Unverified50% confidenceBenchmarkExact time
In benchmarks, the prototype system requires 3-5 times fewer annotations than random sampling because acquisition is concentrated on borderline cases.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/13/2026
First Seen
Valid until: 1/11/2027
Sources
Related Entities
Related Claims
UnverifiedThe experiment has a small sample size (single model, single test, 50 questions), is subject to random fluctuation, and constitutes a heuristic finding rather than a definitive conclusion.73% similarUnverified在机器学习多种子实验中重复次数通常只有5到10次，样本量极小，正态性假设难以验证，因此 Wilcoxon 检验往往是更保守稳健的选择70% similarUnverified该测试样本规模相对有限（5个模型、13个代码库），结论有待更大规模验证68% similarUnverifiedSetting an absolute failure count ceiling for small-sample slices is more robust than setting a percentage threshold and is insensitive to sample size68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1020481API
curl https://kongchang.com/api/v1/knowledge/claims/1020481MCP
get_claim(id=1020481)