PPL
Weijie Su
Papers (4)
CurveRL: Principled Distribution-Aware Context Reweighting for LLM ReasoningWhen Does Model Collapse Occur in Structured Interactive Learning?Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE RoutersUncovering Symmetry Transfer in Large Language Models via Layer-Peeled Optimization