llama.cpp
a22a2358
- Avoid fusing FP8 for --fuse-gate-up-exps as we may have different ws
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
17 days ago
Avoid fusing FP8 for --fuse-gate-up-exps as we may have different ws
References
#27512 - Quant: OCP FP8 E4M3 support
Author
ORippler
Committer
ORippler
Parents
5a5d5cd0
Loading