When Retrieval Metrics Mislead: Measuring Policy Signal in Long-Horizon Tool-Use Agents
This paper shows that standard retrieval metrics can mislead when evaluating policy signal in long-horizon tool-use agents, highlighting a measurement gap in agent evaluation.
This paper shows that standard retrieval metrics can mislead when evaluating policy signal in long-horizon tool-use agents, highlighting a measurement gap in agent evaluation.
Auto-scored: A methodological paper on evaluation metrics for tool-use agents, offering a partial measurement improvement with minor framing risk.