PPL
Trung Le
Papers (5)
Spectral Flattening Is All Muon Needs: How Orthogonalization Controls Learning Rate and ConvergenceBSO: Safety Alignment Is Density Ratio MatchingTokenRatio: Principled Token-Level Preference Optimization via Ratio MatchingSelective Off-Policy Reference Tuning with Plan GuidanceLLM-XTM: Enhancing Cross-Lingual Topic Models with Large Language Models