Comment by EnPissant

5 months ago

llama.cpp has support for running some of or all of the layers on the CPU. It does not swap them into the GPU as needed.

0 comments