👍 374
09/07 20:00
Recursive self-improvement (RSI) requires a concrete mechanism through which an AI system observes its capabilities and converts that evidence into the next round of learning. We present NeoHorse-1, a family of agent-native models developed to explore this path through agentic post-training. Our sys
中文介绍 NeoHorse-1 提出了递归自我改进(RSI)机制,允许 AI 系统观察其能力并将其转化为下一轮学习。该模型采用代理后训练的方法,为实现代理自主学习提供了新的思路。实验表明,NeoHorse-1 在学习效率和自主改进方面具有显著优势。意义:有助于推动 AI 系统在自我学习和适应性方面的发展。
👍 112
09/07 20:00
In this work, we present Gander, an end-to-end model that unifies omni perception, realtime interaction, and agentic capabilities within a single framework. In contrast to turn-based conventional paradigms, Gander continuously receives streaming inputs across multiple modalities, including video, sp
中文介绍 本文介绍了 Gander,一个整合全感知、实时交互和代理能力的端到端模型。不同于传统的轮转交互方法,Gander 可以持续接收多模态流媒体输入,提升交互效果。Gander 的设计可以推动多模态理解和实时响应的进展。意义:对智能助手和交互式系统的应用具有重要影响。
👍 50
09/04 20:00
Most end-to-end autonomous-driving systems learn by imitating human driving logs, leaving their learned behavior constrained by the quality and behavioral coverage of the recorded trajectories. This report presents DriveZero, an end-to-end system that learns driving behavior beyond human demonstrati
中文介绍 DriveZero 提出了一种端到端的自主驾驶系统,能够学习超越人类示范的驾驶行为。该系统通过强化学习,通过多样化的环境和情境训练,提升驾驶智能。实验结果显示,该系统在复杂场景中的表现优于传统模仿学习方法。意义:推动自动驾驶技术的发展,提升汽车智能化水平。
👍 49
09/06 20:00
World-Action Models inherit world knowledge from video-generative priors, and channel it into executable control signals through embodied experience. Existing systems, however, are monolithic: the generative backbone, visual representation, architecture, information flow, inference procedure, and tr
中文介绍 OpenWAM 探索了开放和模块化的系统,旨在系统化地进行世界-动作模型的预训练。通过解耦生成主干和控制信号的生成,该方法提高了模型的可扩展性和灵活性。此研究为智能体在模拟环境中的机器人控制提供了新思路。意义:对人机协作和机器人智能领域具有重要的参考价值。
👍 46
09/03 20:00
World-action models (WAM) predict future states to guide robot actions, enabling learning from both action-free video and action-labeled interaction. Most inherit pretrained video generators, leaving WAM pretraining and scaling underexplored. We introduce Genie Envisioner Act 2.0 (GE-Act 2.0), a wor
中文介绍 GE-Act 2.0 提出了一种改进的世界-动作模型,用于机器人操控的预训练和扩展。该模型结合无动作视频学习和标记交互,提升了机器人的动作预测能力。实验结果表明,在评估任务中表现显著优于传统方法。意义:为机器人智能提升以及物体操作任务的研究提供了新的工具和基线。
👍 44
09/07 20:00
We present Miles v0.1, a full-stack, production-ready system for frontier post-training. Building upon the clean design of slime, Miles designs each stage of the reinforcement-learning (RL) training loop around a single principle: components should be verified, clean, and customizable. With accuracy
中文介绍 Miles v0.1 是一个针对前沿后训练的全栈生产就绪系统。通过设计每个强化学习训练循环的阶段,Miles 在提高准确率的同时,实现了组件的可验证性和可定制性。实验显示,该系统在多种场景下都表现出色。意义:为自主学习和智能体应用提供了更可靠的后训练框架。
👍 43
09/07 20:00
Autoregressive (AR) video diffusion models have shown great potential in real-time video generation. Recent methods distill pretrained bidirectional video diffusion models into causal AR students through Distribution Matching Distillation (DMD), but the generated videos often suffer from over-satura
中文介绍 本研究提出 Mask Forcing 方法,通过双噪声掩膜滚动,提高自回归视频扩散蒸馏的效率。该方法解决了现有模型生成视频时出现的过饱和问题,显著提升了视频生成质量。实验结果显示,在多项生成任务中均取得了更好的效果。意义:对实时视频生成和多媒体创作领域有重要的应用前景。
👍 41
09/03 20:00
Diverse and simulation-ready indoor scenes are essential for interactive entertainment and embodied AI, yet their scalable generation remains challenging. Recent agentic text-to-3D scene pipelines that rely on vision-language models (VLMs) can generate scenes of high fidelity but require costly iter
中文介绍 SceneMosaic 提出了通过混合代理布局进化的方法,以高效多样化地生成适合模拟的室内场景。该方法解决了传统模型对室内场景生成的局限,为交互娱乐和智能体提供了可扩展的解决方案。实验结果表明,其生成效果在视觉和交互性上都优于现有系统。意义:推动室内场景生成和虚拟现实的进步。
👍 32
09/03 20:00
Large Reasoning Models (LRMs) achieve superior problem-solving through extended Chain-of-Thought (CoT) generation, but the resulting key-value (KV) cache grows linearly with sequence length and creates severe memory bottlenecks, often exceeding GPU capacity for long reasoning traces. Existing KV cac
中文介绍 BeaconKV 通过引入信标查询指导的键值缓存压缩方法,解决了大型推理模型推理时的内存瓶颈问题。该方法显著减少了因序列长度增长导致的内存消耗,提高了模型在长推理链上的表现。实验结果显示,其在推理效率上比传统方法提升了量化效果。意义:为大型推理模型的优化提供了新方向,尤其是长序列任务。
👍 30
09/05 20:00
Latent visual reasoning aims to perform multimodal reasoning through hidden-state computation rather than explicit textual chains of thought. However, visual information being present in a latent state does not imply that the model actually relies on that state when producing its answer, especially
中文介绍 本研究探讨了潜在视觉推理的必要性,重点在于通过隐藏状态计算进行多模态推理。研究发现,模型在生成答案时并不总是依赖于潜在状态,表明当前系统在信息使用上的不足。此研究为提高模型的推理效果提供了新的思路。意义:对多模态学习和视觉理解的进展具有重要意义。
👍 27
09/04 20:00
As large language models (LLMs) are increasingly deployed in alignment-sensitive contexts, activation steering has emerged as a lightweight, inference-time alternative to fine-tuning methods (e.g., RLHF, DPO) for behavioral control. However, existing work typically validates steering on isolated beh
中文介绍 Steering Geometry 提出了在人类价值几何中验证大型语言模型(LLM)表现的一种方法。在对齐敏感的上下文中,激活引导作为一种轻量级的推理方法,是对 RLHF 等方法的有效替代。实验结果展示了该方法在多行为控制任务上的优势。意义:对 LLM 行为控制和模型对齐研究具有重要启示。
👍 26
09/06 20:00
Linear attention is increasingly used in frontier language models for efficient long-context inference and constant-memory decoding. Its fixed-size recurrent memory, however, requires an online decision at each token: what to write and how strongly to overwrite existing associations before knowing w
中文介绍 Kalman Delta Networks 提出了基于不确定性的关联记忆方法,旨在提升线性注意力在长上下文推理中的效率。该方法通过修正固定大小记忆的写入决策,提高了模型在不确定性环境中的表现。实验表明,该方法在处理复杂输入时优于传统模型。意义:为长文本处理和记忆模型的研究提供了重要参考。
👍 25
09/06 20:00
Transferring human hand demonstrations to robotic grippers has recently emerged as a cost-effective solution for robot learning. However, existing methods are largely confined to simple, planar tasks and fail to handle complex spatial movements (e.g., intricate trajectories involving rotations or fl
中文介绍 本研究开发了 CosmoH2G 数据集及基线方法,旨在将人类手部演示转移到机器人夹持器上。该方法拓展了机器人学习任务的复杂度,使其能够处理更为复杂的空间动作。实验结果显示其在多任务学习上表现出色。意义:推动机器人在复杂环境下操作的研究和应用。
👍 25
09/06 20:00
Speculative decoding accelerates rollout generation, which dominates the cost of reinforcement learning (RL) post-training. Online co-training can further increase the draft's accuracy, yielding greater speedups. However, scaling this approach to co-training on large models with long contexts poses
中文介绍 Online Draft Co-Training 提出了在大规模长上下文强化学习后训练中使用的预测解码加速方法。该方法通过在线协同训练提高草稿的准确性,显著提升了生成速度。研究显示,在复杂任务中的应用有效减小了训练时间。意义:对强化学习及其应用的效率提升具有重要价值。
👍 24
09/05 20:00
Visual generation is evolving from generative models used through a single invocation into agentic control processes that can plan, select tools, inspect intermediate synthesized outputs, revise failures, and reuse prior experience. In most existing systems, the controller is an LLM or VLM, while vi
中文介绍 Agentic Visual Generation 转变了视觉生成的方式,将其从单次调用的生成模型演化为可以进行代理控制的过程。该方法允许模型规划、选择工具和修正输出,有效提升了生成质量与灵活性。实验结果佐证了其广泛的应用潜力。意义:推动生成模型向智能体化发展的趋势。
👍 22
09/07 20:00
Large language models are increasingly deployed as agents that plan over long horizons and act through external tools. Most agents select actions through unconstrained generation over an accumulating history, leaving implicit the procedural knowledge of what to do, in what order, and under which con
中文介绍 Procedural Graphs 提出了一种自我演化的执行结构,旨在为大型语言模型(LLM)代理提供更好的规划能力。该方法通过明确程序知识提升代理的有效决策能力,实验验证了其在复杂任务上的应用效果。意义:对智能体规划和执行的研究提供了重要参考。
👍 21
09/07 20:00
Large Language Models demonstrate remarkable proficiency in static reasoning, yet training them as autonomous agents through Reinforcement Learning (RL) for long-horizon tasks is often hindered by severe reward sparsity. While conventional agent-side warming up via supervised fine-tuning (SFT) can a
中文介绍 本研究强调环境作为支架的重要性,通过增强反馈来引导自我演化的智能体在长时间任务中的学习。研究提出的强化学习方法在稀疏奖励下表现良好,解决了传统训练中的回报稀疏问题。实验结果证实了其提高训练效率的有效性。意义:对智能体长期任务学习的有效性提升至关重要。
👍 20
09/03 20:00
Vision-Language-Action (VLA) models have shown promising progress in language-conditioned robotic manipulation. However, existing datasets and benchmarks mainly evaluate task completion under predefined settings, offering limited insight into model reasoning under increasing spatial and procedural c
中文介绍 RoboSPA 研究了视觉-语言-动作(VLA)模型在复杂场景和长时间任务中的表现。通过更复杂的数据集和基准,模型能够更全面地展示其推理能力。实验结果显示,模型在处理高复杂性场景时表现出色。意义:为语言条件下的机器人操作开辟了新方向。
👍 20
09/07 20:00
Representing a 3D scene as multi-view images allows 2D VLMs to reason in 3D by reusing priors from pre-training, sidestepping the scarcity of annotated 3D data. However, it produces thousands of redundant visual tokens whose cost grows with every view. Existing visual token pruners fall into two fam
中文介绍 CoVeR 提出了基于覆盖率的令牌修剪方法,以支持 VLM 的多视角三维推理。该方法通过消除冗余视觉令牌,显著提高了处理效率。实验结果表明,这一方法在多视角任务中的表现显著优于现有技术。意义:推动 VLM 在三维理解任务中的应用深度。
👍 18
09/05 20:00
Diffusion-based models enable monocular geometry estimation, yet their pixel-space precision is limited by a shared, under-studied error source: VAE reconstruction degradation. The 8x spatial compression in the VAE encoder-decoder degrades surface normals at object boundaries; even encoding and deco
中文介绍 TransNormal-2 通过边缘感知解码与几何约束流的结合,提升了单眼几何估计的精确性。该模型针对现有方法在像素精度上的不足,提出了有效的解决方案。实验结果显示,其在表面法线估计上表现显著提高。意义:为单目视觉任务的精度提升提供了有力的工具。