[griomnib]

Yeah, that’s what I do: use LLM to help make training data for small models. It’s ao much more efficient, fast, and ergo, scalable.