THE GRITWORKS DATASET LIBRARY
Explore five document datasets, from OCR extraction to US KYC. Download 10 PDFs per pack, each showing a different challenging condition—from blur and cut edges to photocopy artifacts.
Free sample downloads · No sign-up required
PICK YOUR STARTING POINT
Download a sample and try it in your workflow.
For the full dataset, book a call—we’ll confirm fit and share it with your team.
Preview edition: downloads contain illustrative starter packs, not approved GritRender product exports.
Explore invoices, purchase orders, and receipts under challenging conditions. Exercise ingestion, OCR, and extraction workflows with 10 PDF variations.
Bring fictional claim, policy, and repair documents into your sandbox. Explore how document quality affects processing and review workflows.
Work with fictional loan applications, income statements, and account summaries. Explore document handling across 10 challenging PDF conditions.
Explore fictional lab requisitions and specimen labels with scan and capture defects. Exercise extraction and review flows without patient records.
Explore fictional applicant intake, address evidence, and identity summaries. Use negative PDF variants to exercise onboarding document workflows.
BUILT FOR THE LESS-THAN-PERFECT INPUT
Each ZIP contains only PDFs. Use these illustrative variations to explore document handling in your evaluation, sandbox, or test environment.
10 PDF files in a ZIP. No images, JSON, or additional files.
ONE LIBRARY. DIFFERENT STARTING POINTS.
Compare how your pipeline handles challenging inputs. Define reference answers and scoring criteria separately when you need accuracy metrics.
Exercise document workflows with fictional records before introducing approved production data.
Probe ingestion, extraction, error handling, and review flows across degraded document conditions.
FROM FIRST LOOK TO FULL DATASET
The sample is yours to explore.
The full dataset starts with a conversation.
Choose a dataset and download a ZIP containing 10 illustrative PDFs. No account or call needed.
Register your details when booking. Tell us which dataset you need and how your team plans to use it.
We’ll confirm fit and share the full dataset with qualified teams on the call. Free, with no purchase required.
A FEW THINGS TO KNOW
All five illustrative sample packs are free to download immediately. Each ZIP contains exactly 10 PDF files showing different negative document conditions, such as blur, cut edges, photocopy artifacts, and low light. There are no separate image files, JSON answers, or other files in the download.
Choose “Book a call for the full dataset” and register your details in our booking calendar. On the call, we’ll review your use case, confirm the dataset’s scope and fit, and share it with your team. Full datasets are not available through public download links.
Yes. The full standard dataset is free for qualified teams, with no purchase required. Qualification is based on an identifiable team, a concrete evaluation use case, dataset fit, and the agreed usage terms. Additional or custom requirements can be discussed separately.
No. The downloadable starter packs contain fictional records created for this webpage preview. They are illustrative assets, not approved GritRender product exports. They contain no real customer, borrower, claimant, or patient records.
Yes. Use the PDFs with your own document tools for evaluations, sandboxes, and testing. No GritWorks integration is required. These packs do not include ground-truth annotations or scoring tools. For accuracy measurement, define reference answers and scoring criteria separately. These small illustrative packs do not establish production performance.
The illustrative US KYC pack contains fictional applicant intake, address-evidence, and identity-summary documents with negative variations. These are not official identity documents, and the pack is not a compliance or identity-verification benchmark. Confirm the document types and scenarios you need with us on the call.
Bring the document types, required fields, and failure cases to the call. We can discuss how GritWorks products could support your workflow. Please describe requirements without sending sensitive production records.
