https://github.com/woooodyy/earlyrobust
Codes for the EMNLP 2022 paper "Efficient Adversarial Training with Robust Early-Bird Tickets" by Zhiheng Xi, Rui Zheng, Tao Gui, Qi Zhang and Xuanjing Huang.
https://github.com/woooodyy/earlyrobust
Last synced: 22 days ago
JSON representation
Codes for the EMNLP 2022 paper "Efficient Adversarial Training with Robust Early-Bird Tickets" by Zhiheng Xi, Rui Zheng, Tao Gui, Qi Zhang and Xuanjing Huang.
- Host: GitHub
- URL: https://github.com/woooodyy/earlyrobust
- Owner: WooooDyy
- Created: 2022-11-29T16:57:09.000Z (almost 3 years ago)
- Default Branch: master
- Last Pushed: 2022-11-29T17:00:07.000Z (almost 3 years ago)
- Last Synced: 2025-09-21T23:03:39.500Z (22 days ago)
- Language: Python
- Homepage: https://aclanthology.org/2022.emnlp-main.569/
- Size: 4.69 MB
- Stars: 2
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: readme.md
Awesome Lists containing this project
README
## EarlyRobust
- This is the implementation for the paper in EMNLP 2022 "Efficient Adversarial Training with Robust Early-Bird Tickets" by Zhiheng Xi, Rui Zheng, Tao Gui, Qi Zhang and Xuanjing Huang. The codes consist of two main parts: Robust Searching stage and drawing & finetuning stage.
- run searching stage: `sh run_first_stage_imdb.sh`
- run drawing & fine-tuning stage: `sh run_second_stage_imdb.sh`
## Environments
```
textattack==0.3.4
datasets==1.9.0
pytorch==1.9.0
python==3.7.10
transformers==4.4.2
......
```
- We recommend you to install `transformers==4.4.2` by the following steps:
```shell script
cd transformers-4.4.2
pip install -e .
```
## Citation
todo