12 papers
Learning Where Outcomes Change:Credit-Addressable Reasoning for Multimodal Geometry
Jiani Guo, Junjie Wang, Jie Wu +5
Multimodal geometry reasoning requires VLMs to extract precise visual relations and preserve them through multi-step deduction. Existing free-form traces obscure the decisions that…
Route by Kinematics, Act by Observation: Kinematics-Supervised Expert Routing in MoE-Augmented VLA
Tianhang Yang, Yanze Zheng, Junjie Wang +3
While MoE augments VLA via expert specialization, router suffers from ineffective expert routing owing to the kinematic heterogeneity of actions across manipulation tasks and, even…
Simple-OPD: Demystifying Warm-up for On-policy Distillation
Tao Liu, Taiqiang Wu, Mao Zheng +5
On-policy distillation (OPD) trains a student on its own rollouts with token-level supervision from teacher models, but its effectiveness can depend strongly on the warm-up stage b…
Decoupled Residual Quantization for Robust Semantic IDs in Recommendation
Xuesi Wang, Junjie Wang, Ziliang Wang +2
Semantic IDs represent items as shared discrete token sequences and have become a practical tool for recommendation and retrieval. Yet it remains difficult to tell why a tokenizer…
Internalize the Temperature: On-Policy Self-Distillation as Policy Reheater for Reinforcement Learning
Xuewei Yang, Jiachen Yu, Jie Wu +3
Reinforcement learning from verifiable rewards improves the reasoning ability of large language models, but often suffers from entropy collapse, in which increasingly concentrated…
Think-with-Rubrics: From External Evaluator to Internal Reasoning Guidance
Jiachen Yu, Zhihao Xu, Junjie Wang +1
Rubrics have been extensively utilized for evaluating unverifiable, open-ended tasks, with recent research incorporating them into reward systems for reinforcement learning. Howeve…
OrchJail: Jailbreaking Tool-Calling Text-to-Image Agents by Orchestration-Guided Fuzzing
Jianming Chen, Yawen Wang, Junjie Wang +3
Tool-calling text-to-image (T2I) agents can plan and execute multi-step tool chains to accomplish complex generation and editing queries. However, this capability introduces a new…
Velocity-Space 3D Asset Editing
Hao Liu, Yuxuan Lin, Jingfeng Guo +4
Local 3D asset editing requires modifying a target region while preserving the rest. Existing methods enforce locality through mechanisms external to the generator, such as manual…
Mitigating the Reasoning Tax in Vision-Language Fine-Tuning with Input-Adaptive Depth Aggregation
Yiming Ren, Yujiu Yang, Junjie Wang
Supervised fine-tuning (SFT) on visual instruction data often improves perceptual capabilities in vision-language models (VLMs) while degrading reasoning performance, creating a pe…
A Survey of Reasoning in Autonomous Driving Systems: Open Challenges and Emerging Paradigms
Kejin Yu, Yuhan Sun, Taiqiang Wu +5
The development of high-level autonomous driving (AD) is shifting from perception-centric limitations to a more fundamental bottleneck, namely, a deficit in robust and generalizabl…
ProFit: Leveraging High-Value Signals in SFT via Probability-Guided Token Selection
Tao Liu, Taiqiang Wu, Runming Yang +3
Supervised fine-tuning (SFT) is a fundamental post-training strategy to align Large Language Models (LLMs) with human intent. However, traditional SFT often ignores the one-to-many…
Nav- Dual-Relation Reasoning for Generalizable Open-Vocabulary Object-Goal Navigation
Wentao Xiang, Haokang Zhang, Tianhang Yang +10
Object-goal navigation in open-vocabulary settings requires agents to locate novel objects in unseen environments, yet existing approaches suffer from opaque decision-making proces…