Hi, Is there a way of pre-padding a text sequence. The pad_sequence implementation gives post-padding result.
train_data = pad_sequence(train_data, batch_first=True, padding_value=vocab.stoi['<pad>'])
train_data[0]
out: tensor([60, 15, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1])
The result i want to get is:
out: tensor([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 60, 15])