[CamperBob2]

What about the sampling parameters? You can't just run llama-server with no CLI arguments (other than a uselessly-small context size) and expect useful results.