megatron.data.autoaugment.ImageNetPolicy

class megatron.data.autoaugment.ImageNetPolicy(fillcolor=(128, 128, 128))

Bases: object

Definition of an ImageNetPolicy.

Implements a fixed AutoAugment data augmentation policy targeted at ImageNet training by randomly applying at runtime one of the 25 pre-defined data augmentation sub-policies provided in Reference [1].

Usage example as a Pytorch Transform: >>> transform=transforms.Compose([transforms.Resize(256), >>> ImageNetPolicy(), >>> transforms.ToTensor()])

__call__(img)

Define call method for ImageNetPolicy class.

__repr__()

Define repr method for ImageNetPolicy class.