https://github.com/anne-andresen/multi-modal-cuda-c-gan

Raw C/cuda implementation of 3d GAN
https://github.com/anne-andresen/multi-modal-cuda-c-gan

3d 3d-models attention-mechanism c cross-attention cross-attention-c cuda gan gan-models low-level-programming medical-imaging multimodal-deep-learning pytorch transformer-pytorch transformers transformers-c

Last synced: 3 months ago
JSON representation

Raw C/cuda implementation of 3d GAN

• Host: GitHub
• URL: https://github.com/anne-andresen/multi-modal-cuda-c-gan
• Owner: Anne-Andresen
• Created: 2024-05-13T15:04:31.000Z (9 months ago)
• Default Branch: main
• Last Pushed: 2024-07-23T04:05:31.000Z (7 months ago)
• Last Synced: 2024-09-16T10:44:29.896Z (5 months ago)
• Topics: 3d, 3d-models, attention-mechanism, c, cross-attention, cross-attention-c, cuda, gan, gan-models, low-level-programming, medical-imaging, multimodal-deep-learning, pytorch, transformer-pytorch, transformers, transformers-c
• Language: Cuda
• Homepage:
• Size: 156 KB
• Stars: 1
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Multi-Modal-Cuda-C-GAN

## Overview

Welcome to the Multi-Modal-Cuda-C-GAN repository! This project on a 3D deep learning model implemented in CUDA/C, specifically a Hybrid GAN that integrates cross-attention, self-attention, and convolutional blocks within the generator. The model leverages C for high-performance and scalable deep learning solutions.

## Features

- **Self-Attention**: Enhances the generator's capability, implemented in C for flexibility.

- **Cross-Attention Mechanism**: Designed for 3D tensors, suitable for CNN layers. It merges separate input tensors, outputting the same size, facilitating multi-input images and new data introduction during processing. Available in C and C++.

- **Convolutional Blocks**: Core convolution operations for the GAN, implemented in C for efficiency.

- **GAN Structure**: Comprehensive GAN architecture featuring a UNet within the generator, implemented in C for robust performance.

## Current Development

We are actively developing the training script in C, recreating many dependencies typically found in PyTorch from scratch, to ensure optimal performance and customization.

## To-Do List

- Implement GAN training and code iteration in C.

- Update the README with detailed setup instructions and usage examples.

- Optimize nested for loops and arithmetic operations for memory efficiency.

## Getting Started

### Prerequisites

- **C Compiler**: GCC or equivalent

### Installation

1. Clone the repository:

    ```sh

    git clone https://github.com/your-username/Multi-Modal-Cuda-C-GAN.git

    cd Multi-Modal-Cuda-C-GAN

    ```

2. Compile the code:

    ```sh

    gcc -o main main.c -lcuda

    ```

## Contributing

We welcome contributions! Please check the [issues](https://github.com/Anne-Andresen/Multi-Modal-Cuda-C-GAN/issues) for tasks that need assistance or open a new issue to propose enhancements and report bugs.

## License

This project is licensed under the MIT License. See the [LICENSE](https://github.com/Anne-Andresen/Multi-Modal-Cuda-C-GAN/LICENSE) file for details.

## Contact

For any questions or feedback, please contact [[email protected]](mailto:[email protected]).