Long-form factuality in large language models

What changed

Combines LongFact long-form prompts, search-augmented atomic-fact evaluation and F1-at-K using a preferred supported-fact count as a recall surrogate.

Scientific scope

Combines LongFact long-form prompts, search-augmented atomic-fact evaluation and F1-at-K using a preferred supported-fact count as a recall surrogate. The method builds on earlier atomic-claim decomposition. Search results and judge outputs remain fallible; the preferred-length parameter is not exhaustive factual recall. Annotator comparisons concern a sampled disagreement reannotation with different access to sources. Reported search/API costs are historical experiment costs.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. The method builds on earlier atomic-claim decomposition. Search results and judge outputs remain fallible; the preferred-length parameter is not exhaustive factual recall. Annotator comparisons concern a sampled disagreement reannotation with different access to sources. Reported search/API costs are historical experiment costs.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• Long-form factuality in large language models

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction