{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JJPCWYRHVSXEOQE5JASQ4TMTF2","short_pith_number":"pith:JJPCWYRH","schema_version":"1.0","canonical_sha256":"4a5e2b6227acae47409d48250e4d932e8db5439e00cd0b42d504e8a363582bcb","source":{"kind":"arxiv","id":"2506.07165","version":1},"attestation_state":"computed","paper":{"title":"AMoPO: Adaptive Multi-objective Preference Optimization without Reward Models and Reference Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Desheng Wang, Haodong Zhao, Hao Li, Jiansong Chen, Jingqing Ruan, Qi Liu, Tong Xu, Wan Guanglu, Xunliang Cai, Zhi Zheng","submitted_at":"2025-06-08T14:31:06Z","abstract_excerpt":"Existing multi-objective preference alignment methods for large language models (LLMs) face limitations: (1) the inability to effectively balance various preference dimensions, and (2) reliance on auxiliary reward/reference models introduces computational complexity. To address these challenges, we propose Adaptive Multi-objective Preference Optimization (AMoPO), a novel framework that achieves dynamic balance across preference dimensions. By introducing the multi-objective optimization paradigm to use the dimension-aware generation metrics as implicit rewards, AMoPO aligns LLMs with diverse p"},"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":"2506.07165","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-08T14:31:06Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"3f1f30c4aeae65a451434644363d2f26099a5f2e6ad8a08183004781c4b48f45","abstract_canon_sha256":"65bb7dc84714133d296a94f8b3d8f0339af0f7485b645d2bcda65f00a6093e16"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:18:02.887985Z","signature_b64":"Yko3V8Nuntro6n5c8eABr9wrCZLpYtsXHE+80dMEXmYkprC2HSzO6kqlSaWk60p2WviRhZ/DzQkn65G7IYaHCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4a5e2b6227acae47409d48250e4d932e8db5439e00cd0b42d504e8a363582bcb","last_reissued_at":"2026-07-05T11:18:02.887483Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:18:02.887483Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AMoPO: Adaptive Multi-objective Preference Optimization without Reward Models and Reference Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Desheng Wang, Haodong Zhao, Hao Li, Jiansong Chen, Jingqing Ruan, Qi Liu, Tong Xu, Wan Guanglu, Xunliang Cai, Zhi Zheng","submitted_at":"2025-06-08T14:31:06Z","abstract_excerpt":"Existing multi-objective preference alignment methods for large language models (LLMs) face limitations: (1) the inability to effectively balance various preference dimensions, and (2) reliance on auxiliary reward/reference models introduces computational complexity. To address these challenges, we propose Adaptive Multi-objective Preference Optimization (AMoPO), a novel framework that achieves dynamic balance across preference dimensions. By introducing the multi-objective optimization paradigm to use the dimension-aware generation metrics as implicit rewards, AMoPO aligns LLMs with diverse p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.07165","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/2506.07165/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":"2506.07165","created_at":"2026-07-05T11:18:02.887538+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.07165v1","created_at":"2026-07-05T11:18:02.887538+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.07165","created_at":"2026-07-05T11:18:02.887538+00:00"},{"alias_kind":"pith_short_12","alias_value":"JJPCWYRHVSXE","created_at":"2026-07-05T11:18:02.887538+00:00"},{"alias_kind":"pith_short_16","alias_value":"JJPCWYRHVSXEOQE5","created_at":"2026-07-05T11:18:02.887538+00:00"},{"alias_kind":"pith_short_8","alias_value":"JJPCWYRH","created_at":"2026-07-05T11:18:02.887538+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2509.20265","citing_title":"Failure Modes of Maximum Entropy RLHF","ref_index":32,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JJPCWYRHVSXEOQE5JASQ4TMTF2","json":"https://pith.science/pith/JJPCWYRHVSXEOQE5JASQ4TMTF2.json","graph_json":"https://pith.science/api/pith-number/JJPCWYRHVSXEOQE5JASQ4TMTF2/graph.json","events_json":"https://pith.science/api/pith-number/JJPCWYRHVSXEOQE5JASQ4TMTF2/events.json","paper":"https://pith.science/paper/JJPCWYRH"},"agent_actions":{"view_html":"https://pith.science/pith/JJPCWYRHVSXEOQE5JASQ4TMTF2","download_json":"https://pith.science/pith/JJPCWYRHVSXEOQE5JASQ4TMTF2.json","view_paper":"https://pith.science/paper/JJPCWYRH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.07165&json=true","fetch_graph":"https://pith.science/api/pith-number/JJPCWYRHVSXEOQE5JASQ4TMTF2/graph.json","fetch_events":"https://pith.science/api/pith-number/JJPCWYRHVSXEOQE5JASQ4TMTF2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JJPCWYRHVSXEOQE5JASQ4TMTF2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JJPCWYRHVSXEOQE5JASQ4TMTF2/action/storage_attestation","attest_author":"https://pith.science/pith/JJPCWYRHVSXEOQE5JASQ4TMTF2/action/author_attestation","sign_citation":"https://pith.science/pith/JJPCWYRHVSXEOQE5JASQ4TMTF2/action/citation_signature","submit_replication":"https://pith.science/pith/JJPCWYRHVSXEOQE5JASQ4TMTF2/action/replication_record"}},"created_at":"2026-07-05T11:18:02.887538+00:00","updated_at":"2026-07-05T11:18:02.887538+00:00"}