llama.cpp

017f10b5 - fix: crash when calling `llama_state_get_size` on a context without a KV cache (#13542)