llama.cpp
021df05a
- Avoid races in fp8_fallback
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
5 days ago
Avoid races in fp8_fallback
References
osimons/ocp_fp8_cuda
#28413 - CUDA: Add OCP FP8 support
#28898 - ggml: support quant scales for fp8 and nvfp4
#29791 - vulkan: add fp8 and scaled matmul support
Author
ORippler
Parents
fd867ef5
Loading