Federated GPT.

Implementing a federated learning on nanoGPT.

• The experiments and other rough implementations of the code can be found here. fedGPT.
• This repository consists of the packaged code of the above experiments.

Run the Server

cd server/
docker build -t federated-server .
docker run -it -p 8080:8080 federated-server:latest

Run the Client

cd client/
docker build -t federated-client .
docker run -it -p 8080:8080 federated-client:latest

File Structure

.
├── README.md
├── client
│   ├── Dockerfile
│   ├── __init__.py
│   ├── client.py
│   ├── main.py
│   ├── models
│   │   ├── __init__.py
│   │   ├── dataset.py
│   │   ├── engine.py
│   │   └── gpt.py
│   ├── run-client.sh
│   └── utils.py
├── poetry.lock
├── pyproject.toml
└── server
    ├── Dockerfile
    ├── __init__.py
    ├── main.py
    ├── model.pth
    ├── models
    │   ├── __init__.py
    │   ├── dataset.py
    │   ├── engine.py
    │   └── gpt.py
    ├── requirements.txt
    ├── run-server.sh
    ├── strategy.py
    └── utils.py

5 directories, 25 files

Run