Can you provide steps to reproduce so I can see one of these deterministic LLMs running myself? API based or local models.