FATE-LLM: Empowering Large Language Models with Federated Learning
Talkat KubeCon + CloudNativeCon + Open Source Summit China 2023 · on Wednesday, 2023-09-27· at 14:45· 3层 307会议室| 3F Room 307·
summary
Describes FATE, an industrial federated learning platform and its FATE-LLM module that enables training large language models (examples given: ChatGLM and LLaMA) without sharing raw data. Explains the components FATE-LLM adds to handle LLM-specific efficiency and security challenges in the federated setting, and presents real-world experiments, evaluations and a future roadmap.
AI-written summary · about
related talks
20mFwdLLM: Efficient Federated Finetuning of Large Language Models with Perturbed...
34mLFN AI Taskforce: Platform-Driven Fede...
49mA Practical Approach to Using LLMs: Peculiarities and Pitfalls
51mLarge Language Models: Friend, Foe, or Otherwise
47mA Step-by-Step Guide to Securing Large Language Models (LLMs)
01h 00mFedOpt for LLMs
38mUnraveling Multimodality with Large Language Models
01h 01mFederated Learning in the Generative AI Era