Open Models Almanac
Methodsmodel architecturetoken mixersparse attention

single mention · model architecture

QSA micro-block compression at ratio 4

A QSA configuration using compressed QSA with a compression ratio of four.

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

Filed under model architecture › token mixer › sparse attention