Synthetic Form 941 Schedule R - Allocation Schedule for Aggregate Form 941 Filers Data
Synthetic training data — no real PII, fully coherent identities
Generate synthetic Form 941 Schedule R aggregate allocation schedules — 828 fields, the largest form in the SymageDocs library, laid out as 26 client rows of quarterly payroll allocation with every money column split into dollars and cents. Built for benchmarking table extraction at scale.
828
Fields per document
2
Pages
Tax
Category
What this document is
Schedule R (Form 941) is the quarterly allocation schedule attached to an aggregate Form 941. A section 3504 agent, a certified professional employer organization, or another third-party payer files one quarterly return covering many client employers, and Schedule R distributes that return across them: one row per client EIN with employee counts, wages, federal tax, qualified leave wages, Social Security and Medicare bases, tip taxes, payroll tax credits, and COBRA premium assistance. Thirteen client rows appear on each of the two pages.
Why generate synthetically
This is the widest and densest document we generate — more fields than any other form in the library and more annotation relations than any other form in the library — and its content is almost entirely tabular. That combination makes it the definitive stress test for table structure recognition, row segmentation, and wide-grid extraction. It is also unobtainable as real training data: every row on a real Schedule R identifies a client company by EIN, so no filer can share one.
What makes synthetic data useful
Each of the 26 client rows is a separately simulated employer with its own headcount and payroll, so the wage figures vary row to row the way a real client roster does rather than being noise around a single mean. Per-client amounts roll up into the subtotal and combined-total rows that the form actually prints, and the dollars and cents halves of every money column are generated together, so the grid is internally consistent and a model can be scored on whether it reconstructed the arithmetic, not just the glyphs.
Training challenges
Every money column is physically split into a dollars cell and a cents cell, which means a document with hundreds of amounts produces well over a thousand numeric regions, and the dominant failure mode is pairing a row's dollars with the adjacent row's cents. Column headers print once and then govern 13 rows below them, placing labels far outside the local neighborhood most key-value models attend to. Both pages repeat the same header block, filer-type checkboxes, quarter selector, and year, so page classification on appearance alone cannot tell page one from page two. Subtotal, combined-subtotal, and grand-total rows are visually identical to client rows but semantically different, and a model that treats them as clients will double-count the return. About 15% of cells are legitimately blank where a client row is unused, so blank-versus-zero discrimination matters — a zero in an allocation grid is a claim, an empty cell is not.
Generate synthetic Form 941 Schedule R - Allocation Schedule for Aggregate Form 941 Filers data
Start with 500 free credits. No credit card required.
Generate NowWho uses this data
Certified professional employer organizations, section 3504 agents, and other third-party payers who file aggregate quarterly returns and must reconcile them client by client; payroll platforms ingesting allocation schedules during client migration; and table-structure-recognition and document-AI teams who need a genuinely hard production grid to benchmark against rather than an academic table corpus.
Document complexity profile
828 fields across 2 pages — the largest field count in our catalog — comprising 628 currency cells, 156 text fields, 35 numeric counters, and 9 checkbox targets. 1,122 annotation relations, more than any other form we publish, because column headers and printed line labels each govern many cells. Binding logic runs to 623 function calls and 7 conditional bindings at a maximum expression depth of 2, with no arithmetic bindings: per-client allocations are derived upstream from each simulated client's payroll, so the grid balances without expression-tree evaluation at fill time.
Key stats from our synthetic corpus
Quantitative characteristics of the Form 941 Schedule R - Allocation Schedule for Aggregate Form 941 Filers documents our generator produces.
| Metric | Value | Detail |
|---|---|---|
| Client rows per document | 26 | Thirteen client allocation rows print on each of the two pages, for 26 client rows plus the subtotal, combined-subtotal, and grand-total rows that summarize them. |
| Median employees per client row | 7 | The median client company reports 7 compensated employees (p25 3, p75 17), with the largest reaching 4,268. Small clients dominate a typical PEO roster, and the occasional very large client is exactly the row that breaks fixed-width numeric extraction. |
| Median total employees allocated | 267 | The grand-total employee count across all client rows has a median of 267 (p25 162, p75 466, max 9,126). This total must equal the sum of the individual rows, which makes it a ready-made arithmetic-consistency check for evaluating an extractor end to end. |
| Section 3504 agent filers | 62% | 61.8% of synthetic filers check the section 3504 agent box, 21.4% the CPEO box, and 16.9% the other-third-party box. These live in the header block that repeats verbatim on page two. |
| Field fill rate | 85% | 702 of the 828 fields carry a value in a typical document, leaving roughly 15% legitimately blank where the client roster does not fill every row. Distinguishing a blank cell from a zero cell is a real semantic distinction on an allocation schedule. |
How this document co-occurs with others
Rates at which identities in our corpus that produce a Form 941 Schedule R - Allocation Schedule for Aggregate Form 941 Filers also produce other documents.
| Correlation | Rate | Detail |
|---|---|---|
| The aggregate return it attaches to | 100% | Schedule R is never filed alone. Every synthetic filer also produces the Form 941 whose totals the schedule allocates, giving you the canonical attachment-reconciliation pair. |
| Annual sibling schedule | 100% | The same aggregate filers submit the annual FUTA Schedule R. Two grids, two layouts, one underlying client roster — the best available generalization test for a table extraction model. |
| Corrected quarter allocation | 100% | 1.9% of synthetic Schedule R filings mark the return type as 941-X rather than 941, so the corpus includes the corrected-allocation variant alongside the ordinary quarterly one. |
All stats above are corpus-derived: they were computed on a local synthetic corpus of 1,000 generated identities produced by SymageDocs' World Simulation Engine. No real employer, client EIN, or payroll data was used. `make corpus-stats` regenerates the shipped corpus_stats.json for the site's analytics form set, which does not include this edition; reproduce these figures with `python -m symagedocs.corpus_analysis --corpus-size 1000 --form-id irs_f941_schedule_r_standard_2024`.
Frequently asked questions
- What data format do synthetic Form 941 Schedule R documents include?
- Each generated identity produces a rendered PDF plus a structured JSON annotation file with bounding boxes, field types, and ground-truth values for all 828 fields, with each dollars cell, each cents cell, and each row index annotated separately. COCO, YOLO, FUNSD, and BIO/NER exports come from the same job.
- What does each client row contain?
- Client EIN, a wage/tips/compensation type indicator, employee count, and a long run of money columns — total wages, federal taxes withheld, qualified leave wages, taxable Social Security wages, Medicare wages subject to Additional Medicare Tax, combined Social Security and Medicare tax, tax on unreported tips, the qualified small business payroll tax credit, and the sick, family leave, and COBRA credit lines. Each money column contributes a dollars field and a cents field.
- How does labeling work on a grid this large?
- Row index is encoded in the field id, so company_total_wages_dollars_07 is unambiguously the eighth client row. Subtotal and total rows carry their own distinct ids rather than sharing the client row namespace, which lets you score row-type classification and row assignment as separate metrics instead of collapsing the grid into one flat key-value comparison.
- Can I use this data commercially?
- Yes. Every client company, EIN, and payroll figure is synthetic, contains no real business data, and is licensed for commercial use including model training, benchmarking, and redistribution inside your own products.
- How does it compare to the 940 version of Schedule R?
- The 940 schedule is annual and narrower — 15 primary rows and 22 continuation rows with 5 money columns. The 941 schedule is quarterly, wider, and taller per row. Training on one and evaluating on the other is a clean generalization test for whether a table model learned grid structure or memorized a layout.