发布日期: 2025-09-20

更新日期: 2025-09-22

文章字数: 2.2k

阅读时长: 9 分

阅读次数:

⚠️ 以下所有内容总结都来自于大语言模型的能力，如有错误，仅供参考，谨慎使用
🔴 请注意：千万不要用于严肃的学术场景，只能用于论文阅读前的初筛！
💗 如果您觉得我们的项目对您有帮助 ChatPaperFree ，还请您给我们一些鼓励！⭐️ HuggingFace免费体验

2025-09-20 更新

Ask-to-Clarify: Resolving Instruction Ambiguity through Multi-turn Dialogue

Authors:Xingyao Lin, Xinghao Zhu, Tianyi Lu, Sicheng Xie, Hui Zhang, Xipeng Qiu, Zuxuan Wu, Yu-Gang Jiang

The ultimate goal of embodied agents is to create collaborators that can interact with humans, not mere executors that passively follow instructions. This requires agents to communicate, coordinate, and adapt their actions based on human feedback. Recently, advances in VLAs have offered a path toward this goal. However, most current VLA-based embodied agents operate in a one-way mode: they receive an instruction and execute it without feedback. This approach fails in real-world scenarios where instructions are often ambiguous. In this paper, we address this problem with the Ask-to-Clarify framework. Our framework first resolves ambiguous instructions by asking questions in a multi-turn dialogue. Then it generates low-level actions end-to-end. Specifically, the Ask-to-Clarify framework consists of two components, one VLM for collaboration and one diffusion for action. We also introduce a connection module that generates conditions for the diffusion based on the output of the VLM. This module adjusts the observation by instructions to create reliable conditions. We train our framework with a two-stage knowledge-insulation strategy. First, we fine-tune the collaboration component using ambiguity-solving dialogue data to handle ambiguity. Then, we integrate the action component while freezing the collaboration one. This preserves the interaction abilities while fine-tuning the diffusion to generate actions. The training strategy guarantees our framework can first ask questions, then generate actions. During inference, a signal detector functions as a router that helps our framework switch between asking questions and taking actions. We evaluate the Ask-to-Clarify framework in 8 real-world tasks, where it outperforms existing state-of-the-art VLAs. The results suggest that our proposed framework, along with the training strategy, provides a path toward collaborative embodied agents.

最终目标是创建能够与人类互动的合作者，而非仅仅是被动执行指令的执行者。这要求智能体进行通信、协调和根据人类反馈调整其行动。最近，视觉语言能力的进步为实现这一目标提供了途径。然而，当前大多数基于视觉语言能力的实体智能体都处于单向模式：它们接收指令并执行，无需反馈。这种方法在处理指令通常模糊的实际情况时就会失败。在本文中，我们通过“询问以澄清”（Ask-to-Clarify）框架来解决这个问题。该框架首先通过多轮对话提出问题来解决模糊指令，然后生成端到端的低级动作。具体来说，“询问以澄清”框架包含两个组件，一个用于协作的视觉语言模型（VLM）和一个用于行动的分步模型。我们还引入了一个连接模块，该模块根据视觉语言模型的输出生成分步动作的条件。该模块通过调整指令的观察来创建可靠的执行条件。我们使用两阶段的知识绝缘策略来训练我们的框架。首先，我们使用解决模糊对话的数据微调协作组件，以处理模糊性。然后，我们整合行动组件同时冻结协作组件。这保留了交互能力的同时微调分步模型以生成动作。这种训练策略保证了我们的框架能够先提问后生成动作。在推理过程中，一个信号检测器充当路由器，帮助我们的框架在提问和采取行动之间进行切换。我们在8个实际任务中评估了“询问以澄清”框架的性能，它在现有最先进的视觉语言能力上表现出色。结果表明，我们提出的框架及其训练策略为实现协作实体智能体提供了途径。

论文及项目相关链接

PDF 9 pages, 4 figures

Summary
近期，视觉语言代理（VLAs）技术取得进展，为创建能够与人协作的实体代理提供了可能。然而，大多数现有基于VLAs的实体代理仍采用单向模式，无法根据人类反馈调整行动。为解决此问题，本文提出了“询问以澄清”（Ask-to-Clarify）框架，通过多轮对话解决模糊指令问题，并生成低层次行动。通过两阶段知识绝缘策略训练框架，确保先询问问题再采取行动。在8项真实任务中的评估结果表明，该框架优于现有技术，为实现协作式实体代理提供了途径。

Key Takeaways

实体代理的最终目标是创建能够与人互动的合作者，而非仅执行指令的被动执行者。
当前的VLAs技术虽然有进展，但大多仍采用单向模式运作。
“询问以澄清”框架通过多轮对话解决模糊指令问题。
框架包含两个组件：一个用于协作的视觉语言模型（VLM）和一个用于生成行动的动作扩散模型。
引入连接模块，基于VLM的输出生成动作扩散的条件。
采用两阶段知识绝缘策略进行框架训练，确保框架能先询问问题再采取行动。
在真实任务中的评估显示，该框架优于现有技术，为实现协作式实体代理奠定了基础。

Cool Papers

点此查看论文截图

MUSE: MCTS-Driven Red Teaming Framework for Enhanced Multi-Turn Dialogue Safety in Large Language Models

Authors:Siyu Yan, Long Zeng, Xuecheng Wu, Chengcheng Han, Kongcheng Zhang, Chong Peng, Xuezhi Cao, Xunliang Cai, Chenjuan Guo

As large language models~(LLMs) become widely adopted, ensuring their alignment with human values is crucial to prevent jailbreaks where adversaries manipulate models to produce harmful content. While most defenses target single-turn attacks, real-world usage often involves multi-turn dialogues, exposing models to attacks that exploit conversational context to bypass safety measures. We introduce MUSE, a comprehensive framework tackling multi-turn jailbreaks from both attack and defense angles. For attacks, we propose MUSE-A, a method that uses frame semantics and heuristic tree search to explore diverse semantic trajectories. For defense, we present MUSE-D, a fine-grained safety alignment approach that intervenes early in dialogues to reduce vulnerabilities. Extensive experiments on various models show that MUSE effectively identifies and mitigates multi-turn vulnerabilities. Code is available at \href{https://github.com/yansiyu02/MUSE}{https://github.com/yansiyu02/MUSE}.

随着大型语言模型（LLMs）的广泛应用，确保它们与人类价值观的一致性对于防止敌对势力操纵模型产生有害内容至关重要。虽然大多数防御手段针对的是单轮攻击，但现实世界的使用通常涉及多轮对话，使模型暴露在利用对话上下文来绕过安全措施的攻击之下。我们介绍了MUSE，这是一个全面的框架，从攻击和防御两个角度解决多轮突破问题。在攻击方面，我们提出了MUSE-A方法，该方法使用框架语义和启发式树搜索来探索各种语义轨迹。在防御方面，我们提出了MUSE-D这一精细的安全对齐方法，尽早介入对话以减少漏洞。在多种模型上的广泛实验表明，MUSE有效地识别和缓解了多轮漏洞。代码可在https://github.com/yansiyu02/MUSE处获取。

论文及项目相关链接

PDF EMNLP 2025 main conference

摘要

随着大型语言模型（LLMs）的广泛应用，确保其与人类价值观的对齐至关重要，以防止对手利用模型产生有害内容。现有防御措施主要关注单回合攻击，而真实世界的使用情况常涉及多回合对话，暴露模型面临攻击能利用对话上下文绕过安全措施的风险。本文介绍MUSE框架，从攻击和防御两个角度全面解决多回合漏洞问题。在攻击方面，我们提出MUSE-A方法，利用框架语义和启发式树搜索探索多样的语义轨迹。在防御方面，我们推出MUSE-D精细安全对齐方法，在早期对话中干预以降低漏洞风险。在多个模型上的广泛实验表明，MUSE能有效识别并缓解多回合漏洞问题。相关代码可通过链接访问：https://github.com/yansiyu02/MUSE。

关键见解