Zero factorizedtopk accuracy

[thani-ath-nain-khurshid]

I am training a tensorflow recommenders model on user ids and product features like category sale percentage etc following the tutorial at https://www.tensorflow.org/recommenders/examples/deep_recommenders but whatever I do my accuracy as computed by the retrieval task is always 0 Even though loss is decreasing. What can be the cause here?

[mustfkeskin]

Hi you need set compute_metrics = True

return self.task(
        query_embeddings, movie_embeddings, compute_metrics=not training)

But if you do this, the training time will be longer.

[thani-ath-nain-khurshid]

Thanks