{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ROUKFXH5NQTX2J53JZ6TXZ455T","short_pith_number":"pith:ROUKFXH5","schema_version":"1.0","canonical_sha256":"8ba8a2dcfd6c277d27bb4e7d3be79decd91eee969b7cd4ea74d3057382bdaa8f","source":{"kind":"arxiv","id":"2505.17140","version":1},"attestation_state":"computed","paper":{"title":"Data Doping or True Intelligence? Evaluating the Transferability of Injected Knowledge in LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Essa Jan, Fareed Zaffar, Moiz Ali, Muhammad Saram Hassan, Yasir Zaki","submitted_at":"2025-05-22T06:48:33Z","abstract_excerpt":"As the knowledge of large language models (LLMs) becomes outdated over time, there is a growing need for efficient methods to update them, especially when injecting proprietary information. Our study reveals that comprehension-intensive fine-tuning tasks (e.g., question answering and blanks) achieve substantially higher knowledge retention rates (48%) compared to mapping-oriented tasks like translation (17%) or text-to-JSON conversion (20%), despite exposure to identical factual content. We demonstrate that this pattern persists across model architectures and follows scaling laws, with larger "},"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":"2505.17140","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-22T06:48:33Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"f086bd806840afd8e61e1ec6035b6ba7854b6eb012f2f391d0c3c61fc8a9c61d","abstract_canon_sha256":"52edfc35312f56bd79d87b05733cdb4428219a521c372ea2200b5edafd9a21fe"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:07:47.672380Z","signature_b64":"LE9TEkyK5NkPWLa0Vu4UiaGehenIYR18+mDjJyBSzLQmSBjxjoTHTh9xqjT98EAdAJ6nL509rEOn7tBJQpNHCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8ba8a2dcfd6c277d27bb4e7d3be79decd91eee969b7cd4ea74d3057382bdaa8f","last_reissued_at":"2026-07-05T11:07:47.671784Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:07:47.671784Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Data Doping or True Intelligence? Evaluating the Transferability of Injected Knowledge in LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Essa Jan, Fareed Zaffar, Moiz Ali, Muhammad Saram Hassan, Yasir Zaki","submitted_at":"2025-05-22T06:48:33Z","abstract_excerpt":"As the knowledge of large language models (LLMs) becomes outdated over time, there is a growing need for efficient methods to update them, especially when injecting proprietary information. Our study reveals that comprehension-intensive fine-tuning tasks (e.g., question answering and blanks) achieve substantially higher knowledge retention rates (48%) compared to mapping-oriented tasks like translation (17%) or text-to-JSON conversion (20%), despite exposure to identical factual content. We demonstrate that this pattern persists across model architectures and follows scaling laws, with larger "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.17140","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/2505.17140/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":"2505.17140","created_at":"2026-07-05T11:07:47.671847+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.17140v1","created_at":"2026-07-05T11:07:47.671847+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.17140","created_at":"2026-07-05T11:07:47.671847+00:00"},{"alias_kind":"pith_short_12","alias_value":"ROUKFXH5NQTX","created_at":"2026-07-05T11:07:47.671847+00:00"},{"alias_kind":"pith_short_16","alias_value":"ROUKFXH5NQTX2J53","created_at":"2026-07-05T11:07:47.671847+00:00"},{"alias_kind":"pith_short_8","alias_value":"ROUKFXH5","created_at":"2026-07-05T11:07:47.671847+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2511.14774","citing_title":"LiveCLKTBench: Towards Reliable Evaluation of Cross-Lingual Knowledge Transfer in Multilingual LLMs","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ROUKFXH5NQTX2J53JZ6TXZ455T","json":"https://pith.science/pith/ROUKFXH5NQTX2J53JZ6TXZ455T.json","graph_json":"https://pith.science/api/pith-number/ROUKFXH5NQTX2J53JZ6TXZ455T/graph.json","events_json":"https://pith.science/api/pith-number/ROUKFXH5NQTX2J53JZ6TXZ455T/events.json","paper":"https://pith.science/paper/ROUKFXH5"},"agent_actions":{"view_html":"https://pith.science/pith/ROUKFXH5NQTX2J53JZ6TXZ455T","download_json":"https://pith.science/pith/ROUKFXH5NQTX2J53JZ6TXZ455T.json","view_paper":"https://pith.science/paper/ROUKFXH5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.17140&json=true","fetch_graph":"https://pith.science/api/pith-number/ROUKFXH5NQTX2J53JZ6TXZ455T/graph.json","fetch_events":"https://pith.science/api/pith-number/ROUKFXH5NQTX2J53JZ6TXZ455T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ROUKFXH5NQTX2J53JZ6TXZ455T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ROUKFXH5NQTX2J53JZ6TXZ455T/action/storage_attestation","attest_author":"https://pith.science/pith/ROUKFXH5NQTX2J53JZ6TXZ455T/action/author_attestation","sign_citation":"https://pith.science/pith/ROUKFXH5NQTX2J53JZ6TXZ455T/action/citation_signature","submit_replication":"https://pith.science/pith/ROUKFXH5NQTX2J53JZ6TXZ455T/action/replication_record"}},"created_at":"2026-07-05T11:07:47.671847+00:00","updated_at":"2026-07-05T11:07:47.671847+00:00"}