Recommender Systems: An Applied Approach using Deep Learning - Compute Loss

The video tutorial explains how to use TensorFlow Recommenders (TFRS) to train a model by utilizing its loss layers and tasks. It introduces the Retrieval Task object, which simplifies the bundling of loss functions and metric computations. The tutorial then guides viewers through defining the compute_loss function, detailing the use of features as a dictionary containing text and TF tensors. It further explains the creation of user and beer embeddings using user_model and beer_model, respectively. The video concludes with the completion of the Amazon product class and hints at defining the cross TensorFlow callback in the next video.

What is the significance of the built-in wrapper mentioned in the text?

What are the two types of embeddings returned by the model?