[KongchangAI]
Unverified50% confidenceBenchmarkExact time

On OpenCLIP, a blank canvas after sparse pixel edits achieved strong top-1 and top-5 performance in target retrieval tasks.

1
Sources
50%
Confidence
Long-term
Relevance
10/26/2026
First Seen

Sources

SPOOK Reveals Blind Spots in Vision Encoders: How Sparse Pixel Optimization Breaches Semantic Boundaries

Related Entities

OpenCLIPmodel

Related Claims

UnverifiedOn DINOv3, a blank canvas after sparse pixel edits likewise showed high target alignment and near-top prototype rankings.73% similarUnverifiedOpenAI的CLIP模型是早期多模态对齐的里程碑，通过对比学习让图文表示在向量空间中对齐63% similar

Cite This Claim

Stable URI
https://kongchang.com/claim/1098380
API
curl https://kongchang.com/api/v1/knowledge/claims/1098380
MCP
get_claim(id=1098380)