How to create dataloader from a single example in a seq-to-seq model

1

I am building a sequnce-to-sequence model.

For MWE, I have one learning example: [0, 1, 2, 3, 4] --> [0, 1, 2, 3, 4]

How can I create a dataloader from it? I tried

x = torch.arange(5)
y = torch.arange(5)

test_dataloader = torch.utils.data.dataloader.DataLoader((x, y))

But it gives me:

for data in test_dataloader:
    print(data)
    print("----")

tensor([[0, 1, 2, 3, 4]])
----
tensor([[0, 1, 2, 3, 4]])
----

While I expected to get:

(tensor([[0, 1, 2, 3, 4]]), tensor([[0, 1, 2, 3, 4]]))
----

I need this because inside for loop I want to unpack learning samples like

X, y = data

2

Try to use a TensorDataset and pass the data in the shape [batch_size, features] to it.
Once this dataset is created pass it to the DataLoader.