ggml-org/llama.cpp b7779

on GitHub

server : refactor oai_parser_opt, move it to server_chat_params (#18937)

• server_chat_params

• move chat format into CLI

• use meta whenever possible

• clean up, no more chatml fallback