v0.6.0

New Features

• Added compatibility with pytorch version 1.9.0 (#753)
• Added RegNetZ model implementation (#713)
• Redefined RegNetY 64GF to better performing model (#748) 
• Added support for Pytorch AMP + ShardedDDP (#667)
• Added support for cutmix in MixupTransform
• Added support for a 1x1 convolution in the fully connected head (#712) 
• Add support to normalize inputs in the heads (#664) 
• Added support for SiLU activation in RegNets (#711) 
• Added support for FP16 gradient communication in DDP (#709) 
• Added reduction mode none in soft_target_cross_entropy_loss (#714) 
• Added FLOPs computation to ViT (#746)
• Upstreamed parameter schedulers to fvcore