Skip to slide
06 / 53

CH 01 · Chain of Thought, Teaching Models to Think Out Loud · 01/08

Read as one page

Chain of Thought, Teaching Models to Think Out Loud

Asking a large model to show its steps before answering, rather than respond instantly, produces a large jump in reasoning accuracy without any retraining.

8 slides · 4 min read

Contour map for Chain of Thought, Teaching Models to Think Out Loud

In this chapter

  1. 01The problem: answering too fast
  2. 02The fix: ask it to reason step by step
  3. 03How you actually trigger it
  4. 04The catch: it only works when models are big
  5. 05A reliability booster: self-consistency
  6. 06Why this paper mattered
  7. 07The one-sentence takeaway

Paper: Chain-of-Thought Prompting Elicits Reasoning in Large Language Models (2022)

This paper introduced one of the simplest and most powerful tricks in all of AI, and the wonderful part is that it requires no new training at all. You just change how you ask. The idea is to get the model to show its work before giving an answer, exactly like a math teacher asking you to write out the steps. That single change unlocks a dramatic jump in reasoning ability.

Finished this idea?

Your learning progress stays in this browser.