Validation dataset in retrain_from_log()

[harshvardhaniimi]

Hi, I'm trying to retrain a model with LightGBM. Earlier, I trained my model for a period of time and would like the algorithm to pick up where it stopped. retrain_from_log allows that, provided I have the log files saved which I do. However, I'm having trouble using the validation dataset with X_val. I get the following error.

TypeError: fit() got an unexpected keyword argument 'X_val'

The code works well when I'm using automl.fit(); this error only occurs in automl.retrain_from_log(). Any help? Are X_val and Y_val not valid parameters in retrain_fromlog() but only in fit()?

[sonichi]

X_val and Y_val are not valid in retrain_from_log. Please try removing them.

[harshvardhaniimi]

Any reason why it's not implemented? If fit() uses custom X_val and y_val, it's necessary to use it in retrain_from_log() to make the logs useful.

Otherwise, it makes sense to create an issue from this discussion for this feature request. 😄

[sonichi]

To clarify, X_val and y_val can be passed to the underlying estimator if they are supported by the underlying estimator, via fit_kwargs. The LightGBM estimator you are using doesn't support them in the fit().

[harshvardhaniimi]

I'm using them in my fit() function while specifying "estimator_list": ['lgbm'], and they worked well without errors...

[sonichi]

Yes, it's because AutoML.fit() uses X_val and y_val for validation. But they are not passed to LGBMEstimator.fit().