cerebras.modelzoo.data.vision.classification.dataset_factory.VisionClassificationProcessorConfig

class cerebras.modelzoo.data.vision.classification.dataset_factory.VisionClassificationProcessorConfig(*args, **kwargs)

Bases: cerebras.modelzoo.config.data_config.DataConfig

Methods

check_for_deprecated_fields
check_literal_discriminator_field
copy
get_orig_class
get_orig_class_args
model_copy
model_post_init
post_init

Attributes

batch_size	Global batch size for the dataloader
cutmix_alpha	Alpha parameter for the cutmix transform.
data_dir	The path to the data
discriminator
discriminator_value
drop_last	Similar to the PyTorch drop_last setting except that samples that when set to True, samples that would have been dropped at the end of one epoch are yielded at the start of the next epoch so that there is no data loss.
fp16_type
image_size	The size of the images in the dataset
mixed_precision
mixup_alpha	Alpha parameter for the mixup transform.
model_config
noaugment	Indicates to skip augmentation as part of preprocessing.
num_classes	The number of classification classes in the dataset
num_workers	How many subprocesses to use for data loading
persistent_workers	Whether or not to keep workers persistent between epochs.
prefetch_factor	Number of batches loaded in advance by each worker
ra_sampler_num_repeat	Number of repeats for Repeated Augmentation sampler.
sampler	Type of data sampler to use
shuffle	Whether or not to shuffle the dataset.
shuffle_seed	The seed used for deterministic shuffling.
transforms	List of transforms for preprocessing

data_dir = '.'

The path to the data

image_size = [224, 224]

The size of the images in the dataset

num_classes = Ellipsis

The number of classification classes in the dataset

batch_size = 128

Global batch size for the dataloader

shuffle = True

Whether or not to shuffle the dataset.

shuffle_seed = None

The seed used for deterministic shuffling.

drop_last = True

Similar to the PyTorch drop_last setting except that samples that when set to True, samples that would have been dropped at the end of one epoch are yielded at the start of the next epoch so that there is no data loss. This is necessary for a data ordering that is independent of the distributed setup being used.

num_workers = 0

How many subprocesses to use for data loading

prefetch_factor = 10

Number of batches loaded in advance by each worker

persistent_workers = True

Whether or not to keep workers persistent between epochs.

sampler = 'random'

Type of data sampler to use

ra_sampler_num_repeat = 3

Number of repeats for Repeated Augmentation sampler.

mixup_alpha = 0.1

Alpha parameter for the mixup transform.

cutmix_alpha = 0.1

Alpha parameter for the cutmix transform.

noaugment = False

Indicates to skip augmentation as part of preprocessing.

transforms = Ellipsis

List of transforms for preprocessing

previous

cerebras.modelzoo.data.vision.classification.dataset_factory.VisionClassificationProcessor

next

cerebras.modelzoo.data.vision.classification.dataset_factory.VisionSubset