LearnAI

Navigation

Introduction to AI

Machine Learning

Deep Learning

NLP & LLMs

Computer Vision

Generative AI

Reinforcement Learning

AI Ethics & Governance

Tools & Frameworks

General

Introduction to Chain of Density

Chain of Density (CoD) is a prompt-engineering technique designed to create summaries that are increasingly dense with information without increasing the overall length of the summary.

How Chain of Density Works

The goal of CoD is to iteratively generate a summary, identifying “missing entities” from the source text and incorporating them into the previous summary in each step. This process is typically repeated 5 times.

  1. Step 1: Generate an initial sparse summary.
  2. Steps 2-5: Identify 1-3 new, salient entities from the source that are not in the current summary, and rewrite the summary to include them while keeping it concise.

Benefits of CoD

  • Informative: Packs more “meaning” into every sentence.
  • Readable: Maintains a fixed length, avoiding the bloat of long summaries.
  • Controlled: Allows for specific targeting of entities or facts.

When to Use CoD

Use CoD when you need high-quality, dense summaries for news articles, research papers, or any lengthy document where entity extraction is critical.