trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 84,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 1.3622558116912842,
      "learning_rate": 0.00019825664732332884,
      "loss": 1.4496,
      "num_input_tokens_seen": 15664,
      "step": 5
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 0.8986290097236633,
      "learning_rate": 0.00019308737486442045,
      "loss": 1.051,
      "num_input_tokens_seen": 30528,
      "step": 10
    },
    {
      "epoch": 0.5357142857142857,
      "grad_norm": 1.14683997631073,
      "learning_rate": 0.00018467241992282843,
      "loss": 0.9827,
      "num_input_tokens_seen": 43392,
      "step": 15
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.8037276864051819,
      "learning_rate": 0.00017330518718298264,
      "loss": 1.0249,
      "num_input_tokens_seen": 60784,
      "step": 20
    },
    {
      "epoch": 0.8928571428571429,
      "grad_norm": 0.8258629441261292,
      "learning_rate": 0.00015938201855735014,
      "loss": 0.9651,
      "num_input_tokens_seen": 76080,
      "step": 25
    },
    {
      "epoch": 1.0714285714285714,
      "grad_norm": 0.9029279947280884,
      "learning_rate": 0.00014338837391175582,
      "loss": 0.9298,
      "num_input_tokens_seen": 90224,
      "step": 30
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8744432926177979,
      "learning_rate": 0.00012588190451025207,
      "loss": 0.7694,
      "num_input_tokens_seen": 107072,
      "step": 35
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 1.3319528102874756,
      "learning_rate": 0.00010747300935864243,
      "loss": 0.6823,
      "num_input_tokens_seen": 123216,
      "step": 40
    },
    {
      "epoch": 1.6071428571428572,
      "grad_norm": 1.1126036643981934,
      "learning_rate": 8.880355238966923e-05,
      "loss": 0.5681,
      "num_input_tokens_seen": 138416,
      "step": 45
    },
    {
      "epoch": 1.7857142857142856,
      "grad_norm": 1.8243435621261597,
      "learning_rate": 7.052448255890957e-05,
      "loss": 0.6239,
      "num_input_tokens_seen": 152784,
      "step": 50
    },
    {
      "epoch": 1.9642857142857144,
      "grad_norm": 1.5611317157745361,
      "learning_rate": 5.32731371726938e-05,
      "loss": 0.6441,
      "num_input_tokens_seen": 167264,
      "step": 55
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 1.1920043230056763,
      "learning_rate": 3.7651019814126654e-05,
      "loss": 0.5202,
      "num_input_tokens_seen": 181488,
      "step": 60
    },
    {
      "epoch": 2.3214285714285716,
      "grad_norm": 1.1858474016189575,
      "learning_rate": 2.420282768545469e-05,
      "loss": 0.443,
      "num_input_tokens_seen": 198512,
      "step": 65
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2865468263626099,
      "learning_rate": 1.339745962155613e-05,
      "loss": 0.4209,
      "num_input_tokens_seen": 213040,
      "step": 70
    },
    {
      "epoch": 2.678571428571429,
      "grad_norm": 1.2891031503677368,
      "learning_rate": 5.611666969163243e-06,
      "loss": 0.4818,
      "num_input_tokens_seen": 229312,
      "step": 75
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 1.2530392408370972,
      "learning_rate": 1.1169173774871478e-06,
      "loss": 0.5292,
      "num_input_tokens_seen": 244944,
      "step": 80
    }
  ],
  "logging_steps": 5,
  "max_steps": 84,
  "num_input_tokens_seen": 257136,
  "num_train_epochs": 3,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0939806209654784e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}