Summary

With the rise of computing power, deep neural networks, transformers, GANs, and VAEs model the complexity of real-world data much more effectively than previous generations of models, pushing the boundaries of what’s possible with AI algorithms. In this chapter, we explored the recent history of DL and AI and generative models such as LLMs and GPTs, together with the theoretical ideas underpinning them, especially the Transformer architecture. We also explained the basic concepts of models for image generation, such as the Stable Diffusion model, and finally discussed applications beyond text and images, such as sound and video.

The next chapter will explore the tooling of generative models, particularly LLMs, with the LangChain framework, focusing on the fundamentals, the implementation, and the use of this particular tool in exploiting and extending the capability of LLMs.