Open Models Almanac
Methodsmodel architecturemultimodal architecture

single mention · model architecture

Two-stage text-then-multimodal pre-training

First trains the language backbone on text-only data and subsequently adds multimodal training.

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

Also called two-stage pre-training strategy.

Filed under model architecture › multimodal architecture