{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:SQ4IF5TY3BRGKD6F5BMVSVN2P4","short_pith_number":"pith:SQ4IF5TY","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"},"canonical_sha256":"943882f678d862650fc5e8595955ba7f3ca47bbaa704f4b6609af2855da68ed2","source":{"kind":"arxiv","id":"2603.04678","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2603.04678","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"arxiv_version","alias_value":"2603.04678v3","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.04678","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"pith_short_12","alias_value":"SQ4IF5TY3BRG","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"pith_short_16","alias_value":"SQ4IF5TY3BRGKD6F","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"pith_short_8","alias_value":"SQ4IF5TY","created_at":"2026-05-29T02:05:42Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:SQ4IF5TY3BRGKD6F5BMVSVN2P4","target":"record","payload":{"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"},"canonical_sha256":"943882f678d862650fc5e8595955ba7f3ca47bbaa704f4b6609af2855da68ed2","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"},"source_kind":"arxiv","source_id":"2603.04678","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-29T02:05:42Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"YDLJEW7Uan14hMz/1psnXtIIwWMQv9nstyC2U2hiNOA5ofamHpNvN3qE5i7sEGbjoZDOFUfN0iSXXk2d62oYCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-10T06:35:50.725929Z"},"content_sha256":"2a7b3b8c812d1cbd45f8812c3503db990b0afbea727d9792425eebb3287b2230","schema_version":"1.0","event_id":"sha256:2a7b3b8c812d1cbd45f8812c3503db990b0afbea727d9792425eebb3287b2230"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:SQ4IF5TY3BRGKD6F5BMVSVN2P4","target":"graph","payload":{"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"},"verdict_id":"61f392d0-127c-43ad-b170-703d36fdd0a3"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-29T02:05:42Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0hAwF4Hjrt+bwuQeIOFliUdQ8xAfePfbug/tVYTQiNtbReqxy+ih5688TPt/o/DS/81lF0ThmxkJ/NwWABmfBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-10T06:35:50.726960Z"},"content_sha256":"a2f35c0fc180c745857d143d46f1eb0c7ce4a280ff5a85003d30636a9dc69165","schema_version":"1.0","event_id":"sha256:a2f35c0fc180c745857d143d46f1eb0c7ce4a280ff5a85003d30636a9dc69165"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/bundle.json","state_url":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-10T06:35:50Z","links":{"resolver":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4","bundle":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/bundle.json","state":"https://pith.science/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/state.json","well_known_bundle":"https://pith.science/.well-known/pith/SQ4IF5TY3BRGKD6F5BMVSVN2P4/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:SQ4IF5TY3BRGKD6F5BMVSVN2P4","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"75c3b87e8a501f274f185f6199870b7e203119e421ae1f2b71846f8e7dc76e19","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-03-04T23:36:55Z","title_canon_sha256":"09d72fb9a9527f525d87a573e62294c876568a39a6fa759e384c3ba68e37c252"},"schema_version":"1.0","source":{"id":"2603.04678","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2603.04678","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"arxiv_version","alias_value":"2603.04678v3","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.04678","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"pith_short_12","alias_value":"SQ4IF5TY3BRG","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"pith_short_16","alias_value":"SQ4IF5TY3BRGKD6F","created_at":"2026-05-29T02:05:42Z"},{"alias_kind":"pith_short_8","alias_value":"SQ4IF5TY","created_at":"2026-05-29T02:05:42Z"}],"graph_snapshots":[{"event_id":"sha256:a2f35c0fc180c745857d143d46f1eb0c7ce4a280ff5a85003d30636a9dc69165","target":"graph","created_at":"2026-05-29T02:05:42Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"DCO optimizes LLMs for crosslingual knowledge consistency via a structured reward derived directly from the model itself, outperforming prior methods in experiments."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Direct Consistency Optimization derives a reward from the LLM itself to produce consistent answers to the same question across languages."}],"snapshot_sha256":"23b41dd4fd1d4743d6821102c7648dea214a93948081ab4b459126d9f4079a7f"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2603.04678/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"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","authors_text":"Arianna Bisazza, Jirui Qi, Mrinmaya Sachan, Raquel Fern\\'andez, Ryan Cotterell, Tianyu Liu","cross_cats":["cs.AI"],"headline":"Direct Consistency Optimization derives a reward from the LLM itself to produce consistent answers to the same question across languages.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-03-04T23:36:55Z","title":"Post-Training Language Models for Crosslingual Consistency"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.04678","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-15T15:56:49.533431Z","id":"61f392d0-127c-43ad-b170-703d36fdd0a3","model_set":{"reader":"grok-4.3"},"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","pith_extraction_headline":"Direct Consistency Optimization derives a reward from the LLM itself to produce consistent answers to the same question across languages.","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.","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."}},"verdict_id":"61f392d0-127c-43ad-b170-703d36fdd0a3"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:2a7b3b8c812d1cbd45f8812c3503db990b0afbea727d9792425eebb3287b2230","target":"record","created_at":"2026-05-29T02:05:42Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"75c3b87e8a501f274f185f6199870b7e203119e421ae1f2b71846f8e7dc76e19","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-03-04T23:36:55Z","title_canon_sha256":"09d72fb9a9527f525d87a573e62294c876568a39a6fa759e384c3ba68e37c252"},"schema_version":"1.0","source":{"id":"2603.04678","kind":"arxiv","version":3}},"canonical_sha256":"943882f678d862650fc5e8595955ba7f3ca47bbaa704f4b6609af2855da68ed2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"943882f678d862650fc5e8595955ba7f3ca47bbaa704f4b6609af2855da68ed2","first_computed_at":"2026-05-29T02:05:42.407885Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-29T02:05:42.407885Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"keGp/SVUMJ8uItKKIiRbk3kd3dWsVsh6nlitV15oCMlyO2N0bVtb4muTCV7wslqdfQ5QHHjCRS7/T5gDZuXcAQ==","signature_status":"signed_v1","signed_at":"2026-05-29T02:05:42.408602Z","signed_message":"canonical_sha256_bytes"},"source_id":"2603.04678","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:2a7b3b8c812d1cbd45f8812c3503db990b0afbea727d9792425eebb3287b2230","sha256:a2f35c0fc180c745857d143d46f1eb0c7ce4a280ff5a85003d30636a9dc69165"],"state_sha256":"aa6b94786c60dde40862cd5bf71b6cb19a68f412ea8356a93902f8d9fc1fca90"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ZYQEpMsqq0gu3HVYPwL/AHsHqUH4676c+oxPUYZ8uTrLYW75ogDTCvOzNX6I3lsrgjfolhXYxthDJJd+SgPoBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-10T06:35:50.731657Z","bundle_sha256":"990c5499e3ad7d91c285ae9bbda505138c5a2954d4fbc691738ad7cbdc7dc73c"}}