Safety
Clearing the Fog: Towards Installing and Refining Proactive Exploration Capabilities in LLM Agents
arXiv:2608.14339v1 Announce Type: new Abstract: We study proactive exploration in LLM agents, i.e., the ability to explore an environment to acquire information that improves future decision-making. I
arXiv:2608.14339v1 Announce Type: new Abstract: We study proactive exploration in LLM agents, i.e., the ability to explore an environment to acquire information that improves future decision-making. In this regard, we first identify two fundamental bottlenecks that hinder this capability and then propose ours, a novel method designed to instill and refine proactive exploration. Specifically, ours consists of two components: (1) Exploratory Data Construction, which synthesizes exploration-rich trajectories to mitigate the hindsight bias of standard demonstrations; and (2) RL Optimization with Contrastive Signal Guidance, which leverages contrastive trajectory pairs to distinguish productive exploration from redundant wandering. Extensive experiments demonstrate the effectiveness of ours and provide insights into the characteristics of proactive exploration. Our code is available at: https://github.com/GuanZhizhao/SAFARI.
Related
- Internalizing the Future: A Unified Agentic Training Paradigm for World Model Planning
- Exploration and Exploitation Errors Are Measurable for Language Model Agents
- Agentic Transformers Provably Learn to Search via Reinforcement Learning
- SPADER: Step-wise Peer Advantage with Diversity-Aware Exploration Rewards for Multi-Answer Question Answering
Source: arXiv cs.AI | 2026-08-17