VERIFYAI infra
Jev Prompting Techniques
kevinlee@lbki34064963
This project explores 15 prompting techniques on Jev, an AI that does not write tokens, by implementing a chain-of-thought mechanism that asks sub-questions and feeds typed answers back as state.
Ran 15 prompting techniques on Jev by @typesafeai, which never writes a token. "Think step by step", CoVe: noise.
So I built chain-of-thought for it: ask sub-questions first, feed the typed answers back as state, ask again. Rubric memos 72 to 89.
github.com/leepokai/llm-prompt-techniques-