AWhat changed
Extended gated delta updates with finer-grained memory gating and a specialized chunkwise algorithm, integrating the module with latent attention in a hybrid language model.
Scientific scope
Extended gated delta updates with finer-grained memory gating and a specialized chunkwise algorithm, integrating the module with latent attention in a hybrid language model. Inherits Gated DeltaNet, delta-rule fast weights and MLA. Reported cache/performance gains are tested regimes, not universal replacement guarantees or exact dense attention. Individual appendix contributors are not automatically front-page authors.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Inherits Gated DeltaNet, delta-rule fast weights and MLA. Reported cache/performance gains are tested regimes, not universal replacement guarantees or exact dense attention. Individual appendix contributors are not automatically front-page authors.
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.
- Kimi Linear: An Expressive, Efficient Attention Architecture
- Kimi Team: Kimi Delta Attention and hardware-efficient linear mixing — 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