365824308063,
      "grad_norm": Infinity,
      "learning_rate": 0.00017514645337867889,
      "loss": 11.1513,
      "step": 194590
    },
    {
      "epoch": 23.41756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.00017514394336793116,
      "loss": 11.2107,
      "step": 194600
    },
    {
      "epoch": 23.418772563176894,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751414332484316,
      "loss": 11.3053,
      "step": 194610
    },
    {
      "epoch": 23.41997593261131,
      "grad_norm": Infinity,
      "learning_rate": 0.00017513892302018383,
      "loss": 11.0757,
      "step": 194620
    },
    {
      "epoch": 23.42117930204573,
      "grad_norm": Infinity,
      "learning_rate": 0.00017513641268319146,
      "loss": 11.1309,
      "step": 194630
    },
    {
      "epoch": 23.422382671480143,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751339022374582,
      "loss": 11.0912,
      "step": 194640
    },
    {
      "epoch": 23.42358604091456,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751313916829876,
      "loss": 11.2072,
      "step": 194650
    },
    {
      "epoch": 23.42478941034898,
      "grad_norm": Infinity,
      "learning_rate": 0.00017512888101978334,
      "loss": 11.1931,
      "step": 194660
    },
    {
      "epoch": 23.425992779783392,
      "grad_norm": Infinity,
      "learning_rate": 0.00017512637024784904,
      "loss": 11.2085,
      "step": 194670
    },
    {
      "epoch": 23.42719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00017512385936718832,
      "loss": 11.1922,
      "step": 194680
    },
    {
      "epoch": 23.428399518652228,
      "grad_norm": Infinity,
      "learning_rate": 0.00017512134837780483,
      "loss": 11.1161,
      "step": 194690
    },
    {
      "epoch": 23.42960288808664,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751188372797022,
      "loss": 11.1434,
      "step": 194700
    },
    {
      "epoch": 23.43080625752106,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751163260728841,
      "loss": 11.0348,
      "step": 194710
    },
    {
      "epoch": 23.432009626955477,
      "grad_norm": Infinity,
      "learning_rate": 0.00017511381475735407,
      "loss": 11.1551,
      "step": 194720
    },
    {
      "epoch": 23.43321299638989,
      "grad_norm": Infinity,
      "learning_rate": 0.00017511130333311585,
      "loss": 11.1053,
      "step": 194730
    },
    {
      "epoch": 23.43441636582431,
      "grad_norm": Infinity,
      "learning_rate": 0.00017510879180017303,
      "loss": 11.1571,
      "step": 194740
    },
    {
      "epoch": 23.435619735258726,
      "grad_norm": Infinity,
      "learning_rate": 0.0001751062801585292,
      "loss": 11.049,
      "step": 194750
    },
    {
      "epoch": 23.43682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00017510376840818812,
      "loss": 11.1627,
      "step": 194760
    },
    {
      "epoch": 23.438026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00017510125654915328,
      "loss": 11.2024,
      "step": 194770
    },
    {
      "epoch": 23.439229843561975,
      "grad_norm": Infinity,
      "learning_rate": 0.0001750987445814284,
      "loss": 11.1581,
      "step": 194780
    },
    {
      "epoch": 23.44043321299639,
      "grad_norm": Infinity,
      "learning_rate": 0.0001750962325050171,
      "loss": 11.1818,
      "step": 194790
    },
    {
      "epoch": 23.441636582430807,
      "grad_norm": Infinity,
      "learning_rate": 0.00017509372031992298,
      "loss": 11.0888,
      "step": 194800
    },
    {
      "epoch": 23.442839951865224,
      "grad_norm": Infinity,
      "learning_rate": 0.00017509120802614977,
      "loss": 11.2685,
      "step": 194810
    },
    {
      "epoch": 23.444043321299638,
      "grad_norm": Infinity,
      "learning_rate": 0.00017508869562370098,
      "loss": 11.1844,
      "step": 194820
    },
    {
      "epoch": 23.445246690734056,
      "grad_norm": Infinity,
      "learning_rate": 0.00017508618311258033,
      "loss": 11.0798,
      "step": 194830
    },
    {
      "epoch": 23.446450060168473,
      "grad_norm": Infinity,
      "learning_rate": 0.00017508367049279145,
      "loss": 11.2141,
      "step": 194840
    },
    {
      "epoch": 23.447653429602887,
      "grad_norm": Infinity,
      "learning_rate": 0.00017508115776433794,
      "loss": 11.124,
      "step": 194850
    },
    {
      "epoch": 23.448856799037305,
      "grad_norm": Infinity,
      "learning_rate": 0.00017507864492722345,
      "loss": 11.2347,
      "step": 194860
    },
    {
      "epoch": 23.450060168471722,
      "grad_norm": Infinity,
      "learning_rate": 0.00017507613198145165,
      "loss": 11.1683,
      "step": 194870
    },
    {
      "epoch": 23.451263537906136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017507361892702615,
      "loss": 11.1772,
      "step": 194880
    },
    {
      "epoch": 23.452466907340554,
      "grad_norm": Infinity,
      "learning_rate": 0.00017507110576395055,
      "loss": 11.2221,
      "step": 194890
    },
    {
      "epoch": 23.45367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017506859249222858,
      "loss": 11.0778,
      "step": 194900
    },
    {
      "epoch": 23.454873646209386,
      "grad_norm": Infinity,
      "learning_rate": 0.0001750660791118638,
      "loss": 11.1625,
      "step": 194910
    },
    {
      "epoch": 23.456077015643803,
      "grad_norm": Infinity,
      "learning_rate": 0.00017506356562285987,
      "loss": 11.1564,
      "step": 194920
    },
    {
      "epoch": 23.45728038507822,
      "grad_norm": Infinity,
      "learning_rate": 0.0001750610520252204,
      "loss": 11.2514,
      "step": 194930
    },
    {
      "epoch": 23.458483754512635,
      "grad_norm": Infinity,
      "learning_rate": 0.00017505853831894909,
      "loss": 11.1341,
      "step": 194940
    },
    {
      "epoch": 23.459687123947052,
      "grad_norm": Infinity,
      "learning_rate": 0.00017505602450404955,
      "loss": 11.0626,
      "step": 194950
    },
    {
      "epoch": 23.460890493381466,
      "grad_norm": Infinity,
      "learning_rate": 0.0001750535105805254,
      "loss": 11.1028,
      "step": 194960
    },
    {
      "epoch": 23.462093862815884,
      "grad_norm": Infinity,
      "learning_rate": 0.00017505099654838027,
      "loss": 11.1309,
      "step": 194970
    },
    {
      "epoch": 23.4632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00017504848240761786,
      "loss": 11.224,
      "step": 194980
    },
    {
      "epoch": 23.464500601684716,
      "grad_norm": Infinity,
      "learning_rate": 0.00017504596815824176,
      "loss": 11.0937,
      "step": 194990
    },
    {
      "epoch": 23.465703971119133,
      "grad_norm": Infinity,
      "learning_rate": 0.00017504345380025556,
      "loss": 11.0216,
      "step": 195000
    },
    {
      "epoch": 23.46690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.000175040939333663,
      "loss": 11.1226,
      "step": 195010
    },
    {
      "epoch": 23.468110709987965,
      "grad_norm": Infinity,
      "learning_rate": 0.00017503842475846768,
      "loss": 11.1569,
      "step": 195020
    },
    {
      "epoch": 23.469314079422382,
      "grad_norm": Infinity,
      "learning_rate": 0.0001750359100746732,
      "loss": 11.1827,
      "step": 195030
    },
    {
      "epoch": 23.4705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017503339528228325,
      "loss": 11.1777,
      "step": 195040
    },
    {
      "epoch": 23.471720818291214,
      "grad_norm": Infinity,
      "learning_rate": 0.00017503088038130145,
      "loss": 11.007,
      "step": 195050
    },
    {
      "epoch": 23.47292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00017502836537173146,
      "loss": 11.1805,
      "step": 195060
    },
    {
      "epoch": 23.47412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.0001750258502535769,
      "loss": 11.0962,
      "step": 195070
    },
    {
      "epoch": 23.475330926594463,
      "grad_norm": Infinity,
      "learning_rate": 0.00017502333502684137,
      "loss": 11.2764,
      "step": 195080
    },
    {
      "epoch": 23.47653429602888,
      "grad_norm": Infinity,
      "learning_rate": 0.00017502081969152862,
      "loss": 11.0084,
      "step": 195090
    },
    {
      "epoch": 23.477737665463298,
      "grad_norm": Infinity,
      "learning_rate": 0.00017501830424764217,
      "loss": 11.0039,
      "step": 195100
    },
    {
      "epoch": 23.478941034897712,
      "grad_norm": Infinity,
      "learning_rate": 0.0001750157886951857,
      "loss": 11.1472,
      "step": 195110
    },
    {
      "epoch": 23.48014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017501327303416292,
      "loss": 11.1366,
      "step": 195120
    },
    {
      "epoch": 23.481347773766547,
      "grad_norm": Infinity,
      "learning_rate": 0.00017501075726457737,
      "loss": 11.1748,
      "step": 195130
    },
    {
      "epoch": 23.48255114320096,
      "grad_norm": Infinity,
      "learning_rate": 0.00017500824138643275,
      "loss": 11.0998,
      "step": 195140
    },
    {
      "epoch": 23.48375451263538,
      "grad_norm": Infinity,
      "learning_rate": 0.00017500572539973268,
      "loss": 11.223,
      "step": 195150
    },
    {
      "epoch": 23.484957882069796,
      "grad_norm": Infinity,
      "learning_rate": 0.0001750032093044808,
      "loss": 10.9834,
      "step": 195160
    },
    {
      "epoch": 23.48616125150421,
      "grad_norm": Infinity,
      "learning_rate": 0.00017500069310068077,
      "loss": 11.2017,
      "step": 195170
    },
    {
      "epoch": 23.487364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.00017499817678833624,
      "loss": 11.1433,
      "step": 195180
    },
    {
      "epoch": 23.488567990373046,
      "grad_norm": Infinity,
      "learning_rate": 0.0001749956603674508,
      "loss": 11.1318,
      "step": 195190
    },
    {
      "epoch": 23.48977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.00017499314383802813,
      "loss": 11.2182,
      "step": 195200
    },
    {
      "epoch": 23.490974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00017499062720007186,
      "loss": 11.1747,
      "step": 195210
    },
    {
      "epoch": 23.492178098676295,
      "grad_norm": Infinity,
      "learning_rate": 0.00017498811045358565,
      "loss": 11.312,
      "step": 195220
    },
    {
      "epoch": 23.49338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00017498559359857313,
      "loss": 11.1799,
      "step": 195230
    },
    {
      "epoch": 23.494584837545126,
      "grad_norm": Infinity,
      "learning_rate": 0.00017498307663503794,
      "loss": 11.1734,
      "step": 195240
    },
    {
      "epoch": 23.495788206979544,
      "grad_norm": Infinity,
      "learning_rate": 0.00017498055956298374,
      "loss": 11.2187,
      "step": 195250
    },
    {
      "epoch": 23.496991576413958,
      "grad_norm": Infinity,
      "learning_rate": 0.0001749780423824141,
      "loss": 11.1922,
      "step": 195260
    },
    {
      "epoch": 23.498194945848375,
      "grad_norm": Infinity,
      "learning_rate": 0.00017497552509333279,
      "loss": 11.1969,
      "step": 195270
    },
    {
      "epoch": 23.499398315282793,
      "grad_norm": Infinity,
      "learning_rate": 0.00017497300769574333,
      "loss": 11.1444,
      "step": 195280
    },
    {
      "epoch": 23.500601684717207,
      "grad_norm": Infinity,
      "learning_rate": 0.0001749704901896495,
      "loss": 11.1288,
      "step": 195290
    },
    {
      "epoch": 23.501805054151625,
      "grad_norm": Infinity,
      "learning_rate": 0.00017496797257505477,
      "loss": 11.2265,
      "step": 195300
    },
    {
      "epoch": 23.503008423586042,
      "grad_norm": Infinity,
      "learning_rate": 0.00017496545485196291,
      "loss": 11.0784,
      "step": 195310
    },
    {
      "epoch": 23.504211793020456,
      "grad_norm": Infinity,
      "learning_rate": 0.00017496293702037755,
      "loss": 11.2303,
      "step": 195320
    },
    {
      "epoch": 23.505415162454874,
      "grad_norm": Infinity,
      "learning_rate": 0.0001749604190803023,
      "loss": 11.0919,
      "step": 195330
    },
    {
      "epoch": 23.50661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00017495790103174078,
      "loss": 11.0806,
      "step": 195340
    },
    {
      "epoch": 23.507821901323705,
      "grad_norm": Infinity,
      "learning_rate": 0.00017495538287469673,
      "loss": 11.1103,
      "step": 195350
    },
    {
      "epoch": 23.509025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.00017495286460917367,
      "loss": 11.0826,
      "step": 195360
    },
    {
      "epoch": 23.51022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.00017495034623517537,
      "loss": 11.068,
      "step": 195370
    },
    {
      "epoch": 23.511432009626954,
      "grad_norm": Infinity,
      "learning_rate": 0.00017494782775270539,
      "loss": 11.1516,
      "step": 195380
    },
    {
      "epoch": 23.512635379061372,
      "grad_norm": Infinity,
      "learning_rate": 0.0001749453091617674,
      "loss": 11.2196,
      "step": 195390
    },
    {
      "epoch": 23.51383874849579,
      "grad_norm": Infinity,
      "learning_rate": 0.00017494279046236503,
      "loss": 11.1573,
      "step": 195400
    },
    {
      "epoch": 23.515042117930204,
      "grad_norm": Infinity,
      "learning_rate": 0.00017494027165450196,
      "loss": 11.1239,
      "step": 195410
    },
    {
      "epoch": 23.51624548736462,
      "grad_norm": Infinity,
      "learning_rate": 0.0001749377527381818,
      "loss": 11.1621,
      "step": 195420
    },
    {
      "epoch": 23.51744885679904,
      "grad_norm": Infinity,
      "learning_rate": 0.0001749352337134082,
      "loss": 11.1501,
      "step": 195430
    },
    {
      "epoch": 23.518652226233453,
      "grad_norm": Infinity,
      "learning_rate": 0.00017493271458018486,
      "loss": 11.0687,
      "step": 195440
    },
    {
      "epoch": 23.51985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017493019533851536,
      "loss": 11.0757,
      "step": 195450
    },
    {
      "epoch": 23.521058965102288,
      "grad_norm": Infinity,
      "learning_rate": 0.00017492767598840337,
      "loss": 11.1104,
      "step": 195460
    },
    {
      "epoch": 23.522262334536702,
      "grad_norm": Infinity,
      "learning_rate": 0.0001749251565298525,
      "loss": 11.0994,
      "step": 195470
    },
    {
      "epoch": 23.52346570397112,
      "grad_norm": Infinity,
      "learning_rate": 0.00017492263696286645,
      "loss": 11.145,
      "step": 195480
    },
    {
      "epoch": 23.524669073405537,
      "grad_norm": Infinity,
      "learning_rate": 0.00017492011728744886,
      "loss": 11.1268,
      "step": 195490
    },
    {
      "epoch": 23.52587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00017491759750360334,
      "loss": 11.0251,
      "step": 195500
    },
    {
      "epoch": 23.52707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00017491507761133358,
      "loss": 11.0577,
      "step": 195510
    },
    {
      "epoch": 23.528279181708786,
      "grad_norm": Infinity,
      "learning_rate": 0.00017491255761064323,
      "loss": 11.1736,
      "step": 195520
    },
    {
      "epoch": 23.5294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00017491003750153588,
      "loss": 11.2155,
      "step": 195530
    },
    {
      "epoch": 23.530685920577618,
      "grad_norm": Infinity,
      "learning_rate": 0.0001749075172840152,
      "loss": 11.2012,
      "step": 195540
    },
    {
      "epoch": 23.531889290012035,
      "grad_norm": Infinity,
      "learning_rate": 0.00017490499695808488,
      "loss": 11.0345,
      "step": 195550
    },
    {
      "epoch": 23.53309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.00017490247652374852,
      "loss": 11.1722,
      "step": 195560
    },
    {
      "epoch": 23.534296028880867,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748999559810098,
      "loss": 11.2242,
      "step": 195570
    },
    {
      "epoch": 23.535499398315284,
      "grad_norm": Infinity,
      "learning_rate": 0.00017489743532987233,
      "loss": 11.0888,
      "step": 195580
    },
    {
      "epoch": 23.5367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017489491457033977,
      "loss": 11.2027,
      "step": 195590
    },
    {
      "epoch": 23.537906137184116,
      "grad_norm": Infinity,
      "learning_rate": 0.00017489239370241578,
      "loss": 11.2428,
      "step": 195600
    },
    {
      "epoch": 23.53910950661853,
      "grad_norm": Infinity,
      "learning_rate": 0.00017488987272610404,
      "loss": 11.0786,
      "step": 195610
    },
    {
      "epoch": 23.540312876052948,
      "grad_norm": Infinity,
      "learning_rate": 0.00017488735164140814,
      "loss": 11.0887,
      "step": 195620
    },
    {
      "epoch": 23.541516245487365,
      "grad_norm": Infinity,
      "learning_rate": 0.00017488483044833178,
      "loss": 11.1581,
      "step": 195630
    },
    {
      "epoch": 23.54271961492178,
      "grad_norm": Infinity,
      "learning_rate": 0.00017488230914687854,
      "loss": 11.0984,
      "step": 195640
    },
    {
      "epoch": 23.543922984356197,
      "grad_norm": Infinity,
      "learning_rate": 0.00017487978773705215,
      "loss": 11.1414,
      "step": 195650
    },
    {
      "epoch": 23.545126353790614,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748772662188562,
      "loss": 11.1765,
      "step": 195660
    },
    {
      "epoch": 23.54632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00017487474459229434,
      "loss": 11.0267,
      "step": 195670
    },
    {
      "epoch": 23.547533092659446,
      "grad_norm": Infinity,
      "learning_rate": 0.00017487222285737027,
      "loss": 11.1396,
      "step": 195680
    },
    {
      "epoch": 23.548736462093864,
      "grad_norm": Infinity,
      "learning_rate": 0.00017486970101408757,
      "loss": 11.2185,
      "step": 195690
    },
    {
      "epoch": 23.549939831528278,
      "grad_norm": Infinity,
      "learning_rate": 0.00017486717906244997,
      "loss": 11.1946,
      "step": 195700
    },
    {
      "epoch": 23.551143200962695,
      "grad_norm": Infinity,
      "learning_rate": 0.00017486465700246106,
      "loss": 11.1448,
      "step": 195710
    },
    {
      "epoch": 23.552346570397113,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748621348341245,
      "loss": 11.1367,
      "step": 195720
    },
    {
      "epoch": 23.553549939831527,
      "grad_norm": Infinity,
      "learning_rate": 0.00017485961255744395,
      "loss": 11.1005,
      "step": 195730
    },
    {
      "epoch": 23.554753309265944,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748570901724231,
      "loss": 11.1457,
      "step": 195740
    },
    {
      "epoch": 23.555956678700362,
      "grad_norm": Infinity,
      "learning_rate": 0.00017485456767906552,
      "loss": 11.1066,
      "step": 195750
    },
    {
      "epoch": 23.557160048134776,
      "grad_norm": Infinity,
      "learning_rate": 0.00017485204507737488,
      "loss": 11.1473,
      "step": 195760
    },
    {
      "epoch": 23.558363417569193,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748495223673549,
      "loss": 11.1539,
      "step": 195770
    },
    {
      "epoch": 23.55956678700361,
      "grad_norm": Infinity,
      "learning_rate": 0.00017484699954900914,
      "loss": 11.1721,
      "step": 195780
    },
    {
      "epoch": 23.560770156438025,
      "grad_norm": Infinity,
      "learning_rate": 0.00017484447662234132,
      "loss": 11.1648,
      "step": 195790
    },
    {
      "epoch": 23.561973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.00017484195358735504,
      "loss": 11.1405,
      "step": 195800
    },
    {
      "epoch": 23.56317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00017483943044405397,
      "loss": 11.2171,
      "step": 195810
    },
    {
      "epoch": 23.564380264741274,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748369071924418,
      "loss": 11.186,
      "step": 195820
    },
    {
      "epoch": 23.56558363417569,
      "grad_norm": Infinity,
      "learning_rate": 0.00017483438383252215,
      "loss": 11.1565,
      "step": 195830
    },
    {
      "epoch": 23.56678700361011,
      "grad_norm": Infinity,
      "learning_rate": 0.00017483186036429865,
      "loss": 11.0553,
      "step": 195840
    },
    {
      "epoch": 23.567990373044523,
      "grad_norm": Infinity,
      "learning_rate": 0.00017482933678777498,
      "loss": 11.3257,
      "step": 195850
    },
    {
      "epoch": 23.56919374247894,
      "grad_norm": Infinity,
      "learning_rate": 0.00017482681310295475,
      "loss": 11.1756,
      "step": 195860
    },
    {
      "epoch": 23.57039711191336,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748242893098417,
      "loss": 11.0925,
      "step": 195870
    },
    {
      "epoch": 23.571600481347772,
      "grad_norm": Infinity,
      "learning_rate": 0.00017482176540843943,
      "loss": 11.1566,
      "step": 195880
    },
    {
      "epoch": 23.57280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.00017481924139875154,
      "loss": 11.2046,
      "step": 195890
    },
    {
      "epoch": 23.574007220216608,
      "grad_norm": Infinity,
      "learning_rate": 0.00017481671728078178,
      "loss": 11.3324,
      "step": 195900
    },
    {
      "epoch": 23.57521058965102,
      "grad_norm": Infinity,
      "learning_rate": 0.00017481419305453373,
      "loss": 11.0436,
      "step": 195910
    },
    {
      "epoch": 23.57641395908544,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748116687200111,
      "loss": 11.1291,
      "step": 195920
    },
    {
      "epoch": 23.577617328519857,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748091442772175,
      "loss": 11.2869,
      "step": 195930
    },
    {
      "epoch": 23.57882069795427,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748066197261566,
      "loss": 11.2446,
      "step": 195940
    },
    {
      "epoch": 23.58002406738869,
      "grad_norm": Infinity,
      "learning_rate": 0.00017480409506683207,
      "loss": 11.1379,
      "step": 195950
    },
    {
      "epoch": 23.581227436823106,
      "grad_norm": Infinity,
      "learning_rate": 0.0001748015702992475,
      "loss": 11.2747,
      "step": 195960
    },
    {
      "epoch": 23.58243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747990454234066,
      "loss": 10.9951,
      "step": 195970
    },
    {
      "epoch": 23.583634175691937,
      "grad_norm": Infinity,
      "learning_rate": 0.00017479652043931308,
      "loss": 11.202,
      "step": 195980
    },
    {
      "epoch": 23.584837545126355,
      "grad_norm": Infinity,
      "learning_rate": 0.00017479399534697047,
      "loss": 11.2329,
      "step": 195990
    },
    {
      "epoch": 23.58604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00017479147014638248,
      "loss": 11.2264,
      "step": 196000
    },
    {
      "epoch": 23.587244283995187,
      "grad_norm": Infinity,
      "learning_rate": 0.00017478894483755278,
      "loss": 11.0912,
      "step": 196010
    },
    {
      "epoch": 23.588447653429604,
      "grad_norm": Infinity,
      "learning_rate": 0.000174786419420485,
      "loss": 11.2065,
      "step": 196020
    },
    {
      "epoch": 23.589651022864018,
      "grad_norm": Infinity,
      "learning_rate": 0.00017478389389518283,
      "loss": 11.1534,
      "step": 196030
    },
    {
      "epoch": 23.590854392298436,
      "grad_norm": Infinity,
      "learning_rate": 0.00017478136826164987,
      "loss": 11.0833,
      "step": 196040
    },
    {
      "epoch": 23.592057761732853,
      "grad_norm": Infinity,
      "learning_rate": 0.00017477884251988985,
      "loss": 11.0591,
      "step": 196050
    },
    {
      "epoch": 23.593261131167267,
      "grad_norm": Infinity,
      "learning_rate": 0.00017477631666990638,
      "loss": 11.1408,
      "step": 196060
    },
    {
      "epoch": 23.594464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00017477379071170307,
      "loss": 11.1734,
      "step": 196070
    },
    {
      "epoch": 23.595667870036102,
      "grad_norm": Infinity,
      "learning_rate": 0.00017477126464528366,
      "loss": 11.0705,
      "step": 196080
    },
    {
      "epoch": 23.596871239470516,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747687384706518,
      "loss": 11.1252,
      "step": 196090
    },
    {
      "epoch": 23.598074608904934,
      "grad_norm": Infinity,
      "learning_rate": 0.00017476621218781105,
      "loss": 11.1243,
      "step": 196100
    },
    {
      "epoch": 23.59927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00017476368579676517,
      "loss": 11.0919,
      "step": 196110
    },
    {
      "epoch": 23.600481347773766,
      "grad_norm": Infinity,
      "learning_rate": 0.00017476115929751775,
      "loss": 11.0937,
      "step": 196120
    },
    {
      "epoch": 23.601684717208183,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747586326900725,
      "loss": 11.1627,
      "step": 196130
    },
    {
      "epoch": 23.6028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.00017475610597443306,
      "loss": 10.9796,
      "step": 196140
    },
    {
      "epoch": 23.604091456077015,
      "grad_norm": Infinity,
      "learning_rate": 0.00017475357915060303,
      "loss": 11.0526,
      "step": 196150
    },
    {
      "epoch": 23.605294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00017475105221858615,
      "loss": 11.1849,
      "step": 196160
    },
    {
      "epoch": 23.60649819494585,
      "grad_norm": Infinity,
      "learning_rate": 0.00017474852517838604,
      "loss": 11.0206,
      "step": 196170
    },
    {
      "epoch": 23.607701564380264,
      "grad_norm": Infinity,
      "learning_rate": 0.00017474599803000634,
      "loss": 11.2783,
      "step": 196180
    },
    {
      "epoch": 23.60890493381468,
      "grad_norm": Infinity,
      "learning_rate": 0.00017474347077345076,
      "loss": 11.1151,
      "step": 196190
    },
    {
      "epoch": 23.6101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747409434087229,
      "loss": 11.1317,
      "step": 196200
    },
    {
      "epoch": 23.611311672683513,
      "grad_norm": Infinity,
      "learning_rate": 0.00017473841593582646,
      "loss": 11.1793,
      "step": 196210
    },
    {
      "epoch": 23.61251504211793,
      "grad_norm": Infinity,
      "learning_rate": 0.00017473588835476507,
      "loss": 11.1582,
      "step": 196220
    },
    {
      "epoch": 23.613718411552348,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747333606655424,
      "loss": 11.192,
      "step": 196230
    },
    {
      "epoch": 23.614921780986762,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747308328681621,
      "loss": 11.1467,
      "step": 196240
    },
    {
      "epoch": 23.61612515042118,
      "grad_norm": Infinity,
      "learning_rate": 0.00017472830496262788,
      "loss": 11.0942,
      "step": 196250
    },
    {
      "epoch": 23.617328519855597,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747257769489433,
      "loss": 11.078,
      "step": 196260
    },
    {
      "epoch": 23.61853188929001,
      "grad_norm": Infinity,
      "learning_rate": 0.00017472324882711207,
      "loss": 11.1349,
      "step": 196270
    },
    {
      "epoch": 23.61973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.00017472072059713788,
      "loss": 11.1673,
      "step": 196280
    },
    {
      "epoch": 23.620938628158846,
      "grad_norm": Infinity,
      "learning_rate": 0.00017471819225902432,
      "loss": 11.0936,
      "step": 196290
    },
    {
      "epoch": 23.62214199759326,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747156638127751,
      "loss": 11.1077,
      "step": 196300
    },
    {
      "epoch": 23.623345367027678,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747131352583939,
      "loss": 11.0793,
      "step": 196310
    },
    {
      "epoch": 23.624548736462096,
      "grad_norm": Infinity,
      "learning_rate": 0.00017471060659588432,
      "loss": 11.1016,
      "step": 196320
    },
    {
      "epoch": 23.62575210589651,
      "grad_norm": Infinity,
      "learning_rate": 0.00017470807782525008,
      "loss": 11.072,
      "step": 196330
    },
    {
      "epoch": 23.626955475330927,
      "grad_norm": Infinity,
      "learning_rate": 0.00017470554894649475,
      "loss": 11.1738,
      "step": 196340
    },
    {
      "epoch": 23.628158844765345,
      "grad_norm": Infinity,
      "learning_rate": 0.00017470301995962207,
      "loss": 11.1468,
      "step": 196350
    },
    {
      "epoch": 23.62936221419976,
      "grad_norm": Infinity,
      "learning_rate": 0.00017470049086463569,
      "loss": 11.106,
      "step": 196360
    },
    {
      "epoch": 23.630565583634176,
      "grad_norm": Infinity,
      "learning_rate": 0.00017469796166153924,
      "loss": 11.0997,
      "step": 196370
    },
    {
      "epoch": 23.63176895306859,
      "grad_norm": Infinity,
      "learning_rate": 0.00017469543235033643,
      "loss": 11.0373,
      "step": 196380
    },
    {
      "epoch": 23.632972322503008,
      "grad_norm": Infinity,
      "learning_rate": 0.00017469290293103083,
      "loss": 11.1763,
      "step": 196390
    },
    {
      "epoch": 23.634175691937426,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746903734036262,
      "loss": 11.1654,
      "step": 196400
    },
    {
      "epoch": 23.63537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.00017468784376812617,
      "loss": 11.1513,
      "step": 196410
    },
    {
      "epoch": 23.636582430806257,
      "grad_norm": Infinity,
      "learning_rate": 0.00017468531402453437,
      "loss": 11.2204,
      "step": 196420
    },
    {
      "epoch": 23.637785800240675,
      "grad_norm": Infinity,
      "learning_rate": 0.00017468278417285448,
      "loss": 11.1224,
      "step": 196430
    },
    {
      "epoch": 23.63898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00017468025421309017,
      "loss": 11.0977,
      "step": 196440
    },
    {
      "epoch": 23.640192539109506,
      "grad_norm": Infinity,
      "learning_rate": 0.00017467772414524508,
      "loss": 11.1476,
      "step": 196450
    },
    {
      "epoch": 23.641395908543924,
      "grad_norm": Infinity,
      "learning_rate": 0.00017467519396932292,
      "loss": 11.1794,
      "step": 196460
    },
    {
      "epoch": 23.642599277978338,
      "grad_norm": Infinity,
      "learning_rate": 0.00017467266368532727,
      "loss": 11.1345,
      "step": 196470
    },
    {
      "epoch": 23.643802647412755,
      "grad_norm": Infinity,
      "learning_rate": 0.00017467013329326188,
      "loss": 11.1471,
      "step": 196480
    },
    {
      "epoch": 23.645006016847173,
      "grad_norm": Infinity,
      "learning_rate": 0.00017466760279313034,
      "loss": 11.1729,
      "step": 196490
    },
    {
      "epoch": 23.646209386281587,
      "grad_norm": Infinity,
      "learning_rate": 0.00017466507218493637,
      "loss": 11.1896,
      "step": 196500
    },
    {
      "epoch": 23.647412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746625414686836,
      "loss": 11.2276,
      "step": 196510
    },
    {
      "epoch": 23.648616125150422,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746600106443757,
      "loss": 11.1161,
      "step": 196520
    },
    {
      "epoch": 23.649819494584836,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746574797120163,
      "loss": 11.0692,
      "step": 196530
    },
    {
      "epoch": 23.651022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.00017465494867160915,
      "loss": 11.0433,
      "step": 196540
    },
    {
      "epoch": 23.65222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746524175231578,
      "loss": 11.1093,
      "step": 196550
    },
    {
      "epoch": 23.653429602888085,
      "grad_norm": Infinity,
      "learning_rate": 0.00017464988626666601,
      "loss": 11.0656,
      "step": 196560
    },
    {
      "epoch": 23.654632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746473549021374,
      "loss": 11.2172,
      "step": 196570
    },
    {
      "epoch": 23.65583634175692,
      "grad_norm": Infinity,
      "learning_rate": 0.00017464482342957561,
      "loss": 11.099,
      "step": 196580
    },
    {
      "epoch": 23.657039711191334,
      "grad_norm": Infinity,
      "learning_rate": 0.00017464229184898438,
      "loss": 11.0274,
      "step": 196590
    },
    {
      "epoch": 23.658243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.00017463976016036728,
      "loss": 11.1531,
      "step": 196600
    },
    {
      "epoch": 23.65944645006017,
      "grad_norm": Infinity,
      "learning_rate": 0.00017463722836372805,
      "loss": 11.0268,
      "step": 196610
    },
    {
      "epoch": 23.660649819494584,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746346964590703,
      "loss": 11.1401,
      "step": 196620
    },
    {
      "epoch": 23.661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.00017463216444639775,
      "loss": 11.1238,
      "step": 196630
    },
    {
      "epoch": 23.66305655836342,
      "grad_norm": Infinity,
      "learning_rate": 0.000174629632325714,
      "loss": 11.0019,
      "step": 196640
    },
    {
      "epoch": 23.664259927797833,
      "grad_norm": Infinity,
      "learning_rate": 0.00017462710009702275,
      "loss": 11.1642,
      "step": 196650
    },
    {
      "epoch": 23.66546329723225,
      "grad_norm": Infinity,
      "learning_rate": 0.00017462456776032767,
      "loss": 11.1533,
      "step": 196660
    },
    {
      "epoch": 23.666666666666668,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746220353156324,
      "loss": 11.0911,
      "step": 196670
    },
    {
      "epoch": 23.667870036101082,
      "grad_norm": Infinity,
      "learning_rate": 0.00017461950276294064,
      "loss": 11.2071,
      "step": 196680
    },
    {
      "epoch": 23.6690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.000174616970102256,
      "loss": 11.1489,
      "step": 196690
    },
    {
      "epoch": 23.670276774969917,
      "grad_norm": Infinity,
      "learning_rate": 0.00017461443733358224,
      "loss": 11.1009,
      "step": 196700
    },
    {
      "epoch": 23.67148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00017461190445692296,
      "loss": 11.268,
      "step": 196710
    },
    {
      "epoch": 23.67268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.0001746093714722818,
      "loss": 11.14,
      "step": 196720
    },
    {
      "epoch": 23.673886883273166,
      "grad_norm": Infinity,
      "learning_rate": 0.00017460683837966246,
      "loss": 11.1555,
      "step": 196730
    },
    {
      "epoch": 23.67509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.00017460430517906862,
      "loss": 11.0493,
      "step": 196740
    },
    {
      "epoch": 23.676293622141998,
      "grad_norm": Infinity,
      "learning_rate": 0.00017460177187050393,
      "loss": 11.0805,
      "step": 196750
    },
    {
      "epoch": 23.677496991576415,
      "grad_norm": Infinity,
      "learning_rate": 0.00017459923845397206,
      "loss": 11.2198,
      "step": 196760
    },
    {
      "epoch": 23.67870036101083,
      "grad_norm": Infinity,
      "learning_rate": 0.00017459670492947667,
      "loss": 11.041,
      "step": 196770
    },
    {
      "epoch": 23.679903730445247,
      "grad_norm": Infinity,
      "learning_rate": 0.00017459417129702144,
      "loss": 11.1872,
      "step": 196780
    },
    {
      "epoch": 23.681107099879664,
      "grad_norm": Infinity,
      "learning_rate": 0.00017459163755661,
      "loss": 11.0921,
      "step": 196790
    },
    {
      "epoch": 23.68231046931408,
      "grad_norm": Infinity,
      "learning_rate": 0.00017458910370824607,
      "loss": 11.1232,
      "step": 196800
    },
    {
      "epoch": 23.683513838748496,
      "grad_norm": Infinity,
      "learning_rate": 0.00017458656975193328,
      "loss": 11.1452,
      "step": 196810
    },
    {
      "epoch": 23.684717208182914,
      "grad_norm": Infinity,
      "learning_rate": 0.0001745840356876753,
      "loss": 11.143,
      "step": 196820
    },
    {
      "epoch": 23.685920577617328,
      "grad_norm": Infinity,
      "learning_rate": 0.00017458150151547584,
      "loss": 11.1584,
      "step": 196830
    },
    {
      "epoch": 23.687123947051745,
      "grad_norm": Infinity,
      "learning_rate": 0.0001745789672353385,
      "loss": 11.0752,
      "step": 196840
    },
    {
      "epoch": 23.688327316486163,
      "grad_norm": Infinity,
      "learning_rate": 0.000174576432847267,
      "loss": 11.0889,
      "step": 196850
    },
    {
      "epoch": 23.689530685920577,
      "grad_norm": Infinity,
      "learning_rate": 0.00017457389835126498,
      "loss": 11.1347,
      "step": 196860
    },
    {
      "epoch": 23.690734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.00017457136374733612,
      "loss": 11.1619,
      "step": 196870
    },
    {
      "epoch": 23.691937424789412,
      "grad_norm": Infinity,
      "learning_rate": 0.00017456882903548412,
      "loss": 11.1127,
      "step": 196880
    },
    {
      "epoch": 23.693140794223826,
      "grad_norm": Infinity,
      "learning_rate": 0.00017456629421571255,
      "loss": 11.0257,
      "step": 196890
    },
    {
      "epoch": 23.694344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00017456375928802517,
      "loss": 11.1361,
      "step": 196900
    },
    {
      "epoch": 23.69554753309266,
      "grad_norm": Infinity,
      "learning_rate": 0.00017456122425242563,
      "loss": 11.1188,
      "step": 196910
    },
    {
      "epoch": 23.696750902527075,
      "grad_norm": Infinity,
      "learning_rate": 0.00017455868910891758,
      "loss": 11.0865,
      "step": 196920
    },
    {
      "epoch": 23.697954271961493,
      "grad_norm": Infinity,
      "learning_rate": 0.00017455615385750473,
      "loss": 11.0205,
      "step": 196930
    },
    {
      "epoch": 23.69915764139591,
      "grad_norm": Infinity,
      "learning_rate": 0.00017455361849819069,
      "loss": 11.1067,
      "step": 196940
    },
    {
      "epoch": 23.700361010830324,
      "grad_norm": Infinity,
      "learning_rate": 0.00017455108303097912,
      "loss": 11.1015,
      "step": 196950
    },
    {
      "epoch": 23.70156438026474,
      "grad_norm": Infinity,
      "learning_rate": 0.00017454854745587378,
      "loss": 11.0817,
      "step": 196960
    },
    {
      "epoch": 23.70276774969916,
      "grad_norm": Infinity,
      "learning_rate": 0.00017454601177287828,
      "loss": 11.1627,
      "step": 196970
    },
    {
      "epoch": 23.703971119133573,
      "grad_norm": Infinity,
      "learning_rate": 0.0001745434759819963,
      "loss": 11.1831,
      "step": 196980
    },
    {
      "epoch": 23.70517448856799,
      "grad_norm": Infinity,
      "learning_rate": 0.00017454094008323147,
      "loss": 11.1608,
      "step": 196990
    },
    {
      "epoch": 23.706377858002405,
      "grad_norm": Infinity,
      "learning_rate": 0.00017453840407658753,
      "loss": 11.0917,
      "step": 197000
    },
    {
      "epoch": 23.707581227436823,
      "grad_norm": Infinity,
      "learning_rate": 0.0001745358679620681,
      "loss": 11.1608,
      "step": 197010
    },
    {
      "epoch": 23.70878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00017453333173967687,
      "loss": 10.9933,
      "step": 197020
    },
    {
      "epoch": 23.709987966305654,
      "grad_norm": Infinity,
      "learning_rate": 0.0001745307954094175,
      "loss": 11.1136,
      "step": 197030
    },
    {
      "epoch": 23.71119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.00017452825897129368,
      "loss": 11.0798,
      "step": 197040
    },
    {
      "epoch": 23.71239470517449,
      "grad_norm": Infinity,
      "learning_rate": 0.00017452572242530908,
      "loss": 11.1815,
      "step": 197050
    },
    {
      "epoch": 23.713598074608903,
      "grad_norm": Infinity,
      "learning_rate": 0.00017452318577146735,
      "loss": 11.0755,
      "step": 197060
    },
    {
      "epoch": 23.71480144404332,
      "grad_norm": Infinity,
      "learning_rate": 0.00017452064900977217,
      "loss": 11.1482,
      "step": 197070
    },
    {
      "epoch": 23.71600481347774,
      "grad_norm": Infinity,
      "learning_rate": 0.0001745181121402272,
      "loss": 11.1554,
      "step": 197080
    },
    {
      "epoch": 23.717208182912152,
      "grad_norm": Infinity,
      "learning_rate": 0.00017451557516283615,
      "loss": 11.2002,
      "step": 197090
    },
    {
      "epoch": 23.71841155234657,
      "grad_norm": Infinity,
      "learning_rate": 0.00017451303807760265,
      "loss": 11.1313,
      "step": 197100
    },
    {
      "epoch": 23.719614921780988,
      "grad_norm": Infinity,
      "learning_rate": 0.0001745105008845304,
      "loss": 11.1451,
      "step": 197110
    },
    {
      "epoch": 23.7208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00017450796358362304,
      "loss": 11.1946,
      "step": 197120
    },
    {
      "epoch": 23.72202166064982,
      "grad_norm": Infinity,
      "learning_rate": 0.00017450542617488424,
      "loss": 11.2111,
      "step": 197130
    },
    {
      "epoch": 23.723225030084237,
      "grad_norm": Infinity,
      "learning_rate": 0.0001745028886583177,
      "loss": 11.0723,
      "step": 197140
    },
    {
      "epoch": 23.72442839951865,
      "grad_norm": Infinity,
      "learning_rate": 0.00017450035103392715,
      "loss": 11.3497,
      "step": 197150
    },
    {
      "epoch": 23.72563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00017449781330171614,
      "loss": 11.1267,
      "step": 197160
    },
    {
      "epoch": 23.726835138387486,
      "grad_norm": Infinity,
      "learning_rate": 0.00017449527546168841,
      "loss": 11.1037,
      "step": 197170
    },
    {
      "epoch": 23.7280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.00017449273751384765,
      "loss": 11.1939,
      "step": 197180
    },
    {
      "epoch": 23.729241877256317,
      "grad_norm": Infinity,
      "learning_rate": 0.00017449019945819745,
      "loss": 11.2122,
      "step": 197190
    },
    {
      "epoch": 23.730445246690735,
      "grad_norm": Infinity,
      "learning_rate": 0.00017448766129474162,
      "loss": 11.1186,
      "step": 197200
    },
    {
      "epoch": 23.73164861612515,
      "grad_norm": Infinity,
      "learning_rate": 0.00017448512302348367,
      "loss": 11.0483,
      "step": 197210
    },
    {
      "epoch": 23.732851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.0001744825846444274,
      "loss": 11.142,
      "step": 197220
    },
    {
      "epoch": 23.734055354993984,
      "grad_norm": Infinity,
      "learning_rate": 0.00017448004615757646,
      "loss": 11.1176,
      "step": 197230
    },
    {
      "epoch": 23.735258724428398,
      "grad_norm": Infinity,
      "learning_rate": 0.00017447750756293448,
      "loss": 11.1994,
      "step": 197240
    },
    {
      "epoch": 23.736462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.00017447496886050515,
      "loss": 11.1696,
      "step": 197250
    },
    {
      "epoch": 23.737665463297233,
      "grad_norm": Infinity,
      "learning_rate": 0.0001744724300502922,
      "loss": 11.1203,
      "step": 197260
    },
    {
      "epoch": 23.738868832731647,
      "grad_norm": Infinity,
      "learning_rate": 0.00017446989113229918,
      "loss": 11.0952,
      "step": 197270
    },
    {
      "epoch": 23.740072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00017446735210652992,
      "loss": 11.1079,
      "step": 197280
    },
    {
      "epoch": 23.741275571600482,
      "grad_norm": Infinity,
      "learning_rate": 0.00017446481297298796,
      "loss": 11.035,
      "step": 197290
    },
    {
      "epoch": 23.742478941034896,
      "grad_norm": Infinity,
      "learning_rate": 0.00017446227373167704,
      "loss": 11.2231,
      "step": 197300
    },
    {
      "epoch": 23.743682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00017445973438260086,
      "loss": 11.1426,
      "step": 197310
    },
    {
      "epoch": 23.74488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.00017445719492576304,
      "loss": 11.1374,
      "step": 197320
    },
    {
      "epoch": 23.746089049338146,
      "grad_norm": Infinity,
      "learning_rate": 0.00017445465536116726,
      "loss": 11.1727,
      "step": 197330
    },
    {
      "epoch": 23.747292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00017445211568881724,
      "loss": 11.125,
      "step": 197340
    },
    {
      "epoch": 23.74849578820698,
      "grad_norm": Infinity,
      "learning_rate": 0.00017444957590871663,
      "loss": 11.1961,
      "step": 197350
    },
    {
      "epoch": 23.749699157641395,
      "grad_norm": Infinity,
      "learning_rate": 0.00017444703602086906,
      "loss": 11.1705,
      "step": 197360
    },
    {
      "epoch": 23.750902527075812,
      "grad_norm": Infinity,
      "learning_rate": 0.00017444449602527828,
      "loss": 11.0648,
      "step": 197370
    },
    {
      "epoch": 23.75210589651023,
      "grad_norm": Infinity,
      "learning_rate": 0.00017444195592194791,
      "loss": 11.1219,
      "step": 197380
    },
    {
      "epoch": 23.753309265944644,
      "grad_norm": Infinity,
      "learning_rate": 0.00017443941571088168,
      "loss": 11.072,
      "step": 197390
    },
    {
      "epoch": 23.75451263537906,
      "grad_norm": Infinity,
      "learning_rate": 0.0001744368753920832,
      "loss": 11.2714,
      "step": 197400
    },
    {
      "epoch": 23.75571600481348,
      "grad_norm": Infinity,
      "learning_rate": 0.00017443433496555625,
      "loss": 11.0549,
      "step": 197410
    },
    {
      "epoch": 23.756919374247893,
      "grad_norm": Infinity,
      "learning_rate": 0.0001744317944313044,
      "loss": 11.0843,
      "step": 197420
    },
    {
      "epoch": 23.75812274368231,
      "grad_norm": Infinity,
      "learning_rate": 0.00017442925378933135,
      "loss": 11.1402,
      "step": 197430
    },
    {
      "epoch": 23.759326113116728,
      "grad_norm": Infinity,
      "learning_rate": 0.00017442671303964083,
      "loss": 11.1855,
      "step": 197440
    },
    {
      "epoch": 23.760529482551142,
      "grad_norm": Infinity,
      "learning_rate": 0.00017442417218223644,
      "loss": 10.9913,
      "step": 197450
    },
    {
      "epoch": 23.76173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.0001744216312171219,
      "loss": 11.1436,
      "step": 197460
    },
    {
      "epoch": 23.762936221419977,
      "grad_norm": Infinity,
      "learning_rate": 0.00017441909014430094,
      "loss": 11.1198,
      "step": 197470
    },
    {
      "epoch": 23.76413959085439,
      "grad_norm": Infinity,
      "learning_rate": 0.00017441654896377712,
      "loss": 11.1991,
      "step": 197480
    },
    {
      "epoch": 23.76534296028881,
      "grad_norm": Infinity,
      "learning_rate": 0.00017441400767555422,
      "loss": 11.088,
      "step": 197490
    },
    {
      "epoch": 23.766546329723226,
      "grad_norm": Infinity,
      "learning_rate": 0.00017441146627963588,
      "loss": 11.1659,
      "step": 197500
    },
    {
      "epoch": 23.76774969915764,
      "grad_norm": Infinity,
      "learning_rate": 0.00017440892477602575,
      "loss": 11.1147,
      "step": 197510
    },
    {
      "epoch": 23.768953068592058,
      "grad_norm": Infinity,
      "learning_rate": 0.00017440638316472754,
      "loss": 11.0198,
      "step": 197520
    },
    {
      "epoch": 23.770156438026476,
      "grad_norm": Infinity,
      "learning_rate": 0.00017440384144574495,
      "loss": 11.2263,
      "step": 197530
    },
    {
      "epoch": 23.77135980746089,
      "grad_norm": Infinity,
      "learning_rate": 0.0001744012996190816,
      "loss": 11.0934,
      "step": 197540
    },
    {
      "epoch": 23.772563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00017439875768474124,
      "loss": 11.1262,
      "step": 197550
    },
    {
      "epoch": 23.773766546329725,
      "grad_norm": Infinity,
      "learning_rate": 0.00017439621564272745,
      "loss": 11.0539,
      "step": 197560
    },
    {
      "epoch": 23.77496991576414,
      "grad_norm": Infinity,
      "learning_rate": 0.00017439367349304403,
      "loss": 11.1165,
      "step": 197570
    },
    {
      "epoch": 23.776173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00017439113123569454,
      "loss": 11.2575,
      "step": 197580
    },
    {
      "epoch": 23.777376654632974,
      "grad_norm": Infinity,
      "learning_rate": 0.00017438858887068275,
      "loss": 11.1105,
      "step": 197590
    },
    {
      "epoch": 23.778580024067388,
      "grad_norm": Infinity,
      "learning_rate": 0.0001743860463980123,
      "loss": 11.1292,
      "step": 197600
    },
    {
      "epoch": 23.779783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.00017438350381768688,
      "loss": 11.1761,
      "step": 197610
    },
    {
      "epoch": 23.780986762936223,
      "grad_norm": Infinity,
      "learning_rate": 0.00017438096112971016,
      "loss": 11.1877,
      "step": 197620
    },
    {
      "epoch": 23.782190132370637,
      "grad_norm": Infinity,
      "learning_rate": 0.00017437841833408584,
      "loss": 11.1328,
      "step": 197630
    },
    {
      "epoch": 23.783393501805055,
      "grad_norm": Infinity,
      "learning_rate": 0.00017437587543081757,
      "loss": 10.9783,
      "step": 197640
    },
    {
      "epoch": 23.784596871239472,
      "grad_norm": Infinity,
      "learning_rate": 0.00017437333241990906,
      "loss": 11.0719,
      "step": 197650
    },
    {
      "epoch": 23.785800240673886,
      "grad_norm": Infinity,
      "learning_rate": 0.00017437078930136396,
      "loss": 11.1263,
      "step": 197660
    },
    {
      "epoch": 23.787003610108304,
      "grad_norm": Infinity,
      "learning_rate": 0.00017436824607518597,
      "loss": 11.1569,
      "step": 197670
    },
    {
      "epoch": 23.78820697954272,
      "grad_norm": Infinity,
      "learning_rate": 0.00017436570274137876,
      "loss": 11.203,
      "step": 197680
    },
    {
      "epoch": 23.789410348977135,
      "grad_norm": Infinity,
      "learning_rate": 0.000174363159299946,
      "loss": 11.1132,
      "step": 197690
    },
    {
      "epoch": 23.790613718411553,
      "grad_norm": Infinity,
      "learning_rate": 0.00017436061575089141,
      "loss": 11.2019,
      "step": 197700
    },
    {
      "epoch": 23.79181708784597,
      "grad_norm": Infinity,
      "learning_rate": 0.00017435807209421867,
      "loss": 11.126,
      "step": 197710
    },
    {
      "epoch": 23.793020457280385,
      "grad_norm": Infinity,
      "learning_rate": 0.00017435552832993143,
      "loss": 11.1907,
      "step": 197720
    },
    {
      "epoch": 23.794223826714802,
      "grad_norm": Infinity,
      "learning_rate": 0.0001743529844580334,
      "loss": 11.1937,
      "step": 197730
    },
    {
      "epoch": 23.79542719614922,
      "grad_norm": Infinity,
      "learning_rate": 0.0001743504404785282,
      "loss": 11.1883,
      "step": 197740
    },
    {
      "epoch": 23.796630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00017434789639141958,
      "loss": 11.2438,
      "step": 197750
    },
    {
      "epoch": 23.79783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.0001743453521967112,
      "loss": 11.3198,
      "step": 197760
    },
    {
      "epoch": 23.799037304452465,
      "grad_norm": Infinity,
      "learning_rate": 0.00017434280789440672,
      "loss": 11.1848,
      "step": 197770
    },
    {
      "epoch": 23.800240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.00017434026348450986,
      "loss": 11.1922,
      "step": 197780
    },
    {
      "epoch": 23.8014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017433771896702428,
      "loss": 11.047,
      "step": 197790
    },
    {
      "epoch": 23.802647412755714,
      "grad_norm": Infinity,
      "learning_rate": 0.00017433517434195368,
      "loss": 11.1562,
      "step": 197800
    },
    {
      "epoch": 23.803850782190132,
      "grad_norm": Infinity,
      "learning_rate": 0.0001743326296093017,
      "loss": 11.1043,
      "step": 197810
    },
    {
      "epoch": 23.80505415162455,
      "grad_norm": Infinity,
      "learning_rate": 0.0001743300847690721,
      "loss": 11.2207,
      "step": 197820
    },
    {
      "epoch": 23.806257521058964,
      "grad_norm": Infinity,
      "learning_rate": 0.00017432753982126847,
      "loss": 11.1137,
      "step": 197830
    },
    {
      "epoch": 23.80746089049338,
      "grad_norm": Infinity,
      "learning_rate": 0.00017432499476589456,
      "loss": 11.1706,
      "step": 197840
    },
    {
      "epoch": 23.8086642599278,
      "grad_norm": Infinity,
      "learning_rate": 0.00017432244960295402,
      "loss": 11.1371,
      "step": 197850
    },
    {
      "epoch": 23.809867629362213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017431990433245057,
      "loss": 11.3131,
      "step": 197860
    },
    {
      "epoch": 23.81107099879663,
      "grad_norm": Infinity,
      "learning_rate": 0.00017431735895438787,
      "loss": 11.2776,
      "step": 197870
    },
    {
      "epoch": 23.812274368231048,
      "grad_norm": Infinity,
      "learning_rate": 0.0001743148134687696,
      "loss": 11.13,
      "step": 197880
    },
    {
      "epoch": 23.813477737665462,
      "grad_norm": Infinity,
      "learning_rate": 0.00017431226787559942,
      "loss": 11.2197,
      "step": 197890
    },
    {
      "epoch": 23.81468110709988,
      "grad_norm": Infinity,
      "learning_rate": 0.00017430972217488103,
      "loss": 11.1627,
      "step": 197900
    },
    {
      "epoch": 23.815884476534297,
      "grad_norm": Infinity,
      "learning_rate": 0.00017430717636661815,
      "loss": 11.0682,
      "step": 197910
    },
    {
      "epoch": 23.81708784596871,
      "grad_norm": Infinity,
      "learning_rate": 0.00017430463045081446,
      "loss": 11.2374,
      "step": 197920
    },
    {
      "epoch": 23.81829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.0001743020844274736,
      "loss": 11.1987,
      "step": 197930
    },
    {
      "epoch": 23.819494584837546,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742995382965993,
      "loss": 11.0672,
      "step": 197940
    },
    {
      "epoch": 23.82069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742969920581952,
      "loss": 11.2038,
      "step": 197950
    },
    {
      "epoch": 23.821901323706378,
      "grad_norm": Infinity,
      "learning_rate": 0.000174294445712265,
      "loss": 11.1895,
      "step": 197960
    },
    {
      "epoch": 23.823104693140795,
      "grad_norm": Infinity,
      "learning_rate": 0.00017429189925881244,
      "loss": 11.2128,
      "step": 197970
    },
    {
      "epoch": 23.82430806257521,
      "grad_norm": Infinity,
      "learning_rate": 0.00017428935269784112,
      "loss": 11.092,
      "step": 197980
    },
    {
      "epoch": 23.825511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00017428680602935478,
      "loss": 11.1366,
      "step": 197990
    },
    {
      "epoch": 23.826714801444044,
      "grad_norm": Infinity,
      "learning_rate": 0.00017428425925335712,
      "loss": 11.2333,
      "step": 198000
    },
    {
      "epoch": 23.82791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00017428171236985173,
      "loss": 11.0292,
      "step": 198010
    },
    {
      "epoch": 23.829121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742791653788424,
      "loss": 11.1469,
      "step": 198020
    },
    {
      "epoch": 23.830324909747294,
      "grad_norm": Infinity,
      "learning_rate": 0.00017427661828033278,
      "loss": 11.1232,
      "step": 198030
    },
    {
      "epoch": 23.831528279181708,
      "grad_norm": Infinity,
      "learning_rate": 0.00017427407107432655,
      "loss": 11.1536,
      "step": 198040
    },
    {
      "epoch": 23.832731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.00017427152376082742,
      "loss": 11.16,
      "step": 198050
    },
    {
      "epoch": 23.833935018050543,
      "grad_norm": Infinity,
      "learning_rate": 0.00017426897633983902,
      "loss": 11.1687,
      "step": 198060
    },
    {
      "epoch": 23.835138387484957,
      "grad_norm": Infinity,
      "learning_rate": 0.00017426642881136512,
      "loss": 11.1294,
      "step": 198070
    },
    {
      "epoch": 23.836341756919374,
      "grad_norm": Infinity,
      "learning_rate": 0.00017426388117540935,
      "loss": 11.0826,
      "step": 198080
    },
    {
      "epoch": 23.837545126353792,
      "grad_norm": Infinity,
      "learning_rate": 0.00017426133343197538,
      "loss": 11.0565,
      "step": 198090
    },
    {
      "epoch": 23.838748495788206,
      "grad_norm": Infinity,
      "learning_rate": 0.00017425878558106694,
      "loss": 11.1567,
      "step": 198100
    },
    {
      "epoch": 23.839951865222623,
      "grad_norm": Infinity,
      "learning_rate": 0.00017425623762268771,
      "loss": 11.0631,
      "step": 198110
    },
    {
      "epoch": 23.84115523465704,
      "grad_norm": Infinity,
      "learning_rate": 0.00017425368955684137,
      "loss": 11.221,
      "step": 198120
    },
    {
      "epoch": 23.842358604091455,
      "grad_norm": Infinity,
      "learning_rate": 0.00017425114138353161,
      "loss": 11.1611,
      "step": 198130
    },
    {
      "epoch": 23.843561973525873,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742485931027621,
      "loss": 11.079,
      "step": 198140
    },
    {
      "epoch": 23.84476534296029,
      "grad_norm": Infinity,
      "learning_rate": 0.00017424604471453657,
      "loss": 11.0097,
      "step": 198150
    },
    {
      "epoch": 23.845968712394704,
      "grad_norm": Infinity,
      "learning_rate": 0.00017424349621885868,
      "loss": 11.1486,
      "step": 198160
    },
    {
      "epoch": 23.84717208182912,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742409476157321,
      "loss": 11.0677,
      "step": 198170
    },
    {
      "epoch": 23.84837545126354,
      "grad_norm": Infinity,
      "learning_rate": 0.00017423839890516054,
      "loss": 11.1365,
      "step": 198180
    },
    {
      "epoch": 23.849578820697953,
      "grad_norm": Infinity,
      "learning_rate": 0.00017423585008714772,
      "loss": 11.1626,
      "step": 198190
    },
    {
      "epoch": 23.85078219013237,
      "grad_norm": Infinity,
      "learning_rate": 0.00017423330116169726,
      "loss": 11.1317,
      "step": 198200
    },
    {
      "epoch": 23.85198555956679,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742307521288129,
      "loss": 11.1111,
      "step": 198210
    },
    {
      "epoch": 23.853188929001202,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742282029884983,
      "loss": 11.2019,
      "step": 198220
    },
    {
      "epoch": 23.85439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742256537407572,
      "loss": 11.1958,
      "step": 198230
    },
    {
      "epoch": 23.855595667870038,
      "grad_norm": Infinity,
      "learning_rate": 0.00017422310438559323,
      "loss": 11.1331,
      "step": 198240
    },
    {
      "epoch": 23.85679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742205549230101,
      "loss": 11.1346,
      "step": 198250
    },
    {
      "epoch": 23.85800240673887,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742180053530115,
      "loss": 11.0958,
      "step": 198260
    },
    {
      "epoch": 23.859205776173287,
      "grad_norm": Infinity,
      "learning_rate": 0.00017421545567560115,
      "loss": 11.1356,
      "step": 198270
    },
    {
      "epoch": 23.8604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742129058907827,
      "loss": 11.134,
      "step": 198280
    },
    {
      "epoch": 23.86161251504212,
      "grad_norm": Infinity,
      "learning_rate": 0.00017421035599855981,
      "loss": 11.1219,
      "step": 198290
    },
    {
      "epoch": 23.862815884476536,
      "grad_norm": Infinity,
      "learning_rate": 0.00017420780599893627,
      "loss": 11.2187,
      "step": 198300
    },
    {
      "epoch": 23.86401925391095,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742052558919157,
      "loss": 11.1758,
      "step": 198310
    },
    {
      "epoch": 23.865222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.0001742027056775018,
      "loss": 11.1535,
      "step": 198320
    },
    {
      "epoch": 23.866425992779785,
      "grad_norm": Infinity,
      "learning_rate": 0.00017420015535569826,
      "loss": 10.9779,
      "step": 198330
    },
    {
      "epoch": 23.8676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00017419760492650877,
      "loss": 11.1329,
      "step": 198340
    },
    {
      "epoch": 23.868832731648617,
      "grad_norm": Infinity,
      "learning_rate": 0.000174195054389937,
      "loss": 11.2001,
      "step": 198350
    },
    {
      "epoch": 23.870036101083034,
      "grad_norm": Infinity,
      "learning_rate": 0.00017419250374598673,
      "loss": 11.1675,
      "step": 198360
    },
    {
      "epoch": 23.871239470517448,
      "grad_norm": Infinity,
      "learning_rate": 0.00017418995299466156,
      "loss": 11.1709,
      "step": 198370
    },
    {
      "epoch": 23.872442839951866,
      "grad_norm": Infinity,
      "learning_rate": 0.0001741874021359652,
      "loss": 11.0995,
      "step": 198380
    },
    {
      "epoch": 23.87364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.00017418485116990135,
      "loss": 11.0929,
      "step": 198390
    },
    {
      "epoch": 23.874849578820697,
      "grad_norm": Infinity,
      "learning_rate": 0.00017418230009647373,
      "loss": 11.0156,
      "step": 198400
    },
    {
      "epoch": 23.876052948255115,
      "grad_norm": Infinity,
      "learning_rate": 0.000174179748915686,
      "loss": 11.1221,
      "step": 198410
    },
    {
      "epoch": 23.87725631768953,
      "grad_norm": Infinity,
      "learning_rate": 0.00017417719762754185,
      "loss": 11.1425,
      "step": 198420
    },
    {
      "epoch": 23.878459687123947,
      "grad_norm": Infinity,
      "learning_rate": 0.000174174646232045,
      "loss": 11.2075,
      "step": 198430
    },
    {
      "epoch": 23.879663056558364,
      "grad_norm": Infinity,
      "learning_rate": 0.00017417209472919912,
      "loss": 11.0732,
      "step": 198440
    },
    {
      "epoch": 23.880866425992778,
      "grad_norm": Infinity,
      "learning_rate": 0.0001741695431190079,
      "loss": 11.0609,
      "step": 198450
    },
    {
      "epoch": 23.882069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.000174166991401475,
      "loss": 10.9782,
      "step": 198460
    },
    {
      "epoch": 23.883273164861613,
      "grad_norm": Infinity,
      "learning_rate": 0.0001741644395766042,
      "loss": 11.2292,
      "step": 198470
    },
    {
      "epoch": 23.884476534296027,
      "grad_norm": Infinity,
      "learning_rate": 0.00017416188764439913,
      "loss": 11.1384,
      "step": 198480
    },
    {
      "epoch": 23.885679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00017415933560486352,
      "loss": 11.1514,
      "step": 198490
    },
    {
      "epoch": 23.886883273164862,
      "grad_norm": Infinity,
      "learning_rate": 0.00017415678345800102,
      "loss": 11.223,
      "step": 198500
    },
    {
      "epoch": 23.888086642599276,
      "grad_norm": Infinity,
      "learning_rate": 0.00017415423120381536,
      "loss": 10.9908,
      "step": 198510
    },
    {
      "epoch": 23.889290012033694,
      "grad_norm": Infinity,
      "learning_rate": 0.00017415167884231018,
      "loss": 11.1762,
      "step": 198520
    },
    {
      "epoch": 23.89049338146811,
      "grad_norm": Infinity,
      "learning_rate": 0.00017414912637348924,
      "loss": 11.1195,
      "step": 198530
    },
    {
      "epoch": 23.891696750902526,
      "grad_norm": Infinity,
      "learning_rate": 0.00017414657379735622,
      "loss": 11.0932,
      "step": 198540
    },
    {
      "epoch": 23.892900120336943,
      "grad_norm": Infinity,
      "learning_rate": 0.00017414402111391478,
      "loss": 11.0989,
      "step": 198550
    },
    {
      "epoch": 23.89410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017414146832316866,
      "loss": 11.1422,
      "step": 198560
    },
    {
      "epoch": 23.895306859205775,
      "grad_norm": Infinity,
      "learning_rate": 0.0001741389154251215,
      "loss": 11.0829,
      "step": 198570
    },
    {
      "epoch": 23.896510228640192,
      "grad_norm": Infinity,
      "learning_rate": 0.00017413636241977708,
      "loss": 11.1815,
      "step": 198580
    },
    {
      "epoch": 23.89771359807461,
      "grad_norm": Infinity,
      "learning_rate": 0.00017413380930713897,
      "loss": 11.0756,
      "step": 198590
    },
    {
      "epoch": 23.898916967509024,
      "grad_norm": Infinity,
      "learning_rate": 0.00017413125608721096,
      "loss": 11.0634,
      "step": 198600
    },
    {
      "epoch": 23.90012033694344,
      "grad_norm": Infinity,
      "learning_rate": 0.00017412870275999675,
      "loss": 11.2037,
      "step": 198610
    },
    {
      "epoch": 23.90132370637786,
      "grad_norm": Infinity,
      "learning_rate": 0.00017412614932549997,
      "loss": 11.1078,
      "step": 198620
    },
    {
      "epoch": 23.902527075812273,
      "grad_norm": Infinity,
      "learning_rate": 0.00017412359578372437,
      "loss": 11.1544,
      "step": 198630
    },
    {
      "epoch": 23.90373044524669,
      "grad_norm": Infinity,
      "learning_rate": 0.0001741210421346736,
      "loss": 11.2907,
      "step": 198640
    },
    {
      "epoch": 23.904933814681108,
      "grad_norm": Infinity,
      "learning_rate": 0.00017411848837835142,
      "loss": 11.1491,
      "step": 198650
    },
    {
      "epoch": 23.906137184115522,
      "grad_norm": Infinity,
      "learning_rate": 0.00017411593451476147,
      "loss": 11.1781,
      "step": 198660
    },
    {
      "epoch": 23.90734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.00017411338054390747,
      "loss": 11.1698,
      "step": 198670
    },
    {
      "epoch": 23.908543922984357,
      "grad_norm": Infinity,
      "learning_rate": 0.0001741108264657931,
      "loss": 11.1405,
      "step": 198680
    },
    {
      "epoch": 23.90974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00017410827228042205,
      "loss": 11.1147,
      "step": 198690
    },
    {
      "epoch": 23.91095066185319,
      "grad_norm": Infinity,
      "learning_rate": 0.00017410571798779807,
      "loss": 11.0971,
      "step": 198700
    },
    {
      "epoch": 23.912154031287606,
      "grad_norm": Infinity,
      "learning_rate": 0.0001741031635879248,
      "loss": 11.0397,
      "step": 198710
    },
    {
      "epoch": 23.91335740072202,
      "grad_norm": Infinity,
      "learning_rate": 0.00017410060908080598,
      "loss": 11.1268,
      "step": 198720
    },
    {
      "epoch": 23.914560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00017409805446644527,
      "loss": 11.1352,
      "step": 198730
    },
    {
      "epoch": 23.915764139590856,
      "grad_norm": Infinity,
      "learning_rate": 0.00017409549974484637,
      "loss": 11.1539,
      "step": 198740
    },
    {
      "epoch": 23.91696750902527,
      "grad_norm": Infinity,
      "learning_rate": 0.000174092944916013,
      "loss": 11.1898,
      "step": 198750
    },
    {
      "epoch": 23.918170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00017409038997994881,
      "loss": 11.1078,
      "step": 198760
    },
    {
      "epoch": 23.919374247894105,
      "grad_norm": Infinity,
      "learning_rate": 0.00017408783493665756,
      "loss": 11.1321,
      "step": 198770
    },
    {
      "epoch": 23.92057761732852,
      "grad_norm": Infinity,
      "learning_rate": 0.00017408527978614294,
      "loss": 11.1603,
      "step": 198780
    },
    {
      "epoch": 23.921780986762936,
      "grad_norm": Infinity,
      "learning_rate": 0.00017408272452840861,
      "loss": 11.1587,
      "step": 198790
    },
    {
      "epoch": 23.922984356197354,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740801691634583,
      "loss": 11.2889,
      "step": 198800
    },
    {
      "epoch": 23.924187725631768,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740776136912957,
      "loss": 11.1631,
      "step": 198810
    },
    {
      "epoch": 23.925391095066185,
      "grad_norm": Infinity,
      "learning_rate": 0.00017407505811192447,
      "loss": 11.1012,
      "step": 198820
    },
    {
      "epoch": 23.926594464500603,
      "grad_norm": Infinity,
      "learning_rate": 0.00017407250242534836,
      "loss": 11.1792,
      "step": 198830
    },
    {
      "epoch": 23.927797833935017,
      "grad_norm": Infinity,
      "learning_rate": 0.00017406994663157107,
      "loss": 11.0976,
      "step": 198840
    },
    {
      "epoch": 23.929001203369435,
      "grad_norm": Infinity,
      "learning_rate": 0.00017406739073059626,
      "loss": 11.1242,
      "step": 198850
    },
    {
      "epoch": 23.930204572803852,
      "grad_norm": Infinity,
      "learning_rate": 0.00017406483472242763,
      "loss": 11.0359,
      "step": 198860
    },
    {
      "epoch": 23.931407942238266,
      "grad_norm": Infinity,
      "learning_rate": 0.00017406227860706892,
      "loss": 11.0949,
      "step": 198870
    },
    {
      "epoch": 23.932611311672684,
      "grad_norm": Infinity,
      "learning_rate": 0.00017405972238452382,
      "loss": 11.179,
      "step": 198880
    },
    {
      "epoch": 23.9338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.000174057166054796,
      "loss": 11.2209,
      "step": 198890
    },
    {
      "epoch": 23.935018050541515,
      "grad_norm": Infinity,
      "learning_rate": 0.00017405460961788917,
      "loss": 11.3049,
      "step": 198900
    },
    {
      "epoch": 23.936221419975933,
      "grad_norm": Infinity,
      "learning_rate": 0.00017405205307380705,
      "loss": 11.2463,
      "step": 198910
    },
    {
      "epoch": 23.93742478941035,
      "grad_norm": Infinity,
      "learning_rate": 0.00017404949642255333,
      "loss": 11.0641,
      "step": 198920
    },
    {
      "epoch": 23.938628158844764,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740469396641317,
      "loss": 11.1371,
      "step": 198930
    },
    {
      "epoch": 23.939831528279182,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740443827985459,
      "loss": 11.1618,
      "step": 198940
    },
    {
      "epoch": 23.9410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017404182582579953,
      "loss": 11.248,
      "step": 198950
    },
    {
      "epoch": 23.942238267148014,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740392687458964,
      "loss": 11.1252,
      "step": 198960
    },
    {
      "epoch": 23.94344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00017403671155884017,
      "loss": 11.0593,
      "step": 198970
    },
    {
      "epoch": 23.94464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740341542646345,
      "loss": 11.062,
      "step": 198980
    },
    {
      "epoch": 23.945848375451263,
      "grad_norm": Infinity,
      "learning_rate": 0.00017403159686328316,
      "loss": 11.0267,
      "step": 198990
    },
    {
      "epoch": 23.94705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017402903935478982,
      "loss": 10.9414,
      "step": 199000
    },
    {
      "epoch": 23.948255114320098,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740264817391582,
      "loss": 11.092,
      "step": 199010
    },
    {
      "epoch": 23.949458483754512,
      "grad_norm": Infinity,
      "learning_rate": 0.00017402392401639198,
      "loss": 11.0984,
      "step": 199020
    },
    {
      "epoch": 23.95066185318893,
      "grad_norm": Infinity,
      "learning_rate": 0.00017402136618649484,
      "loss": 11.2201,
      "step": 199030
    },
    {
      "epoch": 23.951865222623347,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740188082494705,
      "loss": 11.2339,
      "step": 199040
    },
    {
      "epoch": 23.95306859205776,
      "grad_norm": Infinity,
      "learning_rate": 0.00017401625020532272,
      "loss": 11.2408,
      "step": 199050
    },
    {
      "epoch": 23.95427196149218,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740136920540551,
      "loss": 11.1132,
      "step": 199060
    },
    {
      "epoch": 23.955475330926596,
      "grad_norm": Infinity,
      "learning_rate": 0.00017401113379567144,
      "loss": 11.1311,
      "step": 199070
    },
    {
      "epoch": 23.95667870036101,
      "grad_norm": Infinity,
      "learning_rate": 0.00017400857543017537,
      "loss": 11.2444,
      "step": 199080
    },
    {
      "epoch": 23.957882069795428,
      "grad_norm": Infinity,
      "learning_rate": 0.00017400601695757064,
      "loss": 11.1296,
      "step": 199090
    },
    {
      "epoch": 23.959085439229845,
      "grad_norm": Infinity,
      "learning_rate": 0.00017400345837786092,
      "loss": 11.2244,
      "step": 199100
    },
    {
      "epoch": 23.96028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.0001740008996910499,
      "loss": 11.1838,
      "step": 199110
    },
    {
      "epoch": 23.961492178098677,
      "grad_norm": Infinity,
      "learning_rate": 0.00017399834089714132,
      "loss": 11.1187,
      "step": 199120
    },
    {
      "epoch": 23.96269554753309,
      "grad_norm": Infinity,
      "learning_rate": 0.0001739957819961389,
      "loss": 11.1451,
      "step": 199130
    },
    {
      "epoch": 23.96389891696751,
      "grad_norm": Infinity,
      "learning_rate": 0.0001739932229880463,
      "loss": 11.1041,
      "step": 199140
    },
    {
      "epoch": 23.965102286401926,
      "grad_norm": Infinity,
      "learning_rate": 0.00017399066387286722,
      "loss": 11.1991,
      "step": 199150
    },
    {
      "epoch": 23.96630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00017398810465060543,
      "loss": 11.2304,
      "step": 199160
    },
    {
      "epoch": 23.967509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.00017398554532126453,
      "loss": 11.1529,
      "step": 199170
    },
    {
      "epoch": 23.968712394705175,
      "grad_norm": Infinity,
      "learning_rate": 0.0001739829858848483,
      "loss": 11.2541,
      "step": 199180
    },
    {
      "epoch": 23.96991576413959,
      "grad_norm": Infinity,
      "learning_rate": 0.00017398042634136042,
      "loss": 11.1097,
      "step": 199190
    },
    {
      "epoch": 23.971119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.0001739778666908046,
      "loss": 11.2848,
      "step": 199200
    },
    {
      "epoch": 23.972322503008424,
      "grad_norm": Infinity,
      "learning_rate": 0.00017397530693318453,
      "loss": 11.1338,
      "step": 199210
    },
    {
      "epoch": 23.97352587244284,
      "grad_norm": Infinity,
      "learning_rate": 0.00017397274706850395,
      "loss": 11.2082,
      "step": 199220
    },
    {
      "epoch": 23.974729241877256,
      "grad_norm": Infinity,
      "learning_rate": 0.00017397018709676654,
      "loss": 11.1871,
      "step": 199230
    },
    {
      "epoch": 23.975932611311674,
      "grad_norm": Infinity,
      "learning_rate": 0.00017396762701797598,
      "loss": 11.1451,
      "step": 199240
    },
    {
      "epoch": 23.977135980746088,
      "grad_norm": Infinity,
      "learning_rate": 0.00017396506683213604,
      "loss": 11.168,
      "step": 199250
    },
    {
      "epoch": 23.978339350180505,
      "grad_norm": Infinity,
      "learning_rate": 0.00017396250653925038,
      "loss": 11.2314,
      "step": 199260
    },
    {
      "epoch": 23.979542719614923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017395994613932268,
      "loss": 11.0986,
      "step": 199270
    },
    {
      "epoch": 23.980746089049337,
      "grad_norm": Infinity,
      "learning_rate": 0.0001739573856323567,
      "loss": 11.2303,
      "step": 199280
    },
    {
      "epoch": 23.981949458483754,
      "grad_norm": Infinity,
      "learning_rate": 0.0001739548250183561,
      "loss": 11.0996,
      "step": 199290
    },
    {
      "epoch": 23.983152827918172,
      "grad_norm": Infinity,
      "learning_rate": 0.00017395226429732462,
      "loss": 11.165,
      "step": 199300
    },
    {
      "epoch": 23.984356197352586,
      "grad_norm": Infinity,
      "learning_rate": 0.000173949703469266,
      "loss": 11.1665,
      "step": 199310
    },
    {
      "epoch": 23.985559566787003,
      "grad_norm": Infinity,
      "learning_rate": 0.00017394714253418387,
      "loss": 11.0457,
      "step": 199320
    },
    {
      "epoch": 23.98676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00017394458149208196,
      "loss": 11.1679,
      "step": 199330
    },
    {
      "epoch": 23.987966305655835,
      "grad_norm": Infinity,
      "learning_rate": 0.00017394202034296398,
      "loss": 11.1186,
      "step": 199340
    },
    {
      "epoch": 23.989169675090253,
      "grad_norm": Infinity,
      "learning_rate": 0.00017393945908683364,
      "loss": 11.1849,
      "step": 199350
    },
    {
      "epoch": 23.99037304452467,
      "grad_norm": Infinity,
      "learning_rate": 0.00017393689772369467,
      "loss": 11.1141,
      "step": 199360
    },
    {
      "epoch": 23.991576413959084,
      "grad_norm": Infinity,
      "learning_rate": 0.00017393433625355074,
      "loss": 11.1251,
      "step": 199370
    },
    {
      "epoch": 23.9927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00017393177467640557,
      "loss": 11.302,
      "step": 199380
    },
    {
      "epoch": 23.99398315282792,
      "grad_norm": Infinity,
      "learning_rate": 0.00017392921299226287,
      "loss": 11.1394,
      "step": 199390
    },
    {
      "epoch": 23.995186522262333,
      "grad_norm": Infinity,
      "learning_rate": 0.00017392665120112634,
      "loss": 11.102,
      "step": 199400
    },
    {
      "epoch": 23.99638989169675,
      "grad_norm": Infinity,
      "learning_rate": 0.00017392408930299973,
      "loss": 11.0624,
      "step": 199410
    },
    {
      "epoch": 23.99759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.0001739215272978867,
      "loss": 11.2084,
      "step": 199420
    },
    {
      "epoch": 23.998796630565582,
      "grad_norm": Infinity,
      "learning_rate": 0.00017391896518579092,
      "loss": 11.1525,
      "step": 199430
    },
    {
      "epoch": 24.0,
      "grad_norm": Infinity,
      "learning_rate": 0.00017391640296671618,
      "loss": 10.9994,
      "step": 199440
    },
    {
      "epoch": 24.0,
      "eval_loss": 11.139605522155762,
      "eval_runtime": 120.0591,
      "eval_samples_per_second": 61.528,
      "eval_steps_per_second": 7.696,
      "step": 199440
    },
    {
      "epoch": 24.001203369434418,
      "grad_norm": Infinity,
      "learning_rate": 0.00017391384064066617,
      "loss": 11.0209,
      "step": 199450
    },
    {
      "epoch": 24.00240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.00017391127820764457,
      "loss": 11.2911,
      "step": 199460
    },
    {
      "epoch": 24.00361010830325,
      "grad_norm": Infinity,
      "learning_rate": 0.0001739087156676551,
      "loss": 11.2017,
      "step": 199470
    },
    {
      "epoch": 24.004813477737667,
      "grad_norm": Infinity,
      "learning_rate": 0.00017390615302070148,
      "loss": 11.1332,
      "step": 199480
    },
    {
      "epoch": 24.00601684717208,
      "grad_norm": Infinity,
      "learning_rate": 0.0001739035902667874,
      "loss": 11.189,
      "step": 199490
    },
    {
      "epoch": 24.0072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00017390102740591658,
      "loss": 11.2744,
      "step": 199500
    },
    {
      "epoch": 24.008423586040916,
      "grad_norm": Infinity,
      "learning_rate": 0.00017389846443809271,
      "loss": 11.1867,
      "step": 199510
    },
    {
      "epoch": 24.00962695547533,
      "grad_norm": Infinity,
      "learning_rate": 0.00017389590136331954,
      "loss": 11.0673,
      "step": 199520
    },
    {
      "epoch": 24.010830324909747,
      "grad_norm": Infinity,
      "learning_rate": 0.00017389333818160078,
      "loss": 11.0787,
      "step": 199530
    },
    {
      "epoch": 24.012033694344165,
      "grad_norm": Infinity,
      "learning_rate": 0.00017389077489294007,
      "loss": 11.2569,
      "step": 199540
    },
    {
      "epoch": 24.01323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738882114973412,
      "loss": 11.1808,
      "step": 199550
    },
    {
      "epoch": 24.014440433212997,
      "grad_norm": Infinity,
      "learning_rate": 0.00017388564799480783,
      "loss": 11.1974,
      "step": 199560
    },
    {
      "epoch": 24.015643802647414,
      "grad_norm": Infinity,
      "learning_rate": 0.00017388308438534369,
      "loss": 11.228,
      "step": 199570
    },
    {
      "epoch": 24.016847172081828,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738805206689525,
      "loss": 11.0196,
      "step": 199580
    },
    {
      "epoch": 24.018050541516246,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738779568456379,
      "loss": 11.1351,
      "step": 199590
    },
    {
      "epoch": 24.019253910950663,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738753929154037,
      "loss": 11.1036,
      "step": 199600
    },
    {
      "epoch": 24.020457280385077,
      "grad_norm": Infinity,
      "learning_rate": 0.00017387282887825357,
      "loss": 11.0743,
      "step": 199610
    },
    {
      "epoch": 24.021660649819495,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738702647341912,
      "loss": 11.2657,
      "step": 199620
    },
    {
      "epoch": 24.022864019253912,
      "grad_norm": Infinity,
      "learning_rate": 0.00017386770048322033,
      "loss": 11.1197,
      "step": 199630
    },
    {
      "epoch": 24.024067388688326,
      "grad_norm": Infinity,
      "learning_rate": 0.00017386513612534468,
      "loss": 11.1049,
      "step": 199640
    },
    {
      "epoch": 24.025270758122744,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738625716605679,
      "loss": 11.0684,
      "step": 199650
    },
    {
      "epoch": 24.02647412755716,
      "grad_norm": Infinity,
      "learning_rate": 0.00017386000708889375,
      "loss": 11.21,
      "step": 199660
    },
    {
      "epoch": 24.027677496991576,
      "grad_norm": Infinity,
      "learning_rate": 0.00017385744241032596,
      "loss": 11.1931,
      "step": 199670
    },
    {
      "epoch": 24.028880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738548776248682,
      "loss": 11.1807,
      "step": 199680
    },
    {
      "epoch": 24.03008423586041,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738523127325242,
      "loss": 11.1758,
      "step": 199690
    },
    {
      "epoch": 24.031287605294825,
      "grad_norm": Infinity,
      "learning_rate": 0.00017384974773329768,
      "loss": 11.0785,
      "step": 199700
    },
    {
      "epoch": 24.032490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00017384718262719232,
      "loss": 11.1566,
      "step": 199710
    },
    {
      "epoch": 24.03369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.00017384461741421186,
      "loss": 11.1568,
      "step": 199720
    },
    {
      "epoch": 24.034897713598074,
      "grad_norm": Infinity,
      "learning_rate": 0.00017384205209436003,
      "loss": 11.142,
      "step": 199730
    },
    {
      "epoch": 24.03610108303249,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738394866676405,
      "loss": 11.1465,
      "step": 199740
    },
    {
      "epoch": 24.03730445246691,
      "grad_norm": Infinity,
      "learning_rate": 0.00017383692113405702,
      "loss": 11.1593,
      "step": 199750
    },
    {
      "epoch": 24.038507821901323,
      "grad_norm": Infinity,
      "learning_rate": 0.00017383435549361327,
      "loss": 11.1618,
      "step": 199760
    },
    {
      "epoch": 24.03971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017383178974631297,
      "loss": 11.1748,
      "step": 199770
    },
    {
      "epoch": 24.040914560770158,
      "grad_norm": Infinity,
      "learning_rate": 0.00017382922389215985,
      "loss": 11.2459,
      "step": 199780
    },
    {
      "epoch": 24.042117930204572,
      "grad_norm": Infinity,
      "learning_rate": 0.00017382665793115763,
      "loss": 11.1858,
      "step": 199790
    },
    {
      "epoch": 24.04332129963899,
      "grad_norm": Infinity,
      "learning_rate": 0.00017382409186331,
      "loss": 11.2074,
      "step": 199800
    },
    {
      "epoch": 24.044524669073404,
      "grad_norm": Infinity,
      "learning_rate": 0.00017382152568862067,
      "loss": 11.2532,
      "step": 199810
    },
    {
      "epoch": 24.04572803850782,
      "grad_norm": Infinity,
      "learning_rate": 0.00017381895940709336,
      "loss": 11.2359,
      "step": 199820
    },
    {
      "epoch": 24.04693140794224,
      "grad_norm": Infinity,
      "learning_rate": 0.00017381639301873182,
      "loss": 11.1503,
      "step": 199830
    },
    {
      "epoch": 24.048134777376653,
      "grad_norm": Infinity,
      "learning_rate": 0.00017381382652353973,
      "loss": 11.1877,
      "step": 199840
    },
    {
      "epoch": 24.04933814681107,
      "grad_norm": Infinity,
      "learning_rate": 0.00017381125992152078,
      "loss": 11.1675,
      "step": 199850
    },
    {
      "epoch": 24.050541516245488,
      "grad_norm": Infinity,
      "learning_rate": 0.00017380869321267873,
      "loss": 11.109,
      "step": 199860
    },
    {
      "epoch": 24.051744885679902,
      "grad_norm": Infinity,
      "learning_rate": 0.0001738061263970173,
      "loss": 11.2008,
      "step": 199870
    },
    {
      "epoch": 24.05294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00017380355947454014,
      "loss": 11.0457,
      "step": 199880
    },
    {
      "epoch": 24.054151624548737,
      "grad_norm": Infinity,
      "learning_rate": 0.00017380099244525106,
      "loss": 11.141,
      "step": 199890
    },
    {
      "epoch": 24.05535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00017379842530915366,
      "loss": 11.1414,
      "step": 199900
    },
    {
      "epoch": 24.05655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00017379585806625176,
      "loss": 11.1288,
      "step": 199910
    },
    {
      "epoch": 24.057761732851986,
      "grad_norm": Infinity,
      "learning_rate": 0.00017379329071654903,
      "loss": 11.1121,
      "step": 199920
    },
    {
      "epoch": 24.0589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.00017379072326004918,
      "loss": 11.1482,
      "step": 199930
    },
    {
      "epoch": 24.060168471720818,
      "grad_norm": Infinity,
      "learning_rate": 0.00017378815569675594,
      "loss": 11.2987,
      "step": 199940
    },
    {
      "epoch": 24.061371841155236,
      "grad_norm": Infinity,
      "learning_rate": 0.00017378558802667304,
      "loss": 11.0505,
      "step": 199950
    },
    {
      "epoch": 24.06257521058965,
      "grad_norm": Infinity,
      "learning_rate": 0.00017378302024980414,
      "loss": 11.1084,
      "step": 199960
    },
    {
      "epoch": 24.063778580024067,
      "grad_norm": Infinity,
      "learning_rate": 0.00017378045236615303,
      "loss": 11.1455,
      "step": 199970
    },
    {
      "epoch": 24.064981949458485,
      "grad_norm": Infinity,
      "learning_rate": 0.00017377788437572338,
      "loss": 11.2559,
      "step": 199980
    },
    {
      "epoch": 24.0661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.0001737753162785189,
      "loss": 11.2525,
      "step": 199990
    },
    {
      "epoch": 24.067388688327316,
      "grad_norm": Infinity,
      "learning_rate": 0.00017377274807454333,
      "loss": 11.0958,
      "step": 200000
    },
    {
      "epoch": 24.068592057761734,
      "grad_norm": Infinity,
      "learning_rate": 0.00017377017976380037,
      "loss": 11.1051,
      "step": 200010
    },
    {
      "epoch": 24.069795427196148,
      "grad_norm": Infinity,
      "learning_rate": 0.00017376761134629378,
      "loss": 11.0837,
      "step": 200020
    },
    {
      "epoch": 24.070998796630565,
      "grad_norm": Infinity,
      "learning_rate": 0.00017376504282202722,
      "loss": 11.0301,
      "step": 200030
    },
    {
      "epoch": 24.072202166064983,
      "grad_norm": Infinity,
      "learning_rate": 0.0001737624741910044,
      "loss": 11.0521,
      "step": 200040
    },
    {
      "epoch": 24.073405535499397,
      "grad_norm": Infinity,
      "learning_rate": 0.0001737599054532291,
      "loss": 11.0391,
      "step": 200050
    },
    {
      "epoch": 24.074608904933815,
      "grad_norm": Infinity,
      "learning_rate": 0.00017375733660870504,
      "loss": 11.2403,
      "step": 200060
    },
    {
      "epoch": 24.075812274368232,
      "grad_norm": Infinity,
      "learning_rate": 0.00017375476765743586,
      "loss": 11.0647,
      "step": 200070
    },
    {
      "epoch": 24.077015643802646,
      "grad_norm": Infinity,
      "learning_rate": 0.00017375219859942534,
      "loss": 11.1189,
      "step": 200080
    },
    {
      "epoch": 24.078219013237064,
      "grad_norm": Infinity,
      "learning_rate": 0.00017374962943467716,
      "loss": 11.1609,
      "step": 200090
    },
    {
      "epoch": 24.07942238267148,
      "grad_norm": Infinity,
      "learning_rate": 0.00017374706016319507,
      "loss": 11.208,
      "step": 200100
    },
    {
      "epoch": 24.080625752105895,
      "grad_norm": Infinity,
      "learning_rate": 0.00017374449078498279,
      "loss": 11.1805,
      "step": 200110
    },
    {
      "epoch": 24.081829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.000173741921300044,
      "loss": 11.1044,
      "step": 200120
    },
    {
      "epoch": 24.08303249097473,
      "grad_norm": Infinity,
      "learning_rate": 0.00017373935170838248,
      "loss": 11.0474,
      "step": 200130
    },
    {
      "epoch": 24.084235860409144,
      "grad_norm": Infinity,
      "learning_rate": 0.00017373678201000186,
      "loss": 11.0857,
      "step": 200140
    },
    {
      "epoch": 24.085439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.00017373421220490592,
      "loss": 11.1374,
      "step": 200150
    },
    {
      "epoch": 24.08664259927798,
      "grad_norm": Infinity,
      "learning_rate": 0.0001737316422930984,
      "loss": 11.1366,
      "step": 200160
    },
    {
      "epoch": 24.087845968712394,
      "grad_norm": Infinity,
      "learning_rate": 0.00017372907227458298,
      "loss": 11.0926,
      "step": 200170
    },
    {
      "epoch": 24.08904933814681,
      "grad_norm": Infinity,
      "learning_rate": 0.0001737265021493634,
      "loss": 11.2044,
      "step": 200180
    },
    {
      "epoch": 24.09025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.00017372393191744334,
      "loss": 11.1494,
      "step": 200190
    },
    {
      "epoch": 24.091456077015643,
      "grad_norm": Infinity,
      "learning_rate": 0.00017372136157882658,
      "loss": 11.082,
      "step": 200200
    },
    {
      "epoch": 24.09265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.00017371879113351678,
      "loss": 11.1522,
      "step": 200210
    },
    {
      "epoch": 24.093862815884478,
      "grad_norm": Infinity,
      "learning_rate": 0.0001737162205815177,
      "loss": 11.0132,
      "step": 200220
    },
    {
      "epoch": 24.095066185318892,
      "grad_norm": Infinity,
      "learning_rate": 0.0001737136499228331,
      "loss": 11.0516,
      "step": 200230
    },
    {
      "epoch": 24.09626955475331,
      "grad_norm": Infinity,
      "learning_rate": 0.00017371107915746658,
      "loss": 11.0689,
      "step": 200240
    },
    {
      "epoch": 24.097472924187727,
      "grad_norm": Infinity,
      "learning_rate": 0.00017370850828542196,
      "loss": 11.2285,
      "step": 200250
    },
    {
      "epoch": 24.09867629362214,
      "grad_norm": Infinity,
      "learning_rate": 0.00017370593730670292,
      "loss": 11.1978,
      "step": 200260
    },
    {
      "epoch": 24.09987966305656,
      "grad_norm": Infinity,
      "learning_rate": 0.0001737033662213132,
      "loss": 11.0019,
      "step": 200270
    },
    {
      "epoch": 24.101083032490976,
      "grad_norm": Infinity,
      "learning_rate": 0.00017370079502925648,
      "loss": 11.1283,
      "step": 200280
    },
    {
      "epoch": 24.10228640192539,
      "grad_norm": Infinity,
      "learning_rate": 0.00017369822373053655,
      "loss": 11.1877,
      "step": 200290
    },
    {
      "epoch": 24.103489771359808,
      "grad_norm": Infinity,
      "learning_rate": 0.00017369565232515708,
      "loss": 11.0181,
      "step": 200300
    },
    {
      "epoch": 24.104693140794225,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736930808131218,
      "loss": 11.2241,
      "step": 200310
    },
    {
      "epoch": 24.10589651022864,
      "grad_norm": Infinity,
      "learning_rate": 0.00017369050919443445,
      "loss": 11.1916,
      "step": 200320
    },
    {
      "epoch": 24.107099879663057,
      "grad_norm": Infinity,
      "learning_rate": 0.00017368793746909872,
      "loss": 11.096,
      "step": 200330
    },
    {
      "epoch": 24.108303249097474,
      "grad_norm": Infinity,
      "learning_rate": 0.00017368536563711838,
      "loss": 11.2538,
      "step": 200340
    },
    {
      "epoch": 24.10950661853189,
      "grad_norm": Infinity,
      "learning_rate": 0.00017368279369849713,
      "loss": 10.9995,
      "step": 200350
    },
    {
      "epoch": 24.110709987966306,
      "grad_norm": Infinity,
      "learning_rate": 0.00017368022165323868,
      "loss": 11.0518,
      "step": 200360
    },
    {
      "epoch": 24.111913357400724,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736776495013467,
      "loss": 11.1917,
      "step": 200370
    },
    {
      "epoch": 24.113116726835138,
      "grad_norm": Infinity,
      "learning_rate": 0.00017367507724282504,
      "loss": 11.1399,
      "step": 200380
    },
    {
      "epoch": 24.114320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.00017367250487767733,
      "loss": 11.2126,
      "step": 200390
    },
    {
      "epoch": 24.115523465703973,
      "grad_norm": Infinity,
      "learning_rate": 0.00017366993240590733,
      "loss": 11.1419,
      "step": 200400
    },
    {
      "epoch": 24.116726835138387,
      "grad_norm": Infinity,
      "learning_rate": 0.00017366735982751874,
      "loss": 11.0349,
      "step": 200410
    },
    {
      "epoch": 24.117930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.00017366478714251525,
      "loss": 11.1437,
      "step": 200420
    },
    {
      "epoch": 24.119133574007222,
      "grad_norm": Infinity,
      "learning_rate": 0.00017366221435090068,
      "loss": 11.1644,
      "step": 200430
    },
    {
      "epoch": 24.120336943441636,
      "grad_norm": Infinity,
      "learning_rate": 0.00017365964145267868,
      "loss": 11.1133,
      "step": 200440
    },
    {
      "epoch": 24.121540312876053,
      "grad_norm": Infinity,
      "learning_rate": 0.00017365706844785298,
      "loss": 11.1151,
      "step": 200450
    },
    {
      "epoch": 24.12274368231047,
      "grad_norm": Infinity,
      "learning_rate": 0.00017365449533642737,
      "loss": 11.2123,
      "step": 200460
    },
    {
      "epoch": 24.123947051744885,
      "grad_norm": Infinity,
      "learning_rate": 0.00017365192211840546,
      "loss": 11.2145,
      "step": 200470
    },
    {
      "epoch": 24.125150421179303,
      "grad_norm": Infinity,
      "learning_rate": 0.00017364934879379105,
      "loss": 11.0472,
      "step": 200480
    },
    {
      "epoch": 24.126353790613717,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736467753625879,
      "loss": 11.209,
      "step": 200490
    },
    {
      "epoch": 24.127557160048134,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736442018247996,
      "loss": 11.1361,
      "step": 200500
    },
    {
      "epoch": 24.128760529482552,
      "grad_norm": Infinity,
      "learning_rate": 0.00017364162818043002,
      "loss": 11.0859,
      "step": 200510
    },
    {
      "epoch": 24.129963898916966,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736390544294828,
      "loss": 11.1502,
      "step": 200520
    },
    {
      "epoch": 24.131167268351383,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736364805719617,
      "loss": 11.2071,
      "step": 200530
    },
    {
      "epoch": 24.1323706377858,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736339066078704,
      "loss": 11.225,
      "step": 200540
    },
    {
      "epoch": 24.133574007220215,
      "grad_norm": Infinity,
      "learning_rate": 0.00017363133253721266,
      "loss": 11.143,
      "step": 200550
    },
    {
      "epoch": 24.134777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.00017362875835999223,
      "loss": 11.1317,
      "step": 200560
    },
    {
      "epoch": 24.13598074608905,
      "grad_norm": Infinity,
      "learning_rate": 0.00017362618407621282,
      "loss": 11.1862,
      "step": 200570
    },
    {
      "epoch": 24.137184115523464,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736236096858781,
      "loss": 11.1634,
      "step": 200580
    },
    {
      "epoch": 24.13838748495788,
      "grad_norm": Infinity,
      "learning_rate": 0.00017362103518899185,
      "loss": 11.0827,
      "step": 200590
    },
    {
      "epoch": 24.1395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017361846058555782,
      "loss": 11.1737,
      "step": 200600
    },
    {
      "epoch": 24.140794223826713,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736158858755797,
      "loss": 11.0735,
      "step": 200610
    },
    {
      "epoch": 24.14199759326113,
      "grad_norm": Infinity,
      "learning_rate": 0.0001736133110590612,
      "loss": 11.1858,
      "step": 200620
    },
    {
      "epoch": 24.14320096269555,
      "grad_norm": Infinity,
      "learning_rate": 0.00017361073613600603,
      "loss": 11.052,
      "step": 200630
    },
    {
      "epoch": 24.144404332129962,
      "grad_norm": Infinity,
      "learning_rate": 0.00017360816110641798,
      "loss": 11.1186,
      "step": 200640
    },
    {
      "epoch": 24.14560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00017360558597030075,
      "loss": 11.1394,
      "step": 200650
    },
    {
      "epoch": 24.146811070998798,
      "grad_norm": Infinity,
      "learning_rate": 0.00017360301072765809,
      "loss": 11.0977,
      "step": 200660
    },
    {
      "epoch": 24.14801444043321,
      "grad_norm": Infinity,
      "learning_rate": 0.00017360043537849367,
      "loss": 11.1179,
      "step": 200670
    },
    {
      "epoch": 24.14921780986763,
      "grad_norm": Infinity,
      "learning_rate": 0.00017359785992281126,
      "loss": 11.0894,
      "step": 200680
    },
    {
      "epoch": 24.150421179302047,
      "grad_norm": Infinity,
      "learning_rate": 0.00017359528436061456,
      "loss": 11.1359,
      "step": 200690
    },
    {
      "epoch": 24.15162454873646,
      "grad_norm": Infinity,
      "learning_rate": 0.00017359270869190732,
      "loss": 11.077,
      "step": 200700
    },
    {
      "epoch": 24.15282791817088,
      "grad_norm": Infinity,
      "learning_rate": 0.00017359013291669327,
      "loss": 11.1938,
      "step": 200710
    },
    {
      "epoch": 24.154031287605296,
      "grad_norm": Infinity,
      "learning_rate": 0.00017358755703497615,
      "loss": 11.1356,
      "step": 200720
    },
    {
      "epoch": 24.15523465703971,
      "grad_norm": Infinity,
      "learning_rate": 0.00017358498104675963,
      "loss": 11.0784,
      "step": 200730
    },
    {
      "epoch": 24.156438026474127,
      "grad_norm": Infinity,
      "learning_rate": 0.00017358240495204746,
      "loss": 11.0487,
      "step": 200740
    },
    {
      "epoch": 24.157641395908545,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735798287508434,
      "loss": 11.1477,
      "step": 200750
    },
    {
      "epoch": 24.15884476534296,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735772524431512,
      "loss": 11.2281,
      "step": 200760
    },
    {
      "epoch": 24.160048134777377,
      "grad_norm": Infinity,
      "learning_rate": 0.00017357467602897451,
      "loss": 11.082,
      "step": 200770
    },
    {
      "epoch": 24.161251504211794,
      "grad_norm": Infinity,
      "learning_rate": 0.00017357209950831712,
      "loss": 11.103,
      "step": 200780
    },
    {
      "epoch": 24.162454873646208,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735695228811827,
      "loss": 11.1295,
      "step": 200790
    },
    {
      "epoch": 24.163658243080626,
      "grad_norm": Infinity,
      "learning_rate": 0.00017356694614757508,
      "loss": 11.0448,
      "step": 200800
    },
    {
      "epoch": 24.164861612515043,
      "grad_norm": Infinity,
      "learning_rate": 0.00017356436930749788,
      "loss": 11.2002,
      "step": 200810
    },
    {
      "epoch": 24.166064981949457,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735617923609549,
      "loss": 11.1802,
      "step": 200820
    },
    {
      "epoch": 24.167268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735592153079498,
      "loss": 11.1619,
      "step": 200830
    },
    {
      "epoch": 24.168471720818292,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735566381484864,
      "loss": 11.2427,
      "step": 200840
    },
    {
      "epoch": 24.169675090252706,
      "grad_norm": Infinity,
      "learning_rate": 0.00017355406088256835,
      "loss": 11.0397,
      "step": 200850
    },
    {
      "epoch": 24.170878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00017355148351019944,
      "loss": 11.0204,
      "step": 200860
    },
    {
      "epoch": 24.17208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.00017354890603138337,
      "loss": 11.1485,
      "step": 200870
    },
    {
      "epoch": 24.173285198555956,
      "grad_norm": Infinity,
      "learning_rate": 0.00017354632844612388,
      "loss": 11.1191,
      "step": 200880
    },
    {
      "epoch": 24.174488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.00017354375075442468,
      "loss": 11.2085,
      "step": 200890
    },
    {
      "epoch": 24.17569193742479,
      "grad_norm": Infinity,
      "learning_rate": 0.00017354117295628952,
      "loss": 11.3263,
      "step": 200900
    },
    {
      "epoch": 24.176895306859205,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735385950517221,
      "loss": 11.2333,
      "step": 200910
    },
    {
      "epoch": 24.178098676293622,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735360170407262,
      "loss": 11.0831,
      "step": 200920
    },
    {
      "epoch": 24.17930204572804,
      "grad_norm": Infinity,
      "learning_rate": 0.00017353343892330555,
      "loss": 11.1334,
      "step": 200930
    },
    {
      "epoch": 24.180505415162454,
      "grad_norm": Infinity,
      "learning_rate": 0.00017353086069946385,
      "loss": 11.2668,
      "step": 200940
    },
    {
      "epoch": 24.18170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735282823692048,
      "loss": 11.1141,
      "step": 200950
    },
    {
      "epoch": 24.18291215403129,
      "grad_norm": Infinity,
      "learning_rate": 0.00017352570393253222,
      "loss": 11.0742,
      "step": 200960
    },
    {
      "epoch": 24.184115523465703,
      "grad_norm": Infinity,
      "learning_rate": 0.00017352312538944977,
      "loss": 11.2172,
      "step": 200970
    },
    {
      "epoch": 24.18531889290012,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735205467399612,
      "loss": 11.0903,
      "step": 200980
    },
    {
      "epoch": 24.186522262334538,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735179679840703,
      "loss": 11.1309,
      "step": 200990
    },
    {
      "epoch": 24.187725631768952,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735153891217807,
      "loss": 11.1467,
      "step": 201000
    },
    {
      "epoch": 24.18892900120337,
      "grad_norm": Infinity,
      "learning_rate": 0.00017351281015309618,
      "loss": 11.2023,
      "step": 201010
    },
    {
      "epoch": 24.190132370637787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017351023107802047,
      "loss": 11.1543,
      "step": 201020
    },
    {
      "epoch": 24.1913357400722,
      "grad_norm": Infinity,
      "learning_rate": 0.00017350765189655734,
      "loss": 11.1107,
      "step": 201030
    },
    {
      "epoch": 24.19253910950662,
      "grad_norm": Infinity,
      "learning_rate": 0.00017350507260871045,
      "loss": 11.2184,
      "step": 201040
    },
    {
      "epoch": 24.193742478941036,
      "grad_norm": Infinity,
      "learning_rate": 0.0001735024932144836,
      "loss": 11.2849,
      "step": 201050
    },
    {
      "epoch": 24.19494584837545,
      "grad_norm": Infinity,
      "learning_rate": 0.0001734999137138805,
      "loss": 11.1834,
      "step": 201060
    },
    {
      "epoch": 24.196149217809868,
      "grad_norm": Infinity,
      "learning_rate": 0.00017349733410690484,
      "loss": 11.2253,
      "step": 201070
    },
    {
      "epoch": 24.197352587244286,
      "grad_norm": Infinity,
      "learning_rate": 0.00017349475439356043,
      "loss": 11.1736,
      "step": 201080
    },
    {
      "epoch": 24.1985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.00017349217457385095,
      "loss": 11.222,
      "step": 201090
    },
    {
      "epoch": 24.199759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00017348959464778013,
      "loss": 11.0669,
      "step": 201100
    },
    {
      "epoch": 24.200962695547535,
      "grad_norm": Infinity,
      "learning_rate": 0.00017348701461535176,
      "loss": 11.0729,
      "step": 201110
    },
    {
      "epoch": 24.20216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.00017348443447656956,
      "loss": 11.2728,
      "step": 201120
    },
    {
      "epoch": 24.203369434416366,
      "grad_norm": Infinity,
      "learning_rate": 0.00017348185423143717,
      "loss": 11.0998,
      "step": 201130
    },
    {
      "epoch": 24.204572803850784,
      "grad_norm": Infinity,
      "learning_rate": 0.00017347927387995844,
      "loss": 11.1196,
      "step": 201140
    },
    {
      "epoch": 24.205776173285198,
      "grad_norm": Infinity,
      "learning_rate": 0.000173476693422137,
      "loss": 11.1924,
      "step": 201150
    },
    {
      "epoch": 24.206979542719615,
      "grad_norm": Infinity,
      "learning_rate": 0.0001734741128579767,
      "loss": 11.1145,
      "step": 201160
    },
    {
      "epoch": 24.20818291215403,
      "grad_norm": Infinity,
      "learning_rate": 0.00017347153218748122,
      "loss": 11.0596,
      "step": 201170
    },
    {
      "epoch": 24.209386281588447,
      "grad_norm": Infinity,
      "learning_rate": 0.00017346895141065428,
      "loss": 11.1005,
      "step": 201180
    },
    {
      "epoch": 24.210589651022865,
      "grad_norm": Infinity,
      "learning_rate": 0.0001734663705274996,
      "loss": 11.1415,
      "step": 201190
    },
    {
      "epoch": 24.21179302045728,
      "grad_norm": Infinity,
      "learning_rate": 0.00017346378953802098,
      "loss": 11.1808,
      "step": 201200
    },
    {
      "epoch": 24.212996389891696,
      "grad_norm": Infinity,
      "learning_rate": 0.0001734612084422221,
      "loss": 11.2767,
      "step": 201210
    },
    {
      "epoch": 24.214199759326114,
      "grad_norm": Infinity,
      "learning_rate": 0.00017345862724010676,
      "loss": 11.1925,
      "step": 201220
    },
    {
      "epoch": 24.215403128760528,
      "grad_norm": Infinity,
      "learning_rate": 0.0001734560459316786,
      "loss": 11.135,
      "step": 201230
    },
    {
      "epoch": 24.216606498194945,
      "grad_norm": Infinity,
      "learning_rate": 0.00017345346451694144,
      "loss": 11.0607,
      "step": 201240
    },
    {
      "epoch": 24.217809867629363,
      "grad_norm": Infinity,
      "learning_rate": 0.00017345088299589896,
      "loss": 11.0846,
      "step": 201250
    },
    {
      "epoch": 24.219013237063777,
      "grad_norm": Infinity,
      "learning_rate": 0.00017344830136855493,
      "loss": 11.218,
      "step": 201260
    },
    {
      "epoch": 24.220216606498195,
      "grad_norm": Infinity,
      "learning_rate": 0.00017344571963491304,
      "loss": 11.1366,
      "step": 201270
    },
    {
      "epoch": 24.221419975932612,
      "grad_norm": Infinity,
      "learning_rate": 0.0001734431377949771,
      "loss": 11.113,
      "step": 201280
    },
    {
      "epoch": 24.222623345367026,
      "grad_norm": Infinity,
      "learning_rate": 0.00017344055584875078,
      "loss": 11.1381,
      "step": 201290
    },
    {
      "epoch": 24.223826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.00017343797379623785,
      "loss": 11.1473,
      "step": 201300
    },
    {
      "epoch": 24.22503008423586,
      "grad_norm": Infinity,
      "learning_rate": 0.00017343539163744204,
      "loss": 11.2727,
      "step": 201310
    },
    {
      "epoch": 24.226233453670275,
      "grad_norm": Infinity,
      "learning_rate": 0.00017343280937236712,
      "loss": 11.1165,
      "step": 201320
    },
    {
      "epoch": 24.227436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00017343022700101678,
      "loss": 11.1562,
      "step": 201330
    },
    {
      "epoch": 24.22864019253911,
      "grad_norm": Infinity,
      "learning_rate": 0.00017342764452339476,
      "loss": 11.097,
      "step": 201340
    },
    {
      "epoch": 24.229843561973524,
      "grad_norm": Infinity,
      "learning_rate": 0.0001734250619395048,
      "loss": 11.0585,
      "step": 201350
    },
    {
      "epoch": 24.231046931407942,
      "grad_norm": Infinity,
      "learning_rate": 0.00017342247924935066,
      "loss": 11.1452,
      "step": 201360
    },
    {
      "epoch": 24.23225030084236,
      "grad_norm": Infinity,
      "learning_rate": 0.00017341989645293607,
      "loss": 11.0823,
      "step": 201370
    },
    {
      "epoch": 24.233453670276774,
      "grad_norm": Infinity,
      "learning_rate": 0.00017341731355026476,
      "loss": 11.1407,
      "step": 201380
    },
    {
      "epoch": 24.23465703971119,
      "grad_norm": Infinity,
      "learning_rate": 0.00017341473054134046,
      "loss": 11.0274,
      "step": 201390
    },
    {
      "epoch": 24.23586040914561,
      "grad_norm": Infinity,
      "learning_rate": 0.0001734121474261669,
      "loss": 11.0675,
      "step": 201400
    },
    {
      "epoch": 24.237063778580023,
      "grad_norm": Infinity,
      "learning_rate": 0.00017340956420474788,
      "loss": 11.1382,
      "step": 201410
    },
    {
      "epoch": 24.23826714801444,
      "grad_norm": Infinity,
      "learning_rate": 0.0001734069808770871,
      "loss": 11.1394,
      "step": 201420
    },
    {
      "epoch": 24.239470517448858,
      "grad_norm": Infinity,
      "learning_rate": 0.00017340439744318827,
      "loss": 11.1426,
      "step": 201430
    },
    {
      "epoch": 24.240673886883272,
      "grad_norm": Infinity,
      "learning_rate": 0.00017340181390305514,
      "loss": 11.102,
      "step": 201440
    },
    {
      "epoch": 24.24187725631769,
      "grad_norm": Infinity,
      "learning_rate": 0.00017339923025669148,
      "loss": 11.0605,
      "step": 201450
    },
    {
      "epoch": 24.243080625752107,
      "grad_norm": Infinity,
      "learning_rate": 0.00017339664650410105,
      "loss": 11.1254,
      "step": 201460
    },
    {
      "epoch": 24.24428399518652,
      "grad_norm": Infinity,
      "learning_rate": 0.0001733940626452875,
      "loss": 11.1687,
      "step": 201470
    },
    {
      "epoch": 24.24548736462094,
      "grad_norm": Infinity,
      "learning_rate": 0.00017339147868025462,
      "loss": 11.1794,
      "step": 201480
    },
    {
      "epoch": 24.246690734055356,
      "grad_norm": Infinity,
      "learning_rate": 0.0001733888946090062,
      "loss": 11.0631,
      "step": 201490
    },
    {
      "epoch": 24.24789410348977,
      "grad_norm": Infinity,
      "learning_rate": 0.00017338631043154588,
      "loss": 11.2065,
      "step": 201500
    },
    {
      "epoch": 24.249097472924188,
      "grad_norm": Infinity,
      "learning_rate": 0.00017338372614787747,
      "loss": 11.1384,
      "step": 201510
    },
    {
      "epoch": 24.250300842358605,
      "grad_norm": Infinity,
      "learning_rate": 0.00017338114175800468,
      "loss": 11.1357,
      "step": 201520
    },
    {
      "epoch": 24.25150421179302,
      "grad_norm": Infinity,
      "learning_rate": 0.00017337855726193125,
      "loss": 11.1712,
      "step": 201530
    },
    {
      "epoch": 24.252707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00017337597265966096,
      "loss": 11.1226,
      "step": 201540
    },
    {
      "epoch": 24.253910950661854,
      "grad_norm": Infinity,
      "learning_rate": 0.00017337338795119753,
      "loss": 11.1215,
      "step": 201550
    },
    {
      "epoch": 24.25511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00017337080313654468,
      "loss": 11.1206,
      "step": 201560
    },
    {
      "epoch": 24.256317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00017336821821570614,
      "loss": 11.2056,
      "step": 201570
    },
    {
      "epoch": 24.257521058965104,
      "grad_norm": Infinity,
      "learning_rate": 0.0001733656331886857,
      "loss": 11.1376,
      "step": 201580
    },
    {
      "epoch": 24.258724428399518,
      "grad_norm": Infinity,
      "learning_rate": 0.00017336304805548708,
      "loss": 11.179,
      "step": 201590
    },
    {
      "epoch": 24.259927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.000173360462816114,
      "loss": 11.2423,
      "step": 201600
    },
    {
      "epoch": 24.261131167268353,
      "grad_norm": Infinity,
      "learning_rate": 0.00017335787747057023,
      "loss": 11.2844,
      "step": 201610
    },
    {
      "epoch": 24.262334536702767,
      "grad_norm": Infinity,
      "learning_rate": 0.00017335529201885946,
      "loss": 11.1613,
      "step": 201620
    },
    {
      "epoch": 24.263537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.00017335270646098554,
      "loss": 11.1405,
      "step": 201630
    },
    {
      "epoch": 24.264741275571602,
      "grad_norm": Infinity,
      "learning_rate": 0.0001733501207969521,
      "loss": 11.2263,
      "step": 201640
    },
    {
      "epoch": 24.265944645006016,
      "grad_norm": Infinity,
      "learning_rate": 0.00017334753502676293,
      "loss": 11.1083,
      "step": 201650
    },
    {
      "epoch": 24.267148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00017334494915042178,
      "loss": 11.229,
      "step": 201660
    },
    {
      "epoch": 24.26835138387485,
      "grad_norm": Infinity,
      "learning_rate": 0.00017334236316793238,
      "loss": 11.2218,
      "step": 201670
    },
    {
      "epoch": 24.269554753309265,
      "grad_norm": Infinity,
      "learning_rate": 0.00017333977707929845,
      "loss": 11.0576,
      "step": 201680
    },
    {
      "epoch": 24.270758122743683,
      "grad_norm": Infinity,
      "learning_rate": 0.0001733371908845238,
      "loss": 11.008,
      "step": 201690
    },
    {
      "epoch": 24.2719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00017333460458361207,
      "loss": 11.0592,
      "step": 201700
    },
    {
      "epoch": 24.273164861612514,
      "grad_norm": Infinity,
      "learning_rate": 0.0001733320181765671,
      "loss": 11.2051,
      "step": 201710
    },
    {
      "epoch": 24.27436823104693,
      "grad_norm": Infinity,
      "learning_rate": 0.00017332943166339256,
      "loss": 11.1676,
      "step": 201720
    },
    {
      "epoch": 24.27557160048135,
      "grad_norm": Infinity,
      "learning_rate": 0.00017332684504409228,
      "loss": 11.1241,
      "step": 201730
    },
    {
      "epoch": 24.276774969915763,
      "grad_norm": Infinity,
      "learning_rate": 0.00017332425831866993,
      "loss": 11.2482,
      "step": 201740
    },
    {
      "epoch": 24.27797833935018,
      "grad_norm": Infinity,
      "learning_rate": 0.00017332167148712926,
      "loss": 11.2021,
      "step": 201750
    },
    {
      "epoch": 24.2791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00017331908454947404,
      "loss": 11.3209,
      "step": 201760
    },
    {
      "epoch": 24.280385078219012,
      "grad_norm": Infinity,
      "learning_rate": 0.000173316497505708,
      "loss": 11.2091,
      "step": 201770
    },
    {
      "epoch": 24.28158844765343,
      "grad_norm": Infinity,
      "learning_rate": 0.00017331391035583485,
      "loss": 11.2327,
      "step": 201780
    },
    {
      "epoch": 24.282791817087848,
      "grad_norm": Infinity,
      "learning_rate": 0.0001733113230998584,
      "loss": 11.1139,
      "step": 201790
    },
    {
      "epoch": 24.28399518652226,
      "grad_norm": Infinity,
      "learning_rate": 0.0001733087357377824,
      "loss": 11.1365,
      "step": 201800
    },
    {
      "epoch": 24.28519855595668,
      "grad_norm": Infinity,
      "learning_rate": 0.00017330614826961056,
      "loss": 11.1073,
      "step": 201810
    },
    {
      "epoch": 24.286401925391097,
      "grad_norm": Infinity,
      "learning_rate": 0.0001733035606953466,
      "loss": 11.066,
      "step": 201820
    },
    {
      "epoch": 24.28760529482551,
      "grad_norm": Infinity,
      "learning_rate": 0.00017330097301499427,
      "loss": 11.0502,
      "step": 201830
    },
    {
      "epoch": 24.28880866425993,
      "grad_norm": Infinity,
      "learning_rate": 0.00017329838522855734,
      "loss": 11.2217,
      "step": 201840
    },
    {
      "epoch": 24.290012033694346,
      "grad_norm": Infinity,
      "learning_rate": 0.00017329579733603957,
      "loss": 11.1092,
      "step": 201850
    },
    {
      "epoch": 24.29121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.00017329320933744469,
      "loss": 11.2223,
      "step": 201860
    },
    {
      "epoch": 24.292418772563177,
      "grad_norm": Infinity,
      "learning_rate": 0.00017329062123277642,
      "loss": 11.1845,
      "step": 201870
    },
    {
      "epoch": 24.29362214199759,
      "grad_norm": Infinity,
      "learning_rate": 0.00017328803302203854,
      "loss": 11.2092,
      "step": 201880
    },
    {
      "epoch": 24.29482551143201,
      "grad_norm": Infinity,
      "learning_rate": 0.0001732854447052348,
      "loss": 11.2188,
      "step": 201890
    },
    {
      "epoch": 24.296028880866427,
      "grad_norm": Infinity,
      "learning_rate": 0.00017328285628236888,
      "loss": 11.1232,
      "step": 201900
    },
    {
      "epoch": 24.29723225030084,
      "grad_norm": Infinity,
      "learning_rate": 0.0001732802677534446,
      "loss": 11.1396,
      "step": 201910
    },
    {
      "epoch": 24.29843561973526,
      "grad_norm": Infinity,
      "learning_rate": 0.00017327767911846567,
      "loss": 11.1998,
      "step": 201920
    },
    {
      "epoch": 24.299638989169676,
      "grad_norm": Infinity,
      "learning_rate": 0.00017327509037743587,
      "loss": 11.0563,
      "step": 201930
    },
    {
      "epoch": 24.30084235860409,
      "grad_norm": Infinity,
      "learning_rate": 0.00017327250153035889,
      "loss": 11.0986,
      "step": 201940
    },
    {
      "epoch": 24.302045728038507,
      "grad_norm": Infinity,
      "learning_rate": 0.00017326991257723856,
      "loss": 11.198,
      "step": 201950
    },
    {
      "epoch": 24.303249097472925,
      "grad_norm": Infinity,
      "learning_rate": 0.00017326732351807852,
      "loss": 11.1443,
      "step": 201960
    },
    {
      "epoch": 24.30445246690734,
      "grad_norm": Infinity,
      "learning_rate": 0.0001732647343528826,
      "loss": 11.2522,
      "step": 201970
    },
    {
      "epoch": 24.305655836341757,
      "grad_norm": Infinity,
      "learning_rate": 0.00017326214508165452,
      "loss": 11.0928,
      "step": 201980
    },
    {
      "epoch": 24.306859205776174,
      "grad_norm": Infinity,
      "learning_rate": 0.00017325955570439801,
      "loss": 11.0697,
      "step": 201990
    },
    {
      "epoch": 24.308062575210588,
      "grad_norm": Infinity,
      "learning_rate": 0.0001732569662211169,
      "loss": 11.0921,
      "step": 202000
    },
    {
      "epoch": 24.309265944645006,
      "grad_norm": Infinity,
      "learning_rate": 0.0001732543766318148,
      "loss": 11.1446,
      "step": 202010
    },
    {
      "epoch": 24.310469314079423,
      "grad_norm": Infinity,
      "learning_rate": 0.00017325178693649557,
      "loss": 11.068,
      "step": 202020
    },
    {
      "epoch": 24.311672683513837,
      "grad_norm": Infinity,
      "learning_rate": 0.0001732491971351629,
      "loss": 11.1512,
      "step": 202030
    },
    {
      "epoch": 24.312876052948255,
      "grad_norm": Infinity,
      "learning_rate": 0.00017324660722782057,
      "loss": 11.1301,
      "step": 202040
    },
    {
      "epoch": 24.314079422382672,
      "grad_norm": Infinity,
      "learning_rate": 0.0001732440172144723,
      "loss": 11.1322,
      "step": 202050
    },
    {
      "epoch": 24.315282791817086,
      "grad_norm": Infinity,
      "learning_rate": 0.00017324142709512189,
      "loss": 11.1576,
      "step": 202060
    },
    {
      "epoch": 24.316486161251504,
      "grad_norm": Infinity,
      "learning_rate": 0.000173238836869773,
      "loss": 11.1885,
      "step": 202070
    },
    {
      "epoch": 24.31768953068592,
      "grad_norm": Infinity,
      "learning_rate": 0.00017323624653842943,
      "loss": 11.1461,
      "step": 202080
    },
    {
      "epoch": 24.318892900120336,
      "grad_norm": Infinity,
      "learning_rate": 0.000173233656101095,
      "loss": 11.1913,
      "step": 202090
    },
    {
      "epoch": 24.320096269554753,
      "grad_norm": Infinity,
      "learning_rate": 0.00017323106555777333,
      "loss": 11.0499,
      "step": 202100
    },
    {
      "epoch": 24.32129963898917,
      "grad_norm": Infinity,
      "learning_rate": 0.00017322847490846823,
      "loss": 11.087,
      "step": 202110
    },
    {
      "epoch": 24.322503008423585,
      "grad_norm": Infinity,
      "learning_rate": 0.00017322588415318347,
      "loss": 11.0941,
      "step": 202120
    },
    {
      "epoch": 24.323706377858002,
      "grad_norm": Infinity,
      "learning_rate": 0.00017322329329192278,
      "loss": 11.1832,
      "step": 202130
    },
    {
      "epoch": 24.32490974729242,
      "grad_norm": Infinity,
      "learning_rate": 0.00017322070232468989,
      "loss": 11.0778,
      "step": 202140
    },
    {
      "epoch": 24.326113116726834,
      "grad_norm": Infinity,
      "learning_rate": 0.00017321811125148857,
      "loss": 11.2274,
      "step": 202150
    },
    {
      "epoch": 24.32731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.00017321552007232256,
      "loss": 11.0402,
      "step": 202160
    },
    {
      "epoch": 24.32851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.00017321292878719563,
      "loss": 11.1144,
      "step": 202170
    },
    {
      "epoch": 24.329723225030083,
      "grad_norm": Infinity,
      "learning_rate": 0.0001732103373961115,
      "loss": 11.0195,
      "step": 202180
    },
    {
      "epoch": 24.3309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.00017320774589907396,
      "loss": 11.1208,
      "step": 202190
    },
    {
      "epoch": 24.332129963898918,
      "grad_norm": Infinity,
      "learning_rate": 0.00017320515429608674,
      "loss": 11.2488,
      "step": 202200
    },
    {
      "epoch": 24.333333333333332,
      "grad_norm": Infinity,
      "learning_rate": 0.00017320256258715353,
      "loss": 11.2151,
      "step": 202210
    },
    {
      "epoch": 24.33453670276775,
      "grad_norm": Infinity,
      "learning_rate": 0.00017319997077227818,
      "loss": 11.1032,
      "step": 202220
    },
    {
      "epoch": 24.335740072202167,
      "grad_norm": Infinity,
      "learning_rate": 0.00017319737885146443,
      "loss": 11.1579,
      "step": 202230
    },
    {
      "epoch": 24.33694344163658,
      "grad_norm": Infinity,
      "learning_rate": 0.00017319478682471595,
      "loss": 11.0959,
      "step": 202240
    },
    {
      "epoch": 24.338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00017319219469203655,
      "loss": 11.0999,
      "step": 202250
    },
    {
      "epoch": 24.339350180505416,
      "grad_norm": Infinity,
      "learning_rate": 0.00017318960245343,
      "loss": 11.199,
      "step": 202260
    },
    {
      "epoch": 24.34055354993983,
      "grad_norm": Infinity,
      "learning_rate": 0.0001731870101089,
      "loss": 11.1715,
      "step": 202270
    },
    {
      "epoch": 24.341756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.00017318441765845036,
      "loss": 11.1442,
      "step": 202280
    },
    {
      "epoch": 24.342960288808666,
      "grad_norm": Infinity,
      "learning_rate": 0.00017318182510208475,
      "loss": 11.089,
      "step": 202290
    },
    {
      "epoch": 24.34416365824308,
      "grad_norm": Infinity,
      "learning_rate": 0.00017317923243980702,
      "loss": 11.2527,
      "step": 202300
    },
    {
      "epoch": 24.345367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017317663967162083,
      "loss": 11.1116,
      "step": 202310
    },
    {
      "epoch": 24.346570397111915,
      "grad_norm": Infinity,
      "learning_rate": 0.00017317404679753002,
      "loss": 11.0977,
      "step": 202320
    },
    {
      "epoch": 24.34777376654633,
      "grad_norm": Infinity,
      "learning_rate": 0.00017317145381753828,
      "loss": 11.0884,
      "step": 202330
    },
    {
      "epoch": 24.348977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.00017316886073164939,
      "loss": 11.266,
      "step": 202340
    },
    {
      "epoch": 24.350180505415164,
      "grad_norm": Infinity,
      "learning_rate": 0.00017316626753986704,
      "loss": 11.0885,
      "step": 202350
    },
    {
      "epoch": 24.351383874849578,
      "grad_norm": Infinity,
      "learning_rate": 0.0001731636742421951,
      "loss": 10.9854,
      "step": 202360
    },
    {
      "epoch": 24.352587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00017316108083863723,
      "loss": 11.1087,
      "step": 202370
    },
    {
      "epoch": 24.353790613718413,
      "grad_norm": Infinity,
      "learning_rate": 0.0001731584873291972,
      "loss": 11.1437,
      "step": 202380
    },
    {
      "epoch": 24.354993983152827,
      "grad_norm": Infinity,
      "learning_rate": 0.0001731558937138788,
      "loss": 11.1671,
      "step": 202390
    },
    {
      "epoch": 24.356197352587245,
      "grad_norm": Infinity,
      "learning_rate": 0.00017315329999268578,
      "loss": 11.1036,
      "step": 202400
    },
    {
      "epoch": 24.357400722021662,
      "grad_norm": Infinity,
      "learning_rate": 0.00017315070616562186,
      "loss": 11.1503,
      "step": 202410
    },
    {
      "epoch": 24.358604091456076,
      "grad_norm": Infinity,
      "learning_rate": 0.00017314811223269078,
      "loss": 11.2141,
      "step": 202420
    },
    {
      "epoch": 24.359807460890494,
      "grad_norm": Infinity,
      "learning_rate": 0.00017314551819389636,
      "loss": 11.1071,
      "step": 202430
    },
    {
      "epoch": 24.36101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.00017314292404924228,
      "loss": 11.127,
      "step": 202440
    },
    {
      "epoch": 24.362214199759325,
      "grad_norm": Infinity,
      "learning_rate": 0.00017314032979873236,
      "loss": 11.0803,
      "step": 202450
    },
    {
      "epoch": 24.363417569193743,
      "grad_norm": Infinity,
      "learning_rate": 0.0001731377354423703,
      "loss": 11.0004,
      "step": 202460
    },
    {
      "epoch": 24.36462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.0001731351409801599,
      "loss": 11.15,
      "step": 202470
    },
    {
      "epoch": 24.365824308062574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017313254641210488,
      "loss": 11.1934,
      "step": 202480
    },
    {
      "epoch": 24.367027677496992,
      "grad_norm": Infinity,
      "learning_rate": 0.000173129951738209,
      "loss": 11.1057,
      "step": 202490
    },
    {
      "epoch": 24.36823104693141,
      "grad_norm": Infinity,
      "learning_rate": 0.00017312735695847607,
      "loss": 11.0572,
      "step": 202500
    },
    {
      "epoch": 24.369434416365824,
      "grad_norm": Infinity,
      "learning_rate": 0.00017312476207290975,
      "loss": 11.0955,
      "step": 202510
    },
    {
      "epoch": 24.37063778580024,
      "grad_norm": Infinity,
      "learning_rate": 0.00017312216708151388,
      "loss": 11.1555,
      "step": 202520
    },
    {
      "epoch": 24.37184115523466,
      "grad_norm": Infinity,
      "learning_rate": 0.00017311957198429213,
      "loss": 11.1744,
      "step": 202530
    },
    {
      "epoch": 24.373044524669073,
      "grad_norm": Infinity,
      "learning_rate": 0.00017311697678124837,
      "loss": 11.2549,
      "step": 202540
    },
    {
      "epoch": 24.37424789410349,
      "grad_norm": Infinity,
      "learning_rate": 0.00017311438147238627,
      "loss": 11.1706,
      "step": 202550
    },
    {
      "epoch": 24.375451263537904,
      "grad_norm": Infinity,
      "learning_rate": 0.0001731117860577096,
      "loss": 11.0597,
      "step": 202560
    },
    {
      "epoch": 24.376654632972322,
      "grad_norm": Infinity,
      "learning_rate": 0.00017310919053722213,
      "loss": 11.1124,
      "step": 202570
    },
    {
      "epoch": 24.37785800240674,
      "grad_norm": Infinity,
      "learning_rate": 0.00017310659491092761,
      "loss": 11.1403,
      "step": 202580
    },
    {
      "epoch": 24.379061371841154,
      "grad_norm": Infinity,
      "learning_rate": 0.0001731039991788298,
      "loss": 11.1213,
      "step": 202590
    },
    {
      "epoch": 24.38026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00017310140334093243,
      "loss": 11.1974,
      "step": 202600
    },
    {
      "epoch": 24.38146811070999,
      "grad_norm": Infinity,
      "learning_rate": 0.00017309880739723932,
      "loss": 11.1309,
      "step": 202610
    },
    {
      "epoch": 24.382671480144403,
      "grad_norm": Infinity,
      "learning_rate": 0.00017309621134775418,
      "loss": 11.1348,
      "step": 202620
    },
    {
      "epoch": 24.38387484957882,
      "grad_norm": Infinity,
      "learning_rate": 0.00017309361519248078,
      "loss": 11.1128,
      "step": 202630
    },
    {
      "epoch": 24.385078219013238,
      "grad_norm": Infinity,
      "learning_rate": 0.00017309101893142286,
      "loss": 11.0654,
      "step": 202640
    },
    {
      "epoch": 24.386281588447652,
      "grad_norm": Infinity,
      "learning_rate": 0.00017308842256458416,
      "loss": 11.1745,
      "step": 202650
    },
    {
      "epoch": 24.38748495788207,
      "grad_norm": Infinity,
      "learning_rate": 0.0001730858260919685,
      "loss": 11.0448,
      "step": 202660
    },
    {
      "epoch": 24.388688327316487,
      "grad_norm": Infinity,
      "learning_rate": 0.00017308322951357962,
      "loss": 11.1019,
      "step": 202670
    },
    {
      "epoch": 24.3898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00017308063282942123,
      "loss": 11.297,
      "step": 202680
    },
    {
      "epoch": 24.39109506618532,
      "grad_norm": Infinity,
      "learning_rate": 0.00017307803603949714,
      "loss": 11.1159,
      "step": 202690
    },
    {
      "epoch": 24.392298435619736,
      "grad_norm": Infinity,
      "learning_rate": 0.0001730754391438111,
      "loss": 11.0349,
      "step": 202700
    },
    {
      "epoch": 24.39350180505415,
      "grad_norm": Infinity,
      "learning_rate": 0.00017307284214236685,
      "loss": 11.0906,
      "step": 202710
    },
    {
      "epoch": 24.394705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017307024503516812,
      "loss": 11.1149,
      "step": 202720
    },
    {
      "epoch": 24.395908543922985,
      "grad_norm": Infinity,
      "learning_rate": 0.00017306764782221875,
      "loss": 11.2081,
      "step": 202730
    },
    {
      "epoch": 24.3971119133574,
      "grad_norm": Infinity,
      "learning_rate": 0.00017306505050352245,
      "loss": 11.0717,
      "step": 202740
    },
    {
      "epoch": 24.398315282791817,
      "grad_norm": Infinity,
      "learning_rate": 0.00017306245307908296,
      "loss": 11.116,
      "step": 202750
    },
    {
      "epoch": 24.399518652226234,
      "grad_norm": Infinity,
      "learning_rate": 0.00017305985554890406,
      "loss": 11.136,
      "step": 202760
    },
    {
      "epoch": 24.40072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00017305725791298956,
      "loss": 11.1861,
      "step": 202770
    },
    {
      "epoch": 24.401925391095066,
      "grad_norm": Infinity,
      "learning_rate": 0.00017305466017134312,
      "loss": 10.9928,
      "step": 202780
    },
    {
      "epoch": 24.403128760529484,
      "grad_norm": Infinity,
      "learning_rate": 0.0001730520623239686,
      "loss": 11.1095,
      "step": 202790
    },
    {
      "epoch": 24.404332129963898,
      "grad_norm": Infinity,
      "learning_rate": 0.00017304946437086963,
      "loss": 11.0238,
      "step": 202800
    },
    {
      "epoch": 24.405535499398315,
      "grad_norm": Infinity,
      "learning_rate": 0.00017304686631205011,
      "loss": 11.1042,
      "step": 202810
    },
    {
      "epoch": 24.406738868832733,
      "grad_norm": Infinity,
      "learning_rate": 0.00017304426814751373,
      "loss": 11.1892,
      "step": 202820
    },
    {
      "epoch": 24.407942238267147,
      "grad_norm": Infinity,
      "learning_rate": 0.00017304166987726425,
      "loss": 11.2431,
      "step": 202830
    },
    {
      "epoch": 24.409145607701564,
      "grad_norm": Infinity,
      "learning_rate": 0.00017303907150130544,
      "loss": 11.1347,
      "step": 202840
    },
    {
      "epoch": 24.410348977135982,
      "grad_norm": Infinity,
      "learning_rate": 0.0001730364730196411,
      "loss": 11.1796,
      "step": 202850
    },
    {
      "epoch": 24.411552346570396,
      "grad_norm": Infinity,
      "learning_rate": 0.0001730338744322749,
      "loss": 11.2552,
      "step": 202860
    },
    {
      "epoch": 24.412755716004813,
      "grad_norm": Infinity,
      "learning_rate": 0.00017303127573921066,
      "loss": 11.2265,
      "step": 202870
    },
    {
      "epoch": 24.41395908543923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017302867694045216,
      "loss": 11.2379,
      "step": 202880
    },
    {
      "epoch": 24.415162454873645,
      "grad_norm": Infinity,
      "learning_rate": 0.00017302607803600312,
      "loss": 11.1952,
      "step": 202890
    },
    {
      "epoch": 24.416365824308063,
      "grad_norm": Infinity,
      "learning_rate": 0.0001730234790258673,
      "loss": 11.1413,
      "step": 202900
    },
    {
      "epoch": 24.41756919374248,
      "grad_norm": Infinity,
      "learning_rate": 0.0001730208799100485,
      "loss": 11.0516,
      "step": 202910
    },
    {
      "epoch": 24.418772563176894,
      "grad_norm": Infinity,
      "learning_rate": 0.00017301828068855044,
      "loss": 11.1639,
      "step": 202920
    },
    {
      "epoch": 24.41997593261131,
      "grad_norm": Infinity,
      "learning_rate": 0.0001730156813613769,
      "loss": 11.1749,
      "step": 202930
    },
    {
      "epoch": 24.42117930204573,
      "grad_norm": Infinity,
      "learning_rate": 0.00017301308192853166,
      "loss": 11.1614,
      "step": 202940
    },
    {
      "epoch": 24.422382671480143,
      "grad_norm": Infinity,
      "learning_rate": 0.00017301048239001845,
      "loss": 11.1019,
      "step": 202950
    },
    {
      "epoch": 24.42358604091456,
      "grad_norm": Infinity,
      "learning_rate": 0.00017300788274584106,
      "loss": 11.1456,
      "step": 202960
    },
    {
      "epoch": 24.42478941034898,
      "grad_norm": Infinity,
      "learning_rate": 0.0001730052829960032,
      "loss": 10.9926,
      "step": 202970
    },
    {
      "epoch": 24.425992779783392,
      "grad_norm": Infinity,
      "learning_rate": 0.00017300268314050872,
      "loss": 11.2388,
      "step": 202980
    },
    {
      "epoch": 24.42719614921781,
      "grad_norm": Infinity,
      "learning_rate": 0.00017300008317936132,
      "loss": 11.1645,
      "step": 202990
    },
    {
      "epoch": 24.428399518652228,
      "grad_norm": Infinity,
      "learning_rate": 0.00017299748311256478,
      "loss": 11.2056,
      "step": 203000
    },
    {
      "epoch": 24.42960288808664,
      "grad_norm": Infinity,
      "learning_rate": 0.00017299488294012286,
      "loss": 11.1836,
      "step": 203010
    },
    {
      "epoch": 24.43080625752106,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729922826620393,
      "loss": 11.1311,
      "step": 203020
    },
    {
      "epoch": 24.432009626955477,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729896822783179,
      "loss": 11.0801,
      "step": 203030
    },
    {
      "epoch": 24.43321299638989,
      "grad_norm": Infinity,
      "learning_rate": 0.00017298708178896242,
      "loss": 11.2173,
      "step": 203040
    },
    {
      "epoch": 24.43441636582431,
      "grad_norm": Infinity,
      "learning_rate": 0.00017298448119397656,
      "loss": 11.023,
      "step": 203050
    },
    {
      "epoch": 24.435619735258726,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729818804933642,
      "loss": 11.1907,
      "step": 203060
    },
    {
      "epoch": 24.43682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.000172979279687129,
      "loss": 11.1762,
      "step": 203070
    },
    {
      "epoch": 24.438026474127557,
      "grad_norm": Infinity,
      "learning_rate": 0.00017297667877527482,
      "loss": 11.1403,
      "step": 203080
    },
    {
      "epoch": 24.439229843561975,
      "grad_norm": Infinity,
      "learning_rate": 0.00017297407775780532,
      "loss": 11.1323,
      "step": 203090
    },
    {
      "epoch": 24.44043321299639,
      "grad_norm": Infinity,
      "learning_rate": 0.00017297147663472432,
      "loss": 11.0599,
      "step": 203100
    },
    {
      "epoch": 24.441636582430807,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729688754060356,
      "loss": 11.1402,
      "step": 203110
    },
    {
      "epoch": 24.442839951865224,
      "grad_norm": Infinity,
      "learning_rate": 0.00017296627407174285,
      "loss": 11.1638,
      "step": 203120
    },
    {
      "epoch": 24.444043321299638,
      "grad_norm": Infinity,
      "learning_rate": 0.00017296367263184995,
      "loss": 11.2891,
      "step": 203130
    },
    {
      "epoch": 24.445246690734056,
      "grad_norm": Infinity,
      "learning_rate": 0.00017296107108636055,
      "loss": 10.9729,
      "step": 203140
    },
    {
      "epoch": 24.446450060168473,
      "grad_norm": Infinity,
      "learning_rate": 0.00017295846943527846,
      "loss": 11.0357,
      "step": 203150
    },
    {
      "epoch": 24.447653429602887,
      "grad_norm": Infinity,
      "learning_rate": 0.00017295586767860748,
      "loss": 11.0406,
      "step": 203160
    },
    {
      "epoch": 24.448856799037305,
      "grad_norm": Infinity,
      "learning_rate": 0.00017295326581635137,
      "loss": 11.0706,
      "step": 203170
    },
    {
      "epoch": 24.450060168471722,
      "grad_norm": Infinity,
      "learning_rate": 0.00017295066384851384,
      "loss": 11.1838,
      "step": 203180
    },
    {
      "epoch": 24.451263537906136,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729480617750987,
      "loss": 11.198,
      "step": 203190
    },
    {
      "epoch": 24.452466907340554,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729454595961097,
      "loss": 11.0865,
      "step": 203200
    },
    {
      "epoch": 24.45367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017294285731155062,
      "loss": 11.1278,
      "step": 203210
    },
    {
      "epoch": 24.454873646209386,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729402549214252,
      "loss": 11.2212,
      "step": 203220
    },
    {
      "epoch": 24.456077015643803,
      "grad_norm": Infinity,
      "learning_rate": 0.00017293765242573726,
      "loss": 11.0992,
      "step": 203230
    },
    {
      "epoch": 24.45728038507822,
      "grad_norm": Infinity,
      "learning_rate": 0.00017293504982449046,
      "loss": 11.0595,
      "step": 203240
    },
    {
      "epoch": 24.458483754512635,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729324471176887,
      "loss": 11.135,
      "step": 203250
    },
    {
      "epoch": 24.459687123947052,
      "grad_norm": Infinity,
      "learning_rate": 0.00017292984430533566,
      "loss": 11.2446,
      "step": 203260
    },
    {
      "epoch": 24.460890493381466,
      "grad_norm": Infinity,
      "learning_rate": 0.00017292724138743513,
      "loss": 11.1812,
      "step": 203270
    },
    {
      "epoch": 24.462093862815884,
      "grad_norm": Infinity,
      "learning_rate": 0.00017292463836399087,
      "loss": 11.1719,
      "step": 203280
    },
    {
      "epoch": 24.4632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00017292203523500664,
      "loss": 11.1675,
      "step": 203290
    },
    {
      "epoch": 24.464500601684716,
      "grad_norm": Infinity,
      "learning_rate": 0.00017291943200048625,
      "loss": 11.186,
      "step": 203300
    },
    {
      "epoch": 24.465703971119133,
      "grad_norm": Infinity,
      "learning_rate": 0.00017291682866043344,
      "loss": 11.1694,
      "step": 203310
    },
    {
      "epoch": 24.46690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.00017291422521485195,
      "loss": 11.0342,
      "step": 203320
    },
    {
      "epoch": 24.468110709987965,
      "grad_norm": Infinity,
      "learning_rate": 0.00017291162166374559,
      "loss": 11.2214,
      "step": 203330
    },
    {
      "epoch": 24.469314079422382,
      "grad_norm": Infinity,
      "learning_rate": 0.00017290901800711813,
      "loss": 11.1639,
      "step": 203340
    },
    {
      "epoch": 24.4705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729064142449733,
      "loss": 11.0322,
      "step": 203350
    },
    {
      "epoch": 24.471720818291214,
      "grad_norm": Infinity,
      "learning_rate": 0.0001729038103773149,
      "loss": 11.1796,
      "step": 203360
    },
    {
      "epoch": 24.47292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00017290120640414664,
      "loss": 11.2121,
      "step": 203370
    },
    {
      "epoch": 24.47412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728986023254724,
      "loss": 11.0983,
      "step": 203380
    },
    {
      "epoch": 24.475330926594463,
      "grad_norm": Infinity,
      "learning_rate": 0.00017289599814129586,
      "loss": 11.2589,
      "step": 203390
    },
    {
      "epoch": 24.47653429602888,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728933938516208,
      "loss": 10.9977,
      "step": 203400
    },
    {
      "epoch": 24.477737665463298,
      "grad_norm": Infinity,
      "learning_rate": 0.000172890789456451,
      "loss": 11.1936,
      "step": 203410
    },
    {
      "epoch": 24.478941034897712,
      "grad_norm": Infinity,
      "learning_rate": 0.00017288818495579027,
      "loss": 11.2151,
      "step": 203420
    },
    {
      "epoch": 24.48014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728855803496423,
      "loss": 11.0796,
      "step": 203430
    },
    {
      "epoch": 24.481347773766547,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728829756380109,
      "loss": 11.1623,
      "step": 203440
    },
    {
      "epoch": 24.48255114320096,
      "grad_norm": Infinity,
      "learning_rate": 0.00017288037082089988,
      "loss": 10.9397,
      "step": 203450
    },
    {
      "epoch": 24.48375451263538,
      "grad_norm": Infinity,
      "learning_rate": 0.00017287776589831294,
      "loss": 11.2153,
      "step": 203460
    },
    {
      "epoch": 24.484957882069796,
      "grad_norm": Infinity,
      "learning_rate": 0.00017287516087025387,
      "loss": 11.0897,
      "step": 203470
    },
    {
      "epoch": 24.48616125150421,
      "grad_norm": Infinity,
      "learning_rate": 0.00017287255573672646,
      "loss": 11.2646,
      "step": 203480
    },
    {
      "epoch": 24.487364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728699504977345,
      "loss": 11.2013,
      "step": 203490
    },
    {
      "epoch": 24.488567990373046,
      "grad_norm": Infinity,
      "learning_rate": 0.00017286734515328168,
      "loss": 11.1848,
      "step": 203500
    },
    {
      "epoch": 24.48977135980746,
      "grad_norm": Infinity,
      "learning_rate": 0.00017286473970337186,
      "loss": 11.0456,
      "step": 203510
    },
    {
      "epoch": 24.490974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00017286213414800878,
      "loss": 11.0632,
      "step": 203520
    },
    {
      "epoch": 24.492178098676295,
      "grad_norm": Infinity,
      "learning_rate": 0.00017285952848719615,
      "loss": 11.1796,
      "step": 203530
    },
    {
      "epoch": 24.49338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00017285692272093784,
      "loss": 11.1602,
      "step": 203540
    },
    {
      "epoch": 24.494584837545126,
      "grad_norm": Infinity,
      "learning_rate": 0.00017285431684923753,
      "loss": 11.2746,
      "step": 203550
    },
    {
      "epoch": 24.495788206979544,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728517108720991,
      "loss": 11.0718,
      "step": 203560
    },
    {
      "epoch": 24.496991576413958,
      "grad_norm": Infinity,
      "learning_rate": 0.00017284910478952622,
      "loss": 11.0925,
      "step": 203570
    },
    {
      "epoch": 24.498194945848375,
      "grad_norm": Infinity,
      "learning_rate": 0.00017284649860152272,
      "loss": 11.1142,
      "step": 203580
    },
    {
      "epoch": 24.499398315282793,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728438923080923,
      "loss": 11.2081,
      "step": 203590
    },
    {
      "epoch": 24.500601684717207,
      "grad_norm": Infinity,
      "learning_rate": 0.00017284128590923885,
      "loss": 11.1378,
      "step": 203600
    },
    {
      "epoch": 24.501805054151625,
      "grad_norm": Infinity,
      "learning_rate": 0.00017283867940496604,
      "loss": 11.1431,
      "step": 203610
    },
    {
      "epoch": 24.503008423586042,
      "grad_norm": Infinity,
      "learning_rate": 0.00017283607279527767,
      "loss": 11.1411,
      "step": 203620
    },
    {
      "epoch": 24.504211793020456,
      "grad_norm": Infinity,
      "learning_rate": 0.00017283346608017755,
      "loss": 11.0164,
      "step": 203630
    },
    {
      "epoch": 24.505415162454874,
      "grad_norm": Infinity,
      "learning_rate": 0.00017283085925966937,
      "loss": 11.1921,
      "step": 203640
    },
    {
      "epoch": 24.50661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.000172828252333757,
      "loss": 11.1421,
      "step": 203650
    },
    {
      "epoch": 24.507821901323705,
      "grad_norm": Infinity,
      "learning_rate": 0.00017282564530244417,
      "loss": 11.1104,
      "step": 203660
    },
    {
      "epoch": 24.509025270758123,
      "grad_norm": Infinity,
      "learning_rate": 0.00017282303816573463,
      "loss": 11.0402,
      "step": 203670
    },
    {
      "epoch": 24.51022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728204309236322,
      "loss": 11.1243,
      "step": 203680
    },
    {
      "epoch": 24.511432009626954,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728178235761406,
      "loss": 11.181,
      "step": 203690
    },
    {
      "epoch": 24.512635379061372,
      "grad_norm": Infinity,
      "learning_rate": 0.00017281521612326365,
      "loss": 10.9586,
      "step": 203700
    },
    {
      "epoch": 24.51383874849579,
      "grad_norm": Infinity,
      "learning_rate": 0.00017281260856500508,
      "loss": 11.109,
      "step": 203710
    },
    {
      "epoch": 24.515042117930204,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728100009013687,
      "loss": 11.2251,
      "step": 203720
    },
    {
      "epoch": 24.51624548736462,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728073931323583,
      "loss": 11.1211,
      "step": 203730
    },
    {
      "epoch": 24.51744885679904,
      "grad_norm": Infinity,
      "learning_rate": 0.00017280478525797757,
      "loss": 10.9843,
      "step": 203740
    },
    {
      "epoch": 24.518652226233453,
      "grad_norm": Infinity,
      "learning_rate": 0.0001728021772782304,
      "loss": 11.1796,
      "step": 203750
    },
    {
      "epoch": 24.51985559566787,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727995691931205,
      "loss": 11.1097,
      "step": 203760
    },
    {
      "epoch": 24.521058965102288,
      "grad_norm": Infinity,
      "learning_rate": 0.00017279696100265163,
      "loss": 11.077,
      "step": 203770
    },
    {
      "epoch": 24.522262334536702,
      "grad_norm": Infinity,
      "learning_rate": 0.00017279435270682754,
      "loss": 11.1413,
      "step": 203780
    },
    {
      "epoch": 24.52346570397112,
      "grad_norm": Infinity,
      "learning_rate": 0.00017279174430565213,
      "loss": 11.1108,
      "step": 203790
    },
    {
      "epoch": 24.524669073405537,
      "grad_norm": Infinity,
      "learning_rate": 0.00017278913579912906,
      "loss": 11.1405,
      "step": 203800
    },
    {
      "epoch": 24.52587244283995,
      "grad_norm": Infinity,
      "learning_rate": 0.00017278652718726213,
      "loss": 11.2398,
      "step": 203810
    },
    {
      "epoch": 24.52707581227437,
      "grad_norm": Infinity,
      "learning_rate": 0.00017278391847005514,
      "loss": 11.2712,
      "step": 203820
    },
    {
      "epoch": 24.528279181708786,
      "grad_norm": Infinity,
      "learning_rate": 0.00017278130964751184,
      "loss": 11.123,
      "step": 203830
    },
    {
      "epoch": 24.5294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.00017277870071963602,
      "loss": 11.0383,
      "step": 203840
    },
    {
      "epoch": 24.530685920577618,
      "grad_norm": Infinity,
      "learning_rate": 0.00017277609168643142,
      "loss": 11.0184,
      "step": 203850
    },
    {
      "epoch": 24.531889290012035,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727734825479019,
      "loss": 11.2418,
      "step": 203860
    },
    {
      "epoch": 24.53309265944645,
      "grad_norm": Infinity,
      "learning_rate": 0.00017277087330405119,
      "loss": 11.1657,
      "step": 203870
    },
    {
      "epoch": 24.534296028880867,
      "grad_norm": Infinity,
      "learning_rate": 0.00017276826395488303,
      "loss": 11.1063,
      "step": 203880
    },
    {
      "epoch": 24.535499398315284,
      "grad_norm": Infinity,
      "learning_rate": 0.00017276565450040119,
      "loss": 11.1911,
      "step": 203890
    },
    {
      "epoch": 24.5367027677497,
      "grad_norm": Infinity,
      "learning_rate": 0.00017276304494060954,
      "loss": 11.0958,
      "step": 203900
    },
    {
      "epoch": 24.537906137184116,
      "grad_norm": Infinity,
      "learning_rate": 0.00017276043527551175,
      "loss": 11.1619,
      "step": 203910
    },
    {
      "epoch": 24.53910950661853,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727578255051117,
      "loss": 11.1901,
      "step": 203920
    },
    {
      "epoch": 24.540312876052948,
      "grad_norm": Infinity,
      "learning_rate": 0.00017275521562941308,
      "loss": 11.2114,
      "step": 203930
    },
    {
      "epoch": 24.541516245487365,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727526056484197,
      "loss": 11.1755,
      "step": 203940
    },
    {
      "epoch": 24.54271961492178,
      "grad_norm": Infinity,
      "learning_rate": 0.00017274999556213537,
      "loss": 11.0446,
      "step": 203950
    },
    {
      "epoch": 24.543922984356197,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727473853705638,
      "loss": 11.1149,
      "step": 203960
    },
    {
      "epoch": 24.545126353790614,
      "grad_norm": Infinity,
      "learning_rate": 0.00017274477507370884,
      "loss": 11.1213,
      "step": 203970
    },
    {
      "epoch": 24.54632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.00017274216467157418,
      "loss": 11.1316,
      "step": 203980
    },
    {
      "epoch": 24.547533092659446,
      "grad_norm": Infinity,
      "learning_rate": 0.00017273955416416368,
      "loss": 11.26,
      "step": 203990
    },
    {
      "epoch": 24.548736462093864,
      "grad_norm": Infinity,
      "learning_rate": 0.00017273694355148108,
      "loss": 11.1334,
      "step": 204000
    },
    {
      "epoch": 24.549939831528278,
      "grad_norm": Infinity,
      "learning_rate": 0.00017273433283353015,
      "loss": 11.157,
      "step": 204010
    },
    {
      "epoch": 24.551143200962695,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727317220103147,
      "loss": 11.1606,
      "step": 204020
    },
    {
      "epoch": 24.552346570397113,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727291110818385,
      "loss": 11.1745,
      "step": 204030
    },
    {
      "epoch": 24.553549939831527,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727265000481053,
      "loss": 11.0241,
      "step": 204040
    },
    {
      "epoch": 24.554753309265944,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727238889091189,
      "loss": 11.1493,
      "step": 204050
    },
    {
      "epoch": 24.555956678700362,
      "grad_norm": Infinity,
      "learning_rate": 0.00017272127766488309,
      "loss": 11.1397,
      "step": 204060
    },
    {
      "epoch": 24.557160048134776,
      "grad_norm": Infinity,
      "learning_rate": 0.00017271866631540163,
      "loss": 11.1514,
      "step": 204070
    },
    {
      "epoch": 24.558363417569193,
      "grad_norm": Infinity,
      "learning_rate": 0.00017271605486067831,
      "loss": 11.1527,
      "step": 204080
    },
    {
      "epoch": 24.55956678700361,
      "grad_norm": Infinity,
      "learning_rate": 0.0001727134433007169,
      "loss": 11.1628,
      "step": 204090
    },
    {
      "epoch": 24.560770156438025,
      "grad_norm": Infinity,
      "learning_rate": 0.00017271083163552118,
      "loss": 11.1046,
      "step": 204100
    },
    {
      "epoch": 24.561973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.00017270821986509492,
      "loss": 11.1496,
      "step": 204110
    },
    {
      "epoch": 24.56317689530686,
      "grad_norm": Infinity,
      "learning_rate": 0.00017270560798944196,
      "loss": 10.9967,
      "step": 204120
    },
    {
      "epoch": 24.564380264741274,
      "grad_norm": Infinity,
      "learning_rate": 0.00017270299600856604,
      "loss": 11.1739,
      "step": 204130
    },
    {
      "epoch": 24.56558363417569,
      "grad_norm": Infinity,
      "learning_rate": 0.00017270038392247088,
      "loss": 11.2443,
      "step": 204140
    },
    {
      "epoch": 24.56678700361011,
      "grad_norm": Infinity,
      "learning_rate": 0.00017269777173116034,
      "loss": 11.1713,
      "step": 204150
    },
    {
      "epoch": 24.567990373044523,
      "grad_norm": Infinity,
      "learning_rate": 0.00017269515943463818,
      "loss": 11.2142,
      "step": 204160
    },
    {
      "epoch": 24.56919374247894,
      "grad_norm": Infinity,
      "learning_rate": 0.0001726925470329082,
      "loss": 11.1771,
      "step": 204170
    },
    {
      "epoch": 24.57039711191336,
      "grad_norm": Infinity,
      "learning_rate": 0.00017268993452597412,
      "loss": 11.226,
      "step": 204180
    },
    {
      "epoch": 24.571600481347772,
      "grad_norm": Infinity,
      "learning_rate": 0.00017268732191383975,
      "loss": 11.2209,
      "step": 204190
    },
    {
      "epoch": 24.57280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.0001726847091965089,
      "loss": 11.0481,
      "step": 204200
    },
    {
      "epoch": 24.574007220216608,
      "grad_norm": Infinity,
      "learning_rate": 0.00017268209637398536,
      "loss": 11.1864,
      "step": 204210
    },
    {
      "epoch": 24.57521058965102,
      "grad_norm": Infinity,
      "learning_rate": 0.00017267948344627286,
      "loss": 11.0929,
      "step": 204220
    },
    {
      "epoch": 24.57641395908544,
      "grad_norm": Infinity,
      "learning_rate": 0.0001726768704133752,
      "loss": 11.0729,
      "step": 204230
    },
    {
      "epoch": 24.577617328519857,
      "grad_norm": Infinity,
      "learning_rate": 0.00017267425727529614,
      "loss": 11.1272,
      "step": 204240
    },
    {
      "epoch": 24.57882069795427,
      "grad_norm": Infinity,
      "learning_rate": 0.0001726716440320395,
      "loss": 11.0649,
      "step": 204250
    },
    {
      "epoch": 24.58002406738869,
      "grad_norm": Infinity,
      "learning_rate": 0.00017266903068360908,
      "loss": 11.2332,
      "step": 204260
    },
    {
      "epoch": 24.581227436823106,
      "grad_norm": Infinity,
      "learning_rate": 0.00017266641723000858,
      "loss": 11.1989,
      "step": 204270
    },
    {
      "epoch": 24.58243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.0001726638036712419,
      "loss": 11.1747,
      "step": 204280
    },
    {
      "epoch": 24.583634175691937,
      "grad_norm": Infinity,
      "learning_rate": 0.00017266119000731268,
      "loss": 11.2561,
      "step": 204290
    },
    {
      "epoch": 24.584837545126355,
      "grad_norm": Infinity,
      "learning_rate": 0.00017265857623822485,
      "loss": 11.2219,
      "step": 204300
    },
    {
      "epoch": 24.58604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.00017265596236398206,
      "loss": 11.2434,
      "step": 204310
    },
    {
      "epoch": 24.587244283995187,
      "grad_norm": Infinity,
      "learning_rate": 0.0001726533483845882,
      "loss": 11.0539,
      "step": 204320
    },
    {
      "epoch": 24.588447653429604,
      "grad_norm": Infinity,
      "learning_rate": 0.00017265073430004698,
      "loss": 11.2424,
      "step": 204330
    },
    {
      "epoch": 24.589651022864018,
      "grad_norm": Infinity,
      "learning_rate": 0.0001726481201103622,
      "loss": 11.2003,
      "step": 204340
    },
    {
      "epoch": 24.590854392298436,
      "grad_norm": Infinity,
      "learning_rate": 0.0001726455058155377,
      "loss": 11.1297,
      "step": 204350
    },
    {
      "epoch": 24.592057761732853,
      "grad_norm": Infinity,
      "learning_rate": 0.00017264289141557718,
      "loss": 11.1516,
      "step": 204360
    },
    {
      "epoch": 24.593261131167267,
      "grad_norm": Infinity,
      "learning_rate": 0.00017264027691048448,
      "loss": 11.1794,
      "step": 204370
    },
    {
      "epoch": 24.594464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00017263766230026336,
      "loss": 11.1801,
      "step": 204380
    },
    {
      "epoch": 24.595667870036102,
      "grad_norm": Infinity,
      "learning_rate": 0.00017263504758491762,
      "loss": 11.1979,
      "step": 204390
    },
    {
      "epoch": 24.596871239470516,
      "grad_norm": Infinity,
      "learning_rate": 0.000172632432764451,
      "loss": 11.097,
      "step": 204400
    },
    {
      "epoch": 24.598074608904934,
      "grad_norm": Infinity,
      "learning_rate": 0.00017262981783886734,
      "loss": 11.2459,
      "step": 204410
    },
    {
      "epoch": 24.59927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.0001726272028081704,
      "loss": 11.1592,
      "step": 204420
    },
    {
      "epoch": 24.600481347773766,
      "grad_norm": Infinity,
      "learning_rate": 0.00017262458767236397,
      "loss": 11.2419,
      "step": 204430
    },
    {
      "epoch": 24.601684717208183,
      "grad_norm": Infinity,
      "learning_rate": 0.00017262197243145182,
      "loss": 11.2337,
      "step": 204440
    },
    {
      "epoch": 24.6028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.00017261935708543774,
      "loss": 11.0574,
      "step": 204450
    },
    {
      "epoch": 24.604091456077015,
      "grad_norm": Infinity,
      "learning_rate": 0.00017261674163432554,
      "loss": 11.1017,
      "step": 204460
    },
    {
      "epoch": 24.605294825511432,
      "grad_norm": Infinity,
      "learning_rate": 0.000172614126078119,
      "loss": 11.1663,
      "step": 204470
    },
    {
      "epoch": 24.60649819494585,
      "grad_norm": Infinity,
      "learning_rate": 0.00017261151041682188,
      "loss": 10.9653,
      "step": 204480
    },
    {
      "epoch": 24.607701564380264,
      "grad_norm": Infinity,
      "learning_rate": 0.00017260889465043795,
      "loss": 11.1901,
      "step": 204490
    },
    {
      "epoch": 24.60890493381468,
      "grad_norm": Infinity,
      "learning_rate": 0.00017260627877897105,
      "loss": 11.1646,
      "step": 204500
    },
    {
      "epoch": 24.6101083032491,
      "grad_norm": Infinity,
      "learning_rate": 0.00017260366280242494,
      "loss": 11.0836,
      "step": 204510
    },
    {
      "epoch": 24.611311672683513,
      "grad_norm": Infinity,
      "learning_rate": 0.00017260104672080337,
      "loss": 11.3492,
      "step": 204520
    },
    {
      "epoch": 24.61251504211793,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725984305341102,
      "loss": 11.1114,
      "step": 204530
    },
    {
      "epoch": 24.613718411552348,
      "grad_norm": Infinity,
      "learning_rate": 0.00017259581424234916,
      "loss": 11.2224,
      "step": 204540
    },
    {
      "epoch": 24.614921780986762,
      "grad_norm": Infinity,
      "learning_rate": 0.00017259319784552405,
      "loss": 11.0706,
      "step": 204550
    },
    {
      "epoch": 24.61612515042118,
      "grad_norm": Infinity,
      "learning_rate": 0.00017259058134363867,
      "loss": 11.0646,
      "step": 204560
    },
    {
      "epoch": 24.617328519855597,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725879647366968,
      "loss": 11.1819,
      "step": 204570
    },
    {
      "epoch": 24.61853188929001,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725853480247022,
      "loss": 11.23,
      "step": 204580
    },
    {
      "epoch": 24.61973525872443,
      "grad_norm": Infinity,
      "learning_rate": 0.00017258273120765872,
      "loss": 11.0573,
      "step": 204590
    },
    {
      "epoch": 24.620938628158846,
      "grad_norm": Infinity,
      "learning_rate": 0.00017258011428557008,
      "loss": 11.2542,
      "step": 204600
    },
    {
      "epoch": 24.62214199759326,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725774972584401,
      "loss": 11.1398,
      "step": 204610
    },
    {
      "epoch": 24.623345367027678,
      "grad_norm": Infinity,
      "learning_rate": 0.00017257488012627255,
      "loss": 11.1744,
      "step": 204620
    },
    {
      "epoch": 24.624548736462096,
      "grad_norm": Infinity,
      "learning_rate": 0.00017257226288907126,
      "loss": 11.0909,
      "step": 204630
    },
    {
      "epoch": 24.62575210589651,
      "grad_norm": Infinity,
      "learning_rate": 0.00017256964554683996,
      "loss": 11.1488,
      "step": 204640
    },
    {
      "epoch": 24.626955475330927,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725670280995825,
      "loss": 11.1572,
      "step": 204650
    },
    {
      "epoch": 24.628158844765345,
      "grad_norm": Infinity,
      "learning_rate": 0.00017256441054730258,
      "loss": 11.065,
      "step": 204660
    },
    {
      "epoch": 24.62936221419976,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725617928900041,
      "loss": 11.1162,
      "step": 204670
    },
    {
      "epoch": 24.630565583634176,
      "grad_norm": Infinity,
      "learning_rate": 0.00017255917512769075,
      "loss": 11.0603,
      "step": 204680
    },
    {
      "epoch": 24.63176895306859,
      "grad_norm": Infinity,
      "learning_rate": 0.00017255655726036636,
      "loss": 11.0942,
      "step": 204690
    },
    {
      "epoch": 24.632972322503008,
      "grad_norm": Infinity,
      "learning_rate": 0.00017255393928803473,
      "loss": 11.0301,
      "step": 204700
    },
    {
      "epoch": 24.634175691937426,
      "grad_norm": Infinity,
      "learning_rate": 0.00017255132121069968,
      "loss": 11.0594,
      "step": 204710
    },
    {
      "epoch": 24.63537906137184,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725487030283649,
      "loss": 11.0392,
      "step": 204720
    },
    {
      "epoch": 24.636582430806257,
      "grad_norm": Infinity,
      "learning_rate": 0.00017254608474103424,
      "loss": 11.0384,
      "step": 204730
    },
    {
      "epoch": 24.637785800240675,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725434663487115,
      "loss": 11.3137,
      "step": 204740
    },
    {
      "epoch": 24.63898916967509,
      "grad_norm": Infinity,
      "learning_rate": 0.00017254084785140047,
      "loss": 11.2037,
      "step": 204750
    },
    {
      "epoch": 24.640192539109506,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725382292491049,
      "loss": 11.1143,
      "step": 204760
    },
    {
      "epoch": 24.641395908543924,
      "grad_norm": Infinity,
      "learning_rate": 0.00017253561054182857,
      "loss": 11.2488,
      "step": 204770
    },
    {
      "epoch": 24.642599277978338,
      "grad_norm": Infinity,
      "learning_rate": 0.00017253299172957537,
      "loss": 11.1119,
      "step": 204780
    },
    {
      "epoch": 24.643802647412755,
      "grad_norm": Infinity,
      "learning_rate": 0.00017253037281234897,
      "loss": 11.1652,
      "step": 204790
    },
    {
      "epoch": 24.645006016847173,
      "grad_norm": Infinity,
      "learning_rate": 0.00017252775379015324,
      "loss": 11.2104,
      "step": 204800
    },
    {
      "epoch": 24.646209386281587,
      "grad_norm": Infinity,
      "learning_rate": 0.00017252513466299197,
      "loss": 11.0869,
      "step": 204810
    },
    {
      "epoch": 24.647412755716005,
      "grad_norm": Infinity,
      "learning_rate": 0.00017252251543086887,
      "loss": 11.1367,
      "step": 204820
    },
    {
      "epoch": 24.648616125150422,
      "grad_norm": Infinity,
      "learning_rate": 0.00017251989609378782,
      "loss": 11.2158,
      "step": 204830
    },
    {
      "epoch": 24.649819494584836,
      "grad_norm": Infinity,
      "learning_rate": 0.00017251727665175255,
      "loss": 11.2252,
      "step": 204840
    },
    {
      "epoch": 24.651022864019254,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725146571047669,
      "loss": 11.1481,
      "step": 204850
    },
    {
      "epoch": 24.65222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725120374528346,
      "loss": 11.1409,
      "step": 204860
    },
    {
      "epoch": 24.653429602888085,
      "grad_norm": Infinity,
      "learning_rate": 0.00017250941769595955,
      "loss": 11.1508,
      "step": 204870
    },
    {
      "epoch": 24.654632972322503,
      "grad_norm": Infinity,
      "learning_rate": 0.0001725067978341454,
      "loss": 11.2665,
      "step": 204880
    },
    {
      "epoch": 24.65583634175692,
      "grad_norm": Infinity,
      "learning_rate": 0.00017250417786739605,
      "loss": 11.2422,
      "step": 204890
    },
    {
      "epoch": 24.657039711191334,
      "grad_norm": Infinity,
      "learning_rate": 0.00017250155779571525,
      "loss": 11.1574,
      "step": 204900
    },
    {
      "epoch": 24.658243080625752,
      "grad_norm": Infinity,
      "learning_rate": 0.00017249893761910677,
      "loss": 11.2169,
      "step": 204910
    },
    {
      "epoch": 24.65944645006017,
      "grad_norm": Infinity,
      "learning_rate": 0.00017249631733757444,
      "loss": 10.966,
      "step": 204920
    },
    {
      "epoch": 24.660649819494584,
      "grad_norm": Infinity,
      "learning_rate": 0.00017249369695112208,
      "loss": 11.1227,
      "step": 204930
    },
    {
      "epoch": 24.661853188929,
      "grad_norm": Infinity,
      "learning_rate": 0.0001724910764597534,
      "loss": 11.087,
      "step": 204940
    },
    {
      "epoch": 24.66305655836342,
      "grad_norm": Infinity,
      "learning_rate": 0.00017248845586347223,
      "loss": 11.2986,
      "step": 204950
    },
    {
      "epoch": 24.664259927797833,
      "grad_norm": Infinity,
      "learning_rate": 0.0001724858351622824,
      "loss": 11.0763,
      "step": 204960
    },
    {
      "epoch": 24.66546329723225,
      "grad_norm": Infinity,
      "learning_rate": 0.00017248321435618763,
      "loss": 11.0677,
      "step": 204970
    },
    {
      "epoch": 24.666666666666668,
      "grad_norm": Infinity,
      "learning_rate": 0.00017248059344519178,
      "loss": 11.077,
      "step": 204980
    },
    {
      "epoch": 24.667870036101082,
      "grad_norm": Infinity,
      "learning_rate": 0.00017247797242929863,
      "loss": 11.2168,
      "step": 204990
    },
    {
      "epoch": 24.6690734055355,
      "grad_norm": Infinity,
      "learning_rate": 0.00017247535130851194,
      "loss": 11.1452,
      "step": 205000
    },
    {
      "epoch": 24.670276774969917,
      "grad_norm": Infinity,
      "learning_rate": 0.00017247273008283553,
      "loss": 11.0917,
      "step": 205010
    },
    {
      "epoch": 24.67148014440433,
      "grad_norm": Infinity,
      "learning_rate": 0.00017247010875227317,
      "loss": 11.0937,
      "step": 205020
    },
    {
      "epoch": 24.67268351383875,
      "grad_norm": Infinity,
      "learning_rate": 0.0001724674873168287,
      "loss": 11.1087,
      "step": 205030
    },
    {
      "epoch": 24.673886883273166,
      "grad_norm": Infinity,
      "learning_rate": 0.00017246486577650587,
      "loss": 11.2592,
      "step": 205040
    },
    {
      "epoch": 24.67509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.00017246224413130845,
      "loss": 11.1693,
      "step": 205050
    },
    {
      "epoch": 24.676293622141998,
      "grad_norm": Infinity,
      "learning_rate": 0.00017245962238124035,
      "loss": 11.1687,
      "step": 205060
    },
    {
      "epoch": 24.677496991576415,
      "grad_norm": Infinity,
      "learning_rate": 0.00017245700052630523,
      "loss": 11.0862,
      "step": 205070
    },
    {
      "epoch": 24.67870036101083,
      "grad_norm": Infinity,
      "learning_rate": 0.00017245437856650694,
      "loss": 11.0493,
      "step": 205080
    },
    {
      "epoch": 24.679903730445247,
      "grad_norm": Infinity,
      "learning_rate": 0.00017245175650184932,
      "loss": 11.2379,
      "step": 205090
    },
    {
      "epoch": 24.681107099879664,
      "grad_norm": Infinity,
      "learning_rate": 0.00017244913433233606,
      "loss": 11.0883,
      "step": 205100
    },
    {
      "epoch": 24.68231046931408,
      "grad_norm": Infinity,
      "learning_rate": 0.00017244651205797107,
      "loss": 11.1097,
      "step": 205110
    },
    {
      "epoch": 24.683513838748496,
      "grad_norm": Infinity,
      "learning_rate": 0.00017244388967875804,
      "loss": 11.0599,
      "step": 205120
    },
    {
      "epoch": 24.684717208182914,
      "grad_norm": Infinity,
      "learning_rate": 0.00017244126719470084,
      "loss": 11.2164,
      "step": 205130
    },
    {
      "epoch": 24.685920577617328,
      "grad_norm": Infinity,
      "learning_rate": 0.00017243864460580327,
      "loss": 11.1807,
      "step": 205140
    },
    {
      "epoch": 24.687123947051745,
      "grad_norm": Infinity,
      "learning_rate": 0.00017243602191206904,
      "loss": 11.1286,
      "step": 205150
    },
    {
      "epoch": 24.688327316486163,
      "grad_norm": Infinity,
      "learning_rate": 0.00017243339911350203,
      "loss": 11.1287,
      "step": 205160
    },
    {
      "epoch": 24.689530685920577,
      "grad_norm": Infinity,
      "learning_rate": 0.000172430776210106,
      "loss": 11.126,
      "step": 205170
    },
    {
      "epoch": 24.690734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.0001724281532018848,
      "loss": 11.174,
      "step": 205180
    },
    {
      "epoch": 24.691937424789412,
      "grad_norm": Infinity,
      "learning_rate": 0.00017242553008884216,
      "loss": 11.2406,
      "step": 205190
    },
    {
      "epoch": 24.693140794223826,
      "grad_norm": Infinity,
      "learning_rate": 0.00017242290687098186,
      "loss": 11.1088,
      "step": 205200
    },
    {
      "epoch": 24.694344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.00017242028354830775,
      "loss": 11.1296,
      "step": 205210
    },
    {
      "epoch": 24.69554753309266,
      "grad_norm": Infinity,
      "learning_rate": 0.0001724176601208236,
      "loss": 11.2166,
      "step": 205220
    },
    {
      "epoch": 24.696750902527075,
      "grad_norm": Infinity,
      "learning_rate": 0.00017241503658853326,
      "loss": 11.1493,
      "step": 205230
    },
    {
      "epoch": 24.697954271961493,
      "grad_norm": Infinity,
      "learning_rate": 0.00017241241295144047,
      "loss": 11.2224,
      "step": 205240
    },
    {
      "epoch": 24.69915764139591,
      "grad_norm": Infinity,
      "learning_rate": 0.000172409789209549,
      "loss": 11.1077,
      "step": 205250
    },
    {
      "epoch": 24.700361010830324,
      "grad_norm": Infinity,
      "learning_rate": 0.00017240716536286272,
      "loss": 11.0537,
      "step": 205260
    },
    {
      "epoch": 24.70156438026474,
      "grad_norm": Infinity,
      "learning_rate": 0.00017240454141138537,
      "loss": 11.1137,
      "step": 205270
    },
    {
      "epoch": 24.70276774969916,
      "grad_norm": Infinity,
      "learning_rate": 0.00017240191735512083,
      "loss": 11.1246,
      "step": 205280
    },
    {
      "epoch": 24.703971119133573,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723992931940728,
      "loss": 11.0609,
      "step": 205290
    },
    {
      "epoch": 24.70517448856799,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723966689282451,
      "loss": 11.1098,
      "step": 205300
    },
    {
      "epoch": 24.706377858002405,
      "grad_norm": Infinity,
      "learning_rate": 0.00017239404455764157,
      "loss": 11.1885,
      "step": 205310
    },
    {
      "epoch": 24.707581227436823,
      "grad_norm": Infinity,
      "learning_rate": 0.000172391420082266,
      "loss": 11.1716,
      "step": 205320
    },
    {
      "epoch": 24.70878459687124,
      "grad_norm": Infinity,
      "learning_rate": 0.00017238879550212212,
      "loss": 11.1309,
      "step": 205330
    },
    {
      "epoch": 24.709987966305654,
      "grad_norm": Infinity,
      "learning_rate": 0.00017238617081721382,
      "loss": 11.1303,
      "step": 205340
    },
    {
      "epoch": 24.71119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.00017238354602754487,
      "loss": 11.1221,
      "step": 205350
    },
    {
      "epoch": 24.71239470517449,
      "grad_norm": Infinity,
      "learning_rate": 0.000172380921133119,
      "loss": 11.1669,
      "step": 205360
    },
    {
      "epoch": 24.713598074608903,
      "grad_norm": Infinity,
      "learning_rate": 0.00017237829613394015,
      "loss": 11.037,
      "step": 205370
    },
    {
      "epoch": 24.71480144404332,
      "grad_norm": Infinity,
      "learning_rate": 0.00017237567103001197,
      "loss": 11.1466,
      "step": 205380
    },
    {
      "epoch": 24.71600481347774,
      "grad_norm": Infinity,
      "learning_rate": 0.00017237304582133834,
      "loss": 11.2774,
      "step": 205390
    },
    {
      "epoch": 24.717208182912152,
      "grad_norm": Infinity,
      "learning_rate": 0.00017237042050792306,
      "loss": 11.186,
      "step": 205400
    },
    {
      "epoch": 24.71841155234657,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723677950897699,
      "loss": 11.2295,
      "step": 205410
    },
    {
      "epoch": 24.719614921780988,
      "grad_norm": Infinity,
      "learning_rate": 0.00017236516956688268,
      "loss": 11.126,
      "step": 205420
    },
    {
      "epoch": 24.7208182912154,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723625439392652,
      "loss": 11.0745,
      "step": 205430
    },
    {
      "epoch": 24.72202166064982,
      "grad_norm": Infinity,
      "learning_rate": 0.00017235991820692124,
      "loss": 11.1428,
      "step": 205440
    },
    {
      "epoch": 24.723225030084237,
      "grad_norm": Infinity,
      "learning_rate": 0.00017235729236985463,
      "loss": 11.1674,
      "step": 205450
    },
    {
      "epoch": 24.72442839951865,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723546664280691,
      "loss": 11.2736,
      "step": 205460
    },
    {
      "epoch": 24.72563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723520403815686,
      "loss": 10.9804,
      "step": 205470
    },
    {
      "epoch": 24.726835138387486,
      "grad_norm": Infinity,
      "learning_rate": 0.00017234941423035677,
      "loss": 11.2009,
      "step": 205480
    },
    {
      "epoch": 24.7280385078219,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723467879744375,
      "loss": 11.1039,
      "step": 205490
    },
    {
      "epoch": 24.729241877256317,
      "grad_norm": Infinity,
      "learning_rate": 0.00017234416161381454,
      "loss": 11.1822,
      "step": 205500
    },
    {
      "epoch": 24.730445246690735,
      "grad_norm": Infinity,
      "learning_rate": 0.00017234153514849172,
      "loss": 11.2578,
      "step": 205510
    },
    {
      "epoch": 24.73164861612515,
      "grad_norm": Infinity,
      "learning_rate": 0.00017233890857847286,
      "loss": 11.1938,
      "step": 205520
    },
    {
      "epoch": 24.732851985559567,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723362819037617,
      "loss": 11.1589,
      "step": 205530
    },
    {
      "epoch": 24.734055354993984,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723336551243621,
      "loss": 11.1152,
      "step": 205540
    },
    {
      "epoch": 24.735258724428398,
      "grad_norm": Infinity,
      "learning_rate": 0.00017233102824027786,
      "loss": 11.0362,
      "step": 205550
    },
    {
      "epoch": 24.736462093862816,
      "grad_norm": Infinity,
      "learning_rate": 0.00017232840125151278,
      "loss": 11.0834,
      "step": 205560
    },
    {
      "epoch": 24.737665463297233,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723257741580706,
      "loss": 11.1979,
      "step": 205570
    },
    {
      "epoch": 24.738868832731647,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723231469599552,
      "loss": 11.1567,
      "step": 205580
    },
    {
      "epoch": 24.740072202166065,
      "grad_norm": Infinity,
      "learning_rate": 0.00017232051965717033,
      "loss": 11.1166,
      "step": 205590
    },
    {
      "epoch": 24.741275571600482,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723178922497198,
      "loss": 11.0169,
      "step": 205600
    },
    {
      "epoch": 24.742478941034896,
      "grad_norm": Infinity,
      "learning_rate": 0.00017231526473760747,
      "loss": 11.1624,
      "step": 205610
    },
    {
      "epoch": 24.743682310469314,
      "grad_norm": Infinity,
      "learning_rate": 0.00017231263712083709,
      "loss": 11.0408,
      "step": 205620
    },
    {
      "epoch": 24.74488567990373,
      "grad_norm": Infinity,
      "learning_rate": 0.0001723100093994124,
      "loss": 11.2233,
      "step": 205630
    },
    {
      "epoch": 24.746089049338146,
      "grad_norm": Infinity,
      "learning_rate": 0.00017230738157333737,
      "loss": 11.1705,
      "step": 205640
    },
    {
      "epoch": 24.747292418772563,
      "grad_norm": Infinity,
      "learning_rate": 0.00017230475364261568,
      "loss": 11.0758,
      "step": 205650
    },
    {
      "epoch": 24.74849578820698,
      "grad_norm": Infinity,
      "learning_rate": 0.00017230212560725113,
      "loss": 11.023,
      "step": 205660
    },
    {
      "epoch": 24.749699157641395,
      "grad_norm": Infinity,
      "learning_rate": 0.00017229949746724757,
      "loss": 11.0784,
      "step": 205670
    },
    {
      "epoch": 24.750902527075812,
      "grad_norm": Infinity,
      "learning_rate": 0.00017229686922260877,
      "loss": 11.1199,
      "step": 205680
    },
    {
      "epoch": 24.75210589651023,
      "grad_norm": Infinity,
      "learning_rate": 0.00017229424087333856,
      "loss": 11.1881,
      "step": 205690
    },
    {
      "epoch": 24.753309265944644,
      "grad_norm": Infinity,
      "learning_rate": 0.00017229161241944074,
      "loss": 11.1362,
      "step": 205700
    },
    {
      "epoch": 24.75451263537906,
      "grad_norm": Infinity,
      "learning_rate": 0.00017228898386091913,
      "loss": 11.0152,
      "step": 205710
    },
    {
      "epoch": 24.75571600481348,
      "grad_norm": Infinity,
      "learning_rate": 0.0001722863551977775,
      "loss": 11.1821,
      "step": 205720
    },
    {
      "epoch": 24.756919374247893,
      "grad_norm": Infinity,
      "learning_rate": 0.00017228372643001964,
      "loss": 11.095,
      "step": 205730
    },
    {
      "epoch": 24.75812274368231,
      "grad_norm": Infinity,
      "learning_rate": 0.00017228109755764942,
      "loss": 11.2645,
      "step": 205740
    },
    {
      "epoch": 24.759326113116728,
      "grad_norm": Infinity,
      "learning_rate": 0.00017227846858067057,
      "loss": 11.0042,
      "step": 205750
    },
    {
      "epoch": 24.760529482551142,
      "grad_norm": Infinity,
      "learning_rate": 0.00017227583949908695,
      "loss": 11.0449,
      "step": 205760
    },
    {
      "epoch": 24.76173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00017227321031290234,
      "loss": 11.2319,
      "step": 205770
    },
    {
      "epoch": 24.762936221419977,
      "grad_norm": Infinity,
      "learning_rate": 0.00017227058102212058,
      "loss": 11.1308,
      "step": 205780
    },
    {
      "epoch": 24.76413959085439,
      "grad_norm": Infinity,
      "learning_rate": 0.0001722679516267454,
      "loss": 11.1714,
      "step": 205790
    },
    {
      "epoch": 24.76534296028881,
      "grad_norm": Infinity,
      "learning_rate": 0.00017226532212678066,
      "loss": 11.0361,
      "step": 205800
    },
    {
      "epoch": 24.766546329723226,
      "grad_norm": Infinity,
      "learning_rate": 0.00017226269252223017,
      "loss": 11.2833,
      "step": 205810
    },
    {
      "epoch": 24.76774969915764,
      "grad_norm": Infinity,
      "learning_rate": 0.00017226006281309775,
      "loss": 11.0983,
      "step": 205820
    },
    {
      "epoch": 24.768953068592058,
      "grad_norm": Infinity,
      "learning_rate": 0.00017225743299938712,
      "loss": 11.1982,
      "step": 205830
    },
    {
      "epoch": 24.770156438026476,
      "grad_norm": Infinity,
      "learning_rate": 0.0001722548030811022,
      "loss": 11.2026,
      "step": 205840
    },
    {
      "epoch": 24.77135980746089,
      "grad_norm": Infinity,
      "learning_rate": 0.0001722521730582467,
      "loss": 11.1458,
      "step": 205850
    },
    {
      "epoch": 24.772563176895307,
      "grad_norm": Infinity,
      "learning_rate": 0.00017224954293082448,
      "loss": 11.2007,
      "step": 205860
    },
    {
      "epoch": 24.773766546329725,
      "grad_norm": Infinity,
      "learning_rate": 0.00017224691269883933,
      "loss": 11.1958,
      "step": 205870
    },
    {
      "epoch": 24.77496991576414,
      "grad_norm": Infinity,
      "learning_rate": 0.00017224428236229508,
      "loss": 11.1364,
      "step": 205880
    },
    {
      "epoch": 24.776173285198556,
      "grad_norm": Infinity,
      "learning_rate": 0.00017224165192119553,
      "loss": 11.0171,
      "step": 205890
    },
    {
      "epoch": 24.777376654632974,
      "grad_norm": Infinity,
      "learning_rate": 0.00017223902137554443,
      "loss": 11.0969,
      "step": 205900
    },
    {
      "epoch": 24.778580024067388,
      "grad_norm": Infinity,
      "learning_rate": 0.00017223639072534564,
      "loss": 11.071,
      "step": 205910
    },
    {
      "epoch": 24.779783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.00017223375997060298,
      "loss": 11.232,
      "step": 205920
    },
    {
      "epoch": 24.780986762936223,
      "grad_norm": Infinity,
      "learning_rate": 0.0001722311291113202,
      "loss": 11.0746,
      "step": 205930
    },
    {
      "epoch": 24.782190132370637,
      "grad_norm": Infinity,
      "learning_rate": 0.00017222849814750116,
      "loss": 11.0144,
      "step": 205940
    },
    {
      "epoch": 24.783393501805055,
      "grad_norm": Infinity,
      "learning_rate": 0.00017222586707914966,
      "loss": 11.1086,
      "step": 205950
    },
    {
      "epoch": 24.784596871239472,
      "grad_norm": Infinity,
      "learning_rate": 0.00017222323590626953,
      "loss": 11.2092,
      "step": 205960
    },
    {
      "epoch": 24.785800240673886,
      "grad_norm": Infinity,
      "learning_rate": 0.0001722206046288645,
      "loss": 11.1945,
      "step": 205970
    },
    {
      "epoch": 24.787003610108304,
      "grad_norm": Infinity,
      "learning_rate": 0.00017221797324693843,
      "loss": 11.0865,
      "step": 205980
    },
    {
      "epoch": 24.78820697954272,
      "grad_norm": Infinity,
      "learning_rate": 0.00017221534176049512,
      "loss": 11.1573,
      "step": 205990
    },
    {
      "epoch": 24.789410348977135,
      "grad_norm": Infinity,
      "learning_rate": 0.0001722127101695384,
      "loss": 11.1529,
      "step": 206000
    },
    {
      "epoch": 24.790613718411553,
      "grad_norm": Infinity,
      "learning_rate": 0.00017221007847407205,
      "loss": 11.1101,
      "step": 206010
    },
    {
      "epoch": 24.79181708784597,
      "grad_norm": Infinity,
      "learning_rate": 0.0001722074466740999,
      "loss": 11.0971,
      "step": 206020
    },
    {
      "epoch": 24.793020457280385,
      "grad_norm": Infinity,
      "learning_rate": 0.0001722048147696257,
      "loss": 11.2051,
      "step": 206030
    },
    {
      "epoch": 24.794223826714802,
      "grad_norm": Infinity,
      "learning_rate": 0.00017220218276065333,
      "loss": 11.1422,
      "step": 206040
    },
    {
      "epoch": 24.79542719614922,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721995506471866,
      "loss": 11.1363,
      "step": 206050
    },
    {
      "epoch": 24.796630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00017219691842922926,
      "loss": 11.1196,
      "step": 206060
    },
    {
      "epoch": 24.79783393501805,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721942861067852,
      "loss": 11.1301,
      "step": 206070
    },
    {
      "epoch": 24.799037304452465,
      "grad_norm": Infinity,
      "learning_rate": 0.00017219165367985813,
      "loss": 11.0421,
      "step": 206080
    },
    {
      "epoch": 24.800240673886883,
      "grad_norm": Infinity,
      "learning_rate": 0.00017218902114845197,
      "loss": 11.1288,
      "step": 206090
    },
    {
      "epoch": 24.8014440433213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017218638851257046,
      "loss": 11.0956,
      "step": 206100
    },
    {
      "epoch": 24.802647412755714,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721837557722174,
      "loss": 11.2543,
      "step": 206110
    },
    {
      "epoch": 24.803850782190132,
      "grad_norm": Infinity,
      "learning_rate": 0.00017218112292739663,
      "loss": 11.1714,
      "step": 206120
    },
    {
      "epoch": 24.80505415162455,
      "grad_norm": Infinity,
      "learning_rate": 0.00017217848997811195,
      "loss": 11.1211,
      "step": 206130
    },
    {
      "epoch": 24.806257521058964,
      "grad_norm": Infinity,
      "learning_rate": 0.00017217585692436718,
      "loss": 11.044,
      "step": 206140
    },
    {
      "epoch": 24.80746089049338,
      "grad_norm": Infinity,
      "learning_rate": 0.00017217322376616613,
      "loss": 10.9925,
      "step": 206150
    },
    {
      "epoch": 24.8086642599278,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721705905035126,
      "loss": 11.0859,
      "step": 206160
    },
    {
      "epoch": 24.809867629362213,
      "grad_norm": Infinity,
      "learning_rate": 0.00017216795713641042,
      "loss": 11.3012,
      "step": 206170
    },
    {
      "epoch": 24.81107099879663,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721653236648634,
      "loss": 11.1771,
      "step": 206180
    },
    {
      "epoch": 24.812274368231048,
      "grad_norm": Infinity,
      "learning_rate": 0.00017216269008887532,
      "loss": 11.1421,
      "step": 206190
    },
    {
      "epoch": 24.813477737665462,
      "grad_norm": Infinity,
      "learning_rate": 0.00017216005640845,
      "loss": 11.0431,
      "step": 206200
    },
    {
      "epoch": 24.81468110709988,
      "grad_norm": Infinity,
      "learning_rate": 0.00017215742262359127,
      "loss": 11.1245,
      "step": 206210
    },
    {
      "epoch": 24.815884476534297,
      "grad_norm": Infinity,
      "learning_rate": 0.00017215478873430293,
      "loss": 11.143,
      "step": 206220
    },
    {
      "epoch": 24.81708784596871,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721521547405888,
      "loss": 11.0964,
      "step": 206230
    },
    {
      "epoch": 24.81829121540313,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721495206424527,
      "loss": 11.1142,
      "step": 206240
    },
    {
      "epoch": 24.819494584837546,
      "grad_norm": Infinity,
      "learning_rate": 0.00017214688643989843,
      "loss": 10.9466,
      "step": 206250
    },
    {
      "epoch": 24.82069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00017214425213292977,
      "loss": 11.2159,
      "step": 206260
    },
    {
      "epoch": 24.821901323706378,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721416177215506,
      "loss": 11.0813,
      "step": 206270
    },
    {
      "epoch": 24.823104693140795,
      "grad_norm": Infinity,
      "learning_rate": 0.00017213898320576467,
      "loss": 11.1662,
      "step": 206280
    },
    {
      "epoch": 24.82430806257521,
      "grad_norm": Infinity,
      "learning_rate": 0.00017213634858557586,
      "loss": 11.182,
      "step": 206290
    },
    {
      "epoch": 24.825511432009627,
      "grad_norm": Infinity,
      "learning_rate": 0.00017213371386098791,
      "loss": 11.1403,
      "step": 206300
    },
    {
      "epoch": 24.826714801444044,
      "grad_norm": Infinity,
      "learning_rate": 0.00017213107903200466,
      "loss": 11.1672,
      "step": 206310
    },
    {
      "epoch": 24.82791817087846,
      "grad_norm": Infinity,
      "learning_rate": 0.00017212844409862998,
      "loss": 11.102,
      "step": 206320
    },
    {
      "epoch": 24.829121540312876,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721258090608676,
      "loss": 11.0374,
      "step": 206330
    },
    {
      "epoch": 24.830324909747294,
      "grad_norm": Infinity,
      "learning_rate": 0.00017212317391872138,
      "loss": 11.1039,
      "step": 206340
    },
    {
      "epoch": 24.831528279181708,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721205386721951,
      "loss": 11.1424,
      "step": 206350
    },
    {
      "epoch": 24.832731648616125,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721179033212926,
      "loss": 11.2306,
      "step": 206360
    },
    {
      "epoch": 24.833935018050543,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721152678660177,
      "loss": 11.1831,
      "step": 206370
    },
    {
      "epoch": 24.835138387484957,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721126323063742,
      "loss": 11.0905,
      "step": 206380
    },
    {
      "epoch": 24.836341756919374,
      "grad_norm": Infinity,
      "learning_rate": 0.0001721099966423659,
      "loss": 11.1135,
      "step": 206390
    },
    {
      "epoch": 24.837545126353792,
      "grad_norm": Infinity,
      "learning_rate": 0.00017210736087399666,
      "loss": 11.1013,
      "step": 206400
    },
    {
      "epoch": 24.838748495788206,
      "grad_norm": Infinity,
      "learning_rate": 0.00017210472500127025,
      "loss": 11.274,
      "step": 206410
    },
    {
      "epoch": 24.839951865222623,
      "grad_norm": Infinity,
      "learning_rate": 0.00017210208902419049,
      "loss": 11.1311,
      "step": 206420
    },
    {
      "epoch": 24.84115523465704,
      "grad_norm": Infinity,
      "learning_rate": 0.00017209945294276123,
      "loss": 11.1534,
      "step": 206430
    },
    {
      "epoch": 24.842358604091455,
      "grad_norm": Infinity,
      "learning_rate": 0.00017209681675698624,
      "loss": 11.1009,
      "step": 206440
    },
    {
      "epoch": 24.843561973525873,
      "grad_norm": Infinity,
      "learning_rate": 0.00017209418046686934,
      "loss": 11.2222,
      "step": 206450
    },
    {
      "epoch": 24.84476534296029,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720915440724144,
      "loss": 11.178,
      "step": 206460
    },
    {
      "epoch": 24.845968712394704,
      "grad_norm": Infinity,
      "learning_rate": 0.00017208890757362518,
      "loss": 11.2219,
      "step": 206470
    },
    {
      "epoch": 24.84717208182912,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720862709705055,
      "loss": 11.0898,
      "step": 206480
    },
    {
      "epoch": 24.84837545126354,
      "grad_norm": Infinity,
      "learning_rate": 0.00017208363426305921,
      "loss": 11.2325,
      "step": 206490
    },
    {
      "epoch": 24.849578820697953,
      "grad_norm": Infinity,
      "learning_rate": 0.00017208099745129012,
      "loss": 11.1899,
      "step": 206500
    },
    {
      "epoch": 24.85078219013237,
      "grad_norm": Infinity,
      "learning_rate": 0.000172078360535202,
      "loss": 11.2145,
      "step": 206510
    },
    {
      "epoch": 24.85198555956679,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720757235147987,
      "loss": 11.1796,
      "step": 206520
    },
    {
      "epoch": 24.853188929001202,
      "grad_norm": Infinity,
      "learning_rate": 0.00017207308639008404,
      "loss": 11.1751,
      "step": 206530
    },
    {
      "epoch": 24.85439229843562,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720704491610618,
      "loss": 11.2955,
      "step": 206540
    },
    {
      "epoch": 24.855595667870038,
      "grad_norm": Infinity,
      "learning_rate": 0.00017206781182773587,
      "loss": 11.1943,
      "step": 206550
    },
    {
      "epoch": 24.85679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00017206517439011002,
      "loss": 11.1047,
      "step": 206560
    },
    {
      "epoch": 24.85800240673887,
      "grad_norm": Infinity,
      "learning_rate": 0.00017206253684818806,
      "loss": 11.1515,
      "step": 206570
    },
    {
      "epoch": 24.859205776173287,
      "grad_norm": Infinity,
      "learning_rate": 0.00017205989920197383,
      "loss": 11.0454,
      "step": 206580
    },
    {
      "epoch": 24.8604091456077,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720572614514711,
      "loss": 11.1421,
      "step": 206590
    },
    {
      "epoch": 24.86161251504212,
      "grad_norm": Infinity,
      "learning_rate": 0.00017205462359668375,
      "loss": 11.1458,
      "step": 206600
    },
    {
      "epoch": 24.862815884476536,
      "grad_norm": Infinity,
      "learning_rate": 0.00017205198563761558,
      "loss": 11.0541,
      "step": 206610
    },
    {
      "epoch": 24.86401925391095,
      "grad_norm": Infinity,
      "learning_rate": 0.00017204934757427036,
      "loss": 11.0889,
      "step": 206620
    },
    {
      "epoch": 24.865222623345367,
      "grad_norm": Infinity,
      "learning_rate": 0.00017204670940665196,
      "loss": 11.2054,
      "step": 206630
    },
    {
      "epoch": 24.866425992779785,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720440711347642,
      "loss": 11.1749,
      "step": 206640
    },
    {
      "epoch": 24.8676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.00017204143275861088,
      "loss": 11.1575,
      "step": 206650
    },
    {
      "epoch": 24.868832731648617,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720387942781958,
      "loss": 11.0526,
      "step": 206660
    },
    {
      "epoch": 24.870036101083034,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720361556935228,
      "loss": 11.1868,
      "step": 206670
    },
    {
      "epoch": 24.871239470517448,
      "grad_norm": Infinity,
      "learning_rate": 0.00017203351700459572,
      "loss": 11.2091,
      "step": 206680
    },
    {
      "epoch": 24.872442839951866,
      "grad_norm": Infinity,
      "learning_rate": 0.00017203087821141835,
      "loss": 11.0993,
      "step": 206690
    },
    {
      "epoch": 24.87364620938628,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720282393139945,
      "loss": 11.0359,
      "step": 206700
    },
    {
      "epoch": 24.874849578820697,
      "grad_norm": Infinity,
      "learning_rate": 0.00017202560031232803,
      "loss": 11.096,
      "step": 206710
    },
    {
      "epoch": 24.876052948255115,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720229612064227,
      "loss": 11.1901,
      "step": 206720
    },
    {
      "epoch": 24.87725631768953,
      "grad_norm": Infinity,
      "learning_rate": 0.00017202032199628236,
      "loss": 11.2294,
      "step": 206730
    },
    {
      "epoch": 24.878459687123947,
      "grad_norm": Infinity,
      "learning_rate": 0.00017201768268191087,
      "loss": 11.1205,
      "step": 206740
    },
    {
      "epoch": 24.879663056558364,
      "grad_norm": Infinity,
      "learning_rate": 0.00017201504326331199,
      "loss": 11.0705,
      "step": 206750
    },
    {
      "epoch": 24.880866425992778,
      "grad_norm": Infinity,
      "learning_rate": 0.0001720124037404896,
      "loss": 11.1353,
      "step": 206760
    },
    {
      "epoch": 24.882069795427196,
      "grad_norm": Infinity,
      "learning_rate": 0.00017200976411344744,
      "loss": 11.0561,
      "step": 206770
    },
    {
      "epoch": 24.883273164861613,
      "grad_norm": Infinity,
      "learning_rate": 0.00017200712438218937,
      "loss": 11.1002,
      "step": 206780
    },
    {
      "epoch": 24.884476534296027,
      "grad_norm": Infinity,
      "learning_rate": 0.00017200448454671925,
      "loss": 11.1291,
      "step": 206790
    },
    {
      "epoch": 24.885679903730445,
      "grad_norm": Infinity,
      "learning_rate": 0.00017200184460704082,
      "loss": 11.2082,
      "step": 206800
    },
    {
      "epoch": 24.886883273164862,
      "grad_norm": Infinity,
      "learning_rate": 0.000171999204563158,
      "loss": 11.1146,
      "step": 206810
    },
    {
      "epoch": 24.888086642599276,
      "grad_norm": Infinity,
      "learning_rate": 0.0001719965644150745,
      "loss": 11.1087,
      "step": 206820
    },
    {
      "epoch": 24.889290012033694,
      "grad_norm": Infinity,
      "learning_rate": 0.00017199392416279423,
      "loss": 11.2525,
      "step": 206830
    },
    {
      "epoch": 24.89049338146811,
      "grad_norm": Infinity,
      "learning_rate": 0.00017199128380632095,
      "loss": 11.1837,
      "step": 206840
    },
    {
      "epoch": 24.891696750902526,
      "grad_norm": Infinity,
      "learning_rate": 0.00017198864334565852,
      "loss": 11.148,
      "step": 206850
    },
    {
      "epoch": 24.892900120336943,
      "grad_norm": Infinity,
      "learning_rate": 0.00017198600278081077,
      "loss": 11.1606,
      "step": 206860
    },
    {
      "epoch": 24.89410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017198336211178146,
      "loss": 11.1392,
      "step": 206870
    },
    {
      "epoch": 24.895306859205775,
      "grad_norm": Infinity,
      "learning_rate": 0.00017198072133857445,
      "loss": 11.1581,
      "step": 206880
    },
    {
      "epoch": 24.896510228640192,
      "grad_norm": Infinity,
      "learning_rate": 0.0001719780804611936,
      "loss": 11.1352,
      "step": 206890
    },
    {
      "epoch": 24.89771359807461,
      "grad_norm": Infinity,
      "learning_rate": 0.0001719754394796427,
      "loss": 11.1719,
      "step": 206900
    },
    {
      "epoch": 24.898916967509024,
      "grad_norm": Infinity,
      "learning_rate": 0.00017197279839392557,
      "loss": 11.1446,
      "step": 206910
    },
    {
      "epoch": 24.90012033694344,
      "grad_norm": Infinity,
      "learning_rate": 0.000171970157204046,
      "loss": 11.0143,
      "step": 206920
    },
    {
      "epoch": 24.90132370637786,
      "grad_norm": Infinity,
      "learning_rate": 0.00017196751591000786,
      "loss": 11.2199,
      "step": 206930
    },
    {
      "epoch": 24.902527075812273,
      "grad_norm": Infinity,
      "learning_rate": 0.00017196487451181494,
      "loss": 11.2317,
      "step": 206940
    },
    {
      "epoch": 24.90373044524669,
      "grad_norm": Infinity,
      "learning_rate": 0.00017196223300947112,
      "loss": 11.1998,
      "step": 206950
    },
    {
      "epoch": 24.904933814681108,
      "grad_norm": Infinity,
      "learning_rate": 0.00017195959140298017,
      "loss": 11.2029,
      "step": 206960
    },
    {
      "epoch": 24.906137184115522,
      "grad_norm": Infinity,
      "learning_rate": 0.0001719569496923459,
      "loss": 11.12,
      "step": 206970
    },
    {
      "epoch": 24.90734055354994,
      "grad_norm": Infinity,
      "learning_rate": 0.0001719543078775722,
      "loss": 11.0668,
      "step": 206980
    },
    {
      "epoch": 24.908543922984357,
      "grad_norm": Infinity,
      "learning_rate": 0.00017195166595866278,
      "loss": 11.0521,
      "step": 206990
    },
    {
      "epoch": 24.90974729241877,
      "grad_norm": Infinity,
      "learning_rate": 0.00017194902393562157,
      "loss": 11.1313,
      "step": 207000
    },
    {
      "epoch": 24.91095066185319,
      "grad_norm": Infinity,
      "learning_rate": 0.0001719463818084524,
      "loss": 11.0913,
      "step": 207010
    },
    {
      "epoch": 24.912154031287606,
      "grad_norm": Infinity,
      "learning_rate": 0.00017194373957715899,
      "loss": 11.1522,
      "step": 207020
    },
    {
      "epoch": 24.91335740072202,
      "grad_norm": Infinity,
      "learning_rate": 0.00017194109724174528,
      "loss": 11.0622,
      "step": 207030
    },
    {
      "epoch": 24.914560770156438,
      "grad_norm": Infinity,
      "learning_rate": 0.00017193845480221502,
      "loss": 11.1755,
      "step": 207040
    },
    {
      "epoch": 24.915764139590856,
      "grad_norm": Infinity,
      "learning_rate": 0.00017193581225857205,
      "loss": 11.1641,
      "step": 207050
    },
    {
      "epoch": 24.91696750902527,
      "grad_norm": Infinity,
      "learning_rate": 0.0001719331696108202,
      "loss": 11.2042,
      "step": 207060
    },
    {
      "epoch": 24.918170878459687,
      "grad_norm": Infinity,
      "learning_rate": 0.00017193052685896327,
      "loss": 11.1643,
      "step": 207070
    },
    {
      "epoch": 24.919374247894105,
      "grad_norm": Infinity,
      "learning_rate": 0.00017192788400300515,
      "loss": 11.2261,
      "step": 207080
    },
    {
      "epoch": 24.92057761732852,
      "grad_norm": Infinity,
      "learning_rate": 0.00017192524104294962,
      "loss": 11.1804,
      "step": 207090
    },
    {
      "epoch": 24.921780986762936,
      "grad_norm": Infinity,
      "learning_rate": 0.00017192259797880048,
      "loss": 11.1159,
      "step": 207100
    },
    {
      "epoch": 24.922984356197354,
      "grad_norm": Infinity,
      "learning_rate": 0.00017191995481056157,
      "loss": 11.1054,
      "step": 207110
    },
    {
      "epoch": 24.924187725631768,
      "grad_norm": Infinity,
      "learning_rate": 0.00017191731153823678,
      "loss": 11.0783,
      "step": 207120
    },
    {
      "epoch": 24.925391095066185,
      "grad_norm": Infinity,
      "learning_rate": 0.00017191466816182985,
      "loss": 11.1987,
      "step": 207130
    },
    {
      "epoch": 24.926594464500603,
      "grad_norm": Infinity,
      "learning_rate": 0.00017191202468134467,
      "loss": 11.2727,
      "step": 207140
    },
    {
      "epoch": 24.927797833935017,
      "grad_norm": Infinity,
      "learning_rate": 0.000171909381096785,
      "loss": 11.1519,
      "step": 207150
    },
    {
      "epoch": 24.929001203369435,
      "grad_norm": Infinity,
      "learning_rate": 0.00017190673740815474,
      "loss": 11.1968,
      "step": 207160
    },
    {
      "epoch": 24.930204572803852,
      "grad_norm": Infinity,
      "learning_rate": 0.00017190409361545763,
      "loss": 11.1972,
      "step": 207170
    },
    {
      "epoch": 24.931407942238266,
      "grad_norm": Infinity,
      "learning_rate": 0.00017190144971869758,
      "loss": 11.0516,
      "step": 207180
    },
    {
      "epoch": 24.932611311672684,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718988057178784,
      "loss": 11.0858,
      "step": 207190
    },
    {
      "epoch": 24.9338146811071,
      "grad_norm": Infinity,
      "learning_rate": 0.00017189616161300385,
      "loss": 11.1439,
      "step": 207200
    },
    {
      "epoch": 24.935018050541515,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718935174040778,
      "loss": 11.3689,
      "step": 207210
    },
    {
      "epoch": 24.936221419975933,
      "grad_norm": Infinity,
      "learning_rate": 0.00017189087309110412,
      "loss": 11.0715,
      "step": 207220
    },
    {
      "epoch": 24.93742478941035,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718882286740866,
      "loss": 11.164,
      "step": 207230
    },
    {
      "epoch": 24.938628158844764,
      "grad_norm": Infinity,
      "learning_rate": 0.00017188558415302902,
      "loss": 11.0828,
      "step": 207240
    },
    {
      "epoch": 24.939831528279182,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718829395279353,
      "loss": 11.1284,
      "step": 207250
    },
    {
      "epoch": 24.9410348977136,
      "grad_norm": Infinity,
      "learning_rate": 0.00017188029479880918,
      "loss": 11.0516,
      "step": 207260
    },
    {
      "epoch": 24.942238267148014,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718776499656545,
      "loss": 11.1271,
      "step": 207270
    },
    {
      "epoch": 24.94344163658243,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718750050284752,
      "loss": 11.1813,
      "step": 207280
    },
    {
      "epoch": 24.94464500601685,
      "grad_norm": Infinity,
      "learning_rate": 0.00017187235998727499,
      "loss": 11.0852,
      "step": 207290
    },
    {
      "epoch": 24.945848375451263,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718697148420577,
      "loss": 11.0204,
      "step": 207300
    },
    {
      "epoch": 24.94705174488568,
      "grad_norm": Infinity,
      "learning_rate": 0.00017186706959282721,
      "loss": 11.0849,
      "step": 207310
    },
    {
      "epoch": 24.948255114320098,
      "grad_norm": Infinity,
      "learning_rate": 0.00017186442423958731,
      "loss": 11.1137,
      "step": 207320
    },
    {
      "epoch": 24.949458483754512,
      "grad_norm": Infinity,
      "learning_rate": 0.00017186177878234188,
      "loss": 11.1212,
      "step": 207330
    },
    {
      "epoch": 24.95066185318893,
      "grad_norm": Infinity,
      "learning_rate": 0.00017185913322109467,
      "loss": 11.1826,
      "step": 207340
    },
    {
      "epoch": 24.951865222623347,
      "grad_norm": Infinity,
      "learning_rate": 0.00017185648755584958,
      "loss": 11.1985,
      "step": 207350
    },
    {
      "epoch": 24.95306859205776,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718538417866104,
      "loss": 11.0842,
      "step": 207360
    },
    {
      "epoch": 24.95427196149218,
      "grad_norm": Infinity,
      "learning_rate": 0.000171851195913381,
      "loss": 11.138,
      "step": 207370
    },
    {
      "epoch": 24.955475330926596,
      "grad_norm": Infinity,
      "learning_rate": 0.00017184854993616515,
      "loss": 11.1469,
      "step": 207380
    },
    {
      "epoch": 24.95667870036101,
      "grad_norm": Infinity,
      "learning_rate": 0.00017184590385496673,
      "loss": 11.2099,
      "step": 207390
    },
    {
      "epoch": 24.957882069795428,
      "grad_norm": Infinity,
      "learning_rate": 0.00017184325766978953,
      "loss": 11.1662,
      "step": 207400
    },
    {
      "epoch": 24.959085439229845,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718406113806374,
      "loss": 11.1903,
      "step": 207410
    },
    {
      "epoch": 24.96028880866426,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718379649875142,
      "loss": 11.1444,
      "step": 207420
    },
    {
      "epoch": 24.961492178098677,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718353184904237,
      "loss": 11.1144,
      "step": 207430
    },
    {
      "epoch": 24.96269554753309,
      "grad_norm": Infinity,
      "learning_rate": 0.00017183267188936973,
      "loss": 11.2589,
      "step": 207440
    },
    {
      "epoch": 24.96389891696751,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718300251843562,
      "loss": 11.1186,
      "step": 207450
    },
    {
      "epoch": 24.965102286401926,
      "grad_norm": Infinity,
      "learning_rate": 0.00017182737837538684,
      "loss": 11.0918,
      "step": 207460
    },
    {
      "epoch": 24.96630565583634,
      "grad_norm": Infinity,
      "learning_rate": 0.00017182473146246558,
      "loss": 11.1233,
      "step": 207470
    },
    {
      "epoch": 24.967509025270758,
      "grad_norm": Infinity,
      "learning_rate": 0.00017182208444559618,
      "loss": 11.0938,
      "step": 207480
    },
    {
      "epoch": 24.968712394705175,
      "grad_norm": Infinity,
      "learning_rate": 0.00017181943732478248,
      "loss": 11.2075,
      "step": 207490
    },
    {
      "epoch": 24.96991576413959,
      "grad_norm": Infinity,
      "learning_rate": 0.00017181679010002835,
      "loss": 11.1433,
      "step": 207500
    },
    {
      "epoch": 24.971119133574007,
      "grad_norm": Infinity,
      "learning_rate": 0.00017181414277133756,
      "loss": 11.2439,
      "step": 207510
    },
    {
      "epoch": 24.972322503008424,
      "grad_norm": Infinity,
      "learning_rate": 0.00017181149533871401,
      "loss": 11.0801,
      "step": 207520
    },
    {
      "epoch": 24.97352587244284,
      "grad_norm": Infinity,
      "learning_rate": 0.00017180884780216147,
      "loss": 11.1113,
      "step": 207530
    },
    {
      "epoch": 24.974729241877256,
      "grad_norm": Infinity,
      "learning_rate": 0.00017180620016168382,
      "loss": 11.2429,
      "step": 207540
    },
    {
      "epoch": 24.975932611311674,
      "grad_norm": Infinity,
      "learning_rate": 0.00017180355241728484,
      "loss": 11.1875,
      "step": 207550
    },
    {
      "epoch": 24.977135980746088,
      "grad_norm": Infinity,
      "learning_rate": 0.0001718009045689684,
      "loss": 11.1739,
      "step": 207560
    },
    {
      "epoch": 24.978339350180505,
      "grad_norm": Infinity,
      "learning_rate": 0.00017179825661673833,
      "loss": 11.1301,
      "step": 207570
    },
    {
      "epoch": 24.979542719614923,
      "grad_norm": Infinity,
      "learning_rate": 0.00017179560856059846,
      "loss": 11.2103,
      "step": 207580
    },
    {
      "epoch": 24.980746089049337,
      "grad_norm": Infinity,
      "learning_rate": 0.00017179296040055264,
      "loss": 11.1949,
      "step": 207590
    },
    {
      "epoch": 24.981949458483754,
      "grad_norm": Infinity,
      "learning_rate": 0.00017179031213660466,
      "loss": 11.1181,
      "step": 207600
    },
    {
      "epoch": 24.983152827918172,
      "grad_norm": Infinity,
      "learning_rate": 0.00017178766376875837,
      "loss": 11.145,
      "step": 207610
    },
    {
      "epoch": 24.984356197352586,
      "grad_norm": Infinity,
      "learning_rate": 0.0001717850152970176,
      "loss": 11.1285,
      "step": 207620
    },
    {
      "epoch": 24.985559566787003,
      "grad_norm": Infinity,
      "learning_rate": 0.0001717823667213862,
      "loss": 11.1754,
      "step": 207630
    },
    {
      "epoch": 24.98676293622142,
      "grad_norm": Infinity,
      "learning_rate": 0.000171779718041868,
      "loss": 11.174,
      "step": 207640
    },
    {
      "epoch": 24.987966305655835,
      "grad_norm": Infinity,
      "learning_rate": 0.00017177706925846682,
      "loss": 11.1765,
      "step": 207650
    },
    {
      "epoch": 24.989169675090253,
      "grad_norm": Infinity,
      "learning_rate": 0.0001717744203711865,
      "loss": 11.0419,
      "step": 207660
    },
    {
      "epoch": 24.99037304452467,
      "grad_norm": Infinity,
      "learning_rate": 0.00017177177138003087,
      "loss": 11.1228,
      "step": 207670
    },
    {
      "epoch": 24.991576413959084,
      "grad_norm": Infinity,
      "learning_rate": 0.00017176912228500377,
      "loss": 11.2689,
      "step": 207680
    },
    {
      "epoch": 24.9927797833935,
      "grad_norm": Infinity,
      "learning_rate": 0.00017176647308610906,
      "loss": 11.2195,
      "step": 207690
    },
    {
      "epoch": 24.99398315282792,
      "grad_norm": Infinity,
      "learning_rate": 0.0001717638237833505,
      "loss": 11.0816,
      "step": 207700
    },
    {
      "epoch": 24.995186522262333,
      "grad_norm": Infinity,
      "learning_rate": 0.00017176117437673195,
      "loss": 11.1939,
      "step": 207710
    },
    {
      "epoch": 24.99638989169675,
      "grad_norm": Infinity,
      "learning_rate": 0.00017175852486625733,
      "loss": 11.0975,
      "step": 207720
    },
    {
      "epoch": 24.99759326113117,
      "grad_norm": Infinity,
      "learning_rate": 0.00017175587525193036,
      "loss": 11.0899,
      "step": 207730
    },
    {
      "epoch": 24.998796630565582,
      "grad_norm": Infinity,
      "learning_rate": 0.00017175322553375494,
      "loss": 11.1543,
      "step": 207740
    },
    {
      "epoch": 25.0,
      "grad_norm": Infinity,
      "learning_rate": 0.0001717505757117349,
      "loss": 11.0606,
      "step": 207750
    },
    {
      "epoch": 25.0,
      "eval_loss": 11.139603614807129,
      "eval_runtime": 119.5071,
      "eval_samples_per_second": 61.812,
      "eval_steps_per_second": 7.732,
      "step": 207750
    }
  ],
  "logging_steps": 10,
  "max_steps": 831000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 100,
  "save_steps": 4155,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.3680176217994035e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}