LLMs in practice

So far, this chapter has mainly discussed the theoretical foundations of LLMs. Let’s close this chapter with an overview of the LLM landscape as it stands today, discussing some considerations for choosing an appropriate LLM as well as different techniques to tailor the model’s responses to your needs.

The evolving field of LLMs

Generative AI and LLMs are a rapidly changing field, with new models, frameworks, and research papers on the topic released frequently. Most of the know-how to train an LLM is publicly available, yet at the time of writing, the cost of training a state-of-the-art LLM from scratch is still in the order of tens to hundreds of millions of US dollars, due to the large amount of GPU compute resources needed. This cost puts training your own model out of reach of individuals and most smaller companies, who will have to rely on pre-trained LLMs.

The most competent models as of the time of writing, namely OpenAI’s GPT...