SWAG: adversarially filtered grounded event completion

What changed

Used video-caption contexts and language-model-generated alternatives, iteratively filtering easy negatives with stylistic classifiers and validating choices to build grounded next-event questions.

Scientific scope

Used video-caption contexts and language-model-generated alternatives, iteratively filtering easy negatives with stylistic classifiers and validating choices to build grounded next-event questions. Filtering is relative to the selected classifier family and candidate-generation process, not universal removal of bias. The contribution is the original SWAG resource and filtering procedure, distinct from the already catalogued HellaSwag successor.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Filtering is relative to the selected classifier family and candidate-generation process, not universal removal of bias. The contribution is the original SWAG resource and filtering procedure, distinct from the already catalogued HellaSwag successor.

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.

• SWAG: A Large-Scale Adversarial Dataset for Grounded Commonsense Inference

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