we introduce AttriBench, the first fame- and demographically-balanced quote attribution benchmark dataset.
In its own words: https://arxiv.org/abs/2604.05224, read 2026-10-29
| What receives provenance | a quote and its author |
| Relations carried | authorship (who wrote the quote); suppression (attribution omitted though authorship information is available) op |
| Carrier | the model's answer naming, or not naming, an author |
| Expresses · records · tests | expressesrecordstests |
| Operations tested | quote attribution by 11 widely used LLMs across prompt settings op |
| What survived | large and systematic disparities in attribution accuracy between race, gender and intersectional groups; suppression widespread and unevenly distributed op |
op the operator's own statement · 3p a third-party account · obs our observation · rd our reading. Hover a tag for its quote; each opens below with its source.
| Canonical name | AttriBench |
| Also known as | Attribution Bias in Large Language Models |
| Address | https://huggingface.co/datasets/bermaneh/AttriBench |
| Operator | Eliza Berman, Bella Chang, Daniel B. Neill, Emily Black |
| Kind | benchmark |
| Unit | a quote of 5–30 words with its canonical author |
| Lifecycle | published — accepted at COLM 2026 op |
| Count | 15,620 quotes from 6,292 unique authors across two splits (as of dataset card) op |
| Licence | CC BY 4.0 |
| Versions and dates | v1: 2026-04-06; v2: 2026-09-24; read: 2026-10-29 op |
| Observation | 2026-10-29: read |
4cd45d850843388a…4cd45d850843388a…4cd45d850843388a…4cd45d850843388a…3a243bf599199af1…4cd45d850843388a…4cd45d850843388a…3a243bf599199af1…| AttributionBench | claim support from cited evidence, OSU NLP Group, macro-F1 |
4cd45d850843388a…3a243bf599199af1…Lee Sharks, ed. "AttriBench." Provenance meta-aggregator, Provenance Erasure, compiled 2026-10-29. https://www.provenanceerasure.org/aggregators/attribench
Record https://www.provenanceerasure.org/aggregators/attribench · entity https://www.provenanceerasure.org/aggregators/attribench#entity · the census as data
@misc{pe-attribench2026,
editor = {Sharks, Lee},
title = {AttriBench},
howpublished = {Provenance meta-aggregator, Provenance Erasure},
year = {2026},
note = {Compiled 2026-10-29; every field sourced and dated},
url = {https://www.provenanceerasure.org/aggregators/attribench}
}