AlignSAE is introduced, a method that aligns SAE features with a predefined ontology through a "pre-train, then post-train"curriculum and achieves disentanglement improvements of $9$--$\infty\times$ over standard SAEs, and further supports multi-hop reasoning and a mechanistic probe of grokking-like generalization dyna...
Minglai Yang, Xin-Yu Guo, Zheng-Liang Shi et al.· arXiv.org· 4 citations
CyberArena is a dynamic, symmetric adversarial benchmark built around multi-round attack-and-defense combat, and introduces a dynamical model that maps match dynamics into unified, multidimensional scores across offensive, defensive, and domain-specific capabilities.
Bangya Liu, Jianhong Tu, Yu-Jing Pan et al.· 0 citations
In-context learning (ICL) allows language models to adapt their behavior from instructions and demonstrations in a prompt, but accuracy alone does not reveal how contextual information controls their predictions. We study the acquisition and trainability of ICL with a controlled diagnostic that separates copying a near...
Minglai Yang, Xin-Yu Guo, Ren-Xiong Wang et al.· 0 citations
Long-horizon, semi-verifiable agentic tasks often rely on rubric-based evaluation, but the evidence needed to assess these tasks can be spread across the environment state and the execution trace. A judge may overlook this evidence or accept unsupported claims, including an agent's own reports of success. When rubric v...
Reza Sanayei, Razvan-Gabriel Dumitru, Sun-Jie Hou et al.· 0 citations
Environments are the foundation of recursive self-improvement: they provide the problems agents work on and the feedback used to evaluate progress. Yet constructing challenging research environments with reliable evaluation still depends on domain experts, limiting their scale and disciplinary coverage. We introduce RS...
Ren-Xiong Wang, Darvin Yi, Abril Herrlein et al.· 0 citations
Reliable Enterprise Agent Deployment (READY), a framework for qualifying AI agents for deployment on enterprise workflows, and provides a basis for comparing agent systems, setting oversight requirements, and making evidence-based deployment decisions.
Veronica Chatrath, Bryan Zhu, Jing-Xuan Fan et al.· 0 citations
This work proposes Rubric Dropout, a one-line fix borrowed from neuron dropout that randomly drops a subset of the rubric's criteria before computing the reward, so the policy never optimizes the same rubric twice.
Minglai Yang, Xinyu Guo, Utkarsh Tyagi et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
