This repository has been archived by the owner on Oct 28, 2022. It is now read-only.

Memory Efficient Self Attention

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