世界模型训练偏差常被归咎于环境噪声，本文指出根因在放大器失衡，提出稳定修正法，直击智能体推理要害。
arXiv:2607.01767v1 Announce Type: new Abstract: As agent planning moves from short tool chains toward persistent workflows with thousands or tens of t…
LLM也能当训练环境设计师？这篇论文提出LLM-as-Environment，自动为RL生成多智能体训练环境，终结手工调参。
arXiv:2606.17682v1 Announce Type: new Abstract: Reinforcement learning pipelines for Large Language Model (LLM) training often rely on manually redesi…
入选ICML2026的AutoTool论文，提出动态工具选择与集成方法，显著提升智能体推理能力，获ICCV Workshop最佳论文奖。
arXiv:2512.13278v2 Announce Type: replace-cross Abstract: Agentic reinforcement learning has advanced large language models (LLMs) to reason through l…
利用本体论引导多智能体推理实现大模型文化对齐，ICML 2026 最新成果。
arXiv:2601.21700v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) increasingly support culturally sensitive decision making, yet …
用LLM智能体推理改进符号回归的样本效率，新方法区分候选生成与搜索引导。
arXiv:2606.04360v1 Announce Type: new Abstract: Symbolic regression (SR) discovers compact mathematical expressions from data, yet recent LLM-based ev…
新智能体框架TSQAgent通过专用推理提升时间序列数据质量评级，开创数据质量评估新范式。
arXiv:2606.03629v1 Announce Type: new Abstract: Assessing the quality of time series (TS) data is fundamental yet inherently challenging due to the mu…
提出可验证过程奖励机制，让智能体推理更可信可解释，强化学习新思路。
arXiv:2605.10325v2 Announce Type: replace Abstract: Reinforcement learning from verifiable rewards (RLVR) has improved the reasoning abilities of larg…
提出ATLAS方法，用单个词统一智能体和潜在视觉推理，突破中间状态计算瓶颈
arXiv:2605.15198v1 Announce Type: cross Abstract: Visual reasoning, often interleaved with intermediate visual states, has emerged as a promising dire…