hyperbots-inc/synthetic-doc-classification

Synthetic Enterprise Document Classification

248 page images across 31 enterprise document classes (8 per class). Labels come from Hyperbots' 100-type reference taxonomy and cover finance, procurement, logistics, HR, onboarding and compliance, healthcare, legal and facilities documents. About 20% of pages are stamped and about 20% are degraded scans, so the classifier can't rely on clean pixels alone.

It matches HyperAPI POST /v1/classify, which returns a label, a confidence score and ranked candidates, and supports custom classes.

Classes

ID	Type	Category	doc_type
1	Invoice	A. Payables & Invoices	invoice
10	Tax Invoice	A. Payables & Invoices	tax_invoice
11	Statement of Account / Aging Statement	B. Account Statements & Reconciliation	statement_of_account
13	Credit Card Statement	B. Account Statements & Reconciliation	credit_card_statement
14	Bank Statement	B. Account Statements & Reconciliation	bank_statement
15	Credit Note	C. Credits & Adjustments	credit_note
17	Receipt	D. Payments & Settlements	receipt
20	Remittance Advice	D. Payments & Settlements	remittance_advice
25	Dunning / Collection Letter	E. Collections	dunning_letter
28	Purchase Order	F. Procurement & Ordering	purchase_order
30	Quotation / Estimate	F. Procurement & Ordering	quotation
32	Work Order	F. Procurement & Ordering	work_order
38	Delivery Note / Challan	G. Shipping & Logistics	delivery_note
39	Bill of Lading	G. Shipping & Logistics	bill_of_lading
41	Packing List	G. Shipping & Logistics	packing_list
42	Timesheet	G. Shipping & Logistics	timesheet
43	Expense Report	H. Internal & Employee Documents	expense_report
44	Payroll Statement	H. Internal & Employee Documents	payroll_statement
51	Bank Details Form	I. Onboarding & Compliance	bank_details_form
54	KYC Documentation	I. Onboarding & Compliance	kyc_form
56	Balance Sheet	J. Financial Statements & Reports	balance_sheet
57	Profit & Loss Statement	J. Financial Statements & Reports	profit_and_loss
62	Credit Application Form	K. Credit, Lending & Risk	credit_application
79	Lab Test Report / Result	L. Clinical, Laboratory & Healthcare	lab_report
80	Patient Intake / Registration Form	L. Clinical, Laboratory & Healthcare	patient_intake_form
83	Explanation of Benefits (EOB) / Claim Form	L. Clinical, Laboratory & Healthcare	claim_form
87	Meeting Minutes / Notes	M. Digital & Communication Artifacts	meeting_minutes
90	Non-Disclosure Agreement (NDA)	N. Legal, HR & Administrative	nda
95	Offer Letter / Employment Contract	N. Legal, HR & Administrative	offer_letter
98	Incident / Service & Maintenance Report	O. Operations, Quality & Facilities	maintenance_report
99	Lease Agreement / Utility Bill	O. Operations, Quality & Facilities	utility_bill

Some pairs are deliberately hard: invoice vs tax_invoice, bank_statement vs credit_card_statement vs statement_of_account, purchase_order vs quotation vs work_order, and delivery_note vs packing_list. Titles vary within each class ("Credit Note" / "Credit Memo", "Payslip" / "Earnings Statement" / "Pay Advice").

Columns

Column	Description
image	744 x 1052 JPEG (A4 at about 90 DPI)
label	Class key (doc_type)
taxonomy_id, taxonomy_name, category, financial	Taxonomy metadata
title_on_page	The printed title (don't feed it to the model if you want a fair test)
variant	clean, stamped, degraded or stamped+degraded
ocr_text	Page text, for text-only classifiers

from datasets import load_dataset
ds = load_dataset("hyperbots-inc/synthetic-doc-classification", split="test")
print(ds.unique("label")[:5], ds[0]["variant"])

How the data was made

Every document is generated from a seeded random number generator by docgen.js and rendered on an HTML canvas in Chrome on macOS. Images use the system's print fonts and, for handwriting, the handwriting-style fonts installed with macOS (Bradley Hand, Noteworthy, Marker Felt, Chalkboard SE, Snell Roundhand, Apple Chancery, Savoye LET, Brush Script MT). Labels are taken from the generator itself, so there is no human annotation error. Field and block boxes are recorded at the moment each element is drawn.

All companies, people, addresses, account numbers and amounts are fictional. Phone numbers use ranges reserved for fiction where these exist (US 555-01xx, UK 07700 900xxx, AU 0491 570 1xx). Email addresses use the reserved example.com / example.org domains, and US ID numbers start with 9, which is never issued as an SSN.

To rebuild the dataset, load generator/docgen.js and generator/datasets.js in a browser page and run await DocSets.build('classification'). Rendering depends on installed fonts, so pixels can differ on other systems, but text and labels are identical for the same seed.

Limitations

• The documents are synthetic. Their layouts, vocabulary and value ranges are much narrower than real enterprise traffic. Use them for benchmarking, regression tests and pipeline development, not as a replacement for evaluation on your own documents.
• All content is in English, with values formatted for 9 locales (US, UK, IN, CA, AU, DE, FR, SG, AE).
• Handwriting is rendered from fonts with per-character jitter. It is a stand-in for real handwriting, not a sample of it.

About HyperAPI

HyperAPI is Hyperbots' document intelligence API for enterprise documents. It provides parse, extract, classify, split and redact behind one endpoint and returns schema-validated JSON with a confidence score on every field. It has a REST API and a Python SDK, with an MCP server for agents available on request, and you can start free. See the HyperAPI Space for a walkthrough.

Part of the Hyperbots synthetic document suite

Dataset	Focus
synthetic-enterprise-doc-extraction	Field extraction across 31 enterprise document types
synthetic-doc-parsing	Layout blocks, reading order, Markdown and tables
synthetic-pii-redaction	PII spans and boxes for masking and de-identification
synthetic-doc-classification (this dataset)	31-class document classification
synthetic-doc-splitting	Multi-document bundles with true boundaries
synthetic-handwritten-forms	Forms filled in handwriting-style script
synthetic-stamped-documents	Stamps and seals over printed text
synthetic-bad-scans	Skew, blur, noise, shadows and compression at 3 severities
synthetic-dense-documents	Small type, multi-column layouts and long tables
synthetic-invoice-extraction	Invoice fields and line items (text + HTML)

License and citation

CC BY 4.0. Generator code is MIT.

@misc{hyperbots2026docclassification,
  title  = {Synthetic Enterprise Document Classification},
  author = {Hyperbots Inc.},
  year   = {2026},
  url    = {https://huggingface.co/datasets/hyperbots-inc/synthetic-doc-classification}
}