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

Robotic foundation models are typically evaluated through costly closed-loop rollouts, which motivates cheap offline proxies for model capability. The Platonic Representation Hypothesis (PRH) suggests

Yunfei Xie, Howard H. Qian, Shiyi Lan

Test-time scaling has emerged as a key inference strategy for reasoning models, where a model generates a large number of candidate solutions in the hope of finding at least one correct solution. Runn

Mahdi Salmani, Adel Javanmard, Devansh Gupta

Constructing accurate and computationally tractable approximation spaces is a fundamental task in scientific computing. Partially trainable neural networks bridge fixed random features and fully adapt

Xin He, Huan Lao, Xiliang Lu

We consider online monitoring of an AI system that produces a sequence of decisions. At each round, the system outputs a proposed decision together with a confidence score estimating the trustworthine

Isaac Gibbs, Elise L. Han, Ryan J. Tibshirani

Auditing whether a text-to-image (T2I) model was fine-tuned on a particular image collection is difficult: a model can absorb collection-level visual characteristics without reproducing training image

Sy-Tuyen Ho, Anirudh Satheesh, Soham Chawathe

Wrong automatic routing of an IT support ticket can cost more than human triage. We evaluate cost-aware abstention as an offline, cost-conditional readiness assessment: select a routing policy on cali

Sachin Gupta

Meta-reinforcement learning aims to train agents that adapt quickly to unseen tasks from limited interaction, but this requires representing task evidence over long interaction histories. Classical re

Ali Najar, Hosein Hasani, Amir Kooshan Fattah

Tool-using language model agents rely on environmental facts to decide whether an action is permitted. When a tool response omits a safety-relevant fact, an agent may carry out a prohibited action eve

Qijin Zhang, Hanzhen Wang, Yuehan Zhao

Many recent open-weight language models use n-gram embedding tables to improve performance. These tables store learned representations of short token sequences, such as pairs or triples of consecutive

Stephen Chung, Wenyu Du

Standard implementations of softmax, log-softmax, and cross-entropy have two numerical bugs. The first bug, Softmax Collapse, rounds the target logit's gradient to zero if that one logit is much large

Jonathan T. Barron, Dan Gnanapragasam, Michael Tschannen

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