3 ms·
Unified memory and optimizations in llama.cpp (which Ollama wraps).
by bugglebeetle 3y ago
Unified memory and optimizations in llama.cpp (which Ollama wraps).
- ithkuil 3y agoIs that using the GPU?
- bugglebeetle 3y agoIt can be variably configured. There are details in the repo, but llama.cpp makes use of Metal.