Fetching the paper…

Towards Training Without Depth Limits: Batch Normalization Without Gradient Explosion · Around