llama.cpp
3cb1c348
- metal : try to improve batched decoding
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
1 year ago
metal : try to improve batched decoding
References
gg/pad-kv-cache
#4280 - llama : pad KV cache size
Author
ggerganov
Committer
ggerganov
Parents
3e68df86
Loading