Integrity report for Large language models reorganize representational geometry during in-context learning

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

arXiv:2605.28854 · pith:2026:ZSWHGW2SSJENQZTGYYBFEMWZRF

Paper page arXiv integrity.json bundle.json

Detector runs

Findings

Signed record

The machine-readable record for this paper lives at /pith/ZSWHGW2S/integrity.json. Pith Number bundles also include signed pith.integrity.v1 events where a Pith Number exists.