Unverified50% confidenceFactExact time
Domain shift refers to the distribution change caused by differences in experimental conditions, among which batch effect is the most common one.
1
Sources
50%
Confidence
Long-term
Relevance
10/22/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedResearchers observed that queries from different domains may undergo similar representational shifts during fine-tuning, causing behavioral features to be erroneously activated across domains.68% similarUnverifiedDistribution shift includes three common types: covariate shift (the input distribution changes but the conditional distribution does not), label shift (the label marginal distribution changes) and concept drift (the conditional distribution changes over time).65% similarUnverified概念漂移在不同社区有不同术语：统计学家称dataset shift或distribution shift，机器学习社区区分covariate shift和posterior drift，时间序列领域使用non-stationarity65% similarUnverified概念漂移是数据底层分布随时间发生的自然变化，而数据投毒是攻击者蓄意制造的分布偏移，两者在实际系统中往往同时存在且难以区分63% similarUnverifiedEmergent misalignment spreads through shared representations within the model to behavioral dimensions beyond the fine-tuned domain.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1083985API
curl https://kongchang.com/api/v1/knowledge/claims/1083985MCP
get_claim(id=1083985)