Results for "sparse-attn"

2 skills found

flex-block-attn: an efficient block sparse attention computation library

Pytorch Implementation of the sparse attention from the paper: "Generating Long Sequences with Sparse Transformers"