Papers with Code

paperswithcode.com

Articles de machine learning associés à leurs implémentations et benchmarks — importez la page de l'article et le README ensemble.

It links a paper to its open-source implementation and to leaderboards showing how it compares against other methods on the same benchmark, so you can check a paper's claims against running code rather than the abstract alone. That combination is useful for an ML notebook comparing several approaches to one task. Its coverage leans toward recent, widely reproduced papers, so it's a weak source for older or less popular methods.