This repository contains the code for the paper Rethinking Word-Level Auto-Completion in Computer-Aided Translation, EMNLP 2023
Create the environment using conda:
conda create -n wlac python=3.6
conda activate wlac
pip install -r requirements.txt
export PROJECT_ROOT=path_to_code/WLAC-Joint-Training
Enter the run script directory. For AIOE-BPE model please refer to scripts/aioe_bpe
cd scripts/aioe
Please check all the related files and modify the path to the actual path
bash ./preprocess_training_set.sh
bash ./preprocess_valid_set.sh
For AIOE model:
bash ./train_aioe.sh
For AIOE-Joint model:
bash ./train_aioe_joint.sh
bash ./eval.sh
Please cite as:
@article{chen2023rethinking,
title={Rethinking Word-Level Auto-Completion in Computer-Aided Translation},
author={Chen, Xingyu and Liu, Lemao and Huang, Guoping and Zhang, Zhirui and Yang, Mingming and Shi, Shuming and Wang, Rui},
journal={arXiv preprint arXiv:2310.14523},
year={2023}
}