👍 119
11/01 08:00
Computer graphics is built on code. Large language model (LLM) agents can now generate graphics code and iteratively refine it through various feedback. This capability opens the door to a potentially new paradigm of computer graphics, with graphics algorithms increasingly built around agentic syste
中文介绍 本文提出基于大语言模型代理的计算机图形学新范式。LLM 代理可生成图形代码,并通过多种反馈迭代优化,使图形算法围绕智能体系统构建,提升自动化和生成质量。
👍 95
11/02 21:41
Despite recent advances in CUAs, their development remains constrained by the limited diversity, controllability, and reliability of interaction data. Moreover, conventional optimization pipelines lack iterative feedback, while existing recursive improvement approaches often rely on self-generated e
中文介绍 本文提出 CUA-Master,一种通过开放世界递归改进扩展计算机使用代理(CUA)能力的框架。该方法利用多样化交互数据与迭代反馈机制,突破传统优化管线限制,提升代理在真实环境中的可扩展性和可靠性。
👍 89
11/02 22:15
Rapid advances in cowork agents are shifting human-AI collaboration toward real-world automation, delivering software, designs, and other artifacts. However, such productivity work demands both high-quality delivery on complex productivity tasks and continued monitoring and decision-making over long
中文介绍 本文提出 CoworkOS,一种扩展代理同事(cowork agents)实现真实世界自动化的框架。该系统不仅能高质量完成复杂生产力任务,还支持长期监控与决策,促进人机协作在软件交付和设计等领域的应用。
👍 34
11/01 08:00
Coding agents solve repository issues through long sequences of exploration, editing, and validation. These trajectories are increasingly used as post-training data, so how they are collected shapes what the trained model learns. A common approach is to sample multiple trajectories independently for
中文介绍 本文提出 SWE-Trail,一种历史引导的轨迹收集方法,用于编码代理训练。该方法基于仓库历史信息指导轨迹采样,避免独立采样导致的低效问题,提升训练数据的质量,从而增强编码代理的修复能力。
👍 32
11/02 08:00
Research taste, the judgment of what to read, which problems to pursue, and which methods to try, increasingly shapes the work of AI scientists, yet it remains far less understood than their ability to execute research. We introduce IdeaTaste, a benchmark that scores an LLM by how closely the resear
中文介绍 本文提出 IdeaTaste 基准,评估 LLM 与人类研究品味(研究品味)的匹配程度。该基准通过比较模型与人类研究者对论文、问题和方法的选择偏好,量化模型的研究品味能力,揭示 AI 在科研判断上的潜力与局限。
👍 31
11/02 08:00
Traditional image cropping assumes the photograph already contains the composition worth keeping, and asks only where to crop. A full 360^{circ} scene offers no such guarantee: the viewing direction in a scene has to be discovered before anything can be framed. We present PhotoGuider, a measure-then
中文介绍 本文提出 PhotoGuider,一种面向 360° 场景的构图感知裁剪与取景方法。它先测量场景内容再确定视角,利用可解释的摄影评估机制,自动发现最佳构图方向,解决传统裁剪无法处理全场景构图的问题。
👍 28
11/01 08:00
On-policy distillation (OPD) trains a student on its own generations by matching the teacher's token probabilities. These probabilities entangle two kinds of information: which tokens the teacher prefers, and how confident it is. The latter reflects the teacher's own capacity and calibration, which
中文介绍 本文提出一种仅排序的在线策略蒸馏方法(Order-Only On-Policy Distillation),通过分离教师模型的偏好排序与置信度信息,仅利用排序目标训练学生模型,避免置信度干扰,提升蒸馏效率与学生泛化能力。
👍 25
11/03 02:41
Multimodal large language models (MLLMs) can generate detailed video captions, yet they may overemphasize minor details while overlooking essential context and key events. We argue that not all visual evidence deserves equal attention, as the importance of visual information depends on the content o
中文介绍 本文提出 Reviso 方法,用于视频字幕生成中的视觉证据重加权。该方法依据视觉内容对关键事件的重要性分配注意力,避免 MLLM 过度关注次要细节而忽略关键信息,从而提升视频字幕的准确性和一致性。
👍 21
11/01 11:16
As coding agents achieve increasingly high success rates on software engineering benchmarks, recent efforts have sought to increase task complexity through longer horizons and more extensive code modifications. However, real-world software development often spans multiple interdependent repositories
中文介绍 本文提出 CrossSWE 基准,评估编码代理在跨仓库联合开发中的能力。该基准模拟真实软件工程中多仓库协作场景,涵盖跨仓库依赖、接口变更等挑战,为编码代理在复杂工程环境中的性能提供标准化测试。
👍 20
11/02 08:00
General-purpose robot policies must work across robots, tasks, and environments, yet the variation that makes physical trials informative also makes them difficult to reproduce. We introduce RoboColosseum, a distributed evaluation framework that combines unified policy serving and standardized proto
中文介绍 本文提出 RoboColosseum,一个分布式物理智能评估框架。该框架结合统一策略服务与标准化协议,支持跨机器人、任务和环境的可复现评估,实现大规模在野测试,提升机器人策略的泛化能力。
👍 19
11/03 08:00
Video generation has progressed from isolated clips to coherent short-form audiovisual sequences, and the bottleneck is shifting from rendering to semantic control. This report summarizes the observations and insights on multimodal understanding gained from successive generations of Kling. We argue
中文介绍 本文总结 Kling 视频生成模型在多模态理解方面的观察与见解。报告指出视频生成的瓶颈正从渲染转向语义控制,并强调多模态理解能力对生成连贯音视频序列的重要性,为未来模型设计提供方向。
👍 14
11/02 14:59
Progress in AI is often measured by what a single model can do on its own. Yet teams of AI agents are starting to make breakthroughs and discoveries in science, medicine, mathematics and software engineering. As AI begins to work in teams of agents and people, the ability to collaborate becomes a co
中文介绍 本文提出协作智能(Collaborative Intelligence)概念,用于衡量 AI 代理作为团队成员的协作能力。该框架评估代理在科学、医学、数学和软件工程等领域的团队协作表现,为 AI 在集体任务中的评估提供新维度。
👍 11
11/02 08:00
Speculative decoding accelerates memory-bound LLM decoding with a lightweight drafter, and parallel drafters extend this speedup by drafting a whole block in one forward pass. However, it remains unclear whether their released systems (stock) attain the full speedup that parallel drafting permits. W
中文介绍 本文提出 SpecPilot,一种自主算法发现与系统优化框架,用于推测解码。该方法自动搜索并行起草器的配置和优化策略,以最大化加速比,通过实验验证其性能优于现有系统,并揭示未充分利用的加速潜力。
👍 9
11/01 08:00
Harness optimization can substantially improve LLM agents without modifying model weights, but existing methods primarily optimize response quality. Extending them to joint quality--cost optimization introduces a challenge: applying the deployment preference during intermediate updates can premature
中文介绍 本文提出 ParetoSaddler,一种帕累托引导的代理框架优化方法,用于同时优化响应质量与成本。该方法避免在中间更新时过早应用部署偏好,通过帕累托前沿搜索平衡质量与成本,提升代理在不同预算下的表现。
👍 9
11/02 08:00
Tool-using language-model agents often execute actions that their domain policies forbid. We present LeanGuard, a runtime that checks every proposed tool call against a policy written in Lean4 before the call reaches the environment. Policies combine state predicates with scoped past-time temporal o
中文介绍 本文提出 LeanGuard,一种基于 Lean4 验证的运行时守护机制,用于工具使用代理。该机制在每次工具调用前检查策略约束,结合状态谓词与时态逻辑,防止代理执行被禁止的动作,增强安全性和合规性。
👍 9
10/30 08:00
Unlike autoregressive (AR) models, diffusion language models (DLMs) permit parallel token updates, flexible generation, and bidirectional attention, but leading systems rely on costly pretraining or full-model fine-tuning. We present Build Your Own DLM (BYOD), a generalizable LoRA-only recipe for pa
中文介绍 本文提出 BYOD(Build Your Own DLM),一种仅使用 LoRA 的通用转换方法,用于将自回归模型高效转换为扩散语言模型。该方法无需预训练或全模型微调,实现并行更新和双向注意力,提高生成灵活性与效率。
👍 8
11/03 01:57
Computer programmers chain together the same reusable computational primitives, for example, conditional statements or loops, but in different orders to implement different algorithms. In contrast, existing language model architectures always execute their layers in the same order. We introduce Libr
中文介绍 本文提出库层(Library of Layers),一种运行时自适应排序 Transformer 层的架构。该架构允许模型根据输入动态选择层顺序,提高推理灵活性,为语言模型提供类似程序化的模块化设计,增强任务适应性。
👍 7
10/28 08:00
Jev enables efficient decision-making with large language models (LLMs) through typed querying primitives, allowing the same semantic state to be expressed in different formats, such as binary and multiple-choice queries. However, the consistency of decisions across primitives has not been systemati
中文介绍 本文系统研究 Jev 中跨原语一致性问题。Jev 通过类型化查询原语(如二元和多选查询)表达语义状态,本文分析不同原语下决策的一致性,提出一致性度量方法,为使用 LLM 的高效决策提供可靠保障。
👍 6
10/27 08:00
Existing multi-agent collaboration (MAC) methods have primarily explored a group thinking collaboration topology, where agents independently generate solutions and aggregate their outputs. In this work, we investigate an alternative relay thinking topology, where reasoning ownership is dynamically t
中文介绍 本文提出多智能体中继思维(Multi-Agent Relay Thinking),一种新的协作拓扑。该方法动态转移推理所有权,代理按顺序接力处理问题,与现有独立生成后聚合的群组思维模式相比,显著提高复杂推理任务的性能。
👍 6
10/29 08:00
Predicting future physical states stands as a central challenge in embodied intelligence. Existing video world models leave 3D dynamics implicit, action-based formulations remain tied to specific robot configurations, and 3D motion prediction models use flow-only supervision, providing a limited lea
中文介绍 本文提出 WorldFlow,一种联合视频和 3D 运动预测的预训练框架。通过大规模预训练,该方法将 3D 动力学显式建模,避免隐式表示和动作依赖,实现跨机器人配置的通用预测,提升物理状态预测能力。