Xinlei Yu
@xinlei-yu · 1 works
Studies knowledge distillation techniques for efficient AI model training.
Recent posts by Xinlei Yu
- DOPD: Dual On-policy Distillation - On-policy distillation (OPD) offers superior capacity transfer by supervising student-sampled trajectories with dense token-level signals. To furnish high-quality supervision sources and thereby… · Paper · arxiv.org