{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:YFBVAKCDZEDEWA53FFYKOUN4VR","short_pith_number":"pith:YFBVAKCD","schema_version":"1.0","canonical_sha256":"c143502843c9064b03bb2970a751bcac6a9b97224e1c1859dab991a1ff11a99c","source":{"kind":"arxiv","id":"2406.01660","version":4},"attestation_state":"computed","paper":{"title":"Self-Improving Robust Preference Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Arash Ahmadian, Eugene Choi, Matthieu Geist, Mohammad Gheshlaghi Azar, Oilvier Pietquin","submitted_at":"2024-06-03T17:53:25Z","abstract_excerpt":"Online and offline RLHF methods, such as PPO and DPO, have been highly successful in aligning AI with human preferences. Despite their success, however, these methods suffer from fundamental limitations: (a) Models trained with RLHF can learn from mistakes or negative examples through RL mechanism or contrastive loss during training. However, at inference time, they lack an innate self-improvement mechanism for error corrections. (b) The optimal solution of existing methods is highly task-dependent, making it difficult for them to generalize to new tasks. To address these challenges, we propos"},"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":"2406.01660","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-03T17:53:25Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"4a3110d111fe819e514d8b27789318969f421db63f03d9e17d66305b6147b42c","abstract_canon_sha256":"4401318839c49cc2ae1a4c7be2ef669c4bbcd863f1ab6ce6e7ae8a388ad39ec3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:47:54.606982Z","signature_b64":"165zQslMw5Ntot9U8z93eRY6N+pODpwMh04ZatxmvO3gOVdKyAzNROESiWiahOkLt7hw5fo6p2ZoFllsgjwaAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c143502843c9064b03bb2970a751bcac6a9b97224e1c1859dab991a1ff11a99c","last_reissued_at":"2026-07-05T10:47:54.606500Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:47:54.606500Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Improving Robust Preference Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Arash Ahmadian, Eugene Choi, Matthieu Geist, Mohammad Gheshlaghi Azar, Oilvier Pietquin","submitted_at":"2024-06-03T17:53:25Z","abstract_excerpt":"Online and offline RLHF methods, such as PPO and DPO, have been highly successful in aligning AI with human preferences. Despite their success, however, these methods suffer from fundamental limitations: (a) Models trained with RLHF can learn from mistakes or negative examples through RL mechanism or contrastive loss during training. However, at inference time, they lack an innate self-improvement mechanism for error corrections. (b) The optimal solution of existing methods is highly task-dependent, making it difficult for them to generalize to new tasks. To address these challenges, we propos"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.01660","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/2406.01660/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":"2406.01660","created_at":"2026-07-05T10:47:54.606557+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.01660v4","created_at":"2026-07-05T10:47:54.606557+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.01660","created_at":"2026-07-05T10:47:54.606557+00:00"},{"alias_kind":"pith_short_12","alias_value":"YFBVAKCDZEDE","created_at":"2026-07-05T10:47:54.606557+00:00"},{"alias_kind":"pith_short_16","alias_value":"YFBVAKCDZEDEWA53","created_at":"2026-07-05T10:47:54.606557+00:00"},{"alias_kind":"pith_short_8","alias_value":"YFBVAKCD","created_at":"2026-07-05T10:47:54.606557+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YFBVAKCDZEDEWA53FFYKOUN4VR","json":"https://pith.science/pith/YFBVAKCDZEDEWA53FFYKOUN4VR.json","graph_json":"https://pith.science/api/pith-number/YFBVAKCDZEDEWA53FFYKOUN4VR/graph.json","events_json":"https://pith.science/api/pith-number/YFBVAKCDZEDEWA53FFYKOUN4VR/events.json","paper":"https://pith.science/paper/YFBVAKCD"},"agent_actions":{"view_html":"https://pith.science/pith/YFBVAKCDZEDEWA53FFYKOUN4VR","download_json":"https://pith.science/pith/YFBVAKCDZEDEWA53FFYKOUN4VR.json","view_paper":"https://pith.science/paper/YFBVAKCD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.01660&json=true","fetch_graph":"https://pith.science/api/pith-number/YFBVAKCDZEDEWA53FFYKOUN4VR/graph.json","fetch_events":"https://pith.science/api/pith-number/YFBVAKCDZEDEWA53FFYKOUN4VR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YFBVAKCDZEDEWA53FFYKOUN4VR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YFBVAKCDZEDEWA53FFYKOUN4VR/action/storage_attestation","attest_author":"https://pith.science/pith/YFBVAKCDZEDEWA53FFYKOUN4VR/action/author_attestation","sign_citation":"https://pith.science/pith/YFBVAKCDZEDEWA53FFYKOUN4VR/action/citation_signature","submit_replication":"https://pith.science/pith/YFBVAKCDZEDEWA53FFYKOUN4VR/action/replication_record"}},"created_at":"2026-07-05T10:47:54.606557+00:00","updated_at":"2026-07-05T10:47:54.606557+00:00"}