Title: feat: add Docker support with Dockerfile and .dockerignore

Summary

This PR introduces Docker support for the Audiocraft project by adding a Dockerfile and a .dockerignore file. These additions facilitate containerization, enabling easier deployment and environment consistency across different systems.

Changes Made

1. Dockerfile:

   • Utilizes NVIDIA's CUDA base image with Python 3.9 and cuDNN 8 for GPU acceleration.
   • Installs necessary system dependencies including ffmpeg, build-essential, and git.
   • Sets up Python environment and installs project dependencies from requirements.txt.
   • Copies the project code into the Docker image.
   • Defines the default command to run the training script (train.py).

2. .dockerignore:

   • Excludes unnecessary files and directories to optimize the Docker build context.
   • Prevents inclusion of version control files, Python artifacts, build directories, virtual environments, logs, temporary files, and large data/model directories.

Rationale

• Consistency: Docker ensures that the application runs consistently across different environments by encapsulating all dependencies.
• Ease of Deployment: Facilitates deploying Audiocraft on various platforms without manual environment setup.
• Efficiency: The .dockerignore file optimizes build times and image sizes by excluding non-essential files.

How to Use

1. Build the Docker Image:

   docker build -t audiocraft:latest .

2. Run the Docker Container with GPU Support:

   docker run --gpus all \
     -v /path/to/your/data:/app/data \
     -v /path/to/your/models:/app/models \
     -v /path/to/your/output:/app/output \
     -e AUDIOCRAFT_CACHE_DIR=/app/cache \
     -it audiocraft:latest

Testing

• Verified GPU accessibility within the Docker container using PyTorch.
• Successfully ran the training script (train.py) without issues.

Future Improvements

• Multi-Stage Builds: To further optimize image size by separating build-time and runtime dependencies.
• Environment Variables: Enhance flexibility by parameterizing more configurations.

Kindly test the dockerfile, i await your response