1 code implementation • 7 Oct 2022 • Matthew C. McCallum, Filip Korzeniowski, Sergio Oramas, Fabien Gouyon, Andreas F. Ehmann
We find that restricting the domain of the pre-training dataset to music allows for training with smaller batch sizes while achieving state-of-the-art in unsupervised learning -- and in some cases, supervised learning -- for music understanding.
