unbalanced probabilities in AutoTabPFNClassifier #103
Description
Describe the workflow you want to enable
Describe the workflow you want to enable
Reported by @YuvalRom on tabpfn package
Thank you for this amazing work.
My data is unbalanced and I have around 20% positive samples, in sklearn there is the "class_weight" argument. In TabPFN there is the "balance_probabilites" argument but I don't see one for AutoTabPFN anywhere in the documentation. I tried using it the same way as TabPFN but got an error it doesn't recognize "balance_probabilites".
Is there someway to do this for the automated version?
Thanks
Describe your proposed solution
Maybe we can implement the "balance_probabilites" the same way - It's an easy fix I can help with that
Describe alternatives you've considered, if relevant
No response
Additional context
No response
Impact
None
Describe your proposed solution
See above
Describe alternatives you've considered, if relevant
No response
Additional context
No response
Impact
None