{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:AMVJRKJRPKYFCEXGYQYPFLKZZB","short_pith_number":"pith:AMVJRKJR","schema_version":"1.0","canonical_sha256":"032a98a9317ab05112e6c430f2ad59c854a8b3f4b770752922bc1f08dc21e6e2","source":{"kind":"arxiv","id":"2408.07471","version":4},"attestation_state":"computed","paper":{"title":"Bridging and Modeling Correlations in Pairwise Data for Direct Preference Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bo Huang, Liangyou Li, Lifeng Shang, Ruiming Tang, Wei Wang, Xingshan Zeng, Xin Jiang, Yasheng Wang, Yufei Wang, Yuxin Jiang","submitted_at":"2024-08-14T11:29:47Z","abstract_excerpt":"Direct preference optimization (DPO), a widely adopted offline preference optimization algorithm, aims to align large language models (LLMs) with human-desired behaviors using pairwise preference data. However, the generation of the winning response and the losing response within pairwise data are typically isolated, leading to weak correlations between them as well as suboptimal alignment performance. To address this issue, we propose an effective framework for Bridging and Modeling Correlations in pairwise data, named BMC. Firstly, we increase the consistency and informativeness of the pairw"},"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":"2408.07471","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-08-14T11:29:47Z","cross_cats_sorted":[],"title_canon_sha256":"dd72e82d987d916c4f0447389a106ce9c2b513f0dbb4a28a3a2cf8fd21bb1fd6","abstract_canon_sha256":"1f27c4b8e86aef9ecdf9dac391706603fbbe562256a74ab4e23f71a1fbe754e4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:15:45.187079Z","signature_b64":"4ikEz0vCLvBy+MpfSdIySLwu98gQBwsjqh7r1+gibg3UilGoNZ01FueChXPIJeyYraxjN8P29vE+RgihyZuJCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"032a98a9317ab05112e6c430f2ad59c854a8b3f4b770752922bc1f08dc21e6e2","last_reissued_at":"2026-07-05T10:15:45.186436Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:15:45.186436Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bridging and Modeling Correlations in Pairwise Data for Direct Preference Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bo Huang, Liangyou Li, Lifeng Shang, Ruiming Tang, Wei Wang, Xingshan Zeng, Xin Jiang, Yasheng Wang, Yufei Wang, Yuxin Jiang","submitted_at":"2024-08-14T11:29:47Z","abstract_excerpt":"Direct preference optimization (DPO), a widely adopted offline preference optimization algorithm, aims to align large language models (LLMs) with human-desired behaviors using pairwise preference data. However, the generation of the winning response and the losing response within pairwise data are typically isolated, leading to weak correlations between them as well as suboptimal alignment performance. To address this issue, we propose an effective framework for Bridging and Modeling Correlations in pairwise data, named BMC. Firstly, we increase the consistency and informativeness of the pairw"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.07471","kind":"arxiv","version":4},"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/2408.07471/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":"2408.07471","created_at":"2026-07-05T10:15:45.186583+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.07471v4","created_at":"2026-07-05T10:15:45.186583+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.07471","created_at":"2026-07-05T10:15:45.186583+00:00"},{"alias_kind":"pith_short_12","alias_value":"AMVJRKJRPKYF","created_at":"2026-07-05T10:15:45.186583+00:00"},{"alias_kind":"pith_short_16","alias_value":"AMVJRKJRPKYFCEXG","created_at":"2026-07-05T10:15:45.186583+00:00"},{"alias_kind":"pith_short_8","alias_value":"AMVJRKJR","created_at":"2026-07-05T10:15:45.186583+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.10908","citing_title":"Order Matters: LVLMs as Judges for Temporal Reasoning in Image Sequences","ref_index":25,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AMVJRKJRPKYFCEXGYQYPFLKZZB","json":"https://pith.science/pith/AMVJRKJRPKYFCEXGYQYPFLKZZB.json","graph_json":"https://pith.science/api/pith-number/AMVJRKJRPKYFCEXGYQYPFLKZZB/graph.json","events_json":"https://pith.science/api/pith-number/AMVJRKJRPKYFCEXGYQYPFLKZZB/events.json","paper":"https://pith.science/paper/AMVJRKJR"},"agent_actions":{"view_html":"https://pith.science/pith/AMVJRKJRPKYFCEXGYQYPFLKZZB","download_json":"https://pith.science/pith/AMVJRKJRPKYFCEXGYQYPFLKZZB.json","view_paper":"https://pith.science/paper/AMVJRKJR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.07471&json=true","fetch_graph":"https://pith.science/api/pith-number/AMVJRKJRPKYFCEXGYQYPFLKZZB/graph.json","fetch_events":"https://pith.science/api/pith-number/AMVJRKJRPKYFCEXGYQYPFLKZZB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AMVJRKJRPKYFCEXGYQYPFLKZZB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AMVJRKJRPKYFCEXGYQYPFLKZZB/action/storage_attestation","attest_author":"https://pith.science/pith/AMVJRKJRPKYFCEXGYQYPFLKZZB/action/author_attestation","sign_citation":"https://pith.science/pith/AMVJRKJRPKYFCEXGYQYPFLKZZB/action/citation_signature","submit_replication":"https://pith.science/pith/AMVJRKJRPKYFCEXGYQYPFLKZZB/action/replication_record"}},"created_at":"2026-07-05T10:15:45.186583+00:00","updated_at":"2026-07-05T10:15:45.186583+00:00"}