I am trying to write a NN for estimating a f:R^n --> R^m. My problem is how to train network. I mean if I want to define a dataloader, how to attach X \in R^n to its related Y \in R^m ? Because usually we have an integer form labels and not tensor vectors. But, how to do this? BTW, is Pytorch smart enough if I don't attach the list of outputs e.g. Y to list of inputs X by a DataLoader (as I do not know how to do it) and just write loss(model(x), y), and for batch size of 5 for X, does pytorch itself compare 5 first of X (as input batch) with 5 first y of Y. And then the second five of each list?
Asked
Active
Viewed 21 times
