通过选择性几何控制，提升大模型安全对齐的鲁棒性，为AI防御攻击提供新思路。
arXiv:2602.07340v2 Announce Type: replace Abstract: Safety alignment of large language models remains brittle under domain shift and noisy preference …
OpenAI数学家借助AI破解80年未解几何难题，震惊学界
Nature, Published online: 22 May 2026; doi:10.1038/d41586-026-01651-0 The late Hungarian mathematician Paul Erdős thought he had the last word on a ge…
AI模型破解80年未解的离散几何猜想，数学难题再被AI攻克
An OpenAI model solved the 80-year-old unit distance problem, disproving a major conjecture in discrete geometry and marking a milestone in AI-driven …
OpenAI模型推翻离散几何核心猜想，AI在科学发现中展现惊人潜力
OpenAI Model Disproves Central Conjecture in Discrete Geometry Meta Description: An OpenAI model has disproved a central conjecture in discrete geomet…
AI模型推翻数学猜想，证明离散几何核心猜想错误，引发学界热议。
https://x.com/wtgowers/status/2057175727271800912 , https://xcancel.com/wtgowers/status/2057175727271800912 Comments URL: https://news.ycombinator.com…
AI首次挑战80年几何猜想，数学探索迎来新突破。
IT之家 5 月 21 日消息，OpenAI 称其全新推理模型推导出了一个原创数学证明，推翻了几何学中一道著名的未解猜想。该猜想最早由保罗・埃尔德什于 1946 年提出。 IT之家注意到，OpenAI 已不是第一次放出这般大胆的言论。七个月前，这家人工智能巨头前副总裁凯文・韦尔在社交平台 X 上发文…
用生成式网格遍历技术自动切割接缝，MeshTailor让3D模型处理更智能、更高效。
arXiv:2603.27309v2 Announce Type: replace-cross Abstract: We present MeshTailor, the first mesh-native generative framework for synthesizing edge-alig…
置信度几何方法揭示大语言模型推理中痕迹级正确性的全新框架，附开源代码。
arXiv:2605.16824v1 Announce Type: new Abstract: Large language models (LLMs) generate not only reasoning text, but also token-level confidence traject…
教你利用激活几何诊断转向瓶颈，GRACE框架让概念工程优化更高效。
arXiv:2605.16362v1 Announce Type: new Abstract: Activation steering offers a lightweight way to control LLMs without retraining, but its effectiveness…
从贝叶斯几何视角重新阐释Transformer注意力机制，揭示其内在概率结构。
arXiv:2512.22471v5 Announce Type: replace Abstract: Transformers often appear to perform Bayesian reasoning in context, but verifying this rigorously …
一篇关于几何视图扩散模型的论文，提出景观感知方法提升3D场景重建质量，创新融合空间几何与扩散先验。
arXiv:2605.19865v1 Announce Type: new Abstract: Accurate camera viewpoint estimation under sparse-view conditions remains challenging, particularly in…
被ICML 2026接收，提出在全秩相关矩阵上构建黎曼网络的新几何深度学习方法。
arXiv:2605.19073v1 Announce Type: new Abstract: Representations on the Symmetric Positive Definite (SPD) manifold have garnered significant attention …
新方法Controlla通过图约束潜在几何实现可控性学习，为生成模型中的控制能力提供理论框架。
arXiv:2605.16603v1 Announce Type: new Abstract: Controllable multimodal generation is commonly formulated as an inference-time conditioning problem us…
几何感知的4D视频生成技术，让机器人的操作规划更精准——融合时间一致性与物理动态建模，提升复杂环境交互能力。
arXiv:2507.01099v4 Announce Type: replace-cross Abstract: Understanding and predicting dynamics of the physical world can enhance a robot's ability to…
揭秘大模型内部贝叶斯推断的几何结构，从小模型到生产级LLM的规模扩展规律
arXiv:2512.23752v5 Announce Type: replace Abstract: Recent work has shown that small transformers trained in controlled "wind-tunnel'' settings can im…
提出基于几何感知后验转移的持续多智能体拓扑学习方法，突破灾难性遗忘难题。
arXiv:2605.17361v1 Announce Type: new Abstract: Multi-agent systems (MAS) powered by large language models (LLMs) have emerged as a powerful paradigm …
提出PDI-Bench框架，量化评估生成视频模型的几何一致性，攻克3D结构合理性难题。
arXiv:2605.15185v1 Announce Type: cross Abstract: Generative video models are increasingly studied as implicit world models, yet evaluating whether th…
几何引导的隐藏状态替换，揭秘扩散模型在语言模型中的最佳插入位置，DiHAL创新方案提升性能。
arXiv:2605.14368v1 Announce Type: cross Abstract: Continuous diffusion language models lag behind autoregressive transformers, partly because diffusio…
用矩阵空间抽象局部转移几何，强化学习实现组合泛化新范式。
arXiv:2605.14304v1 Announce Type: cross Abstract: Compositional generalization in sequential decision-making requires identifying which parts of prior…
3D多边形形状的紧凑表示新方法，为高效存储与处理提供新思路。
arXiv:2605.16266v1 Announce Type: cross Abstract: We introduce Patchwork, a new general-purpose shape representation capable of modeling 2D and 3D geo…