{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BNSWY7QMGHTSOFFHKF25FHQNM5","short_pith_number":"pith:BNSWY7QM","schema_version":"1.0","canonical_sha256":"0b656c7e0c31e72714a75175d29e0d67479e372a6f4250ae4146ffb9ee1897d6","source":{"kind":"arxiv","id":"2407.16951","version":1},"attestation_state":"computed","paper":{"title":"Towards Transfer Unlearning: Empirical Evidence of Cross-Domain Bias Mitigation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Huimin Lu, Ichiro Sakata, Junichiro Mori, Masaru Isonuma","submitted_at":"2024-07-24T02:37:42Z","abstract_excerpt":"Large language models (LLMs) often inherit biases from vast amounts of training corpora. Traditional debiasing methods, while effective to some extent, do not completely eliminate memorized biases and toxicity in LLMs. In this paper, we study an unlearning-based approach to debiasing in LLMs by performing gradient ascent on hate speech against minority groups, i.e., minimizing the likelihood of biased or toxic content. Specifically, we propose a mask language modeling unlearning technique, which unlearns the harmful part of the text. This method enables LLMs to selectively forget and disassoci"},"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":"2407.16951","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-24T02:37:42Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"5380ffd4e6faf723444efcc43cbb381db99aec8bace127d3de0564d061476a29","abstract_canon_sha256":"a41b2b0d1388c9d5c6dc231876b9f1652a19f58e43079d0daf568ddd5527a68e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:47:59.759359Z","signature_b64":"mLACjmCXCU4i5AQ6FPPCvBkay8IZeYhJSL7BasHcvR9BYKx3T9mpRP0sHFREDIzTc6eyb4e8s0K2vT8qvThXCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0b656c7e0c31e72714a75175d29e0d67479e372a6f4250ae4146ffb9ee1897d6","last_reissued_at":"2026-07-05T08:47:59.758820Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:47:59.758820Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Transfer Unlearning: Empirical Evidence of Cross-Domain Bias Mitigation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Huimin Lu, Ichiro Sakata, Junichiro Mori, Masaru Isonuma","submitted_at":"2024-07-24T02:37:42Z","abstract_excerpt":"Large language models (LLMs) often inherit biases from vast amounts of training corpora. Traditional debiasing methods, while effective to some extent, do not completely eliminate memorized biases and toxicity in LLMs. In this paper, we study an unlearning-based approach to debiasing in LLMs by performing gradient ascent on hate speech against minority groups, i.e., minimizing the likelihood of biased or toxic content. Specifically, we propose a mask language modeling unlearning technique, which unlearns the harmful part of the text. This method enables LLMs to selectively forget and disassoci"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.16951","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/2407.16951/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":"2407.16951","created_at":"2026-07-05T08:47:59.758876+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.16951v1","created_at":"2026-07-05T08:47:59.758876+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.16951","created_at":"2026-07-05T08:47:59.758876+00:00"},{"alias_kind":"pith_short_12","alias_value":"BNSWY7QMGHTS","created_at":"2026-07-05T08:47:59.758876+00:00"},{"alias_kind":"pith_short_16","alias_value":"BNSWY7QMGHTSOFFH","created_at":"2026-07-05T08:47:59.758876+00:00"},{"alias_kind":"pith_short_8","alias_value":"BNSWY7QM","created_at":"2026-07-05T08:47:59.758876+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/BNSWY7QMGHTSOFFHKF25FHQNM5","json":"https://pith.science/pith/BNSWY7QMGHTSOFFHKF25FHQNM5.json","graph_json":"https://pith.science/api/pith-number/BNSWY7QMGHTSOFFHKF25FHQNM5/graph.json","events_json":"https://pith.science/api/pith-number/BNSWY7QMGHTSOFFHKF25FHQNM5/events.json","paper":"https://pith.science/paper/BNSWY7QM"},"agent_actions":{"view_html":"https://pith.science/pith/BNSWY7QMGHTSOFFHKF25FHQNM5","download_json":"https://pith.science/pith/BNSWY7QMGHTSOFFHKF25FHQNM5.json","view_paper":"https://pith.science/paper/BNSWY7QM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.16951&json=true","fetch_graph":"https://pith.science/api/pith-number/BNSWY7QMGHTSOFFHKF25FHQNM5/graph.json","fetch_events":"https://pith.science/api/pith-number/BNSWY7QMGHTSOFFHKF25FHQNM5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BNSWY7QMGHTSOFFHKF25FHQNM5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BNSWY7QMGHTSOFFHKF25FHQNM5/action/storage_attestation","attest_author":"https://pith.science/pith/BNSWY7QMGHTSOFFHKF25FHQNM5/action/author_attestation","sign_citation":"https://pith.science/pith/BNSWY7QMGHTSOFFHKF25FHQNM5/action/citation_signature","submit_replication":"https://pith.science/pith/BNSWY7QMGHTSOFFHKF25FHQNM5/action/replication_record"}},"created_at":"2026-07-05T08:47:59.758876+00:00","updated_at":"2026-07-05T08:47:59.758876+00:00"}