Unverified50% confidenceFactExact time
In the transfer learning setting, the historical data are collected by an unknown adaptive policy, and the sample size N'_k is a random variable that may be correlated with the arms' reward distributions, which breaks the assumption in traditional statistical estimation that the sample size is fixed and independent of the data.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
Verified数据漂移（Data Drift）指生产环境中输入数据的统计分布随时间发生变化，导致模型接收数据与训练时分布不匹配68% similarUnverifiedRegression tasks require predicting continuous fine-grained numerical values and rely on fine-grained sample information in the history, which is lost when compressed into a compact persona description.67% similarUnverifiedThe stronger error correlation among frontier models is attributed to highly overlapping training data and the shared fitting of human preference data through RLHF, resulting in homogenization of error patterns.66% similarUnverifiedGradient-based methods often overwrite old knowledge when learning new tasks, causing catastrophic forgetting, whereas population-based evolutionary systems naturally maintain an archive of solution diversity.63% similarUnverifiedIncidental mutation is an independent metric measuring the severity of retention failures, counting how many fields that should have remained unchanged the model changed in a single state update.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1046782API
curl https://kongchang.com/api/v1/knowledge/claims/1046782MCP
get_claim(id=1046782)