What does FActScore measure and how does it decompose long-form factual evaluation?

Attempt it

FActScore (Min et al. 2023) decomposes a generated biography or article into atomic facts: the smallest individual claims that can be independently verified. It then checks each atomic fact against a reference knowledge source (e.g., Wikipedia). The score is the fraction of atomic facts that are supported by the reference. This enables fine-grained factual precision measurement rather than coarse holistic scoring of long-form outputs.

Your answer

Quick reference

TL;DR

FActScore breaks a long generation into atomic facts, verifies each against a knowledge source, and scores the fraction supported. It turns vague long-form factuality into a precise, per-claim number.

Memory aid

Sign in for the mnemonic that makes this stick. See Pro pricing.

Easy to grasp

Imagine a friend tells you a long story about a famous person. Instead of grading the whole story with one gut-feel thumbs up or down, you write down every separate claim it makes: born in this city, won that award, married this person. Then you check each claim, one at a time, against a trusted encyclopedia. Your final grade is simple: out of all those little claims, what fraction turned out to be true? That is FActScore. It does not care whether the story sounds confident or reads beautifully. It only counts how many of the small, checkable facts actually hold up against the reference. A polished paragraph with three wrong claims scores worse than a plain one where everything checks out.

Key concepts

FActScore Atomic fact decomposition Faithfulness Hallucination detection

Detailed answer & concept explanation

~8 min read

Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example. Click to expand.

Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example.

Sign in for 5 free unlocks/day or go Pro for unlimited. See Pro pricing.

Quick scan · 2–4 min

Everything important, quickly.

5 min: the decompose, verify, and aggregate pipeline, atomic-fact granularity, precision versus recall, knowledge-source coverage limits, retrieval versus factual error separation, and the estimated LLM verifier variant.

Real-world examples

Real products, models, and research that use this idea.

FActScore is the standard long-form factuality benchmark, used to score biography generation from frontier models like GPT-5.5 and Claude Opus 4.7.
RAGAS faithfulness applies the same claim-decomposition idea, checking each extracted claim against retrieved context rather than Wikipedia.
SAFE, the search-augmented factuality evaluator from the LongFact work, extends FActScore by sending each atomic fact to a live search engine.
Hallucination dashboards in observability tools like Langfuse and Arize Phoenix surface per-claim support rates inspired by atomic-fact scoring.
VeriScore and FActScore-style estimators run an LLM verifier per atomic fact to scale evaluation across thousands of generations per eval pass.

Interviewer follow-ups

What an interviewer would ask next. Try answering before peeking at the approach.

QHow would you adapt FActScore from a static Wikipedia reference to a live retrieval setting like RAG?

Sign in or go Pro to see the approach for this follow-up.

QWhy does FActScore measure precision rather than recall, and how would you add a recall signal?

Sign in or go Pro to see the approach for this follow-up.

QWhat breaks if your atomic-fact decomposer is unreliable, and how would you validate it?

Sign in or go Pro to see the approach for this follow-up.

Don't say this

Red flags and common mistakes that signal junior thinking. Click to expand.

Most common mistake

Treating FActScore as a recall metric. It measures precision, the fraction of stated facts that are supported, and says nothing about which true facts the generation left out.

Sign in or go Pro to see all red flags and common mistakes. See Pro pricing.

60-second cheat sheet

The night-before-the-interview bullets. Scan these on the way to the call.

Sign in for 5 unlocks/day or go Pro for unlimited revision sheets. See Pro pricing.

References

Primary sources. Skim if you want the original framing.