Paper ID: 11298
Title: "Prefer to Classify: Collecting and Modeling Annotators' Disagreements via Preference Learning"

- In this repository, we include the codes for
    - (1) extractive preference generation ("Preference Generation (P2C).ipynb")
    - (2) training with P2C ("train_pref.py")

- Since the training code (2) requires the generated files from, (1) should be conducted before (2)
- For the convenience, we include the training script of P2C in "run.sh"
