https://github.com/smilelab-fl/fedlab-nlp
FedLab-NLP: A Fast and highly Customizable Framework for Federated Learning in Natural Language Processing
https://github.com/smilelab-fl/fedlab-nlp
Last synced: about 2 months ago
JSON representation
FedLab-NLP: A Fast and highly Customizable Framework for Federated Learning in Natural Language Processing
- Host: GitHub
- URL: https://github.com/smilelab-fl/fedlab-nlp
- Owner: SMILELab-FL
- License: apache-2.0
- Created: 2022-07-18T08:36:34.000Z (almost 3 years ago)
- Default Branch: main
- Last Pushed: 2022-08-19T03:12:47.000Z (almost 3 years ago)
- Last Synced: 2024-05-12T00:45:21.007Z (about 1 year ago)
- Language: Python
- Size: 97.7 KB
- Stars: 9
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
FedLab-NLP
致力于联邦自然语言处理发展
1️⃣ 本项目来自 [SMILELab-FL](https://github.com/SMILELab-FL/FedLab-NLP) , 我们主要致力于联邦自然语言处理的发展；
2️⃣ `FedLab-NLP`是一款轻量级、可自由定制联邦学习过程并面向于自然语言处理的代码框架；
3️⃣ `FedLab-NLP`主要实现基于 [FedLab](https://github.com/SMILELab-FL/FedLab) 和 [HuggingFace](https://github.com/huggingface/transformers)，能完成自然语言预训练大模型在联邦学习场景下工作；
4️⃣ 任何使用问题可以联系 :e-mail: [email protected]
## Installation
### 目录组织
我们建议按照如下的文件结构来使用`FedLab-NLP`:
data: 存放数据集，注意该数据集需要已经经过 iid/niid 划分的联邦数据集;
pretrained: 在nlp文件夹里存储从huggingface中下载的预训练模型，如`bert-base-uncased`;
code: 在该文件夹下面使用`FedLab-NLP`;
output: 存储模型输出的记录等。
目录组织如下：
```grapha
├── workspace
│ └── data
| | ├── fedglue
| | └── fedner
│ ├── pretrained
│ │ ├── nlp
│ │ └── cv
│ ├── output
│ └── code
│ └── FedLab-NLP
```
运行路径生成：
```bash
mkdir workspace
cd workspace
mkdir data
mkdir code
mkdir pretrained
cd pretrained
mkdir nlp
cd ..
cd code
```
### 环境安装
建议运行环境的`python`版本为`3.7+`，我们建议使用`pytorch`版本为`1.10+`
```bash
git clone [email protected]:Trustworthy-DL/FedLab-NLP.git
cd FedLab-NLP
pip install -r resquirements.txt
```
## Usage
支持 `fedavg` 联邦学习算法
```bash
bash fed_run.sh {your_file_path}/workspace {task_name} fedavg 10001 {server_gpu_id} {client1_gpu_id} {client2_gpu_id}
```
支持 `centralized` 集中训练学习算法
```bash
bash cen_run.sh {your_file_path}/workspace {task_name} centralized 10001 {server_gpu_id}
```
## Contributing
## License
[Apache License 2.0](https://github.com/SMILELab-FL/FedLab-NLP/blob/main/LICENSE)