Skip to content
conftrace_

softmax-free transformer

1 paper

Explore in graph

Also known as

SOFT

Co-occurring keywords · papers mentioning both

self-attention approximation (1) moore-penrose inverse (1) linear complexity attention (1) linear complexity (1) gaussian kernel (1) vision transformer (1) matrix decomposition (1)

Papers

SOFT: Softmax-free Transformer with Linear Complexity NeurIPS 2021