16 papers
ES-VP : Energy-Shaped Dynamic Visual Prompting for Efficient Model Adaptation
Can Jin, Ying Li, Jingchen Sun +5
Visual prompting (VP) has emerged as a parameter-efficient method for adapting pre-trained models to downstream tasks. However, existing approaches encounter a trade-off between fl…
Latent Reward Steering: An Adaptive Inference-Time Framework that Implicitly Promotes Cognitive Behaviors in Reasoning LLMs
Jiakang Li, Guanyu Zhu, Can Jin +8
Strong reasoning depends not only on model knowledge but also on how effectively cognitive behaviors are deployed during generation. Existing methods often rely on explicit behavio…
Evidence Over Plans: Online Trajectory Verification for Skill Distillation
Yang Zhou, Zihan Dong, Zhenting Wang +7
Agent skills can remarkably improve task success rates by using human-written procedural documents, but their quality is difficult to assess without environment-grounded verificati…
DTop-p MoE: Sparsity-Controlled Dynamic Top-p MoE for Foundation Model Pre-training
Can Jin, Hongwu Peng, Mingcan Xiang +7
Sparse Mixture-of-Experts architectures are essential for scaling model capacity efficiently, yet the standard Top- routing imposes a rigid sparsity pattern that ignores the int…
Weak Critics Make Strong Learners: On-Policy Critique Distillation for Scalable Oversight
Can Jin, Jiakang Li, Rui Wu +3
As large language models become stronger, weak supervisors may fail to provide reliable labels, preferences, or final judgments for complex outputs, limiting both weak-to-strong ge…
DARE: Difficulty-Adaptive Reinforcement Learning with Co-Evolved Difficulty Estimation
Yang Zhou, Can Jin, Zihan Dong +7
Reinforcement learning improves the reasoning ability of large language models but remains costly and sample-inefficient, as many rollouts provide weak learning signals. Difficulty…