This chapter lays the groundwork for understanding large language models (LLMs). We start by defining LLMs and positioning them relative to earlier sequence models, tracing the evolution that led to current architectures. The discussion covers why model size and data scale are fundamental factors influencing LLM behavior and emergent abilities. We will also outline the significant computational and memory constraints inherent in developing these models and introduce the common software frameworks and hardware accelerators involved in the LLM ecosystem.