Unverified50% confidenceSolutionExact time
When using LLM-as-a-judge for safety evaluation, one should explicitly check the judge's refusal rate rather than only looking at the final score.
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedLLM-as-a-Judge 常见评估模式包括单点评分、成对比较以及参考答案对照评分80% similarUnverifiedMandatory checks should be separated from scoring; no matter how high the score on optional check items, it cannot offset a failed mandatory check.74% similarUnverifiedDischargeBench relies on the review reliability of LLM-as-a-Judge, and the degree of its alignment with physicians' actual judgments still requires ongoing validation.72% similarUnverifiedJudge 评估器适合主观维度（如幻觉、语气、指令遵循）的评测，常配合 0-100 分和通过阈值使用。71% similarUnverifiedSubjective judgment criteria (such as whether content is comprehensive) are almost random in LLM evaluation, whereas factual checks (such as whether X exists) rarely flip.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1032801API
curl https://kongchang.com/api/v1/knowledge/claims/1032801MCP
get_claim(id=1032801)