Open Models Almanac
Methodsmodel architecturetoken mixersoftmax attentiongrouped-query attention

single mention · model architecture

Sparse use of grouped-query attention blocks for global context

An architectural arrangement using a small number of GQA blocks to provide global context; the evidence does not specify a distinct GQA mechanism.

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

Filed under model architecture › token mixer › softmax attention › grouped-query attention