New
hikkian/shura.cpp — An inference engine that shares the GPU with the desktop
hikkian/shura.cpp is a llama.cpp fork that tunes the MoE expert cache to manage VRAM on GPUs shared between the desktop and games.
hikkian/shura.cpp
hikkian/shura.cpp is a llama.cpp fork that tunes the MoE expert cache to manage VRAM on GPUs shared between the desktop and games.
View on GitHub ↗- stars
- ★ 3
- Forks
- 0
- Language
- C++
- License
- MIT
Posts
1No matching stories.