CIFAR datasets with label noise
Project description
Noisy CIFAR-10/CIFAR-100
Quickstart
from noisy_cifar import NoisyCIFAR10
from torchvision.datasets import CIFAR10
from torchvision.transforms import ToTensor
root = '/datasets/CIFAR-10'
train_dataset = NoisyCIFAR10(root, 'symmetric', 0.2, transform=ToTensor())
val_dataset = CIFAR10(root, train=False, transform=ToTensor())
Install
pip install noisy-cifar-owaix2quzq
Arguments
root
: path to the original datasetnoise_type
: symmetric / asymmetric / humannoise_level
:- (float) [0.0 ~ 1.0] for symmetric & asymmetric
- (str) {aggregate, random1, random2, random3, worst} for human
random_seed
: default0
transform
: defaultNone
target_transform
: defaultNone
download
: defaultFalse
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
noisy-cifar-owaix2quzq-0.0.1.tar.gz
(429.0 kB
view hashes)
Built Distribution
Close
Hashes for noisy-cifar-owaix2quzq-0.0.1.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | b8280723202ed19e8c26441a073cda45be278b3b34c2d585f9b407fa570e3f3b |
|
MD5 | 23653d149ab51add7893e4ad4a2e08e9 |
|
BLAKE2b-256 | ce5187b94bbb992960c5a6be157d3d124692162ce4c26c82c473af6161993774 |
Close
Hashes for noisy_cifar_owaix2quzq-0.0.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | a1f5aa8e032ac4d07bbe505502d6c976be4be38157f7461ab44240a76373c418 |
|
MD5 | affc07202a9a85790a041add824c1266 |
|
BLAKE2b-256 | 3ad972da2fdb1d8b96b59ba54e3769371d0fb0d3c0ead9e23e1470f26883e51f |