Data → Machine Learning & AI
LLM
A model trained on large text corpora to generate and reason over natural language and code-like sequences.
Motivation
A large language model solves the problem of predicting and generating language-like sequences from learned patterns at scale.
Where it fits
It belongs to AI applications and machine learning systems.
Mental model
An LLM is a trained model that maps context into likely next tokens, enabling conversation, summarization, coding assistance, and reasoning-like behavior.
Important details
Useful systems around LLMs need prompting, retrieval, tool use, safety, evaluation, and monitoring.
Common mistakes
- Treating an LLM as a database.
- Ignoring hallucination, evaluation, and product constraints.