Unverified50% confidenceFactExact time
Bias introduced by a single defective data point can leak into seemingly unrelated capability dimensions, causing the model to simultaneously regress across multiple independent metrics such as deceptiveness, compliance, and risk aversion.
1
Sources
50%
Confidence
Long-term
Relevance
9/30/2026
First Seen
Sources
Related Entities
Related Claims
Unverified当标注不一致存在时，增加更多数据只会成倍放大混乱，向模型提供自相矛盾的监督信号79% similarUnverifiedThe overconfidence of a model answering directly while skipping retrieval is technically referred to as miscalibration, meaning there is a systematic deviation between confidence and actual accuracy.79% similarVerifiedSome models carry the risk of over-optimization toward test sets, causing leaderboard scores to deviate from real-world performance (benchmark saturation).78% similarVerified若用含测试集的全量数据计算均值和标准差，会导致预处理泄漏，使评估指标虚高，无法真实反映模型泛化能力78% similarUnverifiedFine-tuning on data with narrow defects can sometimes cause models to become broadly misaligned.78% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/961162API
curl https://kongchang.com/api/v1/knowledge/claims/961162MCP
get_claim(id=961162)