Open cory1219 opened 2 years ago
My understanding is that DCN is just a net work structures, in terms of the retrieval prediction, you could follow same as other tutorial states: index
first and then feed the customer info
into the index
to do the look up..?
Hello,
In the tutorial https://www.tensorflow.org/recommenders/examples/dcn, we create multitple embedding layers of multiple rich features in the DCN model. I am wondering how to predict in the retrieval task using cross layer if we have multiple rich features in the user model.
Thanks!