llama.cpp
9731be49
- fix sparse FA fp16 underflow bug
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
11 days ago
fix sparse FA fp16 underflow bug
References
0cc4m/vulkan-flash-attn-rdna-opt
#31323 - vulkan: optimize prompt processing Flash Attention for RDNA3 and RDNA4
Author
0cc4m
Parents
d52056be
Loading