Ax Zifan Peng, Yule Liu, Zhen Sun, Mingchen Li, Zeren Luo, Jingyi Zheng, Wenhan Dong, Xinlei He, Xuechao Wang, Yingjie Xue, Shengmin Xu, Xinyi Huang 3/3/2026

JALMBench: Benchmarking Jailbreak Vulnerabilities in Audio Language Models

JALMBench: adversarial benchmark and dataset for evaluating jailbreak vulnerabilities in audio language models with unified evaluation framework.

Ax Hoyeon Chang, Jinho Park, Hanseul Cho, Sohee Yang, Miyoung Ko, Hyeonbin Hwang, Seungpil Won, Dohaeng Lee, Youbin Ahn, Minjoon Seo 3/3/2026

Characterizing Pattern Matching and Its Limits on Compositional Task Structures

Research characterizing how LLMs rely on pattern-matching for compositional tasks and analyzing OOD generalization failures through behavioral studies with controlled task setups.

Ax Alessio Russo, Ryan Welch, Aldo Pacchiano 3/3/2026

In-Context Learning for Pure Exploration

Framework for pure exploration using in-context learning, enabling efficient adaptive data collection for hypothesis testing and best-arm identification tasks.

Ax Chongyi Zheng, Seohong Park, Sergey Levine, Benjamin Eysenbach 3/3/2026

Intention-Conditioned Flow Occupancy Models

Foundation model approach for RL using pre-trained occupancy flow models for intention-conditioned planning, enabling transfer learning in reinforcement learning.

Ax Pavlo Vasylenko, Hugo Pitorro, Andr\'e F. T. Martins, Marcos Treviso 3/3/2026

Long-Context Generalization with Sparse Attention

Sparse attention mechanism for transformers to improve long-context generalization by preventing attention dispersion on non-informative tokens in extended sequences.

Ax Mohammad Hossein Amani, Aryo Lotfi, Nicolas Mario Baldwin, Samy Bengio, Mehrdad Farajtabar, Emmanuel Abbe, Robert West 3/3/2026

RL for Reasoning by Adaptively Revealing Rationales

Framework combining partial supervision with RL for sequence generation tasks, reducing reliance on expert demonstrations while addressing sparse reward challenges in learning.

Ax Yikai Wang, Zhouxia Wang, Zhonghua Wu, Qingyi Tao, Kang Liao, Chen Change Loy 3/3/2026

Next Visual Granularity Generation

Proposes Next Visual Granularity (NVG) framework that generates images through structured sequences with varying token granularity levels.