Intermediate · Generative AI

Self-Consistency

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. A decoding strategy that samples multiple chain-of-thought reasoning paths and returns the majority-voted answer for improved accuracy.

Technical Definition

A decoding strategy that samples multiple chain-of-thought reasoning paths and returns the majority-voted answer for improved accuracy.

How it works

Instead of greedy decoding a single chain of thought, self-consistency (Wang et al., 2022) samples N diverse reasoning paths at higher temperature, then takes the most common final answer. The intuition is that correct reasoning converges on the same answer through different paths, while errors are idiosyncratic. It reliably boosts performance on math and commonsense benchmarks.

Related Concepts

• Chain-of-Thought (CoT) Prompting — Asking models to show step-by-step reasoning before giving a final answer, improving accuracy on complex tasks.
• Reasoning — An AI system's ability to draw logical, multi-step conclusions from given information.
• Tree of Thoughts (ToT) — A prompting strategy that explores multiple reasoning branches as a search tree, evaluating intermediate thoughts and backtracking when needed.