C2W4 ImageDataGenerator normalisation results in 0.0

The greyscale images have 256 levels. I normalised the image values using the rescale argument for ImageDataGenerator. The grader marks me down because the normalised values from the test come out at zero precisely (see below message output).
Yet the model trains and validates so the images cannot be all zero-valued. Please advise?

Thanks

"Details of failed tests for train_val_generators

Failed test case: rescale argument of ImageDataGenerator has incorrect value for train_generator.
Expected:
0.00392156862745098 (1/255),
but got:
0.

Failed test case: rescale argument of ImageDataGenerator has incorrect value for validation_generator.
Expected:
0.00392156862745098 (1/255),
but got:
0."

Please click my name and message your notebook as an attachment.

The first positional argument of ImageDataGenerator.init is featurewise_center. Please use the correct keyword argument to normalize the pixels.