Visual Document Retrieval (VDR) typically operates as text-to-image retrieval using specialized bi-encoders trained to directly embed document images. We revisit a zero-shot generate-and-encode pipeline: a vision–language model first produces a detailed textual description of each document image, which is then embedded by a standard text encoder. On the ViDoRe-v2 benchmark, the method reaches 63.4% nDCG@5, surpassing the strongest specialised multi-vector visual document encoder, and it scales similarly on MIRACL-VISION with broader multilingual coverage. Analysis shows that modern vision–language models capture complex textual and visual cues with sufficient granularity to act as a reusable semantic proxy. By off-loading modality alignment to pretrained vision–language models, our approach removes the need for computationally intensive text-image contrastive training and establishes a strong zero-shot baseline for future VDR systems.

Learned Sparse Retrieval (LSR) models use vocabularies from pre-trained transformers, which often split entities into nonsensical fragments. Splitting entities diminishes retrieval accuracy and limits the model’s ability to incorporate up-to-date world knowledge not included in the training data. In this work, we enhance the LSR vocabulary with Wikipedia concepts and entities, enabling the model to resolve ambiguities more effectively and stay current with evolving knowledge. Central to our approach is a Dynamic Vocabulary (DyVo) head, which leverages existing entity embeddings and an entity retrieval component that identifies entities relevant to a query or document. We use the DyVo head to generate entity weights, which are then merged with word piece weights to create joint representations for efficient indexing and retrieval using an inverted index. In experiments across three entity-rich document ranking datasets, the resulting DyVo model substantially outperforms several state-of-the-art baselines.