CLAVÉ, M. Mathieu (2024) Improvement of a Conversational Agent: Retrieval-Augmented Generation PFE - Project Graduation, ENSTA.

[img]
Preview
PDF
544Kb

Abstract

As part of my final-year project at ENSTA Paris, I completed an internship at CEA CESTA starting in June 2024, following a specialization in Artificial Intelligence. The primary objective of this internship was to develop a Retrieval-Augmented Generation (RAG) system to enhance the performance of a conversational agent based solely on a Large Language Model (LLM). This report presents the work carried out during the internship, beginning with the context, the role of CEA, and the objectives pursued. It also addresses the methods and tools employed, such as leveraging scientific documents to power the RAG system. Subsequently, the report explains the principle of Retrieval-Augmented Generation and details the different processing stages involved. The main challenges encountered are also analyzed. Finally, the solutions implemented to address these challenges are discussed, along with an evaluation of the results obtained and prospects for future improvements.

Item Type:Thesis (PFE - Project Graduation)
Subjects:Information and Communication Sciences and Technologies
ID Code:10492
Deposited By:M. Mathieu CLAVE
Deposited On:09 déc. 2024 17:28
Dernière modification:09 déc. 2024 17:28

Repository Staff Only: item control page