trainer_state.json

{
  "best_metric": 0.9886363636363636,
  "best_model_checkpoint": "swin-base-patch4-window7-224-in22k-finetuned-batch8-22k/checkpoint-120",
  "epoch": 4.848484848484849,
  "eval_steps": 500,
  "global_step": 120,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.40404040404040403,
      "grad_norm": 11.940139770507812,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.74,
      "step": 10
    },
    {
      "epoch": 0.8080808080808081,
      "grad_norm": 8.186209678649902,
      "learning_rate": 4.62962962962963e-05,
      "loss": 0.3752,
      "step": 20
    },
    {
      "epoch": 0.9696969696969697,
      "eval_accuracy": 0.9659090909090909,
      "eval_loss": 0.1119123175740242,
      "eval_runtime": 2.7264,
      "eval_samples_per_second": 64.555,
      "eval_steps_per_second": 4.035,
      "step": 24
    },
    {
      "epoch": 1.2121212121212122,
      "grad_norm": 13.843687057495117,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.1399,
      "step": 30
    },
    {
      "epoch": 1.6161616161616161,
      "grad_norm": 9.353983879089355,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.2189,
      "step": 40
    },
    {
      "epoch": 1.9797979797979797,
      "eval_accuracy": 0.9375,
      "eval_loss": 0.16898323595523834,
      "eval_runtime": 2.9538,
      "eval_samples_per_second": 59.584,
      "eval_steps_per_second": 3.724,
      "step": 49
    },
    {
      "epoch": 2.0202020202020203,
      "grad_norm": 9.988105773925781,
      "learning_rate": 3.240740740740741e-05,
      "loss": 0.0815,
      "step": 50
    },
    {
      "epoch": 2.4242424242424243,
      "grad_norm": 5.987796306610107,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.0405,
      "step": 60
    },
    {
      "epoch": 2.8282828282828283,
      "grad_norm": 2.5615344047546387,
      "learning_rate": 2.314814814814815e-05,
      "loss": 0.0343,
      "step": 70
    },
    {
      "epoch": 2.98989898989899,
      "eval_accuracy": 0.9829545454545454,
      "eval_loss": 0.08864129334688187,
      "eval_runtime": 2.7177,
      "eval_samples_per_second": 64.76,
      "eval_steps_per_second": 4.048,
      "step": 74
    },
    {
      "epoch": 3.2323232323232323,
      "grad_norm": 6.737287998199463,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.0161,
      "step": 80
    },
    {
      "epoch": 3.6363636363636362,
      "grad_norm": 8.469164848327637,
      "learning_rate": 1.388888888888889e-05,
      "loss": 0.009,
      "step": 90
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9829545454545454,
      "eval_loss": 0.09685815125703812,
      "eval_runtime": 2.7931,
      "eval_samples_per_second": 63.012,
      "eval_steps_per_second": 3.938,
      "step": 99
    },
    {
      "epoch": 4.040404040404041,
      "grad_norm": 3.5041699409484863,
      "learning_rate": 9.259259259259259e-06,
      "loss": 0.0054,
      "step": 100
    },
    {
      "epoch": 4.444444444444445,
      "grad_norm": 0.15403775870800018,
      "learning_rate": 4.6296296296296296e-06,
      "loss": 0.0027,
      "step": 110
    },
    {
      "epoch": 4.848484848484849,
      "grad_norm": 0.31535252928733826,
      "learning_rate": 0.0,
      "loss": 0.0015,
      "step": 120
    },
    {
      "epoch": 4.848484848484849,
      "eval_accuracy": 0.9886363636363636,
      "eval_loss": 0.0984315350651741,
      "eval_runtime": 2.7608,
      "eval_samples_per_second": 63.749,
      "eval_steps_per_second": 3.984,
      "step": 120
    },
    {
      "epoch": 4.848484848484849,
      "step": 120,
      "total_flos": 5.991868094431887e+17,
      "train_loss": 0.13874833025814345,
      "train_runtime": 338.048,
      "train_samples_per_second": 23.31,
      "train_steps_per_second": 0.355
    }
  ],
  "logging_steps": 10,
  "max_steps": 120,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 5.991868094431887e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}