Pipeline including StandardScaler and Gaussian Naive Bayes with low accuracy

Question

I have tried to do classification using below pipeline including Gaussian Naive Bayes and Standard Scaler:

GNB_pipeline = Pipeline(steps=[
    ('scale',StandardScaler()),
    ('model', GaussianNB())
])

But when I receive an accuracy which is shocking!:

It is in the condition that when I run StandardScaler and Naive Bayes separately, the accuracy is like below (around 77%):

I know that Gaussian naive bayes doesn't require standardization but I want to know how these pipelines work that doesn't allow naive bayes operate well, please?