Replying...
Intro. Collating known prompting techniques along with up-to-date novel techniques, I've written a prompt that optimized itself over several epochs. The LLM began to have a conversation instead of optimizing further. The result is this OPRO prompt. 📢📢📢IMPORTANT! THIS SEEMS TO FORCE THE MODEL INTO COMPLETION MODE AND NOT COVERSATIONAL! 📢📢📢 (e.g. Begin a sentence and let it finish it for you.)

Recursive OPRO

@mwarg