技术雷达TECH RADAR · GITHUB · ARXIV · HACKER NEWS

As we move through the world and carry out everyday tasks, we encounter objects that may become relevant only later. We are capable of recalling where we left something or what was inside a container,

Shravan Chaudhari, William Paul, Suchi Saria

Modern language models undergo reinforcement learning with verifiable rewards (RLVR) on top of already-trained checkpoints. A key promise of RLVR is the discovery of new reasoning strategies. In princ

Saif Punjwani, Micah Goldblum

Real-time robot control demands enough visual history to infer motion and task progress, but processing that history can delay action. We present Long-WAM, a model-system framework for scaling the con

Wei Huang, Bohan Zhang, Chenzhi Liu

Heavy-tailed noise has been widely observed in modern machine learning, motivating the use of methods like gradient clipping and normalization. While these methods are well understood in centralized s

Aleksandar Armacki, Haoyuan Cai, Ali H. Sayed

Vision-language-action models (VLAs) are strikingly sensitive to instruction phrasing and do not inherit the language robustness of the vision-language models they are built on. A one-word edit can mo

Mikey Watts, Yuchen Cui

GNN-to-MLP distillation aims to retain the predictive accuracy of a message-passing teacher while deploying a graph-free MLP at inference. Existing methods mainly transfer node-wise predictions or use

Zhewei Chen, Hao Zhu, Jiaojiao Jiang

Machine unlearning asks for a deletion algorithm whose output is close to retraining from scratch without the selected forget examples. In this work, we study forget-only unlearning, where the deletio

Luka Radić, Vikrant Singhal, Amartya Sanyal

Latent world models have shown a remarkable ability to predict future states and to plan in the real world. In practice, however, we lack a principled way to estimate how their capabilities scale with

Artem Zholus, Nicolas Beltran-Velez, Jianhao Yuan

Language-model agents are increasingly asked to carry out open-ended scientific research, yet their results are usually graded against a known answer, a rubric, or a language-model reviewer, none of w

Yinling Zhang, Langchen Liu, Dongbin Xiu

Large language models are increasingly applied to tasks grounded in long, heterogeneous information sources. Conventional Retrieval-Augmented Generation (RAG) relies on fixed similarity-based retrieva

Yilun Hao, Krishna Sayana, Isabella Ye

原文 · 讨论

原文 · 讨论

原文 · 讨论

原文 · 讨论

原文 · 讨论

原文 · 讨论

原文 · 讨论

原文 · 讨论

原文 · 讨论

原文 · 讨论