Model techniques map
Techniquesmodel architecturetoken mixersparse attentionblock selection

specific method · filed under model architecture

MiniMax Sparse Attention

A blockwise sparse-attention architecture built on Grouped Query Attention for efficient long-context processing.

Also called MSA (MiniMax Sparse Attention), MiniMax Sparse Attention (MSA).

sources
6
model
1
lab adopt it
1
strongest
core

How sources treat it

One count per evidence span, weakest treatment to strongest.

core 6

Documented in

Further reading

Picked by hand, not extracted: where to read more, not evidence for anything on this page.

Evidence

6 spans quoted from the sources, strongest treatment first.

MSA is a clean and easily extensible new sparse attention architecture. It gives M3 a 1M context window

coremodel architecturein MiniMax-M3MiniMax

We introduce MiniMax Sparse Attention (MSA), a blockwise sparse attention built upon Grouped Query Attention (GQA).

coremodel architecturein MiniMax-M3MiniMax

M3 introduces MiniMax Sparse Attention (MSA) to improve long context efficiency

coremodel architecturein MiniMax-M3MiniMax

The technical foundation is a new attention variant called MiniMax Sparse Attention (MSA).

coremodel architecturein MiniMax-M3MiniMax

M3 is powered by MiniMax Sparse Attention (MSA)

coremodel architecturein MiniMax-M3MiniMax

Powered by the proprietary MiniMax Sparse Attention (MSA) architecture

coremodel architecturein MiniMax-M3MiniMax

Filed alongside

Other methods under model architecture :: token mixer :: sparse attention :: block selection.