Laguerre几何为理解大语言模型中的概念表示提供了精确的数学框架，将概念定义为区域而非单点或方向。
arXiv:2607.10578v1 Announce Type: new Abstract: Existing hypotheses represent a concept in an LLM as a single point, a linear direction, or a Gaussian…
用三角拼图法破解平移平均的鲁棒性难题，为视觉SLAM和3D重建提供全新优化思路。
arXiv:2605.07143v2 Announce Type: replace Abstract: Translation averaging aims to recover camera locations from pairwise relative translation directio…
提出GIFT方法，利用梯度几何信息实现低精度通信，在不牺牲模型精度的前提下显著降低LLM预训练通信开销。
arXiv:2607.07494v1 Announce Type: cross Abstract: Gradient communication is a primary scaling bottleneck in large language model (LLM) pretraining. Co…
从几何先验到照片级逼真，这篇Eurographics 2026 STAR综述系统梳理了构建数字人的全链路核心技术。
arXiv:2607.04341v1 Announce Type: cross Abstract: This state-of-the-art report provides an overview of controllable 3D human avatar creation. We descr…
利用激活几何进行无监督特征挖掘，无需人类标注即可揭示大模型内部表征
arXiv:2607.04222v1 Announce Type: new Abstract: Interpretability methods aim to reveal the features represented inside large language models (LLMs). M…
将几何信息融入目标跟踪模型，G²TAM在Track Anything基础上显著提升鲁棒性与精度，ICML 2026收录的新范式。
arXiv:2607.03789v1 Announce Type: new Abstract: Human spatial understanding arises from jointly perceiving geometry and semantics, enabling consistent…
脑间网络几何建模的前沿研究，揭示多人脑活动交互的结构规律
arXiv:2509.10650v4 Announce Type: replace-cross Abstract: Effective analysis in neuroscience benefits significantly from robust conceptual frameworks.…
论文揭示LLM人格评估中聚合分数忽略结构相关性的双重本质：聚合倾向与框架依赖几何
arXiv:2607.02368v1 Announce Type: cross Abstract: Evaluations of LLM personas via psychometric questionnaires typically rely on aggregate scores, disc…
让大语言模型首次理解3D CAD边界表示，打破自然语言与几何建模的壁垒。
arXiv:2512.16413v2 Announce Type: replace Abstract: Current token-sequence-based Large Language Models (LLMs) struggle to directly process 3D Boundary…
基于几何原理的随机优化方法，为大规模语言模型训练效率提供新思路
arXiv:2510.01878v2 Announce Type: replace Abstract: Low-rank gradient optimization for large language models is currently divided into two categories:…
ICML 2026录用论文，针对LLM强化学习中的rollout采样提出几何优化方法，提升模型鲁棒性。
arXiv:2606.26917v1 Announce Type: cross Abstract: Online reinforcement learning is widely used to align large language models (LLMs) with reward signa…
首次实现视频中物体平移、旋转、缩放等3D几何编辑，突破生成式视频编辑的可靠性瓶颈
arXiv:2606.24225v1 Announce Type: new Abstract: Object-level geometric edits, including translating, rotating, scaling, duplicating, or removing an ob…
让多模态大模型学会看“几何”——用视觉思维链内化推理，实现精确指向定位
arXiv:2606.24539v1 Announce Type: new Abstract: Pointing-based visual grounding requires models to precisely locate target objects by deciphering comp…
自动驾驶MLLM新突破，高效几何感知让大模型更懂空间结构
arXiv:2606.22617v1 Announce Type: new Abstract: Multimodal Large Language Models (MLLMs) have achieved remarkable performance on 2D visual tasks, yet …
LLM增强的A*算法突破几何限制，巧用语言模型引导非几何网络图的最优路径搜索
arXiv:2606.23136v1 Announce Type: cross Abstract: Finding the shortest path in non-geometric network graphs, where edge weights encode arbitrary metri…
最新arXiv论文提出GEOPHYS框架，用几何方法量化物理合理性，为AI生成内容提供可解释性评估。
arXiv:2606.20707v1 Announce Type: cross Abstract: While humans can identify physically implausible events within milliseconds, machine learning approa…
直击FP4低精度训练痛点，首次从几何角度解释收缩偏差的根源，并提出UFP4方案系统性解决该问题。
arXiv:2606.20381v1 Announce Type: new Abstract: FP4 training promises substantial reductions in memory and computation cost for LLM pretraining, yet c…
几何约束让LLM激活引导实现多语义叠加，突破单方向操控瓶颈
arXiv:2606.19946v1 Announce Type: cross Abstract: Activation steering controls model behavior by modifying intermediate hidden states at inference tim…
将热核先验引入流形变分框架，为贝叶斯深度学提供新的几何先验设计思路。
arXiv:2606.18658v1 Announce Type: new Abstract: Learning unsupervised representations of medical imaging cohorts can reveal clinically meaningful prot…
将大模型与视觉代码补全结合，专为航空航天几何设计场景优化，开辟了AI辅助工程的新路径。
arXiv:2606.16806v1 Announce Type: new Abstract: Recent advances in both Large Language Models (LLMs) and Vision Language Models (VLMs) have seen a ste…