Loading Now

Summary of Mdcure: a Scalable Pipeline For Multi-document Instruction-following, by Gabrielle Kaili-may Liu et al.


MDCure: A Scalable Pipeline for Multi-Document Instruction-Following

by Gabrielle Kaili-May Liu, Bowen Shi, Avi Caciularu, Idan Szpektor, Arman Cohan

First submitted to arxiv on: 30 Oct 2024

Categories

  • Main: Computation and Language (cs.CL)
  • Secondary: Machine Learning (cs.LG)

     Abstract of paper      PDF of paper


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
This paper proposes a novel fine-tuning pipeline called MDCure for enhancing the multi-document processing capabilities of Large Language Models (LLMs) without relying on pre-training or human-annotated data. The approach generates high-quality synthetic instruction data from related articles using targeted prompts, and further introduces a multi-objective reward model to filter generated data based on their training utility. The authors fine-tune various LLMs, including FlanT5, Qwen2, and LLAMA3.1 models up to 70B parameters in size, and demonstrate consistent performance improvements over pre-trained baselines and base models by up to 75.5% on a range of multi-document and long-context benchmarks.
Low GrooveSquid.com (original content) Low Difficulty Summary
This paper helps computers understand many documents at once better. Currently, these computers (called Large Language Models or LLMs) struggle with this task because they don’t know how to handle the relationships between different documents, repetitive information, and confusing structures. The researchers created a new way to train these computers using fake data that looks like real articles. They also developed a system to decide which fake data is most helpful for learning about multiple documents. By training different types of LLMs with this new approach, they showed that it can improve the computer’s performance by up to 75.5% on various tasks.

Keywords

* Artificial intelligence  * Fine tuning