LLM Means Large Language Model
An LLM is a type of AI that has learned patterns from huge amounts of language and other information. It can answer questions, explain ideas, summarize, translate, write, compare information, and help with code.
LANGUAGE — patterns involving words, code, ideas, and relationships
MODEL — a trained mathematical system
The Basic Idea: Predict What Comes Next
Suppose you type: “The dog chased the…”
The model considers possible next pieces such as ball, cat, car, or rabbit. It chooses a next piece, adds it, and predicts again.
What Is a Token?
LLMs usually do not work with whole words exactly the way people do. They work with small pieces of text called tokens. A token may be a word, part of a word, punctuation, a number, or another text fragment.
LLMs generate language one token at a time.
Is It Just Autocomplete?
“Advanced autocomplete” is useful as a starting analogy, but it is incomplete. To predict useful continuations across enormous amounts of language, a large model learns complicated relationships involving facts, grammar, code, mathematics, styles, and concepts.
That is why the same basic prediction process can support translation, summarization, coding, explanation, and some forms of reasoning.
Prompt, Context, and Memory
A prompt is the information or instruction you give the model. Context is the information currently available to it—like papers sitting on a desk. Long-term memory is something a larger AI system may save and bring back later.
CONTEXT = what is on the desk now
MEMORY = what can be stored for later
An LLM Is Not the Whole AI System
A model can be connected to search, files, calculators, databases, memory, software tools, and other models.