Preprint
Sep 2026
RandSlot: Learning Compact Visual Document Representations with Random Soft Tokens
Visual document retrieval requires expressive representations to match queries with evidence distributed across text, tables, and page layouts. Multi-vector representations capture fine-grained information, but storing and comparing many vectors introduces substantial retrieval costs. In this paper, we introduce RandSl...
De-Wen Guo, Shi Yu, Ling-Xiao Zhang et al.
· 0 citations