{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:W2V5LJRUHIWMTP2UAFHHC3FKMZ","short_pith_number":"pith:W2V5LJRU","schema_version":"1.0","canonical_sha256":"b6abd5a6343a2cc9bf54014e716caa667d871a53296b0192fd41f9d4d3cfe814","source":{"kind":"arxiv","id":"2607.02757","version":1},"attestation_state":"computed","paper":{"title":"Reinforcement Learning for Data-Efficient Code-Switched ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"cs.CL","authors_text":"Peter Vickers, Ziwei Ye","submitted_at":"2026-07-02T20:44:53Z","abstract_excerpt":"Audio-language models can be prompted for code-switched speech, but their decoding is not optimized for code-switching and often fails at language boundaries. We propose a practical reinforcement learning with verifiable rewards recipe for data-efficient adaptation of audio-language models to code-switched ASR using group relative policy optimization, combining an error rate reward with a script fidelity reward that penalizes wrong writing systems and a two-pass draft-and-refinement procedure. Using Qwen2-Audio as a reproducible testbed across 10 language pairs, training on only TTS code-switc"},"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":"2607.02757","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-07-02T20:44:53Z","cross_cats_sorted":["cs.SD"],"title_canon_sha256":"a5cb1482c85c6e18697901a1f166a5c2f7528538357297eed45854e4f2234351","abstract_canon_sha256":"d7cd90a1362ea49249ed4799adb252e959dd6fc7bea1a49880dcf68435d848df"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T00:16:12.274340Z","signature_b64":"al0eREdGGRwrhMMnBhOLLjvlvXRAANRtZYSn9NYzk4ZPZNZknqp/hoBXXZl7ANn0v0uInMjK+zu1WJR3Zo7gBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b6abd5a6343a2cc9bf54014e716caa667d871a53296b0192fd41f9d4d3cfe814","last_reissued_at":"2026-07-07T00:16:12.272813Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T00:16:12.272813Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reinforcement Learning for Data-Efficient Code-Switched ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"cs.CL","authors_text":"Peter Vickers, Ziwei Ye","submitted_at":"2026-07-02T20:44:53Z","abstract_excerpt":"Audio-language models can be prompted for code-switched speech, but their decoding is not optimized for code-switching and often fails at language boundaries. We propose a practical reinforcement learning with verifiable rewards recipe for data-efficient adaptation of audio-language models to code-switched ASR using group relative policy optimization, combining an error rate reward with a script fidelity reward that penalizes wrong writing systems and a two-pass draft-and-refinement procedure. Using Qwen2-Audio as a reproducible testbed across 10 language pairs, training on only TTS code-switc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.02757","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/2607.02757/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":"2607.02757","created_at":"2026-07-07T00:16:12.273507+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.02757v1","created_at":"2026-07-07T00:16:12.273507+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.02757","created_at":"2026-07-07T00:16:12.273507+00:00"},{"alias_kind":"pith_short_12","alias_value":"W2V5LJRUHIWM","created_at":"2026-07-07T00:16:12.273507+00:00"},{"alias_kind":"pith_short_16","alias_value":"W2V5LJRUHIWMTP2U","created_at":"2026-07-07T00:16:12.273507+00:00"},{"alias_kind":"pith_short_8","alias_value":"W2V5LJRU","created_at":"2026-07-07T00:16:12.273507+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/W2V5LJRUHIWMTP2UAFHHC3FKMZ","json":"https://pith.science/pith/W2V5LJRUHIWMTP2UAFHHC3FKMZ.json","graph_json":"https://pith.science/api/pith-number/W2V5LJRUHIWMTP2UAFHHC3FKMZ/graph.json","events_json":"https://pith.science/api/pith-number/W2V5LJRUHIWMTP2UAFHHC3FKMZ/events.json","paper":"https://pith.science/paper/W2V5LJRU"},"agent_actions":{"view_html":"https://pith.science/pith/W2V5LJRUHIWMTP2UAFHHC3FKMZ","download_json":"https://pith.science/pith/W2V5LJRUHIWMTP2UAFHHC3FKMZ.json","view_paper":"https://pith.science/paper/W2V5LJRU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.02757&json=true","fetch_graph":"https://pith.science/api/pith-number/W2V5LJRUHIWMTP2UAFHHC3FKMZ/graph.json","fetch_events":"https://pith.science/api/pith-number/W2V5LJRUHIWMTP2UAFHHC3FKMZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W2V5LJRUHIWMTP2UAFHHC3FKMZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W2V5LJRUHIWMTP2UAFHHC3FKMZ/action/storage_attestation","attest_author":"https://pith.science/pith/W2V5LJRUHIWMTP2UAFHHC3FKMZ/action/author_attestation","sign_citation":"https://pith.science/pith/W2V5LJRUHIWMTP2UAFHHC3FKMZ/action/citation_signature","submit_replication":"https://pith.science/pith/W2V5LJRUHIWMTP2UAFHHC3FKMZ/action/replication_record"}},"created_at":"2026-07-07T00:16:12.273507+00:00","updated_at":"2026-07-07T00:16:12.273507+00:00"}