Large language model (LLM), explained
Google’s course describes an LLM as a model that predicts a token or a run of tokens, where a token can be a word, part of a word, or a single character. Strung together, those guesses become sentences and whole answers. That is why an LLM can draft a reply, sum up a page or answer a question in plain language.
Most chat tools you have heard of run on an LLM. The model does not look facts up by itself. Google’s course notes that LLMs can make up wrong answers, which is why tools that cite their sources, and a person who checks the work, matter.
How does a large language model write?
It predicts what comes next, one small piece at a time. Google’s course calls LLMs autocomplete that can predict thousands of tokens in a row. Training on a huge amount of text teaches it which words tend to follow which.
Can a large language model be wrong?
Yes. Google lists made up answers, high computing cost and bias among the known problems with LLMs. Grounding the model in real sources and having a person review the output lowers the risk.
Questions
What does LLM stand for?
LLM stands for large language model. It is an AI model trained on a huge amount of text so it can read and write language.
Is ChatGPT a large language model?
ChatGPT is a chat product built on OpenAI’s GPT models, and those models are large language models. Google’s glossary names GPT and Gemini as examples.
Why do large language models make mistakes?
An LLM writes by predicting likely words, not by checking facts. Google’s course says LLMs can hallucinate, which means their answers can contain mistakes.
Where we use this term
Sources
- LLMs: What’s a large language model?. Google Machine Learning Crash Course.
- Machine Learning Glossary. Google for Developers.
Definition checked against these sources and our own pages on .