Thursday 2nd May 2024
This practical aims to cover:
-
Cookiecutter Research. Creating high-quality Python research repositories with minimal effort.
-
Experiment configuration using Hydra. Simplifying the configuration of complex ML experiments.
-
Machine learning at scale using Ray - Google have very kindly granted access to compute from their TPU Research Cloud (TRC) to use for the afternoon. We’ll use this to do some experiments with the same framework used to train ChatGPT.
The 0_shell folder contains misc. quality-of-life suggestions for your machine set-up (shell, IDE etc.) to browse in your own time.
Clone this repository with:
git clone --recurse-submodules https://github.com/joncarter1/effective-ml-practical.git
The --recurse-submodules option is important to ensure that external repository dependencies are also cloned into the externals folder.
I'd then recommend going through Parts 1 to 3 in order, which have their own README files with instructions within their respective folders.