This is the repository for the extended abstract "Eigenpruning", to be presented at the LatinX workshop at NAACL 2024.

The main.py file contains the code to run the experiments described in the paper. These experiments requires poetry to be installed (to handle all the python dependencies).

The implementation of the algorithm is split in different files, all in the training directory.

Although this work is still in progress, we decided to leave this repository as is, to be faithful to the original submission. We'll eventually link the final version of the code here.