GRIT: Generative Role-filler Transformers for Document-level Event Entity Extraction (link) EACL'21
- Python 3.6.10
- Transformers: transformers 2.4.1 installed from source.
- Pytorch-Lightning==0.7.1
- Pytorch==1.4.0
- seqeval
./data/muc
, refer to./data/muc/README.md
for details
- import eval_ceaf,
from eval import eval_ceaf
, readeval.py
for details - eval on preds.out
python eval.py --pred_file model_gtt/preds_s_t.out
- run simple test case:
python test_cases_eval.py
- The encoder-decoder model (code is written based on hugging face transformers/examples/ner/3ee431d
- How to run: see README in model_gtt
If you use materials in this repo helpful, please cite.