Unverified50% confidenceFactExact time
The researchers define the failure mode in which models' language is not correctly bound to visual entities, regions, or locations during visual reasoning as the "reference gap."
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe reference gap is a bottleneck independent of visual perception, explaining some errors previously misjudged as insufficient perception.78% similarUnverifiedThe essence of the reference gap is the manifestation of the coreference resolution problem in visual reasoning, where models are prone to referential drift during multi-step reasoning.78% similarVerifiedThe quality of a visual localization system's predictions is highly dependent on the number of usable geographic cues present in the image, and information-sparse images can put the model in a difficult position.73% similarUnverifiedThe paper proposes that the deficiencies of multimodal large models in spatial, embodied, physical, and temporal reasoning may all stem from the same missing capability—visual transition reasoning.73% similarUnverifiedVision-language models have revealed three core issues on the Winoground and ARO benchmarks: attribute binding errors, weak spatial relationship understanding, and an absence of counting ability.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1113901API
curl https://kongchang.com/api/v1/knowledge/claims/1113901MCP
get_claim(id=1113901)