llama.cpp
5b263dd8
- cuda : unroll Q*K^T loop
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
1 year ago
cuda : unroll Q*K^T loop
References
#5021 - ggml : add Flash Attention
Author
ggerganov
Parents
3b1c4e76
Loading
