Curating the frontier of Artificial Intelligence

FreqDepthKV: Frequency-Guided Depth Sharing for Robust KV Cache Compression in Long-Context LLM Inference

长上下文 LLM 推理越来越受到 KV 缓存, 的内存和带宽成本的限制，但积极的压缩可以消除层规范...

Long-context LLM inference is increasingly limited by the memory and bandwidth cost of KV caches, yet aggressive compression can remove the layer-spec...

Bridging Physical Reasoning and Task Generalization via Visual Action Outcome Reasoning Alignment

视觉语言模型 (VLMs) 很难在交互式物理推理, 中进行泛化，特别是在看不见的任务和环境下。两把钥匙失效了...

Vision-language models (VLMs) struggle to generalize in interactive physical reasoning, particularly under unseen tasks and environments. Two key fail...

DepthWeave-KV: Token-Adaptive Cross-Layer Residual Factorization for Long-Context KV Cache Compression

长上下文语言模型推理越来越受到存储键值缓存所需的内存带宽和容量的限制, 但现有的 co...

Long-context language model inference is increasingly limited by the memory bandwidth and capacity required to store key-value caches, yet existing co...

The Large Cancer Assistant (LCA): A Model-Agnostic Orchestration Framework for Scalable Clinical Decision Support in Oncology

- Objective: 肿瘤学中的多模态深度学习模型目前受到严格耦合数据摄取的整体设计的限制, 临床路线...

- Objective: Multimodal deep learning models in oncology are currently limited by monolithic designs that rigidly couple data ingestion, clinical rout...

Rethinking Indic AI from a Lens of Cultural Heritage Preservation

随着人工智能(AI) 进入印度次大陆, 的不同地区，人们对研究人工智能如何影响......产生了浓厚的兴趣。

As Artificial Intelligence (AI) makes inroads into different parts of the Indian subcontinent, there is significant interest in studying how AI impact...

Graph Sparse Sampling: Breaking the Curse of the Horizon in Continuous MDP Planning

在连续域的不确定性下进行规划对于自治系统, 至关重要，但计算要求较高。基于树的搜索方法，例如...

Planning under uncertainty in continuous domains is essential for autonomous systems, yet computationally demanding. Tree-based search methods such as...

SovereignPA-Bench: Evaluating User-Owned Personal Agents under Evolving Intent, Platform Mediation, and Consent Constraints

个人代理正在成为持久的用户拥有的中介: 他们记住偏好, 过滤平台介导的信息, 使用工具, 并否定...

Personal agents are becoming persistent user-owned intermediaries: they remember preferences, filter platform-mediated information, use tools, and neg...

LLM-as-a-Verifier: A General-Purpose Verification Framework

扩展训练前,、训练后, 和测试时计算已成为提高法学硕士能力的核心范例。在这项工作,中，我们...

Scaling pre-training, post-training, and test-time compute have become the central paradigms for improving the capabilities of LLMs. In this work, we ...

Jesse Thaler named director of the Laboratory for Nuclear Science

“在我自己的粒子物理领域,研究人员正在开发尖端的人工智能算法来处理对撞机实验中的海量数据并......

“In my own field of particle physics, researchers are developing cutting-edge AI algorithms to handle the data deluge from collider experiments and to...

How novice coders can develop AI programs for military applications

在当今的的世界, ChatGPT和Claude等人工智能聊天机器人可以执行许多功能,，例如撰写工作电子邮件和规划工作...

In today's world, artificial intelligence chatbots such as ChatGPT and Claude can perform many functions, such as composing work emails and planning t...