{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:DCIG4EOJ27M24CDLYMR7ILCK7I","short_pith_number":"pith:DCIG4EOJ","schema_version":"1.0","canonical_sha256":"18906e11c9d7d9ae086bc323f42c4afa3f71b2043b610336b18b877423ef0b5d","source":{"kind":"arxiv","id":"2406.08414","version":3},"attestation_state":"computed","paper":{"title":"Discovering Preference Optimization Algorithms with and for Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alex J. Chan, Chris Lu, Claudio Fanconi, Jakob Foerster, Mihaela van der Schaar, Robert Tjarko Lange, Samuel Holt","submitted_at":"2024-06-12T16:58:41Z","abstract_excerpt":"Offline preference optimization is a key method for enhancing and controlling the quality of Large Language Model (LLM) outputs. Typically, preference optimization is approached as an offline supervised learning task using manually-crafted convex loss functions. While these methods are based on theoretical insights, they are inherently constrained by human creativity, so the large search space of possible loss functions remains under explored. We address this by performing LLM-driven objective discovery to automatically discover new state-of-the-art preference optimization algorithms without ("},"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":"2406.08414","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-12T16:58:41Z","cross_cats_sorted":[],"title_canon_sha256":"fb543e0e7d09303e3abc8f00dcce1131c49e999fa1bafd68870b1a1c6967fd86","abstract_canon_sha256":"236688c77c24bf92f443dbb0e66701cbc5d09475e7f0c5ccef427021633e92b7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:30:17.056324Z","signature_b64":"Ch+5YHbT8Gn+v0QtFSu7XyDASOzaNPc1NWlDmTmb3WrTCfFg7ibrLW3RQGKGs2C+L9THiN8AES4zJq58e/UdDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"18906e11c9d7d9ae086bc323f42c4afa3f71b2043b610336b18b877423ef0b5d","last_reissued_at":"2026-07-05T09:30:17.055821Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:30:17.055821Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Discovering Preference Optimization Algorithms with and for Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alex J. Chan, Chris Lu, Claudio Fanconi, Jakob Foerster, Mihaela van der Schaar, Robert Tjarko Lange, Samuel Holt","submitted_at":"2024-06-12T16:58:41Z","abstract_excerpt":"Offline preference optimization is a key method for enhancing and controlling the quality of Large Language Model (LLM) outputs. Typically, preference optimization is approached as an offline supervised learning task using manually-crafted convex loss functions. While these methods are based on theoretical insights, they are inherently constrained by human creativity, so the large search space of possible loss functions remains under explored. We address this by performing LLM-driven objective discovery to automatically discover new state-of-the-art preference optimization algorithms without ("},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.08414","kind":"arxiv","version":3},"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/2406.08414/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":"2406.08414","created_at":"2026-07-05T09:30:17.055879+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.08414v3","created_at":"2026-07-05T09:30:17.055879+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.08414","created_at":"2026-07-05T09:30:17.055879+00:00"},{"alias_kind":"pith_short_12","alias_value":"DCIG4EOJ27M2","created_at":"2026-07-05T09:30:17.055879+00:00"},{"alias_kind":"pith_short_16","alias_value":"DCIG4EOJ27M24CDL","created_at":"2026-07-05T09:30:17.055879+00:00"},{"alias_kind":"pith_short_8","alias_value":"DCIG4EOJ","created_at":"2026-07-05T09:30:17.055879+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.20189","citing_title":"SOLAR: A Self-Optimizing Open-Ended Autonomous Agent for Lifelong Learning and Continual Adaptation","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2408.08435","citing_title":"Automated Design of Agentic Systems","ref_index":177,"is_internal_anchor":false},{"citing_arxiv_id":"2408.06292","citing_title":"The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery","ref_index":68,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DCIG4EOJ27M24CDLYMR7ILCK7I","json":"https://pith.science/pith/DCIG4EOJ27M24CDLYMR7ILCK7I.json","graph_json":"https://pith.science/api/pith-number/DCIG4EOJ27M24CDLYMR7ILCK7I/graph.json","events_json":"https://pith.science/api/pith-number/DCIG4EOJ27M24CDLYMR7ILCK7I/events.json","paper":"https://pith.science/paper/DCIG4EOJ"},"agent_actions":{"view_html":"https://pith.science/pith/DCIG4EOJ27M24CDLYMR7ILCK7I","download_json":"https://pith.science/pith/DCIG4EOJ27M24CDLYMR7ILCK7I.json","view_paper":"https://pith.science/paper/DCIG4EOJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.08414&json=true","fetch_graph":"https://pith.science/api/pith-number/DCIG4EOJ27M24CDLYMR7ILCK7I/graph.json","fetch_events":"https://pith.science/api/pith-number/DCIG4EOJ27M24CDLYMR7ILCK7I/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DCIG4EOJ27M24CDLYMR7ILCK7I/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DCIG4EOJ27M24CDLYMR7ILCK7I/action/storage_attestation","attest_author":"https://pith.science/pith/DCIG4EOJ27M24CDLYMR7ILCK7I/action/author_attestation","sign_citation":"https://pith.science/pith/DCIG4EOJ27M24CDLYMR7ILCK7I/action/citation_signature","submit_replication":"https://pith.science/pith/DCIG4EOJ27M24CDLYMR7ILCK7I/action/replication_record"}},"created_at":"2026-07-05T09:30:17.055879+00:00","updated_at":"2026-07-05T09:30:17.055879+00:00"}