trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 9,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 17.483246106883794,
      "learning_rate": 5e-05,
      "loss": 1.1171,
      "step": 1
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 16.76230573230985,
      "learning_rate": 4.375e-05,
      "loss": 1.105,
      "step": 2
    },
    {
      "epoch": 1.0,
      "grad_norm": 20.19666361790089,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 1.3735,
      "step": 3
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 4.070878513965605,
      "learning_rate": 3.125e-05,
      "loss": 0.9698,
      "step": 4
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 3.8268942352667814,
      "learning_rate": 2.5e-05,
      "loss": 0.7918,
      "step": 5
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.6695479402984423,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.6996,
      "step": 6
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 1.737118961654641,
      "learning_rate": 1.25e-05,
      "loss": 0.5728,
      "step": 7
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 1.1676394567045352,
      "learning_rate": 6.25e-06,
      "loss": 0.5033,
      "step": 8
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9097109004604421,
      "learning_rate": 0.0,
      "loss": 0.4516,
      "step": 9
    },
    {
      "epoch": 3.0,
      "step": 9,
      "total_flos": 3560949186560.0,
      "train_loss": 0.84272125032213,
      "train_runtime": 467.2816,
      "train_samples_per_second": 0.308,
      "train_steps_per_second": 0.019
    }
  ],
  "logging_steps": 1,
  "max_steps": 9,
  "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": 3560949186560.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}