Today I introduce Galois (/ɡælˈwɑː/) which is an open-source auto code completer based on OpenAI GPT-2 .
Galois is trained (finetuned) on a curated list of approximately 45K Python (~470MB) files gathered from the Github. Currently, it just works properly on Python but not bad at other languages (thanks to GPT-2's power).
With this project, I aim to create a Deep Learning Based Auto-completer such that anyone can run it on their own computer easily. I originally inspired by TabNine but wanted to make Galois can be run on endusers computer, freely (as in freedom). Thus anyone (especially developers) can benefit from state-of-the-art NLP technologies freely without sharing the codes (or texts) with a server. To fulfill this purpose we need to do the same what EfficientNet did to achieve state-of-the-art results in many datasets. For now, Galois trained on GPT-2 345M without developing such an efficient deep learning model. Check the planned works here for the future of Galois.
You can try it on usegalois.com. It is licensed under MIT, so you can find the trained model, codes and some instructions on github.com/iedmrc/galois-autocompleter for Galois, and on github.com/iedmrc/galois-web for the online editor. Check the link here for the
docker-compose file .
Any contributions are welcome. The technology behind the Galois has been developed by the community for many years. So clearly we can say that it belongs to the community.
To democratize the AI, let's spread the word to the world! Have it, use it, contribute to it, share it!
With respect to Évariste Galois ([evaʁist ɡalwa]).