Using a Joint-Embedding Predictive Architecture for Symbolic Music Understanding
Rafik Hachana · Bader Rasheed
Abstract
We adapt Joint-Embedding Predictive Architectures to symbolic music by adding new masking methods and loss terms. The model achieves competitive results with much less training compute on a few downstream tasks, while falling short on a few others. This is mainly attributed to a bias towards positional information in the learned representation.
Chat is not available.
Successful Page Load