LEASH：面向计算机使用智能体的感知不确定性的弱到强升级与监督框架
LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents
AI总结：
LEASH是一种无需训练的计算机使用智能体框架，通过三层潜在不确定性机制实现弱智能体按需升级强智能体并持续监督，在OSWorld数据集上提升任务成功率且降低推理成本。
摘要：
计算机使用智能体越来越多地处理长周期任务，全程运行前沿模型成本高昂，而仅使用小型开放权重模型往往无法完成任务。弱到强升级提供了一种折中方案：弱智能体默认执行任务，仅在需要时将控制权交给强智能体。核心挑战在于何时进行升级，这需要一种足够廉价、可在每一步检查的不确定性信号。然而，即使在交接后，强智能体仍可能出错，而强智能体通常使用的黑盒API不暴露内部状态以便监控。为此，我们提出LEASH，这是一种无需训练的框架，仅在弱智能体的潜在不确定性信号显示需要时才启用强智能体，并在之后对其进行监督。LEASH包含三层：(1)用于早期路由的初始任务不确定性；(2)用于运行时升级的步骤级不确定性；(3)交接后用于持续监督强黑盒智能体的影子观察者。我们将潜在不确定性作为所有三层的驱动因素，具体而言，我们将弱模型的隐藏状态与其过去成功和失败轨迹的隐藏状态进行比较，更接近失败轨迹的状态表示更高的不确定性。在弱智能体执行期间，我们的不确定性估计器复用动作生成过程中已产生的激活，因此无需额外的前向传播或专用监控器训练。在OSWorld数据集上针对四种当代模型主干的实验表明，LEASH始终优于独立的弱智能体，且在匹配升级率下，比随机升级的性能提升6.55至11.94个百分点。使用Qwen3.5-35B-A3B时，LEASH的任务成功率为73.66%，每项任务成本为0.4117美元，与始终运行的Claude Opus 5（72.90%）性能相当，同时推理成本降低19.2%。
英文摘要：
Computer-use agents increasingly tackle long-horizon tasks, where running a frontier model at every step is expensive and a small open-weight model alone often fails. Weak-to-strong escalation offers a middle ground where a weak agent acts by default and hands control to a strong one only when needed. The core challenge is deciding when to escalate, which requires an uncertainty signal cheap enough to check at every step. However, even after handoff, the strong agent can still get lost, but a black-box API that the strong agent is usually served exposes no internals to monitor. To this end, we propose LEASH, a training-free framework that unleashes the strong agent only when the weak agent's latent uncertainty signals it is needed, and keeps it on a leash afterward. LEASH has three layers: (1) initial task uncertainty for early routing, (2) step-level uncertainty for runtime escalation, and, after handoff, (3) a shadow observer for continued oversight of the strong black-box agent. Here, we use latent uncertainty as a fuel for all three layers. Specifically, we compare the weak model's hidden states with those from its past successful and failed trajectories, and a state that looks more like the failures signals higher uncertainty. During weak-agent execution, our uncertainty estimator reuses activations already produced during action generation, so it needs no extra forward pass or dedicated monitor training. Experiments on OSWorld across four contemporary model backbones show that LEASH consistently improves over standalone weak agents and outperforms random escalation by 6.55--11.94 points at matched escalation rates. With Qwen3.5-35B-A3B, LEASH achieves 73.66% task success at $0.4117 per task, matching always-on Claude Opus 5 (72.90%) while reducing inference cost by 19.2%.
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。
