Lab 5: Troubleshooting & Testing
In this lab, we'll check out the basic tools required to write clean Python code and see how to write memorization tests for training code in PyTorch Lightning. Then we'll take a deep dive into the trace of a PyTorch training step and use it to debug performance issues in GPU-accelerated code.
Outline
- 00:00 Overview
- 00:51 Linting: pre-commit, black, flake8
- 05:42 Testing: pytest, doctest, memorization testing
- 11:15 Troubleshooting PyTorch performance
- 16:13 A guided tour of a PyTorch trace