Unverified50% confidenceOpinionExact time
For BGE-small, query anchoring remains the dominant factor on the retrieval side, and the injection payload itself has limited impact on ranking metrics.
1
Sources
50%
Confidence
Long-term
Relevance
11/4/2026
First Seen
Sources
Related Entities
Related Claims
Unverified检索策略的效果高度依赖数据特征——在FAQ上BM25可能打败向量检索,在语义模糊的查询上则相反。74% similarUnverifiedDuring HNSW queries, the upper-layer search uses ef = 1, keeping only one candidate and performing pure greedy descent; the bottom layer uses a beam search of size ef.68% similarUnverifiedExposure ranking on microstock platforms is determined by search keyword relevance, asset quality scores and account history performance, and upload volume itself generates almost no traffic.65% similarUnverifiedVector embedding retrieval relies on approximate nearest neighbor search, and results are affected by embedding model version, index parameters, and floating-point errors, potentially producing differences between runs.65% similarUnverifiedThe only helpful placement of orthogonal constraints is to use symmetric positive-definite factors on the query and key projections, but its tiny gain vanishes as the model grows.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1133894API
curl https://kongchang.com/api/v1/knowledge/claims/1133894MCP
get_claim(id=1133894)