Fetching the paper…

Differentially private training of residual networks with scale normalisation · Around