Early-Signal Detector

architecture · first seen 13 Oct 2026

grouped-query attention

Also: GQA

An attention variant where groups of query heads share a single key and value head to reduce KV cache size.

Mentions per week

0219 May11 Aug10 NovPapersCodeDiscussion

PapersCodeDiscussion

Earliest sources

Latest sources

Mentioned together with

KV cacheKV cache compressionsparse attentionmixture of expertsstructured pruningenergy efficiency of LLM inferencetime series foundation modellatent attention