Tom Jacobs

International Conference on Machine Learning Workshop(ICML- W)
HORST: Composing Optimizer Geometries for Sparse Transformer Training

International Conference on Machine Learning workshop (ICML-W)
Continuous Sparsification via Minimizing Movement

International Conference on Machine Learning (ICML)
SparseOpt: Addressing Normalization-induced Gradient Skew in Sparse Training

International Conference on Learning Representations (ICLR)

International Conference on Learning Representations (ICLR)
Never Saddle Down for Reparameterized Steepest Descent as Mirror Flow

International Conference on Learning Representations (ICLR)
Hyperbolic Aware Minimization: Implicit Bias for Sparsity

Conference on Neural Information Processing Systems (NeurIPS)
Pay Attention to Small Weights

Conference on Neural Information Processing Systems (NeurIPS)
The Graphon Limit Hypothesis: Understanding Neural Network Pruning via Infinite Width Analysis

International Conference on Machine Learning (ICML)
Mirror, Mirror of the Flow: How Does Regularization Shape Implicit Bias?

Conference on Neural Information Processing Systems (NeurIPS)