Comment by pontusrehula

If you would setup an RLM, would you set a higher temperature for the root LLM calls and a lower temperature for LLM calls deeper in the recursion?