{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:BSVARF7M2FETILAECBUTQJ73NF","short_pith_number":"pith:BSVARF7M","schema_version":"1.0","canonical_sha256":"0caa0897ecd149342c0410693827fb697c1244b2700ae9d152da382637dff371","source":{"kind":"arxiv","id":"2606.01382","version":1},"attestation_state":"computed","paper":{"title":"Efficient Exploration for Iterative Nash Preference Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Christian Kroer, Tianlong Nan, Tianyi Lin, Xiaopeng Li","submitted_at":"2026-05-31T18:11:26Z","abstract_excerpt":"Preference alignment is central to improving large language models, but standard reward-based formulations can be restrictive when human preferences are cyclic, non-transitive, or otherwise not representable by a scalar reward. Nash Learning from Human Feedback (NLHF) addresses this limitation by modeling alignment as a preference game and targeting a Nash equilibrium rather than a reward maximizer. However, the learning-theoretic foundations of scalable NLHF remain limited. Existing regret guarantees rely on oracle-based methods that estimate a general preference model and solve KL-regularize"},"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":"2606.01382","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-31T18:11:26Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"acdf81bff36a8c29d91f4eaa06b92daedb53c2e64103c11b9e28ffb4e92dd06c","abstract_canon_sha256":"400472037879443693bb2e84029d9204eb1d8a00aef3db379084e7abfbb393f9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:04:31.867170Z","signature_b64":"UB7jRaTV3qCeJ9+6TU/6OOq/v35nqkAd6fpA/MyWmyqeqMWxXyEh0rpllRLs25NS1z3G2elDrZcZrFyzS8ynCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0caa0897ecd149342c0410693827fb697c1244b2700ae9d152da382637dff371","last_reissued_at":"2026-06-02T02:04:31.866807Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:04:31.866807Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Efficient Exploration for Iterative Nash Preference Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Christian Kroer, Tianlong Nan, Tianyi Lin, Xiaopeng Li","submitted_at":"2026-05-31T18:11:26Z","abstract_excerpt":"Preference alignment is central to improving large language models, but standard reward-based formulations can be restrictive when human preferences are cyclic, non-transitive, or otherwise not representable by a scalar reward. Nash Learning from Human Feedback (NLHF) addresses this limitation by modeling alignment as a preference game and targeting a Nash equilibrium rather than a reward maximizer. However, the learning-theoretic foundations of scalable NLHF remain limited. Existing regret guarantees rely on oracle-based methods that estimate a general preference model and solve KL-regularize"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.01382","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/2606.01382/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":"2606.01382","created_at":"2026-06-02T02:04:31.866865+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.01382v1","created_at":"2026-06-02T02:04:31.866865+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.01382","created_at":"2026-06-02T02:04:31.866865+00:00"},{"alias_kind":"pith_short_12","alias_value":"BSVARF7M2FET","created_at":"2026-06-02T02:04:31.866865+00:00"},{"alias_kind":"pith_short_16","alias_value":"BSVARF7M2FETILAE","created_at":"2026-06-02T02:04:31.866865+00:00"},{"alias_kind":"pith_short_8","alias_value":"BSVARF7M","created_at":"2026-06-02T02:04:31.866865+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/BSVARF7M2FETILAECBUTQJ73NF","json":"https://pith.science/pith/BSVARF7M2FETILAECBUTQJ73NF.json","graph_json":"https://pith.science/api/pith-number/BSVARF7M2FETILAECBUTQJ73NF/graph.json","events_json":"https://pith.science/api/pith-number/BSVARF7M2FETILAECBUTQJ73NF/events.json","paper":"https://pith.science/paper/BSVARF7M"},"agent_actions":{"view_html":"https://pith.science/pith/BSVARF7M2FETILAECBUTQJ73NF","download_json":"https://pith.science/pith/BSVARF7M2FETILAECBUTQJ73NF.json","view_paper":"https://pith.science/paper/BSVARF7M","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.01382&json=true","fetch_graph":"https://pith.science/api/pith-number/BSVARF7M2FETILAECBUTQJ73NF/graph.json","fetch_events":"https://pith.science/api/pith-number/BSVARF7M2FETILAECBUTQJ73NF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BSVARF7M2FETILAECBUTQJ73NF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BSVARF7M2FETILAECBUTQJ73NF/action/storage_attestation","attest_author":"https://pith.science/pith/BSVARF7M2FETILAECBUTQJ73NF/action/author_attestation","sign_citation":"https://pith.science/pith/BSVARF7M2FETILAECBUTQJ73NF/action/citation_signature","submit_replication":"https://pith.science/pith/BSVARF7M2FETILAECBUTQJ73NF/action/replication_record"}},"created_at":"2026-06-02T02:04:31.866865+00:00","updated_at":"2026-06-02T02:04:31.866865+00:00"}