Skip to content

Repository files navigation

MedARC Agentic Medical Fact Verifier

An open-source reproduction of Baichuan-M3's medical fact verification system, built by MedARC.

Under active development.

System Sketch

Following Baichuan-M3, the task is split into three models across four steps:

  1. Claim Decomposer breaks input text (answers, documents, reasoning traces, model outputs) into individual medical claims.
  2. Fact Verifier compares each claim against a database of previously fact-checked claims ("Claim X is supported by evidence set Y [under scope Z] as of date T").
  3. For a new claim, a Search Agent is dispatched to find supporting or contradictory evidence from a curated medical corpus.
  4. Results return to the Fact Verifier, which scores the claim on a five-level scale (strongly supported, weakly supported, unclear, weakly unsubstantiated, strongly unsubstantiated) and writes a new entry to the fact database.

Components

ComponentRoleWorkspace
baselineEnd-to-end pipeline on off-the-shelf LLMs (no training)Independent
decomposerLong-form text → atomic medical claimsYes
verifierClaim + evidence → five-level supported↔unsubstantiated scoreYes
searchClaim → supporting / contradictory sourcesYes
datasetsDataset ingestion, construction, and synthetic dataYes
utilsShared helpers used across AMFV packagesYes
trainingTraining experiments and recipes for the aboveIndependent

The Workspace column marks membership in the root uv workspace. Independent packages (baseline, training) are excluded so they can evolve on their own.

About

Agentic Medical Fact Verifier

Resources

Stars

9 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages