Skip to content

Create training & dev sets

In order to move forward with experiments, we need to split the available data into training and validation distributions.

Steps to complete the task might include the following:

  • Coming up with a sampling (and potentially subsampling) method
  • Develop solid reasoning behind splits
  • Potentially create reproducible sampling methods

Additional data sets for validation purposes (last.fm, AOTM, etc.) will be added at later point.

Edited by Andreas Rubin-Schwarz
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information