CI/CD Pipelines for Dockerfiles

This repository contains CI/CD pipelines to build and deploy Docker images for CUDA-enabled environments with PyTorch and LibTorch pre-installed.

Features

• Pre-built Docker images with LibTorch / PyTorch (C++11 ABI)
• CUDA 12.2.2-based environment
• CI/CD automation for reproducible builds

Docker Images

The provided Dockerfiles build images that include:

• PyTorch built from source with CUDA support
• LibTorch pre-installed for C++ inference
• Essential dependencies for deep learning workflows

CI/CD Pipeline

The pipeline automates:

1. Build - Compiles PyTorch and dependencies from source
2. Test - Runs basic validation checks
3. Publish - Pushes built images to a container registry

Usage

Clone the repo and modify the pipeline as needed:

# Build the Docker image locally
DOCKER_BUILDKIT=1 docker build -t my-cuda-pytorch .