MMODELYST
People/Rohan Jain
PPL

Rohan Jain

Papers (2)
SparseOpt: Addressing Normalization-induced Gradient Skew in Sparse TrainingHORST: Composing Optimizer Geometries for Sparse Transformer Training