{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:KWFOTOG4RBPIRODYIA5WRRDSAN","short_pith_number":"pith:KWFOTOG4","schema_version":"1.0","canonical_sha256":"558ae9b8dc885e88b878403b68c472034662ce3e5a558434eeda761332b46341","source":{"kind":"arxiv","id":"2311.04919","version":1},"attestation_state":"computed","paper":{"title":"The Impact of Preference Agreement in Reinforcement Learning from Human Feedback: A Case Study in Summarization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.HC"],"primary_cat":"cs.CL","authors_text":"Hassan Mansoor, Sian Gooding","submitted_at":"2023-11-02T13:21:23Z","abstract_excerpt":"Reinforcement Learning from Human Feedback (RLHF) can be used to capture complex and nuanced properties of text generation quality. As a result, the task of text summarization has been identified as a good candidate for this process. In this paper, we explore how preference agreement impacts the efficacy of RLHF for summarization. We show that sampling human preferences to include a range of annotator agreement results in (1) higher accuracy reward models and (2) alters the characteristics of quality captured. We additionally show improvements in downstream generation when using a reward model"},"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":"2311.04919","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-02T13:21:23Z","cross_cats_sorted":["cs.AI","cs.HC"],"title_canon_sha256":"ed38d819240af66b3e832be2aaad5351c1a7af8fa2dbc90db088346f316d833e","abstract_canon_sha256":"65f3cae1b82d5b84f3a998c6a2af01d81fb08bb71fb3d8b8b025d5c931829ce4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:10:51.784338Z","signature_b64":"NatheDdDohBoEpG12qW+SB0cU18KBBnKziv7sOTC0buem+B8AJUuPHRBdwijzAnD7XOpaesjbabS1Utl0xvoCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"558ae9b8dc885e88b878403b68c472034662ce3e5a558434eeda761332b46341","last_reissued_at":"2026-07-05T07:10:51.783934Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:10:51.783934Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Impact of Preference Agreement in Reinforcement Learning from Human Feedback: A Case Study in Summarization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.HC"],"primary_cat":"cs.CL","authors_text":"Hassan Mansoor, Sian Gooding","submitted_at":"2023-11-02T13:21:23Z","abstract_excerpt":"Reinforcement Learning from Human Feedback (RLHF) can be used to capture complex and nuanced properties of text generation quality. As a result, the task of text summarization has been identified as a good candidate for this process. In this paper, we explore how preference agreement impacts the efficacy of RLHF for summarization. We show that sampling human preferences to include a range of annotator agreement results in (1) higher accuracy reward models and (2) alters the characteristics of quality captured. We additionally show improvements in downstream generation when using a reward model"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.04919","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/2311.04919/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":"2311.04919","created_at":"2026-07-05T07:10:51.783990+00:00"},{"alias_kind":"arxiv_version","alias_value":"2311.04919v1","created_at":"2026-07-05T07:10:51.783990+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.04919","created_at":"2026-07-05T07:10:51.783990+00:00"},{"alias_kind":"pith_short_12","alias_value":"KWFOTOG4RBPI","created_at":"2026-07-05T07:10:51.783990+00:00"},{"alias_kind":"pith_short_16","alias_value":"KWFOTOG4RBPIRODY","created_at":"2026-07-05T07:10:51.783990+00:00"},{"alias_kind":"pith_short_8","alias_value":"KWFOTOG4","created_at":"2026-07-05T07:10:51.783990+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.11072","citing_title":"Multilingual Large Language Models: A Systematic Survey","ref_index":6,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KWFOTOG4RBPIRODYIA5WRRDSAN","json":"https://pith.science/pith/KWFOTOG4RBPIRODYIA5WRRDSAN.json","graph_json":"https://pith.science/api/pith-number/KWFOTOG4RBPIRODYIA5WRRDSAN/graph.json","events_json":"https://pith.science/api/pith-number/KWFOTOG4RBPIRODYIA5WRRDSAN/events.json","paper":"https://pith.science/paper/KWFOTOG4"},"agent_actions":{"view_html":"https://pith.science/pith/KWFOTOG4RBPIRODYIA5WRRDSAN","download_json":"https://pith.science/pith/KWFOTOG4RBPIRODYIA5WRRDSAN.json","view_paper":"https://pith.science/paper/KWFOTOG4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2311.04919&json=true","fetch_graph":"https://pith.science/api/pith-number/KWFOTOG4RBPIRODYIA5WRRDSAN/graph.json","fetch_events":"https://pith.science/api/pith-number/KWFOTOG4RBPIRODYIA5WRRDSAN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KWFOTOG4RBPIRODYIA5WRRDSAN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KWFOTOG4RBPIRODYIA5WRRDSAN/action/storage_attestation","attest_author":"https://pith.science/pith/KWFOTOG4RBPIRODYIA5WRRDSAN/action/author_attestation","sign_citation":"https://pith.science/pith/KWFOTOG4RBPIRODYIA5WRRDSAN/action/citation_signature","submit_replication":"https://pith.science/pith/KWFOTOG4RBPIRODYIA5WRRDSAN/action/replication_record"}},"created_at":"2026-07-05T07:10:51.783990+00:00","updated_at":"2026-07-05T07:10:51.783990+00:00"}