I also don't understand these lines (from here and forward)

Should random seeds check even be here? They are more of a general checks, than svm ones.

Should random seeds check even be here? They are more of a general checks, than svm ones.

Yes, we also validate the random state (see here for example). We use the string "random_state" in the constraints dictionary, which is interpreted as constraining the parameter to either be an instance of RandomState, an int, or None.

Does that answer the question?

Should random seeds check even be here? They are more of a general checks, than svm ones.

Yes, we also validate the random state (see here for example). We use the string "random_state" in the constraints dictionary, which is interpreted as constraining the parameter to either be an instance of RandomState, an int, or None.

Does that answer the question?

So, I need to remove these tests, since parameter validation was added and there is no need in outside random seed validation?

Should random seeds check even be here? They are more of a general checks, than svm ones.

I think I hadn't understand your question on my previous post. If the question is "why are we testing this inside the svm/tests/test_bounds.py file?", the reason is that we are not testing the random seed itself, but that the set_seed_wrap and bounded_rand_int_wrap functions from sklearn.svm._newrand are behaving as expected.

Hope this time I do answer the question! ;)

@ArturoAmorQ Oh, I see, so test_bounds.py contains tests for _bounds,py and _newrand.pyx