Model Architecture & Training MethodsAinslie et al.
GQA
GQA
Why it matters
A compromise between MHA and MQA that allows uptraining of existing checkpoints, and became the industry-standard KV-saving technique
Source — Primary sources first; secondary reports are labelled “Press”
Paperarxiv.orghttps://arxiv.org/abs/2305.13245Added in Week 35, 2026 · See that week