llama.cpp

examples : add --kv-unified to batched example

#18774

Merged