feat: add validation_frequency option to fine-tuning (#811)#816
feat: add validation_frequency option to fine-tuning (#811)#816Anakintano wants to merge 1 commit intoPriorLabs:mainfrom
Conversation
|
Codex usage limits have been reached for code reviews. Please check with the admins of this repo to increase the limits by adding credits. |
- Add `validation_frequency: int | None = 1` parameter to `FinetunedTabPFNBase`, `FinetunedTabPFNClassifier`, and `FinetunedTabPFNRegressor` - When `validation_frequency=None`, validation is disabled entirely (also disables early stopping with a UserWarning) - When `validation_frequency=N`, validation runs every N epochs - Default of 1 preserves existing behaviour (validate every epoch) Closes PriorLabs#811
2e99f23 to
6affc25
Compare
There was a problem hiding this comment.
Code Review
This pull request adds a validation_frequency parameter to control validation during fine-tuning. The implementation is well-structured. I've added one comment to suggest validating the new parameter to prevent potential runtime errors. It would also be beneficial to add unit tests for this new functionality to ensure its correctness and prevent regressions.
|
Thanks @Anakintano for the contribution. |
Resolves #811
Summary
validation_frequency: int | None = 1parameter toFinetunedTabPFNBase,FinetunedTabPFNClassifier, andFinetunedTabPFNRegressorvalidation_frequency=None: disables validation entirely (also disables early stopping, emits UserWarning if early_stopping=True)validation_frequency=N: runs validation every N epochsvalidation_frequency=1(default): preserves existing behaviourTest plan