VER🗣️

The official repo for "VER: Unifying Verbalizing Entities and Relations" (Findings of EMNLP '23)

Introduction

We pretrain VER🗣️: A Unified Model for Verbalizing Entities and Relations. The model takes any entity or entity set as input and generates a sentence to represent the entities and relations.

Requirements

See requirements.txt.

Models

We release the [RE]VER-base models on this link and the [RE]VER-large models on this link.

Data

The WiV data are available on this link.

Preprocessing

You may refer to the preprocessing script on this repo.

Training

bash train.sh

You can find other scripts (e.g., evaluation) on this repo.

Citation

The details of this repo are described in the following paper:

@inproceedings{huang2023ver,
  title={VER: Unifying Verbalizing Entities and Relations},
  author={Huang, Jie and Chang, Kevin Chen-Chuan},
  booktitle={Findings of the Association for Computational Linguistics: EMNLP 2023},
  pages={15700--15710},
  year={2023}
}