{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:SQ4IF5TY3BRGKD6F5BMVSVN2P4","short_pith_number":"pith:SQ4IF5TY","schema_version":"1.0","canonical_sha256":"943882f678d862650fc5e8595955ba7f3ca47bbaa704f4b6609af2855da68ed2","source":{"kind":"arxiv","id":"2603.04678","version":3},"attestation_state":"computed","paper":{"title":"Post-Training Language Models for Crosslingual Consistency","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Direct Consistency Optimization derives a reward from the LLM itself to produce consistent answers to the same question across languages.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Arianna Bisazza, Jirui Qi, Mrinmaya Sachan, Raquel Fern\\'andez, Ryan Cotterell, Tianyu Liu","submitted_at":"2026-03-04T23:36:55Z","abstract_excerpt":"Language models often respond inconsistently to translation-equivalent prompts across languages, undermining the reliability of multilingual systems. To quantify this, we give an information-theoretic definition of crosslingual consistency as a divergence bound between a model's response distribution and its round-trip pushforward across languages. We then introduce penalized consistency optimization (PCO), a post-training procedure that couples this divergence with a Kullback-Leibler penalty to a fixed reference language model. Because direct optimization of PCO requires expensive on-policy r"},"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":"2603.04678","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-03-04T23:36:55Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"09d72fb9a9527f525d87a573e62294c876568a39a6fa759e384c3ba68e37c252","abstract_canon_sha256":"75c3b87e8a501f274f185f6199870b7e203119e421ae1f2b71846f8e7dc76e19"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T02:05:42.408602Z","signature_b64":"keGp/SVUMJ8uItKKIiRbk3kd3dWsVsh6nlitV15oCMlyO2N0bVtb4muTCV7wslqdfQ5QHHjCRS7/T5gDZuXcAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"943882f678d862650fc5e8595955ba7f3ca47bbaa704f4b6609af2855da68ed2","last_reissued_at":"2026-05-29T02:05:42.407885Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T02:05:42.407885Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Post-Training Language Models for Crosslingual Consistency","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Direct Consistency Optimization derives a reward from the LLM itself to produce consistent answers to the same question across languages.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Arianna Bisazza, Jirui Qi, Mrinmaya Sachan, Raquel Fern\\'andez, Ryan Cotterell, Tianyu Liu","submitted_at":"2026-03-04T23:36:55Z","abstract_excerpt":"Language models often respond inconsistently to translation-equivalent prompts across languages, undermining the reliability of multilingual systems. To quantify this, we give an information-theoretic definition of crosslingual consistency as a divergence bound between a model's response distribution and its round-trip pushforward across languages. We then introduce penalized consistency optimization (PCO), a post-training procedure that couples this divergence with a Kullback-Leibler penalty to a fixed reference language model. Because direct optimization of PCO requires expensive on-policy r"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"DCO significantly improves crosslingual consistency across diverse LLMs and outperforms existing methods when training with samples of multiple languages, while complementing DPO when gold labels are available.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a structured reward function derived directly from the LLM itself can reliably enforce true crosslingual knowledge consistency without introducing new inconsistencies or biases during optimization.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"DCO optimizes LLMs for crosslingual knowledge consistency via a structured reward derived directly from the model itself, outperforming prior methods in experiments.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Direct Consistency Optimization derives a reward from the LLM itself to produce consistent answers to the same question across languages.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"23b41dd4fd1d4743d6821102c7648dea214a93948081ab4b459126d9f4079a7f"},"source":{"id":"2603.04678","kind":"arxiv","version":3},"verdict":{"id":"61f392d0-127c-43ad-b170-703d36fdd0a3","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T15:56:49.533431Z","strongest_claim":"DCO significantly improves crosslingual consistency across diverse LLMs and outperforms existing methods when training with samples of multiple languages, while complementing DPO when gold labels are available.","one_line_summary":"DCO optimizes LLMs for crosslingual knowledge consistency via a structured reward derived directly from the model itself, outperforming prior methods in experiments.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a structured reward function derived directly from the LLM itself can reliably enforce true crosslingual knowledge consistency without introducing new inconsistencies or biases during optimization.","pith_extraction_headline":"Direct Consistency Optimization derives a reward from the LLM itself to produce consistent answers to the same question across languages."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.04678/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":"2603.04678","created_at":"2026-05-29T02:05:42.407986+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.04678v3","created_at":"2026-05-29T02:05:42.407986+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.04678","created_at":"2026-05-29T02:05:42.407986+00:00"},{"alias_kind":"pith_short_12","alias_value":"SQ4IF5TY3BRG","created_at":"2026-05-29T02:05:42.407986+00:00"},{"alias_kind":"pith_short_16","alias_value":"SQ4IF5TY3BRGKD6F","created_at":"2026-05-29T02:05:42.407986+00:00"},{"alias_kind":"pith_short_8","alias_value":"SQ4IF5TY","created_at":"2026-05-29T02:05:42.407986+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/SQ4IF5TY3BRGKD6F5BMVSVN2P4","json":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4.json","graph_json":"https://pith.science/api/pith-number/SQ4IF5TY3BRGKD6F5BMVSVN2P4/graph.json","events_json":"https://pith.science/api/pith-number/SQ4IF5TY3BRGKD6F5BMVSVN2P4/events.json","paper":"https://pith.science/paper/SQ4IF5TY"},"agent_actions":{"view_html":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4","download_json":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4.json","view_paper":"https://pith.science/paper/SQ4IF5TY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.04678&json=true","fetch_graph":"https://pith.science/api/pith-number/SQ4IF5TY3BRGKD6F5BMVSVN2P4/graph.json","fetch_events":"https://pith.science/api/pith-number/SQ4IF5TY3BRGKD6F5BMVSVN2P4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/action/storage_attestation","attest_author":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/action/author_attestation","sign_citation":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/action/citation_signature","submit_replication":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/action/replication_record"}},"created_at":"2026-05-29T02:05:42.407986+00:00","updated_at":"2026-05-29T02:05:42.407986+00:00"}