scope.int not consistently returning int value

Hi, I'm facing an issue with hyperopt where the scope.int function does not returns an int but a float instead :

sklearn.utils._param_validation.InvalidParameterError: The 'min_samples_leaf' parameter of RandomForestClassifier must be an int in the range [1, inf) or a float in the range (0.0, 1.0). Got 2.0 instead.

Here is the python code that is using hyperopt and not working for me :

'min_samples_leaf': scope.int(hp.quniform('min_samples_leaf', 1, 20, 1))

Perhaps I'm being mislead by the name "scope.int" but I think it should not return 2.0 but 2 instead ?