llama.cpp

Write a readme on Multi-GPU usage in llama.cpp

#22729

Merged