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
When overfitting on a single batch and defining dataloaders in class, the batch provided to the validation step is different from the batch on the training step. I was told in the slack community that this is NOT the intended behaviour.
Bug description
When overfitting on a single batch and defining dataloaders in class, the batch provided to the validation step is different from the batch on the training step. I was told in the slack community that this is NOT the intended behaviour.
How to reproduce the bug
The val batch is the [10,11] tensor, the train batch is the [0,1] tensor
The text was updated successfully, but these errors were encountered: