This repository contains the code and docs for the "Supercharge your Cloud Development Workflow: Introduction to Docker and GitHub Actions for ML apps" workshop.
To follow along with the workshop, head over to https://supercharge.training.hartree.dev and follow the walkthrough.
The recommended way to follow the tutorial is using GitHub Codespaces to ensure that the development environment is consistent across all attendees - you're welcome to work on your local machine instead, you just need to make sure you install all the necessary dependencies like Docker, Python 3.10 and Poetry.
If you're using GitHub Codespaces, you should already have all the necessary dependencies installed. If not, make sure you've got Python 3.10 and Poetry installed:
python --version
poetry --version
If not, follow the links above to install them.
Next, create a virtual environment and install the dependencies with Poetry:
python -m venv .venv
source ./.venv/bin/activate
Now you're ready to get coding to turn this template into a fully working GPT2-text-generation-as-a-service!
The model we're wrapping up as part of the workshop is the distilgpt2 model, short for "Distilled GPT-2". This is a distilled (i.e. smaller) version of the GPT-2 model which is itself a precursor to more recent closed-source models such as GPT-3 and GPT-3.5 (the latter being used for Chat-GPT).