COMMERCIAL
LABELED
Synthetic tables are clean, and models trained on them fail on spreadsheets people make for themselves, where the header row starts on row four, a notes column wanders, totals are typed by hand and two tables share one sheet. Contributors upload a spreadsheet they built, choose its type and leave the structure as it is; tidied files, stripped headers and blank templates are excluded. Collection is open only in the US, Canada, the UK, Europe, Australia and New Zealand, one per contributor. Household budgets, small-business ledgers, trackers, rosters and inventories: the distribution SpreadsheetBench was built to test. The type tag is the only label; ask about native workbooks if you need formulas.
Regions
US, Canada, UK, Europe, Australia and New Zealand (36 countries)
Updated
What’s in the collection and how it’s delivered.
Data modality
Text
Annotated
Labeled (spreadsheet type)
Exclusivity
Non-exclusive rights
Capture tool
File upload from phone
Use cases
Table understanding on real spreadsheets, spreadsheet QA and formula models, messy-table structure recovery
Format
Resolution
Native PDF, text layer where the source system produced one
Total volume
Available on request
Regions
US, Canada, UK, Europe, Australia and New Zealand (36 countries)
Purity level
99%+ passed QA review
Where this data comes from and how it’s licensed.
Every file comes straight from the contributor who provided it and is rights-cleared before it’s listed, so your license traces back to the source.
Contributor
Provides the data and uploads it straight to Kled.
Kled
Clears the rights before anything is listed.
Your team
Licenses it under the terms below.
License terms
Non-exclusive commercial training license.
Consent
Each contributor opted in to this use when they uploaded.
Specs reflect the current collection and may change as new data is added. Final terms are set in your license agreement.
