Tobias Pitters

Note that this PR solves the issue only for the `__call__` method, `.shap_values` is left unchanged

Nothing's wrong, I suspect your model has four outputs, therefore the shape of the shap values is: `(160, 6, 4)`.

I executed that locally and found that the installed version of tensorflow somehow requires that we import things differently, e.g. instead of ```python from tensorflow.keras.models import Model ``` we need...

We recently had a breaking release that broke this code (see [v.0.45.0](https://github.com/shap/shap/releases/tag/v0.45.0)). Here is how you fix the code ```python class_idx = 0 sample_idx = 0 shap.force_plot(explainer.expected_value[0], shap_values[sample_idx, :, class_idx],...

Hi @ElaineYao, thanks for reaching out. We are happy to use tools that simplify our workflow here. As far as I see, your package is not released yet (just available...

Sure let's schedule a call. Anytime after June 9 will work. Please be aware that I am on Berlin time, so maybe we can find a slot that works for...

already implemented by #4143

Thanks for the report. You are simply trying to plot way too many points, just try to sample the explanation, matplotlib plots tend to be quite slow with a large...

@afafelwafi thanks for pointing this out. @arkanoid87 Maybe I have stated this not well enough but don't see that this is a problem with the shap library but a limitation...

@CousinThrockmorton: may I ask for your opinion here?