trainer_state.json

{
  "best_metric": 0.625,
  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-42",
  "epoch": 2.9473684210526314,
  "eval_steps": 500,
  "global_step": 42,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.7017543859649122,
      "grad_norm": 5.9790940284729,
      "learning_rate": 4.324324324324325e-05,
      "loss": 0.6404,
      "step": 10
    },
    {
      "epoch": 0.9824561403508771,
      "eval_accuracy": 0.62,
      "eval_loss": 0.7281712293624878,
      "eval_runtime": 9.511,
      "eval_samples_per_second": 21.028,
      "eval_steps_per_second": 0.736,
      "step": 14
    },
    {
      "epoch": 1.4035087719298245,
      "grad_norm": 9.003488540649414,
      "learning_rate": 2.9729729729729733e-05,
      "loss": 0.4758,
      "step": 20
    },
    {
      "epoch": 1.9649122807017543,
      "eval_accuracy": 0.57,
      "eval_loss": 0.9752311706542969,
      "eval_runtime": 9.4132,
      "eval_samples_per_second": 21.247,
      "eval_steps_per_second": 0.744,
      "step": 28
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 24.35523223876953,
      "learning_rate": 1.6216216216216218e-05,
      "loss": 0.421,
      "step": 30
    },
    {
      "epoch": 2.807017543859649,
      "grad_norm": 7.268034934997559,
      "learning_rate": 2.702702702702703e-06,
      "loss": 0.3149,
      "step": 40
    },
    {
      "epoch": 2.9473684210526314,
      "eval_accuracy": 0.625,
      "eval_loss": 0.8868895173072815,
      "eval_runtime": 8.4487,
      "eval_samples_per_second": 23.672,
      "eval_steps_per_second": 0.829,
      "step": 42
    },
    {
      "epoch": 2.9473684210526314,
      "step": 42,
      "total_flos": 1.3243270931664077e+17,
      "train_loss": 0.455830332778749,
      "train_runtime": 333.8461,
      "train_samples_per_second": 16.175,
      "train_steps_per_second": 0.126
    }
  ],
  "logging_steps": 10,
  "max_steps": 42,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.3243270931664077e+17,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}