arXiv 论文速递
Modern video games provide a measurable testbed for AI models, combining abilities of visual understanding, instruction decomposition, goal planning, and precise action control ove…
Multi-turn tool-use failures can hinge on a single model call, yet reward variation alone does not reveal which call would benefit from training. When rewards depend on later inter…
视频世界模型在长时程和视角变化下难以保持一致性，本文提出隐式3D感知记忆机制，对可交互动态环境建模有重要推进。
Video world models enable interactive exploration of dynamic environments, yet struggle to respect prior observations over long horizons and across viewpoints. We present WorldCraf…
· 2026-09-21
We present onPanda, an interactive tool for efficiently annotating LLM alignment data and agent trajectories. onPanda adopts token-level correction as its core interaction: while r…
端侧LLM个性化受限于算力，本文用超网络生成LoRA权重，实现高效设备端个性化，对移动端大模型落地有实用价值。
On-device large language models (`LLMs'), e.g. running on mobile phones, are ripe for improvement via personalization. The limited compute resources of mobile devices impose limits…
灵巧操作依赖视觉难以完全观测的接触动力学，本文融合视觉与触觉构建世界-动作模型，提升复杂操作任务的建模能力。
Dexterous manipulation depends on contact dynamics that are often only partially observable from vision. Recent World-Action Models (WAMs) couple predictive video world modeling wi…
· 2026-09-21
Agent harnesses, the external systems that mediate model-environment interaction, can substantially improve agent performance, but their gains remain tied to the harness at deploym…
智能体能力高度依赖外围harness，本文提出正则化递归自我改进方法自动优化harness，为智能体性能提升提供新范式。
An LLM agent's capability is largely magnified by its harness, namely the prompts, control flow, tooling, memory, and context management surrounding the frozen backbone model. Rece…
Agents today often take real-world actions that depend on long-term memory and context recall over time. However, most current memory benchmarks are built for a conversational ques…
· 2026-09-21
As agents are deployed with increased autonomy, even extremely rare events along their stochastic output trajectories can occur and prove catastrophic. Safe deployment therefore do…
· 2026-09-21
多智能体长期交互可能产生非预期的合谋行为，本文系统研究该风险，对LLM智能体安全部署与治理具有警示意义。
LLM agents are increasingly deployed in collaborative settings, yet long-term interaction may give rise to undesirable coordination. We study the emergence of collusion in a long-h…
Scientific workflows often require choosing among known relations before a deterministic calculation can proceed. Whether observations share a culture, treatment or reference stand…