# JSFakes (Dr. Tristan Behrens).

This is a tokenized version of the JS-Fakes dataset by Omar Peracha. The original dataset can be found here:
[js-fakes.git](https://github.com/omarperacha/js-fakes.git)

The representation is four tracks with four bars per track.

## Purpose.

This dataset is a good starting point for Music Generation. You could train GPT-2 on the samples to compose music.

## Contact.

Find me on [LinkedIn](https://www.linkedin.com/in/dr-tristan-behrens-734967a2/) and say hello.

If you find and issue or have a feature request, please contact me. 

Please be so kind and like this dataset if you find it useful.