Unverified50% confidenceFactExact time

DiscoBench最终生成211个测试样本，包含463个歧义实例

1

Sources

50%

Confidence

Long-term

Relevance

7/9/2026

First Seen

Sources

腾讯DiscoBench：评测LLM搜索代理歧义澄清能力的首个基准

bilibili熊二等兵7/4/2026

Related Claims

Unverified本次测试共产生147次模型调用56% similar Unverified模型最终输出一个201×18×4的张量，其中4代表4条车道线，18代表关键行，201代表200个格子位置加1个不存在类别54% similar Unverified叠加cv=5的5折交叉验证后，144种参数组合需训练144×5=720次模型53% similar Unverified基准测试中每个仓库每一边各跑4次取中位数52% similar UnverifiedSol在Terminal Bench 2.1编程测试中获得9019分51% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/396699

API

curl https://kongchang.com/api/v1/knowledge/claims/396699

MCP

get_claim(id=396699)