{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:73CFWKDIPF7O5M4K7LTT4RDQJT","short_pith_number":"pith:73CFWKDI","schema_version":"1.0","canonical_sha256":"fec45b2868797eeeb38afae73e44704cdb5956442920e470cbebcdeef2ce1180","source":{"kind":"arxiv","id":"2607.00247","version":1},"attestation_state":"computed","paper":{"title":"Adaptive Perturbation Selection for Contrastive Audio Decoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SD","authors_text":"Aaron Isidore Grace, Weiran Wang, Zhouyuan Huo","submitted_at":"2026-06-30T22:55:22Z","abstract_excerpt":"Large audio-language models (LALMs) frequently hallucinate by overriding acoustic evidence with language priors. While contrastive decoding (CD) offers training-free mitigation, existing methods rely on blunt perturbations like masking or noise, leaving structured audio transformations unexplored. We explore this design space by evaluating a diverse library of targeted audio perturbations and adaptively selecting the optimal negative branch for each task and example. First, we improve upon earlier prompt engineering by showing that a simple binary yes/no constraint reduces the model's tendency"},"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":"2607.00247","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SD","submitted_at":"2026-06-30T22:55:22Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"affd8e8bf8695d2b5f57517aba4cec9ee27d99e3ca8ea0180d8504a88e0dd53c","abstract_canon_sha256":"6c3dd0c466a707f6b3808d22ba7b577f4b19e46264070c9af493660b47183967"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-02T00:18:40.970393Z","signature_b64":"ItFfEcAMFsPj1DRy5AY/P/7IsbOE7AXktOBkA1sfRp4t4Z6fvfbCAmccfU93jQojDKw7bfrlSC8D9xP7gBLiCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fec45b2868797eeeb38afae73e44704cdb5956442920e470cbebcdeef2ce1180","last_reissued_at":"2026-07-02T00:18:40.969750Z","signature_status":"signed_v1","first_computed_at":"2026-07-02T00:18:40.969750Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Adaptive Perturbation Selection for Contrastive Audio Decoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SD","authors_text":"Aaron Isidore Grace, Weiran Wang, Zhouyuan Huo","submitted_at":"2026-06-30T22:55:22Z","abstract_excerpt":"Large audio-language models (LALMs) frequently hallucinate by overriding acoustic evidence with language priors. While contrastive decoding (CD) offers training-free mitigation, existing methods rely on blunt perturbations like masking or noise, leaving structured audio transformations unexplored. We explore this design space by evaluating a diverse library of targeted audio perturbations and adaptively selecting the optimal negative branch for each task and example. First, we improve upon earlier prompt engineering by showing that a simple binary yes/no constraint reduces the model's tendency"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.00247","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/2607.00247/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":"2607.00247","created_at":"2026-07-02T00:18:40.969848+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.00247v1","created_at":"2026-07-02T00:18:40.969848+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.00247","created_at":"2026-07-02T00:18:40.969848+00:00"},{"alias_kind":"pith_short_12","alias_value":"73CFWKDIPF7O","created_at":"2026-07-02T00:18:40.969848+00:00"},{"alias_kind":"pith_short_16","alias_value":"73CFWKDIPF7O5M4K","created_at":"2026-07-02T00:18:40.969848+00:00"},{"alias_kind":"pith_short_8","alias_value":"73CFWKDI","created_at":"2026-07-02T00:18:40.969848+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/73CFWKDIPF7O5M4K7LTT4RDQJT","json":"https://pith.science/pith/73CFWKDIPF7O5M4K7LTT4RDQJT.json","graph_json":"https://pith.science/api/pith-number/73CFWKDIPF7O5M4K7LTT4RDQJT/graph.json","events_json":"https://pith.science/api/pith-number/73CFWKDIPF7O5M4K7LTT4RDQJT/events.json","paper":"https://pith.science/paper/73CFWKDI"},"agent_actions":{"view_html":"https://pith.science/pith/73CFWKDIPF7O5M4K7LTT4RDQJT","download_json":"https://pith.science/pith/73CFWKDIPF7O5M4K7LTT4RDQJT.json","view_paper":"https://pith.science/paper/73CFWKDI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.00247&json=true","fetch_graph":"https://pith.science/api/pith-number/73CFWKDIPF7O5M4K7LTT4RDQJT/graph.json","fetch_events":"https://pith.science/api/pith-number/73CFWKDIPF7O5M4K7LTT4RDQJT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/73CFWKDIPF7O5M4K7LTT4RDQJT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/73CFWKDIPF7O5M4K7LTT4RDQJT/action/storage_attestation","attest_author":"https://pith.science/pith/73CFWKDIPF7O5M4K7LTT4RDQJT/action/author_attestation","sign_citation":"https://pith.science/pith/73CFWKDIPF7O5M4K7LTT4RDQJT/action/citation_signature","submit_replication":"https://pith.science/pith/73CFWKDIPF7O5M4K7LTT4RDQJT/action/replication_record"}},"created_at":"2026-07-02T00:18:40.969848+00:00","updated_at":"2026-07-02T00:18:40.969848+00:00"}