Wow, this looks really interesting! I've been wanting to dig deeper into fine-tuning LLMs on Apple Silicon, so I'm excited to check this out. Can't wait to see what kind of performance boosts these techniques can provide.