Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix (Fashion|K)MNIST download and MNIST download test #3557

Merged
merged 2 commits into from
Mar 12, 2021
Merged

Fix (Fashion|K)MNIST download and MNIST download test #3557

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7284c1b
add mirrors to (Fashion|K)MNIST
pmeier Mar 12, 2021
f659405
fix download tests for MNIST
pmeier Mar 12, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion test/test_datasets_download.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -249,7 +249,8 @@ def voc():


def mnist():
return collect_download_configs(lambda: datasets.MNIST(ROOT, download=True), name="MNIST")
with unittest.mock.patch.object(datasets.MNIST, "mirrors", datasets.MNIST.mirrors[-1:]):
return collect_download_configs(lambda: datasets.MNIST(ROOT, download=True), name="MNIST")
Comment on lines -252 to +253
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is a hacky fix, but the test infrastructure is currently not able to handle mirrors. During test collection, our tests mock out download_url and record all calls to for each dataset. Afterwards we do head requests against the logged URLs to check if they are accessible. Since we can't know if a request returns an error during collection, the mirror URLs are never collected.

Maybe we should add a mirror functionality to download_url that handles this internally. If we had this, we could add a functionality like this to our tests that warns in case a mirror is down.



def fashion_mnist():
Expand Down
28 changes: 16 additions & 12 deletions torchvision/datasets/mnist.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -206,15 +206,15 @@ class FashionMNIST(MNIST):
target_transform (callable, optional): A function/transform that takes in the
target and transforms it.
"""
mirrors = [
"http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/"
]
Comment on lines +209 to +211
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(Fashion|K)MNIST use the same download logic as MNIST. Thus, we need to also provide the URLs in the same fashion (no pun intended 😉) although we only have a single mirror.


resources = [
("http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-images-idx3-ubyte.gz",
"8d4fb7e6c68d591d4c3dfef9ec88bf0d"),
("http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-labels-idx1-ubyte.gz",
"25c81989df183df01b3e8a0aad5dffbe"),
("http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz",
"bef4ecab320f06d8554ea6380940ec79"),
("http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-labels-idx1-ubyte.gz",
"bb300cfdad3c16e7a12a480ee83cd310")
("train-images-idx3-ubyte.gz", "8d4fb7e6c68d591d4c3dfef9ec88bf0d"),
("train-labels-idx1-ubyte.gz", "25c81989df183df01b3e8a0aad5dffbe"),
("t10k-images-idx3-ubyte.gz", "bef4ecab320f06d8554ea6380940ec79"),
("t10k-labels-idx1-ubyte.gz", "bb300cfdad3c16e7a12a480ee83cd310")
]
classes = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat', 'Sandal',
'Shirt', 'Sneaker', 'Bag', 'Ankle boot']
Expand All @@ -236,11 +236,15 @@ class KMNIST(MNIST):
target_transform (callable, optional): A function/transform that takes in the
target and transforms it.
"""
mirrors = [
"http://codh.rois.ac.jp/kmnist/dataset/kmnist/"
]

resources = [
("http://codh.rois.ac.jp/kmnist/dataset/kmnist/train-images-idx3-ubyte.gz", "bdb82020997e1d708af4cf47b453dcf7"),
("http://codh.rois.ac.jp/kmnist/dataset/kmnist/train-labels-idx1-ubyte.gz", "e144d726b3acfaa3e44228e80efcd344"),
("http://codh.rois.ac.jp/kmnist/dataset/kmnist/t10k-images-idx3-ubyte.gz", "5c965bf0a639b31b8f53240b1b52f4d7"),
("http://codh.rois.ac.jp/kmnist/dataset/kmnist/t10k-labels-idx1-ubyte.gz", "7320c461ea6c1c855c0b718fb2a4b134")
("train-images-idx3-ubyte.gz", "bdb82020997e1d708af4cf47b453dcf7"),
("train-labels-idx1-ubyte.gz", "e144d726b3acfaa3e44228e80efcd344"),
("t10k-images-idx3-ubyte.gz", "5c965bf0a639b31b8f53240b1b52f4d7"),
("t10k-labels-idx1-ubyte.gz", "7320c461ea6c1c855c0b718fb2a4b134")
]
classes = ['o', 'ki', 'su', 'tsu', 'na', 'ha', 'ma', 'ya', 're', 'wo']

Expand Down