Open Models Almanac
Methodsmodel architecturetoken mixersparse attentionsparse attention indexer

single mention · model architecture

Multi-head latent attention with a DSA sparse indexer

Combines multi-head latent attention with a DSA indexer that selects a sparse set of keys.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See sparse attention indexer for the methods filed beside it.

Filed under model architecture › token mixer › sparse attention › sparse attention indexer