Rationale Introspective Explainer Visualization Bugs

[minthigpen]

[cls] and [sep] should not show up. Also the label doesn't change and the anti-rationales, which would have been the neg feat. importance are not passed into the dashboard

[janhavi13]

@imatiach-msft : @enosakhare recommended the fix (similar fix) which was made to unified information explainer. return _create_local_explanation( classification=True, text_explanation=True, local_importance_values=np.array(local_importance_values)[1:-1], method="neural network", model_task="classification", features=self.parsed_sentence[1:-1], classes=classes, predicted_label=predicted_label, true_label=true_label, )

[imatiach-msft]

I'm guessing this is after doing the full run? doing a quick run shows this:

[rebeccaj1211]

@janhavi13 comment about the sentence[1:-1] is the ideal fix since we wanted the dashboard to be agnostic to the input. However, if you wanted to make this change in the dashboard, It's a relatively easy change to make in widget/ExplanationDashboard.py