{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.9998179584034954,
  "eval_steps": 500,
  "global_step": 21972,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.045510399126200335,
      "grad_norm": 2.6165335178375244,
      "learning_rate": 9.887118798361402e-05,
      "loss": 0.8444,
      "step": 250
    },
    {
      "epoch": 0.09102079825240067,
      "grad_norm": 2.7013211250305176,
      "learning_rate": 9.773327264451525e-05,
      "loss": 0.5413,
      "step": 500
    },
    {
      "epoch": 0.136531197378601,
      "grad_norm": 2.3657588958740234,
      "learning_rate": 9.659535730541647e-05,
      "loss": 0.4992,
      "step": 750
    },
    {
      "epoch": 0.18204159650480134,
      "grad_norm": 2.571484088897705,
      "learning_rate": 9.54574419663177e-05,
      "loss": 0.4771,
      "step": 1000
    },
    {
      "epoch": 0.22755199563100167,
      "grad_norm": 2.439297676086426,
      "learning_rate": 9.431952662721894e-05,
      "loss": 0.4604,
      "step": 1250
    },
    {
      "epoch": 0.273062394757202,
      "grad_norm": 2.3595223426818848,
      "learning_rate": 9.318161128812017e-05,
      "loss": 0.452,
      "step": 1500
    },
    {
      "epoch": 0.31857279388340237,
      "grad_norm": 2.526773452758789,
      "learning_rate": 9.204369594902141e-05,
      "loss": 0.4438,
      "step": 1750
    },
    {
      "epoch": 0.3640831930096027,
      "grad_norm": 2.1642537117004395,
      "learning_rate": 9.090578060992263e-05,
      "loss": 0.4338,
      "step": 2000
    },
    {
      "epoch": 0.40959359213580304,
      "grad_norm": 2.5757973194122314,
      "learning_rate": 8.976786527082386e-05,
      "loss": 0.4277,
      "step": 2250
    },
    {
      "epoch": 0.45510399126200335,
      "grad_norm": 2.473083734512329,
      "learning_rate": 8.862994993172509e-05,
      "loss": 0.4175,
      "step": 2500
    },
    {
      "epoch": 0.5006143903882037,
      "grad_norm": 2.8206324577331543,
      "learning_rate": 8.749203459262631e-05,
      "loss": 0.4039,
      "step": 2750
    },
    {
      "epoch": 0.546124789514404,
      "grad_norm": 2.784949779510498,
      "learning_rate": 8.635411925352754e-05,
      "loss": 0.3923,
      "step": 3000
    },
    {
      "epoch": 0.5916351886406044,
      "grad_norm": 3.3137335777282715,
      "learning_rate": 8.521620391442878e-05,
      "loss": 0.3802,
      "step": 3250
    },
    {
      "epoch": 0.6371455877668047,
      "grad_norm": 3.6405413150787354,
      "learning_rate": 8.407828857533e-05,
      "loss": 0.3678,
      "step": 3500
    },
    {
      "epoch": 0.6826559868930051,
      "grad_norm": 3.3154547214508057,
      "learning_rate": 8.294037323623123e-05,
      "loss": 0.3623,
      "step": 3750
    },
    {
      "epoch": 0.7281663860192054,
      "grad_norm": 3.19486141204834,
      "learning_rate": 8.180245789713246e-05,
      "loss": 0.3499,
      "step": 4000
    },
    {
      "epoch": 0.7736767851454057,
      "grad_norm": 3.61464524269104,
      "learning_rate": 8.066454255803368e-05,
      "loss": 0.3429,
      "step": 4250
    },
    {
      "epoch": 0.8191871842716061,
      "grad_norm": 3.3924710750579834,
      "learning_rate": 7.952662721893491e-05,
      "loss": 0.3346,
      "step": 4500
    },
    {
      "epoch": 0.8646975833978064,
      "grad_norm": 3.7511956691741943,
      "learning_rate": 7.838871187983614e-05,
      "loss": 0.3238,
      "step": 4750
    },
    {
      "epoch": 0.9102079825240067,
      "grad_norm": 3.344393253326416,
      "learning_rate": 7.725079654073737e-05,
      "loss": 0.3186,
      "step": 5000
    },
    {
      "epoch": 0.9557183816502071,
      "grad_norm": 3.1730661392211914,
      "learning_rate": 7.61128812016386e-05,
      "loss": 0.3119,
      "step": 5250
    },
    {
      "epoch": 1.0012287807764073,
      "grad_norm": 3.959838628768921,
      "learning_rate": 7.497496586253983e-05,
      "loss": 0.3085,
      "step": 5500
    },
    {
      "epoch": 1.0467391799026078,
      "grad_norm": 3.2908735275268555,
      "learning_rate": 7.383705052344107e-05,
      "loss": 0.2975,
      "step": 5750
    },
    {
      "epoch": 1.092249579028808,
      "grad_norm": 3.441880941390991,
      "learning_rate": 7.26991351843423e-05,
      "loss": 0.2926,
      "step": 6000
    },
    {
      "epoch": 1.1377599781550085,
      "grad_norm": 4.100038051605225,
      "learning_rate": 7.156121984524352e-05,
      "loss": 0.2859,
      "step": 6250
    },
    {
      "epoch": 1.1832703772812088,
      "grad_norm": 3.5992090702056885,
      "learning_rate": 7.042330450614475e-05,
      "loss": 0.2808,
      "step": 6500
    },
    {
      "epoch": 1.228780776407409,
      "grad_norm": 3.3500804901123047,
      "learning_rate": 6.928538916704597e-05,
      "loss": 0.2791,
      "step": 6750
    },
    {
      "epoch": 1.2742911755336095,
      "grad_norm": 3.3923401832580566,
      "learning_rate": 6.814747382794721e-05,
      "loss": 0.2749,
      "step": 7000
    },
    {
      "epoch": 1.3198015746598097,
      "grad_norm": 3.785520315170288,
      "learning_rate": 6.700955848884844e-05,
      "loss": 0.2713,
      "step": 7250
    },
    {
      "epoch": 1.3653119737860102,
      "grad_norm": 3.4256818294525146,
      "learning_rate": 6.587164314974966e-05,
      "loss": 0.2668,
      "step": 7500
    },
    {
      "epoch": 1.4108223729122105,
      "grad_norm": 3.892718553543091,
      "learning_rate": 6.473372781065089e-05,
      "loss": 0.265,
      "step": 7750
    },
    {
      "epoch": 1.4563327720384107,
      "grad_norm": 3.272980213165283,
      "learning_rate": 6.359581247155212e-05,
      "loss": 0.2602,
      "step": 8000
    },
    {
      "epoch": 1.501843171164611,
      "grad_norm": 3.327261209487915,
      "learning_rate": 6.245789713245334e-05,
      "loss": 0.2604,
      "step": 8250
    },
    {
      "epoch": 1.5473535702908114,
      "grad_norm": 3.1805334091186523,
      "learning_rate": 6.131998179335457e-05,
      "loss": 0.2572,
      "step": 8500
    },
    {
      "epoch": 1.592863969417012,
      "grad_norm": 3.5480244159698486,
      "learning_rate": 6.01820664542558e-05,
      "loss": 0.2523,
      "step": 8750
    },
    {
      "epoch": 1.6383743685432122,
      "grad_norm": 3.761888265609741,
      "learning_rate": 5.904415111515703e-05,
      "loss": 0.2491,
      "step": 9000
    },
    {
      "epoch": 1.6838847676694124,
      "grad_norm": 3.0907113552093506,
      "learning_rate": 5.790623577605826e-05,
      "loss": 0.2458,
      "step": 9250
    },
    {
      "epoch": 1.7293951667956127,
      "grad_norm": 3.0482497215270996,
      "learning_rate": 5.676832043695949e-05,
      "loss": 0.2444,
      "step": 9500
    },
    {
      "epoch": 1.7749055659218131,
      "grad_norm": 3.088454008102417,
      "learning_rate": 5.563040509786073e-05,
      "loss": 0.2442,
      "step": 9750
    },
    {
      "epoch": 1.8204159650480136,
      "grad_norm": 3.492504119873047,
      "learning_rate": 5.4492489758761954e-05,
      "loss": 0.24,
      "step": 10000
    },
    {
      "epoch": 1.8659263641742139,
      "grad_norm": 3.4256038665771484,
      "learning_rate": 5.3354574419663187e-05,
      "loss": 0.2381,
      "step": 10250
    },
    {
      "epoch": 1.9114367633004141,
      "grad_norm": 3.163264036178589,
      "learning_rate": 5.221665908056441e-05,
      "loss": 0.2356,
      "step": 10500
    },
    {
      "epoch": 1.9569471624266144,
      "grad_norm": 3.4566102027893066,
      "learning_rate": 5.107874374146564e-05,
      "loss": 0.2356,
      "step": 10750
    },
    {
      "epoch": 2.0024575615528146,
      "grad_norm": 3.404165744781494,
      "learning_rate": 4.9940828402366865e-05,
      "loss": 0.2318,
      "step": 11000
    },
    {
      "epoch": 2.0479679606790153,
      "grad_norm": 3.004925012588501,
      "learning_rate": 4.88029130632681e-05,
      "loss": 0.2259,
      "step": 11250
    },
    {
      "epoch": 2.0934783598052156,
      "grad_norm": 3.1267077922821045,
      "learning_rate": 4.7664997724169324e-05,
      "loss": 0.2253,
      "step": 11500
    },
    {
      "epoch": 2.138988758931416,
      "grad_norm": 3.2789225578308105,
      "learning_rate": 4.652708238507055e-05,
      "loss": 0.2211,
      "step": 11750
    },
    {
      "epoch": 2.184499158057616,
      "grad_norm": 3.55497407913208,
      "learning_rate": 4.5389167045971784e-05,
      "loss": 0.2237,
      "step": 12000
    },
    {
      "epoch": 2.2300095571838163,
      "grad_norm": 3.159508228302002,
      "learning_rate": 4.425125170687301e-05,
      "loss": 0.219,
      "step": 12250
    },
    {
      "epoch": 2.275519956310017,
      "grad_norm": 3.4701318740844727,
      "learning_rate": 4.3113336367774236e-05,
      "loss": 0.2204,
      "step": 12500
    },
    {
      "epoch": 2.3210303554362173,
      "grad_norm": 3.129274606704712,
      "learning_rate": 4.197542102867547e-05,
      "loss": 0.2192,
      "step": 12750
    },
    {
      "epoch": 2.3665407545624175,
      "grad_norm": 3.207860231399536,
      "learning_rate": 4.08375056895767e-05,
      "loss": 0.2179,
      "step": 13000
    },
    {
      "epoch": 2.4120511536886178,
      "grad_norm": 2.966536283493042,
      "learning_rate": 3.969959035047793e-05,
      "loss": 0.2149,
      "step": 13250
    },
    {
      "epoch": 2.457561552814818,
      "grad_norm": 3.199531078338623,
      "learning_rate": 3.8561675011379155e-05,
      "loss": 0.2149,
      "step": 13500
    },
    {
      "epoch": 2.5030719519410187,
      "grad_norm": 2.925417423248291,
      "learning_rate": 3.742375967228038e-05,
      "loss": 0.2127,
      "step": 13750
    },
    {
      "epoch": 2.548582351067219,
      "grad_norm": 3.013047218322754,
      "learning_rate": 3.6285844333181614e-05,
      "loss": 0.21,
      "step": 14000
    },
    {
      "epoch": 2.5940927501934192,
      "grad_norm": 3.139263153076172,
      "learning_rate": 3.514792899408284e-05,
      "loss": 0.2117,
      "step": 14250
    },
    {
      "epoch": 2.6396031493196195,
      "grad_norm": 3.118286371231079,
      "learning_rate": 3.4010013654984066e-05,
      "loss": 0.2113,
      "step": 14500
    },
    {
      "epoch": 2.6851135484458197,
      "grad_norm": 3.226870536804199,
      "learning_rate": 3.28720983158853e-05,
      "loss": 0.2087,
      "step": 14750
    },
    {
      "epoch": 2.7306239475720204,
      "grad_norm": 3.2298526763916016,
      "learning_rate": 3.173418297678653e-05,
      "loss": 0.2083,
      "step": 15000
    },
    {
      "epoch": 2.7761343466982207,
      "grad_norm": 2.83184814453125,
      "learning_rate": 3.059626763768776e-05,
      "loss": 0.2089,
      "step": 15250
    },
    {
      "epoch": 2.821644745824421,
      "grad_norm": 3.196331262588501,
      "learning_rate": 2.9458352298588988e-05,
      "loss": 0.2051,
      "step": 15500
    },
    {
      "epoch": 2.867155144950621,
      "grad_norm": 2.892674446105957,
      "learning_rate": 2.8320436959490214e-05,
      "loss": 0.2062,
      "step": 15750
    },
    {
      "epoch": 2.9126655440768214,
      "grad_norm": 2.7058119773864746,
      "learning_rate": 2.7182521620391444e-05,
      "loss": 0.2044,
      "step": 16000
    },
    {
      "epoch": 2.958175943203022,
      "grad_norm": 2.6910922527313232,
      "learning_rate": 2.604460628129267e-05,
      "loss": 0.2027,
      "step": 16250
    },
    {
      "epoch": 3.0036863423292224,
      "grad_norm": 2.9895167350769043,
      "learning_rate": 2.4906690942193903e-05,
      "loss": 0.2025,
      "step": 16500
    },
    {
      "epoch": 3.0491967414554226,
      "grad_norm": 2.6071224212646484,
      "learning_rate": 2.376877560309513e-05,
      "loss": 0.1976,
      "step": 16750
    },
    {
      "epoch": 3.094707140581623,
      "grad_norm": 2.265737533569336,
      "learning_rate": 2.263086026399636e-05,
      "loss": 0.1969,
      "step": 17000
    },
    {
      "epoch": 3.140217539707823,
      "grad_norm": 2.356515645980835,
      "learning_rate": 2.149294492489759e-05,
      "loss": 0.1967,
      "step": 17250
    },
    {
      "epoch": 3.1857279388340234,
      "grad_norm": 2.7601630687713623,
      "learning_rate": 2.0355029585798818e-05,
      "loss": 0.1972,
      "step": 17500
    },
    {
      "epoch": 3.231238337960224,
      "grad_norm": 3.170037031173706,
      "learning_rate": 1.9217114246700048e-05,
      "loss": 0.197,
      "step": 17750
    },
    {
      "epoch": 3.2767487370864243,
      "grad_norm": 2.934422016143799,
      "learning_rate": 1.8079198907601274e-05,
      "loss": 0.1933,
      "step": 18000
    },
    {
      "epoch": 3.3222591362126246,
      "grad_norm": 2.9783575534820557,
      "learning_rate": 1.6941283568502504e-05,
      "loss": 0.1947,
      "step": 18250
    },
    {
      "epoch": 3.367769535338825,
      "grad_norm": 2.80299711227417,
      "learning_rate": 1.5803368229403733e-05,
      "loss": 0.1938,
      "step": 18500
    },
    {
      "epoch": 3.413279934465025,
      "grad_norm": 2.343477487564087,
      "learning_rate": 1.4665452890304963e-05,
      "loss": 0.1934,
      "step": 18750
    },
    {
      "epoch": 3.458790333591226,
      "grad_norm": 3.13242769241333,
      "learning_rate": 1.352753755120619e-05,
      "loss": 0.1913,
      "step": 19000
    },
    {
      "epoch": 3.504300732717426,
      "grad_norm": 2.3755791187286377,
      "learning_rate": 1.238962221210742e-05,
      "loss": 0.1916,
      "step": 19250
    },
    {
      "epoch": 3.5498111318436263,
      "grad_norm": 3.209934949874878,
      "learning_rate": 1.1251706873008648e-05,
      "loss": 0.1899,
      "step": 19500
    },
    {
      "epoch": 3.5953215309698265,
      "grad_norm": 2.815525531768799,
      "learning_rate": 1.0113791533909878e-05,
      "loss": 0.1911,
      "step": 19750
    },
    {
      "epoch": 3.640831930096027,
      "grad_norm": 3.1206352710723877,
      "learning_rate": 8.975876194811106e-06,
      "loss": 0.1895,
      "step": 20000
    },
    {
      "epoch": 3.686342329222227,
      "grad_norm": 3.1889894008636475,
      "learning_rate": 7.837960855712335e-06,
      "loss": 0.1889,
      "step": 20250
    },
    {
      "epoch": 3.7318527283484277,
      "grad_norm": 3.095888376235962,
      "learning_rate": 6.700045516613563e-06,
      "loss": 0.1895,
      "step": 20500
    },
    {
      "epoch": 3.777363127474628,
      "grad_norm": 2.9188358783721924,
      "learning_rate": 5.562130177514793e-06,
      "loss": 0.1882,
      "step": 20750
    },
    {
      "epoch": 3.8228735266008282,
      "grad_norm": 2.6947951316833496,
      "learning_rate": 4.4242148384160225e-06,
      "loss": 0.1871,
      "step": 21000
    },
    {
      "epoch": 3.8683839257270285,
      "grad_norm": 2.88930606842041,
      "learning_rate": 3.286299499317251e-06,
      "loss": 0.1884,
      "step": 21250
    },
    {
      "epoch": 3.9138943248532287,
      "grad_norm": 3.069244146347046,
      "learning_rate": 2.14838416021848e-06,
      "loss": 0.1888,
      "step": 21500
    },
    {
      "epoch": 3.9594047239794294,
      "grad_norm": 2.7428455352783203,
      "learning_rate": 1.0104688211197086e-06,
      "loss": 0.1869,
      "step": 21750
    },
    {
      "epoch": 3.9998179584034954,
      "step": 21972,
      "total_flos": 1.737238850670485e+18,
      "train_loss": 0.2707560721716333,
      "train_runtime": 41232.1952,
      "train_samples_per_second": 51.159,
      "train_steps_per_second": 0.533
    }
  ],
  "logging_steps": 250,
  "max_steps": 21972,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.737238850670485e+18,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}