Newest papers

WeAct: Dynamic Weight Switching as Agent Actions

Rui Shi, Xunjian Yin, Jiacheng Sang, Zichuan Liu, Shuyan Zhou · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26847 · cited by 0 works (OpenAlex) · open-access PDF

Chinese

长期经纪人必须从交叉路径中学习异质能力。更新共享参数以学习新能力可以降低先前获得的能力,导致灾难性遗忘。随着轨迹的演变,不同的阶段可能需要不同的能力,但传统的联合训练和静态适配器合并在整个执行过程中保持固定的参数配置。我们引入了WeAct(重量级内部行动),它训练了独立的能力特定的LoRA适配器在冻结的背骨上,并将适配器选择添加到代理的行动空间中。这些内部操作使用了模型的原生工具调用界面,以便在共享的环境中组成需求能力。在八个深度研究基准中,Qwen3-32B与WeAct的现有适配器提高了平均精度,平均为6.2%的适配器添加

Machine translation (Meta M2M100); the English original follows.

Abstract

Long-horizon agents must learn heterogeneous capabilities from interleaved trajectories. Updating shared parameters to learn new capabilities can degrade previously acquired ones, resulting in catastrophic forgetting. As a trajectory unfolds, different stages may require different capabilities, yet conventional joint training and static adapter merging maintain a fixed parameter configuration throughout execution. We introduce WeAct (Weight-level Internal Actions), which trains independent capability-specific LoRA adapters over a frozen backbone and adds adapter selection to the agent's action space. These internal actions use the model's native tool-calling interface for on-demand capability composition in a shared context. Across eight deep-research benchmarks, Qwen3-32B with WeAct improves average accuracy by 6.2% relative to full fine-tuning counterparts despite using less compute, and by 27.5% relative to the strongest static LoRA-merging method. WeAct also supports modular continual adaptation, by adding a new code computation adapter to expand the agent's capabilities and updating the existing writing adapter to improve report quality. Both updates leave the backbone and other adapters unchanged, supporting modular continual adaptation in agents.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Adapter (computing), Modular design, Computer science, Train, Computation, Distributed computing

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange