Choromanski and colleagues: Performer positive random-feature attention

What changed

Used positive orthogonal random features to estimate softmax attention with a computation order linear in sequence length at fixed feature count.

Scientific scope

Used positive orthogonal random features to estimate softmax attention with a computation order linear in sequence length at fixed feature count. Kernel estimates, normalized attention and finite-feature error are distinct. No exact equality to dense attention or uniform hardware speedup is claimed.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Kernel estimates, normalized attention and finite-feature error are distinct. No exact equality to dense attention or uniform hardware speedup is claimed.

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.

• Rethinking Attention with Performers
• Choromanski and colleagues: Performer positive random-feature attention — 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