MusiCNN reimplementation in PyTorch - so that it's easier for me to train with dataset subsets for testing limited-data scenarios. Mel spectrogram calculation is done online in GPU (so no dataset preprocessing needed). The dataset(s) are automatically downloaded if they aren't there on your first run.
Pieced together from:
Original inference | Original training | PyTorch implementation | mir_ref
Paper:
@inproceedings{pons2019musicnn,
title={musicnn: pre-trained convolutional neural networks for music audio tagging},
author={Pons, Jordi and Serra, Xavier},
booktitle={Late-breaking/demo session in 20th International Society for Music Information Retrieval Conference (LBD-ISMIR2019)},
year={2019},
}