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
Related Entities
Related Claims
VerifiedEmbedding models typically map text into high-dimensional vector spaces of 768 or 1536 dimensions.74% similarUnverifiedEmbedding vector dimensions typically range from 768 to 3072; higher dimensions provide stronger expressive capability but at greater computational cost.71% similarVerifiedEmbedding models typically compress text into dense vectors of 768 or 1536 dimensions.61% similarUnverifiedwidth: clamp(300px, 50%, 600px) simultaneously defines a minimum value, an ideal value, and a maximum value, allowing an element to adapt at any viewport size.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)