Unverified50% confidenceBenchmarkExact time
The gap between behavioral and causal tests is statistically highly significant (item-matched pairing, 111:14 discordant pairs, p < 1e-15).
1
Sources
50%
Confidence
Long-term
Relevance
9/24/2026
First Seen
Sources
Related Claims
Unverified基准测试面临饱和（Saturation）和数据污染（Data Contamination）两个系统性挑战，当多个主流模型得分超过95%时测试集失去区分能力68% similarUnverified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67% similarUnverified早期研究发现同一问题用不同表述方式提问，模型性能差异可高达数十个百分点67% similarUnverifiedThe research team pointed out that the performance gains from inference-time context management in benchmarks exceeded most system-to-system differences that have been reported.67% 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67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/946174API
curl https://kongchang.com/api/v1/knowledge/claims/946174MCP
get_claim(id=946174)