Tag
ColNanoVDR introduces the first document-free query distillation framework for multi-vector visual document retrieval, using optimal transport to align student and teacher query embeddings, enabling faster encoding while retaining high performance.
BELXTR is a novel multi-vector embedding model for biomedical entity linking that improves upon state-of-the-art methods by leveraging token-level matching, enhancing performance in disambiguation tasks.
This paper introduces MoganColBERT-TR, a late-interaction multi-vector retrieval model for Turkish, which achieves competitive zero-shot performance on Turkish BEIR datasets through distillation training from previous encoders.
This blog post introduces how to train and finetune multi-vector embedding models using the Sentence Transformers library, showcasing its v6.0 update with a new MultiVectorEncoder type and demonstrating superior performance on medical retrieval tasks.
The LateOn model with 140M parameters achieves strong results, and the community is excited about advances in multi-vector models including new CPU indexes and multilingual support.
SMART is a framework that unlocks latent multi-vector capabilities in single-vector models for multimodal retrieval, improving state-of-the-art performance with reduced computational costs via contrastive training and late-interaction inference.