TextVectorization layer widely use in the course for Tensorflow Certificate is not compatible with h5 format

[LubuntuSharp]

Hello,
I am still trying to take the exam next week, I was going to finalize my exercises when I realized (with terror) that the TextVectorization layer widely use in the notebooks is not compatible with h5 format, so you cannot use during the exam which requires to submit model saved this way. I tried with no success to replace with KerasNLP or KerasPreprocessing;

are you aware of this substantial limitation?
could you suggest a workaround, if exists?

Regards,