run-2/checkpoint-32/trainer_state.json

{
  "best_metric": 0.7326732673267327,
  "best_model_checkpoint": "distilhubert-finetuned-not-a-word2/run-2/checkpoint-16",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 32,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.31,
      "grad_norm": 0.8302626013755798,
      "learning_rate": 9.517539799539297e-06,
      "loss": 0.6966,
      "step": 5
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6829280257225037,
      "learning_rate": 1.9035079599078593e-05,
      "loss": 0.6813,
      "step": 10
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5496665835380554,
      "learning_rate": 2.8552619398617893e-05,
      "loss": 0.6451,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_f1": 0.7326732673267327,
      "eval_loss": 0.6860618591308594,
      "eval_runtime": 1.3401,
      "eval_samples_per_second": 47.759,
      "eval_steps_per_second": 5.97,
      "step": 16
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3569624423980713,
      "learning_rate": 2.9610123820788925e-05,
      "loss": 0.6113,
      "step": 20
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.3838456869125366,
      "learning_rate": 2.8552619398617893e-05,
      "loss": 0.5997,
      "step": 25
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.46842828392982483,
      "learning_rate": 2.7495114976446858e-05,
      "loss": 0.6133,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_f1": 0.7326732673267327,
      "eval_loss": 0.719573974609375,
      "eval_runtime": 1.3426,
      "eval_samples_per_second": 47.668,
      "eval_steps_per_second": 5.958,
      "step": 32
    }
  ],
  "logging_steps": 5,
  "max_steps": 160,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 1691273188513872.0,
  "train_batch_size": 12,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 3.045612735852575e-05,
    "per_device_train_batch_size": 12
  }
}