Unverified50% confidenceFactExact time
数据多样性对模型泛化能力的影响往往比数据规模更为根本，仅在特定分布上过度拟合的模型在真实部署时表现骤降
1
Sources
50%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
Related Claims
UnverifiedIf only the data of newly added classes is used to fine-tune an expanded classification head, the model is highly prone to catastrophic forgetting, rapidly losing its ability to recognize the original classes.70% similarUnverified合成数据可能出现模式坍塌导致多样性不足、遗漏尾部分布中稀有模式、在高维数据中引入不真实相关性等质量问题69% similarUnverified错误的归纳偏置会让模型花费容量拟合数据中不存在的虚假模式，导致分布外数据上严重失效67% similarUnverified验证集和测试集不应做任何数据增强，一旦施加变换会引入系统性偏差，破坏对模型泛化能力的无偏估计67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/558817API
curl https://kongchang.com/api/v1/knowledge/claims/558817MCP
get_claim(id=558817)