Nb. days
3
Face-to-face?
Yes
Remote?
Yes
Min. participants
2
Max. participants
8
Place
Toulouse, Paris, etc.
Pricing
2290€ HT
Large Language Models (LLMs) are transforming natural language processing (NLP).
These models, based on the Transformer architecture, offer significant advantages in terms of understanding and generating natural language.
However, these models still have many challenges to overcome. They have limited knowledge of the data they were trained on and can sometimes generate incorrect or fabricated information, known as hallucinations. Not to mention that training and using these LLMs can be expensive in terms of computing power and resources.
Retrieval Augmented Generation (RAG) marks an important step in the evolution of NLP by making it possible to overcome the limitations of traditional language models. Thanks to the integration of external knowledge, these systems make it possible to obtain more precise, factual and relevant answers by relying on up-to-date and verified information.
These models, based on the Transformer architecture, offer significant advantages in terms of understanding and generating natural language.
However, these models still have many challenges to overcome. They have limited knowledge of the data they were trained on and can sometimes generate incorrect or fabricated information, known as hallucinations. Not to mention that training and using these LLMs can be expensive in terms of computing power and resources.
Retrieval Augmented Generation (RAG) marks an important step in the evolution of NLP by making it possible to overcome the limitations of traditional language models. Thanks to the integration of external knowledge, these systems make it possible to obtain more precise, factual and relevant answers by relying on up-to-date and verified information.
By the end of this training, you will understand how to make language models more accurate, reliable, and versatile, enabling you to design and implement RAG systems that enhance chatbots, search engines, virtual assistants, and numerous other applications.
Goals
- Understand the theoretical and practical foundations of LLMs
- Know the challenges and limitations of LLMs
- Master the fundamentals of RAG architecture
- Build high-performance RAG systems, from indexing to generation
- Know how to evaluate and optimize RAG systems using advanced techniques
- Develop LLM-based applications and agents with LangChain
To know the dates of the next training sessions,
to adapt this program to your needs
or to obtain additional information,
contact us!
to adapt this program to your needs
or to obtain additional information,
contact us!
Teaching method
- This training is 70% practical and uses illustrated and didactic exercises.
- A daily evaluation of the acquisition of the knowledge of the day before is carried out.
- A synthesis is proposed at the end of the training.
- An evaluation will be proposed to the trainee at the end of the course.
- A course support will be given to each participant including slides on the theory and exercises.
- A sign-in sheet for each half-day of attendance is provided at the end of the course as well as a certificate of completion if the trainee has attended the entire session.
- A follow-up and an exchange with the participants will take place a few days after the training.
Prerequisite
Knowledge of the Python language is necessary. Basic knowledge of machine learning is required.Targets
- Data Engineers
- Data Scientists
- Développeurs
Evaluation
This training does not require a formal assessment of learningProgram
Large Language Models (LLM)
- History of NLP
- Neural networks for processing text
- LLM Architecture: Transformers
- Attention Mechanism
- Tokens & Embeddings
- Overview of some models
- Learning and optimization methods: Self Supervised Learning, Supervised fine-tuning (SFT), Reinforcement Learning from Human Feedback (RLHF), Group Relative Policy Optimization (GRPO), Fine tuning DPO
- Cost and ecological footprint of LLMs
- From LLMs to SLMs
LangChain
- Presentation of LangChain and its functionalities
- LangChain Ecosystem
Fundamentals of RAG
- Limitations of LLMs
- Definition of key concepts
- Architecture
Injection Pipeline
- Tokens
- Chunking
- Embeddings
- Vector Database
Generation Pipeline
- Search Algorithm
- Memory and context management
- Prompting techniques: zero-shot, few-shot, ReAct, Chain-of-Thought, Tree-of-Thought, etc.
- Categorization of LLMs and suitability for RAG
Evaluation of RAG Systems
- Evaluation principles
- Evaluation metrics
- Evaluation frameworks
Optimization of RAG Systems
- Advanced RAG
- Modular RAG
- AgenticAI
