{
  "best_metric": 0.20909027755260468,
  "best_model_checkpoint": "/root/pretrain_utg4java_02/checkpoint-91689",
  "epoch": 40.0,
  "eval_steps": 500,
  "global_step": 94040,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.21,
      "grad_norm": 1.1109575033187866,
      "learning_rate": 9.98e-05,
      "loss": 0.4973,
      "step": 500
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1452946662902832,
      "learning_rate": 0.0001998,
      "loss": 0.4793,
      "step": 1000
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9271925091743469,
      "learning_rate": 0.0001989273430782459,
      "loss": 0.4999,
      "step": 1500
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9055812358856201,
      "learning_rate": 0.0001978525365434222,
      "loss": 0.4934,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.4301895797252655,
      "eval_runtime": 56.7204,
      "eval_samples_per_second": 331.733,
      "eval_steps_per_second": 2.592,
      "step": 2351
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8988948464393616,
      "learning_rate": 0.00019677773000859847,
      "loss": 0.4818,
      "step": 2500
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9011508822441101,
      "learning_rate": 0.00019570292347377474,
      "loss": 0.4651,
      "step": 3000
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.905346691608429,
      "learning_rate": 0.00019463026655202063,
      "loss": 0.467,
      "step": 3500
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7411425709724426,
      "learning_rate": 0.0001935597592433362,
      "loss": 0.4625,
      "step": 4000
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8985588550567627,
      "learning_rate": 0.00019248495270851249,
      "loss": 0.4572,
      "step": 4500
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.4058227837085724,
      "eval_runtime": 56.7252,
      "eval_samples_per_second": 331.705,
      "eval_steps_per_second": 2.591,
      "step": 4702
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.827357292175293,
      "learning_rate": 0.00019141014617368875,
      "loss": 0.4472,
      "step": 5000
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.736473023891449,
      "learning_rate": 0.00019033533963886502,
      "loss": 0.4372,
      "step": 5500
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.7558779120445251,
      "learning_rate": 0.00018926053310404129,
      "loss": 0.4362,
      "step": 6000
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.7970576882362366,
      "learning_rate": 0.00018818572656921755,
      "loss": 0.4332,
      "step": 6500
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.7692726850509644,
      "learning_rate": 0.00018711092003439382,
      "loss": 0.43,
      "step": 7000
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.38832882046699524,
      "eval_runtime": 56.9476,
      "eval_samples_per_second": 330.409,
      "eval_steps_per_second": 2.581,
      "step": 7053
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.7109490633010864,
      "learning_rate": 0.00018603611349957009,
      "loss": 0.4217,
      "step": 7500
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.7125234603881836,
      "learning_rate": 0.00018496130696474635,
      "loss": 0.4141,
      "step": 8000
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.6914758682250977,
      "learning_rate": 0.00018388650042992262,
      "loss": 0.4169,
      "step": 8500
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.7215930223464966,
      "learning_rate": 0.00018281169389509889,
      "loss": 0.4132,
      "step": 9000
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.3813731372356415,
      "eval_runtime": 56.8812,
      "eval_samples_per_second": 330.794,
      "eval_steps_per_second": 2.584,
      "step": 9404
    },
    {
      "epoch": 4.04,
      "grad_norm": 0.7283121943473816,
      "learning_rate": 0.00018173688736027515,
      "loss": 0.4131,
      "step": 9500
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.6947267055511475,
      "learning_rate": 0.00018066208082545142,
      "loss": 0.4059,
      "step": 10000
    },
    {
      "epoch": 4.47,
      "grad_norm": 0.7544755339622498,
      "learning_rate": 0.0001795872742906277,
      "loss": 0.4023,
      "step": 10500
    },
    {
      "epoch": 4.68,
      "grad_norm": 0.6204919815063477,
      "learning_rate": 0.00017851246775580398,
      "loss": 0.4038,
      "step": 11000
    },
    {
      "epoch": 4.89,
      "grad_norm": 0.7036948800086975,
      "learning_rate": 0.00017743766122098022,
      "loss": 0.4017,
      "step": 11500
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.3638169765472412,
      "eval_runtime": 56.8241,
      "eval_samples_per_second": 331.127,
      "eval_steps_per_second": 2.587,
      "step": 11755
    },
    {
      "epoch": 5.1,
      "grad_norm": 0.62198805809021,
      "learning_rate": 0.0001763628546861565,
      "loss": 0.3921,
      "step": 12000
    },
    {
      "epoch": 5.32,
      "grad_norm": 0.6095067858695984,
      "learning_rate": 0.00017528804815133278,
      "loss": 0.3901,
      "step": 12500
    },
    {
      "epoch": 5.53,
      "grad_norm": 1.0518782138824463,
      "learning_rate": 0.00017421324161650902,
      "loss": 0.3888,
      "step": 13000
    },
    {
      "epoch": 5.74,
      "grad_norm": 0.6801586151123047,
      "learning_rate": 0.00017313843508168531,
      "loss": 0.3859,
      "step": 13500
    },
    {
      "epoch": 5.95,
      "grad_norm": 0.6489918828010559,
      "learning_rate": 0.00017206362854686158,
      "loss": 0.3839,
      "step": 14000
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.35733696818351746,
      "eval_runtime": 56.779,
      "eval_samples_per_second": 331.39,
      "eval_steps_per_second": 2.589,
      "step": 14106
    },
    {
      "epoch": 6.17,
      "grad_norm": 0.6174113154411316,
      "learning_rate": 0.00017098882201203782,
      "loss": 0.3782,
      "step": 14500
    },
    {
      "epoch": 6.38,
      "grad_norm": 0.641315221786499,
      "learning_rate": 0.00016991401547721411,
      "loss": 0.3792,
      "step": 15000
    },
    {
      "epoch": 6.59,
      "grad_norm": 0.6454831957817078,
      "learning_rate": 0.00016883920894239038,
      "loss": 0.3745,
      "step": 15500
    },
    {
      "epoch": 6.81,
      "grad_norm": 0.6489464640617371,
      "learning_rate": 0.00016776440240756665,
      "loss": 0.3729,
      "step": 16000
    },
    {
      "epoch": 7.0,
      "eval_loss": 0.3489345610141754,
      "eval_runtime": 57.0727,
      "eval_samples_per_second": 329.685,
      "eval_steps_per_second": 2.576,
      "step": 16457
    },
    {
      "epoch": 7.02,
      "grad_norm": 0.6929277181625366,
      "learning_rate": 0.0001666938950988822,
      "loss": 0.3718,
      "step": 16500
    },
    {
      "epoch": 7.23,
      "grad_norm": 0.5726625323295593,
      "learning_rate": 0.00016562123817712813,
      "loss": 0.3624,
      "step": 17000
    },
    {
      "epoch": 7.44,
      "grad_norm": 0.603529155254364,
      "learning_rate": 0.0001645464316423044,
      "loss": 0.3664,
      "step": 17500
    },
    {
      "epoch": 7.66,
      "grad_norm": 0.7436888813972473,
      "learning_rate": 0.00016347162510748066,
      "loss": 0.3614,
      "step": 18000
    },
    {
      "epoch": 7.87,
      "grad_norm": 0.6405676007270813,
      "learning_rate": 0.00016239681857265693,
      "loss": 0.3646,
      "step": 18500
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.3384884297847748,
      "eval_runtime": 56.7342,
      "eval_samples_per_second": 331.652,
      "eval_steps_per_second": 2.591,
      "step": 18808
    },
    {
      "epoch": 8.08,
      "grad_norm": 0.660517156124115,
      "learning_rate": 0.0001613220120378332,
      "loss": 0.3623,
      "step": 19000
    },
    {
      "epoch": 8.29,
      "grad_norm": 0.6352570056915283,
      "learning_rate": 0.0001602493551160791,
      "loss": 0.3518,
      "step": 19500
    },
    {
      "epoch": 8.51,
      "grad_norm": 0.6094992756843567,
      "learning_rate": 0.0001591745485812554,
      "loss": 0.3569,
      "step": 20000
    },
    {
      "epoch": 8.72,
      "grad_norm": 0.6646167635917664,
      "learning_rate": 0.00015809974204643166,
      "loss": 0.3567,
      "step": 20500
    },
    {
      "epoch": 8.93,
      "grad_norm": 0.6240518093109131,
      "learning_rate": 0.00015702493551160792,
      "loss": 0.353,
      "step": 21000
    },
    {
      "epoch": 9.0,
      "eval_loss": 0.3339434862136841,
      "eval_runtime": 56.979,
      "eval_samples_per_second": 330.227,
      "eval_steps_per_second": 2.58,
      "step": 21159
    },
    {
      "epoch": 9.15,
      "grad_norm": 0.6990819573402405,
      "learning_rate": 0.0001559501289767842,
      "loss": 0.3491,
      "step": 21500
    },
    {
      "epoch": 9.36,
      "grad_norm": 0.6751831769943237,
      "learning_rate": 0.00015487532244196046,
      "loss": 0.346,
      "step": 22000
    },
    {
      "epoch": 9.57,
      "grad_norm": 0.615511417388916,
      "learning_rate": 0.00015380051590713672,
      "loss": 0.3488,
      "step": 22500
    },
    {
      "epoch": 9.78,
      "grad_norm": 0.5277218222618103,
      "learning_rate": 0.000152725709372313,
      "loss": 0.3466,
      "step": 23000
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.5809108018875122,
      "learning_rate": 0.00015165090283748926,
      "loss": 0.3441,
      "step": 23500
    },
    {
      "epoch": 10.0,
      "eval_loss": 0.324989378452301,
      "eval_runtime": 56.9322,
      "eval_samples_per_second": 330.499,
      "eval_steps_per_second": 2.582,
      "step": 23510
    },
    {
      "epoch": 10.21,
      "grad_norm": 0.5285485982894897,
      "learning_rate": 0.00015058039552880484,
      "loss": 0.3407,
      "step": 24000
    },
    {
      "epoch": 10.42,
      "grad_norm": 0.6268051862716675,
      "learning_rate": 0.00014950558899398108,
      "loss": 0.3389,
      "step": 24500
    },
    {
      "epoch": 10.63,
      "grad_norm": 0.5879684090614319,
      "learning_rate": 0.00014843078245915735,
      "loss": 0.3387,
      "step": 25000
    },
    {
      "epoch": 10.85,
      "grad_norm": 0.5434576869010925,
      "learning_rate": 0.00014735597592433364,
      "loss": 0.3379,
      "step": 25500
    },
    {
      "epoch": 11.0,
      "eval_loss": 0.31874018907546997,
      "eval_runtime": 57.0459,
      "eval_samples_per_second": 329.84,
      "eval_steps_per_second": 2.577,
      "step": 25861
    },
    {
      "epoch": 11.06,
      "grad_norm": 0.6547732949256897,
      "learning_rate": 0.00014628116938950988,
      "loss": 0.3369,
      "step": 26000
    },
    {
      "epoch": 11.27,
      "grad_norm": 0.648601233959198,
      "learning_rate": 0.00014520636285468615,
      "loss": 0.3308,
      "step": 26500
    },
    {
      "epoch": 11.48,
      "grad_norm": 0.6311826109886169,
      "learning_rate": 0.00014413155631986244,
      "loss": 0.3298,
      "step": 27000
    },
    {
      "epoch": 11.7,
      "grad_norm": 0.5571895837783813,
      "learning_rate": 0.00014305674978503868,
      "loss": 0.3292,
      "step": 27500
    },
    {
      "epoch": 11.91,
      "grad_norm": 0.5373964309692383,
      "learning_rate": 0.00014198194325021498,
      "loss": 0.3284,
      "step": 28000
    },
    {
      "epoch": 12.0,
      "eval_loss": 0.3123379051685333,
      "eval_runtime": 56.7858,
      "eval_samples_per_second": 331.351,
      "eval_steps_per_second": 2.589,
      "step": 28212
    },
    {
      "epoch": 12.12,
      "grad_norm": 0.5307437777519226,
      "learning_rate": 0.00014090928632846088,
      "loss": 0.3247,
      "step": 28500
    },
    {
      "epoch": 12.34,
      "grad_norm": 0.6422705054283142,
      "learning_rate": 0.00013983447979363714,
      "loss": 0.3223,
      "step": 29000
    },
    {
      "epoch": 12.55,
      "grad_norm": 0.6680347323417664,
      "learning_rate": 0.00013875967325881344,
      "loss": 0.3223,
      "step": 29500
    },
    {
      "epoch": 12.76,
      "grad_norm": 0.5851155519485474,
      "learning_rate": 0.00013768701633705933,
      "loss": 0.3245,
      "step": 30000
    },
    {
      "epoch": 12.97,
      "grad_norm": 0.544624924659729,
      "learning_rate": 0.0001366122098022356,
      "loss": 0.3252,
      "step": 30500
    },
    {
      "epoch": 13.0,
      "eval_loss": 0.3076089322566986,
      "eval_runtime": 57.0752,
      "eval_samples_per_second": 329.67,
      "eval_steps_per_second": 2.576,
      "step": 30563
    },
    {
      "epoch": 13.19,
      "grad_norm": 0.5343388319015503,
      "learning_rate": 0.00013553740326741187,
      "loss": 0.3185,
      "step": 31000
    },
    {
      "epoch": 13.4,
      "grad_norm": 0.6074999570846558,
      "learning_rate": 0.00013446259673258814,
      "loss": 0.3191,
      "step": 31500
    },
    {
      "epoch": 13.61,
      "grad_norm": 0.6202041506767273,
      "learning_rate": 0.00013338779019776443,
      "loss": 0.3186,
      "step": 32000
    },
    {
      "epoch": 13.82,
      "grad_norm": 0.6058717966079712,
      "learning_rate": 0.00013231298366294067,
      "loss": 0.3134,
      "step": 32500
    },
    {
      "epoch": 14.0,
      "eval_loss": 0.3018127679824829,
      "eval_runtime": 57.0557,
      "eval_samples_per_second": 329.783,
      "eval_steps_per_second": 2.576,
      "step": 32914
    },
    {
      "epoch": 14.04,
      "grad_norm": 0.6439224481582642,
      "learning_rate": 0.00013123817712811694,
      "loss": 0.3161,
      "step": 33000
    },
    {
      "epoch": 14.25,
      "grad_norm": 0.5888046622276306,
      "learning_rate": 0.00013016337059329323,
      "loss": 0.3094,
      "step": 33500
    },
    {
      "epoch": 14.46,
      "grad_norm": 0.6644863486289978,
      "learning_rate": 0.00012908856405846947,
      "loss": 0.3092,
      "step": 34000
    },
    {
      "epoch": 14.67,
      "grad_norm": 0.5649863481521606,
      "learning_rate": 0.00012801375752364576,
      "loss": 0.3117,
      "step": 34500
    },
    {
      "epoch": 14.89,
      "grad_norm": 0.4947267174720764,
      "learning_rate": 0.00012693895098882203,
      "loss": 0.3072,
      "step": 35000
    },
    {
      "epoch": 15.0,
      "eval_loss": 0.29680636525154114,
      "eval_runtime": 57.044,
      "eval_samples_per_second": 329.851,
      "eval_steps_per_second": 2.577,
      "step": 35265
    },
    {
      "epoch": 15.1,
      "grad_norm": 0.6015808582305908,
      "learning_rate": 0.00012586414445399827,
      "loss": 0.3067,
      "step": 35500
    },
    {
      "epoch": 15.31,
      "grad_norm": 0.6003320217132568,
      "learning_rate": 0.00012478933791917456,
      "loss": 0.3027,
      "step": 36000
    },
    {
      "epoch": 15.53,
      "grad_norm": 0.5643883347511292,
      "learning_rate": 0.00012371453138435083,
      "loss": 0.3012,
      "step": 36500
    },
    {
      "epoch": 15.74,
      "grad_norm": 0.5768193602561951,
      "learning_rate": 0.00012264187446259673,
      "loss": 0.303,
      "step": 37000
    },
    {
      "epoch": 15.95,
      "grad_norm": 0.5599430799484253,
      "learning_rate": 0.00012156921754084265,
      "loss": 0.3034,
      "step": 37500
    },
    {
      "epoch": 16.0,
      "eval_loss": 0.2938060164451599,
      "eval_runtime": 56.7422,
      "eval_samples_per_second": 331.605,
      "eval_steps_per_second": 2.591,
      "step": 37616
    },
    {
      "epoch": 16.16,
      "grad_norm": 0.6380453705787659,
      "learning_rate": 0.00012049441100601892,
      "loss": 0.2987,
      "step": 38000
    },
    {
      "epoch": 16.38,
      "grad_norm": 0.5312320590019226,
      "learning_rate": 0.0001194196044711952,
      "loss": 0.2975,
      "step": 38500
    },
    {
      "epoch": 16.59,
      "grad_norm": 0.7027547955513,
      "learning_rate": 0.00011834479793637145,
      "loss": 0.2934,
      "step": 39000
    },
    {
      "epoch": 16.8,
      "grad_norm": 0.5144539475440979,
      "learning_rate": 0.00011726999140154773,
      "loss": 0.2959,
      "step": 39500
    },
    {
      "epoch": 17.0,
      "eval_loss": 0.28754693269729614,
      "eval_runtime": 56.9555,
      "eval_samples_per_second": 330.363,
      "eval_steps_per_second": 2.581,
      "step": 39967
    },
    {
      "epoch": 17.01,
      "grad_norm": 0.5759513974189758,
      "learning_rate": 0.00011619733447979365,
      "loss": 0.2997,
      "step": 40000
    },
    {
      "epoch": 17.23,
      "grad_norm": 0.593640148639679,
      "learning_rate": 0.00011512467755803955,
      "loss": 0.294,
      "step": 40500
    },
    {
      "epoch": 17.44,
      "grad_norm": 0.6821351647377014,
      "learning_rate": 0.00011404987102321583,
      "loss": 0.2904,
      "step": 41000
    },
    {
      "epoch": 17.65,
      "grad_norm": 0.5297681093215942,
      "learning_rate": 0.0001129750644883921,
      "loss": 0.2896,
      "step": 41500
    },
    {
      "epoch": 17.86,
      "grad_norm": 0.5864290595054626,
      "learning_rate": 0.00011190025795356836,
      "loss": 0.2884,
      "step": 42000
    },
    {
      "epoch": 18.0,
      "eval_loss": 0.28015270829200745,
      "eval_runtime": 56.8879,
      "eval_samples_per_second": 330.756,
      "eval_steps_per_second": 2.584,
      "step": 42318
    },
    {
      "epoch": 18.08,
      "grad_norm": 0.5764068961143494,
      "learning_rate": 0.00011082545141874463,
      "loss": 0.2874,
      "step": 42500
    },
    {
      "epoch": 18.29,
      "grad_norm": 0.6073163747787476,
      "learning_rate": 0.0001097506448839209,
      "loss": 0.2828,
      "step": 43000
    },
    {
      "epoch": 18.5,
      "grad_norm": 0.5691092610359192,
      "learning_rate": 0.00010867583834909716,
      "loss": 0.2848,
      "step": 43500
    },
    {
      "epoch": 18.72,
      "grad_norm": 0.5399264097213745,
      "learning_rate": 0.00010760103181427344,
      "loss": 0.2861,
      "step": 44000
    },
    {
      "epoch": 18.93,
      "grad_norm": 0.6221365928649902,
      "learning_rate": 0.0001065262252794497,
      "loss": 0.2839,
      "step": 44500
    },
    {
      "epoch": 19.0,
      "eval_loss": 0.27618157863616943,
      "eval_runtime": 56.9107,
      "eval_samples_per_second": 330.623,
      "eval_steps_per_second": 2.583,
      "step": 44669
    },
    {
      "epoch": 19.14,
      "grad_norm": 0.5080223679542542,
      "learning_rate": 0.00010545141874462596,
      "loss": 0.2791,
      "step": 45000
    },
    {
      "epoch": 19.35,
      "grad_norm": 0.567764163017273,
      "learning_rate": 0.00010437661220980224,
      "loss": 0.2792,
      "step": 45500
    },
    {
      "epoch": 19.57,
      "grad_norm": 0.5838685035705566,
      "learning_rate": 0.00010330180567497851,
      "loss": 0.2768,
      "step": 46000
    },
    {
      "epoch": 19.78,
      "grad_norm": 0.5778998136520386,
      "learning_rate": 0.00010222699914015476,
      "loss": 0.2776,
      "step": 46500
    },
    {
      "epoch": 19.99,
      "grad_norm": 0.615337073802948,
      "learning_rate": 0.0001011543422184007,
      "loss": 0.2785,
      "step": 47000
    },
    {
      "epoch": 20.0,
      "eval_loss": 0.2724262773990631,
      "eval_runtime": 56.852,
      "eval_samples_per_second": 330.965,
      "eval_steps_per_second": 2.586,
      "step": 47020
    },
    {
      "epoch": 20.2,
      "grad_norm": 0.6153652667999268,
      "learning_rate": 0.00010007953568357695,
      "loss": 0.2712,
      "step": 47500
    },
    {
      "epoch": 20.42,
      "grad_norm": 0.6126906275749207,
      "learning_rate": 9.900472914875323e-05,
      "loss": 0.2718,
      "step": 48000
    },
    {
      "epoch": 20.63,
      "grad_norm": 0.5799471139907837,
      "learning_rate": 9.792992261392949e-05,
      "loss": 0.2701,
      "step": 48500
    },
    {
      "epoch": 20.84,
      "grad_norm": 0.5679476261138916,
      "learning_rate": 9.685511607910577e-05,
      "loss": 0.2727,
      "step": 49000
    },
    {
      "epoch": 21.0,
      "eval_loss": 0.26610955595970154,
      "eval_runtime": 56.7263,
      "eval_samples_per_second": 331.698,
      "eval_steps_per_second": 2.591,
      "step": 49371
    },
    {
      "epoch": 21.05,
      "grad_norm": 0.5923852324485779,
      "learning_rate": 9.578245915735168e-05,
      "loss": 0.2694,
      "step": 49500
    },
    {
      "epoch": 21.27,
      "grad_norm": 0.6184679269790649,
      "learning_rate": 9.470765262252796e-05,
      "loss": 0.2649,
      "step": 50000
    },
    {
      "epoch": 21.48,
      "grad_norm": 0.5939807295799255,
      "learning_rate": 9.363284608770421e-05,
      "loss": 0.2697,
      "step": 50500
    },
    {
      "epoch": 21.69,
      "grad_norm": 0.5925255417823792,
      "learning_rate": 9.255803955288048e-05,
      "loss": 0.2652,
      "step": 51000
    },
    {
      "epoch": 21.91,
      "grad_norm": 0.5666438341140747,
      "learning_rate": 9.148323301805676e-05,
      "loss": 0.266,
      "step": 51500
    },
    {
      "epoch": 22.0,
      "eval_loss": 0.266626238822937,
      "eval_runtime": 56.6361,
      "eval_samples_per_second": 332.226,
      "eval_steps_per_second": 2.596,
      "step": 51722
    },
    {
      "epoch": 22.12,
      "grad_norm": 0.5490565299987793,
      "learning_rate": 9.040842648323303e-05,
      "loss": 0.2613,
      "step": 52000
    },
    {
      "epoch": 22.33,
      "grad_norm": 0.6534095406532288,
      "learning_rate": 8.933361994840929e-05,
      "loss": 0.2592,
      "step": 52500
    },
    {
      "epoch": 22.54,
      "grad_norm": 0.568953812122345,
      "learning_rate": 8.82609630266552e-05,
      "loss": 0.2604,
      "step": 53000
    },
    {
      "epoch": 22.76,
      "grad_norm": 0.5968701243400574,
      "learning_rate": 8.718615649183147e-05,
      "loss": 0.2602,
      "step": 53500
    },
    {
      "epoch": 22.97,
      "grad_norm": 0.5995833277702332,
      "learning_rate": 8.61134995700774e-05,
      "loss": 0.2614,
      "step": 54000
    },
    {
      "epoch": 23.0,
      "eval_loss": 0.26461803913116455,
      "eval_runtime": 56.8736,
      "eval_samples_per_second": 330.839,
      "eval_steps_per_second": 2.585,
      "step": 54073
    },
    {
      "epoch": 23.18,
      "grad_norm": 0.577324390411377,
      "learning_rate": 8.503869303525366e-05,
      "loss": 0.2502,
      "step": 54500
    },
    {
      "epoch": 23.39,
      "grad_norm": 0.631363034248352,
      "learning_rate": 8.396603611349956e-05,
      "loss": 0.256,
      "step": 55000
    },
    {
      "epoch": 23.61,
      "grad_norm": 0.5862709879875183,
      "learning_rate": 8.289122957867584e-05,
      "loss": 0.2559,
      "step": 55500
    },
    {
      "epoch": 23.82,
      "grad_norm": 0.5620314478874207,
      "learning_rate": 8.181857265692176e-05,
      "loss": 0.2549,
      "step": 56000
    },
    {
      "epoch": 24.0,
      "eval_loss": 0.2593855559825897,
      "eval_runtime": 57.337,
      "eval_samples_per_second": 328.165,
      "eval_steps_per_second": 2.564,
      "step": 56424
    },
    {
      "epoch": 24.03,
      "grad_norm": 0.5514592528343201,
      "learning_rate": 8.074376612209804e-05,
      "loss": 0.2542,
      "step": 56500
    },
    {
      "epoch": 24.25,
      "grad_norm": 0.6351886987686157,
      "learning_rate": 7.966895958727429e-05,
      "loss": 0.2483,
      "step": 57000
    },
    {
      "epoch": 24.46,
      "grad_norm": 0.6075631380081177,
      "learning_rate": 7.859415305245056e-05,
      "loss": 0.2503,
      "step": 57500
    },
    {
      "epoch": 24.67,
      "grad_norm": 0.5592435002326965,
      "learning_rate": 7.751934651762684e-05,
      "loss": 0.249,
      "step": 58000
    },
    {
      "epoch": 24.88,
      "grad_norm": 0.5757043957710266,
      "learning_rate": 7.64445399828031e-05,
      "loss": 0.2497,
      "step": 58500
    },
    {
      "epoch": 25.0,
      "eval_loss": 0.2535327076911926,
      "eval_runtime": 56.789,
      "eval_samples_per_second": 331.332,
      "eval_steps_per_second": 2.589,
      "step": 58775
    },
    {
      "epoch": 25.1,
      "grad_norm": 0.5995878577232361,
      "learning_rate": 7.536973344797937e-05,
      "loss": 0.245,
      "step": 59000
    },
    {
      "epoch": 25.31,
      "grad_norm": 0.6077148914337158,
      "learning_rate": 7.429492691315564e-05,
      "loss": 0.2461,
      "step": 59500
    },
    {
      "epoch": 25.52,
      "grad_norm": 0.5592058300971985,
      "learning_rate": 7.32201203783319e-05,
      "loss": 0.2433,
      "step": 60000
    },
    {
      "epoch": 25.73,
      "grad_norm": 0.6243628859519958,
      "learning_rate": 7.214531384350817e-05,
      "loss": 0.2444,
      "step": 60500
    },
    {
      "epoch": 25.95,
      "grad_norm": 0.6848371624946594,
      "learning_rate": 7.107050730868444e-05,
      "loss": 0.243,
      "step": 61000
    },
    {
      "epoch": 26.0,
      "eval_loss": 0.250180721282959,
      "eval_runtime": 56.9862,
      "eval_samples_per_second": 330.185,
      "eval_steps_per_second": 2.58,
      "step": 61126
    },
    {
      "epoch": 26.16,
      "grad_norm": 0.5177180171012878,
      "learning_rate": 6.99957007738607e-05,
      "loss": 0.2382,
      "step": 61500
    },
    {
      "epoch": 26.37,
      "grad_norm": 0.6144821047782898,
      "learning_rate": 6.892304385210663e-05,
      "loss": 0.2349,
      "step": 62000
    },
    {
      "epoch": 26.58,
      "grad_norm": 0.5862034559249878,
      "learning_rate": 6.78482373172829e-05,
      "loss": 0.2385,
      "step": 62500
    },
    {
      "epoch": 26.8,
      "grad_norm": 0.5799028277397156,
      "learning_rate": 6.677343078245916e-05,
      "loss": 0.2383,
      "step": 63000
    },
    {
      "epoch": 27.0,
      "eval_loss": 0.24638937413692474,
      "eval_runtime": 56.9276,
      "eval_samples_per_second": 330.525,
      "eval_steps_per_second": 2.582,
      "step": 63477
    },
    {
      "epoch": 27.01,
      "grad_norm": 0.6466034054756165,
      "learning_rate": 6.569862424763543e-05,
      "loss": 0.2375,
      "step": 63500
    },
    {
      "epoch": 27.22,
      "grad_norm": 0.6161571741104126,
      "learning_rate": 6.46238177128117e-05,
      "loss": 0.2337,
      "step": 64000
    },
    {
      "epoch": 27.44,
      "grad_norm": 0.6162907481193542,
      "learning_rate": 6.354901117798796e-05,
      "loss": 0.2328,
      "step": 64500
    },
    {
      "epoch": 27.65,
      "grad_norm": 0.6643475890159607,
      "learning_rate": 6.247420464316423e-05,
      "loss": 0.2316,
      "step": 65000
    },
    {
      "epoch": 27.86,
      "grad_norm": 0.5605382323265076,
      "learning_rate": 6.13993981083405e-05,
      "loss": 0.2325,
      "step": 65500
    },
    {
      "epoch": 28.0,
      "eval_loss": 0.24465161561965942,
      "eval_runtime": 57.0032,
      "eval_samples_per_second": 330.087,
      "eval_steps_per_second": 2.579,
      "step": 65828
    },
    {
      "epoch": 28.07,
      "grad_norm": 0.6108579635620117,
      "learning_rate": 6.032459157351676e-05,
      "loss": 0.231,
      "step": 66000
    },
    {
      "epoch": 28.29,
      "grad_norm": 0.5990898013114929,
      "learning_rate": 5.924978503869304e-05,
      "loss": 0.2269,
      "step": 66500
    },
    {
      "epoch": 28.5,
      "grad_norm": 0.5171389579772949,
      "learning_rate": 5.817712811693895e-05,
      "loss": 0.2281,
      "step": 67000
    },
    {
      "epoch": 28.71,
      "grad_norm": 0.6478968262672424,
      "learning_rate": 5.710232158211523e-05,
      "loss": 0.23,
      "step": 67500
    },
    {
      "epoch": 28.92,
      "grad_norm": 0.5454473495483398,
      "learning_rate": 5.602751504729149e-05,
      "loss": 0.224,
      "step": 68000
    },
    {
      "epoch": 29.0,
      "eval_loss": 0.23835687339305878,
      "eval_runtime": 56.7685,
      "eval_samples_per_second": 331.452,
      "eval_steps_per_second": 2.589,
      "step": 68179
    },
    {
      "epoch": 29.14,
      "grad_norm": 0.5909966826438904,
      "learning_rate": 5.4952708512467756e-05,
      "loss": 0.2237,
      "step": 68500
    },
    {
      "epoch": 29.35,
      "grad_norm": 0.7489617466926575,
      "learning_rate": 5.387790197764403e-05,
      "loss": 0.2218,
      "step": 69000
    },
    {
      "epoch": 29.56,
      "grad_norm": 0.5880895853042603,
      "learning_rate": 5.2803095442820296e-05,
      "loss": 0.2224,
      "step": 69500
    },
    {
      "epoch": 29.77,
      "grad_norm": 0.6126420497894287,
      "learning_rate": 5.1728288907996556e-05,
      "loss": 0.2235,
      "step": 70000
    },
    {
      "epoch": 29.99,
      "grad_norm": 0.7013407349586487,
      "learning_rate": 5.0653482373172836e-05,
      "loss": 0.2189,
      "step": 70500
    },
    {
      "epoch": 30.0,
      "eval_loss": 0.23514553904533386,
      "eval_runtime": 56.6615,
      "eval_samples_per_second": 332.077,
      "eval_steps_per_second": 2.594,
      "step": 70530
    },
    {
      "epoch": 30.2,
      "grad_norm": 0.6700057983398438,
      "learning_rate": 4.9578675838349096e-05,
      "loss": 0.2175,
      "step": 71000
    },
    {
      "epoch": 30.41,
      "grad_norm": 0.5728330016136169,
      "learning_rate": 4.8506018916595015e-05,
      "loss": 0.2169,
      "step": 71500
    },
    {
      "epoch": 30.63,
      "grad_norm": 0.5460176467895508,
      "learning_rate": 4.743121238177128e-05,
      "loss": 0.2159,
      "step": 72000
    },
    {
      "epoch": 30.84,
      "grad_norm": 0.5714312791824341,
      "learning_rate": 4.63585554600172e-05,
      "loss": 0.2131,
      "step": 72500
    },
    {
      "epoch": 31.0,
      "eval_loss": 0.23221717774868011,
      "eval_runtime": 56.8018,
      "eval_samples_per_second": 331.257,
      "eval_steps_per_second": 2.588,
      "step": 72881
    },
    {
      "epoch": 31.05,
      "grad_norm": 0.5165086984634399,
      "learning_rate": 4.528374892519347e-05,
      "loss": 0.2139,
      "step": 73000
    },
    {
      "epoch": 31.26,
      "grad_norm": 0.6647119522094727,
      "learning_rate": 4.4208942390369734e-05,
      "loss": 0.2098,
      "step": 73500
    },
    {
      "epoch": 31.48,
      "grad_norm": 0.6132521629333496,
      "learning_rate": 4.3134135855546e-05,
      "loss": 0.213,
      "step": 74000
    },
    {
      "epoch": 31.69,
      "grad_norm": 0.5501160025596619,
      "learning_rate": 4.2059329320722275e-05,
      "loss": 0.2111,
      "step": 74500
    },
    {
      "epoch": 31.9,
      "grad_norm": 0.6153804659843445,
      "learning_rate": 4.098667239896819e-05,
      "loss": 0.2095,
      "step": 75000
    },
    {
      "epoch": 32.0,
      "eval_loss": 0.22537025809288025,
      "eval_runtime": 56.6384,
      "eval_samples_per_second": 332.213,
      "eval_steps_per_second": 2.595,
      "step": 75232
    },
    {
      "epoch": 32.11,
      "grad_norm": 0.5587669014930725,
      "learning_rate": 3.9911865864144454e-05,
      "loss": 0.2102,
      "step": 75500
    },
    {
      "epoch": 32.33,
      "grad_norm": 0.6313532590866089,
      "learning_rate": 3.883705932932073e-05,
      "loss": 0.2066,
      "step": 76000
    },
    {
      "epoch": 32.54,
      "grad_norm": 0.6371473670005798,
      "learning_rate": 3.7762252794496994e-05,
      "loss": 0.2067,
      "step": 76500
    },
    {
      "epoch": 32.75,
      "grad_norm": 0.6398050785064697,
      "learning_rate": 3.668744625967326e-05,
      "loss": 0.2049,
      "step": 77000
    },
    {
      "epoch": 32.96,
      "grad_norm": 0.6136685013771057,
      "learning_rate": 3.561263972484953e-05,
      "loss": 0.2072,
      "step": 77500
    },
    {
      "epoch": 33.0,
      "eval_loss": 0.22324109077453613,
      "eval_runtime": 56.9404,
      "eval_samples_per_second": 330.451,
      "eval_steps_per_second": 2.582,
      "step": 77583
    },
    {
      "epoch": 33.18,
      "grad_norm": 0.652153730392456,
      "learning_rate": 3.45378331900258e-05,
      "loss": 0.2024,
      "step": 78000
    },
    {
      "epoch": 33.39,
      "grad_norm": 0.5828467011451721,
      "learning_rate": 3.346302665520207e-05,
      "loss": 0.2008,
      "step": 78500
    },
    {
      "epoch": 33.6,
      "grad_norm": 0.5484930872917175,
      "learning_rate": 3.239036973344798e-05,
      "loss": 0.2032,
      "step": 79000
    },
    {
      "epoch": 33.82,
      "grad_norm": 0.5094757676124573,
      "learning_rate": 3.1315563198624247e-05,
      "loss": 0.2012,
      "step": 79500
    },
    {
      "epoch": 34.0,
      "eval_loss": 0.22044067084789276,
      "eval_runtime": 56.7734,
      "eval_samples_per_second": 331.423,
      "eval_steps_per_second": 2.589,
      "step": 79934
    },
    {
      "epoch": 34.03,
      "grad_norm": 0.6236255764961243,
      "learning_rate": 3.0240756663800517e-05,
      "loss": 0.1984,
      "step": 80000
    },
    {
      "epoch": 34.24,
      "grad_norm": 0.6019966006278992,
      "learning_rate": 2.9165950128976787e-05,
      "loss": 0.1958,
      "step": 80500
    },
    {
      "epoch": 34.45,
      "grad_norm": 0.6366199851036072,
      "learning_rate": 2.8093293207222702e-05,
      "loss": 0.1988,
      "step": 81000
    },
    {
      "epoch": 34.67,
      "grad_norm": 0.6756080389022827,
      "learning_rate": 2.701848667239897e-05,
      "loss": 0.1975,
      "step": 81500
    },
    {
      "epoch": 34.88,
      "grad_norm": 0.587624728679657,
      "learning_rate": 2.594368013757524e-05,
      "loss": 0.1938,
      "step": 82000
    },
    {
      "epoch": 35.0,
      "eval_loss": 0.21603567898273468,
      "eval_runtime": 56.8324,
      "eval_samples_per_second": 331.079,
      "eval_steps_per_second": 2.587,
      "step": 82285
    },
    {
      "epoch": 35.09,
      "grad_norm": 0.5856329798698425,
      "learning_rate": 2.4868873602751506e-05,
      "loss": 0.193,
      "step": 82500
    },
    {
      "epoch": 35.3,
      "grad_norm": 0.6045832633972168,
      "learning_rate": 2.3794067067927776e-05,
      "loss": 0.1936,
      "step": 83000
    },
    {
      "epoch": 35.52,
      "grad_norm": 0.5652228593826294,
      "learning_rate": 2.2719260533104043e-05,
      "loss": 0.194,
      "step": 83500
    },
    {
      "epoch": 35.73,
      "grad_norm": 0.6424867510795593,
      "learning_rate": 2.1644453998280313e-05,
      "loss": 0.1896,
      "step": 84000
    },
    {
      "epoch": 35.94,
      "grad_norm": 0.5543941855430603,
      "learning_rate": 2.0571797076526225e-05,
      "loss": 0.1921,
      "step": 84500
    },
    {
      "epoch": 36.0,
      "eval_loss": 0.21675606071949005,
      "eval_runtime": 56.9718,
      "eval_samples_per_second": 330.269,
      "eval_steps_per_second": 2.58,
      "step": 84636
    },
    {
      "epoch": 36.15,
      "grad_norm": 0.6019295454025269,
      "learning_rate": 1.9496990541702495e-05,
      "loss": 0.1884,
      "step": 85000
    },
    {
      "epoch": 36.37,
      "grad_norm": 0.645598292350769,
      "learning_rate": 1.8422184006878762e-05,
      "loss": 0.1888,
      "step": 85500
    },
    {
      "epoch": 36.58,
      "grad_norm": 0.6292818188667297,
      "learning_rate": 1.7347377472055032e-05,
      "loss": 0.1886,
      "step": 86000
    },
    {
      "epoch": 36.79,
      "grad_norm": 0.5169576406478882,
      "learning_rate": 1.62725709372313e-05,
      "loss": 0.1898,
      "step": 86500
    },
    {
      "epoch": 37.0,
      "eval_loss": 0.21486429870128632,
      "eval_runtime": 56.8955,
      "eval_samples_per_second": 330.711,
      "eval_steps_per_second": 2.584,
      "step": 86987
    },
    {
      "epoch": 37.01,
      "grad_norm": 0.680117130279541,
      "learning_rate": 1.5197764402407566e-05,
      "loss": 0.187,
      "step": 87000
    },
    {
      "epoch": 37.22,
      "grad_norm": 0.5357288718223572,
      "learning_rate": 1.4122957867583836e-05,
      "loss": 0.1841,
      "step": 87500
    },
    {
      "epoch": 37.43,
      "grad_norm": 0.5753670334815979,
      "learning_rate": 1.3048151332760103e-05,
      "loss": 0.1829,
      "step": 88000
    },
    {
      "epoch": 37.64,
      "grad_norm": 0.654707133769989,
      "learning_rate": 1.1975494411006018e-05,
      "loss": 0.1857,
      "step": 88500
    },
    {
      "epoch": 37.86,
      "grad_norm": 0.5702338218688965,
      "learning_rate": 1.0900687876182287e-05,
      "loss": 0.1829,
      "step": 89000
    },
    {
      "epoch": 38.0,
      "eval_loss": 0.21089451014995575,
      "eval_runtime": 56.6335,
      "eval_samples_per_second": 332.241,
      "eval_steps_per_second": 2.596,
      "step": 89338
    },
    {
      "epoch": 38.07,
      "grad_norm": 0.6671079993247986,
      "learning_rate": 9.825881341358555e-06,
      "loss": 0.1846,
      "step": 89500
    },
    {
      "epoch": 38.28,
      "grad_norm": 0.5165778398513794,
      "learning_rate": 8.751074806534824e-06,
      "loss": 0.1809,
      "step": 90000
    },
    {
      "epoch": 38.49,
      "grad_norm": 0.6374748945236206,
      "learning_rate": 7.676268271711092e-06,
      "loss": 0.1793,
      "step": 90500
    },
    {
      "epoch": 38.71,
      "grad_norm": 0.5679296851158142,
      "learning_rate": 6.60146173688736e-06,
      "loss": 0.1804,
      "step": 91000
    },
    {
      "epoch": 38.92,
      "grad_norm": 0.5970684885978699,
      "learning_rate": 5.526655202063629e-06,
      "loss": 0.1804,
      "step": 91500
    },
    {
      "epoch": 39.0,
      "eval_loss": 0.20909027755260468,
      "eval_runtime": 56.346,
      "eval_samples_per_second": 333.937,
      "eval_steps_per_second": 2.609,
      "step": 91689
    },
    {
      "epoch": 39.13,
      "grad_norm": 0.5444430112838745,
      "learning_rate": 4.4539982803095445e-06,
      "loss": 0.1804,
      "step": 92000
    },
    {
      "epoch": 39.34,
      "grad_norm": 0.6086540818214417,
      "learning_rate": 3.379191745485813e-06,
      "loss": 0.1786,
      "step": 92500
    },
    {
      "epoch": 39.56,
      "grad_norm": 0.5558965802192688,
      "learning_rate": 2.304385210662081e-06,
      "loss": 0.1758,
      "step": 93000
    },
    {
      "epoch": 39.77,
      "grad_norm": 0.5531004667282104,
      "learning_rate": 1.2295786758383493e-06,
      "loss": 0.1788,
      "step": 93500
    },
    {
      "epoch": 39.98,
      "grad_norm": 0.5611669421195984,
      "learning_rate": 1.5477214101461737e-07,
      "loss": 0.1773,
      "step": 94000
    },
    {
      "epoch": 40.0,
      "eval_loss": 0.20976465940475464,
      "eval_runtime": 56.6405,
      "eval_samples_per_second": 332.2,
      "eval_steps_per_second": 2.595,
      "step": 94040
    },
    {
      "epoch": 40.0,
      "step": 94040,
      "total_flos": 3.6650496018087936e+18,
      "train_loss": 0.2880888063569923,
      "train_runtime": 54838.0441,
      "train_samples_per_second": 109.795,
      "train_steps_per_second": 1.715
    }
  ],
  "logging_steps": 500,
  "max_steps": 94040,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 40,
  "save_steps": 500,
  "total_flos": 3.6650496018087936e+18,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}