1
从LLM到JEPA，中国团队正在把“世界模型”搬进细胞内部
把“世界模型”搬进细胞内部，中国团队用LLM-JEPA架构打造AI虚拟细胞，有望终结新药研发“双十定律”。
四年前LeCun提出JEPA时，大概不会想到，这套架构有一天会被用来理解：一颗细胞。
2
The Missing Knowledge Layer in Cognitive Architectures for AI Agents
指出主流AI认知架构缺失知识层，导致事实与经验混淆，引发认知错误
arXiv:2604.11364v2 Announce Type: replace Abstract: The two most influential cognitive architecture frameworks for AI agents, CoALA [21] and JEPA [12]…
3
V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning
视频自监督学习新突破，V-JEPA 2.1巧妙解锁密集特征，大幅提升视频理解性能。
arXiv:2603.14482v3 Announce Type: replace Abstract: We present V-JEPA 2.1, a family of self-supervised models that learn dense, high-quality visual re…
4
LatentWave: JEPA Pretraining for Wireless Foundation Models
JEPA预训练方法应用于无线通信领域，探索基础模型的新范式
arXiv:2606.06373v1 Announce Type: cross Abstract: Wireless foundation models have emerged as a promising alternative to building separate models for e…
5
VISReg: Variance-Invariance-Sketching Regularization for JEPA training
一文解析VICReg的改进变体VISReg，用方差-不变性-草图正则化防止嵌入坍塌，提升JEPA训练效果
arXiv:2606.02572v1 Announce Type: new Abstract: Self-supervised learning methods prevent embedding collapse via modeling heuristics or explicit regula…
6
DLLM-JEPA: Joint Embedding Predictive Architectures for Masked Diffusion Language Models
结合联合嵌入预测与掩码扩散，提出全新语言模型预训练架构
arXiv:2606.00091v1 Announce Type: cross Abstract: Joint Embedding Predictive Architectures (JEPAs) have reshaped self-supervised representation learni…
7
华为具身大脑一号位做类脑智能世界模型，对标JEPA，获亿元级融资｜硬氪首发
华为具身大脑关键人物朱森华以类脑智能范式打造认知世界模型，获亿元融资，对标Yann LeCun的JEPA路径。
作者｜黄楠 编辑｜袁斯来 硬氪获悉，具身智能大脑公司「具脑磐石」完成新一轮亿元级融资，本轮融资由具备深厚类脑与具身产业背景的顶尖产业资本领投，老股东及多家顶尖基金复投和跟投，多维资本担任独家财务顾问。同时，更新一轮融资也在同步交割中。 资金将重点投入核心技术研发、人才团队扩容及全球化市场拓展，以加速…
8
Representation Without Reward: A JEPA Audit for LLM Fine-Tuning
提出用JEPAs审计LLM微调：预测隐含表示而非输出，以提升任务指标。
arXiv:2605.15394v1 Announce Type: cross Abstract: Joint-embedding predictive architectures (JEPAs) propose that a model should learn more useful abstr…