Large Language Model
A neural network trained on massive text data to predict and generate human language.
What Is a Large Language Model
A large language model (LLM) is a neural network trained on very large collections of text to predict the next token in a sequence. Through this training process, the model learns statistical patterns in grammar, facts, reasoning steps, and writing style, which lets it generate coherent text, answer questions, summarize documents, translate languages, and write code. Most modern LLMs are built on the transformer architecture, which uses an attention mechanism to weigh the relevance of different parts of the input when producing each new token.
How It Works
Training happens in stages. In pretraining, the model processes enormous amounts of text and adjusts billions of internal parameters to reduce prediction error. This produces a base model with broad language ability but no particular alignment to instructions. Additional stages, such as supervised fine-tuning and reinforcement learning from human feedback, teach the model to follow instructions, maintain a helpful tone, and avoid unsafe outputs. Once trained, the model is used through inference: given an input prompt, it generates output one token at a time, each new token conditioned on everything that came before it.
Capabilities and Limits
LLMs are general purpose text processors rather than databases or calculators. They do not store facts in a retrievable, verifiable way; instead, they produce the most statistically plausible continuation of a prompt, which can look confident even when it is factually wrong. Their knowledge is also bounded by a training cutoff date and by the size of the context window, the amount of text they can consider at once. These properties matter for anyone building software around an LLM, since output quality depends heavily on prompt design, the chosen model, and any external data or tools connected to it.
Where LLMs Are Used
Applications range from chat assistants and coding tools to document summarization, data extraction, and autonomous agents that plan and execute multi-step tasks. In agent-based systems, an LLM typically acts as the reasoning engine behind a loop that can call tools, read and write files, or run commands, with the model's output driving each step of the loop.