MetaGAI: A Large-Scale and High-Quality Benchmark for Generative AI Model and Data Card Generation

TL;DR

MetaGAI introduces a large-scale benchmark dataset for evaluating automated generation of Model and Data Cards in Generative AI, addressing scalability and quality issues with a multi-agent framework and comprehensive human validation.

Contribution

The paper presents MetaGAI, a novel benchmark with 2,541 verified document triplets, employing a multi-agent system and human-in-the-loop validation for systematic evaluation of generation methods.

Findings

Sparse Mixture-of-Experts architectures outperform others in cost-quality efficiency.

A fundamental trade-off exists between faithfulness and completeness in generated documents.

MetaGAI enables benchmarking, training, and analysis of automated documentation methods at scale.

Abstract

The rapid proliferation of Generative AI necessitates rigorous documentation standards for transparency and governance. However, manual creation of Model and Data Cards is not scalable, while automated approaches lack large-scale, high-fidelity benchmarks for systematic evaluation. We introduce MetaGAI, a comprehensive benchmark comprising 2,541 verified document triplets constructed through semantic triangulation of academic papers, GitHub repositories, and Hugging Face artifacts. Unlike prior single-source datasets, MetaGAI employs a multi-agent framework with specialized Retriever, Generator, and Editor agents, validated through four-dimensional human-in-the-loop assessment, including human evaluation of editor-refined ground truth. We establish a robust evaluation protocol combining automated metrics with validated LLM-as-a-Judge frameworks. Extensive analysis reveals that sparse Mixture-of-Experts architectures achieve superior cost-quality efficiency, while a fundamental trade-off exists between faithfulness and completeness. MetaGAI provides a foundational testbed for benchmarking, training, and analyzing automated Model and Data Card generation methods at scale. Our data and code are available at: https://github.com/haoxuan-unt2024/MetaGAI-Benchmark.