---
dataset_info:
  features:
  - name: sentence1
    dtype: string
  - name: sentence2
    dtype: string
  - name: labels
    dtype: int64
  - name: metadata
    dtype: string
  splits:
  - name: train
    num_bytes: 12201493
    num_examples: 27422
  - name: test
    num_bytes: 3190343
    num_examples: 6589
  download_size: 5901103
  dataset_size: 15391836
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
license: other
source_datasets:
- irds/antique
- irds/antique_train
- irds/antique_test
---

# antique

All native ANTIQUE judged pairs, retaining relevance grades 1–4.

Join the original query, answer-collection and qrel files without generated
negatives or filtering grades. Preserve native query-disjoint train/test.
Q0/E0/U0 judgment provenance stays in metadata. The original readme permits
research use; Yahoo answer rights are not replaced by a software license.
Original distribution: https://ciir.cs.umass.edu/downloads/Antique/

Original data: [irds/antique](https://huggingface.co/datasets/irds/antique), [irds/antique_train](https://huggingface.co/datasets/irds/antique_train), [irds/antique_test](https://huggingface.co/datasets/irds/antique_test), [original distribution](https://ciir.cs.umass.edu/downloads/Antique/). Repackaged as parquet for [tasksource](https://github.com/sileod/tasksource) by [scripts/repackage_dataset/relevance.py](https://github.com/sileod/tasksource/blob/main/scripts/repackage_dataset/relevance.py#L156).

[Conversion provenance](provenance.json): source revisions, hashes, configuration and exclusions.