Fix crashes during multi-turn conversation due to KV cache serialization#9

JohannesGaessler merged 1 commit intoJohannesGaessler:ggml-meta-backend-8JohannesGaessler/llama.cpp:ggml-meta-backend-8from

gaugarg-nv:fix_kv_serializationgaugarg-nv/llama.cpp:fix_kv_serializationCopy head branch name to clipboard