CIFAR: permanent 'data' and 'targets' fields #594
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
I'm totally ok with merging such changes, but for the next release we need to make sure that we mention that those attributes have been removed, as this is a backwards-incompatible change. I'm mentioning this as a self-reminder. I'll have a closer look to your PR later |
scottclowe
added a commit
to scottclowe/cifar10-fast
that referenced
this pull request
Mar 22, 2019
In the latest release of torchvision (v0.2.2), the fields of the CIFAR10 object containing the data were changed from - both `train_data` and `test_data` to `data` - both `train_labels` and `test_labels` to `targets` https://github.com/pytorch/vision/releases/tag/v0.2.2 These changes were made in PR pytorch/vision#594 Our code is pulling the data out from the old fields, which are not supported in current and future versions of torchvision. To ensure we can support both torchvision <=0.2.1 and >=0.2.2, the fix checks for and uses whichever of these two versions exists (with a preference for the new naming convention if it is available).
TheCBaH
pushed a commit
to TheCBaH/cifar10-fast
that referenced
this pull request
May 5, 2019
In the latest release of torchvision (v0.2.2), the fields of the CIFAR10 object containing the data were changed from - both `train_data` and `test_data` to `data` - both `train_labels` and `test_labels` to `targets` https://github.com/pytorch/vision/releases/tag/v0.2.2 These changes were made in PR pytorch/vision#594 Our code is pulling the data out from the old fields, which are not supported in current and future versions of torchvision. To ensure we can support both torchvision <=0.2.1 and >=0.2.2, the fix checks for and uses whichever of these two versions exists (with a preference for the new naming convention if it is available).
|
I don't think you left the "mention" properly. |
|
@Cipher731 yes, look like I forgot to add the backward-compatibility fixes. Given that this has been in two releases already, I'm going to be considering that this has been dropped already and won't be included anymore. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I should have done it when I fixed MNIST 'data' and 'targets' fields in #578.
I found that people do use dynamic
train_dataandtest_datafields (see here for example), obfuscating their code a lot.