⚡ PyTorch GPU Development Stack with Docker

Isolated, Reproducible NVIDIA GPU-Accelerated Docker Compose Development Environment for PyTorch & Machine Learning.

Overview • Features • Code Architecture • System Flow • Structure • Quick Start • License

---

📌 Overview

A containerized machine learning development template powered by Docker Compose and NVIDIA Container Toolkit. Provides a fully isolated PyTorch environment mapped directly to your local workspace, eliminating host dependency conflicts while providing native GPU acceleration for deep learning and LLM experimentation.

---

✨ Features (Key Outcomes & Capabilities)

Icon	Feature	Outcome & Real Proof
🚀	Native GPU Acceleration	Docker Compose v2 GPU pass-through (capabilities: [gpu]) for high-throughput CUDA compute
📁	Live Host Volume Mapping	Real-time sync between host directory and container /workspace
🛡️	Isolated & Reproducible	Eliminates Python environment conflicts and host system pollution
📓	Interactive Workspace	Ready for interactive CLI, Jupyter, and PyTorch training scripts

---

🔬 Code Architecture & Implementation

🔬 Code Implementation

• docker-compose.yml:
  • build: .: Builds from local Dockerfile.
  • volumes: - .:/workspace: Maps host directory directly into /workspace for real-time live synchronization.
  • deploy.resources.reservations.devices: Configures driver: nvidia, count: all, capabilities: [gpu] for full GPU hardware pass-through.
  • stdin_open: true, tty: true: Enables interactive terminal (docker compose exec) and REPL sessions.
• Python_llm.ipynb: Machine learning and PyTorch GPU verification notebook.

---

📊 System Flow

graph TD
  Host[🖥️ Host OS + NVIDIA Drivers] --> Docker[🐳 Docker Compose Engine]
  Docker --> GPU[⚡ NVIDIA Container Toolkit Pass-Through]
  GPU --> AppContainer[📦 PyTorch ML Container /workspace]
  AppContainer --> Notebook[📓 Python_llm.ipynb Execution]
  AppContainer --> Scripts[🐍 PyTorch Training Scripts]

  classDef primary fill:#10b981,stroke:#059669,stroke-width:2px,color:#fff;
  classDef accent fill:#6366f1,stroke:#4f46e5,stroke-width:2px,color:#fff;
  class Docker,GPU primary;
  class AppContainer,Notebook,Scripts accent;

Loading

---

📁 Project Structure

python_llm/
├── 📁 assets/                 # High-resolution SVG banners
│   └── 🎨 hero.svg
├── 📄 docker-compose.yml      # NVIDIA GPU container stack definition
├── 📄 Dockerfile              # PyTorch + CUDA base image
├── 📄 Python_llm.ipynb        # GPU verification & ML experimentation notebook
└── 📄 README.md               # Documentation

---

🚀 Quick Start

# 1. Build and start GPU container in background
docker compose up -d

# 2. Access interactive bash inside GPU container
docker compose exec app bash

# 3. Verify PyTorch GPU acceleration
python -c "import torch; print('CUDA Available:', torch.cuda.is_available(), '| Device:', torch.cuda.get_device_name(0))" 

---

Released under the MIT License. Crafted with precision by LoNebula