Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

`BOTORCH_MODULAR` with `Surrogate(SingleTaskGP)` doesn't support `ax.plot.feature_importances.plot_feature_importance_by_feature`, is ARD kernel still used?

See original GitHub issue

https://github.com/facebook/Ax/issues/771#issuecomment-1011742002

I tried using e.g. SingleTaskGP, but I was getting an error message that feature importances weren’t supported. However, I’m pretty sure it’s still an ARD kernel based on the docs. Am I missing something basic?

Issue Analytics

State:
Created 2 years ago
Comments:6 (6 by maintainers)

Top GitHub Comments

2reactions

Balandatcommented, Jan 19, 2022

One thing that is clear from this is that we are definitely heavily overloading the term “model” 😃

2reactions

lena-kashtelyancommented, Jan 18, 2022

Building off of @bernardbeckerman’s suggestion above, the way you would apply the contents of BotorchModel.feature_importances, is, given an AxClient, using: ax_client.generation_strategy.model.model.surrogate.model instead of self.model that instance method.

P.S.: We know ax_client.generation_strategy.model.model.surrogate.model is a little ridiculous 😃 There, ax_client.generation_strategy.model gives the model bridge, then ax_client.generation_strategy.model.model gives the BoTorchModel in Ax, then ax_client.generation_strategy.model.model.surrogate.model gives the underlying Model component from BoTorch (e.g. SingleTaskGP).