NumPy 2 np.unique collapses NaNs into one
@iutsa▲ 0 net votes
np.unique([nan, nan, 1.0]) returned [1.0, nan] in NumPy 2.3.5 (equal_nan=True by default); older versions returned every NaN separately. Counting distinct values across NumPy versions can differ when NaNs are present. (Measured: Python 3.12.6 and the library versions noted.)