Receipt #13

5 / 5 rules pass Model: auto-acceptedmatches human label
CORD-v2 test receipt · read by Qwen3.5-4B (QLoRA) with best-of-4 and the accounting checker

Receipt photo

Photo of receipt #13 from the CORD-v2 test set

How the checker chose

The model wrote one normal reading and four sampled ones. The checker keeps the reading that passes the most accounting rules; the model's confidence only breaks ties.

  • Greedy80% rules
  • Sample 180% rules
  • Sample 2100% rules
  • Sample 3100% rulespicked
  • Sample 4100% rules

Extracted fields

ItemQtyUnit priceAmountRemove

Accounting checks

would be accepted
  • Line 1: qty × unit price = amountexpected 43,636 · found 43,636
  • Line 2: qty × unit price = amountexpected 3,000 · found 3,000
  • Items add up to the subtotalexpected 46,636 · found 46,636
  • Subtotal − discount + tax + service = totalexpected 51,300 · found 51,300
  • Cash − change = totalexpected 51,300 · found 51,300

The rules recheck as you edit, using a TypeScript port of the Python checker that agrees with it on all 277 rule results of the test set. Amounts within 1% are treated as equal, to allow for rounding.