{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:YU2U2MWX5NFUXGCDGUU7BFE4YJ","short_pith_number":"pith:YU2U2MWX","schema_version":"1.0","canonical_sha256":"c5354d32d7eb4b4b98433529f0949cc24009cf43ef2a509254c41038004f1eff","source":{"kind":"arxiv","id":"2605.19908","version":1},"attestation_state":"computed","paper":{"title":"Where Does Authorship Signal Emerge in Encoder-Based Language Models?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Florian Cafiero, Francis Kulumba, Guillaume Vimont, Laurent Romary","submitted_at":"2026-05-19T14:37:51Z","abstract_excerpt":"Authorship attribution models fine-tuned with the same pretrained encoder, data, and loss can differ four-fold in performance depending only on their scoring mechanism. We use mechanistic interpretability tools to explain this gap. Stylistic features such as word length, punctuation density, and function-word frequency are equally available at every layer in every model, including in an off-the-shelf control encoder, hence the gap not coming from representation quality. Instead, causal intervention shows that the scorer determines where the encoder consolidates authorship signal. Mean pooling "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.19908","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-19T14:37:51Z","cross_cats_sorted":[],"title_canon_sha256":"3844f410aa8dc7115df8336192e045e516d8f8f12cbbac460d20fbad9e36e0d2","abstract_canon_sha256":"f5b53a80ab82bc89e5e8c2fe7735a026814fe8fd0d89731bff12fcc5b0152ba4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T02:05:54.826642Z","signature_b64":"wrOUYWIPnFX8/FwvBiccv5GjgeT28iSX+dz7Oy0Dk33IK4U+Zg+oijME+xtvtBEy10JmWYJOdbtKgfXc7lWVCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c5354d32d7eb4b4b98433529f0949cc24009cf43ef2a509254c41038004f1eff","last_reissued_at":"2026-05-20T02:05:54.825575Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T02:05:54.825575Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Where Does Authorship Signal Emerge in Encoder-Based Language Models?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Florian Cafiero, Francis Kulumba, Guillaume Vimont, Laurent Romary","submitted_at":"2026-05-19T14:37:51Z","abstract_excerpt":"Authorship attribution models fine-tuned with the same pretrained encoder, data, and loss can differ four-fold in performance depending only on their scoring mechanism. We use mechanistic interpretability tools to explain this gap. Stylistic features such as word length, punctuation density, and function-word frequency are equally available at every layer in every model, including in an off-the-shelf control encoder, hence the gap not coming from representation quality. Instead, causal intervention shows that the scorer determines where the encoder consolidates authorship signal. Mean pooling "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.19908","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.19908/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.19908","created_at":"2026-05-20T02:05:54.825723+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.19908v1","created_at":"2026-05-20T02:05:54.825723+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.19908","created_at":"2026-05-20T02:05:54.825723+00:00"},{"alias_kind":"pith_short_12","alias_value":"YU2U2MWX5NFU","created_at":"2026-05-20T02:05:54.825723+00:00"},{"alias_kind":"pith_short_16","alias_value":"YU2U2MWX5NFUXGCD","created_at":"2026-05-20T02:05:54.825723+00:00"},{"alias_kind":"pith_short_8","alias_value":"YU2U2MWX","created_at":"2026-05-20T02:05:54.825723+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YU2U2MWX5NFUXGCDGUU7BFE4YJ","json":"https://pith.science/pith/YU2U2MWX5NFUXGCDGUU7BFE4YJ.json","graph_json":"https://pith.science/api/pith-number/YU2U2MWX5NFUXGCDGUU7BFE4YJ/graph.json","events_json":"https://pith.science/api/pith-number/YU2U2MWX5NFUXGCDGUU7BFE4YJ/events.json","paper":"https://pith.science/paper/YU2U2MWX"},"agent_actions":{"view_html":"https://pith.science/pith/YU2U2MWX5NFUXGCDGUU7BFE4YJ","download_json":"https://pith.science/pith/YU2U2MWX5NFUXGCDGUU7BFE4YJ.json","view_paper":"https://pith.science/paper/YU2U2MWX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.19908&json=true","fetch_graph":"https://pith.science/api/pith-number/YU2U2MWX5NFUXGCDGUU7BFE4YJ/graph.json","fetch_events":"https://pith.science/api/pith-number/YU2U2MWX5NFUXGCDGUU7BFE4YJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YU2U2MWX5NFUXGCDGUU7BFE4YJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YU2U2MWX5NFUXGCDGUU7BFE4YJ/action/storage_attestation","attest_author":"https://pith.science/pith/YU2U2MWX5NFUXGCDGUU7BFE4YJ/action/author_attestation","sign_citation":"https://pith.science/pith/YU2U2MWX5NFUXGCDGUU7BFE4YJ/action/citation_signature","submit_replication":"https://pith.science/pith/YU2U2MWX5NFUXGCDGUU7BFE4YJ/action/replication_record"}},"created_at":"2026-05-20T02:05:54.825723+00:00","updated_at":"2026-05-20T02:05:54.825723+00:00"}