ACCORD: 面向语言智能体的动作条件上下文接地
ACCORD: Action-Conditioned Contextual Grounding for Language Agents
AI总结：
针对用户指令常因隐含环境假设而欠指定，导致LLM智能体执行失败的问题，提出ACCORD框架，在每次动作前主动探测缺失信息并整合轨迹上下文，无需额外训练，在AppWorld和AlfWorld上显著提升任务完成率。
摘要：
用户指令往往因人类对周围环境的隐含假设而欠指定。对于在信息丰富的数字和物理环境中运行的大型语言模型（LLM）智能体，这些假设无法仅从指令中推断；必须从工具、数据、接口和观察的当前状态中恢复。因此，有效执行要求智能体识别缺失的上下文，将其基于观察到的证据，并带入后续动作。我们表明，当前智能体常常未能做到这一点。它们基于假设而非观察到的细节行动，忽略本可收集的信息，并且未能整合已经返回的证据。基于这一洞察，我们提出ACCORD（动作条件上下文接地），一种简单有效的自适应接地智能体框架。在每次动作前，ACCORD主动探测环境中缺失的信息，并整合来自智能体轨迹中原本会被忽略的相关上下文。无需额外训练或任务成功信号，ACCORD在AppWorld上将任务目标完成率从42.0%提升至62.6%（GPT-5-mini），比强基线高出最多20.6个百分点。这些增益在更强的基模型（Claude-4.5-sonnet上+10.8）、开放权重模型（Qwen3.5-27B-FP8上+10.1）以及具身AlfWorld基准（GPT-5-mini上成功率+7.4）上持续存在。
英文摘要：
User instructions are often underspecified because humans rely on implicit assumptions about the surrounding environment. For large language model (LLM) agents operating in information-rich digital and physical environments, these assumptions cannot be inferred from the instruction alone; they must be recovered from the current state of tools, data, interfaces, and observations. Effective execution therefore requires agents to identify missing context, ground it in observed evidence, and carry it forward into subsequent actions. We show that current agents often fail to do so. They act from assumed rather than observed specifics, overlook information they could have gathered, and fail to incorporate evidence that has already been returned. Building on this insight, we propose ACCORD (Action-Conditioned Contextual Grounding), a simple and effective agent framework for adaptive grounding. Before each action, ACCORD actively probes the environment for missing information and integrates relevant context from the agent's trajectory that would otherwise be overlooked. Requiring no additional training or task-success signals, ACCORD improves task-goal completion on AppWorld by up to +20.6 points with GPT-5-mini, from 42.0% to 62.6%, compared to strong baselines. These gains persist with a substantially stronger base model (+10.8 with Claude-4.5-sonnet), an open-weight model (+10.1 with Qwen3.5-27B-FP8), and on the embodied AlfWorld benchmark (+7.4 success rate with GPT-5-mini).
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
