Skip to content
Module
Module 3 of 5Lesson 2 of 2~14 min

The context window, a powerful but uneven working memory

A long conversation or a large document can make the assistant lose track of an instruction that was perfectly clear. This lesson teaches you what the model actually takes in from an exchange, why its attention fades in places, and how to organize conversations and documents to get reliable answers.

Take this lesson: your first course is freeSee the full syllabus

Lesson objective

By the end of this lesson, you will be able to explain what the model sees of a conversation, why it sometimes overlooks an instruction or a passage in a long document, and to organize your work (conversations, documents, summaries) to keep answers reliable.

Topics covered

  • context window
  • LLM memory
  • long documents and AI
  • answer reliability

In the glossary

Full glossary

Where it fits

What the model knows, and what it sees

What does the model know on its own, what do tools add, and what does it retain from our conversation?

Lessons in this module

  1. What the model knows, since when, and what tools add
  2. The context window, a powerful but uneven working memory (this lesson)

What you will learn in this module

This lesson is part of the course Understand what LLMs do well, and where they fail

  • Locate what the model knows (training data, knowledge cutoff, uneven coverage) and what web search, provided documents and connectors change.
  • Explain the context window as working memory (limit, degradation over long contexts, no memory across conversations without a dedicated feature) and derive working rules from it.

Related courses