TensorFlow Recommenders

Handles the full recommendation pipeline from data preparation with TensorFlow Datasets to deployment, as shown in the quick start example using MovieLens.

Builds on familiar Keras patterns, making model formulation and training accessible to developers already using TensorFlow.

Enables construction of complex recommender models, such as embedding-based retrieval and multi-task learning, with customizable layers and tasks.

Includes built-in tasks like Retrieval with metrics such as FactorizedTopK, simplifying assessment over entire candidate datasets.