Unverified50% confidenceBenchmarkExact time
在人工验证的Track B v1.0标准集上(161个条目),仲裁后的Cohen's kappa值达到0.85
1
Sources
50%
Confidence
Long-term
Relevance
9/25/2026
First Seen
Sources
Related Entities
Related Claims
Unverified在边界匹配的片段上 Cohen's kappa 达到 0.964,但纳入遗漏实体、边界错误和标签修正后宏观 F1 骤降至 0.75359% similarUnverifiedCaveman给出的固定54次运行基准测试结果显示输入Token减少33.2%,且18/18项正确性检查全部通过59% similarUnverifiedQwen3.5-122B-A10B 在 100 次测试中出现 8 次拒答,KL 散度为 0.085657% similarUnverified千问3的235B参数MoE旗舰版在多项权威基准测试中接近甚至超越闭源顶级模型57% similarVerifiedF1分数是Fβ分数在β=1时的特例57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/949059API
curl https://kongchang.com/api/v1/knowledge/claims/949059MCP
get_claim(id=949059)