{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2YPZV2V62NIYNWPYM5LOEEZFD3","short_pith_number":"pith:2YPZV2V6","schema_version":"1.0","canonical_sha256":"d61f9aeabed35186d9f86756e213251efb27f718a8f727e305c388f01813a66d","source":{"kind":"arxiv","id":"2510.05363","version":2},"attestation_state":"computed","paper":{"title":"MHA-RAG: Improving Efficiency, Accuracy, and Consistency by Encoding Exemplars as Soft Prompts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Abhinav Jain, Christopher Jermaine, Thomas Reps, Xinyu Yao","submitted_at":"2025-10-06T20:41:43Z","abstract_excerpt":"Adapting Foundation Models to new domains with limited training data is challenging and computationally expensive. While prior work has demonstrated the effectiveness of using domain-specific exemplars as in-context demonstrations, we investigate whether representing exemplars purely as text is the most efficient, effective, and stable approach. We explore an alternative: representing exemplars as soft prompts with an exemplar order invariant model architecture. To this end, we introduce Multi-Head Attention Retrieval-Augmented Generation (MHA-RAG), a framework with the number of attention hea"},"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":"2510.05363","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-10-06T20:41:43Z","cross_cats_sorted":[],"title_canon_sha256":"5fb2bc594006045d9442ac06143ff16aa56f70dd530314639acf110e5b1e30df","abstract_canon_sha256":"fecd2cea3fbf51c55ee6fd3e756eeaaf57c4021406967d2148401a630744ebc7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-08T01:03:49.219479Z","signature_b64":"pIQaWMXTwFw7rlEimtaRbSc+BpFktgBXzVrFbV0gD+JokFxAb3CYdoQHkdTU87JS0t984zie9Z8VVqlNbbesDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d61f9aeabed35186d9f86756e213251efb27f718a8f727e305c388f01813a66d","last_reissued_at":"2026-06-08T01:03:49.218519Z","signature_status":"signed_v1","first_computed_at":"2026-06-08T01:03:49.218519Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MHA-RAG: Improving Efficiency, Accuracy, and Consistency by Encoding Exemplars as Soft Prompts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Abhinav Jain, Christopher Jermaine, Thomas Reps, Xinyu Yao","submitted_at":"2025-10-06T20:41:43Z","abstract_excerpt":"Adapting Foundation Models to new domains with limited training data is challenging and computationally expensive. While prior work has demonstrated the effectiveness of using domain-specific exemplars as in-context demonstrations, we investigate whether representing exemplars purely as text is the most efficient, effective, and stable approach. We explore an alternative: representing exemplars as soft prompts with an exemplar order invariant model architecture. To this end, we introduce Multi-Head Attention Retrieval-Augmented Generation (MHA-RAG), a framework with the number of attention hea"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.05363","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/2510.05363/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":"2510.05363","created_at":"2026-06-08T01:03:49.218654+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.05363v2","created_at":"2026-06-08T01:03:49.218654+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.05363","created_at":"2026-06-08T01:03:49.218654+00:00"},{"alias_kind":"pith_short_12","alias_value":"2YPZV2V62NIY","created_at":"2026-06-08T01:03:49.218654+00:00"},{"alias_kind":"pith_short_16","alias_value":"2YPZV2V62NIYNWPY","created_at":"2026-06-08T01:03:49.218654+00:00"},{"alias_kind":"pith_short_8","alias_value":"2YPZV2V6","created_at":"2026-06-08T01:03:49.218654+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/2YPZV2V62NIYNWPYM5LOEEZFD3","json":"https://pith.science/pith/2YPZV2V62NIYNWPYM5LOEEZFD3.json","graph_json":"https://pith.science/api/pith-number/2YPZV2V62NIYNWPYM5LOEEZFD3/graph.json","events_json":"https://pith.science/api/pith-number/2YPZV2V62NIYNWPYM5LOEEZFD3/events.json","paper":"https://pith.science/paper/2YPZV2V6"},"agent_actions":{"view_html":"https://pith.science/pith/2YPZV2V62NIYNWPYM5LOEEZFD3","download_json":"https://pith.science/pith/2YPZV2V62NIYNWPYM5LOEEZFD3.json","view_paper":"https://pith.science/paper/2YPZV2V6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.05363&json=true","fetch_graph":"https://pith.science/api/pith-number/2YPZV2V62NIYNWPYM5LOEEZFD3/graph.json","fetch_events":"https://pith.science/api/pith-number/2YPZV2V62NIYNWPYM5LOEEZFD3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2YPZV2V62NIYNWPYM5LOEEZFD3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2YPZV2V62NIYNWPYM5LOEEZFD3/action/storage_attestation","attest_author":"https://pith.science/pith/2YPZV2V62NIYNWPYM5LOEEZFD3/action/author_attestation","sign_citation":"https://pith.science/pith/2YPZV2V62NIYNWPYM5LOEEZFD3/action/citation_signature","submit_replication":"https://pith.science/pith/2YPZV2V62NIYNWPYM5LOEEZFD3/action/replication_record"}},"created_at":"2026-06-08T01:03:49.218654+00:00","updated_at":"2026-06-08T01:03:49.218654+00:00"}