Unverified50% confidenceFactExact time
When out-of-distribution (OOD) failure occurs, learned selectors may persistently ignore key evidence, resulting in answers that appear fluent but omit key facts, increasing the difficulty of troubleshooting.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
Unverified研究团队证明学习分布与真实分布之间的总变差距离可分解为初始化误差、Score误差和求解器离散化误差三项77% 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76% similarUnverified传统回归测试面临核心困境：全量回归耗时过长，而随机选择可能遗漏关键场景75% similarUnverifiedBehavior cloning can only learn the state distribution covered by demonstration data, and error accumulation tends to occur when encountering rare situations.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1011545API
curl https://kongchang.com/api/v1/knowledge/claims/1011545MCP
get_claim(id=1011545)