When using overfit_batches functionality of PyTorch I see the following file python3.7/site-packages/pytorch_lightning/trainer/data_loading.py execute
dataloaders = [deepcopy(train_dataloader) for _ in range(num_loaders)]
And that is resulting in the error:
TypeError: __new__() missing 2 required positional arguments: 'device_type' and 'device_id'
Could someone please help?
The deepcopy and hence the overfit_batches should work fine.
