{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MWAFXI2J2FQBCWYQVFTUYZWMFZ","short_pith_number":"pith:MWAFXI2J","schema_version":"1.0","canonical_sha256":"65805ba349d160115b10a9674c66cc2e535072cb9664968f4f0dd038c27ec707","source":{"kind":"arxiv","id":"2410.07672","version":2},"attestation_state":"computed","paper":{"title":"MACPO: Weak-to-Strong Alignment via Multi-Agent Contrastive Preference Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Dawei Yin, Lingyong Yan, Maarten de Rijke, Pengjie Ren, Yougang Lyu, Zhaochun Ren, Zihan Wang","submitted_at":"2024-10-10T07:29:35Z","abstract_excerpt":"As large language models (LLMs) are rapidly advancing and achieving near-human capabilities on specific tasks, aligning them with human values is becoming more urgent. In scenarios where LLMs outperform humans, we face a weak-to-strong alignment problem where we need to effectively align strong student LLMs through weak supervision generated by weak teachers. Existing alignment methods mainly focus on strong-to-weak alignment and self-alignment settings, and it is impractical to adapt them to the much harder weak-to-strong alignment setting. To fill this gap, we propose a multi-agent contrasti"},"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":"2410.07672","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-10T07:29:35Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c02eb7146816c78a83950cfef93b9f9cbd4d5339c21d202804e82f3d1cbcbd05","abstract_canon_sha256":"8a9f3267afde74c94160c6c19ea940b2c30b13fbb85d558b5912ff3ee8ab01ff"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:22:07.191856Z","signature_b64":"yE4f+7DDcPVn9zK0XNtUZb1b06XnSv4QkV5AxQ7uBETH4sVAwTorZMQzxzxQ8ng2wOr12m+ufTFYADGQ3bZVAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"65805ba349d160115b10a9674c66cc2e535072cb9664968f4f0dd038c27ec707","last_reissued_at":"2026-07-05T10:22:07.191306Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:22:07.191306Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MACPO: Weak-to-Strong Alignment via Multi-Agent Contrastive Preference Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Dawei Yin, Lingyong Yan, Maarten de Rijke, Pengjie Ren, Yougang Lyu, Zhaochun Ren, Zihan Wang","submitted_at":"2024-10-10T07:29:35Z","abstract_excerpt":"As large language models (LLMs) are rapidly advancing and achieving near-human capabilities on specific tasks, aligning them with human values is becoming more urgent. In scenarios where LLMs outperform humans, we face a weak-to-strong alignment problem where we need to effectively align strong student LLMs through weak supervision generated by weak teachers. Existing alignment methods mainly focus on strong-to-weak alignment and self-alignment settings, and it is impractical to adapt them to the much harder weak-to-strong alignment setting. To fill this gap, we propose a multi-agent contrasti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.07672","kind":"arxiv","version":2},"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/2410.07672/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":"2410.07672","created_at":"2026-07-05T10:22:07.191393+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.07672v2","created_at":"2026-07-05T10:22:07.191393+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.07672","created_at":"2026-07-05T10:22:07.191393+00:00"},{"alias_kind":"pith_short_12","alias_value":"MWAFXI2J2FQB","created_at":"2026-07-05T10:22:07.191393+00:00"},{"alias_kind":"pith_short_16","alias_value":"MWAFXI2J2FQBCWYQ","created_at":"2026-07-05T10:22:07.191393+00:00"},{"alias_kind":"pith_short_8","alias_value":"MWAFXI2J","created_at":"2026-07-05T10:22:07.191393+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/MWAFXI2J2FQBCWYQVFTUYZWMFZ","json":"https://pith.science/pith/MWAFXI2J2FQBCWYQVFTUYZWMFZ.json","graph_json":"https://pith.science/api/pith-number/MWAFXI2J2FQBCWYQVFTUYZWMFZ/graph.json","events_json":"https://pith.science/api/pith-number/MWAFXI2J2FQBCWYQVFTUYZWMFZ/events.json","paper":"https://pith.science/paper/MWAFXI2J"},"agent_actions":{"view_html":"https://pith.science/pith/MWAFXI2J2FQBCWYQVFTUYZWMFZ","download_json":"https://pith.science/pith/MWAFXI2J2FQBCWYQVFTUYZWMFZ.json","view_paper":"https://pith.science/paper/MWAFXI2J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.07672&json=true","fetch_graph":"https://pith.science/api/pith-number/MWAFXI2J2FQBCWYQVFTUYZWMFZ/graph.json","fetch_events":"https://pith.science/api/pith-number/MWAFXI2J2FQBCWYQVFTUYZWMFZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MWAFXI2J2FQBCWYQVFTUYZWMFZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MWAFXI2J2FQBCWYQVFTUYZWMFZ/action/storage_attestation","attest_author":"https://pith.science/pith/MWAFXI2J2FQBCWYQVFTUYZWMFZ/action/author_attestation","sign_citation":"https://pith.science/pith/MWAFXI2J2FQBCWYQVFTUYZWMFZ/action/citation_signature","submit_replication":"https://pith.science/pith/MWAFXI2J2FQBCWYQVFTUYZWMFZ/action/replication_record"}},"created_at":"2026-07-05T10:22:07.191393+00:00","updated_at":"2026-07-05T10:22:07.191393+00:00"}