Recommender Systems Benchmarks

There are quite a few tasks that could be good to implement:

Movie (movielens) Recommendation, News (plista dataset), Million song dataset, and probably others.

The Recsys Challenge http://www.recsyschallenge.com is a good set of tasks to try? since they’re often reused and multiple teams participate.

For sure! Recommender systems would be great to have - especially in light of some of the research on reproducibility in this area in the past year (https://arxiv.org/pdf/1907.06902.pdf).

Are you aware of any public repositories that would be good to benchmark for their performance?