Summary of Leveraging Unstructured Text Data For Federated Instruction Tuning Of Large Language Models, by Rui Ye et al.
Leveraging Unstructured Text Data for Federated Instruction Tuning of Large Language Models
by Rui Ye, Rui Ge, Yuchi Fengting, Jingyi Chai, Yanfeng Wang, Siheng Chen
First submitted to arxiv on: 11 Sep 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Federated instruction tuning is a machine learning approach that enables multiple clients to collaborate in fine-tuning a shared large language model (LLM) without sharing raw data. The existing literature requires all clients to have structured instruction-response pairs, which necessitates massive human annotations since clients’ data is usually unstructured text. To address this limitation, we propose FedIT-U2S, a novel framework that can automatically transform unstructured corpus into structured data for federated instruction tuning. FedIT-U2S consists of two key steps: few-shot instruction-tuning data generation and retrieval-based example selection technique. The generated data is then used in a typical federated instruction tuning process. We conduct experiments on three domains (medicine, knowledge, and math) showing that our proposed framework can consistently improve over the base LLM. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine a way for many people to work together to make a language model smarter without sharing their personal data. This is called federated instruction tuning. But right now, it’s hard because each person needs to have special instructions and responses ready. That’s why we created a new way to turn unstructured text into structured instructions and responses. Our method, FedIT-U2S, can be used in many different situations as long as people have valuable text. We tested our method on three topics (medicine, knowledge, and math) and showed that it works better than the usual approach. |
Keywords
» Artificial intelligence » Few shot » Fine tuning » Instruction tuning » Language model » Large language model » Machine learning
