Instructions to use kernels-community/flash-attn2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Kernels
How to use kernels-community/flash-attn2 with Kernels:
# !pip install kernels from kernels import get_kernel # a version (or an explicit revision) is required; see the "Files and versions" tab for the available ones kernel = get_kernel("kernels-community/flash-attn2", version=1) - Notebooks
- Google Colab
- Kaggle
Commit History
-
fix: bump build for torch 2.8
drbh90b3e94
-
feat: bump api on readme
drbh56449c1 -
feat: bump build
drbh09eec95 -
fix: add readme example script
drbhf2b2454 -
fix: bump readme
drbh0518250 -
fix: refactor bindings and update py library for correctness
drbh876ac68
-
fix: add warning for accuracy issue
Verified signature3b8334d
-
fix: adjust comma typos
drbh6c3907a -
fix: simplify and align signatures further
drbh3099d65 -
feat: bump builds
drbh809bfb7 -
fix: align kernel source with latest reference source
drbhc743a32
-
feat: add tag for hfjob build
Verified signature4762963
-
Add license and tag metadata
a741640 -
Build (aarch64)
f849035 -
Build
b58ed97
-
Sync capabilities with upstream
dd2f0f9 -
Update flake
40d139f
-
feat improve readme and library code
drbhd6cc1b0 -
fix: update library for all ops
drbhb833fce -
Update build/torch26-cxx98-cu124-x86_64-linux/flash_attn/__init__.py
Verified signature79e0916 -
feat: bump build
David Holtzdd2a3b2 -
fix: revise bindings and wrapper typing
drbhab4cc6a
-
feat: build for all caps 8 and up
drbhd774688 -
feat: add remaining torch binding def and impls
drbheda872e -
fix: adjust types
drbh4080f9c
-
fix: expand build combinations and include all files
drbhb0d3c12 -
fix: limit build head size and cuda caps for dev
drbh9002ff5
-
feat: pass vars into fwd and include build
drbh39b4aba -
feat: include source and enable build
drbha7165c8 -
initial commit
Verified signaturebdaa00d