🤗 GitHub Repo
Memory optimization and training recipes to extrapolate language models' context length to 1 million tokens, with minimal hardware.
