{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 50.0,
  "eval_steps": 500,
  "global_step": 238700,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.10473397570171764,
      "grad_norm": 1.1384308338165283,
      "learning_rate": 4.989526602429829e-05,
      "loss": 8.9672,
      "step": 500
    },
    {
      "epoch": 0.20946795140343527,
      "grad_norm": 0.9985808730125427,
      "learning_rate": 4.979053204859657e-05,
      "loss": 7.7253,
      "step": 1000
    },
    {
      "epoch": 0.31420192710515293,
      "grad_norm": 1.0521348714828491,
      "learning_rate": 4.968579807289485e-05,
      "loss": 7.5614,
      "step": 1500
    },
    {
      "epoch": 0.41893590280687054,
      "grad_norm": 1.0602227449417114,
      "learning_rate": 4.958106409719313e-05,
      "loss": 7.5037,
      "step": 2000
    },
    {
      "epoch": 0.5236698785085881,
      "grad_norm": 1.6102268695831299,
      "learning_rate": 4.9476330121491414e-05,
      "loss": 7.4595,
      "step": 2500
    },
    {
      "epoch": 0.6284038542103059,
      "grad_norm": 1.335976004600525,
      "learning_rate": 4.9371596145789694e-05,
      "loss": 7.4267,
      "step": 3000
    },
    {
      "epoch": 0.7331378299120235,
      "grad_norm": 1.340728998184204,
      "learning_rate": 4.926686217008798e-05,
      "loss": 7.392,
      "step": 3500
    },
    {
      "epoch": 0.8378718056137411,
      "grad_norm": 1.4520059823989868,
      "learning_rate": 4.916212819438626e-05,
      "loss": 7.3253,
      "step": 4000
    },
    {
      "epoch": 0.9426057813154587,
      "grad_norm": 1.7685532569885254,
      "learning_rate": 4.905760368663595e-05,
      "loss": 7.3204,
      "step": 4500
    },
    {
      "epoch": 1.0473397570171763,
      "grad_norm": 1.396130084991455,
      "learning_rate": 4.8952869710934226e-05,
      "loss": 7.2682,
      "step": 5000
    },
    {
      "epoch": 1.1520737327188941,
      "grad_norm": 1.5962079763412476,
      "learning_rate": 4.884813573523251e-05,
      "loss": 7.2441,
      "step": 5500
    },
    {
      "epoch": 1.2568077084206117,
      "grad_norm": 1.6328166723251343,
      "learning_rate": 4.874340175953079e-05,
      "loss": 7.2216,
      "step": 6000
    },
    {
      "epoch": 1.3615416841223293,
      "grad_norm": 1.8534362316131592,
      "learning_rate": 4.863887725178048e-05,
      "loss": 7.1957,
      "step": 6500
    },
    {
      "epoch": 1.466275659824047,
      "grad_norm": 1.4871692657470703,
      "learning_rate": 4.8534143276078766e-05,
      "loss": 7.1561,
      "step": 7000
    },
    {
      "epoch": 1.5710096355257646,
      "grad_norm": 1.8590672016143799,
      "learning_rate": 4.8429409300377045e-05,
      "loss": 7.1397,
      "step": 7500
    },
    {
      "epoch": 1.6757436112274822,
      "grad_norm": 1.7009446620941162,
      "learning_rate": 4.8324675324675325e-05,
      "loss": 7.1149,
      "step": 8000
    },
    {
      "epoch": 1.7804775869291998,
      "grad_norm": 1.9020010232925415,
      "learning_rate": 4.822015081692501e-05,
      "loss": 7.1215,
      "step": 8500
    },
    {
      "epoch": 1.8852115626309174,
      "grad_norm": 2.912442445755005,
      "learning_rate": 4.811541684122329e-05,
      "loss": 7.0616,
      "step": 9000
    },
    {
      "epoch": 1.989945538332635,
      "grad_norm": 2.654263496398926,
      "learning_rate": 4.801068286552158e-05,
      "loss": 7.0508,
      "step": 9500
    },
    {
      "epoch": 2.0946795140343526,
      "grad_norm": 2.1642003059387207,
      "learning_rate": 4.790594888981986e-05,
      "loss": 7.0251,
      "step": 10000
    },
    {
      "epoch": 2.19941348973607,
      "grad_norm": 1.9420874118804932,
      "learning_rate": 4.7801424382069544e-05,
      "loss": 6.9806,
      "step": 10500
    },
    {
      "epoch": 2.3041474654377883,
      "grad_norm": 2.2306201457977295,
      "learning_rate": 4.769669040636783e-05,
      "loss": 6.9721,
      "step": 11000
    },
    {
      "epoch": 2.4088814411395054,
      "grad_norm": 2.8180389404296875,
      "learning_rate": 4.759195643066611e-05,
      "loss": 6.9582,
      "step": 11500
    },
    {
      "epoch": 2.5136154168412235,
      "grad_norm": 2.387949228286743,
      "learning_rate": 4.748722245496439e-05,
      "loss": 6.9211,
      "step": 12000
    },
    {
      "epoch": 2.618349392542941,
      "grad_norm": 3.3709394931793213,
      "learning_rate": 4.7382697947214076e-05,
      "loss": 6.9183,
      "step": 12500
    },
    {
      "epoch": 2.7230833682446587,
      "grad_norm": 2.567798376083374,
      "learning_rate": 4.727796397151236e-05,
      "loss": 6.8732,
      "step": 13000
    },
    {
      "epoch": 2.8278173439463763,
      "grad_norm": 2.6373414993286133,
      "learning_rate": 4.717322999581064e-05,
      "loss": 6.8658,
      "step": 13500
    },
    {
      "epoch": 2.932551319648094,
      "grad_norm": 2.2950875759124756,
      "learning_rate": 4.706849602010893e-05,
      "loss": 6.8436,
      "step": 14000
    },
    {
      "epoch": 3.0372852953498115,
      "grad_norm": 4.0021514892578125,
      "learning_rate": 4.696397151235861e-05,
      "loss": 6.8496,
      "step": 14500
    },
    {
      "epoch": 3.142019271051529,
      "grad_norm": 3.289193630218506,
      "learning_rate": 4.685923753665689e-05,
      "loss": 6.8047,
      "step": 15000
    },
    {
      "epoch": 3.2467532467532467,
      "grad_norm": 2.9973654747009277,
      "learning_rate": 4.6754503560955175e-05,
      "loss": 6.7651,
      "step": 15500
    },
    {
      "epoch": 3.3514872224549643,
      "grad_norm": 2.9979376792907715,
      "learning_rate": 4.664976958525346e-05,
      "loss": 6.7768,
      "step": 16000
    },
    {
      "epoch": 3.456221198156682,
      "grad_norm": 3.263784885406494,
      "learning_rate": 4.654524507750315e-05,
      "loss": 6.7617,
      "step": 16500
    },
    {
      "epoch": 3.5609551738583995,
      "grad_norm": 3.330116033554077,
      "learning_rate": 4.644051110180143e-05,
      "loss": 6.7417,
      "step": 17000
    },
    {
      "epoch": 3.665689149560117,
      "grad_norm": 3.224337339401245,
      "learning_rate": 4.6335777126099714e-05,
      "loss": 6.7028,
      "step": 17500
    },
    {
      "epoch": 3.7704231252618348,
      "grad_norm": 3.21891450881958,
      "learning_rate": 4.623104315039799e-05,
      "loss": 6.7101,
      "step": 18000
    },
    {
      "epoch": 3.875157100963553,
      "grad_norm": 2.3559324741363525,
      "learning_rate": 4.6126518642647674e-05,
      "loss": 6.6892,
      "step": 18500
    },
    {
      "epoch": 3.97989107666527,
      "grad_norm": 3.1527633666992188,
      "learning_rate": 4.602178466694596e-05,
      "loss": 6.6851,
      "step": 19000
    },
    {
      "epoch": 4.084625052366988,
      "grad_norm": 2.9760189056396484,
      "learning_rate": 4.591705069124424e-05,
      "loss": 6.6704,
      "step": 19500
    },
    {
      "epoch": 4.189359028068705,
      "grad_norm": 2.8135318756103516,
      "learning_rate": 4.5812316715542526e-05,
      "loss": 6.6621,
      "step": 20000
    },
    {
      "epoch": 4.294093003770423,
      "grad_norm": 3.060316324234009,
      "learning_rate": 4.570779220779221e-05,
      "loss": 6.6298,
      "step": 20500
    },
    {
      "epoch": 4.39882697947214,
      "grad_norm": 2.7130279541015625,
      "learning_rate": 4.560305823209049e-05,
      "loss": 6.628,
      "step": 21000
    },
    {
      "epoch": 4.5035609551738585,
      "grad_norm": 3.156386613845825,
      "learning_rate": 4.549853372434017e-05,
      "loss": 6.6423,
      "step": 21500
    },
    {
      "epoch": 4.6082949308755765,
      "grad_norm": 3.039471387863159,
      "learning_rate": 4.539379974863846e-05,
      "loss": 6.6343,
      "step": 22000
    },
    {
      "epoch": 4.713028906577294,
      "grad_norm": 3.976949453353882,
      "learning_rate": 4.5289065772936745e-05,
      "loss": 6.606,
      "step": 22500
    },
    {
      "epoch": 4.817762882279011,
      "grad_norm": 3.310382604598999,
      "learning_rate": 4.5184331797235025e-05,
      "loss": 6.5956,
      "step": 23000
    },
    {
      "epoch": 4.922496857980729,
      "grad_norm": 3.5924322605133057,
      "learning_rate": 4.507959782153331e-05,
      "loss": 6.5965,
      "step": 23500
    },
    {
      "epoch": 5.027230833682447,
      "grad_norm": 2.616468667984009,
      "learning_rate": 4.497486384583159e-05,
      "loss": 6.5788,
      "step": 24000
    },
    {
      "epoch": 5.131964809384164,
      "grad_norm": 3.3178062438964844,
      "learning_rate": 4.487012987012987e-05,
      "loss": 6.5684,
      "step": 24500
    },
    {
      "epoch": 5.236698785085882,
      "grad_norm": 3.7108089923858643,
      "learning_rate": 4.476539589442815e-05,
      "loss": 6.5756,
      "step": 25000
    },
    {
      "epoch": 5.341432760787599,
      "grad_norm": 3.396498918533325,
      "learning_rate": 4.466087138667784e-05,
      "loss": 6.5678,
      "step": 25500
    },
    {
      "epoch": 5.446166736489317,
      "grad_norm": 3.7245748043060303,
      "learning_rate": 4.4556137410976123e-05,
      "loss": 6.5578,
      "step": 26000
    },
    {
      "epoch": 5.5509007121910345,
      "grad_norm": 3.6525135040283203,
      "learning_rate": 4.44514034352744e-05,
      "loss": 6.5385,
      "step": 26500
    },
    {
      "epoch": 5.655634687892753,
      "grad_norm": 3.4302523136138916,
      "learning_rate": 4.434666945957269e-05,
      "loss": 6.5143,
      "step": 27000
    },
    {
      "epoch": 5.76036866359447,
      "grad_norm": 3.762871503829956,
      "learning_rate": 4.4242144951822376e-05,
      "loss": 6.52,
      "step": 27500
    },
    {
      "epoch": 5.865102639296188,
      "grad_norm": 2.8195388317108154,
      "learning_rate": 4.4137410976120656e-05,
      "loss": 6.5213,
      "step": 28000
    },
    {
      "epoch": 5.969836614997905,
      "grad_norm": 3.013187885284424,
      "learning_rate": 4.4032677000418936e-05,
      "loss": 6.5052,
      "step": 28500
    },
    {
      "epoch": 6.074570590699623,
      "grad_norm": 2.9772274494171143,
      "learning_rate": 4.392794302471722e-05,
      "loss": 6.502,
      "step": 29000
    },
    {
      "epoch": 6.17930456640134,
      "grad_norm": 3.2228713035583496,
      "learning_rate": 4.38232090490155e-05,
      "loss": 6.4889,
      "step": 29500
    },
    {
      "epoch": 6.284038542103058,
      "grad_norm": 3.824286937713623,
      "learning_rate": 4.371847507331379e-05,
      "loss": 6.4792,
      "step": 30000
    },
    {
      "epoch": 6.388772517804776,
      "grad_norm": 3.100308656692505,
      "learning_rate": 4.361374109761207e-05,
      "loss": 6.4816,
      "step": 30500
    },
    {
      "epoch": 6.4935064935064934,
      "grad_norm": 3.4449245929718018,
      "learning_rate": 4.350900712191035e-05,
      "loss": 6.4786,
      "step": 31000
    },
    {
      "epoch": 6.5982404692082115,
      "grad_norm": 3.6803085803985596,
      "learning_rate": 4.3404482614160034e-05,
      "loss": 6.4778,
      "step": 31500
    },
    {
      "epoch": 6.702974444909929,
      "grad_norm": 3.6413722038269043,
      "learning_rate": 4.329974863845832e-05,
      "loss": 6.4782,
      "step": 32000
    },
    {
      "epoch": 6.807708420611647,
      "grad_norm": 3.482905626296997,
      "learning_rate": 4.31950146627566e-05,
      "loss": 6.4719,
      "step": 32500
    },
    {
      "epoch": 6.912442396313364,
      "grad_norm": 3.4605376720428467,
      "learning_rate": 4.3090280687054887e-05,
      "loss": 6.4458,
      "step": 33000
    },
    {
      "epoch": 7.017176372015082,
      "grad_norm": 3.814375877380371,
      "learning_rate": 4.2985965647255974e-05,
      "loss": 6.4539,
      "step": 33500
    },
    {
      "epoch": 7.121910347716799,
      "grad_norm": 4.238844871520996,
      "learning_rate": 4.288123167155425e-05,
      "loss": 6.4522,
      "step": 34000
    },
    {
      "epoch": 7.226644323418517,
      "grad_norm": 2.8327670097351074,
      "learning_rate": 4.277649769585253e-05,
      "loss": 6.4383,
      "step": 34500
    },
    {
      "epoch": 7.331378299120234,
      "grad_norm": 3.1451475620269775,
      "learning_rate": 4.267176372015082e-05,
      "loss": 6.4451,
      "step": 35000
    },
    {
      "epoch": 7.436112274821952,
      "grad_norm": 3.6858575344085693,
      "learning_rate": 4.2567239212400506e-05,
      "loss": 6.4313,
      "step": 35500
    },
    {
      "epoch": 7.5408462505236695,
      "grad_norm": 4.258295059204102,
      "learning_rate": 4.2462505236698786e-05,
      "loss": 6.4143,
      "step": 36000
    },
    {
      "epoch": 7.645580226225388,
      "grad_norm": 4.3574676513671875,
      "learning_rate": 4.235777126099707e-05,
      "loss": 6.4061,
      "step": 36500
    },
    {
      "epoch": 7.750314201927106,
      "grad_norm": 3.8001816272735596,
      "learning_rate": 4.225303728529535e-05,
      "loss": 6.3994,
      "step": 37000
    },
    {
      "epoch": 7.855048177628823,
      "grad_norm": 3.487893581390381,
      "learning_rate": 4.214851277754504e-05,
      "loss": 6.4192,
      "step": 37500
    },
    {
      "epoch": 7.95978215333054,
      "grad_norm": 3.9729723930358887,
      "learning_rate": 4.204377880184332e-05,
      "loss": 6.407,
      "step": 38000
    },
    {
      "epoch": 8.064516129032258,
      "grad_norm": 3.4465062618255615,
      "learning_rate": 4.1939044826141604e-05,
      "loss": 6.3867,
      "step": 38500
    },
    {
      "epoch": 8.169250104733976,
      "grad_norm": 3.706404685974121,
      "learning_rate": 4.1834310850439884e-05,
      "loss": 6.3877,
      "step": 39000
    },
    {
      "epoch": 8.273984080435694,
      "grad_norm": 3.8204259872436523,
      "learning_rate": 4.172957687473817e-05,
      "loss": 6.3921,
      "step": 39500
    },
    {
      "epoch": 8.37871805613741,
      "grad_norm": 3.4868948459625244,
      "learning_rate": 4.162505236698786e-05,
      "loss": 6.3729,
      "step": 40000
    },
    {
      "epoch": 8.483452031839128,
      "grad_norm": 3.5007081031799316,
      "learning_rate": 4.152031839128614e-05,
      "loss": 6.3886,
      "step": 40500
    },
    {
      "epoch": 8.588186007540846,
      "grad_norm": 2.937894582748413,
      "learning_rate": 4.1415584415584417e-05,
      "loss": 6.3814,
      "step": 41000
    },
    {
      "epoch": 8.692919983242565,
      "grad_norm": 3.529237985610962,
      "learning_rate": 4.1310850439882696e-05,
      "loss": 6.3722,
      "step": 41500
    },
    {
      "epoch": 8.79765395894428,
      "grad_norm": 3.883575677871704,
      "learning_rate": 4.120611646418098e-05,
      "loss": 6.3655,
      "step": 42000
    },
    {
      "epoch": 8.902387934645999,
      "grad_norm": 4.439103603363037,
      "learning_rate": 4.110159195643067e-05,
      "loss": 6.3673,
      "step": 42500
    },
    {
      "epoch": 9.007121910347717,
      "grad_norm": 4.103298664093018,
      "learning_rate": 4.099685798072895e-05,
      "loss": 6.3659,
      "step": 43000
    },
    {
      "epoch": 9.111855886049435,
      "grad_norm": 3.491204023361206,
      "learning_rate": 4.0892124005027235e-05,
      "loss": 6.3744,
      "step": 43500
    },
    {
      "epoch": 9.216589861751151,
      "grad_norm": 3.441976547241211,
      "learning_rate": 4.0787390029325515e-05,
      "loss": 6.3573,
      "step": 44000
    },
    {
      "epoch": 9.32132383745287,
      "grad_norm": 3.58134126663208,
      "learning_rate": 4.0682656053623795e-05,
      "loss": 6.3407,
      "step": 44500
    },
    {
      "epoch": 9.426057813154587,
      "grad_norm": 3.274592638015747,
      "learning_rate": 4.057813154587348e-05,
      "loss": 6.3373,
      "step": 45000
    },
    {
      "epoch": 9.530791788856305,
      "grad_norm": 4.296390533447266,
      "learning_rate": 4.047339757017177e-05,
      "loss": 6.3499,
      "step": 45500
    },
    {
      "epoch": 9.635525764558023,
      "grad_norm": 3.5000336170196533,
      "learning_rate": 4.036866359447005e-05,
      "loss": 6.3199,
      "step": 46000
    },
    {
      "epoch": 9.74025974025974,
      "grad_norm": 3.4947054386138916,
      "learning_rate": 4.0263929618768334e-05,
      "loss": 6.3474,
      "step": 46500
    },
    {
      "epoch": 9.844993715961458,
      "grad_norm": 3.3658857345581055,
      "learning_rate": 4.0159195643066614e-05,
      "loss": 6.3296,
      "step": 47000
    },
    {
      "epoch": 9.949727691663176,
      "grad_norm": 2.9811642169952393,
      "learning_rate": 4.0054671135316294e-05,
      "loss": 6.345,
      "step": 47500
    },
    {
      "epoch": 10.054461667364894,
      "grad_norm": 4.165875434875488,
      "learning_rate": 3.994993715961458e-05,
      "loss": 6.3209,
      "step": 48000
    },
    {
      "epoch": 10.15919564306661,
      "grad_norm": 3.6118202209472656,
      "learning_rate": 3.9845203183912866e-05,
      "loss": 6.3175,
      "step": 48500
    },
    {
      "epoch": 10.263929618768328,
      "grad_norm": 3.930669069290161,
      "learning_rate": 3.9740469208211146e-05,
      "loss": 6.3231,
      "step": 49000
    },
    {
      "epoch": 10.368663594470046,
      "grad_norm": 3.1688554286956787,
      "learning_rate": 3.963594470046083e-05,
      "loss": 6.309,
      "step": 49500
    },
    {
      "epoch": 10.473397570171764,
      "grad_norm": 3.6746394634246826,
      "learning_rate": 3.953121072475911e-05,
      "loss": 6.3077,
      "step": 50000
    },
    {
      "epoch": 10.57813154587348,
      "grad_norm": 3.5134785175323486,
      "learning_rate": 3.942647674905739e-05,
      "loss": 6.3299,
      "step": 50500
    },
    {
      "epoch": 10.682865521575199,
      "grad_norm": 3.2903287410736084,
      "learning_rate": 3.932174277335568e-05,
      "loss": 6.3178,
      "step": 51000
    },
    {
      "epoch": 10.787599497276917,
      "grad_norm": 3.5344769954681396,
      "learning_rate": 3.921700879765396e-05,
      "loss": 6.3139,
      "step": 51500
    },
    {
      "epoch": 10.892333472978635,
      "grad_norm": 3.5710573196411133,
      "learning_rate": 3.9112274821952245e-05,
      "loss": 6.306,
      "step": 52000
    },
    {
      "epoch": 10.997067448680351,
      "grad_norm": 3.4193336963653564,
      "learning_rate": 3.9007540846250524e-05,
      "loss": 6.3129,
      "step": 52500
    },
    {
      "epoch": 11.101801424382069,
      "grad_norm": 3.683143377304077,
      "learning_rate": 3.890301633850021e-05,
      "loss": 6.3084,
      "step": 53000
    },
    {
      "epoch": 11.206535400083787,
      "grad_norm": 3.214221239089966,
      "learning_rate": 3.879828236279849e-05,
      "loss": 6.302,
      "step": 53500
    },
    {
      "epoch": 11.311269375785505,
      "grad_norm": 3.5691747665405273,
      "learning_rate": 3.869354838709678e-05,
      "loss": 6.3164,
      "step": 54000
    },
    {
      "epoch": 11.416003351487223,
      "grad_norm": 3.2734036445617676,
      "learning_rate": 3.858881441139506e-05,
      "loss": 6.2889,
      "step": 54500
    },
    {
      "epoch": 11.52073732718894,
      "grad_norm": 4.049854278564453,
      "learning_rate": 3.848428990364474e-05,
      "loss": 6.2957,
      "step": 55000
    },
    {
      "epoch": 11.625471302890658,
      "grad_norm": 3.837921380996704,
      "learning_rate": 3.837955592794303e-05,
      "loss": 6.2806,
      "step": 55500
    },
    {
      "epoch": 11.730205278592376,
      "grad_norm": 3.4606828689575195,
      "learning_rate": 3.827482195224131e-05,
      "loss": 6.2896,
      "step": 56000
    },
    {
      "epoch": 11.834939254294094,
      "grad_norm": 4.859198093414307,
      "learning_rate": 3.8170087976539596e-05,
      "loss": 6.273,
      "step": 56500
    },
    {
      "epoch": 11.93967322999581,
      "grad_norm": 4.689023494720459,
      "learning_rate": 3.806535400083787e-05,
      "loss": 6.2776,
      "step": 57000
    },
    {
      "epoch": 12.044407205697528,
      "grad_norm": 4.234752178192139,
      "learning_rate": 3.7960829493087555e-05,
      "loss": 6.282,
      "step": 57500
    },
    {
      "epoch": 12.149141181399246,
      "grad_norm": 3.950773239135742,
      "learning_rate": 3.785609551738584e-05,
      "loss": 6.283,
      "step": 58000
    },
    {
      "epoch": 12.253875157100964,
      "grad_norm": 4.1780548095703125,
      "learning_rate": 3.775136154168412e-05,
      "loss": 6.2635,
      "step": 58500
    },
    {
      "epoch": 12.35860913280268,
      "grad_norm": 3.2049672603607178,
      "learning_rate": 3.764662756598241e-05,
      "loss": 6.2858,
      "step": 59000
    },
    {
      "epoch": 12.463343108504398,
      "grad_norm": 3.863649606704712,
      "learning_rate": 3.7541893590280694e-05,
      "loss": 6.2609,
      "step": 59500
    },
    {
      "epoch": 12.568077084206116,
      "grad_norm": 3.881343364715576,
      "learning_rate": 3.743715961457897e-05,
      "loss": 6.2695,
      "step": 60000
    },
    {
      "epoch": 12.672811059907835,
      "grad_norm": 3.522132635116577,
      "learning_rate": 3.7332635106828654e-05,
      "loss": 6.2523,
      "step": 60500
    },
    {
      "epoch": 12.777545035609553,
      "grad_norm": 4.043595790863037,
      "learning_rate": 3.722790113112694e-05,
      "loss": 6.2546,
      "step": 61000
    },
    {
      "epoch": 12.882279011311269,
      "grad_norm": 3.4860141277313232,
      "learning_rate": 3.712316715542522e-05,
      "loss": 6.2468,
      "step": 61500
    },
    {
      "epoch": 12.987012987012987,
      "grad_norm": 3.9201574325561523,
      "learning_rate": 3.7018433179723506e-05,
      "loss": 6.2615,
      "step": 62000
    },
    {
      "epoch": 13.091746962714705,
      "grad_norm": 3.5582118034362793,
      "learning_rate": 3.6913699204021786e-05,
      "loss": 6.2529,
      "step": 62500
    },
    {
      "epoch": 13.196480938416423,
      "grad_norm": 3.1254093647003174,
      "learning_rate": 3.680917469627147e-05,
      "loss": 6.2418,
      "step": 63000
    },
    {
      "epoch": 13.30121491411814,
      "grad_norm": 4.058616638183594,
      "learning_rate": 3.670444072056975e-05,
      "loss": 6.243,
      "step": 63500
    },
    {
      "epoch": 13.405948889819857,
      "grad_norm": 3.5146963596343994,
      "learning_rate": 3.659970674486803e-05,
      "loss": 6.2595,
      "step": 64000
    },
    {
      "epoch": 13.510682865521575,
      "grad_norm": 3.804818630218506,
      "learning_rate": 3.649497276916632e-05,
      "loss": 6.2438,
      "step": 64500
    },
    {
      "epoch": 13.615416841223293,
      "grad_norm": 3.591214179992676,
      "learning_rate": 3.6390238793464605e-05,
      "loss": 6.2266,
      "step": 65000
    },
    {
      "epoch": 13.72015081692501,
      "grad_norm": 4.973635196685791,
      "learning_rate": 3.6285504817762885e-05,
      "loss": 6.2501,
      "step": 65500
    },
    {
      "epoch": 13.824884792626728,
      "grad_norm": 4.189575672149658,
      "learning_rate": 3.618098031001257e-05,
      "loss": 6.2341,
      "step": 66000
    },
    {
      "epoch": 13.929618768328446,
      "grad_norm": 4.408186912536621,
      "learning_rate": 3.607624633431085e-05,
      "loss": 6.227,
      "step": 66500
    },
    {
      "epoch": 14.034352744030164,
      "grad_norm": 4.066199779510498,
      "learning_rate": 3.597151235860913e-05,
      "loss": 6.2316,
      "step": 67000
    },
    {
      "epoch": 14.139086719731882,
      "grad_norm": 3.8263683319091797,
      "learning_rate": 3.586677838290742e-05,
      "loss": 6.2322,
      "step": 67500
    },
    {
      "epoch": 14.243820695433598,
      "grad_norm": 4.787020206451416,
      "learning_rate": 3.57620444072057e-05,
      "loss": 6.2099,
      "step": 68000
    },
    {
      "epoch": 14.348554671135316,
      "grad_norm": 3.5196545124053955,
      "learning_rate": 3.565751989945538e-05,
      "loss": 6.2425,
      "step": 68500
    },
    {
      "epoch": 14.453288646837034,
      "grad_norm": 4.1746439933776855,
      "learning_rate": 3.555278592375367e-05,
      "loss": 6.2373,
      "step": 69000
    },
    {
      "epoch": 14.558022622538752,
      "grad_norm": 4.07820463180542,
      "learning_rate": 3.544805194805195e-05,
      "loss": 6.2099,
      "step": 69500
    },
    {
      "epoch": 14.662756598240469,
      "grad_norm": 3.400038242340088,
      "learning_rate": 3.534331797235023e-05,
      "loss": 6.216,
      "step": 70000
    },
    {
      "epoch": 14.767490573942187,
      "grad_norm": 4.578042030334473,
      "learning_rate": 3.5238793464599916e-05,
      "loss": 6.2091,
      "step": 70500
    },
    {
      "epoch": 14.872224549643905,
      "grad_norm": 3.6254208087921143,
      "learning_rate": 3.51340594888982e-05,
      "loss": 6.2258,
      "step": 71000
    },
    {
      "epoch": 14.976958525345623,
      "grad_norm": 3.496166467666626,
      "learning_rate": 3.502932551319648e-05,
      "loss": 6.2138,
      "step": 71500
    },
    {
      "epoch": 15.081692501047339,
      "grad_norm": 3.5367865562438965,
      "learning_rate": 3.492459153749477e-05,
      "loss": 6.213,
      "step": 72000
    },
    {
      "epoch": 15.186426476749057,
      "grad_norm": 3.4754440784454346,
      "learning_rate": 3.481985756179305e-05,
      "loss": 6.2153,
      "step": 72500
    },
    {
      "epoch": 15.291160452450775,
      "grad_norm": 4.432271957397461,
      "learning_rate": 3.4715333054042735e-05,
      "loss": 6.2007,
      "step": 73000
    },
    {
      "epoch": 15.395894428152493,
      "grad_norm": 3.8427770137786865,
      "learning_rate": 3.4610599078341014e-05,
      "loss": 6.2071,
      "step": 73500
    },
    {
      "epoch": 15.50062840385421,
      "grad_norm": 3.9617857933044434,
      "learning_rate": 3.4505865102639294e-05,
      "loss": 6.2142,
      "step": 74000
    },
    {
      "epoch": 15.605362379555928,
      "grad_norm": 3.769693613052368,
      "learning_rate": 3.440113112693758e-05,
      "loss": 6.2065,
      "step": 74500
    },
    {
      "epoch": 15.710096355257646,
      "grad_norm": 3.825507402420044,
      "learning_rate": 3.429639715123587e-05,
      "loss": 6.2072,
      "step": 75000
    },
    {
      "epoch": 15.814830330959364,
      "grad_norm": 3.982872724533081,
      "learning_rate": 3.4191872643485554e-05,
      "loss": 6.2003,
      "step": 75500
    },
    {
      "epoch": 15.91956430666108,
      "grad_norm": 3.9958648681640625,
      "learning_rate": 3.408713866778383e-05,
      "loss": 6.1913,
      "step": 76000
    },
    {
      "epoch": 16.0242982823628,
      "grad_norm": 3.947957754135132,
      "learning_rate": 3.398240469208211e-05,
      "loss": 6.2019,
      "step": 76500
    },
    {
      "epoch": 16.129032258064516,
      "grad_norm": 3.8135411739349365,
      "learning_rate": 3.387767071638039e-05,
      "loss": 6.1944,
      "step": 77000
    },
    {
      "epoch": 16.233766233766232,
      "grad_norm": 3.940861701965332,
      "learning_rate": 3.377293674067868e-05,
      "loss": 6.1893,
      "step": 77500
    },
    {
      "epoch": 16.338500209467952,
      "grad_norm": 5.24894905090332,
      "learning_rate": 3.3668412232928366e-05,
      "loss": 6.1984,
      "step": 78000
    },
    {
      "epoch": 16.44323418516967,
      "grad_norm": 4.470870494842529,
      "learning_rate": 3.3563678257226645e-05,
      "loss": 6.1958,
      "step": 78500
    },
    {
      "epoch": 16.547968160871388,
      "grad_norm": 3.699892282485962,
      "learning_rate": 3.345894428152493e-05,
      "loss": 6.1952,
      "step": 79000
    },
    {
      "epoch": 16.652702136573104,
      "grad_norm": 4.136711120605469,
      "learning_rate": 3.335421030582321e-05,
      "loss": 6.1896,
      "step": 79500
    },
    {
      "epoch": 16.75743611227482,
      "grad_norm": 4.904257297515869,
      "learning_rate": 3.324968579807289e-05,
      "loss": 6.1715,
      "step": 80000
    },
    {
      "epoch": 16.86217008797654,
      "grad_norm": 4.219280242919922,
      "learning_rate": 3.314495182237118e-05,
      "loss": 6.1829,
      "step": 80500
    },
    {
      "epoch": 16.966904063678257,
      "grad_norm": 4.426414489746094,
      "learning_rate": 3.3040217846669464e-05,
      "loss": 6.1782,
      "step": 81000
    },
    {
      "epoch": 17.071638039379973,
      "grad_norm": 4.792020797729492,
      "learning_rate": 3.2935483870967744e-05,
      "loss": 6.1675,
      "step": 81500
    },
    {
      "epoch": 17.176372015081693,
      "grad_norm": 3.9796903133392334,
      "learning_rate": 3.283095936321743e-05,
      "loss": 6.179,
      "step": 82000
    },
    {
      "epoch": 17.28110599078341,
      "grad_norm": 4.554388046264648,
      "learning_rate": 3.272622538751571e-05,
      "loss": 6.1756,
      "step": 82500
    },
    {
      "epoch": 17.38583996648513,
      "grad_norm": 4.024316787719727,
      "learning_rate": 3.262149141181399e-05,
      "loss": 6.177,
      "step": 83000
    },
    {
      "epoch": 17.490573942186845,
      "grad_norm": 4.059772968292236,
      "learning_rate": 3.2516757436112276e-05,
      "loss": 6.1818,
      "step": 83500
    },
    {
      "epoch": 17.59530791788856,
      "grad_norm": 4.296391487121582,
      "learning_rate": 3.241223292836196e-05,
      "loss": 6.1866,
      "step": 84000
    },
    {
      "epoch": 17.70004189359028,
      "grad_norm": 4.008220672607422,
      "learning_rate": 3.230749895266024e-05,
      "loss": 6.172,
      "step": 84500
    },
    {
      "epoch": 17.804775869291998,
      "grad_norm": 4.639082908630371,
      "learning_rate": 3.220276497695853e-05,
      "loss": 6.169,
      "step": 85000
    },
    {
      "epoch": 17.909509844993718,
      "grad_norm": 4.635848522186279,
      "learning_rate": 3.209803100125681e-05,
      "loss": 6.1721,
      "step": 85500
    },
    {
      "epoch": 18.014243820695434,
      "grad_norm": 4.662270545959473,
      "learning_rate": 3.199329702555509e-05,
      "loss": 6.1575,
      "step": 86000
    },
    {
      "epoch": 18.11897779639715,
      "grad_norm": 4.280701637268066,
      "learning_rate": 3.1888772517804775e-05,
      "loss": 6.1482,
      "step": 86500
    },
    {
      "epoch": 18.22371177209887,
      "grad_norm": 3.8602380752563477,
      "learning_rate": 3.178403854210306e-05,
      "loss": 6.1565,
      "step": 87000
    },
    {
      "epoch": 18.328445747800586,
      "grad_norm": 4.634263515472412,
      "learning_rate": 3.167930456640134e-05,
      "loss": 6.1587,
      "step": 87500
    },
    {
      "epoch": 18.433179723502302,
      "grad_norm": 4.115392208099365,
      "learning_rate": 3.157457059069963e-05,
      "loss": 6.1436,
      "step": 88000
    },
    {
      "epoch": 18.537913699204022,
      "grad_norm": 3.6665916442871094,
      "learning_rate": 3.146983661499791e-05,
      "loss": 6.1442,
      "step": 88500
    },
    {
      "epoch": 18.64264767490574,
      "grad_norm": 4.444345951080322,
      "learning_rate": 3.1365312107247594e-05,
      "loss": 6.1571,
      "step": 89000
    },
    {
      "epoch": 18.74738165060746,
      "grad_norm": 3.792792558670044,
      "learning_rate": 3.1260578131545873e-05,
      "loss": 6.1535,
      "step": 89500
    },
    {
      "epoch": 18.852115626309175,
      "grad_norm": 3.904019832611084,
      "learning_rate": 3.115584415584415e-05,
      "loss": 6.1501,
      "step": 90000
    },
    {
      "epoch": 18.95684960201089,
      "grad_norm": 4.531284332275391,
      "learning_rate": 3.105111018014244e-05,
      "loss": 6.1592,
      "step": 90500
    },
    {
      "epoch": 19.06158357771261,
      "grad_norm": 3.7976317405700684,
      "learning_rate": 3.0946376204440726e-05,
      "loss": 6.1474,
      "step": 91000
    },
    {
      "epoch": 19.166317553414327,
      "grad_norm": 3.8021469116210938,
      "learning_rate": 3.084185169669041e-05,
      "loss": 6.1408,
      "step": 91500
    },
    {
      "epoch": 19.271051529116047,
      "grad_norm": 4.194758892059326,
      "learning_rate": 3.073711772098869e-05,
      "loss": 6.1476,
      "step": 92000
    },
    {
      "epoch": 19.375785504817763,
      "grad_norm": 4.084668159484863,
      "learning_rate": 3.063238374528697e-05,
      "loss": 6.1443,
      "step": 92500
    },
    {
      "epoch": 19.48051948051948,
      "grad_norm": 4.383222579956055,
      "learning_rate": 3.052764976958525e-05,
      "loss": 6.1422,
      "step": 93000
    },
    {
      "epoch": 19.5852534562212,
      "grad_norm": 4.250995635986328,
      "learning_rate": 3.042312526183494e-05,
      "loss": 6.1375,
      "step": 93500
    },
    {
      "epoch": 19.689987431922916,
      "grad_norm": 4.78529691696167,
      "learning_rate": 3.0318391286133225e-05,
      "loss": 6.1368,
      "step": 94000
    },
    {
      "epoch": 19.794721407624632,
      "grad_norm": 3.4997754096984863,
      "learning_rate": 3.0213657310431504e-05,
      "loss": 6.1432,
      "step": 94500
    },
    {
      "epoch": 19.89945538332635,
      "grad_norm": 4.723648548126221,
      "learning_rate": 3.0108923334729787e-05,
      "loss": 6.1263,
      "step": 95000
    },
    {
      "epoch": 20.004189359028068,
      "grad_norm": 3.930859088897705,
      "learning_rate": 3.0004398826979474e-05,
      "loss": 6.1387,
      "step": 95500
    },
    {
      "epoch": 20.108923334729788,
      "grad_norm": 4.286599159240723,
      "learning_rate": 2.9899664851277754e-05,
      "loss": 6.1187,
      "step": 96000
    },
    {
      "epoch": 20.213657310431504,
      "grad_norm": 3.8475680351257324,
      "learning_rate": 2.9794930875576037e-05,
      "loss": 6.1312,
      "step": 96500
    },
    {
      "epoch": 20.31839128613322,
      "grad_norm": 4.844906806945801,
      "learning_rate": 2.9690196899874323e-05,
      "loss": 6.1457,
      "step": 97000
    },
    {
      "epoch": 20.42312526183494,
      "grad_norm": 4.691315174102783,
      "learning_rate": 2.958567239212401e-05,
      "loss": 6.1351,
      "step": 97500
    },
    {
      "epoch": 20.527859237536656,
      "grad_norm": 6.15250825881958,
      "learning_rate": 2.9480938416422286e-05,
      "loss": 6.1275,
      "step": 98000
    },
    {
      "epoch": 20.632593213238373,
      "grad_norm": 3.8872599601745605,
      "learning_rate": 2.9376204440720573e-05,
      "loss": 6.1275,
      "step": 98500
    },
    {
      "epoch": 20.737327188940093,
      "grad_norm": 4.541051864624023,
      "learning_rate": 2.9271470465018852e-05,
      "loss": 6.1472,
      "step": 99000
    },
    {
      "epoch": 20.84206116464181,
      "grad_norm": 4.556408405303955,
      "learning_rate": 2.9166736489317135e-05,
      "loss": 6.1369,
      "step": 99500
    },
    {
      "epoch": 20.94679514034353,
      "grad_norm": 4.5567498207092285,
      "learning_rate": 2.9062211981566822e-05,
      "loss": 6.1148,
      "step": 100000
    },
    {
      "epoch": 21.051529116045245,
      "grad_norm": 4.647518634796143,
      "learning_rate": 2.8957478005865102e-05,
      "loss": 6.1281,
      "step": 100500
    },
    {
      "epoch": 21.15626309174696,
      "grad_norm": 4.372421741485596,
      "learning_rate": 2.8852744030163388e-05,
      "loss": 6.1226,
      "step": 101000
    },
    {
      "epoch": 21.26099706744868,
      "grad_norm": 4.270533084869385,
      "learning_rate": 2.874801005446167e-05,
      "loss": 6.123,
      "step": 101500
    },
    {
      "epoch": 21.365731043150397,
      "grad_norm": 3.5596370697021484,
      "learning_rate": 2.8643485546711358e-05,
      "loss": 6.1402,
      "step": 102000
    },
    {
      "epoch": 21.470465018852117,
      "grad_norm": 5.230384826660156,
      "learning_rate": 2.8538751571009638e-05,
      "loss": 6.1199,
      "step": 102500
    },
    {
      "epoch": 21.575198994553833,
      "grad_norm": 3.9881417751312256,
      "learning_rate": 2.843401759530792e-05,
      "loss": 6.1244,
      "step": 103000
    },
    {
      "epoch": 21.67993297025555,
      "grad_norm": 4.617568016052246,
      "learning_rate": 2.83292836196062e-05,
      "loss": 6.1154,
      "step": 103500
    },
    {
      "epoch": 21.78466694595727,
      "grad_norm": 4.641009330749512,
      "learning_rate": 2.8224759111855887e-05,
      "loss": 6.113,
      "step": 104000
    },
    {
      "epoch": 21.889400921658986,
      "grad_norm": 4.005772113800049,
      "learning_rate": 2.812002513615417e-05,
      "loss": 6.1163,
      "step": 104500
    },
    {
      "epoch": 21.994134897360702,
      "grad_norm": 4.2611799240112305,
      "learning_rate": 2.801529116045245e-05,
      "loss": 6.1023,
      "step": 105000
    },
    {
      "epoch": 22.098868873062422,
      "grad_norm": 4.568357467651367,
      "learning_rate": 2.7910557184750736e-05,
      "loss": 6.1169,
      "step": 105500
    },
    {
      "epoch": 22.203602848764138,
      "grad_norm": 4.323103427886963,
      "learning_rate": 2.780603267700042e-05,
      "loss": 6.1226,
      "step": 106000
    },
    {
      "epoch": 22.308336824465858,
      "grad_norm": 4.507444381713867,
      "learning_rate": 2.77012987012987e-05,
      "loss": 6.1052,
      "step": 106500
    },
    {
      "epoch": 22.413070800167574,
      "grad_norm": 4.301244735717773,
      "learning_rate": 2.7596564725596985e-05,
      "loss": 6.0944,
      "step": 107000
    },
    {
      "epoch": 22.51780477586929,
      "grad_norm": 4.984853267669678,
      "learning_rate": 2.749183074989527e-05,
      "loss": 6.1125,
      "step": 107500
    },
    {
      "epoch": 22.62253875157101,
      "grad_norm": 4.682931423187256,
      "learning_rate": 2.7387096774193548e-05,
      "loss": 6.1158,
      "step": 108000
    },
    {
      "epoch": 22.727272727272727,
      "grad_norm": 4.494015693664551,
      "learning_rate": 2.7282572266443235e-05,
      "loss": 6.1035,
      "step": 108500
    },
    {
      "epoch": 22.832006702974446,
      "grad_norm": 3.880779981613159,
      "learning_rate": 2.717783829074152e-05,
      "loss": 6.1084,
      "step": 109000
    },
    {
      "epoch": 22.936740678676163,
      "grad_norm": 4.154653072357178,
      "learning_rate": 2.7073104315039798e-05,
      "loss": 6.0945,
      "step": 109500
    },
    {
      "epoch": 23.04147465437788,
      "grad_norm": 5.443271160125732,
      "learning_rate": 2.6968370339338084e-05,
      "loss": 6.1016,
      "step": 110000
    },
    {
      "epoch": 23.1462086300796,
      "grad_norm": 4.298133373260498,
      "learning_rate": 2.686384583158777e-05,
      "loss": 6.103,
      "step": 110500
    },
    {
      "epoch": 23.250942605781315,
      "grad_norm": 4.379884243011475,
      "learning_rate": 2.675911185588605e-05,
      "loss": 6.0814,
      "step": 111000
    },
    {
      "epoch": 23.35567658148303,
      "grad_norm": 6.175398349761963,
      "learning_rate": 2.6654377880184333e-05,
      "loss": 6.1088,
      "step": 111500
    },
    {
      "epoch": 23.46041055718475,
      "grad_norm": 4.121715068817139,
      "learning_rate": 2.6549643904482613e-05,
      "loss": 6.0966,
      "step": 112000
    },
    {
      "epoch": 23.565144532886467,
      "grad_norm": 5.040287494659424,
      "learning_rate": 2.6444909928780896e-05,
      "loss": 6.098,
      "step": 112500
    },
    {
      "epoch": 23.669878508588187,
      "grad_norm": 4.766879081726074,
      "learning_rate": 2.6340175953079182e-05,
      "loss": 6.0957,
      "step": 113000
    },
    {
      "epoch": 23.774612484289904,
      "grad_norm": 5.87930965423584,
      "learning_rate": 2.623565144532887e-05,
      "loss": 6.1089,
      "step": 113500
    },
    {
      "epoch": 23.87934645999162,
      "grad_norm": 5.318653583526611,
      "learning_rate": 2.613091746962715e-05,
      "loss": 6.0761,
      "step": 114000
    },
    {
      "epoch": 23.98408043569334,
      "grad_norm": 4.465319633483887,
      "learning_rate": 2.6026183493925432e-05,
      "loss": 6.0826,
      "step": 114500
    },
    {
      "epoch": 24.088814411395056,
      "grad_norm": 4.640571594238281,
      "learning_rate": 2.592144951822371e-05,
      "loss": 6.0805,
      "step": 115000
    },
    {
      "epoch": 24.193548387096776,
      "grad_norm": 4.252554416656494,
      "learning_rate": 2.5816925010473398e-05,
      "loss": 6.0701,
      "step": 115500
    },
    {
      "epoch": 24.298282362798492,
      "grad_norm": 4.704644203186035,
      "learning_rate": 2.571219103477168e-05,
      "loss": 6.0936,
      "step": 116000
    },
    {
      "epoch": 24.40301633850021,
      "grad_norm": 4.601324558258057,
      "learning_rate": 2.560745705906996e-05,
      "loss": 6.0758,
      "step": 116500
    },
    {
      "epoch": 24.507750314201928,
      "grad_norm": 4.380444526672363,
      "learning_rate": 2.5502723083368247e-05,
      "loss": 6.0871,
      "step": 117000
    },
    {
      "epoch": 24.612484289903644,
      "grad_norm": 4.119806289672852,
      "learning_rate": 2.5397989107666527e-05,
      "loss": 6.102,
      "step": 117500
    },
    {
      "epoch": 24.71721826560536,
      "grad_norm": 3.9712698459625244,
      "learning_rate": 2.5293464599916217e-05,
      "loss": 6.0999,
      "step": 118000
    },
    {
      "epoch": 24.82195224130708,
      "grad_norm": 5.146612167358398,
      "learning_rate": 2.5188730624214497e-05,
      "loss": 6.0947,
      "step": 118500
    },
    {
      "epoch": 24.926686217008797,
      "grad_norm": 4.406741142272949,
      "learning_rate": 2.508399664851278e-05,
      "loss": 6.0829,
      "step": 119000
    },
    {
      "epoch": 25.031420192710517,
      "grad_norm": 5.4739766120910645,
      "learning_rate": 2.497926267281106e-05,
      "loss": 6.0598,
      "step": 119500
    },
    {
      "epoch": 25.136154168412233,
      "grad_norm": 4.6231794357299805,
      "learning_rate": 2.4874738165060746e-05,
      "loss": 6.072,
      "step": 120000
    },
    {
      "epoch": 25.24088814411395,
      "grad_norm": 4.47750186920166,
      "learning_rate": 2.477000418935903e-05,
      "loss": 6.0664,
      "step": 120500
    },
    {
      "epoch": 25.34562211981567,
      "grad_norm": 5.023014068603516,
      "learning_rate": 2.4665270213657312e-05,
      "loss": 6.0894,
      "step": 121000
    },
    {
      "epoch": 25.450356095517385,
      "grad_norm": 5.687644004821777,
      "learning_rate": 2.4560536237955595e-05,
      "loss": 6.0936,
      "step": 121500
    },
    {
      "epoch": 25.555090071219105,
      "grad_norm": 4.534958362579346,
      "learning_rate": 2.4455802262253878e-05,
      "loss": 6.0794,
      "step": 122000
    },
    {
      "epoch": 25.65982404692082,
      "grad_norm": 5.563751697540283,
      "learning_rate": 2.435127775450356e-05,
      "loss": 6.081,
      "step": 122500
    },
    {
      "epoch": 25.764558022622538,
      "grad_norm": 4.613626956939697,
      "learning_rate": 2.4246543778801845e-05,
      "loss": 6.0982,
      "step": 123000
    },
    {
      "epoch": 25.869291998324258,
      "grad_norm": 4.645818710327148,
      "learning_rate": 2.4141809803100128e-05,
      "loss": 6.0665,
      "step": 123500
    },
    {
      "epoch": 25.974025974025974,
      "grad_norm": 4.9156928062438965,
      "learning_rate": 2.4037075827398407e-05,
      "loss": 6.0674,
      "step": 124000
    },
    {
      "epoch": 26.07875994972769,
      "grad_norm": 4.7342305183410645,
      "learning_rate": 2.3932551319648094e-05,
      "loss": 6.0741,
      "step": 124500
    },
    {
      "epoch": 26.18349392542941,
      "grad_norm": 4.607081413269043,
      "learning_rate": 2.3827817343946377e-05,
      "loss": 6.0626,
      "step": 125000
    },
    {
      "epoch": 26.288227901131126,
      "grad_norm": 4.820442199707031,
      "learning_rate": 2.372308336824466e-05,
      "loss": 6.0936,
      "step": 125500
    },
    {
      "epoch": 26.392961876832846,
      "grad_norm": 4.549975395202637,
      "learning_rate": 2.3618349392542943e-05,
      "loss": 6.0804,
      "step": 126000
    },
    {
      "epoch": 26.497695852534562,
      "grad_norm": 4.722150802612305,
      "learning_rate": 2.351382488479263e-05,
      "loss": 6.0555,
      "step": 126500
    },
    {
      "epoch": 26.60242982823628,
      "grad_norm": 4.2948408126831055,
      "learning_rate": 2.340909090909091e-05,
      "loss": 6.0626,
      "step": 127000
    },
    {
      "epoch": 26.707163803938,
      "grad_norm": 4.246878623962402,
      "learning_rate": 2.3304356933389193e-05,
      "loss": 6.0577,
      "step": 127500
    },
    {
      "epoch": 26.811897779639715,
      "grad_norm": 4.165809154510498,
      "learning_rate": 2.3199622957687476e-05,
      "loss": 6.0608,
      "step": 128000
    },
    {
      "epoch": 26.916631755341434,
      "grad_norm": 4.3159894943237305,
      "learning_rate": 2.309488898198576e-05,
      "loss": 6.0806,
      "step": 128500
    },
    {
      "epoch": 27.02136573104315,
      "grad_norm": 4.15300989151001,
      "learning_rate": 2.2990364474235442e-05,
      "loss": 6.0654,
      "step": 129000
    },
    {
      "epoch": 27.126099706744867,
      "grad_norm": 4.730154991149902,
      "learning_rate": 2.2885630498533725e-05,
      "loss": 6.0567,
      "step": 129500
    },
    {
      "epoch": 27.230833682446587,
      "grad_norm": 4.300974369049072,
      "learning_rate": 2.2780896522832008e-05,
      "loss": 6.0449,
      "step": 130000
    },
    {
      "epoch": 27.335567658148303,
      "grad_norm": 4.148283958435059,
      "learning_rate": 2.2676162547130288e-05,
      "loss": 6.0491,
      "step": 130500
    },
    {
      "epoch": 27.44030163385002,
      "grad_norm": 4.9924421310424805,
      "learning_rate": 2.2571638039379974e-05,
      "loss": 6.0491,
      "step": 131000
    },
    {
      "epoch": 27.54503560955174,
      "grad_norm": 5.713706016540527,
      "learning_rate": 2.246690406367826e-05,
      "loss": 6.0396,
      "step": 131500
    },
    {
      "epoch": 27.649769585253456,
      "grad_norm": 5.007369518280029,
      "learning_rate": 2.236217008797654e-05,
      "loss": 6.0378,
      "step": 132000
    },
    {
      "epoch": 27.754503560955175,
      "grad_norm": 4.500640392303467,
      "learning_rate": 2.2257436112274823e-05,
      "loss": 6.0313,
      "step": 132500
    },
    {
      "epoch": 27.85923753665689,
      "grad_norm": 4.709275722503662,
      "learning_rate": 2.2152702136573107e-05,
      "loss": 6.0278,
      "step": 133000
    },
    {
      "epoch": 27.963971512358608,
      "grad_norm": 4.891386032104492,
      "learning_rate": 2.204817762882279e-05,
      "loss": 6.0267,
      "step": 133500
    },
    {
      "epoch": 28.068705488060328,
      "grad_norm": 4.82666540145874,
      "learning_rate": 2.1943443653121073e-05,
      "loss": 5.986,
      "step": 134000
    },
    {
      "epoch": 28.173439463762044,
      "grad_norm": 4.489607810974121,
      "learning_rate": 2.1838709677419356e-05,
      "loss": 6.0209,
      "step": 134500
    },
    {
      "epoch": 28.278173439463764,
      "grad_norm": 4.719301700592041,
      "learning_rate": 2.173397570171764e-05,
      "loss": 5.9998,
      "step": 135000
    },
    {
      "epoch": 28.38290741516548,
      "grad_norm": 5.639565467834473,
      "learning_rate": 2.1629451193967322e-05,
      "loss": 5.9881,
      "step": 135500
    },
    {
      "epoch": 28.487641390867196,
      "grad_norm": 4.745512008666992,
      "learning_rate": 2.1524717218265605e-05,
      "loss": 6.0002,
      "step": 136000
    },
    {
      "epoch": 28.592375366568916,
      "grad_norm": 5.661725997924805,
      "learning_rate": 2.141998324256389e-05,
      "loss": 5.9967,
      "step": 136500
    },
    {
      "epoch": 28.697109342270632,
      "grad_norm": 5.3391194343566895,
      "learning_rate": 2.131524926686217e-05,
      "loss": 6.0024,
      "step": 137000
    },
    {
      "epoch": 28.80184331797235,
      "grad_norm": 5.1614089012146,
      "learning_rate": 2.1210724759111858e-05,
      "loss": 5.9992,
      "step": 137500
    },
    {
      "epoch": 28.90657729367407,
      "grad_norm": 5.429248332977295,
      "learning_rate": 2.110599078341014e-05,
      "loss": 5.9929,
      "step": 138000
    },
    {
      "epoch": 29.011311269375785,
      "grad_norm": 5.1270012855529785,
      "learning_rate": 2.100125680770842e-05,
      "loss": 5.9883,
      "step": 138500
    },
    {
      "epoch": 29.116045245077505,
      "grad_norm": 5.027891159057617,
      "learning_rate": 2.0896522832006704e-05,
      "loss": 5.963,
      "step": 139000
    },
    {
      "epoch": 29.22077922077922,
      "grad_norm": 5.712099552154541,
      "learning_rate": 2.079199832425639e-05,
      "loss": 5.9811,
      "step": 139500
    },
    {
      "epoch": 29.325513196480937,
      "grad_norm": 4.954220294952393,
      "learning_rate": 2.0687264348554674e-05,
      "loss": 5.9808,
      "step": 140000
    },
    {
      "epoch": 29.430247172182657,
      "grad_norm": 5.713419437408447,
      "learning_rate": 2.0582530372852953e-05,
      "loss": 5.9887,
      "step": 140500
    },
    {
      "epoch": 29.534981147884373,
      "grad_norm": 4.683711528778076,
      "learning_rate": 2.0477796397151236e-05,
      "loss": 5.9612,
      "step": 141000
    },
    {
      "epoch": 29.63971512358609,
      "grad_norm": 5.164538383483887,
      "learning_rate": 2.0373271889400923e-05,
      "loss": 5.993,
      "step": 141500
    },
    {
      "epoch": 29.74444909928781,
      "grad_norm": 5.386078357696533,
      "learning_rate": 2.0268537913699203e-05,
      "loss": 5.9735,
      "step": 142000
    },
    {
      "epoch": 29.849183074989526,
      "grad_norm": 4.4406418800354,
      "learning_rate": 2.016380393799749e-05,
      "loss": 5.9672,
      "step": 142500
    },
    {
      "epoch": 29.953917050691246,
      "grad_norm": 5.029815673828125,
      "learning_rate": 2.0059069962295772e-05,
      "loss": 5.961,
      "step": 143000
    },
    {
      "epoch": 30.058651026392962,
      "grad_norm": 4.666591167449951,
      "learning_rate": 1.9954335986594052e-05,
      "loss": 5.9505,
      "step": 143500
    },
    {
      "epoch": 30.163385002094678,
      "grad_norm": 6.975547790527344,
      "learning_rate": 1.9849602010892335e-05,
      "loss": 5.956,
      "step": 144000
    },
    {
      "epoch": 30.268118977796398,
      "grad_norm": 4.687684535980225,
      "learning_rate": 1.974507750314202e-05,
      "loss": 5.9475,
      "step": 144500
    },
    {
      "epoch": 30.372852953498114,
      "grad_norm": 5.594231605529785,
      "learning_rate": 1.96403435274403e-05,
      "loss": 5.9496,
      "step": 145000
    },
    {
      "epoch": 30.477586929199834,
      "grad_norm": 4.879722595214844,
      "learning_rate": 1.9535609551738584e-05,
      "loss": 5.9577,
      "step": 145500
    },
    {
      "epoch": 30.58232090490155,
      "grad_norm": 5.470447540283203,
      "learning_rate": 1.9430875576036867e-05,
      "loss": 5.9672,
      "step": 146000
    },
    {
      "epoch": 30.687054880603267,
      "grad_norm": 5.818385124206543,
      "learning_rate": 1.932614160033515e-05,
      "loss": 5.9501,
      "step": 146500
    },
    {
      "epoch": 30.791788856304986,
      "grad_norm": 5.907487392425537,
      "learning_rate": 1.9221617092584834e-05,
      "loss": 5.9458,
      "step": 147000
    },
    {
      "epoch": 30.896522832006703,
      "grad_norm": 4.739224433898926,
      "learning_rate": 1.911688311688312e-05,
      "loss": 5.935,
      "step": 147500
    },
    {
      "epoch": 31.00125680770842,
      "grad_norm": 4.57131814956665,
      "learning_rate": 1.90121491411814e-05,
      "loss": 5.945,
      "step": 148000
    },
    {
      "epoch": 31.10599078341014,
      "grad_norm": 5.128586769104004,
      "learning_rate": 1.8907415165479683e-05,
      "loss": 5.9494,
      "step": 148500
    },
    {
      "epoch": 31.210724759111855,
      "grad_norm": 4.871676921844482,
      "learning_rate": 1.880289065772937e-05,
      "loss": 5.9415,
      "step": 149000
    },
    {
      "epoch": 31.315458734813575,
      "grad_norm": 5.380068778991699,
      "learning_rate": 1.8698156682027652e-05,
      "loss": 5.939,
      "step": 149500
    },
    {
      "epoch": 31.42019271051529,
      "grad_norm": 5.430812835693359,
      "learning_rate": 1.8593422706325932e-05,
      "loss": 5.9276,
      "step": 150000
    },
    {
      "epoch": 31.524926686217007,
      "grad_norm": 4.7710442543029785,
      "learning_rate": 1.8488688730624215e-05,
      "loss": 5.9413,
      "step": 150500
    },
    {
      "epoch": 31.629660661918727,
      "grad_norm": 5.183919906616211,
      "learning_rate": 1.8383954754922498e-05,
      "loss": 5.9257,
      "step": 151000
    },
    {
      "epoch": 31.734394637620444,
      "grad_norm": 4.851598739624023,
      "learning_rate": 1.827943024717218e-05,
      "loss": 5.9251,
      "step": 151500
    },
    {
      "epoch": 31.839128613322163,
      "grad_norm": 4.835882663726807,
      "learning_rate": 1.8174696271470464e-05,
      "loss": 5.92,
      "step": 152000
    },
    {
      "epoch": 31.94386258902388,
      "grad_norm": 5.428823947906494,
      "learning_rate": 1.8069962295768748e-05,
      "loss": 5.9146,
      "step": 152500
    },
    {
      "epoch": 32.0485965647256,
      "grad_norm": 6.11329984664917,
      "learning_rate": 1.796522832006703e-05,
      "loss": 5.9179,
      "step": 153000
    },
    {
      "epoch": 32.15333054042731,
      "grad_norm": 4.836859226226807,
      "learning_rate": 1.7860703812316717e-05,
      "loss": 5.9189,
      "step": 153500
    },
    {
      "epoch": 32.25806451612903,
      "grad_norm": 4.598475456237793,
      "learning_rate": 1.7755969836615e-05,
      "loss": 5.9207,
      "step": 154000
    },
    {
      "epoch": 32.36279849183075,
      "grad_norm": 4.638394832611084,
      "learning_rate": 1.765123586091328e-05,
      "loss": 5.915,
      "step": 154500
    },
    {
      "epoch": 32.467532467532465,
      "grad_norm": 5.637279987335205,
      "learning_rate": 1.7546501885211563e-05,
      "loss": 5.9228,
      "step": 155000
    },
    {
      "epoch": 32.572266443234184,
      "grad_norm": 4.516068458557129,
      "learning_rate": 1.7441767909509846e-05,
      "loss": 5.9322,
      "step": 155500
    },
    {
      "epoch": 32.677000418935904,
      "grad_norm": 4.652084827423096,
      "learning_rate": 1.7337243401759533e-05,
      "loss": 5.9395,
      "step": 156000
    },
    {
      "epoch": 32.78173439463762,
      "grad_norm": 5.667607307434082,
      "learning_rate": 1.7232509426057812e-05,
      "loss": 5.9172,
      "step": 156500
    },
    {
      "epoch": 32.88646837033934,
      "grad_norm": 4.980391025543213,
      "learning_rate": 1.7127775450356095e-05,
      "loss": 5.934,
      "step": 157000
    },
    {
      "epoch": 32.99120234604106,
      "grad_norm": 4.39646053314209,
      "learning_rate": 1.702304147465438e-05,
      "loss": 5.914,
      "step": 157500
    },
    {
      "epoch": 33.095936321742776,
      "grad_norm": 5.263533115386963,
      "learning_rate": 1.6918516966904062e-05,
      "loss": 5.9094,
      "step": 158000
    },
    {
      "epoch": 33.20067029744449,
      "grad_norm": 4.661126136779785,
      "learning_rate": 1.6813782991202348e-05,
      "loss": 5.9189,
      "step": 158500
    },
    {
      "epoch": 33.30540427314621,
      "grad_norm": 4.935306549072266,
      "learning_rate": 1.670904901550063e-05,
      "loss": 5.8944,
      "step": 159000
    },
    {
      "epoch": 33.41013824884793,
      "grad_norm": 5.82065486907959,
      "learning_rate": 1.660431503979891e-05,
      "loss": 5.892,
      "step": 159500
    },
    {
      "epoch": 33.51487222454964,
      "grad_norm": 4.6220927238464355,
      "learning_rate": 1.6499581064097194e-05,
      "loss": 5.9022,
      "step": 160000
    },
    {
      "epoch": 33.61960620025136,
      "grad_norm": 5.109046936035156,
      "learning_rate": 1.639505655634688e-05,
      "loss": 5.8933,
      "step": 160500
    },
    {
      "epoch": 33.72434017595308,
      "grad_norm": 5.230437278747559,
      "learning_rate": 1.6290322580645164e-05,
      "loss": 5.9107,
      "step": 161000
    },
    {
      "epoch": 33.829074151654794,
      "grad_norm": 6.466080188751221,
      "learning_rate": 1.6185588604943443e-05,
      "loss": 5.9077,
      "step": 161500
    },
    {
      "epoch": 33.933808127356514,
      "grad_norm": 4.655428409576416,
      "learning_rate": 1.6080854629241726e-05,
      "loss": 5.9186,
      "step": 162000
    },
    {
      "epoch": 34.038542103058234,
      "grad_norm": 6.136354923248291,
      "learning_rate": 1.597612065354001e-05,
      "loss": 5.8815,
      "step": 162500
    },
    {
      "epoch": 34.143276078759946,
      "grad_norm": 5.668376445770264,
      "learning_rate": 1.587138667783829e-05,
      "loss": 5.8872,
      "step": 163000
    },
    {
      "epoch": 34.248010054461666,
      "grad_norm": 5.579314708709717,
      "learning_rate": 1.5766862170087976e-05,
      "loss": 5.8966,
      "step": 163500
    },
    {
      "epoch": 34.352744030163386,
      "grad_norm": 5.474893569946289,
      "learning_rate": 1.5662128194386262e-05,
      "loss": 5.8865,
      "step": 164000
    },
    {
      "epoch": 34.457478005865106,
      "grad_norm": 4.853377342224121,
      "learning_rate": 1.5557394218684542e-05,
      "loss": 5.8718,
      "step": 164500
    },
    {
      "epoch": 34.56221198156682,
      "grad_norm": 4.8684563636779785,
      "learning_rate": 1.5452660242982825e-05,
      "loss": 5.8896,
      "step": 165000
    },
    {
      "epoch": 34.66694595726854,
      "grad_norm": 4.851233959197998,
      "learning_rate": 1.5347926267281108e-05,
      "loss": 5.8908,
      "step": 165500
    },
    {
      "epoch": 34.77167993297026,
      "grad_norm": 6.647628307342529,
      "learning_rate": 1.524319229157939e-05,
      "loss": 5.8972,
      "step": 166000
    },
    {
      "epoch": 34.87641390867197,
      "grad_norm": 5.826745986938477,
      "learning_rate": 1.5138667783829074e-05,
      "loss": 5.8852,
      "step": 166500
    },
    {
      "epoch": 34.98114788437369,
      "grad_norm": 5.109675407409668,
      "learning_rate": 1.5033933808127357e-05,
      "loss": 5.8971,
      "step": 167000
    },
    {
      "epoch": 35.08588186007541,
      "grad_norm": 5.743017196655273,
      "learning_rate": 1.4929199832425639e-05,
      "loss": 5.8892,
      "step": 167500
    },
    {
      "epoch": 35.19061583577712,
      "grad_norm": 4.862270355224609,
      "learning_rate": 1.482446585672392e-05,
      "loss": 5.8976,
      "step": 168000
    },
    {
      "epoch": 35.29534981147884,
      "grad_norm": 5.532686233520508,
      "learning_rate": 1.4719941348973607e-05,
      "loss": 5.8716,
      "step": 168500
    },
    {
      "epoch": 35.40008378718056,
      "grad_norm": 5.498019695281982,
      "learning_rate": 1.4615207373271891e-05,
      "loss": 5.8768,
      "step": 169000
    },
    {
      "epoch": 35.504817762882276,
      "grad_norm": 4.7324042320251465,
      "learning_rate": 1.4510473397570173e-05,
      "loss": 5.8739,
      "step": 169500
    },
    {
      "epoch": 35.609551738583995,
      "grad_norm": 4.973413944244385,
      "learning_rate": 1.4405739421868456e-05,
      "loss": 5.8801,
      "step": 170000
    },
    {
      "epoch": 35.714285714285715,
      "grad_norm": 4.977658271789551,
      "learning_rate": 1.430121491411814e-05,
      "loss": 5.8644,
      "step": 170500
    },
    {
      "epoch": 35.819019689987435,
      "grad_norm": 5.551715850830078,
      "learning_rate": 1.4196480938416424e-05,
      "loss": 5.8789,
      "step": 171000
    },
    {
      "epoch": 35.92375366568915,
      "grad_norm": 5.135740756988525,
      "learning_rate": 1.4091746962714705e-05,
      "loss": 5.8862,
      "step": 171500
    },
    {
      "epoch": 36.02848764139087,
      "grad_norm": 5.068655967712402,
      "learning_rate": 1.3987012987012987e-05,
      "loss": 5.879,
      "step": 172000
    },
    {
      "epoch": 36.13322161709259,
      "grad_norm": 5.393857479095459,
      "learning_rate": 1.388227901131127e-05,
      "loss": 5.8544,
      "step": 172500
    },
    {
      "epoch": 36.2379555927943,
      "grad_norm": 5.854538440704346,
      "learning_rate": 1.3777545035609551e-05,
      "loss": 5.8824,
      "step": 173000
    },
    {
      "epoch": 36.34268956849602,
      "grad_norm": 5.566401481628418,
      "learning_rate": 1.3673020527859238e-05,
      "loss": 5.8587,
      "step": 173500
    },
    {
      "epoch": 36.44742354419774,
      "grad_norm": 6.091250896453857,
      "learning_rate": 1.3568286552157519e-05,
      "loss": 5.8624,
      "step": 174000
    },
    {
      "epoch": 36.55215751989945,
      "grad_norm": 4.826417922973633,
      "learning_rate": 1.3463552576455804e-05,
      "loss": 5.879,
      "step": 174500
    },
    {
      "epoch": 36.65689149560117,
      "grad_norm": 5.28770637512207,
      "learning_rate": 1.3358818600754087e-05,
      "loss": 5.8632,
      "step": 175000
    },
    {
      "epoch": 36.76162547130289,
      "grad_norm": 5.072086811065674,
      "learning_rate": 1.3254084625052368e-05,
      "loss": 5.8698,
      "step": 175500
    },
    {
      "epoch": 36.866359447004605,
      "grad_norm": 6.194067001342773,
      "learning_rate": 1.3149560117302053e-05,
      "loss": 5.8701,
      "step": 176000
    },
    {
      "epoch": 36.971093422706325,
      "grad_norm": 5.250491142272949,
      "learning_rate": 1.3044826141600336e-05,
      "loss": 5.855,
      "step": 176500
    },
    {
      "epoch": 37.075827398408045,
      "grad_norm": 4.9726080894470215,
      "learning_rate": 1.2940092165898618e-05,
      "loss": 5.8613,
      "step": 177000
    },
    {
      "epoch": 37.180561374109764,
      "grad_norm": 5.526548385620117,
      "learning_rate": 1.28353581901969e-05,
      "loss": 5.8519,
      "step": 177500
    },
    {
      "epoch": 37.28529534981148,
      "grad_norm": 5.5989861488342285,
      "learning_rate": 1.2730624214495182e-05,
      "loss": 5.8642,
      "step": 178000
    },
    {
      "epoch": 37.3900293255132,
      "grad_norm": 5.138686180114746,
      "learning_rate": 1.2625890238793465e-05,
      "loss": 5.852,
      "step": 178500
    },
    {
      "epoch": 37.49476330121492,
      "grad_norm": 5.0514326095581055,
      "learning_rate": 1.252115626309175e-05,
      "loss": 5.8484,
      "step": 179000
    },
    {
      "epoch": 37.59949727691663,
      "grad_norm": 4.9300360679626465,
      "learning_rate": 1.241642228739003e-05,
      "loss": 5.849,
      "step": 179500
    },
    {
      "epoch": 37.70423125261835,
      "grad_norm": 5.487224102020264,
      "learning_rate": 1.2311897779639716e-05,
      "loss": 5.8562,
      "step": 180000
    },
    {
      "epoch": 37.80896522832007,
      "grad_norm": 5.826539516448975,
      "learning_rate": 1.2207163803937999e-05,
      "loss": 5.8665,
      "step": 180500
    },
    {
      "epoch": 37.91369920402178,
      "grad_norm": 5.733819961547852,
      "learning_rate": 1.2102639296187684e-05,
      "loss": 5.8569,
      "step": 181000
    },
    {
      "epoch": 38.0184331797235,
      "grad_norm": 4.917960166931152,
      "learning_rate": 1.1997905320485967e-05,
      "loss": 5.8395,
      "step": 181500
    },
    {
      "epoch": 38.12316715542522,
      "grad_norm": 5.337119102478027,
      "learning_rate": 1.1893171344784248e-05,
      "loss": 5.854,
      "step": 182000
    },
    {
      "epoch": 38.227901131126934,
      "grad_norm": 5.299139022827148,
      "learning_rate": 1.178843736908253e-05,
      "loss": 5.8433,
      "step": 182500
    },
    {
      "epoch": 38.332635106828654,
      "grad_norm": 5.900153160095215,
      "learning_rate": 1.1683703393380813e-05,
      "loss": 5.8535,
      "step": 183000
    },
    {
      "epoch": 38.437369082530374,
      "grad_norm": 6.776584625244141,
      "learning_rate": 1.15791788856305e-05,
      "loss": 5.8454,
      "step": 183500
    },
    {
      "epoch": 38.542103058232094,
      "grad_norm": 6.258368015289307,
      "learning_rate": 1.1474444909928783e-05,
      "loss": 5.8354,
      "step": 184000
    },
    {
      "epoch": 38.64683703393381,
      "grad_norm": 5.288670539855957,
      "learning_rate": 1.1369710934227064e-05,
      "loss": 5.8458,
      "step": 184500
    },
    {
      "epoch": 38.751571009635526,
      "grad_norm": 5.596650123596191,
      "learning_rate": 1.1264976958525345e-05,
      "loss": 5.8387,
      "step": 185000
    },
    {
      "epoch": 38.856304985337246,
      "grad_norm": 5.121638774871826,
      "learning_rate": 1.1160452450775032e-05,
      "loss": 5.8268,
      "step": 185500
    },
    {
      "epoch": 38.96103896103896,
      "grad_norm": 4.5758442878723145,
      "learning_rate": 1.1055718475073313e-05,
      "loss": 5.83,
      "step": 186000
    },
    {
      "epoch": 39.06577293674068,
      "grad_norm": 5.161282539367676,
      "learning_rate": 1.0950984499371596e-05,
      "loss": 5.8544,
      "step": 186500
    },
    {
      "epoch": 39.1705069124424,
      "grad_norm": 4.628884315490723,
      "learning_rate": 1.084625052366988e-05,
      "loss": 5.8474,
      "step": 187000
    },
    {
      "epoch": 39.27524088814411,
      "grad_norm": 5.854598045349121,
      "learning_rate": 1.074151654796816e-05,
      "loss": 5.8501,
      "step": 187500
    },
    {
      "epoch": 39.37997486384583,
      "grad_norm": 5.315525054931641,
      "learning_rate": 1.0636782572266444e-05,
      "loss": 5.8265,
      "step": 188000
    },
    {
      "epoch": 39.48470883954755,
      "grad_norm": 6.078185081481934,
      "learning_rate": 1.0532048596564727e-05,
      "loss": 5.8419,
      "step": 188500
    },
    {
      "epoch": 39.589442815249264,
      "grad_norm": 5.223086357116699,
      "learning_rate": 1.0427524088814412e-05,
      "loss": 5.8197,
      "step": 189000
    },
    {
      "epoch": 39.69417679095098,
      "grad_norm": 5.235757827758789,
      "learning_rate": 1.0322790113112695e-05,
      "loss": 5.8245,
      "step": 189500
    },
    {
      "epoch": 39.7989107666527,
      "grad_norm": 5.124643325805664,
      "learning_rate": 1.0218056137410976e-05,
      "loss": 5.839,
      "step": 190000
    },
    {
      "epoch": 39.90364474235442,
      "grad_norm": 5.613321304321289,
      "learning_rate": 1.011332216170926e-05,
      "loss": 5.844,
      "step": 190500
    },
    {
      "epoch": 40.008378718056136,
      "grad_norm": 5.873430252075195,
      "learning_rate": 1.0008588186007542e-05,
      "loss": 5.837,
      "step": 191000
    },
    {
      "epoch": 40.113112693757856,
      "grad_norm": 5.089309215545654,
      "learning_rate": 9.903854210305824e-06,
      "loss": 5.8384,
      "step": 191500
    },
    {
      "epoch": 40.217846669459576,
      "grad_norm": 7.3569817543029785,
      "learning_rate": 9.79932970255551e-06,
      "loss": 5.8232,
      "step": 192000
    },
    {
      "epoch": 40.32258064516129,
      "grad_norm": 6.024489402770996,
      "learning_rate": 9.694595726853792e-06,
      "loss": 5.8292,
      "step": 192500
    },
    {
      "epoch": 40.42731462086301,
      "grad_norm": 5.7150983810424805,
      "learning_rate": 9.589861751152073e-06,
      "loss": 5.8614,
      "step": 193000
    },
    {
      "epoch": 40.53204859656473,
      "grad_norm": 4.717107772827148,
      "learning_rate": 9.485127775450356e-06,
      "loss": 5.8092,
      "step": 193500
    },
    {
      "epoch": 40.63678257226644,
      "grad_norm": 4.9722490310668945,
      "learning_rate": 9.380603267700043e-06,
      "loss": 5.8231,
      "step": 194000
    },
    {
      "epoch": 40.74151654796816,
      "grad_norm": 5.593094825744629,
      "learning_rate": 9.275869291998326e-06,
      "loss": 5.8339,
      "step": 194500
    },
    {
      "epoch": 40.84625052366988,
      "grad_norm": 5.731310844421387,
      "learning_rate": 9.171135316296607e-06,
      "loss": 5.8381,
      "step": 195000
    },
    {
      "epoch": 40.95098449937159,
      "grad_norm": 5.072065353393555,
      "learning_rate": 9.066401340594889e-06,
      "loss": 5.8367,
      "step": 195500
    },
    {
      "epoch": 41.05571847507331,
      "grad_norm": 5.219040870666504,
      "learning_rate": 8.961667364893172e-06,
      "loss": 5.8234,
      "step": 196000
    },
    {
      "epoch": 41.16045245077503,
      "grad_norm": 5.844238758087158,
      "learning_rate": 8.856933389191455e-06,
      "loss": 5.8347,
      "step": 196500
    },
    {
      "epoch": 41.26518642647675,
      "grad_norm": 6.088447093963623,
      "learning_rate": 8.752199413489736e-06,
      "loss": 5.8178,
      "step": 197000
    },
    {
      "epoch": 41.369920402178465,
      "grad_norm": 5.14108943939209,
      "learning_rate": 8.647465437788019e-06,
      "loss": 5.8248,
      "step": 197500
    },
    {
      "epoch": 41.474654377880185,
      "grad_norm": 5.424249172210693,
      "learning_rate": 8.542940930037704e-06,
      "loss": 5.8113,
      "step": 198000
    },
    {
      "epoch": 41.579388353581905,
      "grad_norm": 4.888121604919434,
      "learning_rate": 8.43841642228739e-06,
      "loss": 5.8111,
      "step": 198500
    },
    {
      "epoch": 41.68412232928362,
      "grad_norm": 4.9909515380859375,
      "learning_rate": 8.333682446585672e-06,
      "loss": 5.8276,
      "step": 199000
    },
    {
      "epoch": 41.78885630498534,
      "grad_norm": 5.032175540924072,
      "learning_rate": 8.228948470883955e-06,
      "loss": 5.8332,
      "step": 199500
    },
    {
      "epoch": 41.89359028068706,
      "grad_norm": 5.116880416870117,
      "learning_rate": 8.124214495182238e-06,
      "loss": 5.8233,
      "step": 200000
    },
    {
      "epoch": 41.99832425638877,
      "grad_norm": 5.235647678375244,
      "learning_rate": 8.019689987431923e-06,
      "loss": 5.8297,
      "step": 200500
    },
    {
      "epoch": 42.10305823209049,
      "grad_norm": 5.445380210876465,
      "learning_rate": 7.914956011730206e-06,
      "loss": 5.8159,
      "step": 201000
    },
    {
      "epoch": 42.20779220779221,
      "grad_norm": 4.979036331176758,
      "learning_rate": 7.810222036028488e-06,
      "loss": 5.809,
      "step": 201500
    },
    {
      "epoch": 42.31252618349392,
      "grad_norm": 5.359362602233887,
      "learning_rate": 7.70548806032677e-06,
      "loss": 5.8346,
      "step": 202000
    },
    {
      "epoch": 42.41726015919564,
      "grad_norm": 5.264519214630127,
      "learning_rate": 7.600754084625053e-06,
      "loss": 5.8089,
      "step": 202500
    },
    {
      "epoch": 42.52199413489736,
      "grad_norm": 5.985982894897461,
      "learning_rate": 7.496020108923335e-06,
      "loss": 5.8192,
      "step": 203000
    },
    {
      "epoch": 42.626728110599075,
      "grad_norm": 5.505626201629639,
      "learning_rate": 7.391286133221617e-06,
      "loss": 5.8095,
      "step": 203500
    },
    {
      "epoch": 42.731462086300795,
      "grad_norm": 5.069738388061523,
      "learning_rate": 7.286552157519899e-06,
      "loss": 5.8186,
      "step": 204000
    },
    {
      "epoch": 42.836196062002514,
      "grad_norm": 6.004745960235596,
      "learning_rate": 7.182027649769586e-06,
      "loss": 5.8136,
      "step": 204500
    },
    {
      "epoch": 42.940930037704234,
      "grad_norm": 6.299502372741699,
      "learning_rate": 7.077293674067868e-06,
      "loss": 5.8213,
      "step": 205000
    },
    {
      "epoch": 43.04566401340595,
      "grad_norm": 6.302718162536621,
      "learning_rate": 6.97255969836615e-06,
      "loss": 5.8075,
      "step": 205500
    },
    {
      "epoch": 43.15039798910767,
      "grad_norm": 5.921250343322754,
      "learning_rate": 6.867825722664433e-06,
      "loss": 5.8276,
      "step": 206000
    },
    {
      "epoch": 43.25513196480939,
      "grad_norm": 5.123110771179199,
      "learning_rate": 6.763091746962715e-06,
      "loss": 5.7965,
      "step": 206500
    },
    {
      "epoch": 43.3598659405111,
      "grad_norm": 5.187294006347656,
      "learning_rate": 6.658357771260998e-06,
      "loss": 5.8137,
      "step": 207000
    },
    {
      "epoch": 43.46459991621282,
      "grad_norm": 5.407510757446289,
      "learning_rate": 6.55362379555928e-06,
      "loss": 5.8305,
      "step": 207500
    },
    {
      "epoch": 43.56933389191454,
      "grad_norm": 5.892600059509277,
      "learning_rate": 6.449099287808966e-06,
      "loss": 5.8167,
      "step": 208000
    },
    {
      "epoch": 43.67406786761625,
      "grad_norm": 5.39382266998291,
      "learning_rate": 6.344365312107248e-06,
      "loss": 5.8185,
      "step": 208500
    },
    {
      "epoch": 43.77880184331797,
      "grad_norm": 5.608034133911133,
      "learning_rate": 6.23963133640553e-06,
      "loss": 5.8051,
      "step": 209000
    },
    {
      "epoch": 43.88353581901969,
      "grad_norm": 6.069722652435303,
      "learning_rate": 6.1348973607038125e-06,
      "loss": 5.8101,
      "step": 209500
    },
    {
      "epoch": 43.988269794721404,
      "grad_norm": 5.938599109649658,
      "learning_rate": 6.0301633850020955e-06,
      "loss": 5.807,
      "step": 210000
    },
    {
      "epoch": 44.093003770423124,
      "grad_norm": 5.808456897735596,
      "learning_rate": 5.925429409300378e-06,
      "loss": 5.8246,
      "step": 210500
    },
    {
      "epoch": 44.197737746124844,
      "grad_norm": 5.229996681213379,
      "learning_rate": 5.820904901550063e-06,
      "loss": 5.8029,
      "step": 211000
    },
    {
      "epoch": 44.302471721826564,
      "grad_norm": 5.295706748962402,
      "learning_rate": 5.716170925848346e-06,
      "loss": 5.8094,
      "step": 211500
    },
    {
      "epoch": 44.407205697528276,
      "grad_norm": 5.649194240570068,
      "learning_rate": 5.611436950146628e-06,
      "loss": 5.811,
      "step": 212000
    },
    {
      "epoch": 44.511939673229996,
      "grad_norm": 6.5928521156311035,
      "learning_rate": 5.50670297444491e-06,
      "loss": 5.7974,
      "step": 212500
    },
    {
      "epoch": 44.616673648931716,
      "grad_norm": 6.246605396270752,
      "learning_rate": 5.401968998743192e-06,
      "loss": 5.8011,
      "step": 213000
    },
    {
      "epoch": 44.72140762463343,
      "grad_norm": 5.312093734741211,
      "learning_rate": 5.2972350230414745e-06,
      "loss": 5.7819,
      "step": 213500
    },
    {
      "epoch": 44.82614160033515,
      "grad_norm": 5.348554611206055,
      "learning_rate": 5.19271051529116e-06,
      "loss": 5.8027,
      "step": 214000
    },
    {
      "epoch": 44.93087557603687,
      "grad_norm": 5.95352029800415,
      "learning_rate": 5.087976539589443e-06,
      "loss": 5.8046,
      "step": 214500
    },
    {
      "epoch": 45.03560955173858,
      "grad_norm": 5.978014945983887,
      "learning_rate": 4.983242563887726e-06,
      "loss": 5.8021,
      "step": 215000
    },
    {
      "epoch": 45.1403435274403,
      "grad_norm": 5.595849990844727,
      "learning_rate": 4.878508588186008e-06,
      "loss": 5.7996,
      "step": 215500
    },
    {
      "epoch": 45.24507750314202,
      "grad_norm": 5.570345401763916,
      "learning_rate": 4.77377461248429e-06,
      "loss": 5.7973,
      "step": 216000
    },
    {
      "epoch": 45.34981147884373,
      "grad_norm": 5.320748805999756,
      "learning_rate": 4.669040636782573e-06,
      "loss": 5.7886,
      "step": 216500
    },
    {
      "epoch": 45.45454545454545,
      "grad_norm": 4.676185607910156,
      "learning_rate": 4.564516129032258e-06,
      "loss": 5.7874,
      "step": 217000
    },
    {
      "epoch": 45.55927943024717,
      "grad_norm": 5.7768473625183105,
      "learning_rate": 4.45978215333054e-06,
      "loss": 5.7875,
      "step": 217500
    },
    {
      "epoch": 45.66401340594889,
      "grad_norm": 5.668895244598389,
      "learning_rate": 4.355048177628823e-06,
      "loss": 5.8121,
      "step": 218000
    },
    {
      "epoch": 45.768747381650606,
      "grad_norm": 5.033557891845703,
      "learning_rate": 4.2503142019271055e-06,
      "loss": 5.8137,
      "step": 218500
    },
    {
      "epoch": 45.873481357352325,
      "grad_norm": 6.15772819519043,
      "learning_rate": 4.145580226225388e-06,
      "loss": 5.8137,
      "step": 219000
    },
    {
      "epoch": 45.978215333054045,
      "grad_norm": 6.617910861968994,
      "learning_rate": 4.04084625052367e-06,
      "loss": 5.8079,
      "step": 219500
    },
    {
      "epoch": 46.08294930875576,
      "grad_norm": 5.210205554962158,
      "learning_rate": 3.936112274821952e-06,
      "loss": 5.8189,
      "step": 220000
    },
    {
      "epoch": 46.18768328445748,
      "grad_norm": 5.630945205688477,
      "learning_rate": 3.831587767071639e-06,
      "loss": 5.795,
      "step": 220500
    },
    {
      "epoch": 46.2924172601592,
      "grad_norm": 5.8690032958984375,
      "learning_rate": 3.7268537913699205e-06,
      "loss": 5.8019,
      "step": 221000
    },
    {
      "epoch": 46.39715123586091,
      "grad_norm": 5.787112712860107,
      "learning_rate": 3.6221198156682027e-06,
      "loss": 5.7901,
      "step": 221500
    },
    {
      "epoch": 46.50188521156263,
      "grad_norm": 5.568469524383545,
      "learning_rate": 3.5173858399664853e-06,
      "loss": 5.803,
      "step": 222000
    },
    {
      "epoch": 46.60661918726435,
      "grad_norm": 5.671326637268066,
      "learning_rate": 3.4126518642647675e-06,
      "loss": 5.7931,
      "step": 222500
    },
    {
      "epoch": 46.71135316296606,
      "grad_norm": 5.4085307121276855,
      "learning_rate": 3.30791788856305e-06,
      "loss": 5.7903,
      "step": 223000
    },
    {
      "epoch": 46.81608713866778,
      "grad_norm": 5.7440571784973145,
      "learning_rate": 3.203393380812736e-06,
      "loss": 5.7897,
      "step": 223500
    },
    {
      "epoch": 46.9208211143695,
      "grad_norm": 5.144542217254639,
      "learning_rate": 3.098659405111018e-06,
      "loss": 5.8145,
      "step": 224000
    },
    {
      "epoch": 47.02555509007122,
      "grad_norm": 5.842213153839111,
      "learning_rate": 2.9939254294093008e-06,
      "loss": 5.8046,
      "step": 224500
    },
    {
      "epoch": 47.130289065772935,
      "grad_norm": 6.161410331726074,
      "learning_rate": 2.8891914537075826e-06,
      "loss": 5.7965,
      "step": 225000
    },
    {
      "epoch": 47.235023041474655,
      "grad_norm": 6.173724174499512,
      "learning_rate": 2.784457478005865e-06,
      "loss": 5.7854,
      "step": 225500
    },
    {
      "epoch": 47.339757017176375,
      "grad_norm": 5.132796287536621,
      "learning_rate": 2.679932970255551e-06,
      "loss": 5.791,
      "step": 226000
    },
    {
      "epoch": 47.44449099287809,
      "grad_norm": 6.053417205810547,
      "learning_rate": 2.5751989945538332e-06,
      "loss": 5.7935,
      "step": 226500
    },
    {
      "epoch": 47.54922496857981,
      "grad_norm": 5.08466911315918,
      "learning_rate": 2.470465018852116e-06,
      "loss": 5.8006,
      "step": 227000
    },
    {
      "epoch": 47.65395894428153,
      "grad_norm": 6.060305595397949,
      "learning_rate": 2.365731043150398e-06,
      "loss": 5.7839,
      "step": 227500
    },
    {
      "epoch": 47.75869291998324,
      "grad_norm": 5.808520317077637,
      "learning_rate": 2.2609970674486806e-06,
      "loss": 5.7933,
      "step": 228000
    },
    {
      "epoch": 47.86342689568496,
      "grad_norm": 5.413971424102783,
      "learning_rate": 2.156472559698366e-06,
      "loss": 5.7985,
      "step": 228500
    },
    {
      "epoch": 47.96816087138668,
      "grad_norm": 6.4786529541015625,
      "learning_rate": 2.0517385839966487e-06,
      "loss": 5.8059,
      "step": 229000
    },
    {
      "epoch": 48.07289484708839,
      "grad_norm": 5.606147289276123,
      "learning_rate": 1.947004608294931e-06,
      "loss": 5.7934,
      "step": 229500
    },
    {
      "epoch": 48.17762882279011,
      "grad_norm": 5.827240467071533,
      "learning_rate": 1.8422706325932133e-06,
      "loss": 5.8007,
      "step": 230000
    },
    {
      "epoch": 48.28236279849183,
      "grad_norm": 5.678854465484619,
      "learning_rate": 1.7375366568914957e-06,
      "loss": 5.8038,
      "step": 230500
    },
    {
      "epoch": 48.38709677419355,
      "grad_norm": 5.42138671875,
      "learning_rate": 1.632802681189778e-06,
      "loss": 5.7976,
      "step": 231000
    },
    {
      "epoch": 48.491830749895264,
      "grad_norm": 6.03090238571167,
      "learning_rate": 1.5280687054880603e-06,
      "loss": 5.7874,
      "step": 231500
    },
    {
      "epoch": 48.596564725596984,
      "grad_norm": 4.660221099853516,
      "learning_rate": 1.4233347297863427e-06,
      "loss": 5.7742,
      "step": 232000
    },
    {
      "epoch": 48.701298701298704,
      "grad_norm": 6.032063961029053,
      "learning_rate": 1.3188102220360285e-06,
      "loss": 5.7866,
      "step": 232500
    },
    {
      "epoch": 48.80603267700042,
      "grad_norm": 6.296925067901611,
      "learning_rate": 1.2142857142857144e-06,
      "loss": 5.7922,
      "step": 233000
    },
    {
      "epoch": 48.91076665270214,
      "grad_norm": 5.2991437911987305,
      "learning_rate": 1.1095517385839968e-06,
      "loss": 5.7873,
      "step": 233500
    },
    {
      "epoch": 49.015500628403856,
      "grad_norm": 6.130777835845947,
      "learning_rate": 1.0048177628822792e-06,
      "loss": 5.8074,
      "step": 234000
    },
    {
      "epoch": 49.12023460410557,
      "grad_norm": 6.305094242095947,
      "learning_rate": 9.000837871805614e-07,
      "loss": 5.7847,
      "step": 234500
    },
    {
      "epoch": 49.22496857980729,
      "grad_norm": 6.156949996948242,
      "learning_rate": 7.953498114788438e-07,
      "loss": 5.7741,
      "step": 235000
    },
    {
      "epoch": 49.32970255550901,
      "grad_norm": 6.475966930389404,
      "learning_rate": 6.908253037285296e-07,
      "loss": 5.8083,
      "step": 235500
    },
    {
      "epoch": 49.43443653121072,
      "grad_norm": 5.687895774841309,
      "learning_rate": 5.86091328026812e-07,
      "loss": 5.7976,
      "step": 236000
    },
    {
      "epoch": 49.53917050691244,
      "grad_norm": 5.615401744842529,
      "learning_rate": 4.813573523250943e-07,
      "loss": 5.7907,
      "step": 236500
    },
    {
      "epoch": 49.64390448261416,
      "grad_norm": 6.177160263061523,
      "learning_rate": 3.7662337662337666e-07,
      "loss": 5.792,
      "step": 237000
    },
    {
      "epoch": 49.74863845831588,
      "grad_norm": 5.604287624359131,
      "learning_rate": 2.71889400921659e-07,
      "loss": 5.7837,
      "step": 237500
    },
    {
      "epoch": 49.853372434017594,
      "grad_norm": 4.826539039611816,
      "learning_rate": 1.673648931713448e-07,
      "loss": 5.7808,
      "step": 238000
    },
    {
      "epoch": 49.95810640971931,
      "grad_norm": 5.6649250984191895,
      "learning_rate": 6.263091746962715e-08,
      "loss": 5.7888,
      "step": 238500
    },
    {
      "epoch": 50.0,
      "step": 238700,
      "total_flos": 5161725447936000.0,
      "train_loss": 6.140905278960581,
      "train_runtime": 7883.0646,
      "train_samples_per_second": 484.444,
      "train_steps_per_second": 30.28
    }
  ],
  "logging_steps": 500,
  "max_steps": 238700,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5161725447936000.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}