Amit Sharma

Thanks for raising this @shreyakhandelwal07 Agree, this can be a useful feature for classification models. However, it may require specific logic for handling strings in many places. We will look...

The method works only for differentiable models like neural networks. It is not included separately, but if you set the diversity loss coefficient to be zero for the tf or...

@aayesha-coder @drishyamlabs As of v0.5, we have included support for non-differentiable models using the parameter `backend="sklearn"` for the `Model` class. All sklearn classifiers/regressors are supported. In addition, since DiCE only...

one question: why is the test_dice_random.py file deleted? Is the idea that those tests are moved to explainer base?

paging @soundarya98 will be good to resolve this.

Slightly late to this discussion, let me chime in. 1. The original issue about supporting post-hoc filtering is a good one. It isn't required if the model supports categorical variables,...

@etheng-jiao if you are using tensorflow, then need to set the backend to `TF1` (for tensorflow v1) or `TF2`. `m = dice_ml.Model(model=ann_model,backend='TF1')`

Did the fix above work for you @etheng-jiao ?

yeah, this is a work-in-progress for TF models. At this point, it might be end-June by which we can support this.

It's unclear why CF generation is successful for the raw data but not the z-scored ones. Is it possible to share a minimum working example so we can debug? Perhaps...