Awesome
IRNet
Code for our ACL'19 accepted paper: Towards Complex Text-to-SQL in Cross-Domain Database with Intermediate Representation
<p align='center'> <img src='https://zhanzecheng.github.io/architecture.png' width="91%"/> </p>Environment Setup
Python3.6
Pytorch 0.4.0
or higher
Install Python dependency via pip install -r requirements.txt
when the environment of Python and Pytorch is setup.
Running Code
Data preparation
- Download Glove Embedding and put
glove.42B.300d
under./data/
directory - Download Pretrained IRNet and put
IRNet_pretrained.model
under./saved_model/
directory - Download preprocessed train/dev datasets from here and put
train.json
,dev.json
andtables.json
under./data/
directory
Generating train/dev data by yourself
You could process the origin Spider Data by your own. Download and put train.json
, dev.json
and
tables.json
under ./data/
directory and follow the instruction on ./preprocess/
Training
Run train.sh
to train IRNet.
sh train.sh [GPU_ID] [SAVE_FOLD]
Testing
Run eval.sh
to eval IRNet.
sh eval.sh [GPU_ID] [OUTPUT_FOLD]
Evaluation
You could follow the general evaluation process in Spider Page
Results
Model | Dev <br /> Exact Set Match <br />Accuracy | Test<br /> Exact Set Match <br />Accuracy |
---|---|---|
IRNet | 53.2 | 46.7 |
IRNet+BERT(base) | 61.9 | 54.7 |
Citation
If you use IRNet, please cite the following work.
@inproceedings{GuoIRNet2019,
author={Jiaqi Guo and Zecheng Zhan and Yan Gao and Yan Xiao and Jian-Guang Lou and Ting Liu and Dongmei Zhang},
title={Towards Complex Text-to-SQL in Cross-Domain Database with Intermediate Representation},
booktitle={Proceeding of the 57th Annual Meeting of the Association for Computational Linguistics (ACL)},
year={2019},
organization={Association for Computational Linguistics}
}
Thanks
We would like to thank Tao Yu and Bo Pang for running evaluations on our submitted models. We are also grateful to the flexible semantic parser TranX that inspires our works.
Contributing
This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.microsoft.com.
When you submit a pull request, a CLA-bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., label, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repositories using our CLA.
This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.