LightGBM error - train() got unexpected keyword argument

I was uploading basically the same script which worked few weeks ago but this time it didn’t work.

run id: #2978

lightgbm.train function has changed bewteen 3.x and 4.x.
The early_stopping_rounds does not seems to exists anymore.

You must have an older version installed on your computer/colab, but the runner always install the latest version.


Thank you @enzo,
2 follow up questions:
1 - is there a way how to control that? (e.g. if I want that specific version of library would be used)
2 - if I have uploaded few weeks ago model (let’s say it is latest upload and at the time it ran properly which makes it to be used in oos period) - will there be issue in oos period?

Many thanks in advance!

Its by using the CLI instead of the notebook.

I will try to see if I could extract comment so you could put like:

import lightgbm # =3.5.2
import lightgbm # >=4

We are soon gonna introduce a selector with which you will be able to chose what submission you want, with which train frequency and which runner type (CPU vs GPU).

We are going to make an announcement when one or both of these features will be ready.
(I will work on the first one today, the second one is almost ready but still need discussion with ADIA Lab and the Team)

Hey @prime-paulius,

The version specifiers are here:

Don’t hesitate to give me feedbacks!

Waiting for some before making an announcement to make sure everything is perfect!

Thank you