今日速览
| 序号 | 标题 | 来源 | 日期 | 主题 | 推荐等级 |
|---|---|---|---|---|---|
| 1 | Beacon: Knowing When and How to Perform Agentic Visual Reasoning | arXiv | 2026-07-30 | RAG | 高 |
| 2 | MANTA: Multi-Agent Network Topology Adaptation for Self-Evolving Multi-Agent Systems | arXiv | 2026-07-30 | RAG | 高 |
重点论文与技术动态
1. Beacon: Knowing When and How to Perform Agentic Visual Reasoning
- 来源：arXiv
- 日期：2026-07-30
- 作者/机构：Qixun Wang, Yang Shi, Letian Cheng, Zhuoran Zhang, Yan He, Yuqi Tang
- 主题标签：
RAG,arXiv - 推荐等级：高
- 分类：cs.CV
一句话结论
Beacon模型通过优化工具调用的适应性和工具效果，显著提升多模态大语言模型在复杂任务中的成功率。
核心内容
- 提出Mode Adaptiveness (MA) 和 Tool Effect (TE) 两个量化维度：MA确保模型仅在必要时调用工具以避免计算浪费，TE确保工具在模型无法解决的问题上有效扩展能力。
- 现有agentic视觉推理模型在MA上表现不足，工具使用在难例上的性能增益被易例上的错误引入所抵消。
- Beacon引入Necessity-Aware Adaptive Reward和Hint-Guided Capability Expansion机制，实现更强整体性能、改进的MA和真实的工具增益。
方法与数据
- 方法：核心机制为Necessity-Aware Adaptive Reward和Hint-Guided Capability Expansion。
- 数据：摘要未明确。
价值判断
- 值得关注：解决了agentic视觉推理的效率瓶颈，通过实证分析证明工具调用策略对任务成功率的关键影响。
- 可复用点：MA和TE评估框架可直接推广至其他多模态模型的工具集成优化。
- 局限/待核查：摘要未明确。
摘要
The fundamental goal of agentic visual reasoning is to improve the success rate of multimodal large language models (MLLMs) on complex tasks, rather than merely equipping them with a sophisticated yet inefficient reasoning paradigm. In this work, we rethink agentic visual reasoning through two key dimensions of tool use: Mode Adaptiveness (MA) and Tool Effect (TE). Mode Adaptiveness characterizes whether an MLLM can recognize when tools are truly necessary and invoke them accordingly, thereby avoiding unnecessary computational overhead while improving performance on challenging problems that require tool assistance. Tool Effect characterizes the actual impact of tool use: tools should extend the model's capabilities on problems unsolvable through text-only reasoning, while avoiding additional errors on problems that the model can already solve without tools. We conduct a comprehensive analysis to quantify these two properties and empirically reveal that existing agentic visual reasoning models exhibit limited Mode Adaptiveness, while the gains produced by tool use on hard examples are largely offset by the harm introduced on easy examples that the models can already solve. Motivated by these observations, we propose Beacon, a novel agentic visual reasoning model that achieves stronger overall performance, improved Mode Adaptiveness, and genuine tool-induced performance gains. At the core of Beacon are the Necessity-Aware Adaptive Reward and the Hint-Guided Capability Expansion mechanism in the reinforcement learning stage, which respectively encourage adaptive tool invocation based on task necessity and strengthen the model's tool-use capability on the most challenging problems. Extensive experiments across diverse benchmarks demonstrate the strong overall performance of Beacon and its substantial improvements in both Mode Adaptiveness and Tool Effect.2. MANTA: Multi-Agent Network Topology Adaptation for Self-Evolving Multi-Agent Systems
- 来源：arXiv
- 日期：2026-07-30
- 作者/机构：Mao-xun Huang, Jerry Wang, Yi-Cheng Lai, Zhengxin Zhang, Claire Cardie, Hen-Hsen Huang
- 主题标签：
RAG,arXiv - 推荐等级：高
- 分类：cs.AI
一句话结论
MANTA通过推理时通信拓扑自适应进化，显著提升多智能体系统性能。
MANTA框架在推理过程中动态调整通信结构，通过任务条件初始化和协作痕迹监控实施有限更新，优化代理角色、通信链接等，同时保留任务接口和代理预算。在信息寻求、工具使用、规划、工作流执行及数学推理五个基准测试中，平均得分74.0，比最强基线高5.8个百分点。
核心内容
- 通信拓扑在推理时自适应进化，突破传统固定设计或离线优化的限制。
- 初始化任务条件拓扑，监控协作痕迹，应用有限更新以动态调整代理角色、通信链接及执行顺序。
- 在五个基准测试中平均得分74.0，优于最强基线5.8个百分点，尤其在PlanCraft任务中表现最佳。
方法与数据
- 方法：基于任务条件初始化拓扑，通过协作痕迹监控实施有限结构更新。
- 数据：评估在五个基准测试（信息寻求、工具使用、规划、工作流执行、数学推理）上。
价值判断
- 值得关注：推理时自适应进化首次扩展至协作架构本身，为多智能体系统提供动态优化新范式。
- 可复用点：框架可迁移至其他多智能体场景，实现通信结构的实时优化。
- 局限/待核查：摘要未明确提及局限。