Integrity report for One prompt is not enough: Instruction Sensitivity Undermines Embedding Model Evaluation

A machine-verified record of the checks Pith has run against this paper: detector runs, findings, signed bundle events, and canonical identifiers.

arXiv:2605.22544 · pith:2026:Q24F4WTDUUZJBKFTRIHYLH3REB

0Critical

1Advisory

7Detectors run

2026-05-25Last checked

Paper page arXiv integrity.json bundle.json

Detector runs

doi_compliance completed v1.0.0 · findings 1 · 2026-05-25 06:48:35.211524+00:00

doi_title_agreement completed v1.0.0 · findings 0 · 2026-05-25 06:32:40.219296+00:00

claim_evidence completed v1.0.0 · findings 0 · 2026-05-25 01:23:38.643642+00:00

citation_quote_validity completed v0.1.0 · findings 0 · 2026-05-24 05:52:55.721089+00:00

shingle_duplication completed v0.1.0 · findings 0 · 2026-05-22 21:51:53.206059+00:00

cited_work_retraction completed v1.0.0 · findings 0 · 2026-05-22 05:54:07.233057+00:00

ai_meta_artifact skipped v1.0.0 · findings 0 · 2026-05-22 01:33:41.878895+00:00

Findings

No public integrity findings for this paper.

Signed record

The machine-readable record for this paper lives at /pith/Q24F4WTD/integrity.json. Pith Number bundles also include signed pith.integrity.v1 events where a Pith Number exists.