Siméoni et al.: DINOv3 Gram-anchored dense visual features

What changed

Scaled self-supervised image representation learning and introduced Gram anchoring to stabilize dense features during long training, with adaptable model sizes and resolution strategies.

Scientific scope

Scaled self-supervised image representation learning and introduced Gram anchoring to stabilize dense features during long training, with adaptable model sizes and resolution strategies. One representation-training suite, not separate achievements for every downstream application or model size. Frozen-backbone results can still require trained task heads; no clinical, biological or ecological result is included.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. One representation-training suite, not separate achievements for every downstream application or model size. Frozen-backbone results can still require trained task heads; no clinical, biological or ecological result is included.

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.

• DINOv3
• Siméoni et al.: DINOv3 Gram-anchored dense visual features — official research 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