🧪 Pytorch Lightning UV Template

A modern Deep Learning experiment template with PyTorch Lightning

Click on Use this template to initialize new repository.

✨ Features

• 🚀 UV Environment Management - Fast and efficient dependency management
• 🎯 Typer CLI - Modern command line interface
• ⚙️ YAML Config - Flexible experiment configuration
• 🔋 Lightning Components
  • DataModule for clean data handling
  • Model with built-in training logic
  • Trainer with all the bells and whistles
• 📊 Weights & Biases Integration
  • Experiment tracking and visualization
  • Hyperparameter optimization with sweeps
  • Dataset analysis and exploration
• 🎨 Clean Project Structure
  • Modular and maintainable codebase
  • Easy to extend and customize

🛠️ Installation

# install dependencies
uv sync --dev
# install project as a package
uv pip install -e .

Note

1. remember to replace your_wandb_entity with your actual W&B entity in the config files and config.py
2. uv run before bash scripts to ensure the environment is activated

📊 Dataset Analysis

Explore and analyze your dataset with built-in EDA tools:

python -m expt.main -c config/resnet.yml --eda

📈 View EDA Example

🚀 Training

Start training your model with a single command:

python -m expt.main -c config/resnet.yml --train

🔍 View Training Details

Configuration Overview

Training Progress

Training Summary

W&B Dashboard

📊 Evaluation

Evaluate your trained model:

python -m expt.main -c config/resnet.yml --eval --run-id n8fjnlyi

📊 View Evaluation Results

🎛️ Hyperparameter Tuning

Optimize your model with W&B Sweeps:

python -m expt.main -c config/resnet.yml --sweep --sweep-config config/sweep/mlp.yml

📈 View Sweep Results

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgements

• PyTorch Lightning
• Weights & Biases
• Typer
• UV

---

Made with ❤️ for the ML community