scipy edition requirement

[liuyunhui123]

Hi there :)
I happen to use your tool to analyze my code, with my SciPy being 1.1.0 (in pip I got "Requirement already satisfied: scipy>=0.19.1 in __MY_DISK_PATH__ (from scikit-learn->dnn-tip==0.1.0) (1.1.0)").
However, when I use your LSA, error occurs:

Traceback (most recent call last):
  File ".\run_sas.py", line 123, in <module>
    sa = LSA(train_ats)
  File "__MY_DISK_PATH__\venv_p367_x\lib\site-packages\dnn_tip\surprise.py", line 444, in __init__
    self.kde = self._create_gaussian_kde(activations)
  File "__MY_DISK_PATH__\venv_p367_x\lib\site-packages\dnn_tip\surprise.py", line 475, in _create_gaussian_kde
    return StableGaussianKDE(cleaned_activations.transpose())
  File "__MY_DISK_PATH__\venv_p367_x\lib\site-packages\dnn_tip\stable_kde.py", line 22, in __init__
    dataset.astype(np.float64), bw_method, weights
TypeError: __init__() takes from 2 to 3 positional arguments but 4 were given

And I traced to my \scipy\stats\kde.py and find that mine has NOT a parameter of 'weights'.
That is, the weighted kde is brought to scipy after 1.2.0, see: https://github.com/scipy/scipy/releases/tag/v1.2.0

Hope this can help to upgrade your requirement :) Or could you specify a stable scipy edition?
Best wishes.