Comment by Avicebron

Do you have some sort of automatic test suite for what's generated by the LLM prior to release? Just to ensure what it returns won't break downstream?