{"as_of":"2026-07-21T18:35:00Z","canonical_paper_id":"4CAA2XJ2","canonical_url":"https://pith.science/paper/4CAA2XJ2/integrity","context_digest":"sha256:e9fa5902b181b18f6e7ca9244a26044aa853e4ff87919d447f3b0a7980565de3","coverage_summary":[],"findings":[],"findings_empty_copy":"No checks are listed for this paper yet.","findings_observation_state":"no_listed_checks","links":{"arxiv":"https://arxiv.org/abs/2111.08181","bundle_json":"/pith/4CAA2XJ2A2XT4VAJO2DVZWL532/bundle.json","citation_record":"/paper/4CAA2XJ2/citation-record","evidence":"/evidence","html":"/paper/4CAA2XJ2/integrity","integrity_json":"/paper/4CAA2XJ2/integrity.json","json":"/paper/4CAA2XJ2/integrity.json","legacy_integrity_json":"/pith/4CAA2XJ2/integrity.json","paper":"/paper/4CAA2XJ2","reference_changes":"/flags?citing=4CAA2XJ2"},"paper":{"arxiv_id":"2111.08181","latest_version":1,"primary_cat":"cs.CL","title":"Adversarially Constructed Evaluation Sets Are More Challenging, but May Not Be Fair"},"pith_number":"pith:2021:4CAA2XJ2A2XT4VAJO2DVZWL532","pith_short":"4CAA2XJ2A2XT4VAJO2DVZWL532","record":{"arxiv_id":"2111.08181","claim":"As of an unrecorded date, Pith completed 0 of 0 listed checks against arXiv:2111.08181. No listed check completed, so this record makes no findings claim.","counts":{"completed":0,"failed":0,"listed":0,"not_collected":0,"not_requested":0,"partial":0,"public_findings_from_completed":0,"skipped":0,"unavailable":0,"withheld":0},"coverage":[],"observed_at":null,"refusal":"This is a record of named checks, not a clean-status claim or a paper verdict.","schema":"pith.integrity.v1"},"schema":"pith.paper-integrity-record.v1","species":"LEDGER"}