{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9785407725321889,
  "eval_steps": 300,
  "global_step": 116,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08583690987124463,
      "grad_norm": 1.0348295145965056,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.63,
      "mean_token_accuracy": 0.8028752787393729,
      "step": 5
    },
    {
      "epoch": 0.17167381974248927,
      "grad_norm": 0.7946003237280415,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.6382,
      "mean_token_accuracy": 0.7991655463152572,
      "step": 10
    },
    {
      "epoch": 0.2575107296137339,
      "grad_norm": 0.7838232140655009,
      "learning_rate": 4.989741394042727e-05,
      "loss": 0.6105,
      "mean_token_accuracy": 0.8056573072191984,
      "step": 15
    },
    {
      "epoch": 0.34334763948497854,
      "grad_norm": 0.6462660097267386,
      "learning_rate": 4.92735454356513e-05,
      "loss": 0.606,
      "mean_token_accuracy": 0.8061553724053724,
      "step": 20
    },
    {
      "epoch": 0.4291845493562232,
      "grad_norm": 0.5037099303179456,
      "learning_rate": 4.8096988312782174e-05,
      "loss": 0.6115,
      "mean_token_accuracy": 0.8043439717150109,
      "step": 25
    },
    {
      "epoch": 0.5150214592274678,
      "grad_norm": 0.6195545306913001,
      "learning_rate": 4.639453180753619e-05,
      "loss": 0.5917,
      "mean_token_accuracy": 0.8100629578754578,
      "step": 30
    },
    {
      "epoch": 0.6008583690987125,
      "grad_norm": 0.5391133807181806,
      "learning_rate": 4.420493945100702e-05,
      "loss": 0.5984,
      "mean_token_accuracy": 0.8075732600732601,
      "step": 35
    },
    {
      "epoch": 0.6866952789699571,
      "grad_norm": 0.4989577694128677,
      "learning_rate": 4.157806645601988e-05,
      "loss": 0.5906,
      "mean_token_accuracy": 0.8098496642246642,
      "step": 40
    },
    {
      "epoch": 0.7725321888412017,
      "grad_norm": 0.44652843585336766,
      "learning_rate": 3.857372455503697e-05,
      "loss": 0.5855,
      "mean_token_accuracy": 0.8110664682539681,
      "step": 45
    },
    {
      "epoch": 0.8583690987124464,
      "grad_norm": 0.440623988108488,
      "learning_rate": 3.526032013631893e-05,
      "loss": 0.5825,
      "mean_token_accuracy": 0.8121346519902609,
      "step": 50
    },
    {
      "epoch": 0.944206008583691,
      "grad_norm": 0.4801584677439355,
      "learning_rate": 3.1713296686859426e-05,
      "loss": 0.5732,
      "mean_token_accuracy": 0.8148958333333332,
      "step": 55
    },
    {
      "epoch": 1.0171673819742488,
      "grad_norm": 0.8143816005601359,
      "learning_rate": 2.8013417006383076e-05,
      "loss": 0.54,
      "mean_token_accuracy": 0.8251169871573535,
      "step": 60
    },
    {
      "epoch": 1.1030042918454936,
      "grad_norm": 0.6905303409845385,
      "learning_rate": 2.4244924304977785e-05,
      "loss": 0.4399,
      "mean_token_accuracy": 0.8520162575203752,
      "step": 65
    },
    {
      "epoch": 1.1888412017167382,
      "grad_norm": 0.5001697685249674,
      "learning_rate": 2.0493624054652357e-05,
      "loss": 0.4337,
      "mean_token_accuracy": 0.8533382229436546,
      "step": 70
    },
    {
      "epoch": 1.2746781115879828,
      "grad_norm": 0.4316206882931221,
      "learning_rate": 1.6844930269478274e-05,
      "loss": 0.4279,
      "mean_token_accuracy": 0.8551797161172161,
      "step": 75
    },
    {
      "epoch": 1.3605150214592274,
      "grad_norm": 0.38409385416618463,
      "learning_rate": 1.3381920698905787e-05,
      "loss": 0.4156,
      "mean_token_accuracy": 0.8587591575091574,
      "step": 80
    },
    {
      "epoch": 1.4463519313304722,
      "grad_norm": 0.39576280367792716,
      "learning_rate": 1.0183445215899584e-05,
      "loss": 0.4236,
      "mean_token_accuracy": 0.8561346275221495,
      "step": 85
    },
    {
      "epoch": 1.5321888412017168,
      "grad_norm": 0.3840880238814308,
      "learning_rate": 7.3223304703363135e-06,
      "loss": 0.4158,
      "mean_token_accuracy": 0.859030630897102,
      "step": 90
    },
    {
      "epoch": 1.6180257510729614,
      "grad_norm": 0.33692841072311475,
      "learning_rate": 4.86372168622635e-06,
      "loss": 0.4231,
      "mean_token_accuracy": 0.8563792442426716,
      "step": 95
    },
    {
      "epoch": 1.703862660944206,
      "grad_norm": 0.32222211768399056,
      "learning_rate": 2.8635993586697553e-06,
      "loss": 0.4168,
      "mean_token_accuracy": 0.8585276149893059,
      "step": 100
    },
    {
      "epoch": 1.7896995708154506,
      "grad_norm": 0.3378812679021034,
      "learning_rate": 1.3675046241339918e-06,
      "loss": 0.402,
      "mean_token_accuracy": 0.8626781898656898,
      "step": 105
    },
    {
      "epoch": 1.8755364806866952,
      "grad_norm": 0.3155391223091413,
      "learning_rate": 4.095023263214121e-07,
      "loss": 0.4084,
      "mean_token_accuracy": 0.8610657051282053,
      "step": 110
    },
    {
      "epoch": 1.9613733905579398,
      "grad_norm": 0.30286260681832744,
      "learning_rate": 1.1405387761664887e-08,
      "loss": 0.4117,
      "mean_token_accuracy": 0.8597779304029303,
      "step": 115
    },
    {
      "epoch": 1.9785407725321889,
      "mean_token_accuracy": 0.8636809371184371,
      "step": 116,
      "total_flos": 120709586092032.0,
      "train_loss": 0.5114190966404718,
      "train_runtime": 2623.0811,
      "train_samples_per_second": 5.684,
      "train_steps_per_second": 0.044
    }
  ],
  "logging_steps": 5,
  "max_steps": 116,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 120709586092032.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}