{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:5ZQE73KCUYSEBXWUXDK4PIXUR5","short_pith_number":"pith:5ZQE73KC","schema_version":"1.0","canonical_sha256":"ee604fed42a62440ded4b8d5c7a2f48f7fe82d33f78d05621b65ea520adbd331","source":{"kind":"arxiv","id":"2606.17967","version":1},"attestation_state":"computed","paper":{"title":"Learning task-specific subspaces via interventional post-training of speech foundation models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jack Cox, Jon Barker","submitted_at":"2026-06-16T14:18:20Z","abstract_excerpt":"Speech foundation models, pre-trained on large corpora of unlabelled speech data, produce general-purpose representations which are useful across tasks. However, these representations encode information about salient speech variables in a distributed manner, while downstream speech tasks rely on only some of this variability. In this work, we propose a post-training refinement approach using interventional contrastive learning. By leveraging an interventional dataset and multi-part contrastive loss, we learn a transformation from the entangled representation space of speech foundation models i"},"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":"2606.17967","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-16T14:18:20Z","cross_cats_sorted":[],"title_canon_sha256":"39f5d9b77eb0921ce276bbbceb43ec439862caf04ab4982bf5c9c9226a007289","abstract_canon_sha256":"958fd9d0ddb3bd99ce282033bfe543b2382c2610b4598f68157076ce32af97f9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:10:44.757820Z","signature_b64":"FnABpr5xGEu3Hv0t3/Z68W+ABo0+W12y6zOc3NcVgdfSv6ji0XYo5WJZWVHPMQlm9oA/sv6O1SQ306OVAIcICg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ee604fed42a62440ded4b8d5c7a2f48f7fe82d33f78d05621b65ea520adbd331","last_reissued_at":"2026-06-19T16:10:44.757484Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:10:44.757484Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning task-specific subspaces via interventional post-training of speech foundation models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jack Cox, Jon Barker","submitted_at":"2026-06-16T14:18:20Z","abstract_excerpt":"Speech foundation models, pre-trained on large corpora of unlabelled speech data, produce general-purpose representations which are useful across tasks. However, these representations encode information about salient speech variables in a distributed manner, while downstream speech tasks rely on only some of this variability. In this work, we propose a post-training refinement approach using interventional contrastive learning. By leveraging an interventional dataset and multi-part contrastive loss, we learn a transformation from the entangled representation space of speech foundation models i"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.17967","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/2606.17967/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":"2606.17967","created_at":"2026-06-19T16:10:44.757544+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.17967v1","created_at":"2026-06-19T16:10:44.757544+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.17967","created_at":"2026-06-19T16:10:44.757544+00:00"},{"alias_kind":"pith_short_12","alias_value":"5ZQE73KCUYSE","created_at":"2026-06-19T16:10:44.757544+00:00"},{"alias_kind":"pith_short_16","alias_value":"5ZQE73KCUYSEBXWU","created_at":"2026-06-19T16:10:44.757544+00:00"},{"alias_kind":"pith_short_8","alias_value":"5ZQE73KC","created_at":"2026-06-19T16:10:44.757544+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2606.17967","citing_title":"Learning task-specific subspaces via interventional post-training of speech foundation models","ref_index":1,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5ZQE73KCUYSEBXWUXDK4PIXUR5","json":"https://pith.science/pith/5ZQE73KCUYSEBXWUXDK4PIXUR5.json","graph_json":"https://pith.science/api/pith-number/5ZQE73KCUYSEBXWUXDK4PIXUR5/graph.json","events_json":"https://pith.science/api/pith-number/5ZQE73KCUYSEBXWUXDK4PIXUR5/events.json","paper":"https://pith.science/paper/5ZQE73KC"},"agent_actions":{"view_html":"https://pith.science/pith/5ZQE73KCUYSEBXWUXDK4PIXUR5","download_json":"https://pith.science/pith/5ZQE73KCUYSEBXWUXDK4PIXUR5.json","view_paper":"https://pith.science/paper/5ZQE73KC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.17967&json=true","fetch_graph":"https://pith.science/api/pith-number/5ZQE73KCUYSEBXWUXDK4PIXUR5/graph.json","fetch_events":"https://pith.science/api/pith-number/5ZQE73KCUYSEBXWUXDK4PIXUR5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5ZQE73KCUYSEBXWUXDK4PIXUR5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5ZQE73KCUYSEBXWUXDK4PIXUR5/action/storage_attestation","attest_author":"https://pith.science/pith/5ZQE73KCUYSEBXWUXDK4PIXUR5/action/author_attestation","sign_citation":"https://pith.science/pith/5ZQE73KCUYSEBXWUXDK4PIXUR5/action/citation_signature","submit_replication":"https://pith.science/pith/5ZQE73KCUYSEBXWUXDK4PIXUR5/action/replication_record"}},"created_at":"2026-06-19T16:10:44.757544+00:00","updated_at":"2026-06-19T16:10:44.757544+00:00"}