{
  "best_global_step": 1234,
  "best_metric": 2.5843265056610107,
  "best_model_checkpoint": "models/gemma270m-sft-fqa/checkpoint-1234",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1851,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08115236356258876,
      "grad_norm": 13.660741806030273,
      "learning_rate": 8.75e-05,
      "loss": 4.983,
      "step": 50
    },
    {
      "epoch": 0.16230472712517752,
      "grad_norm": 11.402396202087402,
      "learning_rate": 9.985847186947076e-05,
      "loss": 4.0619,
      "step": 100
    },
    {
      "epoch": 0.24345709068776628,
      "grad_norm": 10.74139404296875,
      "learning_rate": 9.93391274588681e-05,
      "loss": 3.7985,
      "step": 150
    },
    {
      "epoch": 0.32460945425035503,
      "grad_norm": 11.64829158782959,
      "learning_rate": 9.844218899490987e-05,
      "loss": 3.6412,
      "step": 200
    },
    {
      "epoch": 0.4057618178129438,
      "grad_norm": 10.44972038269043,
      "learning_rate": 9.717452077876346e-05,
      "loss": 3.4894,
      "step": 250
    },
    {
      "epoch": 0.48691418137553255,
      "grad_norm": 10.880023002624512,
      "learning_rate": 9.554582431924548e-05,
      "loss": 3.3765,
      "step": 300
    },
    {
      "epoch": 0.5680665449381214,
      "grad_norm": 10.3041353225708,
      "learning_rate": 9.356856408683826e-05,
      "loss": 3.3129,
      "step": 350
    },
    {
      "epoch": 0.6492189085007101,
      "grad_norm": 9.135042190551758,
      "learning_rate": 9.125787212266508e-05,
      "loss": 3.1973,
      "step": 400
    },
    {
      "epoch": 0.7303712720632989,
      "grad_norm": 10.27686595916748,
      "learning_rate": 8.863143223245465e-05,
      "loss": 3.1488,
      "step": 450
    },
    {
      "epoch": 0.8115236356258876,
      "grad_norm": 9.036040306091309,
      "learning_rate": 8.570934465176278e-05,
      "loss": 3.0712,
      "step": 500
    },
    {
      "epoch": 0.8926759991884764,
      "grad_norm": 9.041040420532227,
      "learning_rate": 8.251397221817252e-05,
      "loss": 2.9367,
      "step": 550
    },
    {
      "epoch": 0.9738283627510651,
      "grad_norm": 8.38169002532959,
      "learning_rate": 7.906976922772278e-05,
      "loss": 2.8638,
      "step": 600
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.8626081943511963,
      "eval_runtime": 47.5672,
      "eval_samples_per_second": 184.224,
      "eval_steps_per_second": 5.76,
      "step": 617
    },
    {
      "epoch": 1.0535605599513085,
      "grad_norm": 10.622579574584961,
      "learning_rate": 7.54030942853327e-05,
      "loss": 2.1546,
      "step": 650
    },
    {
      "epoch": 1.1347129235138973,
      "grad_norm": 8.288007736206055,
      "learning_rate": 7.154200858148418e-05,
      "loss": 1.7795,
      "step": 700
    },
    {
      "epoch": 1.2158652870764861,
      "grad_norm": 8.66895580291748,
      "learning_rate": 6.751606113895808e-05,
      "loss": 1.7836,
      "step": 750
    },
    {
      "epoch": 1.297017650639075,
      "grad_norm": 8.41976261138916,
      "learning_rate": 6.335606267313821e-05,
      "loss": 1.7458,
      "step": 800
    },
    {
      "epoch": 1.3781700142016637,
      "grad_norm": 9.445528030395508,
      "learning_rate": 5.9093849796538034e-05,
      "loss": 1.763,
      "step": 850
    },
    {
      "epoch": 1.4593223777642523,
      "grad_norm": 9.660811424255371,
      "learning_rate": 5.476204137210122e-05,
      "loss": 1.7357,
      "step": 900
    },
    {
      "epoch": 1.5404747413268411,
      "grad_norm": 8.14544677734375,
      "learning_rate": 5.039378887991235e-05,
      "loss": 1.6956,
      "step": 950
    },
    {
      "epoch": 1.62162710488943,
      "grad_norm": 8.195582389831543,
      "learning_rate": 4.602252270777026e-05,
      "loss": 1.7108,
      "step": 1000
    },
    {
      "epoch": 1.7027794684520186,
      "grad_norm": 9.13266372680664,
      "learning_rate": 4.1681696307270946e-05,
      "loss": 1.662,
      "step": 1050
    },
    {
      "epoch": 1.7839318320146074,
      "grad_norm": 8.996973037719727,
      "learning_rate": 3.7404530173382555e-05,
      "loss": 1.6397,
      "step": 1100
    },
    {
      "epoch": 1.8650841955771962,
      "grad_norm": 8.205750465393066,
      "learning_rate": 3.322375760684524e-05,
      "loss": 1.6252,
      "step": 1150
    },
    {
      "epoch": 1.946236559139785,
      "grad_norm": 9.282258033752441,
      "learning_rate": 2.917137420508569e-05,
      "loss": 1.5914,
      "step": 1200
    },
    {
      "epoch": 2.0,
      "eval_loss": 2.5843265056610107,
      "eval_runtime": 46.6973,
      "eval_samples_per_second": 187.656,
      "eval_steps_per_second": 5.868,
      "step": 1234
    },
    {
      "epoch": 2.0259687563400286,
      "grad_norm": 9.058813095092773,
      "learning_rate": 2.5278392998800394e-05,
      "loss": 1.2762,
      "step": 1250
    },
    {
      "epoch": 2.107121119902617,
      "grad_norm": 9.824188232421875,
      "learning_rate": 2.1574607108156425e-05,
      "loss": 0.6675,
      "step": 1300
    },
    {
      "epoch": 2.188273483465206,
      "grad_norm": 8.858758926391602,
      "learning_rate": 1.808836173500954e-05,
      "loss": 0.6402,
      "step": 1350
    },
    {
      "epoch": 2.2694258470277946,
      "grad_norm": 9.899801254272461,
      "learning_rate": 1.484633723609154e-05,
      "loss": 0.6364,
      "step": 1400
    },
    {
      "epoch": 2.3505782105903834,
      "grad_norm": 8.896689414978027,
      "learning_rate": 1.1873344937315002e-05,
      "loss": 0.6221,
      "step": 1450
    },
    {
      "epoch": 2.4317305741529722,
      "grad_norm": 8.655142784118652,
      "learning_rate": 9.192137251835742e-06,
      "loss": 0.6147,
      "step": 1500
    },
    {
      "epoch": 2.512882937715561,
      "grad_norm": 10.144364356994629,
      "learning_rate": 6.823233555045849e-06,
      "loss": 0.6095,
      "step": 1550
    },
    {
      "epoch": 2.59403530127815,
      "grad_norm": 9.760300636291504,
      "learning_rate": 4.784763149082405e-06,
      "loss": 0.6137,
      "step": 1600
    },
    {
      "epoch": 2.675187664840738,
      "grad_norm": 9.731839179992676,
      "learning_rate": 3.0923265186494188e-06,
      "loss": 0.6282,
      "step": 1650
    },
    {
      "epoch": 2.7563400284033275,
      "grad_norm": 8.688471794128418,
      "learning_rate": 1.7588759399675525e-06,
      "loss": 0.5927,
      "step": 1700
    },
    {
      "epoch": 2.837492391965916,
      "grad_norm": 8.816551208496094,
      "learning_rate": 7.94616356554978e-07,
      "loss": 0.6152,
      "step": 1750
    },
    {
      "epoch": 2.9186447555285047,
      "grad_norm": 9.027681350708008,
      "learning_rate": 2.0692728044079957e-07,
      "loss": 0.5801,
      "step": 1800
    },
    {
      "epoch": 2.9997971190910935,
      "grad_norm": 9.38516616821289,
      "learning_rate": 3.06316503556614e-10,
      "loss": 0.5927,
      "step": 1850
    },
    {
      "epoch": 3.0,
      "eval_loss": 3.1965487003326416,
      "eval_runtime": 46.5876,
      "eval_samples_per_second": 188.097,
      "eval_steps_per_second": 5.881,
      "step": 1851
    }
  ],
  "logging_steps": 50,
  "max_steps": 1851,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0937701855920128e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}