Chaos in the Text: Irrelevant Text Hurts Mixed-Modality Retrieval More Than Images
Yubo Sun · hf · 2026-10-09
A systematic study finds dense retrievers degrade sharply on mixed-modality corpora: performance follows a V-shaped curve as image documents are swapped for semantically equivalent text, with irrelevant text hurting more than irrelevant images—a phenomenon the authors call Chaos in the Text, driven by a modality preference that scores text higher. They propose Trident, which builds text, image, and fused views as co-equal positives optimized via Multi-Positive View InfoNCE, improving retrieval on both CLIP- and VLM-based architectures.
Related event: 'Text Chaos': Irrelevant Text Hurts Multimodal Retrieval More Than Images(2 posts)→
More from Research
- RFD4-Proteina's 'underrated' docs detail partial diffusion for protein redesign — AllThingsApx · 2026-10-23
- MPI-Sec PhD Thesis on Vulnerability Detection Benchmarks Includes Pre-LLM Classic "Top Score on the Wrong Exam" — mboehme_ · 2026-10-23
- LiveNerf's 29-day data finds no nerfing of Claude Opus 5.5 — TheOnlyVibemaster · 2026-10-23
- New paper makes 9 falsifiable predictions toward a major AI-enabled terror attack by 2027-2028 — StephenLCasper · 2026-10-23
- HumanoidPose recovers humanoid robots' full-body joint states from RGB video alone — cholz · 2026-10-23
- Sophont's Holon-S1 predicts genes and biomarkers from routine pathology slides — iScienceLuvr · 2026-10-23