Unverified50% confidenceFactExact time
The statistical premise of drift detection is that the two sets of observations being compared must be comparable samples from the same distribution; directly comparing raw counts across versions produces false positives.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
Unverified对同一测试用例执行多次调用统计成功比例得到通过率，比二元的通过/失败判断更能反映模型真实表现76% similarUnverified多重比较问题指同时测试多个模型变体时，即使每个单独检验显著性水平设为0.05，整体假阳性率也会急剧膨胀74% similarUnverified综合多个独立来源的评测结果比单一自报基准分数更能反映模型的实际水平74% similarUnverifiedMcNemar 检验关注两个模型在每个样本上的分歧情况，只有判断结果不一致的样本（b和c单元格）具有信息量，两者都对或都错的样本被视为无关信息73% similarUnverifiedThe comparative experiment did not disclose specific dataset sizes, evaluation metrics, or details of comparison fairness, all of which affect the generalizability of its conclusions.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1013846API
curl https://kongchang.com/api/v1/knowledge/claims/1013846MCP
get_claim(id=1013846)