Counter-Attractor Prompting

Counter-Attractor Prompting starts with a task’s known failure modes, then provides examples of better behavior and a structured self-audit. The hypothesis is that these examples guide the model more effectively than prohibitions alone. The planned comparison is against plain and prohibition-style prompts for the same task. A detailed write-up is in preparation; no public quantitative results are available yet.

In a businessTeams that use a language model for a repeated task need instructions aimed at its specific failures and a way to test revisions. This is a prompting protocol and a review design for that job. It is still a proposal without quantitative evidence.

Technical skills: Prompt design · Failure-mode analysis · Experimental design (proposed)

My contribution

I developed the prompting protocol around task failure modes. The project page describes the approach; a detailed experimental write-up is still in preparation.

Result and limit

No public quantitative result is presented yet. Treat this as work in progress.