{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:WSYXKACKAMZ554Q4DZ63CSSNTY","short_pith_number":"pith:WSYXKACK","schema_version":"1.0","canonical_sha256":"b4b175004a0333def21c1e7db14a4d9e2f93dea519d9945ee092a4c209cc25ed","source":{"kind":"arxiv","id":"2606.06286","version":1},"attestation_state":"computed","paper":{"title":"LLMs Can Leak Training Data But Do They Want To? A Propensity-Aware Evaluation of Memorization in LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Gianluca Barmina, Lukas Galke Poech, Peter Schneider-Kamp","submitted_at":"2026-06-04T15:25:24Z","abstract_excerpt":"Large language models can reproduce training data, but existing memorization evaluations mostly measure whether models can be forced to do so, rather than whether they do so under ordinary use. We introduce PropMe, a propensity-aware framework for memorization evaluation that contrasts prefix-based capability attacks with non-adversarial evaluations. We propose a metric transformation that, applied to existing functions, allows to create propensity metrics. We further introduce SimpleTrace, a lightweight tracing pipeline built on infini-gram that deterministically attributes model generations "},"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":"2606.06286","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-04T15:25:24Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1541308ab731dc5c3228b34e6be5e98954dd27ea3bbaff64ac5eb5f4741fcb4c","abstract_canon_sha256":"e4ee217d28ba62573776051440ab1fa2f9d13c4cc6b97c1875e3992e5f7d713e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-05T01:15:40.571443Z","signature_b64":"+9pFau1hfusz53iWYGtSPrV+c9SZ8NFNbxsIjp0U4aHeTntSgLq+dKRMftwcNc0JZH3U0yLjylO/F3+bjMf0DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b4b175004a0333def21c1e7db14a4d9e2f93dea519d9945ee092a4c209cc25ed","last_reissued_at":"2026-06-05T01:15:40.571092Z","signature_status":"signed_v1","first_computed_at":"2026-06-05T01:15:40.571092Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLMs Can Leak Training Data But Do They Want To? A Propensity-Aware Evaluation of Memorization in LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Gianluca Barmina, Lukas Galke Poech, Peter Schneider-Kamp","submitted_at":"2026-06-04T15:25:24Z","abstract_excerpt":"Large language models can reproduce training data, but existing memorization evaluations mostly measure whether models can be forced to do so, rather than whether they do so under ordinary use. We introduce PropMe, a propensity-aware framework for memorization evaluation that contrasts prefix-based capability attacks with non-adversarial evaluations. We propose a metric transformation that, applied to existing functions, allows to create propensity metrics. We further introduce SimpleTrace, a lightweight tracing pipeline built on infini-gram that deterministically attributes model generations "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.06286","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/2606.06286/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":"2606.06286","created_at":"2026-06-05T01:15:40.571150+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.06286v1","created_at":"2026-06-05T01:15:40.571150+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.06286","created_at":"2026-06-05T01:15:40.571150+00:00"},{"alias_kind":"pith_short_12","alias_value":"WSYXKACKAMZ5","created_at":"2026-06-05T01:15:40.571150+00:00"},{"alias_kind":"pith_short_16","alias_value":"WSYXKACKAMZ554Q4","created_at":"2026-06-05T01:15:40.571150+00:00"},{"alias_kind":"pith_short_8","alias_value":"WSYXKACK","created_at":"2026-06-05T01:15:40.571150+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/WSYXKACKAMZ554Q4DZ63CSSNTY","json":"https://pith.science/pith/WSYXKACKAMZ554Q4DZ63CSSNTY.json","graph_json":"https://pith.science/api/pith-number/WSYXKACKAMZ554Q4DZ63CSSNTY/graph.json","events_json":"https://pith.science/api/pith-number/WSYXKACKAMZ554Q4DZ63CSSNTY/events.json","paper":"https://pith.science/paper/WSYXKACK"},"agent_actions":{"view_html":"https://pith.science/pith/WSYXKACKAMZ554Q4DZ63CSSNTY","download_json":"https://pith.science/pith/WSYXKACKAMZ554Q4DZ63CSSNTY.json","view_paper":"https://pith.science/paper/WSYXKACK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.06286&json=true","fetch_graph":"https://pith.science/api/pith-number/WSYXKACKAMZ554Q4DZ63CSSNTY/graph.json","fetch_events":"https://pith.science/api/pith-number/WSYXKACKAMZ554Q4DZ63CSSNTY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WSYXKACKAMZ554Q4DZ63CSSNTY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WSYXKACKAMZ554Q4DZ63CSSNTY/action/storage_attestation","attest_author":"https://pith.science/pith/WSYXKACKAMZ554Q4DZ63CSSNTY/action/author_attestation","sign_citation":"https://pith.science/pith/WSYXKACKAMZ554Q4DZ63CSSNTY/action/citation_signature","submit_replication":"https://pith.science/pith/WSYXKACKAMZ554Q4DZ63CSSNTY/action/replication_record"}},"created_at":"2026-06-05T01:15:40.571150+00:00","updated_at":"2026-06-05T01:15:40.571150+00:00"}