{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:WBXRJO4VUT5XR427UJAZ4J5CJK","short_pith_number":"pith:WBXRJO4V","schema_version":"1.0","canonical_sha256":"b06f14bb95a4fb78f35fa2419e27a24ab00dc34c1c6e687533bea80b6f526bfb","source":{"kind":"arxiv","id":"2406.14673","version":2},"attestation_state":"computed","paper":{"title":"Insights into LLM Long-Context Failures: When Transformers Know but Don't Tell","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Adam Byerly, Daniel Khashabi, Kuai Yu, Muhan Gao, Taiming Lu","submitted_at":"2024-06-20T18:50:44Z","abstract_excerpt":"Large Language Models (LLMs) exhibit positional bias, struggling to utilize information from the middle or end of long contexts. Our study explores LLMs' long-context reasoning by probing their hidden representations. We find that while LLMs encode the position of target information, they often fail to leverage this in generating accurate responses. This reveals a disconnect between information retrieval and utilization, a \"know but don't tell\" phenomenon. We further analyze the relationship between extraction time and final accuracy, offering insights into the underlying mechanics of transfor"},"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":"2406.14673","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-20T18:50:44Z","cross_cats_sorted":[],"title_canon_sha256":"51cc307a2c35723cdf2b23ff18229f901002fc87e122de3858b648e7efc3f3ce","abstract_canon_sha256":"ebe30d017de69902798c2cd463f559702a532ff53ad264d1c010aedc5c4576f0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:16:18.567017Z","signature_b64":"uTnXV/xPUyCX2cDCG5rLTw2gMDVzM2XBEymOMhTdX3dDOUKsg8Cui4bqChJFe/OxO4F0+ZPA45h6NdK7GCBBCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b06f14bb95a4fb78f35fa2419e27a24ab00dc34c1c6e687533bea80b6f526bfb","last_reissued_at":"2026-07-05T09:16:18.566585Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:16:18.566585Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Insights into LLM Long-Context Failures: When Transformers Know but Don't Tell","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Adam Byerly, Daniel Khashabi, Kuai Yu, Muhan Gao, Taiming Lu","submitted_at":"2024-06-20T18:50:44Z","abstract_excerpt":"Large Language Models (LLMs) exhibit positional bias, struggling to utilize information from the middle or end of long contexts. Our study explores LLMs' long-context reasoning by probing their hidden representations. We find that while LLMs encode the position of target information, they often fail to leverage this in generating accurate responses. This reveals a disconnect between information retrieval and utilization, a \"know but don't tell\" phenomenon. We further analyze the relationship between extraction time and final accuracy, offering insights into the underlying mechanics of transfor"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.14673","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/2406.14673/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":"2406.14673","created_at":"2026-07-05T09:16:18.566637+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.14673v2","created_at":"2026-07-05T09:16:18.566637+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.14673","created_at":"2026-07-05T09:16:18.566637+00:00"},{"alias_kind":"pith_short_12","alias_value":"WBXRJO4VUT5X","created_at":"2026-07-05T09:16:18.566637+00:00"},{"alias_kind":"pith_short_16","alias_value":"WBXRJO4VUT5XR427","created_at":"2026-07-05T09:16:18.566637+00:00"},{"alias_kind":"pith_short_8","alias_value":"WBXRJO4V","created_at":"2026-07-05T09:16:18.566637+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18158","citing_title":"The Measurement Gap in the Automation of EU Law: Benchmarking Doctrinal Legal Reasoning under the EU AI Act","ref_index":70,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WBXRJO4VUT5XR427UJAZ4J5CJK","json":"https://pith.science/pith/WBXRJO4VUT5XR427UJAZ4J5CJK.json","graph_json":"https://pith.science/api/pith-number/WBXRJO4VUT5XR427UJAZ4J5CJK/graph.json","events_json":"https://pith.science/api/pith-number/WBXRJO4VUT5XR427UJAZ4J5CJK/events.json","paper":"https://pith.science/paper/WBXRJO4V"},"agent_actions":{"view_html":"https://pith.science/pith/WBXRJO4VUT5XR427UJAZ4J5CJK","download_json":"https://pith.science/pith/WBXRJO4VUT5XR427UJAZ4J5CJK.json","view_paper":"https://pith.science/paper/WBXRJO4V","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.14673&json=true","fetch_graph":"https://pith.science/api/pith-number/WBXRJO4VUT5XR427UJAZ4J5CJK/graph.json","fetch_events":"https://pith.science/api/pith-number/WBXRJO4VUT5XR427UJAZ4J5CJK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WBXRJO4VUT5XR427UJAZ4J5CJK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WBXRJO4VUT5XR427UJAZ4J5CJK/action/storage_attestation","attest_author":"https://pith.science/pith/WBXRJO4VUT5XR427UJAZ4J5CJK/action/author_attestation","sign_citation":"https://pith.science/pith/WBXRJO4VUT5XR427UJAZ4J5CJK/action/citation_signature","submit_replication":"https://pith.science/pith/WBXRJO4VUT5XR427UJAZ4J5CJK/action/replication_record"}},"created_at":"2026-07-05T09:16:18.566637+00:00","updated_at":"2026-07-05T09:16:18.566637+00:00"}