We're extremely excited to announce that Hugging Face's AI Agent Course 🤗 is now available in French!
The content, which is completely free, ad-free and aimed at everyone, is divided into 4 units (+ 3 bonuses) 📚
The first is a “theoretical” introduction to AI agents 🤵 and covers:
- The basics: what is an agent? What is a LLM? What is a tool? What are special messages and tokens?
- The Thought-Action-Observation cycle and the ReAct approach ;
- How to create your first agent?
The second is an introduction to Python 🐍 libraries for implementing agents, with:
- A focus on Hugging Face's smolagents: CodeAgent, ToolCallingAgent, multi-agent system, vision agent, ...
- A focus on LlamaIndex from the company of the same name: LlamaHub, Components & Tools, agentic workflow, ...
- A focus on LangChain's LangGraph: building blocks, document analysis, ...
The third focuses on agentic RAG 🐶:
- How it works ;
- How to implement it in each of the 3 libraries via a use case (organizing a party).
The last one is about the final project 🎓:
- A description of the GAIA leaderboard used for the assessment ;
- How to get your certificate if you pass the project.
3 bonuses 🎁 are also available to show you how to:
- Finetuning an LLM to make function calls ;
- Observe and evaluate an agent ;
- Create an agent to play Pokémon.
All the content can be viewed through the 74 pages of the Hugging Face website: https://huggingface.co/learn/agents-course/fr/ and the 16 Jupyter notebooks 📓 : https://huggingface.co/agents-course/notebooks/tree/main/fr/
Many thanks to Kim NOEL and Loïck BOURDOIS for their translation efforts.
Many thanks also to the course instructors: Ben BURTENSHAW, Sergio PANIEGO BLANCO, Joffrey THOMAS and Thomas SIMONINI.
We hope you find it useful and enjoyable.
Good reading!