How to evaluate the performance of MERT on the melody extraction task using MARBLE benchmark?

[XaryLee]

Hi! Thank you for releasing this benchmark as an open-source resource, which provides a method for the evaluation of music audio representation models.

I want to evaluate the performance of MERT on the melody extraction task using MARBLE benchmark, as is mentioned in the paper of the benchmark. According to the paper, the model performances are evaluated on the MelodyDB dataset. However, I cannot find the corresponding codes and datasets in this repository. Therefore, I am wondering how to implement the benchmark on this task.