Use an LLM to generate reasonable code snippets in a hardware-aware manner for Diffusers. - View it on GitHub