Reimers and Gurevych: Sentence-BERT reusable sentence embeddings

What changed

Fine-tuned shared BERT-based sentence encoders with siamese and triplet structures so embeddings can be compared without a cross-encoder pass per pair.

Scientific scope

Fine-tuned shared BERT-based sentence encoders with siamese and triplet structures so embeddings can be compared without a cross-encoder pass per pair. Pairwise similarity still has a search/comparison cost; encoder savings and benchmark accuracy do not prove universally equivalent cross-encoder rankings.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Pairwise similarity still has a search/comparison cost; encoder savings and benchmark accuracy do not prove universally equivalent cross-encoder rankings.

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.

• Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
• Reimers and Gurevych: Sentence-BERT reusable sentence embeddings — author or proceedings record

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