Error while creating the dataset

audio
1

I have a train input of size [3082086, 7, 30], that is 3082086 frames, a window of 7 frames, each single frame contains 30 numbers.
The train labels is of size [3082086, 1], that is one label for every window of 7 frames.
I tried to stack them to create a train set as follow:
trainset = torch.hstack(train_input, train_labels)
…but I got an error RuntimeError: Sizes of tensors must match except in dimension 1. Got 30 and 1 in dimension 2 (The offending index is 1)

Any way I can create the dataset with the above dimensions?

2

You dont want to mix inputs and labels into the same tensor, it doesnt make sense to merge them. Instead create a dataset, have separate tensors self.inputs, self.labels in the datasets __init__ and In the __getitem__(idx), just return self.inputs[idx], self.labels[idx]

1 Like
3

@bsridatta That did it. Thanks mate.

1 Like