Marco Tulio Correia Ribeiro
Marco Tulio Correia Ribeiro
Can you run the example notebooks? Can you share an example (with data) so I can look at it in more detail? Thanks,
Is it the case that there is extreme class imbalance? If so, the empty anchor just means that almost all perturbations we try have no impact in the prediction, even...
The threshold is a hyperparameter (`threshold`), with default value 0.95. I think the empty anchor is the explanation I want in this case though, as it tells me that something...
No ongoing work at the moment, I'm rather busy with other stuff. I think [this](https://github.com/marcotcr/anchor/blob/master/anchor/anchor_image.py) should work (or almost work), the interface is just not great.
If precision is still not 1 with all the features, it means that the discretization is too coarse. Try using `discretizer=decile` or providing your own discretizer.
It may very well be the case that the model is too 'jumpy', in which case a full anchor is the right thing even if it's not useful (this is...
Can you share an example @iamshnoo? I think there is either a bug or (more likely) you're not accounting for token healing issues. For example, if you run `lm +...