Jiacheng Xu

Skywork-Reward-V2: Scaling Preference Data Curation via Human-AI Synergy

Adversarial Test Case Generation via Reinforcement Learning Extends Scaling Laws

Focus-Then-Reuse: Fast Adaptation in Visual Perturbation Environments

Policy Regularization on Globally Accessible States in Cross-Dynamics Reinforcement Learning

Incentivizing LLMs to Self-Verify Their Answers

ASOR: Anchor State Oriented Regularization for Policy Optimization under Dynamics Shift

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