https://github.com/shub-garg/vision-transformer-vit-for-mnist

This repository implements a Vision Transformer (ViT) to classify handwritten digits from the MNIST dataset. The project includes model definition, training scripts, and visualization of results, including correct/incorrect predictions and a confusion matrix.
https://github.com/shub-garg/vision-transformer-vit-for-mnist

artificial-intelligence deeplearning mnist-classification project transformers vision-transformer

Last synced: about 7 hours ago
JSON representation

This repository implements a Vision Transformer (ViT) to classify handwritten digits from the MNIST dataset. The project includes model definition, training scripts, and visualization of results, including correct/incorrect predictions and a confusion matrix.

• Host: GitHub
• URL: https://github.com/shub-garg/vision-transformer-vit-for-mnist
• Owner: shub-garg
• License: mit
• Created: 2024-05-29T02:20:17.000Z (8 months ago)
• Default Branch: main
• Last Pushed: 2024-05-29T02:24:36.000Z (8 months ago)
• Last Synced: 2024-05-29T15:58:07.973Z (8 months ago)
• Topics: artificial-intelligence, deeplearning, mnist-classification, project, transformers, vision-transformer
• Language: Jupyter Notebook
• Homepage:
• Size: 188 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Vision Transformer for MNIST

This repository contains an implementation of a Vision Transformer (ViT) for the MNIST dataset. The project demonstrates the ability to classify handwritten digits using a Transformer-based architecture.

## Features

- **Dataset**: Utilizes the MNIST dataset with 70,000 grayscale images of handwritten digits (60,000 for training and 10,000 for testing).

- **Model Architecture**: Implements a Vision Transformer with customizable parameters such as image size, patch size, number of classes, and more.

- **Training**: Includes training scripts with detailed loss tracking and evaluation metrics.

- **Visualization**: Provides visualizations of correct and incorrect predictions, as well as a confusion matrix for performance analysis.

## Setup and Usage

### Clone the Repository

```bash

git clone https://github.com/yourusername/ViT_MNIST.git

cd ViT_MNIST

```

### Launch Jupyter Notebook

```bash

jupyter notebook Vision_Transformer_for_MNIST.ipynb

```

### Run All Cells

Execute all cells in the notebook to preprocess data, define models, and start the training process.

## Example Plots

### Correct and Incorrect Predictions:



  



### Confusion Matrix: