Why is the calculation of cross_val_score differently to when i calculate it myself?

Question

Ok so here is the example code:

# imports
from numpy import mean
from sklearn.datasets import make_classification
from sklearn.model_selection import cross_valida