This repository contains materials for a SciPy 2023 workshop on modern deep learning with PyTorch. It targets researchers and practitioners looking to understand and implement advanced deep learning techniques, including multi-GPU training and fine-tuning large language models, using PyTorch. The workshop aims to provide a practical, hands-on introduction to PyTorch's capabilities and modern training paradigms.

How It Works

The workshop covers foundational deep learning concepts, PyTorch API fundamentals, and practical training strategies. It emphasizes multi-GPU training paradigms and fine-tuning large language models (transformers), showcasing how PyTorch facilitates these advanced workflows. The approach focuses on bridging the gap between traditional machine learning and current deep learning practices.

Quick Start & Requirements

• Installation: Download the repository. Follow the provided guides for Python setup and library installation (./00-1_python-setup-guide, ./00-2_python-libraries-for-workshop).
• Prerequisites: Python, PyTorch, and other specified libraries. GPU and CUDA are recommended for advanced training paradigms.
• Resources: No specific resource footprint is detailed, but deep learning tasks generally require significant computational resources.
• Links: Official Tutorial Schedule

Highlighted Details

• Introduction to deep learning strengths and use cases.
• Deep dive into the PyTorch API.
• Training deep neural networks with multi-GPU paradigms.
• Fine-tuning large language models (transformers).

Maintenance & Community

• The repository is associated with the SciPy 2023 conference.
• Questions can be directed via the GitHub Discussion page.

Licensing & Compatibility

• The repository content is likely subject to the terms of the SciPy conference and any specific licenses applied to the code and slides. No explicit license is mentioned in the README.

Limitations & Caveats

The materials were prepared for a specific workshop date (July 10, 2023) and may not reflect the latest PyTorch features or best practices. The README does not specify the exact license for the workshop materials, which could impact commercial use or redistribution.