Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
WEN Hao e966049b67 | 5 months ago | |
---|---|---|
.. | ||
Datasets | 5 months ago | |
EvalBox | 6 months ago | |
Models | 5 months ago | |
configs | 6 months ago | |
test | 7 months ago | |
utils | 7 months ago | |
.gitignore | 7 months ago | |
Dockerfile | 1 year ago | |
ITC.py | 7 months ago | |
README.md | 1 year ago | |
TestingText.py | 7 months ago | |
__init__.py | 2 years ago | |
cli.py | 7 months ago | |
const.py | 7 months ago | |
no-test.Dockerfile | 1 year ago | |
post_docker_build.py | 7 months ago | |
requirements-apt.txt | 7 months ago | |
requirements-no-torch.txt | 7 months ago | |
requirements.txt | 7 months ago |
参见本项目文档
python TestingText.py /path/to/config.yml
配置文件的示例可见ExampleConfigFile.yml
python cli.py -r text_fooler --language zh -m roberta_dianping -d jd_binary --subset test
可以用如下命令查看命令行参数
python cli.py --help
一些重要的参数:
--time-out TIME_OUT
:单个样本处理时间上限,单位:分钟 (default: 5)--ignore-errors
:忽略所有错误。默认关闭。关闭情况下,如果算法内部发生错误,则直接报错 (default: False)详细参数列表:
本模块已构建好了Docker镜像,并已推送至Docker Hub. 用户可以直接拉取该镜像
docker pull wenh06/ai-testing-text:latest
并通过如下命令启动、进入该镜像进行使用
docker run -it --shm-size=10240m --gpus all -v ~/ai-testing-text-config-files:/AI-Testing/text/config_files -v ~/ai-testing-text-output:/AI-Testing/text/log wenh06/ai-testing-text bash
以上命令中的~/ai-testing-text-config-files
为需要用到的配置文件存放的文件夹,~/ai-testing-text-output
为算法执行日志,输出结果等存放的文件夹。
text_fooler:
max_candidates: 30
min_cos_sim: 0.5
use_threshold: 0.8
攻击方法默认参数配置文件:
本项目为AI-Testing开源项目的算法内核,是一个用于 计算机视觉、文本自然语言处理、音频处理 等多个领域的对抗攻击全流程评测算法学习研究的 Python库。
https://numbda.cs.tsinghua.edu.cn/AI-Testing/
Python Pickle Text Dockerfile Jupyter Notebook other
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》