{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:KIVSEI3CXEBJKDMRECVKNCUNWC","short_pith_number":"pith:KIVSEI3C","schema_version":"1.0","canonical_sha256":"522b222362b902950d9120aaa68a8db08d3acf7570e5c6cc804a604ee9bb76d0","source":{"kind":"arxiv","id":"2402.11525","version":3},"attestation_state":"computed","paper":{"title":"Advancing Translation Preference Modeling with RLHF: A Step Towards Cost-Effective Solution","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Can Zu, Jun Zhao, Lu Chen, Nuo Xu, Qi Zhang, Rui Zheng, Shihan Dou, Sixian Li, Tao Gui, Wenjuan Qin, Xuanjing Huang, Zhihao Zhang","submitted_at":"2024-02-18T09:51:49Z","abstract_excerpt":"Faithfulness, expressiveness, and elegance is the constant pursuit in machine translation. However, traditional metrics like \\textit{BLEU} do not strictly align with human preference of translation quality. In this paper, we explore leveraging reinforcement learning with human feedback (\\textit{RLHF}) to improve translation quality. It is non-trivial to collect a large high-quality dataset of human comparisons between translations, especially for low-resource languages. To address this issue, we propose a cost-effective preference learning strategy, optimizing reward models by distinguishing b"},"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":"2402.11525","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-18T09:51:49Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"27fc468c5607fbebb5b32d63c4913ca2817bb434c82328f6af8608e10b4807f9","abstract_canon_sha256":"b4b6d46008fcd350bd345fea611a4dcf778843ecc7005eb94c112abec059cb01"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:49:56.207559Z","signature_b64":"c8xKUeRy5ok3bUlBM1ETWSR5erlAnfV4IPN5iIf2dV9Uh1f3qbPvBIWORvzcAhARH8gfUOgDaki1Nw4YkByZDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"522b222362b902950d9120aaa68a8db08d3acf7570e5c6cc804a604ee9bb76d0","last_reissued_at":"2026-07-05T07:49:56.206915Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:49:56.206915Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Advancing Translation Preference Modeling with RLHF: A Step Towards Cost-Effective Solution","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Can Zu, Jun Zhao, Lu Chen, Nuo Xu, Qi Zhang, Rui Zheng, Shihan Dou, Sixian Li, Tao Gui, Wenjuan Qin, Xuanjing Huang, Zhihao Zhang","submitted_at":"2024-02-18T09:51:49Z","abstract_excerpt":"Faithfulness, expressiveness, and elegance is the constant pursuit in machine translation. However, traditional metrics like \\textit{BLEU} do not strictly align with human preference of translation quality. In this paper, we explore leveraging reinforcement learning with human feedback (\\textit{RLHF}) to improve translation quality. It is non-trivial to collect a large high-quality dataset of human comparisons between translations, especially for low-resource languages. To address this issue, we propose a cost-effective preference learning strategy, optimizing reward models by distinguishing b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.11525","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/2402.11525/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":"2402.11525","created_at":"2026-07-05T07:49:56.206988+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.11525v3","created_at":"2026-07-05T07:49:56.206988+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.11525","created_at":"2026-07-05T07:49:56.206988+00:00"},{"alias_kind":"pith_short_12","alias_value":"KIVSEI3CXEBJ","created_at":"2026-07-05T07:49:56.206988+00:00"},{"alias_kind":"pith_short_16","alias_value":"KIVSEI3CXEBJKDMR","created_at":"2026-07-05T07:49:56.206988+00:00"},{"alias_kind":"pith_short_8","alias_value":"KIVSEI3C","created_at":"2026-07-05T07:49:56.206988+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2504.01919","citing_title":"Bridging the Linguistic Divide: A Survey on Leveraging Large Language Models for Machine Translation","ref_index":97,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KIVSEI3CXEBJKDMRECVKNCUNWC","json":"https://pith.science/pith/KIVSEI3CXEBJKDMRECVKNCUNWC.json","graph_json":"https://pith.science/api/pith-number/KIVSEI3CXEBJKDMRECVKNCUNWC/graph.json","events_json":"https://pith.science/api/pith-number/KIVSEI3CXEBJKDMRECVKNCUNWC/events.json","paper":"https://pith.science/paper/KIVSEI3C"},"agent_actions":{"view_html":"https://pith.science/pith/KIVSEI3CXEBJKDMRECVKNCUNWC","download_json":"https://pith.science/pith/KIVSEI3CXEBJKDMRECVKNCUNWC.json","view_paper":"https://pith.science/paper/KIVSEI3C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.11525&json=true","fetch_graph":"https://pith.science/api/pith-number/KIVSEI3CXEBJKDMRECVKNCUNWC/graph.json","fetch_events":"https://pith.science/api/pith-number/KIVSEI3CXEBJKDMRECVKNCUNWC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KIVSEI3CXEBJKDMRECVKNCUNWC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KIVSEI3CXEBJKDMRECVKNCUNWC/action/storage_attestation","attest_author":"https://pith.science/pith/KIVSEI3CXEBJKDMRECVKNCUNWC/action/author_attestation","sign_citation":"https://pith.science/pith/KIVSEI3CXEBJKDMRECVKNCUNWC/action/citation_signature","submit_replication":"https://pith.science/pith/KIVSEI3CXEBJKDMRECVKNCUNWC/action/replication_record"}},"created_at":"2026-07-05T07:49:56.206988+00:00","updated_at":"2026-07-05T07:49:56.206988+00:00"}