{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:NRKPXRUEZTSLQEYDL4TY2MTJND","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"09d914937117c78ab329abaf67f1d6e02c14dae08841a17410929cf43269947a","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-12-05T21:50:22Z","title_canon_sha256":"182ac3e49c2769bffa390ca94b42c0826b59ca522a2b4bda6a7f8b0589261ef8"},"schema_version":"1.0","source":{"id":"2412.04628","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2412.04628","created_at":"2026-07-05T11:24:09Z"},{"alias_kind":"arxiv_version","alias_value":"2412.04628v4","created_at":"2026-07-05T11:24:09Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.04628","created_at":"2026-07-05T11:24:09Z"},{"alias_kind":"pith_short_12","alias_value":"NRKPXRUEZTSL","created_at":"2026-07-05T11:24:09Z"},{"alias_kind":"pith_short_16","alias_value":"NRKPXRUEZTSLQEYD","created_at":"2026-07-05T11:24:09Z"},{"alias_kind":"pith_short_8","alias_value":"NRKPXRUE","created_at":"2026-07-05T11:24:09Z"}],"graph_snapshots":[{"event_id":"sha256:3d8f050cbfcda69f3eeaf10713795c966236cb3bf409f34a18b65a3f4729788e","target":"graph","created_at":"2026-07-05T11:24:09Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2412.04628/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Direct Preference Optimization (DPO) has become a popular approach for aligning language models using pairwise preferences. However, in practical post-training pipelines, on-policy generation typically yields multiple candidate responses per prompt, which are scored by a reward model to guide learning. In this setting, we propose $\\textbf{Multi-Preference Optimization (MPO)}$, a generalization of DPO that optimizes over entire sets of responses by extending the Bradley-Terry model to groupwise comparisons between chosen and rejected sets. To further enhance learning, MPO employs deviation-base","authors_text":"Chetan Bansal, Nagarajan Natarajan, Rahul Madhavan, Saravan Rajmohan, Taneesh Gupta, Xuchao Zhang","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-12-05T21:50:22Z","title":"Multi-Preference Optimization: Generalizing DPO via Set-Level Contrasts"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.04628","kind":"arxiv","version":4},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:d33609732a41482bd4f083905ad1c3a989d226c8c954a0d8d18ef7c5c6af00a2","target":"record","created_at":"2026-07-05T11:24:09Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"09d914937117c78ab329abaf67f1d6e02c14dae08841a17410929cf43269947a","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-12-05T21:50:22Z","title_canon_sha256":"182ac3e49c2769bffa390ca94b42c0826b59ca522a2b4bda6a7f8b0589261ef8"},"schema_version":"1.0","source":{"id":"2412.04628","kind":"arxiv","version":4}},"canonical_sha256":"6c54fbc684cce4b813035f278d326968d197afb80d5ff686d40288fab00b17a4","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"6c54fbc684cce4b813035f278d326968d197afb80d5ff686d40288fab00b17a4","first_computed_at":"2026-07-05T11:24:09.785839Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:24:09.785839Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"QYWziJZrJcFSa+lTW4/jc7wECFqvVmmqDaXenS8dalUW+BufMjL8D16fSgG8kcPRQWSksWPDcrd0trkx3kD6CQ==","signature_status":"signed_v1","signed_at":"2026-07-05T11:24:09.786394Z","signed_message":"canonical_sha256_bytes"},"source_id":"2412.04628","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d33609732a41482bd4f083905ad1c3a989d226c8c954a0d8d18ef7c5c6af00a2","sha256:3d8f050cbfcda69f3eeaf10713795c966236cb3bf409f34a18b65a3f4729788e"],"state_sha256":"05b06253f06ce0393d2712e7a577efd279e13d2b02d71afb437b4da5bb99d494"}