BWhat changed
Fused blockwise exact-attention computation with feedforward computation on completed query blocks, reducing intermediate activation memory without materializing whole-sequence feedforward intermediates.
Scientific scope
Fused blockwise exact-attention computation with feedforward computation on completed query blocks, reducing intermediate activation memory without materializing whole-sequence feedforward intermediates. Inherits online softmax, FlashAttention and memory-efficient exact attention. The contribution extends blockwise fusion to feedforward activations, not invention of exact attention or subquadratic dense-attention compute. Tested context-length gains depend on model, checkpointing and accelerator memory.
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 online softmax, FlashAttention and memory-efficient exact attention. The contribution extends blockwise fusion to feedforward activations, not invention of exact attention or subquadratic dense-attention compute. Tested context-length gains depend on model, checkpointing and accelerator memory.
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.
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