Appgithub-com-thu-ml-spargeattn

Download Pinokio

Sign in Create accountJoin

Sign in Create accountJoin

SpargeAttn

github.com/thu-ml/spargeattnUpdated 8mo ago

[ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.

Loading community details…

Loading project owner…

Loading launchers…

Loading check-ins…

Community

Sort byCreate post

Ask or share about SpargeAttn…

Loading...

Loading references…

Loading project activity…

Pinokio

SpargeAttn on Pinokio

0

Projects using this repository

No references listed yet.

Check-ins (0)

Recent commits

reduce repo size

Jintao Zhang8 months agoae5b629

Add new article reference for SpargeAttention2

Jintao Zhang8 months ago6fb72d6

fix cuda launch error on devices < sm89

whx100310 months agobfd980b

Merge pull request #62 from t0saki/main

Jintao Zhang10 months agod097a13

Modify CogVideoX model path in inference example

jt-zhang10 months agoc01e14e

Please use `spas_sage2_attn_meansim_topk_cuda` and `block_sparse_sage2_attn_cuda` APIs

Jintao Zhang10 months ago9906f67

fix sm90 pad_size

whx100310 months ago01c4b48

Please use `spas_sage2_attn_meansim_topk_cuda` and `block_sparse_sage2_attn_cuda` APIs

Jintao Zhang10 months ago2e6e89e

Please use `spas_sage2_attn_meansim_topk_cuda` and `block_sparse_sage2_attn_cuda` APIs

Jintao Zhang10 months ago62532ec

Please use `spas_sage2_attn_meansim_topk_cuda` and `block_sparse_sage2_attn_cuda` API

Jintao Zhang10 months agobaf8739