用熵激活控制大语言模型智能体
Controlling Large Language Model Agents with Entropic Activation Steering
AI总结：
研究LLM智能体探索行为的控制问题，提出表征层面的激活控制方法EAST，通过影响高级动作和调节思想不确定性来有效操纵探索，并证明控制向量的跨任务泛化性，揭示了智能体在表征空间显式编码动作不确定性。
摘要：
大语言模型（LLM）的兴起促使人们对其作为上下文学习智能体的兴趣日益浓厚。智能体行为的核心是探索能力，即主动收集环境信息的能力。但 LLM 智能体如何探索，我们如何控制其探索行为？为回答这些问题，我们采取表征层面的视角，并引入 Entropic Activation Steering (EAST)，一种用于上下文 LLM 智能体的激活控制方法。首先，我们证明 EAST 能够通过直接影响从 LLM 输出中解析出的高级动作，有效地操纵 LLM 智能体的探索，这与 token 级别的温度采样形成对比。其次，我们揭示了应用此控制如何调节 LLM 思想中表现出的不确定性，从而引导智能体采取更具探索性的动作。最后，我们证明 EAST 获得的控制向量在不同任务变体间具有泛化能力。总而言之，这些结果表明 LLM 智能体在其表征空间中明确编码了对其动作的不确定性。我们的工作为理解 LLM 智能体的功能以及有效控制其决策行为铺平了道路。
英文摘要：
The rise of large language models (LLMs) has prompted increasing interest in their use as in-context learning agents. At the core of agentic behavior is the capacity for exploration, or the ability to actively gather information about the environment. But how do LLM agents explore, and how can we control their exploratory behaviors? To answer these questions, we take a representation-level perspective, and introduce Entropic Activation Steering (EAST), an activation steering method for in-context LLM agents. Firstly, we demonstrate that EAST can effectively manipulate an LLM agent's exploration by directly affecting the high-level actions parsed from the outputs of the LLM, in contrast to token-level temperature sampling. Secondly, we reveal how applying this control modulates the uncertainty exhibited in the LLM's thoughts, guiding the agent towards more exploratory actions. Finally, we demonstrate that the steering vectors obtained by EAST generalize across task variants. In total, these results show that LLM agents explicitly encode uncertainty over their actions in their representation space. Our work paves the way for a new understanding of the functioning of LLM agents and to effective control of their decision-making behaviors.
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。
