Summary
In the previous chapters, the main question was how to find information and how to deliver it effectively to an LLM. In such cases, the model is a passive agent that receives information and responds. With this chapter, we are trying to move away from this paradigm, toward an idea where an agent explores an environment, learns through this exploration, performs actions, and learns from the feedback that the environment provides to it. In this view, the model is an active component that interacts with the environment and can modify it. This view is also much closer to how we humans learn. In our exploration of the external world, we receive feedback that guides us in our learning. Although much of the world has been noted in texts, the real world cannot be reduced to a textual description. Therefore, an agent cannot learn certain knowledge and skills without interacting with the world. RL is a field of artificial intelligence that focuses on an agent’s interactions with...
