NVIDIA 如何评估 AI 智能体：从工具调用到任务完成
How to Evaluate AI Agents From Tool Calls to Task Completion
NVIDIA 发布文章阐述 AI 智能体评估方法，指出上线智能体的关键问题是能否在真实环境中连续完成数十次顺序工具调用并在某步失败后恢复，而仅评估模型回答是否"听起来对"无法反映任务是否真正完成。因此智能体评估正从为单次函数调用打分，转向为整个任务打分。
When you ship an AI agent, the key question is whether it can execute a chain of work across dozens of sequential tool calls against a live environment, and...
When you ship an AI agent, the key question is whether it can execute a chain of work across dozens of sequential tool calls against a live environment, and recover when a step fails. Scoring whether the model sounds right tells you almost nothing about whether the work finished. That gap is why agent evaluation has had to evolve from scoring a single function call to scoring an entire task…
来源：NVIDIA Developer Blog · developer.nvidia.com