Fetching the paper…

LOT: Layer-wise Orthogonal Training on Improving $\ell_2$ Certified Robustness · Around