4,175 views
Presentation by: Arnaud PICHERY, Aurélien Coquard ???? Summary: 45 minutes to understand (a little) how these algorithms manage to write poems or answer questions better than your grandmother. Everyone has nothing else on their lips: Generative AI. Among the most captivating models in this sphere are LLM and RAG (Retrieval-Augmented Generation). This technical talk aims to reveal the mechanisms and fundamental principles that drive these powerful AI architectures. Diving into Large-Scale Language Models (LLM) * Explanation of LLMs and how they work. * Overview of neural network architectures, such as Transformers, that support LLMs. * Exploration of LLM use cases: text writing, code generation, language translation, and more. RAG Approaches: Fundamentals and Innovations * Introduction to the concept of RAG and its importance. * How RAG merges the power of LLM with information retrieval techniques Image generation * Overview of how DALL-e and Midjourney work Recorded in April 2024 in Paris, Palais des Congrès, Porte Maillot. ???? To stay informed about Devoxx France news, follow us on linkedIn : / devoxx-france , twitter : / devoxxfr or visit our website https://www.devoxx.fr/