{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.5003750937734432,
  "eval_steps": 1000,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_loss": 1.6887853145599365,
      "eval_ppl": 5.4129,
      "eval_runtime": 167.3526,
      "eval_samples_per_second": 4.362,
      "eval_steps_per_second": 0.872,
      "memory/device_reserved (GiB)": 139.12,
      "memory/max_active (GiB)": 18.94,
      "memory/max_allocated (GiB)": 18.94,
      "step": 0
    },
    {
      "epoch": 0.018754688672168042,
      "grad_norm": 1.415561556816101,
      "learning_rate": 4.8e-05,
      "loss": 1.6848,
      "memory/device_reserved (GiB)": 139.11,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 5.3914,
      "step": 25,
      "tokens_per_second_per_gpu": 16277.76,
      "total_tokens": 1723633
    },
    {
      "epoch": 0.037509377344336084,
      "grad_norm": 0.33179354667663574,
      "learning_rate": 9.8e-05,
      "loss": 0.9839,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 2.6749,
      "step": 50,
      "tokens_per_second_per_gpu": 4303.21,
      "total_tokens": 2175386
    },
    {
      "epoch": 0.056264066016504126,
      "grad_norm": 0.17453454434871674,
      "learning_rate": 0.000148,
      "loss": 0.8002,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 2.226,
      "step": 75,
      "tokens_per_second_per_gpu": 3776.03,
      "total_tokens": 2623712
    },
    {
      "epoch": 0.07501875468867217,
      "grad_norm": 0.19318008422851562,
      "learning_rate": 0.00019800000000000002,
      "loss": 0.7218,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 2.0581,
      "step": 100,
      "tokens_per_second_per_gpu": 4252.49,
      "total_tokens": 3072519
    },
    {
      "epoch": 0.09377344336084022,
      "grad_norm": 0.18435686826705933,
      "learning_rate": 0.00019998127418269004,
      "loss": 0.6759,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.9658,
      "step": 125,
      "tokens_per_second_per_gpu": 4303.31,
      "total_tokens": 3523983
    },
    {
      "epoch": 0.11252813203300825,
      "grad_norm": 0.19870473444461823,
      "learning_rate": 0.00019992195096972548,
      "loss": 0.6703,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.9548,
      "step": 150,
      "tokens_per_second_per_gpu": 4260.86,
      "total_tokens": 3973452
    },
    {
      "epoch": 0.1312828207051763,
      "grad_norm": 0.20499658584594727,
      "learning_rate": 0.0001998220219574743,
      "loss": 0.6381,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8929,
      "step": 175,
      "tokens_per_second_per_gpu": 4288.64,
      "total_tokens": 4423763
    },
    {
      "epoch": 0.15003750937734434,
      "grad_norm": 0.18934418261051178,
      "learning_rate": 0.00019968152775460537,
      "loss": 0.6383,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8933,
      "step": 200,
      "tokens_per_second_per_gpu": 4244.79,
      "total_tokens": 4872365
    },
    {
      "epoch": 0.16879219804951237,
      "grad_norm": 0.1827855408191681,
      "learning_rate": 0.00019950052545447352,
      "loss": 0.6347,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8865,
      "step": 225,
      "tokens_per_second_per_gpu": 4252.71,
      "total_tokens": 5319322
    },
    {
      "epoch": 0.18754688672168043,
      "grad_norm": 0.16483066976070404,
      "learning_rate": 0.00019927908861191827,
      "loss": 0.6392,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.895,
      "step": 250,
      "tokens_per_second_per_gpu": 3772.0,
      "total_tokens": 5768644
    },
    {
      "epoch": 0.20630157539384847,
      "grad_norm": 0.17186357080936432,
      "learning_rate": 0.00019901730721337302,
      "loss": 0.614,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8478,
      "step": 275,
      "tokens_per_second_per_gpu": 4281.82,
      "total_tokens": 6220751
    },
    {
      "epoch": 0.2250562640660165,
      "grad_norm": 0.18073013424873352,
      "learning_rate": 0.00019871528764029667,
      "loss": 0.6196,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8582,
      "step": 300,
      "tokens_per_second_per_gpu": 4234.51,
      "total_tokens": 6668111
    },
    {
      "epoch": 0.24381095273818454,
      "grad_norm": 0.19639697670936584,
      "learning_rate": 0.00019837315262594306,
      "loss": 0.6181,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8554,
      "step": 325,
      "tokens_per_second_per_gpu": 4261.44,
      "total_tokens": 7117439
    },
    {
      "epoch": 0.2625656414103526,
      "grad_norm": 0.1670486479997635,
      "learning_rate": 0.00019799104120548492,
      "loss": 0.6141,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.848,
      "step": 350,
      "tokens_per_second_per_gpu": 4298.97,
      "total_tokens": 7569060
    },
    {
      "epoch": 0.2813203300825206,
      "grad_norm": 0.17752495408058167,
      "learning_rate": 0.00019756910865951377,
      "loss": 0.6075,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8358,
      "step": 375,
      "tokens_per_second_per_gpu": 4256.6,
      "total_tokens": 8017630
    },
    {
      "epoch": 0.30007501875468867,
      "grad_norm": 0.2000180035829544,
      "learning_rate": 0.00019710752645093747,
      "loss": 0.6108,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8419,
      "step": 400,
      "tokens_per_second_per_gpu": 4245.46,
      "total_tokens": 8464998
    },
    {
      "epoch": 0.31882970742685673,
      "grad_norm": 0.17395919561386108,
      "learning_rate": 0.00019660648215530206,
      "loss": 0.5966,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8159,
      "step": 425,
      "tokens_per_second_per_gpu": 3758.92,
      "total_tokens": 8914723
    },
    {
      "epoch": 0.33758439609902474,
      "grad_norm": 0.18785236775875092,
      "learning_rate": 0.00019606617938456572,
      "loss": 0.6099,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.8402,
      "step": 450,
      "tokens_per_second_per_gpu": 4200.53,
      "total_tokens": 9359638
    },
    {
      "epoch": 0.3563390847711928,
      "grad_norm": 0.17702797055244446,
      "learning_rate": 0.0001954868377043559,
      "loss": 0.5922,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.808,
      "step": 475,
      "tokens_per_second_per_gpu": 4265.36,
      "total_tokens": 9810837
    },
    {
      "epoch": 0.37509377344336087,
      "grad_norm": 0.19927558302879333,
      "learning_rate": 0.00019486869254474337,
      "loss": 0.5759,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7787,
      "step": 500,
      "tokens_per_second_per_gpu": 4276.25,
      "total_tokens": 10261446
    },
    {
      "epoch": 0.3938484621155289,
      "grad_norm": 0.1908370852470398,
      "learning_rate": 0.0001942119951045692,
      "loss": 0.584,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7932,
      "step": 525,
      "tokens_per_second_per_gpu": 4272.28,
      "total_tokens": 10707841
    },
    {
      "epoch": 0.41260315078769694,
      "grad_norm": 0.2064146101474762,
      "learning_rate": 0.00019351701224936383,
      "loss": 0.5791,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7844,
      "step": 550,
      "tokens_per_second_per_gpu": 4250.37,
      "total_tokens": 11155384
    },
    {
      "epoch": 0.43135783945986494,
      "grad_norm": 0.26748332381248474,
      "learning_rate": 0.0001927840264028995,
      "loss": 0.5758,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7786,
      "step": 575,
      "tokens_per_second_per_gpu": 4256.55,
      "total_tokens": 11601192
    },
    {
      "epoch": 0.450112528132033,
      "grad_norm": 0.17514832317829132,
      "learning_rate": 0.00019201333543242036,
      "loss": 0.5791,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7844,
      "step": 600,
      "tokens_per_second_per_gpu": 3770.83,
      "total_tokens": 12048477
    },
    {
      "epoch": 0.46886721680420107,
      "grad_norm": 0.22069169580936432,
      "learning_rate": 0.00019120525252759647,
      "loss": 0.5803,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7866,
      "step": 625,
      "tokens_per_second_per_gpu": 4179.31,
      "total_tokens": 12488141
    },
    {
      "epoch": 0.4876219054763691,
      "grad_norm": 0.20555566251277924,
      "learning_rate": 0.00019036010607325138,
      "loss": 0.5716,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7711,
      "step": 650,
      "tokens_per_second_per_gpu": 4209.96,
      "total_tokens": 12934358
    },
    {
      "epoch": 0.5063765941485371,
      "grad_norm": 0.19018156826496124,
      "learning_rate": 0.00018947823951591478,
      "loss": 0.5608,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7521,
      "step": 675,
      "tokens_per_second_per_gpu": 4226.4,
      "total_tokens": 13378983
    },
    {
      "epoch": 0.5251312828207052,
      "grad_norm": 0.17173859477043152,
      "learning_rate": 0.00018856001122425416,
      "loss": 0.5667,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7624,
      "step": 700,
      "tokens_per_second_per_gpu": 4265.57,
      "total_tokens": 13829519
    },
    {
      "epoch": 0.5438859714928732,
      "grad_norm": 0.17706550657749176,
      "learning_rate": 0.0001876057943434428,
      "loss": 0.565,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7594,
      "step": 725,
      "tokens_per_second_per_gpu": 4281.61,
      "total_tokens": 14281879
    },
    {
      "epoch": 0.5626406601650412,
      "grad_norm": 0.18528586626052856,
      "learning_rate": 0.00018661597664352284,
      "loss": 0.5666,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7623,
      "step": 750,
      "tokens_per_second_per_gpu": 4229.32,
      "total_tokens": 14725919
    },
    {
      "epoch": 0.5813953488372093,
      "grad_norm": 0.16790929436683655,
      "learning_rate": 0.00018559096036182516,
      "loss": 0.5633,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7565,
      "step": 775,
      "tokens_per_second_per_gpu": 3775.0,
      "total_tokens": 15175146
    },
    {
      "epoch": 0.6001500375093773,
      "grad_norm": 0.17511805891990662,
      "learning_rate": 0.00018453116203951005,
      "loss": 0.5664,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7619,
      "step": 800,
      "tokens_per_second_per_gpu": 4218.07,
      "total_tokens": 15619901
    },
    {
      "epoch": 0.6189047261815454,
      "grad_norm": 0.19853387773036957,
      "learning_rate": 0.0001834370123522954,
      "loss": 0.5646,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7587,
      "step": 825,
      "tokens_per_second_per_gpu": 4230.84,
      "total_tokens": 16066102
    },
    {
      "epoch": 0.6376594148537135,
      "grad_norm": 0.18872258067131042,
      "learning_rate": 0.00018230895593544056,
      "loss": 0.552,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7367,
      "step": 850,
      "tokens_per_second_per_gpu": 4222.33,
      "total_tokens": 16510696
    },
    {
      "epoch": 0.6564141035258815,
      "grad_norm": 0.9702818989753723,
      "learning_rate": 0.0001811474512030578,
      "loss": 0.5607,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7519,
      "step": 875,
      "tokens_per_second_per_gpu": 4200.39,
      "total_tokens": 16953918
    },
    {
      "epoch": 0.6751687921980495,
      "grad_norm": 0.17479568719863892,
      "learning_rate": 0.00017995297016182405,
      "loss": 0.564,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7577,
      "step": 900,
      "tokens_per_second_per_gpu": 4210.15,
      "total_tokens": 17396453
    },
    {
      "epoch": 0.6939234808702176,
      "grad_norm": 0.1948954463005066,
      "learning_rate": 0.0001787259982191692,
      "loss": 0.5511,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7352,
      "step": 925,
      "tokens_per_second_per_gpu": 4237.98,
      "total_tokens": 17841287
    },
    {
      "epoch": 0.7126781695423856,
      "grad_norm": 0.19541053473949432,
      "learning_rate": 0.00017746703398601872,
      "loss": 0.5532,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7388,
      "step": 950,
      "tokens_per_second_per_gpu": 3725.33,
      "total_tokens": 18283596
    },
    {
      "epoch": 0.7314328582145536,
      "grad_norm": 0.1818365603685379,
      "learning_rate": 0.0001761765890741701,
      "loss": 0.5521,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7369,
      "step": 975,
      "tokens_per_second_per_gpu": 4211.63,
      "total_tokens": 18726722
    },
    {
      "epoch": 0.7501875468867217,
      "grad_norm": 0.1838025599718094,
      "learning_rate": 0.00017485518788838705,
      "loss": 0.5511,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7352,
      "step": 1000,
      "tokens_per_second_per_gpu": 3962.4,
      "total_tokens": 19167258
    },
    {
      "epoch": 0.7501875468867217,
      "eval_loss": 0.540988564491272,
      "eval_ppl": 1.7177,
      "eval_runtime": 138.0264,
      "eval_samples_per_second": 5.289,
      "eval_steps_per_second": 1.058,
      "memory/device_reserved (GiB)": 139.02,
      "memory/max_active (GiB)": 19.1,
      "memory/max_allocated (GiB)": 19.1,
      "step": 1000
    },
    {
      "epoch": 0.7689422355588897,
      "grad_norm": 0.2199818342924118,
      "learning_rate": 0.00017350336741329413,
      "loss": 0.549,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7315,
      "step": 1025,
      "tokens_per_second_per_gpu": 4129.73,
      "total_tokens": 20870820
    },
    {
      "epoch": 0.7876969242310577,
      "grad_norm": 0.19783177971839905,
      "learning_rate": 0.0001721216769951596,
      "loss": 0.5615,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7533,
      "step": 1050,
      "tokens_per_second_per_gpu": 4243.63,
      "total_tokens": 21317982
    },
    {
      "epoch": 0.8064516129032258,
      "grad_norm": 0.1678430140018463,
      "learning_rate": 0.00017071067811865476,
      "loss": 0.5557,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7432,
      "step": 1075,
      "tokens_per_second_per_gpu": 4092.04,
      "total_tokens": 21754087
    },
    {
      "epoch": 0.8252063015753939,
      "grad_norm": 0.16523879766464233,
      "learning_rate": 0.00016927094417868048,
      "loss": 0.556,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7437,
      "step": 1100,
      "tokens_per_second_per_gpu": 4187.02,
      "total_tokens": 22198779
    },
    {
      "epoch": 0.8439609902475619,
      "grad_norm": 0.18177717924118042,
      "learning_rate": 0.00016780306024735382,
      "loss": 0.5468,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7277,
      "step": 1125,
      "tokens_per_second_per_gpu": 4198.97,
      "total_tokens": 22639769
    },
    {
      "epoch": 0.8627156789197299,
      "grad_norm": 0.17299720644950867,
      "learning_rate": 0.0001663076228362492,
      "loss": 0.554,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7402,
      "step": 1150,
      "tokens_per_second_per_gpu": 3762.13,
      "total_tokens": 23086742
    },
    {
      "epoch": 0.881470367591898,
      "grad_norm": 0.19112971425056458,
      "learning_rate": 0.00016478523965399085,
      "loss": 0.5434,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7219,
      "step": 1175,
      "tokens_per_second_per_gpu": 4205.37,
      "total_tokens": 23528106
    },
    {
      "epoch": 0.900225056264066,
      "grad_norm": 0.17930163443088531,
      "learning_rate": 0.00016323652935929536,
      "loss": 0.5362,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7095,
      "step": 1200,
      "tokens_per_second_per_gpu": 4228.83,
      "total_tokens": 23974427
    },
    {
      "epoch": 0.918979744936234,
      "grad_norm": 0.18718039989471436,
      "learning_rate": 0.00016166212130956382,
      "loss": 0.5533,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.739,
      "step": 1225,
      "tokens_per_second_per_gpu": 4211.64,
      "total_tokens": 24415919
    },
    {
      "epoch": 0.9377344336084021,
      "grad_norm": 0.17105573415756226,
      "learning_rate": 0.0001600626553051268,
      "loss": 0.5492,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7319,
      "step": 1250,
      "tokens_per_second_per_gpu": 4183.86,
      "total_tokens": 24854345
    },
    {
      "epoch": 0.9564891222805701,
      "grad_norm": 0.1733955442905426,
      "learning_rate": 0.0001584387813292454,
      "loss": 0.5348,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7071,
      "step": 1275,
      "tokens_per_second_per_gpu": 4172.93,
      "total_tokens": 25292647
    },
    {
      "epoch": 0.9752438109527382,
      "grad_norm": 0.1858205944299698,
      "learning_rate": 0.00015679115928397401,
      "loss": 0.5527,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7379,
      "step": 1300,
      "tokens_per_second_per_gpu": 4226.34,
      "total_tokens": 25733591
    },
    {
      "epoch": 0.9939984996249063,
      "grad_norm": 0.1944192498922348,
      "learning_rate": 0.00015512045872199276,
      "loss": 0.5311,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.7008,
      "step": 1325,
      "tokens_per_second_per_gpu": 3655.12,
      "total_tokens": 26164528
    },
    {
      "epoch": 1.0127531882970742,
      "grad_norm": 0.18358173966407776,
      "learning_rate": 0.00015342735857451777,
      "loss": 0.5145,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6728,
      "step": 1350,
      "tokens_per_second_per_gpu": 4227.25,
      "total_tokens": 26610460
    },
    {
      "epoch": 1.0315078769692423,
      "grad_norm": 0.1853465735912323,
      "learning_rate": 0.00015171254687540038,
      "loss": 0.5081,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6621,
      "step": 1375,
      "tokens_per_second_per_gpu": 4318.88,
      "total_tokens": 27064008
    },
    {
      "epoch": 1.0502625656414104,
      "grad_norm": 0.18925060331821442,
      "learning_rate": 0.0001499767204815273,
      "loss": 0.5185,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6795,
      "step": 1400,
      "tokens_per_second_per_gpu": 4324.01,
      "total_tokens": 27516590
    },
    {
      "epoch": 1.0690172543135783,
      "grad_norm": 0.20961470901966095,
      "learning_rate": 0.00014822058478963532,
      "loss": 0.5234,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6878,
      "step": 1425,
      "tokens_per_second_per_gpu": 4319.64,
      "total_tokens": 27970075
    },
    {
      "epoch": 1.0877719429857464,
      "grad_norm": 0.1982697695493698,
      "learning_rate": 0.0001464448534496555,
      "loss": 0.5169,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6768,
      "step": 1450,
      "tokens_per_second_per_gpu": 4267.88,
      "total_tokens": 28419716
    },
    {
      "epoch": 1.1065266316579145,
      "grad_norm": 0.1925143301486969,
      "learning_rate": 0.00014465024807470376,
      "loss": 0.5197,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6815,
      "step": 1475,
      "tokens_per_second_per_gpu": 4264.53,
      "total_tokens": 28866312
    },
    {
      "epoch": 1.1252813203300824,
      "grad_norm": 0.18788637220859528,
      "learning_rate": 0.0001428374979478349,
      "loss": 0.5204,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6827,
      "step": 1500,
      "tokens_per_second_per_gpu": 3779.33,
      "total_tokens": 29315968
    },
    {
      "epoch": 1.1440360090022506,
      "grad_norm": 0.18954145908355713,
      "learning_rate": 0.00014100733972568038,
      "loss": 0.5164,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.676,
      "step": 1525,
      "tokens_per_second_per_gpu": 4282.57,
      "total_tokens": 29766723
    },
    {
      "epoch": 1.1627906976744187,
      "grad_norm": 0.19003146886825562,
      "learning_rate": 0.00013916051713908924,
      "loss": 0.5095,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6645,
      "step": 1550,
      "tokens_per_second_per_gpu": 4290.76,
      "total_tokens": 30218573
    },
    {
      "epoch": 1.1815453863465866,
      "grad_norm": 0.18279583752155304,
      "learning_rate": 0.00013729778069089437,
      "loss": 0.522,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6854,
      "step": 1575,
      "tokens_per_second_per_gpu": 4300.13,
      "total_tokens": 30669810
    },
    {
      "epoch": 1.2003000750187547,
      "grad_norm": 0.18783092498779297,
      "learning_rate": 0.00013541988735092672,
      "loss": 0.5003,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6492,
      "step": 1600,
      "tokens_per_second_per_gpu": 4271.27,
      "total_tokens": 31117586
    },
    {
      "epoch": 1.2190547636909228,
      "grad_norm": 0.199558824300766,
      "learning_rate": 0.00013352760024840175,
      "loss": 0.5115,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6678,
      "step": 1625,
      "tokens_per_second_per_gpu": 4248.14,
      "total_tokens": 31562224
    },
    {
      "epoch": 1.2378094523630907,
      "grad_norm": 0.19465653598308563,
      "learning_rate": 0.00013162168836180246,
      "loss": 0.4967,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6433,
      "step": 1650,
      "tokens_per_second_per_gpu": 4286.24,
      "total_tokens": 32011071
    },
    {
      "epoch": 1.2565641410352588,
      "grad_norm": 0.2054641842842102,
      "learning_rate": 0.00012970292620638574,
      "loss": 0.5172,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6773,
      "step": 1675,
      "tokens_per_second_per_gpu": 3733.1,
      "total_tokens": 32452490
    },
    {
      "epoch": 1.275318829707427,
      "grad_norm": 0.19450411200523376,
      "learning_rate": 0.00012777209351943862,
      "loss": 0.5149,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6735,
      "step": 1700,
      "tokens_per_second_per_gpu": 4251.33,
      "total_tokens": 32899103
    },
    {
      "epoch": 1.2940735183795948,
      "grad_norm": 0.19844166934490204,
      "learning_rate": 0.0001258299749434123,
      "loss": 0.5205,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6829,
      "step": 1725,
      "tokens_per_second_per_gpu": 4240.57,
      "total_tokens": 33344569
    },
    {
      "epoch": 1.312828207051763,
      "grad_norm": 0.19240470230579376,
      "learning_rate": 0.00012387735970706312,
      "loss": 0.5033,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6542,
      "step": 1750,
      "tokens_per_second_per_gpu": 4267.65,
      "total_tokens": 33790426
    },
    {
      "epoch": 1.331582895723931,
      "grad_norm": 0.18220192193984985,
      "learning_rate": 0.00012191504130472937,
      "loss": 0.5103,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6658,
      "step": 1775,
      "tokens_per_second_per_gpu": 4237.08,
      "total_tokens": 34233908
    },
    {
      "epoch": 1.350337584396099,
      "grad_norm": 0.20157551765441895,
      "learning_rate": 0.00011994381717387514,
      "loss": 0.5192,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6807,
      "step": 1800,
      "tokens_per_second_per_gpu": 4244.09,
      "total_tokens": 34678691
    },
    {
      "epoch": 1.369092273068267,
      "grad_norm": 0.17189238965511322,
      "learning_rate": 0.00011796448837103129,
      "loss": 0.5011,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6505,
      "step": 1825,
      "tokens_per_second_per_gpu": 4277.26,
      "total_tokens": 35125624
    },
    {
      "epoch": 1.387846961740435,
      "grad_norm": 0.19443106651306152,
      "learning_rate": 0.00011597785924626616,
      "loss": 0.4994,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6477,
      "step": 1850,
      "tokens_per_second_per_gpu": 3766.52,
      "total_tokens": 35568850
    },
    {
      "epoch": 1.406601650412603,
      "grad_norm": 0.1810811311006546,
      "learning_rate": 0.00011398473711631764,
      "loss": 0.5083,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6625,
      "step": 1875,
      "tokens_per_second_per_gpu": 4204.76,
      "total_tokens": 36009980
    },
    {
      "epoch": 1.4253563390847712,
      "grad_norm": 0.19805970788002014,
      "learning_rate": 0.00011198593193651958,
      "loss": 0.5141,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6721,
      "step": 1900,
      "tokens_per_second_per_gpu": 4270.21,
      "total_tokens": 36457032
    },
    {
      "epoch": 1.4441110277569393,
      "grad_norm": 0.1936168372631073,
      "learning_rate": 0.00010998225597165628,
      "loss": 0.5045,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6562,
      "step": 1925,
      "tokens_per_second_per_gpu": 4275.24,
      "total_tokens": 36905590
    },
    {
      "epoch": 1.4628657164291072,
      "grad_norm": 0.19065748155117035,
      "learning_rate": 0.00010797452346587798,
      "loss": 0.5025,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6528,
      "step": 1950,
      "tokens_per_second_per_gpu": 4285.81,
      "total_tokens": 37354436
    },
    {
      "epoch": 1.4816204051012754,
      "grad_norm": 0.18647657334804535,
      "learning_rate": 0.0001059635503118125,
      "loss": 0.5102,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6656,
      "step": 1975,
      "tokens_per_second_per_gpu": 4259.76,
      "total_tokens": 37801500
    },
    {
      "epoch": 1.5003750937734432,
      "grad_norm": 0.21211788058280945,
      "learning_rate": 0.00010395015371900663,
      "loss": 0.5052,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 25.53,
      "memory/max_allocated (GiB)": 25.53,
      "ppl": 1.6573,
      "step": 2000,
      "tokens_per_second_per_gpu": 4250.7,
      "total_tokens": 38244936
    },
    {
      "epoch": 1.5003750937734432,
      "eval_loss": 0.5063687562942505,
      "eval_ppl": 1.6593,
      "eval_runtime": 141.112,
      "eval_samples_per_second": 5.173,
      "eval_steps_per_second": 1.035,
      "memory/device_reserved (GiB)": 139.06,
      "memory/max_active (GiB)": 19.1,
      "memory/max_allocated (GiB)": 19.1,
      "step": 2000
    }
  ],
  "logging_steps": 25,
  "max_steps": 3996,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.62796004179968e+18,
  "train_batch_size": 5,
  "trial_name": null,
  "trial_params": null
}