Convolutional Neural Network for MNIST Dataset

This repository contains the code for training a Convolutional Neural Network (CNN) that classifies handwritten digits from on the MNIST dataset.
The code is organized into separate files to promote modularity and reusability.

Files

• model.py: This file contains the neural network architecture.
• utils.py: This file contains utility functions for data loading, training and evaluation.
• S5.ipynb: This Jupyter notebook contains the code for training and testing the CNN.
• README.md: This file (provides instructions and overview).

Requirements

• Python 3.x
• PyTorch 1.x
• torchvision
• matplotlib
• tqdm

Usage

1. Clone this repository.
2. Install required libraries: pip install torch torchvision matplotlib tqdm
3. Open the Jupyter Notebook S5.ipynb and follow the instructions within the notebook to train and test the CNN, and see the summary.

Key Features

• Data augmentation for training data.
• Clear separation of model architecture and training logic.
• Visual tracking of training and validation accuracy and loss.
• Optional model summary using torchsummary.

License

This project is licensed under the MIT License.