Skip to content

GorkemP/pytorch_classification

Repository files navigation

PyTorch Classification

Minimal and clean training and evaluation codes for baseline performances.

Code Features

  • Learning Rate Scheduling is implemented with torch.optim.lr_Scheduler
  • Tensorboard visualization is added
  • Early stopping is implemented
  • Best parameters for validation accuracy is saved
  • Confusion matrix for validation set is generated
  • Optuna hyper parameter search framework is used to find best parameters

Confusion Matrix

Training and Validation Sets Accuracies for ResNet18

accuracies

About

Minimal and clean CIFAR10 and ImageNet classification

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages