Dataset · Hugging Face

ServiceNow-AI/SynthDocBench

Name: SynthDocBench
Creator: ServiceNow-AI
License: mit
Keywords: task_categories:visual-question-answering, task_categories:document-question-answering, language:en, license:mit, license:llama3, size_categories:1K<n<10K, format:parquet, format:optimized-parquet, modality:document, modality:tabular, modality:text, library:datasets, library:pandas, library:polars, library:mlcroissant

Built With Llama! SynthDocBench SynthDocBench is a fully synthetic benchmark for evaluating vision-language models (VLMs) on complex, multi-page PDF documents. Documents are generated end-to-end by an LLM pipeline that produces realistic multi-page reports with embedded D3.js charts, rich layouts, and deterministically grounded ground-truth answers — enabling controlled, noise-free evaluation impossible with real-world corpora. Paper: SynthDocBench: A Controlled… See the full description on the dataset page: https://huggingface.co/datasets/ServiceNow-AI/SynthDocBench.

View on Hugging FaceReport an issue

Author / organization	ServiceNow-AI
License (as declared)	mit
Downloads	1.6K (as reported by Hugging Face)
Likes	4
Created	26 March 2026
Last updated at source	23 May 2026

benchmarkvision-languagedocument-understandingsyntheticchartsd3built-with-llamallama

Embed this entry

<iframe src="https://silytics.si/embed/item/eb1a3a49-691f-4524-8716-8e1aa6aadaa5" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

Popularity, last 2 days

1,629 now (-40 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.

More from ServiceNow-AI

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.