Oh I’m sorry to bring such a presentist idea to such a cool project, but if we ever get LLMs inferencing cheaply on consumer hardware and capable of efficient continuous learning, this insane format might be the perfect way to share your unique tamagotchi of expertise in a specific area