Research

Retrieval-Augmented Generation (RAG), a framework that enhances pre-trained language models with external knowledge retrieval to improve accuracy and performance on complex, knowledge-intensive NLP tasks. Read more →

A comprehensive survey of Retrieval-Augmented Generation (RAG), detailing its evolution, current landscape, and future directions in enhancing language models with retrieval mechanisms for improved accuracy in knowledge-intensive tasks. Read more →

A comprehensive survey of trustworthiness in Retrieval-Augmented Generation (RAG) systems, evaluating key dimensions such as factuality, robustness, fairness, transparency, accountability, and privacy, and providing practical insights for enhancing RAG systems in real-world applications. Read more →

This paper introduces C-FedRAG, a system that integrates confidential computing techniques into federated retrieval-augmented generation workflows, enabling secure and scalable connections across decentralized data providers while ensuring context confidentiality. Read more →

Recent methods for privately adapting closed Large Language Models (LLMs) and concludes that open-source LLMs offer superior privacy protection, performance, and cost-effectiveness for enterprise AI applications. Read more →