llama.cpp

server : enable KV cache defrag by default

#10233

Merged