Mistral: Light-weight library for mixture-of-experts (MoE) training

2 点作者 georgehill超过 1 年前

1 comment

verdverm超过 1 年前

This is a fork of <a href="https://github.com/stanford-futuredata/megablocks">https://github.com/stanford-futuredata/megablocks</a><p>should link to the original when possible, per HN posting guidelines