trainer_state.json

{
  "best_metric": 0.936830835117773,
  "best_model_checkpoint": "swin-tiny-patch4-window7-224-Kaggle_test_20231117/checkpoint-22",
  "epoch": 2.966292134831461,
  "eval_steps": 500,
  "global_step": 66,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.45,
      "learning_rate": 9.491525423728815e-05,
      "loss": 3.4547,
      "step": 10
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.796610169491526e-05,
      "loss": 0.2786,
      "step": 20
    },
    {
      "epoch": 0.99,
      "eval_accuracy": 0.936830835117773,
      "eval_loss": 0.21843616664409637,
      "eval_runtime": 399.4037,
      "eval_samples_per_second": 2.338,
      "eval_steps_per_second": 0.075,
      "step": 22
    },
    {
      "epoch": 1.35,
      "learning_rate": 6.101694915254238e-05,
      "loss": 0.1855,
      "step": 30
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.4067796610169495e-05,
      "loss": 0.1598,
      "step": 40
    },
    {
      "epoch": 1.98,
      "eval_accuracy": 0.9346895074946466,
      "eval_loss": 0.18257774412631989,
      "eval_runtime": 16.3859,
      "eval_samples_per_second": 57.0,
      "eval_steps_per_second": 1.831,
      "step": 44
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.711864406779661e-05,
      "loss": 0.1355,
      "step": 50
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.016949152542373e-05,
      "loss": 0.1352,
      "step": 60
    },
    {
      "epoch": 2.97,
      "eval_accuracy": 0.9336188436830836,
      "eval_loss": 0.2057112157344818,
      "eval_runtime": 16.5678,
      "eval_samples_per_second": 56.375,
      "eval_steps_per_second": 1.811,
      "step": 66
    },
    {
      "epoch": 2.97,
      "step": 66,
      "total_flos": 2.145105467415429e+17,
      "train_loss": 0.6672632369128141,
      "train_runtime": 1831.5188,
      "train_samples_per_second": 4.622,
      "train_steps_per_second": 0.036
    }
  ],
  "logging_steps": 10,
  "max_steps": 66,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2.145105467415429e+17,
  "trial_name": null,
  "trial_params": null
}