You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Realized MNISTDataModule was ignoring batch_size parameter. I found a closed issue (#171) referring to that without a fix.
While fixing it myself (PR #331), I found more datamodules had this problem too - MNISTDataModule, BinaryMNISTDataModule, FashionMNISTDataModule, SklearnDataModule, SSLImagenetDataModule.
I could take care of that. My question is, is this signature from MNIST in use anymore?:
Cool, I fixed the signatures too. Please @ananyahjha93 have a look at the PR here #331
Should I assign somebody for review myself? I'm a bit lost on that.
馃悰 Bug
Realized
MNISTDataModule
was ignoringbatch_size
parameter. I found a closed issue (#171) referring to that without a fix.While fixing it myself (PR #331), I found more datamodules had this problem too -
MNISTDataModule
,BinaryMNISTDataModule
,FashionMNISTDataModule
,SklearnDataModule
,SSLImagenetDataModule
.I could take care of that. My question is, is this signature from MNIST in use anymore?:
That is basically where the bug comes from. Other datamodules working fine, like
CIFAR10DataModule
, are simply:Same thing goes for
val_dataloader
andtest_dataloader
.To Reproduce
Steps to reproduce the behavior:
MNISTDataModule
to use a batch size anything else than 32.Code sample
Expected behavior
Batch sizes of 64.
Printing
fine
instead of assert error.Environment
conda
,pip
, source): condaAdditional context
The text was updated successfully, but these errors were encountered: