Unverified50% confidenceFactExact time
The cosine similarity between two embedding vectors is meaningful only when they are produced by the same mapping function (the same model); changing the model (a new checkpoint, a new tokenizer, or a change in normalization) produces embeddings with the same shape but a different space.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedEmbedding models do not output text; they only convert inputs into vectors, whose meaning lies in their relative relationships to one another (the embedding space).69% similarVerified余弦相似度是衡量两个向量方向接近程度的指标，值域在-1到1之间，越接近1表示语义越相似65% similarUnverified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63% similarUnverified研究者在分析多语言BERT（mBERT）内部表示时发现，即便没有平行语料的明确监督，模型也能自发学到不同语言相同概念在向量空间中高度相似的对齐结构63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1007795API
curl https://kongchang.com/api/v1/knowledge/claims/1007795MCP
get_claim(id=1007795)