Unverified50% confidenceFactExact time
RLHF标注员之间的一致性（inter-annotator agreement）通常只有70-80%
1
Sources
50%
Confidence
Long-term
Relevance
8/5/2026
First Seen
Sources
Related Claims
Unverified研究表明经过良好校准的LLM裁判与人工标注的一致性可达80%以上57% similarUnverifiedOn the comprehensive benchmark, IMG Sign achieves 81.02% while cosine similarity achieves 79.49%; on LFW the figures are 96.27% and 95.53% respectively55% similarUnverified主流ANN算法包括HNSW（分层可导航小世界图）和IVF-PQ（倒排索引+乘积量化），可在毫秒级响应下将召回率维持在90%以上54% similarUnverifiedResearch shows that the agreement rate between LLM judges and human scores is approximately 80%, approaching but not yet surpassing the level of human inter-annotator agreement (approximately 81%).52% similarUnverified研究表明，超过80%的审稿人在rebuttal后不会实质性更改评分52% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/689241API
curl https://kongchang.com/api/v1/knowledge/claims/689241MCP
get_claim(id=689241)