CoCoNot

Allen AI CoCoNot scores whether a chat model withholds answers on 1,001 contextual noncompliance prompts, plus a 379-item contrast set for over-refusal.

Also known as: Coconot, Contextually, Comply Not, The Art of Saying No

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categorysafety
Subcategorycontextual noncompliance and over-refusal, beyond unsafe queries
Page statusactive
Metriccompliance_rate (percent; original: lower is better; contrast: higher is better)
Directionlower_is_better
Unit%
Dataset size1001
Dataset licenceODC-By (LICENSE.md); card also links Allen AI IMPACT-LR, which 404'd
PublisherAllen Institute for AI (with University of Washington, Microsoft Research, Samaya AI)

What it measures

CoCoNot (Contextually, Comply Not) tests when a chat model should not comply. The original test set is 1,001 English prompts across five taxonomy categories: incomplete, unsupported, indeterminate, safety, and humanizing requests. A judge labels the reply ACCEPTABLE (noncompliant) or UNACCEPTABLE (compliant). A 379-prompt contrast set asks the reverse: the model should answer, and the judge labels COMPLIANCE versus NONCOMPLIANCE. It is not jailbreak success and not [xstest](xstest.md), which only probes over-refusal on superficially unsafe wording.

Task format

Single-turn English prompt in, free-text reply out. inspect_evals defaults to subset=original, no system prompt, temperature 0, max_tokens 256, one epoch. Optional Tulu-style system prompt. Default judge openai/gpt-3.5-turbo.

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub