DS1. LLM and Agent Fundamentals (incomplete)

This session provides a foundational vocabulary for interacting with large language models, covering just enough theory to ground hands-on work. We'll begin with the core parameters that steer model behavior, such as the context window, temperature, and top_p for controlling output, along with practical knobs like max_tokens and stop sequences. We will also explore the basics of tokenization for effective prompt and response budgeting and differentiate between system and user prompts. Building on this foundation, we'll examine advanced techniques, including Retrieval-Augmented Generation (RAG) for grounding models in external data, function calling for enabling tool use, and the concept of agents that combine these capabilities to autonomously execute complex tasks.