Let's try to create an agentic architecture to generate a fight, having the generation of the fight made by an LLM!
To execute the Agentic_fighters, you can do it in 2 ways:
- run the src/main.py file from the root directory
- move the Exercise 5 notebook to the root directory, and run it
An agent is a system that perceives its environment, makes decisions and takes actions autonomously.
Instead of programming a fixed control flow, we sometimes want LLM systems that can choose their own control flow to solve more complex problems:
- an LLM can decide between two potential routes
- an LLM can decide which of the many tools to call
- an LLM can decide whether the generated response is sufficient or if more work is needed
| Content | Time estimate | Description |
|---|---|---|
| Exercise 0 | 10 minutes | Getting your python environment ready |
| Exercise 1 | 10 minutes | LangChain basics |
| Exercise 2 | 10 minutes | Creating our own agent |
| Exercise 3 | 20 minutes | LangGraph for simplicity |
| Exercise 4 | 40 minutes | Architectures |
| Exercise 5 | ¿?¿?¿?¿? | Going further... AGENTIC FIGHTERS! |
- A Python 3.11 environment
- An OPENAI_API_KEY
- A TAVILY_API_KEY for exercise 3 (created for free in https://tavily.com/)
- Python
- Basic SQL