Unverified50% confidenceCautionExact time
Running K-Means directly on high-dimensional text embeddings may cause clustering results to be dominated by a small number of vectors with extremely large magnitudes due to differences in document embedding magnitudes.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
Unverified仅仅改变输入文本的措辞和结构，就能让同一个模型的性能产生数十个百分点的差异67% similarUnverified大语言模型在生成摘要时天然倾向于反映评论中的多数意见，出现频率越高的信息在模型内部表征中权重越大66% similarUnverified轻量模型在多语言覆盖和长文本处理上普遍存在短板66% similarUnverifiedMainstream text embedding models are typically trained based on contrastive learning, mapping semantically similar text pairs to nearby positions in vector space and pushing unrelated text pairs further apart.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/901370API
curl https://kongchang.com/api/v1/knowledge/claims/901370MCP
get_claim(id=901370)