{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:NBEJD5XJZU5OLOZQJRZN7YPKFU","short_pith_number":"pith:NBEJD5XJ","schema_version":"1.0","canonical_sha256":"684891f6e9cd3ae5bb304c72dfe1ea2d10efb34a3d759f08647511417f209b48","source":{"kind":"arxiv","id":"2404.16811","version":2},"attestation_state":"computed","paper":{"title":"Make Your LLM Fully Utilize the Context","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jian-Guang Lou, Nanning Zheng, Shengnan An, Zeqi Lin, Zexiong Ma","submitted_at":"2024-04-25T17:55:14Z","abstract_excerpt":"While many contemporary large language models (LLMs) can process lengthy input, they still struggle to fully utilize information within the long context, known as the lost-in-the-middle challenge. We hypothesize that it stems from insufficient explicit supervision during the long-context training, which fails to emphasize that any position in a long context can hold crucial information. Based on this intuition, our study presents information-intensive (IN2) training, a purely data-driven solution to overcome lost-in-the-middle. Specifically, IN2 training leverages a synthesized long-context qu"},"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":"2404.16811","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-25T17:55:14Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"cab341e21447575246b97efa3bfe007113d73b2cd2da518a29a038ad0c30fdc5","abstract_canon_sha256":"5f976cc59368867538fcda6e1568203b9e6765173b7799c5070c5d9d120fb4c3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:12:32.729220Z","signature_b64":"7SVUYVGfsHIuGoOJFBEOYS3ylr4RD2PufE6F3tr71BFq/9BJSkya5ULXWdo1zi4+KHN9Om7nkzs01qOi+clKBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"684891f6e9cd3ae5bb304c72dfe1ea2d10efb34a3d759f08647511417f209b48","last_reissued_at":"2026-07-05T08:12:32.728752Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:12:32.728752Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Make Your LLM Fully Utilize the Context","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jian-Guang Lou, Nanning Zheng, Shengnan An, Zeqi Lin, Zexiong Ma","submitted_at":"2024-04-25T17:55:14Z","abstract_excerpt":"While many contemporary large language models (LLMs) can process lengthy input, they still struggle to fully utilize information within the long context, known as the lost-in-the-middle challenge. We hypothesize that it stems from insufficient explicit supervision during the long-context training, which fails to emphasize that any position in a long context can hold crucial information. Based on this intuition, our study presents information-intensive (IN2) training, a purely data-driven solution to overcome lost-in-the-middle. Specifically, IN2 training leverages a synthesized long-context qu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.16811","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/2404.16811/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":"2404.16811","created_at":"2026-07-05T08:12:32.728807+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.16811v2","created_at":"2026-07-05T08:12:32.728807+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.16811","created_at":"2026-07-05T08:12:32.728807+00:00"},{"alias_kind":"pith_short_12","alias_value":"NBEJD5XJZU5O","created_at":"2026-07-05T08:12:32.728807+00:00"},{"alias_kind":"pith_short_16","alias_value":"NBEJD5XJZU5OLOZQ","created_at":"2026-07-05T08:12:32.728807+00:00"},{"alias_kind":"pith_short_8","alias_value":"NBEJD5XJ","created_at":"2026-07-05T08:12:32.728807+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12422","citing_title":"Creating and Evaluating K-12 GenAI Assessment Graders Through Context Engineering","ref_index":241,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27705","citing_title":"Mitigating Position Bias in Transformers via Layer-Specific Positional Embedding Scaling","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2509.11295","citing_title":"The Prompt Engineering Report Distilled: Quick Start Guide for Life Sciences","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2509.11295","citing_title":"The Prompt Engineering Report Distilled: Quick Start Guide for Life Sciences","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2511.12676","citing_title":"BridgeEQA: Virtual Embodied Agents for Real Bridge Inspections","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2410.10813","citing_title":"LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06456","citing_title":"Context-Aware Dialectal Arabic Machine Translation with Interactive Region and Register Selection","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NBEJD5XJZU5OLOZQJRZN7YPKFU","json":"https://pith.science/pith/NBEJD5XJZU5OLOZQJRZN7YPKFU.json","graph_json":"https://pith.science/api/pith-number/NBEJD5XJZU5OLOZQJRZN7YPKFU/graph.json","events_json":"https://pith.science/api/pith-number/NBEJD5XJZU5OLOZQJRZN7YPKFU/events.json","paper":"https://pith.science/paper/NBEJD5XJ"},"agent_actions":{"view_html":"https://pith.science/pith/NBEJD5XJZU5OLOZQJRZN7YPKFU","download_json":"https://pith.science/pith/NBEJD5XJZU5OLOZQJRZN7YPKFU.json","view_paper":"https://pith.science/paper/NBEJD5XJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.16811&json=true","fetch_graph":"https://pith.science/api/pith-number/NBEJD5XJZU5OLOZQJRZN7YPKFU/graph.json","fetch_events":"https://pith.science/api/pith-number/NBEJD5XJZU5OLOZQJRZN7YPKFU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NBEJD5XJZU5OLOZQJRZN7YPKFU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NBEJD5XJZU5OLOZQJRZN7YPKFU/action/storage_attestation","attest_author":"https://pith.science/pith/NBEJD5XJZU5OLOZQJRZN7YPKFU/action/author_attestation","sign_citation":"https://pith.science/pith/NBEJD5XJZU5OLOZQJRZN7YPKFU/action/citation_signature","submit_replication":"https://pith.science/pith/NBEJD5XJZU5OLOZQJRZN7YPKFU/action/replication_record"}},"created_at":"2026-07-05T08:12:32.728807+00:00","updated_at":"2026-07-05T08:12:32.728807+00:00"}