{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JEXNKHTGR7PQQODPPG4TYOSM4D","short_pith_number":"pith:JEXNKHTG","schema_version":"1.0","canonical_sha256":"492ed51e668fdf08386f79b93c3a4ce0d9ba6e4df324bef9243141d26cee3230","source":{"kind":"arxiv","id":"2507.05686","version":1},"attestation_state":"computed","paper":{"title":"Smoothie-Qwen: Post-Hoc Smoothing to Reduce Language Bias in Multilingual LLMs","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jemin Kim, JungYup Lee, Sang Park, Seungjae Lee, SeungWon Ji","submitted_at":"2025-07-08T05:30:51Z","abstract_excerpt":"Multilingual large language models (LLMs) often exhibit language confusion, a tendency to generate responses in a dominant language irrespective of the prompt's language. To address this, we propose Smoothie-Qwen, a lightweight, post-hoc method that mitigates language bias without retraining. This technique selectively adjusts token-level output probabilities to effectively suppress undesired language generation. Applied to the Qwen model, our method reduces unintended Chinese output by over 95% while preserving task accuracy on multilingual benchmarks. This work provides a practical and effic"},"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":"2507.05686","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-08T05:30:51Z","cross_cats_sorted":[],"title_canon_sha256":"33149bf163856ca3fa6ba5bc2f76fd7342005100e042ab66baafe122742838e6","abstract_canon_sha256":"67e2231b3daf3fac78e01ee5402f23ee1d938cef4325c39787d0c34d2bad240d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:33:34.805410Z","signature_b64":"NufJCcVQ5B1bV95YTtSxETfnuGX4KPBfyHs8AlUfhF52CDUxAQofHhMCFZE0yzzGX8ahffQ6y9KkqfBlOeQsDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"492ed51e668fdf08386f79b93c3a4ce0d9ba6e4df324bef9243141d26cee3230","last_reissued_at":"2026-07-05T11:33:34.804852Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:33:34.804852Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Smoothie-Qwen: Post-Hoc Smoothing to Reduce Language Bias in Multilingual LLMs","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jemin Kim, JungYup Lee, Sang Park, Seungjae Lee, SeungWon Ji","submitted_at":"2025-07-08T05:30:51Z","abstract_excerpt":"Multilingual large language models (LLMs) often exhibit language confusion, a tendency to generate responses in a dominant language irrespective of the prompt's language. To address this, we propose Smoothie-Qwen, a lightweight, post-hoc method that mitigates language bias without retraining. This technique selectively adjusts token-level output probabilities to effectively suppress undesired language generation. Applied to the Qwen model, our method reduces unintended Chinese output by over 95% while preserving task accuracy on multilingual benchmarks. This work provides a practical and effic"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.05686","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/2507.05686/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":"2507.05686","created_at":"2026-07-05T11:33:34.804923+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.05686v1","created_at":"2026-07-05T11:33:34.804923+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.05686","created_at":"2026-07-05T11:33:34.804923+00:00"},{"alias_kind":"pith_short_12","alias_value":"JEXNKHTGR7PQ","created_at":"2026-07-05T11:33:34.804923+00:00"},{"alias_kind":"pith_short_16","alias_value":"JEXNKHTGR7PQQODP","created_at":"2026-07-05T11:33:34.804923+00:00"},{"alias_kind":"pith_short_8","alias_value":"JEXNKHTG","created_at":"2026-07-05T11:33:34.804923+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/JEXNKHTGR7PQQODPPG4TYOSM4D","json":"https://pith.science/pith/JEXNKHTGR7PQQODPPG4TYOSM4D.json","graph_json":"https://pith.science/api/pith-number/JEXNKHTGR7PQQODPPG4TYOSM4D/graph.json","events_json":"https://pith.science/api/pith-number/JEXNKHTGR7PQQODPPG4TYOSM4D/events.json","paper":"https://pith.science/paper/JEXNKHTG"},"agent_actions":{"view_html":"https://pith.science/pith/JEXNKHTGR7PQQODPPG4TYOSM4D","download_json":"https://pith.science/pith/JEXNKHTGR7PQQODPPG4TYOSM4D.json","view_paper":"https://pith.science/paper/JEXNKHTG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.05686&json=true","fetch_graph":"https://pith.science/api/pith-number/JEXNKHTGR7PQQODPPG4TYOSM4D/graph.json","fetch_events":"https://pith.science/api/pith-number/JEXNKHTGR7PQQODPPG4TYOSM4D/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JEXNKHTGR7PQQODPPG4TYOSM4D/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JEXNKHTGR7PQQODPPG4TYOSM4D/action/storage_attestation","attest_author":"https://pith.science/pith/JEXNKHTGR7PQQODPPG4TYOSM4D/action/author_attestation","sign_citation":"https://pith.science/pith/JEXNKHTGR7PQQODPPG4TYOSM4D/action/citation_signature","submit_replication":"https://pith.science/pith/JEXNKHTGR7PQQODPPG4TYOSM4D/action/replication_record"}},"created_at":"2026-07-05T11:33:34.804923+00:00","updated_at":"2026-07-05T11:33:34.804923+00:00"}