maciejkula

spotlight: Build PyTorch Recommender Models

Spotlight is a Python library for prototyping recommendation models, from matrix factorization to sequence-based ranking. It includes losses, dataset utilities, and evaluation tools, but its inactive maintenance makes it a better fit for research or existing projects than new production dependencies.

View on GitHub Compare
Stars
3k
Forks
416
Stars, last 30 days
-4
Last commit
3.8 years ago
Contributors
9

Our take

Inactiveno commits in over a year

Health score

24/100

How it is scored
Activity
19
Community
34
Issues
21
Pull requests
20

-4 stars in the last 30 days

Spotlight is a focused research and prototyping toolkit, but its long inactivity and highly concentrated contribution history make it a risky choice for a new, maintained production dependency.

Good fit if

  • You need Python/PyTorch examples or components for explicit, implicit, or sequential recommendation experiments.
  • You can work with the existing code and maintain or adapt it yourself, given there have been no recent commits or releases.
  • You want an MIT-licensed starting point for research or a prototype.

Look elsewhere if

  • You need an actively maintained recommender library, since the project is marked inactive and has had no recent issue closures or merged pull requests.
  • Your team cannot absorb maintenance risk from a project where most contributions come from one person.
  • You need capabilities beyond the model types, datasets, and utilities documented by the project.
All health signals
Last commit2022-12-21 (3.8 years ago)
Commits, last 90 days0
Releases, last 12 months0 (latest v0.1.6, 2019-09-08)
Contributors9 (top contributor: 84% of commits)
Issues closed, last 90 days0
Pull requests merged, last 90 days0
Project age9 years

Checked on 2026-10-08 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

3,044

Forks

416

Change since Nov 17, 2026: 0 stars·0 forks

Overview

Spotlight is a PyTorch library for building and evaluating recommendation models. It covers both explicit ratings and implicit interactions, helping researchers and developers compare approaches without assembling every modeling and dataset component from scratch.

Its model choices range from factorization methods to sequence models for next-item prediction. The project is most useful for experimentation and prototyping; it has been inactive for years, so adopting it in a new production system carries maintenance risk.

Key Features

  • Explicit-feedback factorization models for rating prediction.
  • Implicit-feedback factorization models with pairwise ranking losses such as BPR.
  • Sequence recommenders using pooling, LSTMs, or causal convolutions.
  • Utilities to load MovieLens datasets and generate synthetic interaction data.
  • Train-test split helpers for interaction and sequence workflows.
  • Evaluation metrics including RMSE, MRR, and sequence MRR.
  • PyTorch-based building blocks intended to support experimentation with recommendation models.

Use Cases

  • A recommender-systems researcher can compare rating-prediction and ranking approaches on MovieLens data.
  • A developer exploring next-item recommendation can prototype pooling, recurrent, or convolutional sequence models.
  • An educator can demonstrate matrix factorization, pairwise ranking, and sequence prediction with a Python codebase.
  • A team investigating recommendation algorithms can generate synthetic interactions to test model behavior on controlled data.

What you need

Detected in the repository

  • Python (from setup.py)
  • A test suite

License in plain words

MITpermissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the copyright and license notice
  • Share your changes: no

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

Install from the project's Conda channel:

conda install -c maciejkula -c pytorch spotlight

See the README and documentation for model and dataset examples.

Considerations

The repository is inactive: it has had no commits in over a year, its latest release dates to 2019, and recent issue and pull-request activity is absent. While nine contributors are recorded, most of the code comes from one person, which may make fixes and compatibility updates difficult to sustain. The project uses Python and provides tests, but the supplied information does not specify supported Python or PyTorch versions. It is MIT licensed; review the license terms when incorporating it into a product.

Found this useful?

Share it with someone who would like spotlight.

Comparisons