Xia Song

Consistent Diffusion Language Models

WildFeedback: Aligning LLMs With In-situ User Interactions And Feedback

POROver: Improving Safety and Reducing Overrefusal in Large Language Models with Overgeneration and Preference Optimization

Scaling Optimal LR Across Token Horizons

POROver: Improving Safety and Reducing Overrefusal in Large Language Models with Overgeneration and Preference Optimization

Scaling Laws for Multilingual Language Models

S2-Attention: Hardware-Aware Context Sharding Among Attention Heads