Event ·Multi-source

DolphinBench 发布：以动作正确性评测智能体记忆系统

Deshraj Yadav 团队发布 DolphinBench，一个针对智能体记忆的基准测试。它不再用问答形式打分，而是在长历史对话后检验智能体实际执行的动作是否正确。榜单同时记录准确率、成本和延迟三项指标，用于绘制记忆系统的 Pareto 前沿。测试页面在 dolphinbench.ai 开放。

Current read

做智能体的朋友看看这个新基准 DolphinBench，考的是长对话后动作对不对，还把准确率、成本、延迟放一张榜上，选记忆系统时挺实用。

3 sources73° AI heatLast updated 2026/9/22 18:06:41

Evidence

3 sources

Primary sourceTaranjeet

2026/9/22 18:06:41View source

Related source 1mem0

2026/9/22 16:41:25View source

Related source 2arXiv cs.AI

2026/9/21 17:54:35View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story