{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:TRM3KP4NL57CBNSDDKJUC5TEVD","short_pith_number":"pith:TRM3KP4N","schema_version":"1.0","canonical_sha256":"9c59b53f8d5f7e20b6431a93417664a8c4802301af783083cd1f67460230a885","source":{"kind":"arxiv","id":"2311.08685","version":3},"attestation_state":"computed","paper":{"title":"Safer-Instruct: Aligning Language Models with Automated Preference Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jieyu Zhao, Kai Chen, Taiwei Shi","submitted_at":"2023-11-15T04:22:22Z","abstract_excerpt":"Reinforcement learning from human feedback (RLHF) is a vital strategy for enhancing model capability in language models. However, annotating preference data for RLHF is a resource-intensive and creativity-demanding process, while existing automatic generation methods face limitations in data diversity and quality. In response, we present Safer-Instruct, a novel pipeline for automatically constructing large-scale preference data. Our approach leverages reversed instruction tuning, instruction induction, and expert model evaluation to efficiently generate high-quality preference data without hum"},"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":"2311.08685","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-15T04:22:22Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"54248f96b87e83f31571c15c7d56c02d2f345a0b07406872980a571b7225cd53","abstract_canon_sha256":"2e82bd6ce887ecbc989803307395f8466b49f7225e563641208572b8ece37970"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:02:47.875688Z","signature_b64":"cKm0GsMLiyjdQrgVmf1JHE8vTxPURl7h0WtljnHMAwVQmCSYOfErfWXsJQu4CNwULgn0QjC+FOnYtqaJJ2uYCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9c59b53f8d5f7e20b6431a93417664a8c4802301af783083cd1f67460230a885","last_reissued_at":"2026-07-05T08:02:47.875187Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:02:47.875187Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Safer-Instruct: Aligning Language Models with Automated Preference Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jieyu Zhao, Kai Chen, Taiwei Shi","submitted_at":"2023-11-15T04:22:22Z","abstract_excerpt":"Reinforcement learning from human feedback (RLHF) is a vital strategy for enhancing model capability in language models. However, annotating preference data for RLHF is a resource-intensive and creativity-demanding process, while existing automatic generation methods face limitations in data diversity and quality. In response, we present Safer-Instruct, a novel pipeline for automatically constructing large-scale preference data. Our approach leverages reversed instruction tuning, instruction induction, and expert model evaluation to efficiently generate high-quality preference data without hum"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.08685","kind":"arxiv","version":3},"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/2311.08685/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":"2311.08685","created_at":"2026-07-05T08:02:47.875247+00:00"},{"alias_kind":"arxiv_version","alias_value":"2311.08685v3","created_at":"2026-07-05T08:02:47.875247+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.08685","created_at":"2026-07-05T08:02:47.875247+00:00"},{"alias_kind":"pith_short_12","alias_value":"TRM3KP4NL57C","created_at":"2026-07-05T08:02:47.875247+00:00"},{"alias_kind":"pith_short_16","alias_value":"TRM3KP4NL57CBNSD","created_at":"2026-07-05T08:02:47.875247+00:00"},{"alias_kind":"pith_short_8","alias_value":"TRM3KP4N","created_at":"2026-07-05T08:02:47.875247+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.24536","citing_title":"Generating Place-Based Compromises Between Two Points of View","ref_index":62,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TRM3KP4NL57CBNSDDKJUC5TEVD","json":"https://pith.science/pith/TRM3KP4NL57CBNSDDKJUC5TEVD.json","graph_json":"https://pith.science/api/pith-number/TRM3KP4NL57CBNSDDKJUC5TEVD/graph.json","events_json":"https://pith.science/api/pith-number/TRM3KP4NL57CBNSDDKJUC5TEVD/events.json","paper":"https://pith.science/paper/TRM3KP4N"},"agent_actions":{"view_html":"https://pith.science/pith/TRM3KP4NL57CBNSDDKJUC5TEVD","download_json":"https://pith.science/pith/TRM3KP4NL57CBNSDDKJUC5TEVD.json","view_paper":"https://pith.science/paper/TRM3KP4N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2311.08685&json=true","fetch_graph":"https://pith.science/api/pith-number/TRM3KP4NL57CBNSDDKJUC5TEVD/graph.json","fetch_events":"https://pith.science/api/pith-number/TRM3KP4NL57CBNSDDKJUC5TEVD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TRM3KP4NL57CBNSDDKJUC5TEVD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TRM3KP4NL57CBNSDDKJUC5TEVD/action/storage_attestation","attest_author":"https://pith.science/pith/TRM3KP4NL57CBNSDDKJUC5TEVD/action/author_attestation","sign_citation":"https://pith.science/pith/TRM3KP4NL57CBNSDDKJUC5TEVD/action/citation_signature","submit_replication":"https://pith.science/pith/TRM3KP4NL57CBNSDDKJUC5TEVD/action/replication_record"}},"created_at":"2026-07-05T08:02:47.875247+00:00","updated_at":"2026-07-05T08:02:47.875247+00:00"}