{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4SFF2AUYZVXNIBXSQJYP7WAUJG","short_pith_number":"pith:4SFF2AUY","schema_version":"1.0","canonical_sha256":"e48a5d0298cd6ed406f28270ffd814498dc42f300b9dfae27f7959b209a347eb","source":{"kind":"arxiv","id":"2509.22854","version":2},"attestation_state":"computed","paper":{"title":"Train Once, Reuse Everywhere: Generalizable Implicit In-Context Learning by Routing Attention","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jiaqian Li, Ligong Han, Ruixiang Tang, Wenya Wang, Yanshu Li","submitted_at":"2025-09-26T19:05:45Z","abstract_excerpt":"Implicit in-context learning (ICL) has newly emerged as a promising paradigm that simulates ICL behaviors in the representation space of large language models (LLMs), aiming to attain few-shot performance at zero-shot cost. However, existing approaches largely rely on injecting shift vectors into residual flows, which are typically constructed from labeled demonstrations or task-specific alignment. Such designs fall short of utilizing the structural mechanisms underlying ICL and suffer from limited generalizability. To address this, we propose In-Context Routing (ICR), a novel implicit ICL met"},"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":"2509.22854","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-09-26T19:05:45Z","cross_cats_sorted":[],"title_canon_sha256":"916a70bf0b200595fb649ab03b7676311b288fa5750e3f49ff9c5f9d8b85ffe3","abstract_canon_sha256":"3e14eb463dc6b103fef1b4ca9576325aa61cbaed1409745a4f16f00658cb7efb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-03T02:05:42.471433Z","signature_b64":"EFgtWp+BBPFq5Q4KNO0A+xm27yiBsw7ZRrD8EB1vFafyITW4X5Qhdg23LUdhreBwu62UomzneHmrkm2CzOZeAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e48a5d0298cd6ed406f28270ffd814498dc42f300b9dfae27f7959b209a347eb","last_reissued_at":"2026-06-03T02:05:42.470938Z","signature_status":"signed_v1","first_computed_at":"2026-06-03T02:05:42.470938Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Train Once, Reuse Everywhere: Generalizable Implicit In-Context Learning by Routing Attention","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jiaqian Li, Ligong Han, Ruixiang Tang, Wenya Wang, Yanshu Li","submitted_at":"2025-09-26T19:05:45Z","abstract_excerpt":"Implicit in-context learning (ICL) has newly emerged as a promising paradigm that simulates ICL behaviors in the representation space of large language models (LLMs), aiming to attain few-shot performance at zero-shot cost. However, existing approaches largely rely on injecting shift vectors into residual flows, which are typically constructed from labeled demonstrations or task-specific alignment. Such designs fall short of utilizing the structural mechanisms underlying ICL and suffer from limited generalizability. To address this, we propose In-Context Routing (ICR), a novel implicit ICL met"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.22854","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/2509.22854/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":"2509.22854","created_at":"2026-06-03T02:05:42.470996+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.22854v2","created_at":"2026-06-03T02:05:42.470996+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.22854","created_at":"2026-06-03T02:05:42.470996+00:00"},{"alias_kind":"pith_short_12","alias_value":"4SFF2AUYZVXN","created_at":"2026-06-03T02:05:42.470996+00:00"},{"alias_kind":"pith_short_16","alias_value":"4SFF2AUYZVXNIBXS","created_at":"2026-06-03T02:05:42.470996+00:00"},{"alias_kind":"pith_short_8","alias_value":"4SFF2AUY","created_at":"2026-06-03T02:05:42.470996+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.20730","citing_title":"Distributional Alignment as a Criterion for Designing Task Vectors in In-Context Learning","ref_index":26,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4SFF2AUYZVXNIBXSQJYP7WAUJG","json":"https://pith.science/pith/4SFF2AUYZVXNIBXSQJYP7WAUJG.json","graph_json":"https://pith.science/api/pith-number/4SFF2AUYZVXNIBXSQJYP7WAUJG/graph.json","events_json":"https://pith.science/api/pith-number/4SFF2AUYZVXNIBXSQJYP7WAUJG/events.json","paper":"https://pith.science/paper/4SFF2AUY"},"agent_actions":{"view_html":"https://pith.science/pith/4SFF2AUYZVXNIBXSQJYP7WAUJG","download_json":"https://pith.science/pith/4SFF2AUYZVXNIBXSQJYP7WAUJG.json","view_paper":"https://pith.science/paper/4SFF2AUY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.22854&json=true","fetch_graph":"https://pith.science/api/pith-number/4SFF2AUYZVXNIBXSQJYP7WAUJG/graph.json","fetch_events":"https://pith.science/api/pith-number/4SFF2AUYZVXNIBXSQJYP7WAUJG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4SFF2AUYZVXNIBXSQJYP7WAUJG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4SFF2AUYZVXNIBXSQJYP7WAUJG/action/storage_attestation","attest_author":"https://pith.science/pith/4SFF2AUYZVXNIBXSQJYP7WAUJG/action/author_attestation","sign_citation":"https://pith.science/pith/4SFF2AUYZVXNIBXSQJYP7WAUJG/action/citation_signature","submit_replication":"https://pith.science/pith/4SFF2AUYZVXNIBXSQJYP7WAUJG/action/replication_record"}},"created_at":"2026-06-03T02:05:42.470996+00:00","updated_at":"2026-06-03T02:05:42.470996+00:00"}