👍 138
11/11 08:00
System One models promise fast, general-purpose decision-making without lengthy token generation. However, understanding their capability spectrum and decision reliability requires more than aggregate performance on current benchmarks. Existing evaluations often provide limited insight into the brea
中文介绍 本文提出SO-Arena基准,系统评估System One模型的能力谱系与决策可靠性。该基准超越了现有聚合性能指标,旨在深入揭示这些快速通用模型在无需长文本生成下的决策稳定性与能力边界。
👍 46
11/12 08:00
Video generation models produce visually realistic videos but still struggle with physical dynamics and causal consistency. Learning from generated videos requires fine-grained physical supervision, yet reliable quality judgments, correction pairs, and controlled variations are difficult to obtain.
中文介绍 本文提出PhysLoop框架,利用LLM的物理知识实现视频生成模型的自改进。通过生成模型与LLM的交互循环,模型可从生成视频中获取物理监督信号,提升视频生成的物理动态与因果一致性,无需人工标注修正对。
👍 43
11/11 08:00
In this report, we introduce the Reciprocal World Model (RWM) and present Duo v1, a full-duplex world model that continuously perceives audiovisual inputs while responding to users in real time. The RWM treats intelligence and physical reality as equally fundamental to what constitutes a world. Exis
中文介绍 本文提出Duo v1,一种全双工世界模型,能持续感知视听输入并实时响应用户。其核心是互惠世界模型(RWM),将智能与物理现实视为同等基础,实现交互式感知与响应的统一框架。
👍 37
11/11 08:00
Recent work suggests that pretrained image generators encode real-world understanding beyond image synthesis. Vision Banana exploits this by framing vision tasks as conditional image editing. However, its data, weights, code, and implementation details are not public. We present Open Vision Banana (
中文介绍 本文开源了Open Vision Banana,一个完全可复现的生成式视觉感知模型。它利用预训练图像生成器的内在理解,将视觉任务转化为条件图像编辑,并公开了数据、权重、代码与实现细节,促进可复现性。
👍 35
11/12 08:00
Decision models answer a question with a typed output, such as a choice among options with a probability for each, so one model can choose an agent's actions, evaluate actions already taken, or score every action to teach a policy with reinforcement learning. Prior work has studied them mainly as ac
中文介绍 本文系统研究长时程智能体中的决策模型,将其分为行动者、评判者与教师三类角色,并分析它们在不同训练与推理设置下的表现。研究揭示了决策模型在智能体长期任务中的多功能性和潜在权衡。
👍 29
11/04 08:00
Hyper-Connections (HC) architectures are increasingly adopted in frontier language models, yet their components remain insufficiently understood and introduce non-negligible training overhead. In this work, we systematically study the HC architecture to understand its components and identify simpler
中文介绍 本文系统研究超连接(HC)架构的组件,旨在简化设计并降低训练开销。通过深入分析,作者识别出可简化的组件,并提出了更稳定、更具可扩展性的简化版本,在保持性能的同时减少计算成本。
👍 25
11/12 08:00
Decoder-only Large View Synthesis Models (LVSMs) with KV-cache have recently achieved state-of-the-art quality by regressing novel views from neural networks without reconstructing 3D geometry, while Test-Time Training (TTT) layers save computation but sacrifice quality. We propose Test-Time Baking
中文介绍 本文提出测试时烘焙(TTB)方法,用于提升仅解码器视角合成模型(LVSM)的渲染效率。TTB在测试时对模型进行烘焙,避免了TTT层的质量损失,同时保持KV-cache的快速推理能力,实现了高质量与高效率的平衡。
👍 21
11/05 08:00
Self-improving embodied agents are increasingly being developed and compared on robot simulation benchmarks. Several recent works expose the benchmark's success criteria to the learning agent, a practice we call Criteria Exposure (CE). These criteria, originally written to verify policies trained by
中文介绍 本文探讨了具身自我改进智能体中标准暴露(CE)现象,即学习智能体接触到基准的成功标准。研究发现CE会错误地引导学习,使智能体学习任务而非验证器,导致泛化能力受损。作者提出方法缓解此问题。
👍 18
11/12 08:00
How multimodal language models should encode visual input remains up for debate. Most models use a pretrained vision encoder, while some recent recipes train the encoder from scratch with the language model or remove it entirely. Vision encoders bundle together three design choices: a visual prior f
中文介绍 本文解构了多模态视觉编码的三个设计选择:视觉先验、训练目标和与语言模型的耦合。通过系统实验,作者分析了各组件对多模态模型性能的影响,为视觉编码器的设计与选择提供指导。
👍 16
11/13 08:00
The release of GPT-6 Astra marks the emergence of a new paradigm in robotics: agentic robotics. Within weeks, frontier models with agentic harnesses matched or exceeded task-specific policies on many embodied benchmarks. Instead of training a dedicated policy for each family of tasks, an agentic rob
中文介绍 本文论述了GPT-6 Astra标志的智能体机器人新范式:智能体机器人。文中展示,配备智能体框架的前沿模型在多项具身基准上匹配或超过任务专用策略,无需为每类任务训练专用策略,预示着机器人系统的新时代。
👍 16
11/12 08:00
Multimodal coding agents decouple 4D world construction from visual generation, repurposing video models from world generators toward generative renderers. This division of labor raises a fundamental question: if code builds the world, can video models just render it? A reliable generative renderer
中文介绍 本文提出WorldDraft框架,探讨多模态编码智能体是否能将视频模型作为生成式渲染器,仅负责渲染由代码构建的世界。研究检验了生成式渲染器的可靠性与能力,为4D世界构建与视觉生成的解耦提供了实证。
👍 14
11/07 08:00
Self-evolving agents improve their own skills, memory, and harness code from experience without retraining the base LLM, and group-evolving agents go further by sharing these lessons across agents. Yet most such systems overlook that sharing across users and heterogeneous agents requires deciding wh
中文介绍 本文提出多粒度记忆提取方法,用于多用户、多智能体交互中的经验共享。该方法决定哪些经验应被提取与共享,以适应不同智能体与用户的异质性,提升群体进化智能体的学习效率。
👍 14
11/12 08:00
Second-order optimizers (e.g., Shampoo, SOAP) have emerged as compelling alternatives to Adam/AdamW and promise faster per-step convergence than AdamW in transformer training. However, they still remain costly in practice, and even SOAP still pays substantial optimizer-state memory and eigendecompos
中文介绍 本文提出因子化单边Adam-Moment(FOAM)优化器,通过分解Shampoo/SOAP的统计量,大幅降低优化器状态内存与特征分解开销。FOAM在Transformer训练中保持SOAP的收敛速度,同时显著提升实践可行性。
👍 10
11/13 08:00
We introduce Gamer, a generalist agent framework built for long-horizon tasks in complex interactive environments. Its end-to-end architecture combines perception, memory management, and multi-step reasoning. A shared training and deployment pipeline supports interaction with both conventional 2D gr
中文介绍 本文提出Gamer,一个通用视觉-语言-动作基础模型,专为复杂交互环境中的长时程任务设计。其端到端架构整合感知、记忆管理与多步推理,并支持在传统2D游戏及3D模拟中的训练与部署。
👍 8
11/12 08:00
It is common to train Large Language Models (LLMs) for agentic tasks with Reinforcement Learning (RL) from scalar rewards. However, this throws away the fine-grained information about the environment that the agent experienced during the trial (e.g., error logs). To utilize this rich information, re
中文介绍 本文提出经验交接强化学习(EHRL),用于递归自我改进。与传统标量奖励不同,EHRL利用智能体试错过程中的丰富信息(如错误日志)作为反馈,通过交接机制实现更高效、更具信息的强化学习。
👍 8
11/05 08:00
Large language models have made rapid progress in agentic reasoning and tool use, raising expectations for agents that can take over realistic productivity work from humans. However, current agents still struggle to finish such tasks, where success requires following long-horizon professional workfl
中文介绍 本文提出Agent-R^2框架,通过真实来源数据合成,扩展智能体在现实生产力任务中的能力。该方法利用专业工作流程数据生成训练样本,使智能体能够遵循长时程职业流程,显著提升任务成功率。
👍 8
11/04 08:00
Subliminal learning is a phenomenon in large language models (LLMs) where a teacher passes a hidden preference to a student through training data unrelated to that preference. We study whether subliminal learning also occurs in vision-language models (VLMs). We finetune a teacher VLM to prefer a spe
中文介绍 本文研究视觉语言模型中的阈下学习现象,即教师通过无关训练数据向学生传递隐藏偏好。作者通过微调教师VLM以偏好特定答案,并在与学生共享的无关数据中观察到该偏好的传递,证实了阈下学习在VLM中同样存在。
👍 7
11/12 08:00
Existing adaptive-reasoning methods shorten the explicit reasoning a model writes, or skip it by switching modes; either way the answer is still written out token by token. Rather than deciding how much to think, we explore a different paradigm, zero-or-T thinking: a single model either answers from
中文介绍 本文提出零或一思考(zero-or-T thinking)范式,混合决策模型要么直接回答,要么执行一步推理。通过学习决定何时推理,模型在保持精度的同时减少推理开销,实现效率与质量的平衡。
👍 7
11/12 08:00
While search agents demonstrate impressive capabilities in multi-step question answering, their robustness to poor-quality evidence remains under-explored. Such evidence is rare in benchmark corpora, and resulting failures are hard to reproduce. We propose DeepStress, a dynamic stress-testing framew
中文介绍 本文提出DeepStress,一个动态压力测试框架,用于评估深度搜索智能体在低质量证据下的鲁棒性。通过动态生成误导性证据,DeepStress能够复现并暴露智能体在证据不理想时的失败模式,弥补现有基准的不足。
👍 5
11/11 08:00
GPT-6 Astra can drive a robot arm directly through a fixed harness, and scores of such robot-use agents on RoboDojo have risen quickly since. Reported systems differ in harness and test subset and often change several designs at once, so it is hard to tell which choices produce the gains. We take th
中文介绍 本文对GPT-6机器人使用智能体进行受控研究,以识别不同设计与训练选择对性能的影响。通过控制变量对比,作者分析了固定harness下智能体在RoboDojo上的表现,为机器人使用智能体的设计提供实证依据。