My local LLM kept talking itself in circles until I changed two settings

Two settings changed everything about how my local model behaves