llama.cpp

context : allow cache-less context for embeddings

#13108

Merged