Efficient ResNet for CIFAR-10 Classification

This project implements a custom ResNet model optimized to stay under 5 million parameters while maintaining high classification accuracy on the CIFAR-10 dataset. It was submitted as part of a Kaggle competition.

Key Features

• Lightweight ResNet architecture
• Parameter count constrained below 5M
• Achieved strong test accuracy on CIFAR-10

Dataset

• CIFAR-10 (10-class image classification)

Requirements

• Python
• PyTorch
• Torchvision

How to Run

pip install torch torchvision
jupyter notebook ResNet_CIFAR10.ipynb