Exogenous variable inclusion to dataloader

I am new at PyTorch, and try to model time series problem with 1-D CNN. I have to provide 3 different inputs, which are load, temperature, day dummies, separately to the network. However, I did not find any clue to provide this type of setting to the data loader in pytorch. Which way I can try to provide this setting to data loader?


If your dataset’s __getitem__ method returna a tuple of three torch.Tensors, DataLoader will generate a tuple of three minibatches of those tensors.

Thank you. I will give a shot with your recommendation.