search

autogluon / autogluon

Fast and Accurate ML in 3 Lines of Code
https://auto.gluon.ai/
Apache License 2.0
8.04k stars 928 forks source link

[Tabular] Support for Raising an Error When No Models Were Trained Successfully During fit() #4389

Closed LennartPurucker closed 3 months ago

LennartPurucker commented 3 months ago

This PR enables users to make AutoGluon raise an error if no models are trained successfully during fit.

Without this, people sometimes write their own code to catch this or ignore it and then believe AutoGluon is broken, although it is most likely a result of a broken environment or their memory being insufficient.

So far, this defaults to False, but I would argue this should default to True in the future.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

LennartPurucker commented 3 months ago

@Innixma, I think you need to re-trigger the CI with your authentication. Otherwise, I am lost in what the error message could be.

Innixma commented 3 months ago

@Innixma, I think you need to re-trigger the CI with your authentication. Otherwise, I am lost in what the error message could be.

Fun fact, the CI failed because all of GitHub was down yesterday for over 30 minutes: https://www.githubstatus.com/incidents/kz4khcgdsfdv

github-actions[bot] commented 3 months ago

Job PR-4389-5380bb4 is done. Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-4389/5380bb4/index.html