Yang et al.: hardware-efficient gated linear attention

What changed

Derived a chunk-parallel gated linear-attention form that permits data-dependent matrix-memory forgetting while using ordinary tensor-core operations for training.

Scientific scope

Derived a chunk-parallel gated linear-attention form that permits data-dependent matrix-memory forgetting while using ordinary tensor-core operations for training. Original December2023 version controls method and claims; later faster implementation metrics are excluded. Linear attention and gating are inherited; no unconditional equality with dense softmax attention or accuracy guarantee is implied.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Original December2023 version controls method and claims; later faster implementation metrics are excluded. Linear attention and gating are inherited; no unconditional equality with dense softmax attention or accuracy guarantee is implied.

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.

• Gated Linear Attention Transformers with Hardware-Efficient Training
• Yang et al.: hardware-efficient gated linear attention — official method source

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