Izacard et al.: contrastively trained unsupervised dense retrieval

What changed

Trained dense text retrieval through contrastive learning without labeled query-document pairs, then tested cross-domain retrieval and optional supervised adaptation.

Scientific scope

Trained dense text retrieval through contrastive learning without labeled query-document pairs, then tested cross-domain retrieval and optional supervised adaptation. Contrastive learning and dense encoders predate Contriever. Unsupervised means no relevance labels in that stage, not absence of a text corpus or all downstream supervision. No universal advantage over sparse retrieval.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Contrastive learning and dense encoders predate Contriever. Unsupervised means no relevance labels in that stage, not absence of a text corpus or all downstream supervision. No universal advantage over sparse retrieval.

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.

• Towards Unsupervised Dense Information Retrieval with Contrastive Learning

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