{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:R36JBDSDNAGT2N4QBSQOJT6UQ4","short_pith_number":"pith:R36JBDSD","schema_version":"1.0","canonical_sha256":"8efc908e43680d3d37900ca0e4cfd4871f983e550b37e5feef0f86d5e6984035","source":{"kind":"arxiv","id":"2412.10079","version":1},"attestation_state":"computed","paper":{"title":"Lost in the Middle, and In-Between: Enhancing Language Models' Ability to Reason Over Long Contexts in Multi-Hop QA","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ankush Raut, George Arthur Baker, Katharina von der Wense, Lawrence E Hunter, Sagi Shaier","submitted_at":"2024-12-13T12:13:19Z","abstract_excerpt":"Previous work finds that recent long-context language models fail to make equal use of information in the middle of their inputs, preferring pieces of information located at the tail ends which creates an undue bias in situations where we would like models to be equally capable of using different parts of the input. Thus far, the problem has mainly only been considered in settings with single pieces of critical information, leading us to question what happens when multiple necessary pieces of information are spread out over the inputs. Here, we demonstrate the effects of the \"lost in the middl"},"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":"2412.10079","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-12-13T12:13:19Z","cross_cats_sorted":[],"title_canon_sha256":"22e3b5b111cd976f9f4864ca177bce453abee18ccd46c287e11bdf3c258fb892","abstract_canon_sha256":"6e978891ccaa8a896a39432ae5243f89b192f417970ebfacc3b5416d70b7874f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:48:46.765371Z","signature_b64":"YOmVy0p16sH0he9x2zCyw56W6LESQZ0aYjM2ESTiSTWRwKIIDVb+qnVO4dADWNCMxOPlvbnJodO1rT4ea9pGCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8efc908e43680d3d37900ca0e4cfd4871f983e550b37e5feef0f86d5e6984035","last_reissued_at":"2026-07-05T09:48:46.764946Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:48:46.764946Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Lost in the Middle, and In-Between: Enhancing Language Models' Ability to Reason Over Long Contexts in Multi-Hop QA","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ankush Raut, George Arthur Baker, Katharina von der Wense, Lawrence E Hunter, Sagi Shaier","submitted_at":"2024-12-13T12:13:19Z","abstract_excerpt":"Previous work finds that recent long-context language models fail to make equal use of information in the middle of their inputs, preferring pieces of information located at the tail ends which creates an undue bias in situations where we would like models to be equally capable of using different parts of the input. Thus far, the problem has mainly only been considered in settings with single pieces of critical information, leading us to question what happens when multiple necessary pieces of information are spread out over the inputs. Here, we demonstrate the effects of the \"lost in the middl"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.10079","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/2412.10079/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":"2412.10079","created_at":"2026-07-05T09:48:46.765007+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.10079v1","created_at":"2026-07-05T09:48:46.765007+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.10079","created_at":"2026-07-05T09:48:46.765007+00:00"},{"alias_kind":"pith_short_12","alias_value":"R36JBDSDNAGT","created_at":"2026-07-05T09:48:46.765007+00:00"},{"alias_kind":"pith_short_16","alias_value":"R36JBDSDNAGT2N4Q","created_at":"2026-07-05T09:48:46.765007+00:00"},{"alias_kind":"pith_short_8","alias_value":"R36JBDSD","created_at":"2026-07-05T09:48:46.765007+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09709","citing_title":"IS-CoT: Breaking the Long-form Generation Collapse via Interleaved Structural Thinking","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11470","citing_title":"The Periodic Table of LLM Reasoning: A Structured Survey of Reasoning Paradigms, Methods, and Failure Modes","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2601.12499","citing_title":"Failure Modes in Multi-Hop QA: The Weakest Link Effect and the Recognition Bottleneck","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12370","citing_title":"Context Convergence Improves Answering Inferential Questions","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08766","citing_title":"UserGPT Technical Report","ref_index":55,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/R36JBDSDNAGT2N4QBSQOJT6UQ4","json":"https://pith.science/pith/R36JBDSDNAGT2N4QBSQOJT6UQ4.json","graph_json":"https://pith.science/api/pith-number/R36JBDSDNAGT2N4QBSQOJT6UQ4/graph.json","events_json":"https://pith.science/api/pith-number/R36JBDSDNAGT2N4QBSQOJT6UQ4/events.json","paper":"https://pith.science/paper/R36JBDSD"},"agent_actions":{"view_html":"https://pith.science/pith/R36JBDSDNAGT2N4QBSQOJT6UQ4","download_json":"https://pith.science/pith/R36JBDSDNAGT2N4QBSQOJT6UQ4.json","view_paper":"https://pith.science/paper/R36JBDSD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.10079&json=true","fetch_graph":"https://pith.science/api/pith-number/R36JBDSDNAGT2N4QBSQOJT6UQ4/graph.json","fetch_events":"https://pith.science/api/pith-number/R36JBDSDNAGT2N4QBSQOJT6UQ4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R36JBDSDNAGT2N4QBSQOJT6UQ4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R36JBDSDNAGT2N4QBSQOJT6UQ4/action/storage_attestation","attest_author":"https://pith.science/pith/R36JBDSDNAGT2N4QBSQOJT6UQ4/action/author_attestation","sign_citation":"https://pith.science/pith/R36JBDSDNAGT2N4QBSQOJT6UQ4/action/citation_signature","submit_replication":"https://pith.science/pith/R36JBDSDNAGT2N4QBSQOJT6UQ4/action/replication_record"}},"created_at":"2026-07-05T09:48:46.765007+00:00","updated_at":"2026-07-05T09:48:46.765007+00:00"}