Unverified50% confidenceFactExact time
The 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.
1
Sources
50%
Confidence
Long-term
Relevance
7/24/2026
First Seen
Sources
DeepSeek神秘撤稿论文：用视觉基元突破AI多模态推理瓶颈
bilibili枫叶边城7/11/2026
Related Claims
Unverified自注意力机制的计算路径长度与序列长度无关，任意两个位置之间的信息传递都是直接的，解决了长序列建模中的梯度消失问题70% similarUnverifiedWhen text instructions precede image input, the Transformer's cross-attention mechanism causes the model to have already formed clear query vectors when encoding the image, preferentially activating visual regions semantically related to the target.68% similarUnverifiedThe paper aggressively compresses images, reflecting the view that the bottleneck lies not in pixels but in the referencing mechanism, which runs counter to the traditional thinking on perceptual gap.67% similarUnverified将稀疏可解释性方法拓展到视觉领域面临独特挑战，因为视觉特征具有更强的空间相关性、层次性和组合性67% similarUnverifiedStyle drift is an inherent challenge in image generation with diffusion models: even using exactly the same text prompt, two generated results may differ significantly in visual style.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/609001API
curl https://kongchang.com/api/v1/knowledge/claims/609001MCP
get_claim(id=609001)