llama.cpp
ggml-zendnn : adaptive fallback to CPU backend for small batch sizes
#22681
Merged

Loading