{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7GSOPSVQGM6ESP4ZCJ2CLI5O37","short_pith_number":"pith:7GSOPSVQ","schema_version":"1.0","canonical_sha256":"f9a4e7cab0333c493f99127425a3aedfd2992f283e7c03fb6c7f981ea7d15aec","source":{"kind":"arxiv","id":"2505.23911","version":1},"attestation_state":"computed","paper":{"title":"One Task Vector is not Enough: A Large-Scale Study for In-Context Learning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Elena Tutubalina, Ivan Oseledets, Pavel Tikhonov","submitted_at":"2025-05-29T18:05:12Z","abstract_excerpt":"In-context learning (ICL) enables Large Language Models (LLMs) to adapt to new tasks using few examples, with task vectors - specific hidden state activations - hypothesized to encode task information. Existing studies are limited by small-scale benchmarks, restricting comprehensive analysis. We introduce QuiteAFew, a novel dataset of 3,096 diverse few-shot tasks, each with 30 input-output pairs derived from the Alpaca dataset. Experiments with Llama-3-8B on QuiteAFew reveal: (1) task vector performance peaks at an intermediate layer (e.g., 15th), (2) effectiveness varies significantly by task"},"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":"2505.23911","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-29T18:05:12Z","cross_cats_sorted":[],"title_canon_sha256":"45c59d33e233cd226d40341524f873a995dca230dfd0a929c3561489428c7e13","abstract_canon_sha256":"ff34defbb1df52f53d6f00222dc904f07a88a99e933d59be1749a5c93c41c3d1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:12:37.602731Z","signature_b64":"kMcBjqtd/GL+L+LR4bdCBLvxQoo+dDHIv9/wJBQ6iUKSisUab2wXzP+lYNbR2ERcI40oC+EN0kbSN2X+JQ1kCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f9a4e7cab0333c493f99127425a3aedfd2992f283e7c03fb6c7f981ea7d15aec","last_reissued_at":"2026-07-05T11:12:37.602219Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:12:37.602219Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"One Task Vector is not Enough: A Large-Scale Study for In-Context Learning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Elena Tutubalina, Ivan Oseledets, Pavel Tikhonov","submitted_at":"2025-05-29T18:05:12Z","abstract_excerpt":"In-context learning (ICL) enables Large Language Models (LLMs) to adapt to new tasks using few examples, with task vectors - specific hidden state activations - hypothesized to encode task information. Existing studies are limited by small-scale benchmarks, restricting comprehensive analysis. We introduce QuiteAFew, a novel dataset of 3,096 diverse few-shot tasks, each with 30 input-output pairs derived from the Alpaca dataset. Experiments with Llama-3-8B on QuiteAFew reveal: (1) task vector performance peaks at an intermediate layer (e.g., 15th), (2) effectiveness varies significantly by task"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.23911","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/2505.23911/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":"2505.23911","created_at":"2026-07-05T11:12:37.602289+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.23911v1","created_at":"2026-07-05T11:12:37.602289+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.23911","created_at":"2026-07-05T11:12:37.602289+00:00"},{"alias_kind":"pith_short_12","alias_value":"7GSOPSVQGM6E","created_at":"2026-07-05T11:12:37.602289+00:00"},{"alias_kind":"pith_short_16","alias_value":"7GSOPSVQGM6ESP4Z","created_at":"2026-07-05T11:12:37.602289+00:00"},{"alias_kind":"pith_short_8","alias_value":"7GSOPSVQ","created_at":"2026-07-05T11:12:37.602289+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/7GSOPSVQGM6ESP4ZCJ2CLI5O37","json":"https://pith.science/pith/7GSOPSVQGM6ESP4ZCJ2CLI5O37.json","graph_json":"https://pith.science/api/pith-number/7GSOPSVQGM6ESP4ZCJ2CLI5O37/graph.json","events_json":"https://pith.science/api/pith-number/7GSOPSVQGM6ESP4ZCJ2CLI5O37/events.json","paper":"https://pith.science/paper/7GSOPSVQ"},"agent_actions":{"view_html":"https://pith.science/pith/7GSOPSVQGM6ESP4ZCJ2CLI5O37","download_json":"https://pith.science/pith/7GSOPSVQGM6ESP4ZCJ2CLI5O37.json","view_paper":"https://pith.science/paper/7GSOPSVQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.23911&json=true","fetch_graph":"https://pith.science/api/pith-number/7GSOPSVQGM6ESP4ZCJ2CLI5O37/graph.json","fetch_events":"https://pith.science/api/pith-number/7GSOPSVQGM6ESP4ZCJ2CLI5O37/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7GSOPSVQGM6ESP4ZCJ2CLI5O37/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7GSOPSVQGM6ESP4ZCJ2CLI5O37/action/storage_attestation","attest_author":"https://pith.science/pith/7GSOPSVQGM6ESP4ZCJ2CLI5O37/action/author_attestation","sign_citation":"https://pith.science/pith/7GSOPSVQGM6ESP4ZCJ2CLI5O37/action/citation_signature","submit_replication":"https://pith.science/pith/7GSOPSVQGM6ESP4ZCJ2CLI5O37/action/replication_record"}},"created_at":"2026-07-05T11:12:37.602289+00:00","updated_at":"2026-07-05T11:12:37.602289+00:00"}