{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:U5AUYIJBUCFIPJFPM3PSSTMWLJ","short_pith_number":"pith:U5AUYIJB","schema_version":"1.0","canonical_sha256":"a7414c2121a08a87a4af66df294d965a75368749743d4bbecd9175c1938c71d3","source":{"kind":"arxiv","id":"2505.10063","version":1},"attestation_state":"computed","paper":{"title":"CAFE: Retrieval Head-based Coarse-to-Fine Information Seeking to Enhance Multi-Document QA Capability","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Han Peng, Jinhao Jiang, Lei Fang, Wayne Xin Zhao, Zican Dong","submitted_at":"2025-05-15T08:05:12Z","abstract_excerpt":"Advancements in Large Language Models (LLMs) have extended their input context length, yet they still struggle with retrieval and reasoning in long-context inputs. Existing methods propose to utilize the prompt strategy and retrieval head to alleviate this limitation. However, they still face challenges in balancing retrieval precision and recall, impacting their efficacy in answering questions. To address this, we introduce $\\textbf{CAFE}$, a two-stage coarse-to-fine method to enhance multi-document question-answering capacities. By gradually eliminating the negative impacts of background and"},"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":"2505.10063","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-15T08:05:12Z","cross_cats_sorted":[],"title_canon_sha256":"be71ab6108cfe9f4e893854207cf1778b547903f23b431315e0f2a52579659f4","abstract_canon_sha256":"44df78332c7413e607d38782d73315e186fdbf673b8bf6a8df66550d5a8f38a2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:03:32.646205Z","signature_b64":"qz6aw59QfwzhMcfsQps6Ty8CMyd6qikb1vVk/T7U+uvsf0CwrVQjpt8QfRe4QtN9ArVtPtu5tgL90omEFO4PDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a7414c2121a08a87a4af66df294d965a75368749743d4bbecd9175c1938c71d3","last_reissued_at":"2026-07-05T11:03:32.645764Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:03:32.645764Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CAFE: Retrieval Head-based Coarse-to-Fine Information Seeking to Enhance Multi-Document QA Capability","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Han Peng, Jinhao Jiang, Lei Fang, Wayne Xin Zhao, Zican Dong","submitted_at":"2025-05-15T08:05:12Z","abstract_excerpt":"Advancements in Large Language Models (LLMs) have extended their input context length, yet they still struggle with retrieval and reasoning in long-context inputs. Existing methods propose to utilize the prompt strategy and retrieval head to alleviate this limitation. However, they still face challenges in balancing retrieval precision and recall, impacting their efficacy in answering questions. To address this, we introduce $\\textbf{CAFE}$, a two-stage coarse-to-fine method to enhance multi-document question-answering capacities. By gradually eliminating the negative impacts of background and"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.10063","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/2505.10063/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":"2505.10063","created_at":"2026-07-05T11:03:32.645823+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.10063v1","created_at":"2026-07-05T11:03:32.645823+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.10063","created_at":"2026-07-05T11:03:32.645823+00:00"},{"alias_kind":"pith_short_12","alias_value":"U5AUYIJBUCFI","created_at":"2026-07-05T11:03:32.645823+00:00"},{"alias_kind":"pith_short_16","alias_value":"U5AUYIJBUCFIPJFP","created_at":"2026-07-05T11:03:32.645823+00:00"},{"alias_kind":"pith_short_8","alias_value":"U5AUYIJB","created_at":"2026-07-05T11:03:32.645823+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.07365","citing_title":"LongReD: Mitigating Short-Text Degradation of Long-Context Large Language Models via Restoration Distillation","ref_index":54,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/U5AUYIJBUCFIPJFPM3PSSTMWLJ","json":"https://pith.science/pith/U5AUYIJBUCFIPJFPM3PSSTMWLJ.json","graph_json":"https://pith.science/api/pith-number/U5AUYIJBUCFIPJFPM3PSSTMWLJ/graph.json","events_json":"https://pith.science/api/pith-number/U5AUYIJBUCFIPJFPM3PSSTMWLJ/events.json","paper":"https://pith.science/paper/U5AUYIJB"},"agent_actions":{"view_html":"https://pith.science/pith/U5AUYIJBUCFIPJFPM3PSSTMWLJ","download_json":"https://pith.science/pith/U5AUYIJBUCFIPJFPM3PSSTMWLJ.json","view_paper":"https://pith.science/paper/U5AUYIJB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.10063&json=true","fetch_graph":"https://pith.science/api/pith-number/U5AUYIJBUCFIPJFPM3PSSTMWLJ/graph.json","fetch_events":"https://pith.science/api/pith-number/U5AUYIJBUCFIPJFPM3PSSTMWLJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/U5AUYIJBUCFIPJFPM3PSSTMWLJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/U5AUYIJBUCFIPJFPM3PSSTMWLJ/action/storage_attestation","attest_author":"https://pith.science/pith/U5AUYIJBUCFIPJFPM3PSSTMWLJ/action/author_attestation","sign_citation":"https://pith.science/pith/U5AUYIJBUCFIPJFPM3PSSTMWLJ/action/citation_signature","submit_replication":"https://pith.science/pith/U5AUYIJBUCFIPJFPM3PSSTMWLJ/action/replication_record"}},"created_at":"2026-07-05T11:03:32.645823+00:00","updated_at":"2026-07-05T11:03:32.645823+00:00"}