TensorFlow Recommenders: Quickstart

View on TensorFlow.org View source on GitHub Download notebook

In this tutorial, we build a simple matrix factorization model using the MovieLens 100K dataset with TFRS. We can use this model to recommend movies for a given user.

Import TFRS

First, install and import TFRS:

pip install -q tensorflow-recommenders
pip install -q --upgrade tensorflow-datasets
from typing import Dict, Text

import numpy as np
import tensorflow as tf

import tensorflow_datasets as tfds
import tensorflow_recommenders as tfrs

Read the data

# Ratings data.
ratings = tfds.load('movielens/100k-ratings', split="train")
# Features of all the available movies.
movies = tfds.load('movielens/100k-movies', split="train")

# Select the basic features.
ratings = ratings.map(lambda x: {
    "movie_title": x["movie_title"],
    "user_id": x["user_id"]
})
movies = movies.map(lambda x: x["movie_title"])
Downloading and preparing dataset movielens/100k-ratings/0.1.0 (download: 4.70 MiB, generated: 32.41 MiB, total: 37.10 MiB) to /home/kbuilder/tensorflow_datasets/movielens/100k-ratings/0.1.0...
Shuffling and writing examples to /home/kbuilder/tensorflow_datasets/movielens/100k-ratings/0.1.0.incompleteZHA953/movielens-train.tfrecord
Dataset movielens downloaded and prepared to /home/kbuilder/tensorflow_datasets/movielens/100k-ratings/0.1.0. Subsequent calls will reuse this data.
Downloading and preparing dataset movielens/100k-movies/0.1.0 (download: 4.70 MiB, generated: 150.35 KiB, total: 4.84 MiB) to /home/kbuilder/tensorflow_datasets/movielens/100k-movies/0.1.0...
Shuffling and writing examples to /home/kbuilder/tensorflow_datasets/movielens/100k-movies/0.1.0.incompleteBULZB9/movielens-train.tfrecord
Dataset movielens downloaded and prepared to /home/kbuilder/tensorflow_datasets/movielens/100k-movies/0.1.0. Subsequent calls will reuse this data.

Build vocabularies to convert user ids and movie titles into integer indices for embedding layers:

user_ids_vocabulary = tf.keras.layers.experimental.preprocessing.StringLookup(mask_token=None)
user_ids_vocabulary.adapt(ratings.map(lambda x: x["user_id"]))

movie_titles_vocabulary = tf.keras.layers.experimental.preprocessing.StringLookup(mask_token=None)
movie_titles_vocabulary.adapt(movies)

Define a model

We can define a TFRS model by inheriting from tfrs.Model and implementing the compute_loss method:

class MovieLensModel(tfrs.Model):
  # We derive from a custom base class to help reduce boilerplate. Under the hood,
  # these are still plain Keras Models.

  def __init__(
      self,
      user_model: tf.keras.Model,
      movie_model: tf.keras.Model,
      task: tfrs.tasks.Retrieval):
    super().__init__()

    # Set up user and movie representations.
    self.user_model = user_model
    self.movie_model = movie_model

    # Set up a retrieval task.
    self.task = task

  def compute_loss(self, features: Dict[Text, tf.Tensor], training=False) -> tf.Tensor:
    # Define how the loss is computed.

    user_embeddings = self.user_model(features["user_id"])
    movie_embeddings = self.movie_model(features["movie_title"])

    return self.task(user_embeddings, movie_embeddings)

Define the two models and the retrieval task.

# Define user and movie models.
user_model = tf.keras.Sequential([
    user_ids_vocabulary,
    tf.keras.layers.Embedding(user_ids_vocabulary.vocab_size(), 64)
])
movie_model = tf.keras.Sequential([
    movie_titles_vocabulary,
    tf.keras.layers.Embedding(movie_titles_vocabulary.vocab_size(), 64)
])

# Define your objectives.
task = tfrs.tasks.Retrieval(metrics=tfrs.metrics.FactorizedTopK(
    movies.batch(128).map(movie_model)
  )
)

Fit and evaluate it.

Create the model, train it, and generate predictions:

# Create a retrieval model.
model = MovieLensModel(user_model, movie_model, task)
model.compile(optimizer=tf.keras.optimizers.Adagrad(0.5))

# Train for 3 epochs.
model.fit(ratings.batch(4096), epochs=3)

# Use brute-force search to set up retrieval using the trained representations.
index = tfrs.layers.ann.BruteForce(model.user_model)
index.index(movies.batch(100).map(model.movie_model), movies)

# Get some recommendations.
_, titles = index(np.array(["42"]))
print(f"Top 3 recommendations for user 42: {titles[0, :3]}")
Epoch 1/3
25/25 [==============================] - 3s 120ms/step - factorized_top_k: 0.0291 - factorized_top_k/top_1_categorical_accuracy: 9.0000e-05 - factorized_top_k/top_5_categorical_accuracy: 0.0014 - factorized_top_k/top_10_categorical_accuracy: 0.0047 - factorized_top_k/top_50_categorical_accuracy: 0.0424 - factorized_top_k/top_100_categorical_accuracy: 0.0970 - loss: 33101.8998 - regularization_loss: 0.0000e+00 - total_loss: 33101.8998
Epoch 2/3
25/25 [==============================] - 3s 114ms/step - factorized_top_k: 0.0668 - factorized_top_k/top_1_categorical_accuracy: 1.4000e-04 - factorized_top_k/top_5_categorical_accuracy: 0.0047 - factorized_top_k/top_10_categorical_accuracy: 0.0140 - factorized_top_k/top_50_categorical_accuracy: 0.1053 - factorized_top_k/top_100_categorical_accuracy: 0.2099 - loss: 31015.9617 - regularization_loss: 0.0000e+00 - total_loss: 31015.9617
Epoch 3/3
25/25 [==============================] - 3s 114ms/step - factorized_top_k: 0.0886 - factorized_top_k/top_1_categorical_accuracy: 4.1000e-04 - factorized_top_k/top_5_categorical_accuracy: 0.0076 - factorized_top_k/top_10_categorical_accuracy: 0.0223 - factorized_top_k/top_50_categorical_accuracy: 0.1438 - factorized_top_k/top_100_categorical_accuracy: 0.2689 - loss: 30420.4044 - regularization_loss: 0.0000e+00 - total_loss: 30420.4044
Top 3 recommendations for user 42: [b'Rent-a-Kid (1995)' b'Farewell to Arms, A (1932)'
 b'Winnie the Pooh and the Blustery Day (1968)']