update documentation to reflect IMDB output

[acxz]

When attempting to use the IMDB api, I got results that were different from what the docs suggested. This PR attempts to update the docs with the correct output of the IMDB api.

[Nayef211]

@acxz I think this might have been an oversight when the dataset was implemented. To be consistent with our other text classification dataset, we do want to return labels as integers.

If you want to take this on, it would be as simple as adding a MAP_LABELS similar to how we do it in the QNLI dataset. https://github.com/pytorch/text/blob/fd49d42c6c90fc4b556b63227f3509292bbdba11/torchtext/datasets/qnli.py#L40

And then we can do a simple label lookup in the _modify_res method like so https://github.com/pytorch/text/blob/fd49d42c6c90fc4b556b63227f3509292bbdba11/torchtext/datasets/qnli.py#L59

[acxz]

Ah I see, I'll take a stab at this in the upcoming weekend. Thanks for the implementation pointers.

[Nayef211]

@acxz lmk if you still plan on taking this on, otherwise I'm happy to submit a PR for it this week 😄

[acxz]

Yeah, actually if you could take it on that would be great. Ended up traveling this weekend and this week is looking a bit busy. Thanks for that!

[joecummings]

Closed in favor of #1914