FATE-LLM

FATE-LLM is a framework to support federated learning for large language models(LLMs).

Design Principle

• Federated learning for heterogeneous large model and small model.
• Promote training efficiency of federated LLMs using Parameter-Efficient methods.
• Protect the IP of LLMs using FedIPR.
• Protect data privacy during training and inference through privacy preserving mechanisms.

Deployment

Standalone deployment

Please refer to FATE-Standalone deployment.

• To deploy FATE-LLM v2.0, deploy FATE-Standalone with version >= 2.1, then make a new directory {fate_install}/fate_llm and clone the code into it, install the python requirements, and add {fate_install}/fate_llm/python to PYTHONPATH
• To deploy FATE-LLM v1.x, deploy FATE-Standalone with 1.11.3 <= version < 2.0, then copy directory python/fate_llm to {fate_install}/fate/python/fate_llm

Cluster deployment

Use FATE-LLM deployment packages to deploy, refer to FATE-Cluster deployment for more deployment details.

Quick Start

• Federated ChatGLM3-6B Training
• Builtin Models In PELLM
• Offsite Tuning Tutorial
• FedKSeed