12 papers
WorldCycle: Self-Verifiable Reinforcement Learning for Long-Horizon Video World Models
Bohai Gu, Yueyang Yuan, Taiyi Wu +9
Interactive video world models are essential for long-horizon planning and exploration, yet they suffer from compounding errors. Post-training methods such as reinforcement learnin…
Decomposing and Measuring Evaluation Awareness
Changling Li, Terry Jingchen Zhang, Jie Zhang +4
Frontier language models sometimes recognize that they are under evaluation and adjust their behavior which can undermine validity of benchmark results. Yet the field studies it wi…
WorldCraft: From Camera Navigation to Object Manipulation in Interactive Video World Models
Bohai Gu, Taiyi Wu, Yueyang Yuan +9
Recent video world models enable interactive camera navigation but provide limited control over individual objects. We study persistent object-state control in autoregressive video…
Reasoning-Trace Collapse: Evaluating the Loss of Explicit Reasoning During Fine-Tuning
Lukas Twist, Helen Yannakoudakis, Jie M. Zhang
Explicit reasoning models are trained to produce intermediate reasoning traces before final answers, but downstream fine-tuning is often performed on ordinary instruction--response…
PRISM: A Benchmark for Programmatic Spatial-Temporal Reasoning
Qiran Zhang, Yuheng Wang, Runde Yang +9
Programmatic video generation through code offers geometric precision and temporal coherence beyond pixel-level diffusion models, yet rigorously evaluating whether language models…
Beyond Execution: Static-Analysis Rewards and Hint-Conditioned Diffusion RL for Code Generation
Shuyin Ouyang, Zhaozhi Qian, Faroq AL-Tam +2
Reinforcement Learning (RL) is an important paradigm for aligning Diffusion Language Models (DLMs) toward functional correctness in code generation. However, these models often enc…
EchoDistill: Robust Large Audio Language Models via Noisy-to-Clean Self-Distillation
Liang Lin, Chunxi Luo, Kaiwen Luo +12
Large Audio Language Models (LALMs) remain vulnerable to acoustic noise, which can obscure task-relevant evidence and produce unreliable responses. We propose EchoDistill, a noisy-…
TTVS: Boosting Self-Exploring Reinforcement Learning via Test-time Variational Synthesis
Sikai Bai, Haoxi Li, Jie Zhang +2
Despite significant advances in Large Reasoning Models (LRMs) driven by reinforcement learning with verifiable rewards (RLVR), this paradigm is fundamentally limited in specialized…
HALO: A Unified Vision-Language-Action Model for Embodied Multimodal Chain-of-Thought Reasoning
Quanxin Shou, Fangqi Zhu, Shawn Chen +9
Vision-Language-Action (VLA) models have shown strong performance in robotic manipulation, but often struggle in long-horizon or out-of-distribution scenarios due to the lack of ex…
DiEP: Adaptive Mixture-of-Experts Compression through Differentiable Expert Pruning
Sikai Bai, Haoxi Li, Jie Zhang +2
Despite the significant breakthrough of Mixture-of-Experts (MoE), the increasing scale of these MoE models presents huge memory and storage challenges. Existing MoE pruning methods…
CoRE: Enhancing Metacognition with Label-free Self-evaluation in LRMs
Haoxi Li, Sikai Bai, Jie Zhang +1
Large reasoning models (LRMs) have demonstrated impressive capabilities in domains like mathematics and program synthesis. Despite their strong performance, LRMs often exhibit over…
Think How to Think: Mitigating Overthinking with Autonomous Difficulty Cognition in Large Reasoning Models
Yongjiang Liu, Haoxi Li, Xiaosong Ma +2
Recent Large Reasoning Models (LRMs) excel at complex reasoning tasks but often suffer from overthinking, generating overly long and redundant reasoning trajectories. To explore it…