{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:J5RHQWZI72HOYKHLQRJ4AMMUOQ","short_pith_number":"pith:J5RHQWZI","schema_version":"1.0","canonical_sha256":"4f62785b28fe8eec28eb8453c03194742a84a5b0226e6378ec6644a2559befd9","source":{"kind":"arxiv","id":"2607.02966","version":1},"attestation_state":"computed","paper":{"title":"Distill Where the Student Goes: Teacher-Regularized RL for English-Evidence Cross-Lingual RAG","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Haotian Zhou, Siqi Liu, Weiran Huang, Xin Zhang, Xiting Wang, Zhihao Wen","submitted_at":"2026-07-03T05:17:48Z","abstract_excerpt":"Cross-lingual retrieval-augmented generation (RAG) is often deployed in an English-evidence regime, where users query in diverse languages but retrieved passages remain English. In this setting, generation can fail despite strong base models: English evidence induces language drift (English or code-switching outputs) and models use evidence unreliably when producing non-English answers. We attribute these failures to two post-training challenges: (i) errors are prefix-dependent, so fixed-trajectory supervision suffers from prefix mismatch; and (ii) sequence-level (partly discrete / judge-based"},"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":"2607.02966","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-03T05:17:48Z","cross_cats_sorted":[],"title_canon_sha256":"51d8d6067ed88729c9846139a7dd97a43125160a10ce1a8e67a467c88da18342","abstract_canon_sha256":"4d805e848c4027507149645f1f2444818164d97c7b4f5c238ac19ca6a6c951da"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T01:16:38.076565Z","signature_b64":"We8uz/6h64uW4JrRjeDC+r7IeQKYBuVrEHYSODGmqP1VyOqcKancdPWHfTA6C+WmKoNFEbiZDU5xIJLkYEQcCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4f62785b28fe8eec28eb8453c03194742a84a5b0226e6378ec6644a2559befd9","last_reissued_at":"2026-07-07T01:16:38.076108Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T01:16:38.076108Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Distill Where the Student Goes: Teacher-Regularized RL for English-Evidence Cross-Lingual RAG","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Haotian Zhou, Siqi Liu, Weiran Huang, Xin Zhang, Xiting Wang, Zhihao Wen","submitted_at":"2026-07-03T05:17:48Z","abstract_excerpt":"Cross-lingual retrieval-augmented generation (RAG) is often deployed in an English-evidence regime, where users query in diverse languages but retrieved passages remain English. In this setting, generation can fail despite strong base models: English evidence induces language drift (English or code-switching outputs) and models use evidence unreliably when producing non-English answers. We attribute these failures to two post-training challenges: (i) errors are prefix-dependent, so fixed-trajectory supervision suffers from prefix mismatch; and (ii) sequence-level (partly discrete / judge-based"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.02966","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/2607.02966/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":"2607.02966","created_at":"2026-07-07T01:16:38.076169+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.02966v1","created_at":"2026-07-07T01:16:38.076169+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.02966","created_at":"2026-07-07T01:16:38.076169+00:00"},{"alias_kind":"pith_short_12","alias_value":"J5RHQWZI72HO","created_at":"2026-07-07T01:16:38.076169+00:00"},{"alias_kind":"pith_short_16","alias_value":"J5RHQWZI72HOYKHL","created_at":"2026-07-07T01:16:38.076169+00:00"},{"alias_kind":"pith_short_8","alias_value":"J5RHQWZI","created_at":"2026-07-07T01:16:38.076169+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/J5RHQWZI72HOYKHLQRJ4AMMUOQ","json":"https://pith.science/pith/J5RHQWZI72HOYKHLQRJ4AMMUOQ.json","graph_json":"https://pith.science/api/pith-number/J5RHQWZI72HOYKHLQRJ4AMMUOQ/graph.json","events_json":"https://pith.science/api/pith-number/J5RHQWZI72HOYKHLQRJ4AMMUOQ/events.json","paper":"https://pith.science/paper/J5RHQWZI"},"agent_actions":{"view_html":"https://pith.science/pith/J5RHQWZI72HOYKHLQRJ4AMMUOQ","download_json":"https://pith.science/pith/J5RHQWZI72HOYKHLQRJ4AMMUOQ.json","view_paper":"https://pith.science/paper/J5RHQWZI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.02966&json=true","fetch_graph":"https://pith.science/api/pith-number/J5RHQWZI72HOYKHLQRJ4AMMUOQ/graph.json","fetch_events":"https://pith.science/api/pith-number/J5RHQWZI72HOYKHLQRJ4AMMUOQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J5RHQWZI72HOYKHLQRJ4AMMUOQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J5RHQWZI72HOYKHLQRJ4AMMUOQ/action/storage_attestation","attest_author":"https://pith.science/pith/J5RHQWZI72HOYKHLQRJ4AMMUOQ/action/author_attestation","sign_citation":"https://pith.science/pith/J5RHQWZI72HOYKHLQRJ4AMMUOQ/action/citation_signature","submit_replication":"https://pith.science/pith/J5RHQWZI72HOYKHLQRJ4AMMUOQ/action/replication_record"}},"created_at":"2026-07-07T01:16:38.076169+00:00","updated_at":"2026-07-07T01:16:38.076169+00:00"}