{
  "best_metric": 0.03396870195865631,
  "best_model_checkpoint": "text-performance-longformer/checkpoint-7374",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 7374,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01017087062652563,
      "grad_norm": 13.638067245483398,
      "learning_rate": 1.6260162601626018e-06,
      "loss": 0.3817,
      "step": 25
    },
    {
      "epoch": 0.02034174125305126,
      "grad_norm": 6.143240928649902,
      "learning_rate": 3.3197831978319785e-06,
      "loss": 0.2287,
      "step": 50
    },
    {
      "epoch": 0.030512611879576892,
      "grad_norm": 1.5986428260803223,
      "learning_rate": 5.013550135501355e-06,
      "loss": 0.0691,
      "step": 75
    },
    {
      "epoch": 0.04068348250610252,
      "grad_norm": 0.9884291887283325,
      "learning_rate": 6.707317073170733e-06,
      "loss": 0.0543,
      "step": 100
    },
    {
      "epoch": 0.050854353132628156,
      "grad_norm": 3.724142074584961,
      "learning_rate": 8.401084010840109e-06,
      "loss": 0.0619,
      "step": 125
    },
    {
      "epoch": 0.061025223759153785,
      "grad_norm": 2.291118860244751,
      "learning_rate": 1.0094850948509485e-05,
      "loss": 0.0523,
      "step": 150
    },
    {
      "epoch": 0.07119609438567942,
      "grad_norm": 2.3093833923339844,
      "learning_rate": 1.1788617886178862e-05,
      "loss": 0.0541,
      "step": 175
    },
    {
      "epoch": 0.08136696501220504,
      "grad_norm": 3.3519299030303955,
      "learning_rate": 1.348238482384824e-05,
      "loss": 0.0515,
      "step": 200
    },
    {
      "epoch": 0.09153783563873068,
      "grad_norm": 2.0906357765197754,
      "learning_rate": 1.5176151761517615e-05,
      "loss": 0.0529,
      "step": 225
    },
    {
      "epoch": 0.10170870626525631,
      "grad_norm": 1.1683790683746338,
      "learning_rate": 1.6802168021680217e-05,
      "loss": 0.0573,
      "step": 250
    },
    {
      "epoch": 0.11187957689178193,
      "grad_norm": 1.7971270084381104,
      "learning_rate": 1.8495934959349594e-05,
      "loss": 0.0495,
      "step": 275
    },
    {
      "epoch": 0.12205044751830757,
      "grad_norm": 3.8092596530914307,
      "learning_rate": 2.018970189701897e-05,
      "loss": 0.0462,
      "step": 300
    },
    {
      "epoch": 0.1322213181448332,
      "grad_norm": 2.3589837551116943,
      "learning_rate": 2.1883468834688347e-05,
      "loss": 0.044,
      "step": 325
    },
    {
      "epoch": 0.14239218877135884,
      "grad_norm": 1.4069643020629883,
      "learning_rate": 2.3577235772357724e-05,
      "loss": 0.0458,
      "step": 350
    },
    {
      "epoch": 0.15256305939788445,
      "grad_norm": 0.8279024362564087,
      "learning_rate": 2.5271002710027104e-05,
      "loss": 0.0457,
      "step": 375
    },
    {
      "epoch": 0.16273393002441008,
      "grad_norm": 0.7554256916046143,
      "learning_rate": 2.696476964769648e-05,
      "loss": 0.0431,
      "step": 400
    },
    {
      "epoch": 0.17290480065093572,
      "grad_norm": 0.7221837043762207,
      "learning_rate": 2.8658536585365854e-05,
      "loss": 0.0455,
      "step": 425
    },
    {
      "epoch": 0.18307567127746135,
      "grad_norm": 1.551680088043213,
      "learning_rate": 3.035230352303523e-05,
      "loss": 0.0443,
      "step": 450
    },
    {
      "epoch": 0.193246541903987,
      "grad_norm": 1.2256766557693481,
      "learning_rate": 3.204607046070461e-05,
      "loss": 0.0423,
      "step": 475
    },
    {
      "epoch": 0.20341741253051263,
      "grad_norm": 1.9404680728912354,
      "learning_rate": 3.373983739837399e-05,
      "loss": 0.0454,
      "step": 500
    },
    {
      "epoch": 0.21358828315703823,
      "grad_norm": 1.481037974357605,
      "learning_rate": 3.5433604336043364e-05,
      "loss": 0.0549,
      "step": 525
    },
    {
      "epoch": 0.22375915378356387,
      "grad_norm": 1.0439728498458862,
      "learning_rate": 3.712737127371274e-05,
      "loss": 0.0474,
      "step": 550
    },
    {
      "epoch": 0.2339300244100895,
      "grad_norm": 2.524623394012451,
      "learning_rate": 3.882113821138211e-05,
      "loss": 0.0442,
      "step": 575
    },
    {
      "epoch": 0.24410089503661514,
      "grad_norm": 0.46617817878723145,
      "learning_rate": 4.051490514905149e-05,
      "loss": 0.0416,
      "step": 600
    },
    {
      "epoch": 0.25427176566314075,
      "grad_norm": 0.9644585251808167,
      "learning_rate": 4.220867208672087e-05,
      "loss": 0.0459,
      "step": 625
    },
    {
      "epoch": 0.2644426362896664,
      "grad_norm": 0.7916778922080994,
      "learning_rate": 4.390243902439025e-05,
      "loss": 0.0453,
      "step": 650
    },
    {
      "epoch": 0.274613506916192,
      "grad_norm": 0.41167861223220825,
      "learning_rate": 4.5596205962059624e-05,
      "loss": 0.0447,
      "step": 675
    },
    {
      "epoch": 0.2847843775427177,
      "grad_norm": 2.670891046524048,
      "learning_rate": 4.7289972899729e-05,
      "loss": 0.0436,
      "step": 700
    },
    {
      "epoch": 0.2949552481692433,
      "grad_norm": 0.5455201268196106,
      "learning_rate": 4.898373983739837e-05,
      "loss": 0.046,
      "step": 725
    },
    {
      "epoch": 0.3051261187957689,
      "grad_norm": 0.477638304233551,
      "learning_rate": 4.9924653405666065e-05,
      "loss": 0.0461,
      "step": 750
    },
    {
      "epoch": 0.31529698942229456,
      "grad_norm": 1.8583427667617798,
      "learning_rate": 4.973628691983123e-05,
      "loss": 0.046,
      "step": 775
    },
    {
      "epoch": 0.32546786004882017,
      "grad_norm": 1.4675369262695312,
      "learning_rate": 4.954792043399639e-05,
      "loss": 0.043,
      "step": 800
    },
    {
      "epoch": 0.33563873067534583,
      "grad_norm": 0.5187409520149231,
      "learning_rate": 4.9359553948161545e-05,
      "loss": 0.0465,
      "step": 825
    },
    {
      "epoch": 0.34580960130187144,
      "grad_norm": 1.1287797689437866,
      "learning_rate": 4.917118746232671e-05,
      "loss": 0.0395,
      "step": 850
    },
    {
      "epoch": 0.35598047192839705,
      "grad_norm": 1.6059693098068237,
      "learning_rate": 4.8982820976491866e-05,
      "loss": 0.0463,
      "step": 875
    },
    {
      "epoch": 0.3661513425549227,
      "grad_norm": 0.5075823664665222,
      "learning_rate": 4.8794454490657024e-05,
      "loss": 0.0431,
      "step": 900
    },
    {
      "epoch": 0.3763222131814483,
      "grad_norm": 0.4540039896965027,
      "learning_rate": 4.860608800482219e-05,
      "loss": 0.0444,
      "step": 925
    },
    {
      "epoch": 0.386493083807974,
      "grad_norm": 0.3790999948978424,
      "learning_rate": 4.8417721518987346e-05,
      "loss": 0.0408,
      "step": 950
    },
    {
      "epoch": 0.3966639544344996,
      "grad_norm": 0.35743555426597595,
      "learning_rate": 4.82293550331525e-05,
      "loss": 0.0412,
      "step": 975
    },
    {
      "epoch": 0.40683482506102525,
      "grad_norm": 0.9319044351577759,
      "learning_rate": 4.804098854731766e-05,
      "loss": 0.0383,
      "step": 1000
    },
    {
      "epoch": 0.41700569568755086,
      "grad_norm": 0.9398106336593628,
      "learning_rate": 4.785262206148282e-05,
      "loss": 0.0439,
      "step": 1025
    },
    {
      "epoch": 0.42717656631407647,
      "grad_norm": 0.5996136665344238,
      "learning_rate": 4.766425557564798e-05,
      "loss": 0.0455,
      "step": 1050
    },
    {
      "epoch": 0.43734743694060213,
      "grad_norm": 2.0928783416748047,
      "learning_rate": 4.747588908981314e-05,
      "loss": 0.0451,
      "step": 1075
    },
    {
      "epoch": 0.44751830756712774,
      "grad_norm": 0.42079484462738037,
      "learning_rate": 4.7287522603978304e-05,
      "loss": 0.0476,
      "step": 1100
    },
    {
      "epoch": 0.4576891781936534,
      "grad_norm": 1.7878057956695557,
      "learning_rate": 4.709915611814346e-05,
      "loss": 0.0434,
      "step": 1125
    },
    {
      "epoch": 0.467860048820179,
      "grad_norm": 0.3113216459751129,
      "learning_rate": 4.691078963230862e-05,
      "loss": 0.042,
      "step": 1150
    },
    {
      "epoch": 0.4780309194467046,
      "grad_norm": 0.319442480802536,
      "learning_rate": 4.6722423146473784e-05,
      "loss": 0.0409,
      "step": 1175
    },
    {
      "epoch": 0.4882017900732303,
      "grad_norm": 0.9614000916481018,
      "learning_rate": 4.653405666063894e-05,
      "loss": 0.0391,
      "step": 1200
    },
    {
      "epoch": 0.4983726606997559,
      "grad_norm": 0.6299770474433899,
      "learning_rate": 4.63456901748041e-05,
      "loss": 0.0414,
      "step": 1225
    },
    {
      "epoch": 0.5085435313262815,
      "grad_norm": 0.8669236898422241,
      "learning_rate": 4.615732368896926e-05,
      "loss": 0.0388,
      "step": 1250
    },
    {
      "epoch": 0.5187144019528072,
      "grad_norm": 0.6755848526954651,
      "learning_rate": 4.596895720313442e-05,
      "loss": 0.0421,
      "step": 1275
    },
    {
      "epoch": 0.5288852725793328,
      "grad_norm": 0.5094274282455444,
      "learning_rate": 4.5780590717299585e-05,
      "loss": 0.0408,
      "step": 1300
    },
    {
      "epoch": 0.5390561432058584,
      "grad_norm": 1.5034645795822144,
      "learning_rate": 4.559222423146474e-05,
      "loss": 0.0467,
      "step": 1325
    },
    {
      "epoch": 0.549227013832384,
      "grad_norm": 1.6210927963256836,
      "learning_rate": 4.54038577456299e-05,
      "loss": 0.0443,
      "step": 1350
    },
    {
      "epoch": 0.5593978844589097,
      "grad_norm": 0.4528130292892456,
      "learning_rate": 4.5215491259795064e-05,
      "loss": 0.041,
      "step": 1375
    },
    {
      "epoch": 0.5695687550854354,
      "grad_norm": 0.6066830158233643,
      "learning_rate": 4.5027124773960215e-05,
      "loss": 0.0427,
      "step": 1400
    },
    {
      "epoch": 0.5797396257119609,
      "grad_norm": 0.2950328588485718,
      "learning_rate": 4.483875828812538e-05,
      "loss": 0.0417,
      "step": 1425
    },
    {
      "epoch": 0.5899104963384866,
      "grad_norm": 0.425102561712265,
      "learning_rate": 4.4650391802290537e-05,
      "loss": 0.0391,
      "step": 1450
    },
    {
      "epoch": 0.6000813669650122,
      "grad_norm": 0.5589340329170227,
      "learning_rate": 4.4462025316455694e-05,
      "loss": 0.0406,
      "step": 1475
    },
    {
      "epoch": 0.6102522375915378,
      "grad_norm": 3.0561866760253906,
      "learning_rate": 4.427365883062086e-05,
      "loss": 0.0403,
      "step": 1500
    },
    {
      "epoch": 0.6204231082180635,
      "grad_norm": 0.6851157546043396,
      "learning_rate": 4.4085292344786016e-05,
      "loss": 0.0405,
      "step": 1525
    },
    {
      "epoch": 0.6305939788445891,
      "grad_norm": 0.5816906690597534,
      "learning_rate": 4.389692585895118e-05,
      "loss": 0.0465,
      "step": 1550
    },
    {
      "epoch": 0.6407648494711147,
      "grad_norm": 1.0224462747573853,
      "learning_rate": 4.370855937311634e-05,
      "loss": 0.0419,
      "step": 1575
    },
    {
      "epoch": 0.6509357200976403,
      "grad_norm": 0.4752540588378906,
      "learning_rate": 4.3520192887281495e-05,
      "loss": 0.0405,
      "step": 1600
    },
    {
      "epoch": 0.661106590724166,
      "grad_norm": 1.1999990940093994,
      "learning_rate": 4.333182640144666e-05,
      "loss": 0.041,
      "step": 1625
    },
    {
      "epoch": 0.6712774613506917,
      "grad_norm": 0.40787383913993835,
      "learning_rate": 4.314345991561182e-05,
      "loss": 0.0411,
      "step": 1650
    },
    {
      "epoch": 0.6814483319772172,
      "grad_norm": 0.26120448112487793,
      "learning_rate": 4.2955093429776974e-05,
      "loss": 0.0401,
      "step": 1675
    },
    {
      "epoch": 0.6916192026037429,
      "grad_norm": 0.5405380725860596,
      "learning_rate": 4.276672694394214e-05,
      "loss": 0.0387,
      "step": 1700
    },
    {
      "epoch": 0.7017900732302685,
      "grad_norm": 0.2800443768501282,
      "learning_rate": 4.2578360458107296e-05,
      "loss": 0.0411,
      "step": 1725
    },
    {
      "epoch": 0.7119609438567941,
      "grad_norm": 0.5752384662628174,
      "learning_rate": 4.2389993972272454e-05,
      "loss": 0.0404,
      "step": 1750
    },
    {
      "epoch": 0.7221318144833198,
      "grad_norm": 0.778127133846283,
      "learning_rate": 4.220162748643762e-05,
      "loss": 0.0428,
      "step": 1775
    },
    {
      "epoch": 0.7323026851098454,
      "grad_norm": 1.7695764303207397,
      "learning_rate": 4.2013261000602775e-05,
      "loss": 0.0353,
      "step": 1800
    },
    {
      "epoch": 0.7424735557363711,
      "grad_norm": 0.2690475285053253,
      "learning_rate": 4.182489451476794e-05,
      "loss": 0.0404,
      "step": 1825
    },
    {
      "epoch": 0.7526444263628966,
      "grad_norm": 0.506496012210846,
      "learning_rate": 4.163652802893309e-05,
      "loss": 0.0402,
      "step": 1850
    },
    {
      "epoch": 0.7628152969894223,
      "grad_norm": 0.7220098972320557,
      "learning_rate": 4.1448161543098255e-05,
      "loss": 0.0372,
      "step": 1875
    },
    {
      "epoch": 0.772986167615948,
      "grad_norm": 1.1999934911727905,
      "learning_rate": 4.125979505726341e-05,
      "loss": 0.0429,
      "step": 1900
    },
    {
      "epoch": 0.7831570382424735,
      "grad_norm": 0.5055158734321594,
      "learning_rate": 4.107142857142857e-05,
      "loss": 0.0367,
      "step": 1925
    },
    {
      "epoch": 0.7933279088689992,
      "grad_norm": 0.5015272498130798,
      "learning_rate": 4.0883062085593734e-05,
      "loss": 0.0376,
      "step": 1950
    },
    {
      "epoch": 0.8034987794955248,
      "grad_norm": 0.6382879018783569,
      "learning_rate": 4.069469559975889e-05,
      "loss": 0.0387,
      "step": 1975
    },
    {
      "epoch": 0.8136696501220505,
      "grad_norm": 1.4380210638046265,
      "learning_rate": 4.050632911392405e-05,
      "loss": 0.0433,
      "step": 2000
    },
    {
      "epoch": 0.823840520748576,
      "grad_norm": 0.46868982911109924,
      "learning_rate": 4.031796262808921e-05,
      "loss": 0.0409,
      "step": 2025
    },
    {
      "epoch": 0.8340113913751017,
      "grad_norm": 0.47722935676574707,
      "learning_rate": 4.012959614225437e-05,
      "loss": 0.0367,
      "step": 2050
    },
    {
      "epoch": 0.8441822620016274,
      "grad_norm": 0.66898113489151,
      "learning_rate": 3.9941229656419535e-05,
      "loss": 0.0426,
      "step": 2075
    },
    {
      "epoch": 0.8543531326281529,
      "grad_norm": 0.640957772731781,
      "learning_rate": 3.975286317058469e-05,
      "loss": 0.0374,
      "step": 2100
    },
    {
      "epoch": 0.8645240032546786,
      "grad_norm": 2.0676088333129883,
      "learning_rate": 3.956449668474985e-05,
      "loss": 0.0391,
      "step": 2125
    },
    {
      "epoch": 0.8746948738812043,
      "grad_norm": 0.3028632402420044,
      "learning_rate": 3.9376130198915014e-05,
      "loss": 0.0379,
      "step": 2150
    },
    {
      "epoch": 0.8848657445077298,
      "grad_norm": 3.03582763671875,
      "learning_rate": 3.918776371308017e-05,
      "loss": 0.0396,
      "step": 2175
    },
    {
      "epoch": 0.8950366151342555,
      "grad_norm": 0.5792508721351624,
      "learning_rate": 3.899939722724533e-05,
      "loss": 0.0393,
      "step": 2200
    },
    {
      "epoch": 0.9052074857607811,
      "grad_norm": 0.6225530505180359,
      "learning_rate": 3.8811030741410494e-05,
      "loss": 0.0413,
      "step": 2225
    },
    {
      "epoch": 0.9153783563873068,
      "grad_norm": 1.2196191549301147,
      "learning_rate": 3.862266425557565e-05,
      "loss": 0.0411,
      "step": 2250
    },
    {
      "epoch": 0.9255492270138324,
      "grad_norm": 1.1619917154312134,
      "learning_rate": 3.843429776974081e-05,
      "loss": 0.0372,
      "step": 2275
    },
    {
      "epoch": 0.935720097640358,
      "grad_norm": 0.4266558289527893,
      "learning_rate": 3.8245931283905966e-05,
      "loss": 0.0382,
      "step": 2300
    },
    {
      "epoch": 0.9458909682668837,
      "grad_norm": 0.2716640830039978,
      "learning_rate": 3.8057564798071124e-05,
      "loss": 0.0364,
      "step": 2325
    },
    {
      "epoch": 0.9560618388934092,
      "grad_norm": 0.3579002618789673,
      "learning_rate": 3.787673297166968e-05,
      "loss": 0.0397,
      "step": 2350
    },
    {
      "epoch": 0.9662327095199349,
      "grad_norm": 0.3124788999557495,
      "learning_rate": 3.768836648583484e-05,
      "loss": 0.0367,
      "step": 2375
    },
    {
      "epoch": 0.9764035801464606,
      "grad_norm": 0.7987418174743652,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.0425,
      "step": 2400
    },
    {
      "epoch": 0.9865744507729862,
      "grad_norm": 0.623573899269104,
      "learning_rate": 3.731163351416516e-05,
      "loss": 0.0407,
      "step": 2425
    },
    {
      "epoch": 0.9967453213995118,
      "grad_norm": 0.6313973665237427,
      "learning_rate": 3.712326702833032e-05,
      "loss": 0.0359,
      "step": 2450
    },
    {
      "epoch": 1.0,
      "eval_explained_variance": 0.22635483741760254,
      "eval_loss": 0.03618196025490761,
      "eval_mae": 0.1520702838897705,
      "eval_mse": 0.03618059679865837,
      "eval_r2": 0.22514164447784424,
      "eval_rmse": 0.19021197858877967,
      "eval_runtime": 39.05,
      "eval_samples_per_second": 503.405,
      "eval_steps_per_second": 7.887,
      "step": 2458
    },
    {
      "epoch": 1.0069161920260374,
      "grad_norm": 0.5591532588005066,
      "learning_rate": 3.693490054249548e-05,
      "loss": 0.0361,
      "step": 2475
    },
    {
      "epoch": 1.017087062652563,
      "grad_norm": 0.35737472772598267,
      "learning_rate": 3.674653405666064e-05,
      "loss": 0.035,
      "step": 2500
    },
    {
      "epoch": 1.0272579332790888,
      "grad_norm": 1.0595606565475464,
      "learning_rate": 3.6558167570825805e-05,
      "loss": 0.0374,
      "step": 2525
    },
    {
      "epoch": 1.0374288039056143,
      "grad_norm": 0.42890357971191406,
      "learning_rate": 3.636980108499096e-05,
      "loss": 0.0366,
      "step": 2550
    },
    {
      "epoch": 1.0475996745321399,
      "grad_norm": 0.41043972969055176,
      "learning_rate": 3.618143459915612e-05,
      "loss": 0.0377,
      "step": 2575
    },
    {
      "epoch": 1.0577705451586656,
      "grad_norm": 0.5887218117713928,
      "learning_rate": 3.5993068113321284e-05,
      "loss": 0.0391,
      "step": 2600
    },
    {
      "epoch": 1.0679414157851912,
      "grad_norm": 0.245023712515831,
      "learning_rate": 3.5804701627486435e-05,
      "loss": 0.032,
      "step": 2625
    },
    {
      "epoch": 1.0781122864117167,
      "grad_norm": 0.4018308222293854,
      "learning_rate": 3.56163351416516e-05,
      "loss": 0.0401,
      "step": 2650
    },
    {
      "epoch": 1.0882831570382425,
      "grad_norm": 1.287443995475769,
      "learning_rate": 3.5427968655816756e-05,
      "loss": 0.0366,
      "step": 2675
    },
    {
      "epoch": 1.098454027664768,
      "grad_norm": 0.9983925223350525,
      "learning_rate": 3.5239602169981914e-05,
      "loss": 0.0408,
      "step": 2700
    },
    {
      "epoch": 1.1086248982912936,
      "grad_norm": 1.045338749885559,
      "learning_rate": 3.505123568414708e-05,
      "loss": 0.0411,
      "step": 2725
    },
    {
      "epoch": 1.1187957689178194,
      "grad_norm": 0.2734503746032715,
      "learning_rate": 3.4862869198312236e-05,
      "loss": 0.0373,
      "step": 2750
    },
    {
      "epoch": 1.128966639544345,
      "grad_norm": 0.4091496169567108,
      "learning_rate": 3.46745027124774e-05,
      "loss": 0.0362,
      "step": 2775
    },
    {
      "epoch": 1.1391375101708707,
      "grad_norm": 1.026307225227356,
      "learning_rate": 3.448613622664256e-05,
      "loss": 0.036,
      "step": 2800
    },
    {
      "epoch": 1.1493083807973963,
      "grad_norm": 0.5424162149429321,
      "learning_rate": 3.4297769740807715e-05,
      "loss": 0.0362,
      "step": 2825
    },
    {
      "epoch": 1.1594792514239218,
      "grad_norm": 0.4070860743522644,
      "learning_rate": 3.410940325497288e-05,
      "loss": 0.0338,
      "step": 2850
    },
    {
      "epoch": 1.1696501220504476,
      "grad_norm": 0.861303448677063,
      "learning_rate": 3.392103676913804e-05,
      "loss": 0.0371,
      "step": 2875
    },
    {
      "epoch": 1.1798209926769732,
      "grad_norm": 0.6860642433166504,
      "learning_rate": 3.3732670283303194e-05,
      "loss": 0.0369,
      "step": 2900
    },
    {
      "epoch": 1.1899918633034987,
      "grad_norm": 0.6697980761528015,
      "learning_rate": 3.354430379746836e-05,
      "loss": 0.0379,
      "step": 2925
    },
    {
      "epoch": 1.2001627339300245,
      "grad_norm": 0.6728507876396179,
      "learning_rate": 3.3355937311633516e-05,
      "loss": 0.038,
      "step": 2950
    },
    {
      "epoch": 1.21033360455655,
      "grad_norm": 0.4306727349758148,
      "learning_rate": 3.316757082579868e-05,
      "loss": 0.0381,
      "step": 2975
    },
    {
      "epoch": 1.2205044751830756,
      "grad_norm": 0.8999012112617493,
      "learning_rate": 3.297920433996384e-05,
      "loss": 0.0393,
      "step": 3000
    },
    {
      "epoch": 1.2306753458096014,
      "grad_norm": 0.358962744474411,
      "learning_rate": 3.2790837854128995e-05,
      "loss": 0.0374,
      "step": 3025
    },
    {
      "epoch": 1.240846216436127,
      "grad_norm": 0.8197824954986572,
      "learning_rate": 3.260247136829416e-05,
      "loss": 0.0391,
      "step": 3050
    },
    {
      "epoch": 1.2510170870626527,
      "grad_norm": 0.6671149730682373,
      "learning_rate": 3.241410488245931e-05,
      "loss": 0.0357,
      "step": 3075
    },
    {
      "epoch": 1.2611879576891782,
      "grad_norm": 0.8932905197143555,
      "learning_rate": 3.2225738396624475e-05,
      "loss": 0.038,
      "step": 3100
    },
    {
      "epoch": 1.2713588283157038,
      "grad_norm": 0.3033260405063629,
      "learning_rate": 3.203737191078963e-05,
      "loss": 0.0364,
      "step": 3125
    },
    {
      "epoch": 1.2815296989422293,
      "grad_norm": 0.6071414351463318,
      "learning_rate": 3.184900542495479e-05,
      "loss": 0.0325,
      "step": 3150
    },
    {
      "epoch": 1.2917005695687551,
      "grad_norm": 0.28337907791137695,
      "learning_rate": 3.1660638939119954e-05,
      "loss": 0.0373,
      "step": 3175
    },
    {
      "epoch": 1.3018714401952807,
      "grad_norm": 0.5393190979957581,
      "learning_rate": 3.147227245328511e-05,
      "loss": 0.0362,
      "step": 3200
    },
    {
      "epoch": 1.3120423108218064,
      "grad_norm": 1.3214200735092163,
      "learning_rate": 3.128390596745027e-05,
      "loss": 0.0368,
      "step": 3225
    },
    {
      "epoch": 1.322213181448332,
      "grad_norm": 0.5053825974464417,
      "learning_rate": 3.109553948161543e-05,
      "loss": 0.0375,
      "step": 3250
    },
    {
      "epoch": 1.3323840520748575,
      "grad_norm": 0.4198523759841919,
      "learning_rate": 3.090717299578059e-05,
      "loss": 0.0404,
      "step": 3275
    },
    {
      "epoch": 1.342554922701383,
      "grad_norm": 0.9187168478965759,
      "learning_rate": 3.0718806509945755e-05,
      "loss": 0.0374,
      "step": 3300
    },
    {
      "epoch": 1.3527257933279089,
      "grad_norm": 0.3942495882511139,
      "learning_rate": 3.053044002411091e-05,
      "loss": 0.0362,
      "step": 3325
    },
    {
      "epoch": 1.3628966639544344,
      "grad_norm": 0.6447917819023132,
      "learning_rate": 3.0342073538276073e-05,
      "loss": 0.038,
      "step": 3350
    },
    {
      "epoch": 1.3730675345809602,
      "grad_norm": 0.3163827061653137,
      "learning_rate": 3.015370705244123e-05,
      "loss": 0.0369,
      "step": 3375
    },
    {
      "epoch": 1.3832384052074858,
      "grad_norm": 0.3072253465652466,
      "learning_rate": 2.9965340566606392e-05,
      "loss": 0.0385,
      "step": 3400
    },
    {
      "epoch": 1.3934092758340113,
      "grad_norm": 0.4808538556098938,
      "learning_rate": 2.9776974080771553e-05,
      "loss": 0.0378,
      "step": 3425
    },
    {
      "epoch": 1.403580146460537,
      "grad_norm": 0.2448228895664215,
      "learning_rate": 2.9588607594936714e-05,
      "loss": 0.0337,
      "step": 3450
    },
    {
      "epoch": 1.4137510170870626,
      "grad_norm": 0.42077022790908813,
      "learning_rate": 2.940024110910187e-05,
      "loss": 0.0359,
      "step": 3475
    },
    {
      "epoch": 1.4239218877135884,
      "grad_norm": 0.48862871527671814,
      "learning_rate": 2.921187462326703e-05,
      "loss": 0.0378,
      "step": 3500
    },
    {
      "epoch": 1.434092758340114,
      "grad_norm": 0.5912586450576782,
      "learning_rate": 2.9023508137432186e-05,
      "loss": 0.0384,
      "step": 3525
    },
    {
      "epoch": 1.4442636289666395,
      "grad_norm": 0.4059402644634247,
      "learning_rate": 2.8835141651597347e-05,
      "loss": 0.0381,
      "step": 3550
    },
    {
      "epoch": 1.454434499593165,
      "grad_norm": 0.3919837176799774,
      "learning_rate": 2.8646775165762508e-05,
      "loss": 0.036,
      "step": 3575
    },
    {
      "epoch": 1.4646053702196908,
      "grad_norm": 0.2935680150985718,
      "learning_rate": 2.845840867992767e-05,
      "loss": 0.0368,
      "step": 3600
    },
    {
      "epoch": 1.4747762408462164,
      "grad_norm": 0.7148743867874146,
      "learning_rate": 2.8270042194092826e-05,
      "loss": 0.0386,
      "step": 3625
    },
    {
      "epoch": 1.4849471114727422,
      "grad_norm": 1.314514398574829,
      "learning_rate": 2.8081675708257987e-05,
      "loss": 0.0375,
      "step": 3650
    },
    {
      "epoch": 1.4951179820992677,
      "grad_norm": 1.6261988878250122,
      "learning_rate": 2.7893309222423148e-05,
      "loss": 0.0376,
      "step": 3675
    },
    {
      "epoch": 1.5052888527257933,
      "grad_norm": 1.105427861213684,
      "learning_rate": 2.770494273658831e-05,
      "loss": 0.0354,
      "step": 3700
    },
    {
      "epoch": 1.5154597233523188,
      "grad_norm": 0.5577530264854431,
      "learning_rate": 2.7516576250753466e-05,
      "loss": 0.0362,
      "step": 3725
    },
    {
      "epoch": 1.5256305939788446,
      "grad_norm": 0.49557003378868103,
      "learning_rate": 2.7328209764918627e-05,
      "loss": 0.0389,
      "step": 3750
    },
    {
      "epoch": 1.5358014646053704,
      "grad_norm": 0.8892014622688293,
      "learning_rate": 2.7139843279083788e-05,
      "loss": 0.0379,
      "step": 3775
    },
    {
      "epoch": 1.545972335231896,
      "grad_norm": 0.5090736150741577,
      "learning_rate": 2.695147679324895e-05,
      "loss": 0.0359,
      "step": 3800
    },
    {
      "epoch": 1.5561432058584215,
      "grad_norm": 0.8963241577148438,
      "learning_rate": 2.6763110307414107e-05,
      "loss": 0.0421,
      "step": 3825
    },
    {
      "epoch": 1.566314076484947,
      "grad_norm": 0.3889683485031128,
      "learning_rate": 2.6574743821579268e-05,
      "loss": 0.0362,
      "step": 3850
    },
    {
      "epoch": 1.5764849471114726,
      "grad_norm": 0.6879289150238037,
      "learning_rate": 2.638637733574443e-05,
      "loss": 0.0367,
      "step": 3875
    },
    {
      "epoch": 1.5866558177379984,
      "grad_norm": 1.1574759483337402,
      "learning_rate": 2.619801084990959e-05,
      "loss": 0.0387,
      "step": 3900
    },
    {
      "epoch": 1.5968266883645241,
      "grad_norm": 1.3032798767089844,
      "learning_rate": 2.6009644364074747e-05,
      "loss": 0.0366,
      "step": 3925
    },
    {
      "epoch": 1.6069975589910497,
      "grad_norm": 1.0193997621536255,
      "learning_rate": 2.58212778782399e-05,
      "loss": 0.0325,
      "step": 3950
    },
    {
      "epoch": 1.6171684296175752,
      "grad_norm": 1.767223834991455,
      "learning_rate": 2.5632911392405062e-05,
      "loss": 0.0393,
      "step": 3975
    },
    {
      "epoch": 1.6273393002441008,
      "grad_norm": 1.016648530960083,
      "learning_rate": 2.5444544906570223e-05,
      "loss": 0.0333,
      "step": 4000
    },
    {
      "epoch": 1.6375101708706266,
      "grad_norm": 2.0735578536987305,
      "learning_rate": 2.5256178420735384e-05,
      "loss": 0.0355,
      "step": 4025
    },
    {
      "epoch": 1.647681041497152,
      "grad_norm": 0.8982949256896973,
      "learning_rate": 2.506781193490054e-05,
      "loss": 0.0369,
      "step": 4050
    },
    {
      "epoch": 1.6578519121236779,
      "grad_norm": 0.324400395154953,
      "learning_rate": 2.4879445449065702e-05,
      "loss": 0.0358,
      "step": 4075
    },
    {
      "epoch": 1.6680227827502034,
      "grad_norm": 0.32701972126960754,
      "learning_rate": 2.4691078963230863e-05,
      "loss": 0.0336,
      "step": 4100
    },
    {
      "epoch": 1.678193653376729,
      "grad_norm": 1.151262640953064,
      "learning_rate": 2.4502712477396024e-05,
      "loss": 0.0392,
      "step": 4125
    },
    {
      "epoch": 1.6883645240032545,
      "grad_norm": 0.5716719627380371,
      "learning_rate": 2.431434599156118e-05,
      "loss": 0.0383,
      "step": 4150
    },
    {
      "epoch": 1.6985353946297803,
      "grad_norm": 0.8748169541358948,
      "learning_rate": 2.4125979505726342e-05,
      "loss": 0.0344,
      "step": 4175
    },
    {
      "epoch": 1.708706265256306,
      "grad_norm": 0.25271666049957275,
      "learning_rate": 2.3937613019891503e-05,
      "loss": 0.0351,
      "step": 4200
    },
    {
      "epoch": 1.7188771358828316,
      "grad_norm": 0.23297059535980225,
      "learning_rate": 2.3749246534056664e-05,
      "loss": 0.0337,
      "step": 4225
    },
    {
      "epoch": 1.7290480065093572,
      "grad_norm": 0.3409133851528168,
      "learning_rate": 2.356088004822182e-05,
      "loss": 0.0346,
      "step": 4250
    },
    {
      "epoch": 1.7392188771358827,
      "grad_norm": 0.822523832321167,
      "learning_rate": 2.337251356238698e-05,
      "loss": 0.0374,
      "step": 4275
    },
    {
      "epoch": 1.7493897477624083,
      "grad_norm": 1.9754129648208618,
      "learning_rate": 2.318414707655214e-05,
      "loss": 0.0409,
      "step": 4300
    },
    {
      "epoch": 1.759560618388934,
      "grad_norm": 0.49358049035072327,
      "learning_rate": 2.29957805907173e-05,
      "loss": 0.0379,
      "step": 4325
    },
    {
      "epoch": 1.7697314890154598,
      "grad_norm": 0.6075097918510437,
      "learning_rate": 2.280741410488246e-05,
      "loss": 0.0358,
      "step": 4350
    },
    {
      "epoch": 1.7799023596419854,
      "grad_norm": 0.5666526556015015,
      "learning_rate": 2.261904761904762e-05,
      "loss": 0.0337,
      "step": 4375
    },
    {
      "epoch": 1.790073230268511,
      "grad_norm": 0.7485412955284119,
      "learning_rate": 2.243068113321278e-05,
      "loss": 0.037,
      "step": 4400
    },
    {
      "epoch": 1.8002441008950365,
      "grad_norm": 0.585403323173523,
      "learning_rate": 2.224231464737794e-05,
      "loss": 0.0379,
      "step": 4425
    },
    {
      "epoch": 1.8104149715215623,
      "grad_norm": 0.7822312712669373,
      "learning_rate": 2.2053948161543102e-05,
      "loss": 0.0368,
      "step": 4450
    },
    {
      "epoch": 1.8205858421480878,
      "grad_norm": 0.3547162413597107,
      "learning_rate": 2.186558167570826e-05,
      "loss": 0.038,
      "step": 4475
    },
    {
      "epoch": 1.8307567127746136,
      "grad_norm": 0.5509994626045227,
      "learning_rate": 2.1677215189873417e-05,
      "loss": 0.0371,
      "step": 4500
    },
    {
      "epoch": 1.8409275834011392,
      "grad_norm": 1.1359673738479614,
      "learning_rate": 2.1488848704038578e-05,
      "loss": 0.0321,
      "step": 4525
    },
    {
      "epoch": 1.8510984540276647,
      "grad_norm": 0.7981705069541931,
      "learning_rate": 2.130048221820374e-05,
      "loss": 0.0369,
      "step": 4550
    },
    {
      "epoch": 1.8612693246541903,
      "grad_norm": 0.3582057058811188,
      "learning_rate": 2.11121157323689e-05,
      "loss": 0.0412,
      "step": 4575
    },
    {
      "epoch": 1.871440195280716,
      "grad_norm": 0.9928992986679077,
      "learning_rate": 2.0923749246534057e-05,
      "loss": 0.0352,
      "step": 4600
    },
    {
      "epoch": 1.8816110659072418,
      "grad_norm": 0.48575785756111145,
      "learning_rate": 2.0735382760699218e-05,
      "loss": 0.035,
      "step": 4625
    },
    {
      "epoch": 1.8917819365337674,
      "grad_norm": 0.5365208387374878,
      "learning_rate": 2.054701627486438e-05,
      "loss": 0.0379,
      "step": 4650
    },
    {
      "epoch": 1.901952807160293,
      "grad_norm": 1.141358494758606,
      "learning_rate": 2.0358649789029536e-05,
      "loss": 0.0355,
      "step": 4675
    },
    {
      "epoch": 1.9121236777868185,
      "grad_norm": 0.43180742859840393,
      "learning_rate": 2.0170283303194694e-05,
      "loss": 0.0354,
      "step": 4700
    },
    {
      "epoch": 1.922294548413344,
      "grad_norm": 0.7140740752220154,
      "learning_rate": 1.9981916817359855e-05,
      "loss": 0.0355,
      "step": 4725
    },
    {
      "epoch": 1.9324654190398698,
      "grad_norm": 0.30647122859954834,
      "learning_rate": 1.9793550331525016e-05,
      "loss": 0.0371,
      "step": 4750
    },
    {
      "epoch": 1.9426362896663956,
      "grad_norm": 0.42196792364120483,
      "learning_rate": 1.9605183845690177e-05,
      "loss": 0.0356,
      "step": 4775
    },
    {
      "epoch": 1.9528071602929211,
      "grad_norm": 0.6331903338432312,
      "learning_rate": 1.9416817359855334e-05,
      "loss": 0.0352,
      "step": 4800
    },
    {
      "epoch": 1.9629780309194467,
      "grad_norm": 0.7057808637619019,
      "learning_rate": 1.9228450874020495e-05,
      "loss": 0.0364,
      "step": 4825
    },
    {
      "epoch": 1.9731489015459722,
      "grad_norm": 0.49434205889701843,
      "learning_rate": 1.9040084388185656e-05,
      "loss": 0.0347,
      "step": 4850
    },
    {
      "epoch": 1.983319772172498,
      "grad_norm": 0.3139288127422333,
      "learning_rate": 1.8851717902350817e-05,
      "loss": 0.0364,
      "step": 4875
    },
    {
      "epoch": 1.9934906427990235,
      "grad_norm": 0.3922992944717407,
      "learning_rate": 1.8663351416515974e-05,
      "loss": 0.036,
      "step": 4900
    },
    {
      "epoch": 2.0,
      "eval_explained_variance": 0.25461888313293457,
      "eval_loss": 0.03482421860098839,
      "eval_mae": 0.14410310983657837,
      "eval_mse": 0.0348237045109272,
      "eval_r2": 0.25420135259628296,
      "eval_rmse": 0.18661110500430353,
      "eval_runtime": 39.031,
      "eval_samples_per_second": 503.652,
      "eval_steps_per_second": 7.891,
      "step": 4916
    },
    {
      "epoch": 2.0036615134255493,
      "grad_norm": 0.5134842395782471,
      "learning_rate": 1.8474984930681132e-05,
      "loss": 0.0344,
      "step": 4925
    },
    {
      "epoch": 2.013832384052075,
      "grad_norm": 0.3377295136451721,
      "learning_rate": 1.8286618444846293e-05,
      "loss": 0.0336,
      "step": 4950
    },
    {
      "epoch": 2.0240032546786004,
      "grad_norm": 0.3855837285518646,
      "learning_rate": 1.8098251959011453e-05,
      "loss": 0.0316,
      "step": 4975
    },
    {
      "epoch": 2.034174125305126,
      "grad_norm": 0.4808228313922882,
      "learning_rate": 1.7909885473176614e-05,
      "loss": 0.0347,
      "step": 5000
    },
    {
      "epoch": 2.0443449959316515,
      "grad_norm": 0.6781342029571533,
      "learning_rate": 1.7721518987341772e-05,
      "loss": 0.0366,
      "step": 5025
    },
    {
      "epoch": 2.0545158665581775,
      "grad_norm": 0.5457364320755005,
      "learning_rate": 1.7533152501506933e-05,
      "loss": 0.0326,
      "step": 5050
    },
    {
      "epoch": 2.064686737184703,
      "grad_norm": 1.1539140939712524,
      "learning_rate": 1.7344786015672094e-05,
      "loss": 0.032,
      "step": 5075
    },
    {
      "epoch": 2.0748576078112286,
      "grad_norm": 0.7635537981987,
      "learning_rate": 1.7156419529837255e-05,
      "loss": 0.0318,
      "step": 5100
    },
    {
      "epoch": 2.085028478437754,
      "grad_norm": 0.8772742748260498,
      "learning_rate": 1.6968053044002412e-05,
      "loss": 0.0337,
      "step": 5125
    },
    {
      "epoch": 2.0951993490642797,
      "grad_norm": 0.45236992835998535,
      "learning_rate": 1.677968655816757e-05,
      "loss": 0.031,
      "step": 5150
    },
    {
      "epoch": 2.1053702196908057,
      "grad_norm": 0.5050310492515564,
      "learning_rate": 1.659132007233273e-05,
      "loss": 0.031,
      "step": 5175
    },
    {
      "epoch": 2.1155410903173313,
      "grad_norm": 0.442862331867218,
      "learning_rate": 1.640295358649789e-05,
      "loss": 0.0342,
      "step": 5200
    },
    {
      "epoch": 2.125711960943857,
      "grad_norm": 0.5236470103263855,
      "learning_rate": 1.6214587100663052e-05,
      "loss": 0.0372,
      "step": 5225
    },
    {
      "epoch": 2.1358828315703824,
      "grad_norm": 0.9813937544822693,
      "learning_rate": 1.602622061482821e-05,
      "loss": 0.0326,
      "step": 5250
    },
    {
      "epoch": 2.146053702196908,
      "grad_norm": 0.349025696516037,
      "learning_rate": 1.583785412899337e-05,
      "loss": 0.0346,
      "step": 5275
    },
    {
      "epoch": 2.1562245728234335,
      "grad_norm": 0.35612091422080994,
      "learning_rate": 1.564948764315853e-05,
      "loss": 0.0342,
      "step": 5300
    },
    {
      "epoch": 2.1663954434499595,
      "grad_norm": 0.5912727117538452,
      "learning_rate": 1.5461121157323692e-05,
      "loss": 0.0324,
      "step": 5325
    },
    {
      "epoch": 2.176566314076485,
      "grad_norm": 0.2870270609855652,
      "learning_rate": 1.5272754671488847e-05,
      "loss": 0.0342,
      "step": 5350
    },
    {
      "epoch": 2.1867371847030106,
      "grad_norm": 0.3680706322193146,
      "learning_rate": 1.5084388185654007e-05,
      "loss": 0.0329,
      "step": 5375
    },
    {
      "epoch": 2.196908055329536,
      "grad_norm": 0.9814783930778503,
      "learning_rate": 1.4896021699819168e-05,
      "loss": 0.0293,
      "step": 5400
    },
    {
      "epoch": 2.2070789259560617,
      "grad_norm": 0.7239277958869934,
      "learning_rate": 1.4707655213984328e-05,
      "loss": 0.0346,
      "step": 5425
    },
    {
      "epoch": 2.2172497965825873,
      "grad_norm": 0.44417452812194824,
      "learning_rate": 1.4519288728149488e-05,
      "loss": 0.0339,
      "step": 5450
    },
    {
      "epoch": 2.2274206672091132,
      "grad_norm": 0.3636336326599121,
      "learning_rate": 1.4330922242314648e-05,
      "loss": 0.0327,
      "step": 5475
    },
    {
      "epoch": 2.237591537835639,
      "grad_norm": 0.2732349634170532,
      "learning_rate": 1.4142555756479809e-05,
      "loss": 0.032,
      "step": 5500
    },
    {
      "epoch": 2.2477624084621644,
      "grad_norm": 0.820342481136322,
      "learning_rate": 1.3954189270644968e-05,
      "loss": 0.0318,
      "step": 5525
    },
    {
      "epoch": 2.25793327908869,
      "grad_norm": 0.31075552105903625,
      "learning_rate": 1.3765822784810129e-05,
      "loss": 0.0311,
      "step": 5550
    },
    {
      "epoch": 2.2681041497152155,
      "grad_norm": 0.8737571835517883,
      "learning_rate": 1.3577456298975286e-05,
      "loss": 0.0356,
      "step": 5575
    },
    {
      "epoch": 2.2782750203417415,
      "grad_norm": 0.9981245994567871,
      "learning_rate": 1.3389089813140445e-05,
      "loss": 0.0333,
      "step": 5600
    },
    {
      "epoch": 2.288445890968267,
      "grad_norm": 0.5384612679481506,
      "learning_rate": 1.3200723327305606e-05,
      "loss": 0.0316,
      "step": 5625
    },
    {
      "epoch": 2.2986167615947926,
      "grad_norm": 0.6893337965011597,
      "learning_rate": 1.3012356841470765e-05,
      "loss": 0.0324,
      "step": 5650
    },
    {
      "epoch": 2.308787632221318,
      "grad_norm": 0.449916273355484,
      "learning_rate": 1.2823990355635926e-05,
      "loss": 0.0325,
      "step": 5675
    },
    {
      "epoch": 2.3189585028478437,
      "grad_norm": 0.38824161887168884,
      "learning_rate": 1.2635623869801086e-05,
      "loss": 0.0339,
      "step": 5700
    },
    {
      "epoch": 2.329129373474369,
      "grad_norm": 0.7458836436271667,
      "learning_rate": 1.2447257383966246e-05,
      "loss": 0.0355,
      "step": 5725
    },
    {
      "epoch": 2.339300244100895,
      "grad_norm": 0.47954612970352173,
      "learning_rate": 1.2258890898131404e-05,
      "loss": 0.0323,
      "step": 5750
    },
    {
      "epoch": 2.3494711147274208,
      "grad_norm": 0.42400848865509033,
      "learning_rate": 1.2070524412296565e-05,
      "loss": 0.0302,
      "step": 5775
    },
    {
      "epoch": 2.3596419853539463,
      "grad_norm": 1.189965009689331,
      "learning_rate": 1.1882157926461724e-05,
      "loss": 0.0338,
      "step": 5800
    },
    {
      "epoch": 2.369812855980472,
      "grad_norm": 0.5762277841567993,
      "learning_rate": 1.1693791440626885e-05,
      "loss": 0.0332,
      "step": 5825
    },
    {
      "epoch": 2.3799837266069974,
      "grad_norm": 0.5994691848754883,
      "learning_rate": 1.1505424954792044e-05,
      "loss": 0.0364,
      "step": 5850
    },
    {
      "epoch": 2.390154597233523,
      "grad_norm": 0.9533575773239136,
      "learning_rate": 1.1317058468957203e-05,
      "loss": 0.0326,
      "step": 5875
    },
    {
      "epoch": 2.400325467860049,
      "grad_norm": 0.4238649308681488,
      "learning_rate": 1.1128691983122364e-05,
      "loss": 0.034,
      "step": 5900
    },
    {
      "epoch": 2.4104963384865745,
      "grad_norm": 0.8726415038108826,
      "learning_rate": 1.0940325497287523e-05,
      "loss": 0.0327,
      "step": 5925
    },
    {
      "epoch": 2.4206672091131,
      "grad_norm": 0.5922726988792419,
      "learning_rate": 1.0751959011452683e-05,
      "loss": 0.0337,
      "step": 5950
    },
    {
      "epoch": 2.4308380797396256,
      "grad_norm": 0.3707614839076996,
      "learning_rate": 1.0563592525617842e-05,
      "loss": 0.0338,
      "step": 5975
    },
    {
      "epoch": 2.441008950366151,
      "grad_norm": 0.4853639602661133,
      "learning_rate": 1.0375226039783003e-05,
      "loss": 0.0317,
      "step": 6000
    },
    {
      "epoch": 2.451179820992677,
      "grad_norm": 0.8022235631942749,
      "learning_rate": 1.0186859553948162e-05,
      "loss": 0.032,
      "step": 6025
    },
    {
      "epoch": 2.4613506916192027,
      "grad_norm": 0.8553130030632019,
      "learning_rate": 9.998493068113323e-06,
      "loss": 0.0312,
      "step": 6050
    },
    {
      "epoch": 2.4715215622457283,
      "grad_norm": 0.4112774431705475,
      "learning_rate": 9.81012658227848e-06,
      "loss": 0.0349,
      "step": 6075
    },
    {
      "epoch": 2.481692432872254,
      "grad_norm": 0.8546609282493591,
      "learning_rate": 9.621760096443641e-06,
      "loss": 0.0332,
      "step": 6100
    },
    {
      "epoch": 2.4918633034987794,
      "grad_norm": 0.8445001840591431,
      "learning_rate": 9.4333936106088e-06,
      "loss": 0.0324,
      "step": 6125
    },
    {
      "epoch": 2.5020341741253054,
      "grad_norm": 0.4580422043800354,
      "learning_rate": 9.245027124773961e-06,
      "loss": 0.0346,
      "step": 6150
    },
    {
      "epoch": 2.5122050447518305,
      "grad_norm": 0.6121585369110107,
      "learning_rate": 9.05666063893912e-06,
      "loss": 0.0345,
      "step": 6175
    },
    {
      "epoch": 2.5223759153783565,
      "grad_norm": 0.5637044906616211,
      "learning_rate": 8.86829415310428e-06,
      "loss": 0.0315,
      "step": 6200
    },
    {
      "epoch": 2.532546786004882,
      "grad_norm": 0.6579483151435852,
      "learning_rate": 8.67992766726944e-06,
      "loss": 0.0345,
      "step": 6225
    },
    {
      "epoch": 2.5427176566314076,
      "grad_norm": 0.30682843923568726,
      "learning_rate": 8.499095840867993e-06,
      "loss": 0.0334,
      "step": 6250
    },
    {
      "epoch": 2.552888527257933,
      "grad_norm": 1.7261478900909424,
      "learning_rate": 8.310729355033153e-06,
      "loss": 0.0337,
      "step": 6275
    },
    {
      "epoch": 2.5630593978844587,
      "grad_norm": 0.7609931826591492,
      "learning_rate": 8.122362869198312e-06,
      "loss": 0.0329,
      "step": 6300
    },
    {
      "epoch": 2.5732302685109847,
      "grad_norm": 1.1947487592697144,
      "learning_rate": 7.933996383363473e-06,
      "loss": 0.0338,
      "step": 6325
    },
    {
      "epoch": 2.5834011391375102,
      "grad_norm": 0.5045105814933777,
      "learning_rate": 7.745629897528632e-06,
      "loss": 0.0336,
      "step": 6350
    },
    {
      "epoch": 2.593572009764036,
      "grad_norm": 0.8998399972915649,
      "learning_rate": 7.557263411693792e-06,
      "loss": 0.0334,
      "step": 6375
    },
    {
      "epoch": 2.6037428803905613,
      "grad_norm": 0.3800385594367981,
      "learning_rate": 7.368896925858952e-06,
      "loss": 0.0306,
      "step": 6400
    },
    {
      "epoch": 2.613913751017087,
      "grad_norm": 0.35073891282081604,
      "learning_rate": 7.180530440024111e-06,
      "loss": 0.0342,
      "step": 6425
    },
    {
      "epoch": 2.624084621643613,
      "grad_norm": 0.35614126920700073,
      "learning_rate": 6.992163954189271e-06,
      "loss": 0.0317,
      "step": 6450
    },
    {
      "epoch": 2.6342554922701384,
      "grad_norm": 1.0959842205047607,
      "learning_rate": 6.8037974683544305e-06,
      "loss": 0.0328,
      "step": 6475
    },
    {
      "epoch": 2.644426362896664,
      "grad_norm": 0.9010970592498779,
      "learning_rate": 6.6154309825195905e-06,
      "loss": 0.0364,
      "step": 6500
    },
    {
      "epoch": 2.6545972335231895,
      "grad_norm": 0.8300909996032715,
      "learning_rate": 6.42706449668475e-06,
      "loss": 0.0312,
      "step": 6525
    },
    {
      "epoch": 2.664768104149715,
      "grad_norm": 0.7244754433631897,
      "learning_rate": 6.23869801084991e-06,
      "loss": 0.0319,
      "step": 6550
    },
    {
      "epoch": 2.674938974776241,
      "grad_norm": 1.3230552673339844,
      "learning_rate": 6.05033152501507e-06,
      "loss": 0.0328,
      "step": 6575
    },
    {
      "epoch": 2.685109845402766,
      "grad_norm": 0.437537282705307,
      "learning_rate": 5.861965039180229e-06,
      "loss": 0.0325,
      "step": 6600
    },
    {
      "epoch": 2.695280716029292,
      "grad_norm": 0.4210902154445648,
      "learning_rate": 5.673598553345389e-06,
      "loss": 0.0362,
      "step": 6625
    },
    {
      "epoch": 2.7054515866558178,
      "grad_norm": 0.3914755880832672,
      "learning_rate": 5.485232067510549e-06,
      "loss": 0.0329,
      "step": 6650
    },
    {
      "epoch": 2.7156224572823433,
      "grad_norm": 0.9759465456008911,
      "learning_rate": 5.296865581675708e-06,
      "loss": 0.0343,
      "step": 6675
    },
    {
      "epoch": 2.725793327908869,
      "grad_norm": 0.34633737802505493,
      "learning_rate": 5.108499095840868e-06,
      "loss": 0.0308,
      "step": 6700
    },
    {
      "epoch": 2.7359641985353944,
      "grad_norm": 0.5408746600151062,
      "learning_rate": 4.9201326100060275e-06,
      "loss": 0.0285,
      "step": 6725
    },
    {
      "epoch": 2.7461350691619204,
      "grad_norm": 0.3921310007572174,
      "learning_rate": 4.7317661241711876e-06,
      "loss": 0.0307,
      "step": 6750
    },
    {
      "epoch": 2.756305939788446,
      "grad_norm": 0.6094385981559753,
      "learning_rate": 4.543399638336348e-06,
      "loss": 0.0303,
      "step": 6775
    },
    {
      "epoch": 2.7664768104149715,
      "grad_norm": 0.5900077819824219,
      "learning_rate": 4.355033152501508e-06,
      "loss": 0.0355,
      "step": 6800
    },
    {
      "epoch": 2.776647681041497,
      "grad_norm": 0.4339945912361145,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.0351,
      "step": 6825
    },
    {
      "epoch": 2.7868185516680226,
      "grad_norm": 0.9042001962661743,
      "learning_rate": 3.978300180831827e-06,
      "loss": 0.033,
      "step": 6850
    },
    {
      "epoch": 2.7969894222945486,
      "grad_norm": 0.5715941190719604,
      "learning_rate": 3.789933694996986e-06,
      "loss": 0.0325,
      "step": 6875
    },
    {
      "epoch": 2.807160292921074,
      "grad_norm": 0.40120917558670044,
      "learning_rate": 3.601567209162146e-06,
      "loss": 0.0324,
      "step": 6900
    },
    {
      "epoch": 2.8173311635475997,
      "grad_norm": 0.636159360408783,
      "learning_rate": 3.413200723327306e-06,
      "loss": 0.0311,
      "step": 6925
    },
    {
      "epoch": 2.8275020341741253,
      "grad_norm": 0.79677414894104,
      "learning_rate": 3.2248342374924654e-06,
      "loss": 0.0298,
      "step": 6950
    },
    {
      "epoch": 2.837672904800651,
      "grad_norm": 0.6220082640647888,
      "learning_rate": 3.036467751657625e-06,
      "loss": 0.0315,
      "step": 6975
    },
    {
      "epoch": 2.847843775427177,
      "grad_norm": 0.4538786709308624,
      "learning_rate": 2.848101265822785e-06,
      "loss": 0.032,
      "step": 7000
    },
    {
      "epoch": 2.858014646053702,
      "grad_norm": 0.44975048303604126,
      "learning_rate": 2.6597347799879447e-06,
      "loss": 0.0314,
      "step": 7025
    },
    {
      "epoch": 2.868185516680228,
      "grad_norm": 0.2438650280237198,
      "learning_rate": 2.4713682941531043e-06,
      "loss": 0.0322,
      "step": 7050
    },
    {
      "epoch": 2.8783563873067535,
      "grad_norm": 0.9189873337745667,
      "learning_rate": 2.2830018083182644e-06,
      "loss": 0.0315,
      "step": 7075
    },
    {
      "epoch": 2.888527257933279,
      "grad_norm": 0.31788191199302673,
      "learning_rate": 2.094635322483424e-06,
      "loss": 0.0287,
      "step": 7100
    },
    {
      "epoch": 2.8986981285598046,
      "grad_norm": 0.7033805847167969,
      "learning_rate": 1.9062688366485836e-06,
      "loss": 0.0331,
      "step": 7125
    },
    {
      "epoch": 2.90886899918633,
      "grad_norm": 0.3187176287174225,
      "learning_rate": 1.7179023508137434e-06,
      "loss": 0.0349,
      "step": 7150
    },
    {
      "epoch": 2.919039869812856,
      "grad_norm": 0.3502849042415619,
      "learning_rate": 1.529535864978903e-06,
      "loss": 0.0314,
      "step": 7175
    },
    {
      "epoch": 2.9292107404393817,
      "grad_norm": 0.38132113218307495,
      "learning_rate": 1.3411693791440627e-06,
      "loss": 0.0307,
      "step": 7200
    },
    {
      "epoch": 2.9393816110659072,
      "grad_norm": 0.335792601108551,
      "learning_rate": 1.1528028933092225e-06,
      "loss": 0.0329,
      "step": 7225
    },
    {
      "epoch": 2.949552481692433,
      "grad_norm": 0.43150436878204346,
      "learning_rate": 9.644364074743821e-07,
      "loss": 0.0279,
      "step": 7250
    },
    {
      "epoch": 2.9597233523189583,
      "grad_norm": 0.43568554520606995,
      "learning_rate": 7.76069921639542e-07,
      "loss": 0.0297,
      "step": 7275
    },
    {
      "epoch": 2.9698942229454843,
      "grad_norm": 0.2997362017631531,
      "learning_rate": 5.877034358047017e-07,
      "loss": 0.0315,
      "step": 7300
    },
    {
      "epoch": 2.98006509357201,
      "grad_norm": 0.555476725101471,
      "learning_rate": 3.993369499698613e-07,
      "loss": 0.0296,
      "step": 7325
    },
    {
      "epoch": 2.9902359641985354,
      "grad_norm": 0.31480032205581665,
      "learning_rate": 2.1097046413502108e-07,
      "loss": 0.0334,
      "step": 7350
    },
    {
      "epoch": 3.0,
      "eval_explained_variance": 0.273209810256958,
      "eval_loss": 0.03396870195865631,
      "eval_mae": 0.14112502336502075,
      "eval_mse": 0.03396843746304512,
      "eval_r2": 0.2725181579589844,
      "eval_rmse": 0.18430528332916862,
      "eval_runtime": 39.0131,
      "eval_samples_per_second": 503.882,
      "eval_steps_per_second": 7.895,
      "step": 7374
    }
  ],
  "logging_steps": 25,
  "max_steps": 7374,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.01
      },
      "attributes": {
        "early_stopping_patience_counter": 2
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.937435559513293e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}