PPL
Dawei Li
Papers (4)
EMA-Nesterov: Stabilizing Nesterov's Lookahead for Accelerated Deep Learning OptimizationRevisiting the Adam-SGD Gap in LLM Pre-Training: The Role of Large Effective Learning RatesTo See is Not to Learn: Protecting Multimodal Data from Unauthorized Fine-Tuning of Large Vision-Language ModelCornerstones or Stumbling Blocks? Deciphering the Rock Tokens in On-Policy Distillation