Open Models Almanac
Methodsmodel architecturetoken mixersoftmax attention

single mention · model architecture

Shared self-attention across modalities

Uses a shared self-attention mechanism to couple distinct vision-language and action transformer pathways.

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

Also called shared self-attention mechanism.

Filed under model architecture › token mixer › softmax attention