AI Dictionary Ep 04: What is an LLM? (Large Language Model) Explained

Welcome to Episode 04 of the AI Dictionary! In this video, we answer the question: What is an LLM? (Large Language Models), explained in under two minutes!

An LLM is a neural network trained on huge amounts of text. Its job is surprisingly simple: it predicts the next token (the next word or part of a word), one step at a time.

In this quick lesson, we cover:
• What makes it "Large"? Billions of parameters (learned weights), training data at internet scale, and the ability to generalize to brand new tasks without being retrained.
• How it generates text: Your prompt is split into tokens. The Transformer architecture runs attention across them, outputs probabilities for the next token, samples one, and loops to generate language word by word.
• Old NLP vs. LLM: While old NLP systems needed one specific model per task and were easily broken by new phrasing, an LLM is a single model that generalizes to many tasks and generates fluent language.
• Where you meet LLMs: They are the models inside GPT, Claude, Gemini, and Llama. They are the friendly face of chat assistants and the brain inside RAG systems and AI agents.

Next episode: What is an embedding? Subscribe to continue mastering AI terms!

#LLM #LargeLanguageModel #ArtificialIntelligence #AIDictionary #MachineLearning

Видео AI Dictionary Ep 04: What is an LLM? (Large Language Model) Explained канала AImagic