What's an LLM?

Definition

• LLM = large language model: an AI model trained on huge amounts of text data 0:15.
• Key shift from previous ML waves: you usually don't train these models yourself — you take an available model, maybe fine-tune it, but generally use it as is 0:15.
• Model sources 0:15:
  • Commercial: OpenAI, Anthropic, Google.
  • Openly available: Mistral, Llama.

How you interact with it

• You write a prompt — a set of text instructions — and the model outputs text in return 0:15 0:35.

Task types

• Generative (open-ended output): extract information, write summaries, generate code, write emails, translate 0:35.
• Predictive (simpler, closed output): classify text by sentiment or topic 0:35.