Model
ColPali
A multimodal embedding model designed for document retrieval that directly inputs screenshot images of document pages into a vision-language model. It generates multiple local vectors at the token level for each image patch, bypassing OCR and layout parsing. This approach simplifies the indexing pipeline at the cost of larger index storage requirements.
Timeline (last 90 days)
Oct 8
ColPali is a representative visual retrieval solution that renders document pages as images and uses a vision-language model to extract patch-level multi-vector representations, bypassing the OCR step.
Unverified50%