Developing AI applications with LLMs

LLMs are complex, large-scale models, and are dominated by the ChatGPT models hosted by OpenAI. This has changed the paradigm of AI application development by shifting back to API-based access to an LLM and reducing machine learning development and deployment operations.

With LLMs, we reference these large models as foundational models. These are large-scale general-purpose models that have been pre-trained with vast amounts of web-scale data and are capable of thinking outside the box for a wide range of tasks. ChatGPT 3.5 and ChatGPT4 are well-known and popular foundational models hosted by OpenAI and are available for further training and fine-tuning via APIs. AI application developers interact with APIs without the burden (and flexibility) of MLOps. The following diagram summarizes a basic LLM application topology:

Figure 13.1 – Basic LLM application with the third-party model (for example, OpenAI)

Note...