研究人员已将“任务不敏感性”确定为大型语言模型作为代理时，在分布外泛化能力较弱的关键原因。当模型将学到的模式应用于新的、相似的任务时，即使指令被损坏或语义改变，也会发生这种现象。为解决此问题，提出了一种名为任务扰动 NLL 优化（Task-Perturbed NLL Optimization）的新方法，该方法作为一种正则化器，确保行为更依赖于任务指令。评估表明，此干预措施在保持对任务相关信息的关注的同时，提高了任务敏感性和泛化能力。 AI
影响 这项研究可能带来更强大、更可靠的 AI 代理，能够处理更广泛的任务而不会降低性能。
排序理由 该集群包含一篇学术论文，详细介绍了一种提高 LLM 代理性能的新方法。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2606.26918v1 Announce Type: new Abstract: Large language models can serve as capable long-horizon agents, but their out-of-distribution (OOD) generalization remains weak. We identify a key source of this failure as task insensitivity: when faced with similar but distinct ta…
Large language models can serve as capable long-horizon agents, but their out-of-distribution (OOD) generalization remains weak. We identify a key source of this failure as task insensitivity: when faced with similar but distinct tasks, models might apply patterns learned during …
Large language models can serve as capable long-horizon agents, but their out-of-distribution (OOD) generalization remains weak. We identify a key source of this failure as task insensitivity: when faced with similar but distinct tasks, models might apply patterns learned during …