vivekmig

Hi @ahmadajal , here is a workaround to obtain explanations that require gradient. We essentially need to override the default gradient function, to add an additional parameter create_graph=True, which enables...

Hi @CleonWong, yes, we are working on a Lime tutorial, it will be released soon :)

Hi @m30m , thanks for the contribution! It is good to note this information regarding the sanity check results. Since different methods have different disadvantages, we think it might be...

Thanks for reporting this issue @katjahauser ! We have added a similar warning to LRP in #911 .

Hi @ndalton12 , this is definitely not expected, it seems like there is some memory leak causing this issue. I attempted to reproduce this with the latest versions of Captum...

Interesting, thanks for the detailed information! It does sound like it could be related to PyTorch lightning since the original module doesn't result in the issue, but would need a...

Hi @zhj123169 , it's possible that the default regularization term with Lasso regression may be too high, causing the attribution values to become 0. You can try providing interpretable_model as...

Also, looks like the model file isn't included in this PR, we need to update that as well. The current model is [boston_model.pt](https://github.com/pytorch/captum/blob/master/tutorials/models/boston_model.pt)

Hi @HCWDavid , the issue with visualize_timeseries_attr is likely due to the method being recently added, so it was not included in the previous pip / conda releases, and will...

Closing this issue, feel free to open another one if there are further issues.