---
date: '2024-02-12'
description: constructive communication with llms through quality prompts, including chain-of-thought and least-to-most prompting strategies.
id: prompt engineering
modified: 2026-06-05 15:08:07 GMT-04:00
tags:
  - seed
  - ml
title: Prompt engineering
created: '2024-02-12'
published: '2024-02-12'
pageLayout: default
slug: thoughts/prompt-engineering
permalink: https://aarnphm.xyz/thoughts/prompt-engineering.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
A constructive way to form communications with [[thoughts/LLMs]]. As we improve the quality of prompts, we can expect better results from the models. Similar to [[thoughts/linguistic]], a good prompt is a good form of communication with the system.

This is different from [[thoughts/zero-shot learning|zero-shot prompting]]

## CoT prompting

See also: [[thoughts/NLP]]

You can think of it as explaining a big topics to a five years old. You break down topic into smaller, logic parts that mimics a train of thoughts.

## Least-to-most prompting

Prompted to first list the sub-problems to a problem, then solve them in sequence.