https://github.com/declare-lab/sat
Code for the EMNLP 2022 Findings short paper "SAT: Improving Semi-Supervised Text Classification with Simple Instance-Adaptive Self-Training"
https://github.com/declare-lab/sat
Last synced: about 1 year ago
JSON representation
Code for the EMNLP 2022 Findings short paper "SAT: Improving Semi-Supervised Text Classification with Simple Instance-Adaptive Self-Training"
- Host: GitHub
- URL: https://github.com/declare-lab/sat
- Owner: declare-lab
- Created: 2022-10-20T16:09:45.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2023-02-25T16:44:51.000Z (over 3 years ago)
- Last Synced: 2025-03-27T18:21:48.953Z (about 1 year ago)
- Language: Jupyter Notebook
- Homepage:
- Size: 63.9 MB
- Stars: 13
- Watchers: 2
- Forks: 4
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# SAT: Improving Semi-Supervised Text Classification with Simple Instance-Adaptive Self-Training
This repository contains the official implementation code of the EMNLP 2022 Findings short paper [SAT: Improving Semi-Supervised Text Classification with Simple Instance-Adaptive Self-Training](https://arxiv.org/pdf/2210.12653v1.pdf).
## Usage
1. Set up the environment
```
conda create -n sat python==3.7.5
conda activate sat
cd SAT/
pip3 install -r requirements.txt
```
2. Training
```
cd src/
bash run.sh
```
The parameters.txt shows a list of hyper-parameters.
## Citation
Please cite our paper if you find our work useful for your research:
```bibtex
@inproceedings{chen-etal-2022-sat,
title = "{SAT}: Improving Semi-Supervised Text Classification with Simple Instance-Adaptive Self-Training",
author = "Chen, Hui and Han, Wei and Poria, Soujanya",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022",
month = dec,
year = "2022",
address = "Abu Dhabi, United Arab Emirates",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.findings-emnlp.456",
pages = "6141--6146",
}
```
## Contact
Should you have any questions, feel free to contact [chchenhui1996@gmail.com](chchenhui1996@gmail.com).