Making EfficientNet More Efficient: Exploring Batch-Independent Normalization, Group Convolutions and Reduced Resolution Training

Dominic Masters, Antoine Labatie, Zach Eaton-Rosen, Carlo Luschi