Unverified85% confidenceFactTime unknown
现代Embedding模型通常基于BERT或其变体架构，输出768维至4096维不等的稠密向量
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
DeepSeek+RAGFlow本地部署AI知识库教程（2025最新）
bilibili大模型基础知识
Related Entities
Related Claims
VerifiedEmbedding模型通常将文本映射到768或1536维的高维向量空间74% similarUnverifiedEmbedding向量维度通常在768到3072之间，维度越高表达能力越强但计算成本也越高71% similarVerified嵌入模型通常将文本压缩为768或1536维的稠密向量61% similarUnverified适用面积估算公式：CADR(m³/h) × 0.07 ~ 0.12 = 适用面积(㎡)。层高2.8米时建议按0.08系数计算，即CADR 400的机器适用约32㎡58% similarUnverifiedwidth: clamp(300px, 50%, 600px)同时定义了最小值、理想值和最大值，让元素在任何视口下适应57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/18840API
curl https://kongchang.com/api/v1/knowledge/claims/18840MCP
get_claim(id=18840)