It will depend on the type of data you are using, and then what you are planning on doing with it. Various open source models will take various modalities (e.g.

It will depend on the type of data you are using, and then what you are planning on doing with it. Various open source models will take various modalities (e.g. images of models), and will create an embedding/vector for them (even if just as an internal representation). You can take these, and store them in a vector DB. This may help https://www.marqo.ai/blog/how-to-implement-text-to-image-search-on-marqo-in-5-lines-of-code https://www.marqo.ai/blog/how-to-implement-text-to-image-sea...