//
Search across all documentation pages
6 pages in this section.
Claude predicts each next token using patterns learned from massive training datasets.
A first walkthrough of what Claude is, how it answers, and core AI vocabulary.
A glossary covering tokens, transformers, inference, and fine-tuning, the core vocabulary of LLMs.
Contrasts deterministic code logic with Claude's probabilistic, pattern-based approach to generating answers.
Clarifies why Claude reasons and drafts responses instead of retrieving pre-written, indexed answers.
Enumerates practical habits for evaluating, piloting, and responsibly rolling out Claude across a team.