Maybe related with #15761.

Describe the issue linked to the documentation

I have noticed that when the default is None for some parameter or attribute, the meaning is included only in some cases.

For instance, for the fit method in the class sklearn.tree.DecisionTreeClassifier, sample_weight=None is documented as:

sample_weight : array-like of shape (n_samples,), default=None
Sample weights. If None, then samples are equally weighted.

However, for the score method it is:

sample_weight : array-like of shape (n_samples,), default=None
Sample weights.

It is okay like that or should be specified always?