Xin Liu

9 papers Peng Cheng Laboratory

• DAPO: An Open-Source LLM Reinforcement Learning System at Scale
• DUO: No Compromise to Accuracy Degradation
• Enhancing Safety in Reinforcement Learning with Human Feedback via Rectified Policy Optimization
• Learn and Ensemble Bridge Adapters for Multi-domain Task Incremental Learning
• RADAR: Benchmarking Language Models on Imperfect Tabular Data
• SensorLM: Learning the Language of Wearable Sensors
• SpecEM: Training-Free LLM Ensembling via Iterative Drafting, Verification, and Online Feedback
• Think-RM: Enabling Long-Horizon Reasoning in Generative Reward Models
• Videos are Sample-Efficient Supervisions: Behavior Cloning from Videos via Latent Representations