Zhilong Zhang

ADM-v2: Pursuing Full-Horizon Roll-out in Dynamics Models for Offline Policy Learning and Evaluation

Hierarchical Value-Decomposed Offline Reinforcement Learning for Whole-Body Control

On the Representation Degradation in Vision-Language-Action Models

Provably Efficient Policy-Reward Co-Pretraining for Adversarial Imitation Learning

Planning with Unified Multimodal Models

Any-step Dynamics Model Improves Future Predictions for Online and Offline Reinforcement Learning

Semantic Temporal Abstraction via Vision-Language Model Guidance for Efficient Reinforcement Learning

Improving Reward Model Generalization from Adversarial Process Enhanced Preferences

Diffusion-Guided Safe Policy Optimization From Cost-Label-Free Offline Dataset

Whale-X: Learning Scalable Embodied World Models with Enhanced Generalizability