COA Mapper v1

The first open model for chart-of-accounts mapping: given a ledger account from QuickBooks Desktop or Sage 50 (name + type), retrieve the correct target account in the client's new QuickBooks Online chart of accounts.

A 33M-parameter embedding model (BAAI/bge-small-en-v1.5, fine-tuned) β€” small enough to run on CPU, in a browser, or inside a migration tool.

Results β€” COA-Map-Bench v1

Test split of gyaanbyte/coa-mapping-synthetic: 137 mappings from 4 held-out companies (2 QuickBooks Desktop, 2 Sage 50) never seen in training. Candidates = the distinct target accounts of the same company (i.e. mapping into the client's new QBO COA). A top-1 hit implies the correct QBO type and detail type.

Method acc@1 acc@5 MRR
TF-IDF (char 2–4-grams, cosine) 0.745 0.949 0.824
bge-small-en-v1.5 zero-shot 0.796 0.956 0.868
qwen3:8b zero-shot (picks from the candidate list) 0.788 β€” β€”
coa-mapper v1 (this model) 0.883 0.964 0.924

Read the numbers honestly: the benchmark is synthetic, and account names carry strong lexical signal (TF-IDF already gets 74.5%). The fine-tune's gain lives in the hard 25% β€” abbreviations, near-duplicates to consolidate, dumping-ground splits, type/detail-type disambiguation. An adjudicated benchmark subset is planned for v1.1.

Try it

πŸ§ͺ Live demo β€” runs in your browser via Transformers.js (int8 ONNX included in this repo under onnx/; nothing is uploaded).

Usage

from sentence_transformers import SentenceTransformer
m = SentenceTransformer("gyaanbyte/coa-mapper")

# Source accounts as "name (type) [system]"; targets as "name β€” type / detail type"
src = ["Accum Depr - Equipment (fixed asset) [quickbooks desktop]"]
targets = [
    "Accumulated Depreciation β€” Fixed Assets / Accumulated Depreciation",
    "Equipment β€” Fixed Assets / Machinery & Equipment",
    "Depreciation Expense β€” Other Expense / Depreciation",
]
S = m.encode(src, normalize_embeddings=True)
T = m.encode(targets, normalize_embeddings=True)
print(targets[(S @ T.T).argmax()])

In production we use it as retrieval β†’ LLM re-rank of the top-5 β†’ confidence threshold β†’ human review lane: the goal is hours saved per migration at $0 reconciliation variance, not unattended automation.

Training

  • Base: BAAI/bge-small-en-v1.5 Β· loss: MultipleNegativesRankingLoss (in-batch negatives)
  • Data: 1,025 (source β†’ target) pairs, train split of the dataset above
  • 4 epochs, batch 32, lr 3e-5, fp16 β€” 12 seconds on one RTX 5060 laptop GPU
  • Training/eval code: coa_train.py (open)

Limitations

  • Trained and evaluated on synthetic data; real ledgers are messier and more idiosyncratic
  • Target side is QuickBooks Online only; sources are QuickBooks Desktop and Sage 50
  • English, US-GAAP-oriented account structures
  • Candidate pool must be supplied (the target COA); this model does not invent accounts

Who's building this

GyaanByte Labs β€” accounting data-migration specialists (6+ production migrations, 200,000+ records, $0 variance). The mapper is the engine behind our migration service; the reusable parts are published openly.

Downloads last month
45
Safetensors
Model size
33.4M params
Tensor type
F32
Β·
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for gyaanbyte/coa-mapper

Quantized
(44)
this model

Dataset used to train gyaanbyte/coa-mapper

Space using gyaanbyte/coa-mapper 1

Evaluation results