IntelLabs / IntelLabs/Xe-Forge
The optimized FlashAttention kernels
Open
- Dominant language
- Python
- Stars
- 21
- Forks
- 8
- PR merge metrics
- No merged PRs in 30d
Description
Hi, the README/paper report strong FlashAttention forward results on Intel Arc Pro B70, but I only found `examples/attention/1_FlashAttention_Fwd.py` and the YAML benchmark spec.
Are the optimized FlashAttention kernels used for the reported results checked into this repo? I’d like to use them for testing. Thanks!
Contributor guide
Assessment
This issue has not been assessed yet.