This is the repository for relation-agnostic (entity-based) fact linking on ComFact benckmark.

Start with creating a python 3.6 venv and installing requirements.txt.

The ComFact dataset can be downloaded from this link, please place data/ under this root directory.

python process_nlu_head_tail_link.py

bash train_baseline_entity_linker.sh

Our trained entity linker based on DeBERTa (large) can be downloaded from this link, please place all-deberta-large-nlu-entity/ under the runs/ folder.

bash run_baseline_head_entity_link.sh
bash run_baseline_tail_entity_link.sh
python merge_linking.py