transformers-interpret icon indicating copy to clipboard operation
transformers-interpret copied to clipboard

Explainers for two input classification problem

Open isspek opened this issue 4 years ago • 2 comments

Thanks for the project.

My question is that if we have an input composed of two sentences for a classification problem, which should be separated with a [SEP] token, how can we set up the explainers? As I understand from the source code, it encodes single text.

isspek avatar Jul 28 '21 08:07 isspek

Hi @isspek

Thanks for using the package! So this is certainly doable as I do it for the QA explainer but I just hadn't thought to add the functionality for the classification explainer. I think it would only be a matter of adding an additional optional param to the explainer call method but I'll have to have a think on how to do it.

I should have some time to have a look at it over the weekend and see what's involved.

cdpierse avatar Jul 28 '21 21:07 cdpierse

Good question! I wonder how to do :)

savasy avatar Sep 14 '21 11:09 savasy

Closed by #99

cdpierse avatar Aug 23 '22 15:08 cdpierse