Monarch: Expressive Structured Matrices for Efficient and Accurate TrainingLarge neural networks excel in many domains, but they are expensive to train and fine-tune. A popular approach to reduce their compute or memory requirements is to replace dense weight matrices...https://arxiv.org/abs/2204.00595arxiv.orghttps://arxiv.org/pdf/2204.00595.pdf
Seonglae ChoSeonglae ChoSeonglae ChoSeonglae Cho
