A pytorch implementation of the paper: Self-Attention Enhanced Selective Gate with Entity-Aware Embedding for Distantly Supervised Relation Extraction.
- python==3.8
- pytorch==1.6
- numpy==1.19.1
- tqdm==4.48.2
- scikit_learn==0.23.2
Download the dataset from here, and unzip it under ./data/
.
python main.py
AUC | P@100 | P@200 | P@300 | Mean |
---|---|---|---|---|
0.452 | 0.810 | 0.790 | 0.763 | 0.772 |
PCNN and SAN DO NOT share the same entity-aware embedding layer, and the lambda
values for PCNN and SAN are 0.05 and 1.0 respectively (confirmed by the authors).