Retrieval Augmented Generation (RAG) systems have revolutionized the way Large-Language Models (LLMs) integrate domain-specific and time-sensitive data to enhance output. The evolution from basic setups to more complex forms of RAG is evident. This paper presents a robust dataset creation and evaluation workflow for comparing different RAG strategies and introduces a Boolean agent RAG setup that allows LLMs to decide when to query external databases.

• Comprehensive dataset creation for RAG comparison
• Evaluation of a novel Boolean agent RAG setup
• Publication of code and dataset for public use
• RAG advancements to reduce redundant information retrieval
• Fine-tuning LLM outputs with domain-specific data

The insights from this research are significant, illustrating how refined RAG systems can optimize LLM performance, especially in applications requiring quick, accurate, and context-aware responses. These advancements present numerous applications ranging from real-time decision-making systems to educational tools.