{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:GI22ES2ARSSLU2GVMSVGAQ25T5","short_pith_number":"pith:GI22ES2A","schema_version":"1.0","canonical_sha256":"3235a24b408ca4ba68d564aa60435d9f5cd227629d2c75e405f4b17f499759ba","source":{"kind":"arxiv","id":"2509.02830","version":1},"attestation_state":"computed","paper":{"title":"SSVD: Structured SVD for Parameter-Efficient Fine-Tuning and Benchmarking under Domain Shift in ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.CL","authors_text":"Hugo Van hamme, Pu Wang, Shinji Watanabe","submitted_at":"2025-09-02T20:51:17Z","abstract_excerpt":"Parameter-efficient fine-tuning (PEFT) has emerged as a scalable solution for adapting large foundation models. While low-rank adaptation (LoRA) is widely used in speech applications, its state-of-the-art variants, e.g., VeRA, DoRA, PiSSA, and SVFT, are developed mainly for language and vision tasks, with limited validation in speech. This work presents the first comprehensive integration and benchmarking of these PEFT methods within ESPnet. We further introduce structured SVD-guided (SSVD) fine-tuning, which selectively rotates input-associated right singular vectors while keeping output-asso"},"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":"2509.02830","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-09-02T20:51:17Z","cross_cats_sorted":["eess.AS"],"title_canon_sha256":"a22e60fdf4dbac816f7349dbee79a5e698e6302308715a687496b75bb1ab0d2e","abstract_canon_sha256":"78ca0fe6fcf0aa194bb639df2fbb5a28559840e868475e3a4feab7ce45e6a153"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:03:58.260521Z","signature_b64":"nfetNuIQfb4kCZa8wN2bOEN+7HjCm0C5GsC58zuAafqUftO1+awRJmFD9Fr+S5/GUHYAg/tzjqbwXiI9Y3VlDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3235a24b408ca4ba68d564aa60435d9f5cd227629d2c75e405f4b17f499759ba","last_reissued_at":"2026-07-05T12:03:58.260084Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:03:58.260084Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SSVD: Structured SVD for Parameter-Efficient Fine-Tuning and Benchmarking under Domain Shift in ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.CL","authors_text":"Hugo Van hamme, Pu Wang, Shinji Watanabe","submitted_at":"2025-09-02T20:51:17Z","abstract_excerpt":"Parameter-efficient fine-tuning (PEFT) has emerged as a scalable solution for adapting large foundation models. While low-rank adaptation (LoRA) is widely used in speech applications, its state-of-the-art variants, e.g., VeRA, DoRA, PiSSA, and SVFT, are developed mainly for language and vision tasks, with limited validation in speech. This work presents the first comprehensive integration and benchmarking of these PEFT methods within ESPnet. We further introduce structured SVD-guided (SSVD) fine-tuning, which selectively rotates input-associated right singular vectors while keeping output-asso"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.02830","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/2509.02830/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":"2509.02830","created_at":"2026-07-05T12:03:58.260136+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.02830v1","created_at":"2026-07-05T12:03:58.260136+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.02830","created_at":"2026-07-05T12:03:58.260136+00:00"},{"alias_kind":"pith_short_12","alias_value":"GI22ES2ARSSL","created_at":"2026-07-05T12:03:58.260136+00:00"},{"alias_kind":"pith_short_16","alias_value":"GI22ES2ARSSLU2GV","created_at":"2026-07-05T12:03:58.260136+00:00"},{"alias_kind":"pith_short_8","alias_value":"GI22ES2A","created_at":"2026-07-05T12:03:58.260136+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.10464","citing_title":"GC-LoRA: Gated Convolutional LoRA for Parameter-Efficient Acoustic Adaptation","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09342","citing_title":"Parameter-Efficient Continual Learning for Automatic Speech Recognition","ref_index":43,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GI22ES2ARSSLU2GVMSVGAQ25T5","json":"https://pith.science/pith/GI22ES2ARSSLU2GVMSVGAQ25T5.json","graph_json":"https://pith.science/api/pith-number/GI22ES2ARSSLU2GVMSVGAQ25T5/graph.json","events_json":"https://pith.science/api/pith-number/GI22ES2ARSSLU2GVMSVGAQ25T5/events.json","paper":"https://pith.science/paper/GI22ES2A"},"agent_actions":{"view_html":"https://pith.science/pith/GI22ES2ARSSLU2GVMSVGAQ25T5","download_json":"https://pith.science/pith/GI22ES2ARSSLU2GVMSVGAQ25T5.json","view_paper":"https://pith.science/paper/GI22ES2A","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.02830&json=true","fetch_graph":"https://pith.science/api/pith-number/GI22ES2ARSSLU2GVMSVGAQ25T5/graph.json","fetch_events":"https://pith.science/api/pith-number/GI22ES2ARSSLU2GVMSVGAQ25T5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GI22ES2ARSSLU2GVMSVGAQ25T5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GI22ES2ARSSLU2GVMSVGAQ25T5/action/storage_attestation","attest_author":"https://pith.science/pith/GI22ES2ARSSLU2GVMSVGAQ25T5/action/author_attestation","sign_citation":"https://pith.science/pith/GI22ES2ARSSLU2GVMSVGAQ25T5/action/citation_signature","submit_replication":"https://pith.science/pith/GI22ES2ARSSLU2GVMSVGAQ25T5/action/replication_record"}},"created_at":"2026-07-05T12:03:58.260136+00:00","updated_at":"2026-07-05T12:03:58.260136+00:00"}