Hacker News· paidev·· 6 天前AI 评分22
JevBench：面向类型化决策模型的开放基准，覆盖 Clef、pplx-decider 等 15 个以上模型
JevBench: Benchmarking decision models (Clef, pplx-decider and 15 more)
AI 导读
JevBench 是一个面向类型化决策模型的开放基准，提供论文 PDF、数据集与 GitHub 代码，覆盖 Clef、pplx-decider 等 15 个以上模型。榜单以 0 到 100 的均分展示每个模型在各数据集上的表现，并用条形区间标出最差到最好的模型，悬停单元格可查看区间与覆盖率，点击数据集名可生成图表。
正文 · 原文
Open benchmark for typed decision models.
Hover a cell for its interval and coverage. Click a dataset name to chart it.
Models
click a model for scores and where to get it
Average: mean score of every model on that dataset, 0 to 100. Bar: worst to best model. Scales differ, so compare datasets of the same metric most closely.
Submit
来源：Hacker News · leanmcp.github.io