LSTM in Pytorch: how to add/change sequence length dimension?

Can someone help me to modify the code to add a sequence dimension? The code and description are provided here. Thank you.

In PyTorch LSTM model, you need not/can’t specify sequence length. Here sequence length of the model depends on the input data. As in your code you have specified ‘batch_first=True’, the data dimension should be (batch size, seq length, input feature dimension ). In your case the data given to LSTM model should have a shape of (200,4,16839) .

What would be the shape of my target label, (200, 4, 3)?
I think dataloader takes care of the batch size. What shapes do I supply for the train and target to the dataloader then?