Get Started
Home
Topics
Search
Library
Research questionHow can cross-modal generative retrieval avoid hallucinating visual details missing from concise text queries?Concise text queries often omit fine-grained visual details that autoregressive decoders must nevertheless predict. Forced predictions during constrained identifier decoding can cause irrelevant visual candidates to outrank relevant ones.
Computer Vision
Information Retrieval
Machine Learning
Multimodal Models
Latest papersRecent research connected to this question, newest first.WIDE: Wildcard Inference with Dynamic Expansion for Cross-Modal Generative RetrievalThe source describes WIDE for this autoregressive cross-modal generative retrieval setting and reports experiments on the M-BEIR benchmark, using wildcard expansion and hybrid reranking to address decoder blind spots. The supplied evidence is limited to the abstract and its reported benchmark results.research paper · Sep 3, 2026
Related questions
How can generative retrieval represent both queries and items for direct matching in existing inverted indexes?How can document retrieval find the correct record when large collections share nearly identical visual templates?How can visual-document RAG adapt page retrieval to each query without hurting answer accuracy?How can image-text retrieval focus on caption-described attributes while ignoring unmentioned visual information?