Gradient Descent Optimizes Infinite-Depth ReLU Implicit Networks with Linear Widths

Open in new window