{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9975031210986267,
  "eval_steps": 500,
  "global_step": 12800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.007802746566791511,
      "grad_norm": 0.09780355542898178,
      "learning_rate": 4.9923533083645446e-05,
      "loss": 2.1517,
      "step": 50
    },
    {
      "epoch": 0.015605493133583021,
      "grad_norm": 0.1110800951719284,
      "learning_rate": 4.984550561797753e-05,
      "loss": 2.0733,
      "step": 100
    },
    {
      "epoch": 0.023408239700374533,
      "grad_norm": 0.11351309716701508,
      "learning_rate": 4.9767478152309616e-05,
      "loss": 1.9999,
      "step": 150
    },
    {
      "epoch": 0.031210986267166042,
      "grad_norm": 0.12184558063745499,
      "learning_rate": 4.96894506866417e-05,
      "loss": 1.911,
      "step": 200
    },
    {
      "epoch": 0.03901373283395755,
      "grad_norm": 0.12001396715641022,
      "learning_rate": 4.9611423220973786e-05,
      "loss": 1.8957,
      "step": 250
    },
    {
      "epoch": 0.04681647940074907,
      "grad_norm": 0.13359545171260834,
      "learning_rate": 4.9533395755305875e-05,
      "loss": 1.881,
      "step": 300
    },
    {
      "epoch": 0.054619225967540576,
      "grad_norm": 0.177729532122612,
      "learning_rate": 4.9455368289637956e-05,
      "loss": 1.8426,
      "step": 350
    },
    {
      "epoch": 0.062421972534332085,
      "grad_norm": 0.15538156032562256,
      "learning_rate": 4.9377340823970044e-05,
      "loss": 1.8528,
      "step": 400
    },
    {
      "epoch": 0.0702247191011236,
      "grad_norm": 0.1490369588136673,
      "learning_rate": 4.9299313358302126e-05,
      "loss": 1.8604,
      "step": 450
    },
    {
      "epoch": 0.0780274656679151,
      "grad_norm": 0.15821540355682373,
      "learning_rate": 4.9221285892634214e-05,
      "loss": 1.8454,
      "step": 500
    },
    {
      "epoch": 0.08583021223470662,
      "grad_norm": 0.15961939096450806,
      "learning_rate": 4.9143258426966296e-05,
      "loss": 1.8235,
      "step": 550
    },
    {
      "epoch": 0.09363295880149813,
      "grad_norm": 0.18454231321811676,
      "learning_rate": 4.906523096129838e-05,
      "loss": 1.8245,
      "step": 600
    },
    {
      "epoch": 0.10143570536828964,
      "grad_norm": 0.19588346779346466,
      "learning_rate": 4.8987203495630466e-05,
      "loss": 1.8332,
      "step": 650
    },
    {
      "epoch": 0.10923845193508115,
      "grad_norm": 0.20523308217525482,
      "learning_rate": 4.890917602996255e-05,
      "loss": 1.8176,
      "step": 700
    },
    {
      "epoch": 0.11704119850187265,
      "grad_norm": 0.2048981934785843,
      "learning_rate": 4.883114856429463e-05,
      "loss": 1.7751,
      "step": 750
    },
    {
      "epoch": 0.12484394506866417,
      "grad_norm": 0.19311609864234924,
      "learning_rate": 4.875312109862672e-05,
      "loss": 1.8034,
      "step": 800
    },
    {
      "epoch": 0.13264669163545567,
      "grad_norm": 0.21123754978179932,
      "learning_rate": 4.86750936329588e-05,
      "loss": 1.7859,
      "step": 850
    },
    {
      "epoch": 0.1404494382022472,
      "grad_norm": 0.270113080739975,
      "learning_rate": 4.859706616729089e-05,
      "loss": 1.8121,
      "step": 900
    },
    {
      "epoch": 0.1482521847690387,
      "grad_norm": 0.23597899079322815,
      "learning_rate": 4.8519038701622975e-05,
      "loss": 1.8213,
      "step": 950
    },
    {
      "epoch": 0.1560549313358302,
      "grad_norm": 0.2787221372127533,
      "learning_rate": 4.844101123595506e-05,
      "loss": 1.8048,
      "step": 1000
    },
    {
      "epoch": 0.16385767790262173,
      "grad_norm": 0.2505972981452942,
      "learning_rate": 4.8362983770287145e-05,
      "loss": 1.7799,
      "step": 1050
    },
    {
      "epoch": 0.17166042446941324,
      "grad_norm": 0.2511955797672272,
      "learning_rate": 4.828495630461923e-05,
      "loss": 1.7758,
      "step": 1100
    },
    {
      "epoch": 0.17946317103620474,
      "grad_norm": 0.2740708291530609,
      "learning_rate": 4.8206928838951315e-05,
      "loss": 1.7817,
      "step": 1150
    },
    {
      "epoch": 0.18726591760299627,
      "grad_norm": 0.30172184109687805,
      "learning_rate": 4.81289013732834e-05,
      "loss": 1.7665,
      "step": 1200
    },
    {
      "epoch": 0.19506866416978777,
      "grad_norm": 0.2734954059123993,
      "learning_rate": 4.8050873907615485e-05,
      "loss": 1.7673,
      "step": 1250
    },
    {
      "epoch": 0.20287141073657927,
      "grad_norm": 0.2837677597999573,
      "learning_rate": 4.797284644194757e-05,
      "loss": 1.7539,
      "step": 1300
    },
    {
      "epoch": 0.21067415730337077,
      "grad_norm": 0.287616491317749,
      "learning_rate": 4.7894818976279655e-05,
      "loss": 1.7845,
      "step": 1350
    },
    {
      "epoch": 0.2184769038701623,
      "grad_norm": 0.35649779438972473,
      "learning_rate": 4.7816791510611737e-05,
      "loss": 1.7629,
      "step": 1400
    },
    {
      "epoch": 0.2262796504369538,
      "grad_norm": 0.3466143012046814,
      "learning_rate": 4.7738764044943825e-05,
      "loss": 1.7637,
      "step": 1450
    },
    {
      "epoch": 0.2340823970037453,
      "grad_norm": 0.29961591958999634,
      "learning_rate": 4.7660736579275906e-05,
      "loss": 1.7775,
      "step": 1500
    },
    {
      "epoch": 0.24188514357053684,
      "grad_norm": 0.3261962831020355,
      "learning_rate": 4.7582709113607995e-05,
      "loss": 1.7651,
      "step": 1550
    },
    {
      "epoch": 0.24968789013732834,
      "grad_norm": 0.31850898265838623,
      "learning_rate": 4.7504681647940076e-05,
      "loss": 1.7601,
      "step": 1600
    },
    {
      "epoch": 0.25749063670411987,
      "grad_norm": 0.3334502577781677,
      "learning_rate": 4.7426654182272165e-05,
      "loss": 1.7783,
      "step": 1650
    },
    {
      "epoch": 0.26529338327091134,
      "grad_norm": 0.3199051320552826,
      "learning_rate": 4.7348626716604246e-05,
      "loss": 1.7492,
      "step": 1700
    },
    {
      "epoch": 0.2730961298377029,
      "grad_norm": 0.3381502628326416,
      "learning_rate": 4.7270599250936335e-05,
      "loss": 1.7592,
      "step": 1750
    },
    {
      "epoch": 0.2808988764044944,
      "grad_norm": 0.33800897002220154,
      "learning_rate": 4.7192571785268416e-05,
      "loss": 1.7403,
      "step": 1800
    },
    {
      "epoch": 0.2887016229712859,
      "grad_norm": 0.3942790925502777,
      "learning_rate": 4.71145443196005e-05,
      "loss": 1.7955,
      "step": 1850
    },
    {
      "epoch": 0.2965043695380774,
      "grad_norm": 0.3514516353607178,
      "learning_rate": 4.7036516853932586e-05,
      "loss": 1.7453,
      "step": 1900
    },
    {
      "epoch": 0.30430711610486894,
      "grad_norm": 0.3649444878101349,
      "learning_rate": 4.695848938826467e-05,
      "loss": 1.7584,
      "step": 1950
    },
    {
      "epoch": 0.3121098626716604,
      "grad_norm": 0.33589330315589905,
      "learning_rate": 4.6880461922596756e-05,
      "loss": 1.7369,
      "step": 2000
    },
    {
      "epoch": 0.31991260923845194,
      "grad_norm": 0.3481082320213318,
      "learning_rate": 4.680243445692884e-05,
      "loss": 1.7281,
      "step": 2050
    },
    {
      "epoch": 0.32771535580524347,
      "grad_norm": 0.34206053614616394,
      "learning_rate": 4.6724406991260926e-05,
      "loss": 1.7266,
      "step": 2100
    },
    {
      "epoch": 0.33551810237203494,
      "grad_norm": 0.3519572615623474,
      "learning_rate": 4.664637952559301e-05,
      "loss": 1.7365,
      "step": 2150
    },
    {
      "epoch": 0.3433208489388265,
      "grad_norm": 0.35082659125328064,
      "learning_rate": 4.6568352059925096e-05,
      "loss": 1.7411,
      "step": 2200
    },
    {
      "epoch": 0.351123595505618,
      "grad_norm": 0.3652777373790741,
      "learning_rate": 4.649032459425718e-05,
      "loss": 1.7466,
      "step": 2250
    },
    {
      "epoch": 0.3589263420724095,
      "grad_norm": 0.39173486828804016,
      "learning_rate": 4.6412297128589266e-05,
      "loss": 1.7443,
      "step": 2300
    },
    {
      "epoch": 0.366729088639201,
      "grad_norm": 0.47386667132377625,
      "learning_rate": 4.633426966292135e-05,
      "loss": 1.7276,
      "step": 2350
    },
    {
      "epoch": 0.37453183520599254,
      "grad_norm": 0.3906112015247345,
      "learning_rate": 4.6256242197253436e-05,
      "loss": 1.7194,
      "step": 2400
    },
    {
      "epoch": 0.382334581772784,
      "grad_norm": 0.4129493832588196,
      "learning_rate": 4.6178214731585524e-05,
      "loss": 1.7203,
      "step": 2450
    },
    {
      "epoch": 0.39013732833957554,
      "grad_norm": 0.4499260485172272,
      "learning_rate": 4.6100187265917605e-05,
      "loss": 1.7268,
      "step": 2500
    },
    {
      "epoch": 0.397940074906367,
      "grad_norm": 0.4021989703178406,
      "learning_rate": 4.6022159800249694e-05,
      "loss": 1.7532,
      "step": 2550
    },
    {
      "epoch": 0.40574282147315854,
      "grad_norm": 0.3842841386795044,
      "learning_rate": 4.5944132334581775e-05,
      "loss": 1.7195,
      "step": 2600
    },
    {
      "epoch": 0.4135455680399501,
      "grad_norm": 0.3462275266647339,
      "learning_rate": 4.5866104868913864e-05,
      "loss": 1.7269,
      "step": 2650
    },
    {
      "epoch": 0.42134831460674155,
      "grad_norm": 0.39410606026649475,
      "learning_rate": 4.5788077403245945e-05,
      "loss": 1.7192,
      "step": 2700
    },
    {
      "epoch": 0.4291510611735331,
      "grad_norm": 0.4126410484313965,
      "learning_rate": 4.5710049937578034e-05,
      "loss": 1.7421,
      "step": 2750
    },
    {
      "epoch": 0.4369538077403246,
      "grad_norm": 0.36984720826148987,
      "learning_rate": 4.5632022471910115e-05,
      "loss": 1.7101,
      "step": 2800
    },
    {
      "epoch": 0.4447565543071161,
      "grad_norm": 0.3527930974960327,
      "learning_rate": 4.5553995006242203e-05,
      "loss": 1.7276,
      "step": 2850
    },
    {
      "epoch": 0.4525593008739076,
      "grad_norm": 0.4303431510925293,
      "learning_rate": 4.5475967540574285e-05,
      "loss": 1.7292,
      "step": 2900
    },
    {
      "epoch": 0.46036204744069914,
      "grad_norm": 0.43609175086021423,
      "learning_rate": 4.5397940074906367e-05,
      "loss": 1.7167,
      "step": 2950
    },
    {
      "epoch": 0.4681647940074906,
      "grad_norm": 0.43081963062286377,
      "learning_rate": 4.5319912609238455e-05,
      "loss": 1.7214,
      "step": 3000
    },
    {
      "epoch": 0.47596754057428214,
      "grad_norm": 0.4468071162700653,
      "learning_rate": 4.5241885143570536e-05,
      "loss": 1.7375,
      "step": 3050
    },
    {
      "epoch": 0.4837702871410737,
      "grad_norm": 0.4700297713279724,
      "learning_rate": 4.516385767790262e-05,
      "loss": 1.7245,
      "step": 3100
    },
    {
      "epoch": 0.49157303370786515,
      "grad_norm": 0.4510380029678345,
      "learning_rate": 4.5085830212234706e-05,
      "loss": 1.7303,
      "step": 3150
    },
    {
      "epoch": 0.4993757802746567,
      "grad_norm": 0.4726191461086273,
      "learning_rate": 4.5007802746566795e-05,
      "loss": 1.7146,
      "step": 3200
    },
    {
      "epoch": 0.5071785268414482,
      "grad_norm": 0.4493762254714966,
      "learning_rate": 4.4929775280898876e-05,
      "loss": 1.7214,
      "step": 3250
    },
    {
      "epoch": 0.5149812734082397,
      "grad_norm": 0.425413578748703,
      "learning_rate": 4.4851747815230965e-05,
      "loss": 1.6912,
      "step": 3300
    },
    {
      "epoch": 0.5227840199750312,
      "grad_norm": 0.49669891595840454,
      "learning_rate": 4.4773720349563046e-05,
      "loss": 1.7082,
      "step": 3350
    },
    {
      "epoch": 0.5305867665418227,
      "grad_norm": 0.5002058148384094,
      "learning_rate": 4.4695692883895134e-05,
      "loss": 1.7119,
      "step": 3400
    },
    {
      "epoch": 0.5383895131086143,
      "grad_norm": 0.5210412740707397,
      "learning_rate": 4.4617665418227216e-05,
      "loss": 1.7139,
      "step": 3450
    },
    {
      "epoch": 0.5461922596754057,
      "grad_norm": 0.42913952469825745,
      "learning_rate": 4.4539637952559304e-05,
      "loss": 1.7083,
      "step": 3500
    },
    {
      "epoch": 0.5539950062421972,
      "grad_norm": 0.45524224638938904,
      "learning_rate": 4.4461610486891386e-05,
      "loss": 1.715,
      "step": 3550
    },
    {
      "epoch": 0.5617977528089888,
      "grad_norm": 0.47845831513404846,
      "learning_rate": 4.4383583021223474e-05,
      "loss": 1.7194,
      "step": 3600
    },
    {
      "epoch": 0.5696004993757803,
      "grad_norm": 0.4836772680282593,
      "learning_rate": 4.4305555555555556e-05,
      "loss": 1.7009,
      "step": 3650
    },
    {
      "epoch": 0.5774032459425718,
      "grad_norm": 0.4728386700153351,
      "learning_rate": 4.4227528089887644e-05,
      "loss": 1.7128,
      "step": 3700
    },
    {
      "epoch": 0.5852059925093633,
      "grad_norm": 0.4861275255680084,
      "learning_rate": 4.4149500624219726e-05,
      "loss": 1.7089,
      "step": 3750
    },
    {
      "epoch": 0.5930087390761548,
      "grad_norm": 0.5425216555595398,
      "learning_rate": 4.4071473158551814e-05,
      "loss": 1.7238,
      "step": 3800
    },
    {
      "epoch": 0.6008114856429463,
      "grad_norm": 0.4091911315917969,
      "learning_rate": 4.3993445692883896e-05,
      "loss": 1.683,
      "step": 3850
    },
    {
      "epoch": 0.6086142322097379,
      "grad_norm": 0.38465380668640137,
      "learning_rate": 4.3915418227215984e-05,
      "loss": 1.6991,
      "step": 3900
    },
    {
      "epoch": 0.6164169787765293,
      "grad_norm": 0.4543341398239136,
      "learning_rate": 4.383739076154807e-05,
      "loss": 1.7105,
      "step": 3950
    },
    {
      "epoch": 0.6242197253433208,
      "grad_norm": 0.46507373452186584,
      "learning_rate": 4.3759363295880154e-05,
      "loss": 1.6935,
      "step": 4000
    },
    {
      "epoch": 0.6320224719101124,
      "grad_norm": 0.4509834349155426,
      "learning_rate": 4.368133583021224e-05,
      "loss": 1.6869,
      "step": 4050
    },
    {
      "epoch": 0.6398252184769039,
      "grad_norm": 0.5607530474662781,
      "learning_rate": 4.3603308364544324e-05,
      "loss": 1.7125,
      "step": 4100
    },
    {
      "epoch": 0.6476279650436954,
      "grad_norm": 0.4719931483268738,
      "learning_rate": 4.3525280898876405e-05,
      "loss": 1.677,
      "step": 4150
    },
    {
      "epoch": 0.6554307116104869,
      "grad_norm": 0.4882090091705322,
      "learning_rate": 4.344725343320849e-05,
      "loss": 1.6981,
      "step": 4200
    },
    {
      "epoch": 0.6632334581772784,
      "grad_norm": 0.5204262733459473,
      "learning_rate": 4.3369225967540575e-05,
      "loss": 1.6905,
      "step": 4250
    },
    {
      "epoch": 0.6710362047440699,
      "grad_norm": 0.5416198372840881,
      "learning_rate": 4.329119850187266e-05,
      "loss": 1.6874,
      "step": 4300
    },
    {
      "epoch": 0.6788389513108615,
      "grad_norm": 0.484465628862381,
      "learning_rate": 4.3213171036204745e-05,
      "loss": 1.6967,
      "step": 4350
    },
    {
      "epoch": 0.686641697877653,
      "grad_norm": 0.4895637333393097,
      "learning_rate": 4.313514357053683e-05,
      "loss": 1.7088,
      "step": 4400
    },
    {
      "epoch": 0.6944444444444444,
      "grad_norm": 0.5524686574935913,
      "learning_rate": 4.3057116104868915e-05,
      "loss": 1.6826,
      "step": 4450
    },
    {
      "epoch": 0.702247191011236,
      "grad_norm": 0.5131920576095581,
      "learning_rate": 4.2979088639200997e-05,
      "loss": 1.6932,
      "step": 4500
    },
    {
      "epoch": 0.7100499375780275,
      "grad_norm": 0.5526320934295654,
      "learning_rate": 4.2901061173533085e-05,
      "loss": 1.7017,
      "step": 4550
    },
    {
      "epoch": 0.717852684144819,
      "grad_norm": 0.5150460600852966,
      "learning_rate": 4.282303370786517e-05,
      "loss": 1.6903,
      "step": 4600
    },
    {
      "epoch": 0.7256554307116105,
      "grad_norm": 0.5039018988609314,
      "learning_rate": 4.2745006242197255e-05,
      "loss": 1.7134,
      "step": 4650
    },
    {
      "epoch": 0.733458177278402,
      "grad_norm": 0.48572325706481934,
      "learning_rate": 4.266697877652934e-05,
      "loss": 1.6976,
      "step": 4700
    },
    {
      "epoch": 0.7412609238451935,
      "grad_norm": 0.5142014026641846,
      "learning_rate": 4.2588951310861425e-05,
      "loss": 1.6834,
      "step": 4750
    },
    {
      "epoch": 0.7490636704119851,
      "grad_norm": 0.4606572985649109,
      "learning_rate": 4.251092384519351e-05,
      "loss": 1.6793,
      "step": 4800
    },
    {
      "epoch": 0.7568664169787765,
      "grad_norm": 0.46960020065307617,
      "learning_rate": 4.2432896379525595e-05,
      "loss": 1.6828,
      "step": 4850
    },
    {
      "epoch": 0.764669163545568,
      "grad_norm": 0.4922361671924591,
      "learning_rate": 4.235486891385768e-05,
      "loss": 1.7022,
      "step": 4900
    },
    {
      "epoch": 0.7724719101123596,
      "grad_norm": 0.5289677381515503,
      "learning_rate": 4.2276841448189764e-05,
      "loss": 1.6903,
      "step": 4950
    },
    {
      "epoch": 0.7802746566791511,
      "grad_norm": 0.5616611838340759,
      "learning_rate": 4.219881398252185e-05,
      "loss": 1.6915,
      "step": 5000
    },
    {
      "epoch": 0.7880774032459426,
      "grad_norm": 0.4942910075187683,
      "learning_rate": 4.2120786516853934e-05,
      "loss": 1.6813,
      "step": 5050
    },
    {
      "epoch": 0.795880149812734,
      "grad_norm": 0.5219690203666687,
      "learning_rate": 4.204275905118602e-05,
      "loss": 1.6731,
      "step": 5100
    },
    {
      "epoch": 0.8036828963795256,
      "grad_norm": 0.4913477897644043,
      "learning_rate": 4.1964731585518104e-05,
      "loss": 1.7001,
      "step": 5150
    },
    {
      "epoch": 0.8114856429463171,
      "grad_norm": 0.49623045325279236,
      "learning_rate": 4.188670411985019e-05,
      "loss": 1.686,
      "step": 5200
    },
    {
      "epoch": 0.8192883895131086,
      "grad_norm": 0.5660040974617004,
      "learning_rate": 4.1808676654182274e-05,
      "loss": 1.6794,
      "step": 5250
    },
    {
      "epoch": 0.8270911360799001,
      "grad_norm": 0.5747349262237549,
      "learning_rate": 4.173064918851436e-05,
      "loss": 1.6847,
      "step": 5300
    },
    {
      "epoch": 0.8348938826466916,
      "grad_norm": 0.5206372737884521,
      "learning_rate": 4.1652621722846444e-05,
      "loss": 1.6719,
      "step": 5350
    },
    {
      "epoch": 0.8426966292134831,
      "grad_norm": 0.5366120934486389,
      "learning_rate": 4.1574594257178526e-05,
      "loss": 1.6696,
      "step": 5400
    },
    {
      "epoch": 0.8504993757802747,
      "grad_norm": 0.6354568600654602,
      "learning_rate": 4.1496566791510614e-05,
      "loss": 1.6657,
      "step": 5450
    },
    {
      "epoch": 0.8583021223470662,
      "grad_norm": 0.5194640159606934,
      "learning_rate": 4.1418539325842695e-05,
      "loss": 1.6731,
      "step": 5500
    },
    {
      "epoch": 0.8661048689138576,
      "grad_norm": 0.43448275327682495,
      "learning_rate": 4.1340511860174784e-05,
      "loss": 1.6639,
      "step": 5550
    },
    {
      "epoch": 0.8739076154806492,
      "grad_norm": 0.5286650657653809,
      "learning_rate": 4.1262484394506865e-05,
      "loss": 1.677,
      "step": 5600
    },
    {
      "epoch": 0.8817103620474407,
      "grad_norm": 0.49659380316734314,
      "learning_rate": 4.1184456928838954e-05,
      "loss": 1.6758,
      "step": 5650
    },
    {
      "epoch": 0.8895131086142322,
      "grad_norm": 0.5224044322967529,
      "learning_rate": 4.1106429463171035e-05,
      "loss": 1.6528,
      "step": 5700
    },
    {
      "epoch": 0.8973158551810237,
      "grad_norm": 0.510977566242218,
      "learning_rate": 4.1028401997503124e-05,
      "loss": 1.6869,
      "step": 5750
    },
    {
      "epoch": 0.9051186017478152,
      "grad_norm": 0.5862101912498474,
      "learning_rate": 4.0950374531835205e-05,
      "loss": 1.68,
      "step": 5800
    },
    {
      "epoch": 0.9129213483146067,
      "grad_norm": 0.5646480321884155,
      "learning_rate": 4.0872347066167293e-05,
      "loss": 1.6749,
      "step": 5850
    },
    {
      "epoch": 0.9207240948813983,
      "grad_norm": 0.5872883200645447,
      "learning_rate": 4.0794319600499375e-05,
      "loss": 1.6661,
      "step": 5900
    },
    {
      "epoch": 0.9285268414481898,
      "grad_norm": 0.5308676958084106,
      "learning_rate": 4.0716292134831463e-05,
      "loss": 1.6747,
      "step": 5950
    },
    {
      "epoch": 0.9363295880149812,
      "grad_norm": 0.5872898101806641,
      "learning_rate": 4.0638264669163545e-05,
      "loss": 1.6719,
      "step": 6000
    },
    {
      "epoch": 0.9441323345817728,
      "grad_norm": 0.6066872477531433,
      "learning_rate": 4.056023720349563e-05,
      "loss": 1.6746,
      "step": 6050
    },
    {
      "epoch": 0.9519350811485643,
      "grad_norm": 0.5329908132553101,
      "learning_rate": 4.048220973782772e-05,
      "loss": 1.6789,
      "step": 6100
    },
    {
      "epoch": 0.9597378277153558,
      "grad_norm": 0.4528316855430603,
      "learning_rate": 4.04041822721598e-05,
      "loss": 1.6767,
      "step": 6150
    },
    {
      "epoch": 0.9675405742821473,
      "grad_norm": 0.5394971966743469,
      "learning_rate": 4.032615480649189e-05,
      "loss": 1.6797,
      "step": 6200
    },
    {
      "epoch": 0.9753433208489388,
      "grad_norm": 0.6735771298408508,
      "learning_rate": 4.024812734082397e-05,
      "loss": 1.6842,
      "step": 6250
    },
    {
      "epoch": 0.9831460674157303,
      "grad_norm": 0.48064225912094116,
      "learning_rate": 4.017009987515606e-05,
      "loss": 1.6767,
      "step": 6300
    },
    {
      "epoch": 0.9909488139825219,
      "grad_norm": 0.49285510182380676,
      "learning_rate": 4.009207240948814e-05,
      "loss": 1.666,
      "step": 6350
    },
    {
      "epoch": 0.9987515605493134,
      "grad_norm": 0.5762596726417542,
      "learning_rate": 4.001404494382023e-05,
      "loss": 1.6766,
      "step": 6400
    },
    {
      "epoch": 1.006554307116105,
      "grad_norm": 0.4971337616443634,
      "learning_rate": 3.993601747815231e-05,
      "loss": 1.6727,
      "step": 6450
    },
    {
      "epoch": 1.0143570536828963,
      "grad_norm": 0.5485156178474426,
      "learning_rate": 3.9857990012484394e-05,
      "loss": 1.6842,
      "step": 6500
    },
    {
      "epoch": 1.0221598002496879,
      "grad_norm": 0.4900757968425751,
      "learning_rate": 3.977996254681648e-05,
      "loss": 1.6661,
      "step": 6550
    },
    {
      "epoch": 1.0299625468164795,
      "grad_norm": 0.4844594895839691,
      "learning_rate": 3.9701935081148564e-05,
      "loss": 1.6601,
      "step": 6600
    },
    {
      "epoch": 1.0377652933832708,
      "grad_norm": 0.6316475868225098,
      "learning_rate": 3.9623907615480646e-05,
      "loss": 1.6481,
      "step": 6650
    },
    {
      "epoch": 1.0455680399500624,
      "grad_norm": 0.6341625452041626,
      "learning_rate": 3.9545880149812734e-05,
      "loss": 1.6656,
      "step": 6700
    },
    {
      "epoch": 1.053370786516854,
      "grad_norm": 0.6208726763725281,
      "learning_rate": 3.9467852684144816e-05,
      "loss": 1.6379,
      "step": 6750
    },
    {
      "epoch": 1.0611735330836454,
      "grad_norm": 0.605767011642456,
      "learning_rate": 3.9389825218476904e-05,
      "loss": 1.6537,
      "step": 6800
    },
    {
      "epoch": 1.068976279650437,
      "grad_norm": 0.5422509908676147,
      "learning_rate": 3.931179775280899e-05,
      "loss": 1.6689,
      "step": 6850
    },
    {
      "epoch": 1.0767790262172285,
      "grad_norm": 0.6556758880615234,
      "learning_rate": 3.9233770287141074e-05,
      "loss": 1.6619,
      "step": 6900
    },
    {
      "epoch": 1.08458177278402,
      "grad_norm": 0.5463916659355164,
      "learning_rate": 3.915574282147316e-05,
      "loss": 1.6584,
      "step": 6950
    },
    {
      "epoch": 1.0923845193508115,
      "grad_norm": 0.660961925983429,
      "learning_rate": 3.9077715355805244e-05,
      "loss": 1.6813,
      "step": 7000
    },
    {
      "epoch": 1.100187265917603,
      "grad_norm": 0.5504911541938782,
      "learning_rate": 3.899968789013733e-05,
      "loss": 1.6482,
      "step": 7050
    },
    {
      "epoch": 1.1079900124843944,
      "grad_norm": 0.5077877044677734,
      "learning_rate": 3.8921660424469414e-05,
      "loss": 1.6508,
      "step": 7100
    },
    {
      "epoch": 1.115792759051186,
      "grad_norm": 0.5340275764465332,
      "learning_rate": 3.88436329588015e-05,
      "loss": 1.6685,
      "step": 7150
    },
    {
      "epoch": 1.1235955056179776,
      "grad_norm": 0.615564227104187,
      "learning_rate": 3.8765605493133584e-05,
      "loss": 1.6617,
      "step": 7200
    },
    {
      "epoch": 1.131398252184769,
      "grad_norm": 0.600592315196991,
      "learning_rate": 3.868757802746567e-05,
      "loss": 1.6478,
      "step": 7250
    },
    {
      "epoch": 1.1392009987515606,
      "grad_norm": 0.606829047203064,
      "learning_rate": 3.8609550561797754e-05,
      "loss": 1.6627,
      "step": 7300
    },
    {
      "epoch": 1.1470037453183521,
      "grad_norm": 0.5715992450714111,
      "learning_rate": 3.853152309612984e-05,
      "loss": 1.6724,
      "step": 7350
    },
    {
      "epoch": 1.1548064918851435,
      "grad_norm": 0.5475966334342957,
      "learning_rate": 3.8453495630461923e-05,
      "loss": 1.6762,
      "step": 7400
    },
    {
      "epoch": 1.162609238451935,
      "grad_norm": 0.5486684441566467,
      "learning_rate": 3.837546816479401e-05,
      "loss": 1.6418,
      "step": 7450
    },
    {
      "epoch": 1.1704119850187267,
      "grad_norm": 0.5656526684761047,
      "learning_rate": 3.829744069912609e-05,
      "loss": 1.6623,
      "step": 7500
    },
    {
      "epoch": 1.178214731585518,
      "grad_norm": 0.471967875957489,
      "learning_rate": 3.821941323345818e-05,
      "loss": 1.6339,
      "step": 7550
    },
    {
      "epoch": 1.1860174781523096,
      "grad_norm": 0.5814192891120911,
      "learning_rate": 3.814138576779026e-05,
      "loss": 1.6626,
      "step": 7600
    },
    {
      "epoch": 1.1938202247191012,
      "grad_norm": 0.5809513926506042,
      "learning_rate": 3.806335830212235e-05,
      "loss": 1.6521,
      "step": 7650
    },
    {
      "epoch": 1.2016229712858926,
      "grad_norm": 0.564431369304657,
      "learning_rate": 3.798533083645443e-05,
      "loss": 1.633,
      "step": 7700
    },
    {
      "epoch": 1.2094257178526842,
      "grad_norm": 0.5864349007606506,
      "learning_rate": 3.7907303370786515e-05,
      "loss": 1.6724,
      "step": 7750
    },
    {
      "epoch": 1.2172284644194757,
      "grad_norm": 0.5883368849754333,
      "learning_rate": 3.78292759051186e-05,
      "loss": 1.6661,
      "step": 7800
    },
    {
      "epoch": 1.225031210986267,
      "grad_norm": 0.5318378806114197,
      "learning_rate": 3.7751248439450685e-05,
      "loss": 1.6703,
      "step": 7850
    },
    {
      "epoch": 1.2328339575530587,
      "grad_norm": 0.6735764741897583,
      "learning_rate": 3.767322097378277e-05,
      "loss": 1.6558,
      "step": 7900
    },
    {
      "epoch": 1.2406367041198503,
      "grad_norm": 0.5900487303733826,
      "learning_rate": 3.7595193508114855e-05,
      "loss": 1.6472,
      "step": 7950
    },
    {
      "epoch": 1.2484394506866416,
      "grad_norm": 0.4971151649951935,
      "learning_rate": 3.751716604244694e-05,
      "loss": 1.6357,
      "step": 8000
    },
    {
      "epoch": 1.2562421972534332,
      "grad_norm": 0.6045508980751038,
      "learning_rate": 3.7439138576779024e-05,
      "loss": 1.6556,
      "step": 8050
    },
    {
      "epoch": 1.2640449438202248,
      "grad_norm": 0.5860553979873657,
      "learning_rate": 3.736111111111111e-05,
      "loss": 1.6659,
      "step": 8100
    },
    {
      "epoch": 1.2718476903870162,
      "grad_norm": 0.5339462161064148,
      "learning_rate": 3.7283083645443194e-05,
      "loss": 1.6707,
      "step": 8150
    },
    {
      "epoch": 1.2796504369538078,
      "grad_norm": 0.5763932466506958,
      "learning_rate": 3.720505617977528e-05,
      "loss": 1.6694,
      "step": 8200
    },
    {
      "epoch": 1.2874531835205993,
      "grad_norm": 0.5927013754844666,
      "learning_rate": 3.712702871410737e-05,
      "loss": 1.6258,
      "step": 8250
    },
    {
      "epoch": 1.2952559300873907,
      "grad_norm": 0.6203845739364624,
      "learning_rate": 3.704900124843945e-05,
      "loss": 1.6459,
      "step": 8300
    },
    {
      "epoch": 1.3030586766541823,
      "grad_norm": 0.5438473224639893,
      "learning_rate": 3.697097378277154e-05,
      "loss": 1.6352,
      "step": 8350
    },
    {
      "epoch": 1.3108614232209739,
      "grad_norm": 0.6493474245071411,
      "learning_rate": 3.689294631710362e-05,
      "loss": 1.6643,
      "step": 8400
    },
    {
      "epoch": 1.3186641697877652,
      "grad_norm": 0.5788607597351074,
      "learning_rate": 3.681491885143571e-05,
      "loss": 1.6753,
      "step": 8450
    },
    {
      "epoch": 1.3264669163545568,
      "grad_norm": 0.5591830015182495,
      "learning_rate": 3.673689138576779e-05,
      "loss": 1.6316,
      "step": 8500
    },
    {
      "epoch": 1.3342696629213484,
      "grad_norm": 0.4842735230922699,
      "learning_rate": 3.665886392009988e-05,
      "loss": 1.6699,
      "step": 8550
    },
    {
      "epoch": 1.3420724094881398,
      "grad_norm": 0.6692916750907898,
      "learning_rate": 3.658083645443196e-05,
      "loss": 1.6249,
      "step": 8600
    },
    {
      "epoch": 1.3498751560549314,
      "grad_norm": 0.5876237154006958,
      "learning_rate": 3.650280898876405e-05,
      "loss": 1.6576,
      "step": 8650
    },
    {
      "epoch": 1.357677902621723,
      "grad_norm": 0.6215786933898926,
      "learning_rate": 3.642478152309613e-05,
      "loss": 1.6332,
      "step": 8700
    },
    {
      "epoch": 1.3654806491885143,
      "grad_norm": 0.54453444480896,
      "learning_rate": 3.634675405742822e-05,
      "loss": 1.6456,
      "step": 8750
    },
    {
      "epoch": 1.373283395755306,
      "grad_norm": 0.6348562836647034,
      "learning_rate": 3.62687265917603e-05,
      "loss": 1.6476,
      "step": 8800
    },
    {
      "epoch": 1.3810861423220975,
      "grad_norm": 0.6236295700073242,
      "learning_rate": 3.6190699126092384e-05,
      "loss": 1.6676,
      "step": 8850
    },
    {
      "epoch": 1.3888888888888888,
      "grad_norm": 0.5054611563682556,
      "learning_rate": 3.611267166042447e-05,
      "loss": 1.6506,
      "step": 8900
    },
    {
      "epoch": 1.3966916354556804,
      "grad_norm": 0.582488477230072,
      "learning_rate": 3.6034644194756553e-05,
      "loss": 1.6413,
      "step": 8950
    },
    {
      "epoch": 1.404494382022472,
      "grad_norm": 0.6589245796203613,
      "learning_rate": 3.595661672908864e-05,
      "loss": 1.6824,
      "step": 9000
    },
    {
      "epoch": 1.4122971285892634,
      "grad_norm": 0.5584797859191895,
      "learning_rate": 3.587858926342072e-05,
      "loss": 1.6685,
      "step": 9050
    },
    {
      "epoch": 1.420099875156055,
      "grad_norm": 0.6652534604072571,
      "learning_rate": 3.580056179775281e-05,
      "loss": 1.6362,
      "step": 9100
    },
    {
      "epoch": 1.4279026217228465,
      "grad_norm": 0.6449257731437683,
      "learning_rate": 3.572253433208489e-05,
      "loss": 1.6543,
      "step": 9150
    },
    {
      "epoch": 1.435705368289638,
      "grad_norm": 0.6358399391174316,
      "learning_rate": 3.564450686641698e-05,
      "loss": 1.6464,
      "step": 9200
    },
    {
      "epoch": 1.4435081148564295,
      "grad_norm": 0.6031101942062378,
      "learning_rate": 3.556647940074906e-05,
      "loss": 1.6435,
      "step": 9250
    },
    {
      "epoch": 1.451310861423221,
      "grad_norm": 0.5363774299621582,
      "learning_rate": 3.548845193508115e-05,
      "loss": 1.6728,
      "step": 9300
    },
    {
      "epoch": 1.4591136079900124,
      "grad_norm": 0.6634340286254883,
      "learning_rate": 3.541042446941323e-05,
      "loss": 1.6638,
      "step": 9350
    },
    {
      "epoch": 1.466916354556804,
      "grad_norm": 0.6200147867202759,
      "learning_rate": 3.533239700374532e-05,
      "loss": 1.6537,
      "step": 9400
    },
    {
      "epoch": 1.4747191011235956,
      "grad_norm": 0.5800793766975403,
      "learning_rate": 3.52543695380774e-05,
      "loss": 1.6549,
      "step": 9450
    },
    {
      "epoch": 1.482521847690387,
      "grad_norm": 0.5839795470237732,
      "learning_rate": 3.517634207240949e-05,
      "loss": 1.6571,
      "step": 9500
    },
    {
      "epoch": 1.4903245942571786,
      "grad_norm": 0.5768577456474304,
      "learning_rate": 3.509831460674157e-05,
      "loss": 1.679,
      "step": 9550
    },
    {
      "epoch": 1.4981273408239701,
      "grad_norm": 0.5268595218658447,
      "learning_rate": 3.502028714107366e-05,
      "loss": 1.6461,
      "step": 9600
    },
    {
      "epoch": 1.5059300873907615,
      "grad_norm": 0.6356619000434875,
      "learning_rate": 3.494225967540574e-05,
      "loss": 1.6312,
      "step": 9650
    },
    {
      "epoch": 1.513732833957553,
      "grad_norm": 0.5722295641899109,
      "learning_rate": 3.486423220973783e-05,
      "loss": 1.6418,
      "step": 9700
    },
    {
      "epoch": 1.5215355805243447,
      "grad_norm": 0.5974990129470825,
      "learning_rate": 3.478620474406992e-05,
      "loss": 1.6526,
      "step": 9750
    },
    {
      "epoch": 1.529338327091136,
      "grad_norm": 0.6584164500236511,
      "learning_rate": 3.4708177278402e-05,
      "loss": 1.6749,
      "step": 9800
    },
    {
      "epoch": 1.5371410736579276,
      "grad_norm": 0.6195454001426697,
      "learning_rate": 3.463014981273409e-05,
      "loss": 1.6406,
      "step": 9850
    },
    {
      "epoch": 1.5449438202247192,
      "grad_norm": 0.5923195481300354,
      "learning_rate": 3.455212234706617e-05,
      "loss": 1.6471,
      "step": 9900
    },
    {
      "epoch": 1.5527465667915106,
      "grad_norm": 0.59232097864151,
      "learning_rate": 3.447409488139825e-05,
      "loss": 1.6595,
      "step": 9950
    },
    {
      "epoch": 1.5605493133583022,
      "grad_norm": 0.5838867425918579,
      "learning_rate": 3.439606741573034e-05,
      "loss": 1.6449,
      "step": 10000
    },
    {
      "epoch": 1.5683520599250937,
      "grad_norm": 0.6070720553398132,
      "learning_rate": 3.431803995006242e-05,
      "loss": 1.6376,
      "step": 10050
    },
    {
      "epoch": 1.576154806491885,
      "grad_norm": 0.5864161849021912,
      "learning_rate": 3.4240012484394504e-05,
      "loss": 1.6555,
      "step": 10100
    },
    {
      "epoch": 1.5839575530586767,
      "grad_norm": 0.6388084292411804,
      "learning_rate": 3.416198501872659e-05,
      "loss": 1.6522,
      "step": 10150
    },
    {
      "epoch": 1.5917602996254683,
      "grad_norm": 0.5700277090072632,
      "learning_rate": 3.4083957553058674e-05,
      "loss": 1.6347,
      "step": 10200
    },
    {
      "epoch": 1.5995630461922596,
      "grad_norm": 0.6094324588775635,
      "learning_rate": 3.400593008739076e-05,
      "loss": 1.6439,
      "step": 10250
    },
    {
      "epoch": 1.6073657927590512,
      "grad_norm": 0.6227761507034302,
      "learning_rate": 3.3927902621722844e-05,
      "loss": 1.6567,
      "step": 10300
    },
    {
      "epoch": 1.6151685393258428,
      "grad_norm": 0.6303547024726868,
      "learning_rate": 3.384987515605493e-05,
      "loss": 1.6221,
      "step": 10350
    },
    {
      "epoch": 1.6229712858926342,
      "grad_norm": 0.6025314927101135,
      "learning_rate": 3.3771847690387014e-05,
      "loss": 1.6371,
      "step": 10400
    },
    {
      "epoch": 1.6307740324594258,
      "grad_norm": 0.683813214302063,
      "learning_rate": 3.36938202247191e-05,
      "loss": 1.6364,
      "step": 10450
    },
    {
      "epoch": 1.6385767790262173,
      "grad_norm": 0.763845682144165,
      "learning_rate": 3.361579275905119e-05,
      "loss": 1.6274,
      "step": 10500
    },
    {
      "epoch": 1.6463795255930087,
      "grad_norm": 0.6550936698913574,
      "learning_rate": 3.353776529338327e-05,
      "loss": 1.6433,
      "step": 10550
    },
    {
      "epoch": 1.6541822721598003,
      "grad_norm": 0.6213018894195557,
      "learning_rate": 3.345973782771536e-05,
      "loss": 1.6405,
      "step": 10600
    },
    {
      "epoch": 1.6619850187265919,
      "grad_norm": 0.6199821829795837,
      "learning_rate": 3.338171036204744e-05,
      "loss": 1.6221,
      "step": 10650
    },
    {
      "epoch": 1.6697877652933832,
      "grad_norm": 0.6940792798995972,
      "learning_rate": 3.330368289637953e-05,
      "loss": 1.6064,
      "step": 10700
    },
    {
      "epoch": 1.6775905118601748,
      "grad_norm": 0.5895411968231201,
      "learning_rate": 3.322565543071161e-05,
      "loss": 1.6246,
      "step": 10750
    },
    {
      "epoch": 1.6853932584269664,
      "grad_norm": 0.584697425365448,
      "learning_rate": 3.31476279650437e-05,
      "loss": 1.6164,
      "step": 10800
    },
    {
      "epoch": 1.6931960049937578,
      "grad_norm": 0.5995935201644897,
      "learning_rate": 3.306960049937578e-05,
      "loss": 1.6303,
      "step": 10850
    },
    {
      "epoch": 1.7009987515605494,
      "grad_norm": 0.6184208989143372,
      "learning_rate": 3.299157303370787e-05,
      "loss": 1.6226,
      "step": 10900
    },
    {
      "epoch": 1.708801498127341,
      "grad_norm": 0.6035963892936707,
      "learning_rate": 3.291354556803995e-05,
      "loss": 1.6519,
      "step": 10950
    },
    {
      "epoch": 1.7166042446941323,
      "grad_norm": 0.6514495611190796,
      "learning_rate": 3.283551810237204e-05,
      "loss": 1.6311,
      "step": 11000
    },
    {
      "epoch": 1.724406991260924,
      "grad_norm": 0.681898832321167,
      "learning_rate": 3.275749063670412e-05,
      "loss": 1.6493,
      "step": 11050
    },
    {
      "epoch": 1.7322097378277155,
      "grad_norm": 0.641394317150116,
      "learning_rate": 3.267946317103621e-05,
      "loss": 1.6562,
      "step": 11100
    },
    {
      "epoch": 1.7400124843945068,
      "grad_norm": 0.6565654277801514,
      "learning_rate": 3.260143570536829e-05,
      "loss": 1.6193,
      "step": 11150
    },
    {
      "epoch": 1.7478152309612984,
      "grad_norm": 0.6501032114028931,
      "learning_rate": 3.252340823970037e-05,
      "loss": 1.6294,
      "step": 11200
    },
    {
      "epoch": 1.75561797752809,
      "grad_norm": 0.5766665935516357,
      "learning_rate": 3.244538077403246e-05,
      "loss": 1.6236,
      "step": 11250
    },
    {
      "epoch": 1.7634207240948814,
      "grad_norm": 0.6376942992210388,
      "learning_rate": 3.236735330836454e-05,
      "loss": 1.6592,
      "step": 11300
    },
    {
      "epoch": 1.771223470661673,
      "grad_norm": 0.6972282528877258,
      "learning_rate": 3.228932584269663e-05,
      "loss": 1.6173,
      "step": 11350
    },
    {
      "epoch": 1.7790262172284645,
      "grad_norm": 0.6171312928199768,
      "learning_rate": 3.221129837702871e-05,
      "loss": 1.6414,
      "step": 11400
    },
    {
      "epoch": 1.786828963795256,
      "grad_norm": 0.715184211730957,
      "learning_rate": 3.21332709113608e-05,
      "loss": 1.637,
      "step": 11450
    },
    {
      "epoch": 1.7946317103620475,
      "grad_norm": 0.6628192663192749,
      "learning_rate": 3.205524344569288e-05,
      "loss": 1.6102,
      "step": 11500
    },
    {
      "epoch": 1.802434456928839,
      "grad_norm": 0.6058287024497986,
      "learning_rate": 3.197721598002497e-05,
      "loss": 1.6512,
      "step": 11550
    },
    {
      "epoch": 1.8102372034956304,
      "grad_norm": 0.6275887489318848,
      "learning_rate": 3.189918851435705e-05,
      "loss": 1.6435,
      "step": 11600
    },
    {
      "epoch": 1.818039950062422,
      "grad_norm": 0.7389242053031921,
      "learning_rate": 3.182116104868914e-05,
      "loss": 1.6234,
      "step": 11650
    },
    {
      "epoch": 1.8258426966292136,
      "grad_norm": 0.649131715297699,
      "learning_rate": 3.174313358302122e-05,
      "loss": 1.6332,
      "step": 11700
    },
    {
      "epoch": 1.833645443196005,
      "grad_norm": 0.5898476839065552,
      "learning_rate": 3.166510611735331e-05,
      "loss": 1.6299,
      "step": 11750
    },
    {
      "epoch": 1.8414481897627963,
      "grad_norm": 0.617365837097168,
      "learning_rate": 3.158707865168539e-05,
      "loss": 1.608,
      "step": 11800
    },
    {
      "epoch": 1.8492509363295881,
      "grad_norm": 0.6347021460533142,
      "learning_rate": 3.150905118601748e-05,
      "loss": 1.6168,
      "step": 11850
    },
    {
      "epoch": 1.8570536828963795,
      "grad_norm": 0.6479565501213074,
      "learning_rate": 3.143102372034957e-05,
      "loss": 1.6089,
      "step": 11900
    },
    {
      "epoch": 1.8648564294631709,
      "grad_norm": 0.6168213486671448,
      "learning_rate": 3.135299625468165e-05,
      "loss": 1.63,
      "step": 11950
    },
    {
      "epoch": 1.8726591760299627,
      "grad_norm": 0.5773766040802002,
      "learning_rate": 3.127496878901374e-05,
      "loss": 1.6183,
      "step": 12000
    },
    {
      "epoch": 1.880461922596754,
      "grad_norm": 0.5600804686546326,
      "learning_rate": 3.119694132334582e-05,
      "loss": 1.6393,
      "step": 12050
    },
    {
      "epoch": 1.8882646691635454,
      "grad_norm": 0.623058557510376,
      "learning_rate": 3.111891385767791e-05,
      "loss": 1.6112,
      "step": 12100
    },
    {
      "epoch": 1.8960674157303372,
      "grad_norm": 0.5952323079109192,
      "learning_rate": 3.104088639200999e-05,
      "loss": 1.6319,
      "step": 12150
    },
    {
      "epoch": 1.9038701622971286,
      "grad_norm": 0.7055174112319946,
      "learning_rate": 3.096285892634208e-05,
      "loss": 1.6275,
      "step": 12200
    },
    {
      "epoch": 1.91167290886392,
      "grad_norm": 0.5625096559524536,
      "learning_rate": 3.088483146067416e-05,
      "loss": 1.6345,
      "step": 12250
    },
    {
      "epoch": 1.9194756554307117,
      "grad_norm": 0.5937293767929077,
      "learning_rate": 3.080680399500624e-05,
      "loss": 1.6197,
      "step": 12300
    },
    {
      "epoch": 1.927278401997503,
      "grad_norm": 0.6606655120849609,
      "learning_rate": 3.072877652933833e-05,
      "loss": 1.6201,
      "step": 12350
    },
    {
      "epoch": 1.9350811485642945,
      "grad_norm": 0.6392807960510254,
      "learning_rate": 3.065074906367041e-05,
      "loss": 1.6309,
      "step": 12400
    },
    {
      "epoch": 1.9428838951310863,
      "grad_norm": 0.7471784353256226,
      "learning_rate": 3.057272159800249e-05,
      "loss": 1.6131,
      "step": 12450
    },
    {
      "epoch": 1.9506866416978776,
      "grad_norm": 0.6735255718231201,
      "learning_rate": 3.0494694132334585e-05,
      "loss": 1.6514,
      "step": 12500
    },
    {
      "epoch": 1.958489388264669,
      "grad_norm": 0.6417968273162842,
      "learning_rate": 3.0416666666666666e-05,
      "loss": 1.6245,
      "step": 12550
    },
    {
      "epoch": 1.9662921348314608,
      "grad_norm": 0.6633313894271851,
      "learning_rate": 3.0338639200998755e-05,
      "loss": 1.6267,
      "step": 12600
    },
    {
      "epoch": 1.9740948813982522,
      "grad_norm": 0.6218631267547607,
      "learning_rate": 3.026061173533084e-05,
      "loss": 1.6472,
      "step": 12650
    },
    {
      "epoch": 1.9818976279650435,
      "grad_norm": 0.594956636428833,
      "learning_rate": 3.018258426966292e-05,
      "loss": 1.6298,
      "step": 12700
    },
    {
      "epoch": 1.9897003745318353,
      "grad_norm": 0.5888795852661133,
      "learning_rate": 3.010455680399501e-05,
      "loss": 1.626,
      "step": 12750
    },
    {
      "epoch": 1.9975031210986267,
      "grad_norm": 0.6327818632125854,
      "learning_rate": 3.002652933832709e-05,
      "loss": 1.625,
      "step": 12800
    }
  ],
  "logging_steps": 50,
  "max_steps": 32040,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.7153416265151283e+17,
  "train_batch_size": 10,
  "trial_name": null,
  "trial_params": null
}