Tensorflow recommenders

Hi,
I went through the tutorials for the recommenders and what I’m missing it how to implement “candidate” features into the model.

For example lets say the model takes as an input user_id (or other user/features) and returns media_title. How should I add other media /candidate features like genre to the training - but keeping the input the same for the model ?

Thanks.

1 Like

This tutorial two features, title look up and embedding, to a candidate model (MovieModel).

Google for ‘tensorflow recommenders side features’

getting same issue and errors, any solution, it says me to NotImplementedError: When subclassing the Model class, you should implement a call method.

you need to implement call method as the error says, which is something like this:

    def call(self, input):
        return tf.concat([
            self.title_text_embedding(input['title']),
            self.body_text_embedding(input['body']),
        ], axis=1)