llama.cpp releases
· Infrastructure
b10380: chat : fix muse-glimmer detection of tool calls after EOM (#26879)
chat : fix muse-glimmer swallowing a trailing tool call into content Muse Glimmer routinely answers the user and calls a tool in a single generation. The template terminates a message with when more messages follow in the same turn and only at the end of the turn, so the answer is closed by and