Dataset · modusaudit
Financial statement audit benchmark
modusaudit/FinancialAuditBench
A set of tasks to evaluate AI agents in performing financial statement audits.
What's inside
The dataset contains 90 tasks across 6 synthetic audit projects in manufacturing and HR services in English. Agents use supporting documents to fill out or verify tabular workpapers.
What you can do with it
- Evaluate and test autonomous AI agents on financial tasks
- Develop systems to automate accounting and audit analysis
- Languages
- English
- Size
- up to 1K rows
- Downloads this month
- 7.5K
- Likes
- 5
How to load it
from datasets import load_dataset
ds = load_dataset("modusaudit/FinancialAuditBench")