Skip to main content

BelegBench — Benchmark für die Extraktion Schweizer QR-Rechnungen (100 % synthetisch, DE/FR/IT)

Project description

Belegant

BelegBench

🇨🇭 Schweiz · 🇩🇪 DE · 🇫🇷 FR · 🇮🇹 IT · 🇦🇹 AT

BelegBench ist der Benchmark hinter Belegant-4B — der Schweizer On-Prem Beleg-AI. Von den Machern von German-OCR.

Benchmark für die Extraktion Schweizer QR-Rechnungen — 100 % synthetisch, DE / FR / IT.

Bisher gab es kein öffentliches Benchmark für Schweizer QR-Rechnungen mit CH-Spezifika (QR-IBAN, QR-Referenz, MWST 8.1 / 2.6 / 3.8 %, CHE-UID, Apostroph-Tausender). BelegBench schliesst die Lücke — inkl. der Metrik, die für Treuhänder zählt: Halluzinationsrate.

Ergebnisse (v1, n = 300, fairer 4B-Vergleich)

Modell Field-EM Halluzinationsrate Parse-Fehler
Belegant-4B 84.7 % 0.0 % 3/300
Qwen3.5-4B (roh) 75.8 % 1.2 % 15/100
gemma3:4b (roh) 16.5 % 15.9 % 76/100

Trilingual: FR 81.4 %, IT 83.0 %. Viral-Grafik: belegbench/viral_table.html.

Der Datensatz ist zu 100 % synthetisch (300 Belege, DE / FR / IT) — kein Datenschutzproblem. Die aggregierten Ergebnisse je Modell liegen unter belegbench/results/.

Benchmark laufen lassen

python belegbench/run_eval.py --model belegant-4b --mode belegant --n 300

Metriken: Field-Level Exact Match, Halluzinationsrate, checksummen-validierte Accuracy, Verify-Catch-Rate, Latenz.

Lizenz

Apache-2.0 · von den Machern von German-OCR

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

belegbench-0.1.1.tar.gz (7.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

belegbench-0.1.1-py3-none-any.whl (7.7 kB view details)

Uploaded Python 3

File details

Details for the file belegbench-0.1.1.tar.gz.

File metadata

  • Download URL: belegbench-0.1.1.tar.gz
  • Upload date:
  • Size: 7.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for belegbench-0.1.1.tar.gz
Algorithm Hash digest
SHA256 2a6643ba1bb7aff87e6052601afb7d0068d21efac186394e3f90c88caffa053c
MD5 b803980f5f3c7ed6c9d7b4b184c5a53c
BLAKE2b-256 090e653187c1dd63d70486837690b58ca2d43011daf8fa384151d5e91faeec64

See more details on using hashes here.

Provenance

The following attestation bundles were made for belegbench-0.1.1.tar.gz:

Publisher: publish.yml on Keyvanhardani/belegbench

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file belegbench-0.1.1-py3-none-any.whl.

File metadata

  • Download URL: belegbench-0.1.1-py3-none-any.whl
  • Upload date:
  • Size: 7.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for belegbench-0.1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 5ce566a8b7c57a3ec11977756ee825c3a8c6d5856f29b3f15fa8071a36b626ba
MD5 b0cadc4dfbca075d8c2d792d3ee31911
BLAKE2b-256 b06184b122aae2b49b3c1f423dcc8b1ec2e84603c690c340ee9ee48d23cb6dc3

See more details on using hashes here.

Provenance

The following attestation bundles were made for belegbench-0.1.1-py3-none-any.whl:

Publisher: publish.yml on Keyvanhardani/belegbench

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page