llama.cpp
56e45a23
- metal : optimize softmax for C > 32
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
1 year ago
metal : optimize softmax for C > 32
References
#5021 - ggml : add Flash Attention
Author
ggerganov
Parents
41d136b6
Loading