I saw the same problem on the dev chat, about how to use TabularPandas and create a dataloader with weighted_dataloader. This is what I did:
to = TabularPandas(train_valid, procs=[Categorify], cat_names=cat_var, cont_names=, y_names='active', splits=splits,y_block=CategoryBlock) dls = to.weighted_dataloaders(bs=64,wgts=weights)
It seems my dataloader failed with this message
Could not do one pass in your dataloader, there is something wrong in it
Anybody managed to create a weighted_dataloader for tabular data?