llama.cpp
perplexity : faster HellaSwag via batching
#5017
Merged

Loading