trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 60,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 1.9183009817804988,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 3.4904,
      "step": 1
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 1.4669764740494382,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 3.6113,
      "step": 5
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 2.3102441270892657,
      "learning_rate": 1.973044870579824e-05,
      "loss": 3.5363,
      "step": 10
    },
    {
      "epoch": 1.0,
      "eval_loss": 3.6004714965820312,
      "eval_runtime": 0.3916,
      "eval_samples_per_second": 45.968,
      "eval_steps_per_second": 7.661,
      "step": 12
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.2536552391599605,
      "learning_rate": 1.866025403784439e-05,
      "loss": 3.4578,
      "step": 15
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.8379504173528296,
      "learning_rate": 1.686241637868734e-05,
      "loss": 3.516,
      "step": 20
    },
    {
      "epoch": 2.0,
      "eval_loss": 3.560063362121582,
      "eval_runtime": 0.395,
      "eval_samples_per_second": 45.564,
      "eval_steps_per_second": 7.594,
      "step": 24
    },
    {
      "epoch": 2.0833333333333335,
      "grad_norm": 0.8826743196650887,
      "learning_rate": 1.4487991802004625e-05,
      "loss": 3.429,
      "step": 25
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9643551070955534,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 3.4933,
      "step": 30
    },
    {
      "epoch": 2.9166666666666665,
      "grad_norm": 0.800770231101502,
      "learning_rate": 8.839070858747697e-06,
      "loss": 3.4836,
      "step": 35
    },
    {
      "epoch": 3.0,
      "eval_loss": 3.534764289855957,
      "eval_runtime": 0.5846,
      "eval_samples_per_second": 30.789,
      "eval_steps_per_second": 5.131,
      "step": 36
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 0.8243232107653643,
      "learning_rate": 6.039202339608432e-06,
      "loss": 3.4287,
      "step": 40
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.0751974758407474,
      "learning_rate": 3.5721239031346067e-06,
      "loss": 3.5007,
      "step": 45
    },
    {
      "epoch": 4.0,
      "eval_loss": 3.523860454559326,
      "eval_runtime": 0.3897,
      "eval_samples_per_second": 46.188,
      "eval_steps_per_second": 7.698,
      "step": 48
    },
    {
      "epoch": 4.166666666666667,
      "grad_norm": 0.7780096021946287,
      "learning_rate": 1.6451218858706374e-06,
      "loss": 3.4041,
      "step": 50
    },
    {
      "epoch": 4.583333333333333,
      "grad_norm": 0.9901654284407495,
      "learning_rate": 4.2010487684511105e-07,
      "loss": 3.4807,
      "step": 55
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.3948344306556273,
      "learning_rate": 0.0,
      "loss": 3.4187,
      "step": 60
    },
    {
      "epoch": 5.0,
      "eval_loss": 3.5221145153045654,
      "eval_runtime": 0.3973,
      "eval_samples_per_second": 45.3,
      "eval_steps_per_second": 7.55,
      "step": 60
    },
    {
      "epoch": 5.0,
      "step": 60,
      "total_flos": 18203285913600.0,
      "train_loss": 3.477997573216756,
      "train_runtime": 36.1944,
      "train_samples_per_second": 12.295,
      "train_steps_per_second": 1.658
    }
  ],
  "logging_steps": 5,
  "max_steps": 60,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "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": 18203285913600.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}