flash-attn-4

Flash Attention CUTE (CUDA Template Engine) implementation

Installation

In a virtualenv (see these instructions if you need to create one):

pip3 install flash-attn-4

Dependencies

  • None

Releases

Version Released Bullseye
Python 3.9
Bookworm
Python 3.11
Trixie
Python 3.13
Files
4.0.0b16 pre-release 2026-06-03      
4.0.0b15 pre-release 2026-05-27      
4.0.0b14 pre-release 2026-05-20      
4.0.0b13 pre-release 2026-05-13      
4.0.0b12 pre-release 2026-05-06      
4.0.0b11 pre-release 2026-04-29      
4.0.0b10 pre-release 2026-04-22      
4.0.0b9 pre-release 2026-04-15      
4.0.0b8 pre-release 2026-04-08      
4.0.0b7 pre-release 2026-04-01      
4.0.0b5 pre-release 2026-03-23      
4.0.0b4 pre-release 2026-03-05      
4.0.0b3 pre-release 2026-03-05      
0.0.1 yanked 2026-02-09      

Issues with this package?

Page last updated 2026-06-03 11:13:08 UTC