DA-GPN

Latent Graph Learning with Dual-channel Attention for Relation Extraction

Required Packages

• Python (tested on 3.8.12)
• CUDA (tested on 11.1)
• PyTorch (tested on 1.8.1)
• Transformers (tested on 3.4.0)
• ujson
• tqdm

TACRED Dataset

The TACRED dataset can be obtained from this link. The TACREV and Re-TACRED dataset can be obtained following the instructions in Tacrev and Re-TACRED, respectively. The expected structure of files is:

DA-GPN
 |-- dataset
 |    |-- tacred
 |    |    |-- train.json        
 |    |    |-- dev.json
 |    |    |-- test.json
 |    |    |-- dev_rev.json
 |    |    |-- test_rev.json
 |    |-- retacred
 |    |    |-- train.json        
 |    |    |-- dev.json
 |    |    |-- test.json

Training and Evaluation

Train the DA-GPN model:

>> sh run_tacred.sh    # TACRED and TACREV
>> sh run_retacred.sh  # Re-TACRED

The results on TACRED and TACREV can be obtained in one run as they share the same training set. We use Roberta large as the backbone of BERT module.

DialogRE Dataset

This DialogRE dataset can be downloaded at: https://github.com/nlpdata/dialogre. You can download and unzip BERT-base-uncased from https://github.com/google-research/bert

>> sh run_dialog.sh    # Dialog

Note: We perform our experiments on GTX 3090 card.

Part of the code is adapted from An Improved Baseline for Sentence-level Relation Extraction.