Document Packing Impacts LLMs' Multi-Hop Reasoning

This ArXiv paper likely explores how different document organization strategies affect the ability of Large Language Models (LLMs) to perform multi-hop reasoning. The research offers insights into optimizing input formatting for improved performance on complex reasoning tasks.

• •Focuses on how document structure impacts LLM reasoning.
• •Investigates multi-hop reasoning capabilities.
• •Potentially reveals ways to improve LLM performance.