Runtime fallback: ensure /models/model.gguf exists, then exec llama-server 63adad0 Javedalam commited on Sep 3, 2025