AONE-NLP / FSRE-SaCon

Dataset and code for Synergistic Anchored Contrastive Pre-training for Few-Shot Relation Extraction.
8 stars 1 forks source link

Synergistic Anchored Contrastive Pre-training for Few-Shot Relation Extraction

Overview

The Dataset and code for paper Synergistic Anchored Contrastive Pre-training for Few-Shot Relation Extraction, accepted by AAAI 2024.

1703063577738

Requirements

GPU=NVIDIA A100 Tensor Core
Python=3.7
Pytorch=1.13.0

Pre-training

cd pre-train/code
bash train.sh

Fine-tuning

Step1. Select the down-stream baseline and train

cd fine-tune
bash run_train.sh

Step2. Select the down-stream baseline and test

cd fine-tune
bash run_test.sh

Citation

@inproceedings{luo2024synergistic,
  title={Synergistic Anchored Contrastive Pre-training for Few-Shot Relation Extraction},
  author={Luo, Da and Gan, Yanglei and Hou, Rui and Lin, Run and Liu, Qiao and Cai, Yuxiang and Gao, Wannian},
  booktitle={Proceedings of the 38th AAAI Conference on Artificial Intelligence},
  volume={38},
  number={17},
  pages={18742--18750},
  year={2024}
}