AttributionBench

Name: AttributionBench
Creator: OSU NLP Group (The Ohio State University)

Provenance meta-aggregator / Answer engines

Answer enginesbenchmarkpublished

AttributionBench is a benchmark from OSU NLP Group (The Ohio State University). The object it attaches provenance to: a generated claim and the evidence it cites.

Visit AttributionBench ↗Run an erasure audit

we present AttributionBench, a comprehensive benchmark compiled from various existing attribution datasets.

In its own words: https://osu-nlp-group.github.io/AttributionBench/, read 2026-10-29

What it carries

What receives provenance	a generated claim and the evidence it cites
Relations carried	support (is the claim fully supported by its cited evidence) op
Carrier	citations in a generated answer
Expresses · records · tests	expressesrecordstests
Operations tested	generation with citations by search-augmented models
What survived	a fine-tuned GPT-3.5 reaches about 80% macro-F1 on the binary task (project abstract) 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.

Entity

Canonical name	AttributionBench
Also known as	AttributionBench (OSU)
Address	https://osu-nlp-group.github.io/AttributionBench/
Operator	OSU NLP Group (The Ohio State University)
Kind	benchmark
Unit	a claim–evidence pair with a binary attribution label
Lifecycle	published — arXiv:2402.15089
Data access	Hugging Face dataset
Licence	apache-2.0 (dataset card) op
Observation	2026-10-29: read

Relations

Neighbours

AttriBench

Evidenceevery field's quote, its source and the hash of the saved copy5

self description op
we present AttributionBench, a comprehensive benchmark compiled from various existing attribution datasets.
https://osu-nlp-group.github.io/AttributionBench/ · read 2026-10-29 · saved copy sha256 0bb85b9388499eb5…
relation types op
whether every claim within the generated responses is fully supported by its cited evidence
https://osu-nlp-group.github.io/AttributionBench/ · read 2026-10-29 · saved copy sha256 0bb85b9388499eb5…
retention results op
even a fine-tuned GPT-3.5 only achieves around 80% macro-F1 under a binary classification formulation.
https://osu-nlp-group.github.io/AttributionBench/ · read 2026-10-29 · saved copy sha256 0bb85b9388499eb5…
license op
license: apache-2.0
https://huggingface.co/datasets/osunlp/AttributionBench/raw/main/README.md · read 2026-10-29 · saved copy sha256 b90e7e73959c0b91…
aggregation op
We constructed this dataset from multiple existing data sources in a unified format
https://huggingface.co/datasets/osunlp/AttributionBench/raw/main/README.md · read 2026-10-29 · saved copy sha256 b90e7e73959c0b91…

Disambiguationthe names it can be confused with1

AttriBench

quote attribution, author, suppression, Berman et al., 2026

Negative tagswhat it is not2

NOT AttriBench: it asks whether evidence supports a claim, not who wrote a quote
evidence support is not credit to the original author

Sourcesevery source read, with its date and hash2

https://osu-nlp-group.github.io/AttributionBench/
read 2026-10-29 · HTTP 200 · sha256 0bb85b9388499eb5…
https://huggingface.co/datasets/osunlp/AttributionBench/raw/main/README.md
read 2026-10-29 · HTTP 200 · sha256 b90e7e73959c0b91…

Not yet checkedwhat remains to read1

the constituent datasets one by one (ExpertQA, BEGIN, HAGRID, AttrScore and others named on the card)

Cite this record

Lee Sharks, ed. "AttributionBench." Provenance meta-aggregator, Provenance Erasure, compiled 2026-10-29. https://www.provenanceerasure.org/aggregators/attributionbench

Record https://www.provenanceerasure.org/aggregators/attributionbench · entity https://www.provenanceerasure.org/aggregators/attributionbench#entity · the census as data

@misc{pe-attributionbench2026,
  editor = {Sharks, Lee},
  title = {AttributionBench},
  howpublished = {Provenance meta-aggregator, Provenance Erasure},
  year = {2026},
  note = {Compiled 2026-10-29; every field sourced and dated},
  url = {https://www.provenanceerasure.org/aggregators/attributionbench}
}

Editor: Lee Sharks (ORCID 0009-0000-1599-0703). This page is the provenance meta-aggregator's record of the resource above: its reading and fields are ours, and each field carries its source, its basis and the hash of the copy read. Compiled 2026-10-29. CC BY 4.0.