spellbrush/PupuJEPA

Frequency-Aware Self-Supervised Music Representation Learning

This is the official Hugging Face model repository for the paper "Frequency-Aware Self-Supervised Music Representation Learning".

• Project Page: https://www.yichenggu.com/PupuJEPA/
• GitHub Repository: https://github.com/sizigi/PupuJEPA

Model Checkpoints

This repository contains the following checkpoints:

Model Name	Directory	Description
PupuJEPA-Tiny	./PupuJEPAV2_25Hz_tiny/*	5M parameter tiny version of PupuJEPA.
PupuJEPA-Small	./PupuJEPAV2_25Hz_small/*	22M parameter small version of PupuJEPA.
PupuJEPA-Base	./PupuJEPAV2_25Hz_base/*	86M parameter base version of PupuJEPA.
PupuJEPA-Large	./PupuJEPAV2_25Hz_large/*	307M parameter large version of PupuJEPA.
PupuJEPA-Huge	./PupuJEPAV2_25Hz_huge/*	632M parameter huge version of PupuJEPA.
PupuJEPA-giant	./PupuJEPAV2_25Hz_giant/*	1.1B parameter giant version of PupuJEPA.

How to use

You need to put the pretrained models in:

  PupuJEPA/experiments

of our official repository.

Citation

@article{PupuJEPA,
  title        = {Frequency-Aware Self-Supervised Music Representation Learning},
  author       = {Yicheng Gu and Junan Zhang and Jerry Li and Zhizheng Wu and Lauri Juvela},
  year         = {2026},
  journal      = {arXiv:2606.25713},
}