TensorFlow site shows that the Tokenizer, and the whole tf.keras.preprocessing APIs, are deprecated. Why are we still learning to use it?
And what about the TensorFlow Developer exam: are we expected not to use the deprecated APIs in the exam?
The tensorflow exam required the candidate to know Tokenizer API. Here’s the exam handbook. Do check with tensorflow exam help center if this is still the case and please let us know.
I did check the exam handbook but it doesn’t mention the Tokenizer. It just says “Prepare text to use in TensorFlow models”. I am just wondering if it will be marked as “incorrect” if we still use the Tokenizer in the exam instead of the tf.keras.layers.TextVectorization (with which the Tokenizer is replaced as per the API documentation)!?
Same question is valid for CNNs (Course 2). It looks like the ImageDataGenerator API is also deprecated.
2 Likes
Here’s a pdf of setting up your environment where they state that 2.10.x is used for the exam. The staff have been notified to update the notebooks.
Do note that 2.7.x was used around aug 8, 2022 and this update of 2.10.x was announced only this jan.
That said, what did the google tensorflow certificate help center tell you regarding usage of deprecated apis?