llama.cpp

output normalize embedding in '/v1/embeddings'

#5956

Merged