Datasets:
Tasks:
Document Question Answering
Size:
n<1K
Tags:
benchmark
document-ai
information-extraction
structured-extraction
key-information-extraction
ocr
License:
File size: 1,372 Bytes
ca66b51 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 | # Contributing
DocuBench is intended to be reproducible benchmark infrastructure. Contributions should preserve that property.
## Development Setup
```bash
python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install -e ".[dev]"
```
## Checks
Run these before opening a pull request:
```bash
docubench validate
docubench score
pytest
```
## Result Submissions
Add new system outputs under `results/<system_name>/`. Each `<doc_id>.json` file should include a top-level `data` object. See [`docs/submissions.md`](docs/submissions.md).
If your system uses a prompt or run configuration, commit it under [`prompts/`](prompts) alongside the existing baselines so the result set is reproducible and auditable.
## Leaderboard Data
The Hugging Face Space in [`space/`](space) renders `space/leaderboard.json`. After changing results or labels, refresh it so the leaderboard stays in sync:
```bash
docubench report
python3 space/build_data.py
```
`tests/test_space.py` fails if the committed `space/leaderboard.json` is stale.
## Dataset Changes
Changes to documents, schemas, labels, or scoring behavior can change benchmark meaning. Include:
- the reason for the change
- affected document IDs
- before/after scoring impact if applicable
- source and license details for any new document
Scoring changes should be treated as benchmark-version changes.
|