{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:KFT2M7CEYM4NW7ZMER773ZSQFJ","short_pith_number":"pith:KFT2M7CE","schema_version":"1.0","canonical_sha256":"5167a67c44c338db7f2c247ffde6502a65a1e72d10e36d91543de03ad517c43b","source":{"kind":"arxiv","id":"2311.02248","version":2},"attestation_state":"computed","paper":{"title":"COSMIC: Data Efficient Instruction-tuning For Speech In-Context Learning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","eess.AS"],"primary_cat":"cs.CL","authors_text":"Jian Wu, Jing Pan, Jinyu Li, Shujie Liu, Sunit Sivasankaran, Yashesh Gaur, Zhuo Chen","submitted_at":"2023-11-03T21:47:03Z","abstract_excerpt":"We present a cost-effective method to integrate speech into a large language model (LLM), resulting in a Contextual Speech Model with Instruction-following/in-context-learning Capabilities (COSMIC) multi-modal LLM. Using GPT-3.5, we generate Speech Comprehension Test Question-Answer (SQA) pairs from speech transcriptions for supervised instruction tuning. With under 30 million trainable parameters and only 450 hours of English speech data, COSMIC demonstrates emerging capabilities in instruction-following and in-context learning. Equipped with such capabilities, COSMIC achieves a maximum 33.18"},"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":"2311.02248","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-03T21:47:03Z","cross_cats_sorted":["cs.AI","eess.AS"],"title_canon_sha256":"565dcdb8861283389dcc4b43332cdfbe536dc96d28959bd8c835812fb3862291","abstract_canon_sha256":"1c8f8b10e1f2982e83aadab83faa8de448a960f567e520cdfc0af66908114645"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:31:44.095996Z","signature_b64":"2ZdW2PSfXq0utkhq060cFqhMDXV6Wl++BBf6xnqmr8Dq7r+uYGO69HYo7O6SGIGqaFPBrntG1t94S6FZwFL9Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5167a67c44c338db7f2c247ffde6502a65a1e72d10e36d91543de03ad517c43b","last_reissued_at":"2026-07-05T08:31:44.095444Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:31:44.095444Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"COSMIC: Data Efficient Instruction-tuning For Speech In-Context Learning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","eess.AS"],"primary_cat":"cs.CL","authors_text":"Jian Wu, Jing Pan, Jinyu Li, Shujie Liu, Sunit Sivasankaran, Yashesh Gaur, Zhuo Chen","submitted_at":"2023-11-03T21:47:03Z","abstract_excerpt":"We present a cost-effective method to integrate speech into a large language model (LLM), resulting in a Contextual Speech Model with Instruction-following/in-context-learning Capabilities (COSMIC) multi-modal LLM. Using GPT-3.5, we generate Speech Comprehension Test Question-Answer (SQA) pairs from speech transcriptions for supervised instruction tuning. With under 30 million trainable parameters and only 450 hours of English speech data, COSMIC demonstrates emerging capabilities in instruction-following and in-context learning. Equipped with such capabilities, COSMIC achieves a maximum 33.18"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.02248","kind":"arxiv","version":2},"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/2311.02248/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":"2311.02248","created_at":"2026-07-05T08:31:44.095495+00:00"},{"alias_kind":"arxiv_version","alias_value":"2311.02248v2","created_at":"2026-07-05T08:31:44.095495+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.02248","created_at":"2026-07-05T08:31:44.095495+00:00"},{"alias_kind":"pith_short_12","alias_value":"KFT2M7CEYM4N","created_at":"2026-07-05T08:31:44.095495+00:00"},{"alias_kind":"pith_short_16","alias_value":"KFT2M7CEYM4NW7ZM","created_at":"2026-07-05T08:31:44.095495+00:00"},{"alias_kind":"pith_short_8","alias_value":"KFT2M7CE","created_at":"2026-07-05T08:31:44.095495+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.02615","citing_title":"FSA-GRPO: Teaching Auditory LLMs to Use Few-shot Demonstrations","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KFT2M7CEYM4NW7ZMER773ZSQFJ","json":"https://pith.science/pith/KFT2M7CEYM4NW7ZMER773ZSQFJ.json","graph_json":"https://pith.science/api/pith-number/KFT2M7CEYM4NW7ZMER773ZSQFJ/graph.json","events_json":"https://pith.science/api/pith-number/KFT2M7CEYM4NW7ZMER773ZSQFJ/events.json","paper":"https://pith.science/paper/KFT2M7CE"},"agent_actions":{"view_html":"https://pith.science/pith/KFT2M7CEYM4NW7ZMER773ZSQFJ","download_json":"https://pith.science/pith/KFT2M7CEYM4NW7ZMER773ZSQFJ.json","view_paper":"https://pith.science/paper/KFT2M7CE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2311.02248&json=true","fetch_graph":"https://pith.science/api/pith-number/KFT2M7CEYM4NW7ZMER773ZSQFJ/graph.json","fetch_events":"https://pith.science/api/pith-number/KFT2M7CEYM4NW7ZMER773ZSQFJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KFT2M7CEYM4NW7ZMER773ZSQFJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KFT2M7CEYM4NW7ZMER773ZSQFJ/action/storage_attestation","attest_author":"https://pith.science/pith/KFT2M7CEYM4NW7ZMER773ZSQFJ/action/author_attestation","sign_citation":"https://pith.science/pith/KFT2M7CEYM4NW7ZMER773ZSQFJ/action/citation_signature","submit_replication":"https://pith.science/pith/KFT2M7CEYM4NW7ZMER773ZSQFJ/action/replication_record"}},"created_at":"2026-07-05T08:31:44.095495+00:00","updated_at":"2026-07-05T08:31:44.095495+00:00"}