RFs: fastai's set_rf_samples & sklearn's max_samples

Hey, I’m currently going through the intro to machine learning for coders lessons…

From lesson 2, the fastai set_rf_samples class sets the number of rows to be randomly sampled from the training data for the trees to be trained on.

Sklearn seem to have introduced this functionality into their RandomForrestRegressor class with the max_samples parameter: https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.ExtraTreesClassifier.html

However, the two are giving different RMSE and R^2 results - the fastai method being better in both cases.

Does any one know if there is a difference in these two methods and if so, what is it?