Unverified50% confidenceFactExact time
In semantic space, semantically similar inputs are mapped to positions close to each other, while semantically dissimilar inputs are pushed apart.
1
Sources
50%
Confidence
Long-term
Relevance
8/8/2026
First Seen
Sources
嵌入向量时代：数据科学从特征工程到表征选择的范式转移
Related Claims
Unverified提示词注入分为直接注入和间接注入两种形式，因语言模型无法区分数据与指令，凡进入上下文窗口的文本都可能被执行69% similarUnverifiedSynonym normalization can be achieved by using sentence embedding models such as Sentence-BERT to map words into dense vectors, and then computing semantic proximity via cosine similarity.69% similarUnverified经过 Attention 处理后得到的向量称为 Contextualized Embedding，能结合上下文让相同 Token 根据语境得到不同表示68% similarUnverified代码的语义更多体现在结构关系而非字面文本，一个函数改动的影响范围必须沿调用图遍历而非仅靠语义相似度捕获68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/711026API
curl https://kongchang.com/api/v1/knowledge/claims/711026MCP
get_claim(id=711026)