onnxruntime
0ccc9b0f
- [webgpu] Apply template to flash attention (#25722)
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
140 days ago
[webgpu] Apply template to flash attention (#25722) ### Description This PR applies template to flash attention, and simplifies the `is_unidirectional` check in shader. ### Motivation and Context See above.
References
#25722 - [webgpu] Apply template to flash attention
Author
daijh
Parents
4f6ae14e
Loading