PPL
Li Shen
Papers (5)
Provable Sparse Inversion and Token Relabel Enhanced One-shot Federated Learning with ViTsSimReg: Achieving Higher Performance in the Pretraining via Embedding Similarity RegularizationA Semantic-Sampling Framework for Evaluating Calibration in Open-Ended Question AnsweringRevealing Modular Gradient Noise Imbalance in LLMs: Calibrating Adam via Signal-to-Noise RatioELAS: Efficient Pre-Training of Low-Rank Large Language Models via 2:4 Activation Sparsity