Open Models Almanac
Methodsmodel architecturetoken mixer

single mention · model architecture

Factorized spatial-temporal attention

Attention is separated into intra-frame spatial and inter-frame temporal passes, with temporal pooling compressing tokens over time.

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

Filed under model architecture › token mixer