scikit-learn wrappers

[Donatien Hainaut]

Hi,

the scikit-learn wrappers (keras.wrappers) in tensorflow 2.18 are not compatible anymore with scikit-learn 1.6.1. There are very useful for applying interpretability tools from scikit-learn to keras models. The issue seems to come from the system of tags implemented in the recent version of scikit-learn. The error message is:

AttributeError: 'super' object has no attribute '__sklearn_tags__'

The only solution that I've found is to downgrade scikit-learn to version 1.5.1. Maybe that some of you have found another solution?

Many thanks!

Donatien

 

[François Chollet]

Do you have a notebook to reproduce this issue?

--
You received this message because you are subscribed to the Google Groups "Keras-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to keras-users...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/keras-users/05d14442-f312-4fce-aadd-8ed13a953e97n%40googlegroups.com.

[Hainaut]

Cc to the group.

Le lun. 21 avr. 2025, 19:54, Hainaut <das.h...@gmail.com> a écrit :

Dear François,

Many thanks for your message. Unfortunately, I do not have a notebook (just a py file). I've done several tests and finally reinstalled Anaconda, installed tensorflow 2.19 and scikit-learn 1.6.1. With these versions, I do not have anymore this bug. Probably a problem of versioning on my pc. Sorry for disturbing.

Kind regards 

Donatien Hainaut 

[Rahul Reddy M.L]

Try upgrading all the dependencies may be they are outdated.

Just check