Unverified50% confidenceSolutionExact time
Using a large-scale, diverse prompt library when evaluating models or LoRA capabilities is more objective than using a few handpicked prompts, as it avoids selection bias.
1
Sources
50%
Confidence
Long-term
Relevance
9/22/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedA common limitation shared by mainstream bias benchmarks is that they measure a model's explicit choices in specific controlled contexts, rather than implicit biases that emerge in open-ended generation, long-text reasoning, or complex role-playing scenarios.72% similarUnverifiedThere is a systematic bias in annotator preferences, with higher scores given to polite, clearly structured, and risk-averse responses.71% similarUnverified根据待评内容类型通过语义相似度检索动态选择最相关的锚定样本，在实践中可将裁判-人工一致性提升5-15个百分点70% similarUnverified少量高质量的偏好数据往往比大量低质量数据能产生更好的对齐效果70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/939794API
curl https://kongchang.com/api/v1/knowledge/claims/939794MCP
get_claim(id=939794)