/Memory-Efficient-Self-Attention

Unofficial PyTorch implementation of "Self-Attention does not Need O(n^2) Memory".

Primary LanguageDockerfileMIT LicenseMIT

Stargazers