Model techniques map
Techniquesmodel architecturetoken mixersparse attentionsparse attention indexer

implementation detail · filed under model architecture

Index Branch value head

The value head associated with the Index Branch; the evidence notes it may be omitted for efficiency.

Also called index value head.

source
1
model
1
labs adopt it
0
strongest
not used

How sources treat it

One count per evidence span, weakest treatment to strongest.

not used 1

Documented in

Evidence

1 span quoted from the sources, strongest treatment first.

drops the index value head on efficiency grounds

not usedmodel architecturein MiniMax Sparse AttentionMiniMax

Filed alongside

Other methods under model architecture :: token mixer :: sparse attention :: sparse attention indexer.