{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:VKG2MYUFVNFP23MBF4WUFGUL4U","short_pith_number":"pith:VKG2MYUF","schema_version":"1.0","canonical_sha256":"aa8da66285ab4afd6d812f2d429a8be52bc9f33ec7efeeeacb26c5e1331e36c6","source":{"kind":"arxiv","id":"2406.09827","version":3},"attestation_state":"computed","paper":{"title":"A Training-free Sub-quadratic Cost Transformer Model Serving Framework With Hierarchically Pruned Attention","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV","cs.DC","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bumsik Kim, Geon Park, Heejun Lee, Hyemin Lee, Jaduk Suh, Jina Kim, Myeongjae Jeon, Sung Ju Hwang, Wonyoung Jeong, Youngwan Lee","submitted_at":"2024-06-14T08:32:45Z","abstract_excerpt":"In modern large language models (LLMs), increasing the context length is crucial for improving comprehension and coherence in long-context, multi-modal, and retrieval-augmented language generation. While many recent transformer models attempt to extend their context length over a million tokens, they remain impractical due to the quadratic time and space complexities. Although recent works on linear and sparse attention mechanisms can achieve this goal, their real-world applicability is often limited by the need to re-train from scratch and significantly worse performance. In response, we prop"},"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":"2406.09827","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-14T08:32:45Z","cross_cats_sorted":["cs.CV","cs.DC","cs.LG"],"title_canon_sha256":"d81901b7bc19f77af2338820e41455a3741b93cffe7fb4da5fdb7e74ebdddfa7","abstract_canon_sha256":"d27e389deb210b5254a4b142084c2439399c7b601e537f09f5ae3c6daea40829"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:04:13.784299Z","signature_b64":"HBH7fjpB5GV2u9KOAKELdSL5HdwtEohsAWcq96n2O+O5fU2DCiaJSUa/eSQIDH5ZeaQ9+6HWuzTE+UBsDPFaBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aa8da66285ab4afd6d812f2d429a8be52bc9f33ec7efeeeacb26c5e1331e36c6","last_reissued_at":"2026-07-05T10:04:13.783854Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:04:13.783854Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Training-free Sub-quadratic Cost Transformer Model Serving Framework With Hierarchically Pruned Attention","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV","cs.DC","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bumsik Kim, Geon Park, Heejun Lee, Hyemin Lee, Jaduk Suh, Jina Kim, Myeongjae Jeon, Sung Ju Hwang, Wonyoung Jeong, Youngwan Lee","submitted_at":"2024-06-14T08:32:45Z","abstract_excerpt":"In modern large language models (LLMs), increasing the context length is crucial for improving comprehension and coherence in long-context, multi-modal, and retrieval-augmented language generation. While many recent transformer models attempt to extend their context length over a million tokens, they remain impractical due to the quadratic time and space complexities. Although recent works on linear and sparse attention mechanisms can achieve this goal, their real-world applicability is often limited by the need to re-train from scratch and significantly worse performance. In response, we prop"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.09827","kind":"arxiv","version":3},"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/2406.09827/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":"2406.09827","created_at":"2026-07-05T10:04:13.783916+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.09827v3","created_at":"2026-07-05T10:04:13.783916+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.09827","created_at":"2026-07-05T10:04:13.783916+00:00"},{"alias_kind":"pith_short_12","alias_value":"VKG2MYUFVNFP","created_at":"2026-07-05T10:04:13.783916+00:00"},{"alias_kind":"pith_short_16","alias_value":"VKG2MYUFVNFP23MB","created_at":"2026-07-05T10:04:13.783916+00:00"},{"alias_kind":"pith_short_8","alias_value":"VKG2MYUF","created_at":"2026-07-05T10:04:13.783916+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.08910","citing_title":"InfiniteHiP: Extending Language Model Context Up to 3 Million Tokens on a Single GPU","ref_index":18,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VKG2MYUFVNFP23MBF4WUFGUL4U","json":"https://pith.science/pith/VKG2MYUFVNFP23MBF4WUFGUL4U.json","graph_json":"https://pith.science/api/pith-number/VKG2MYUFVNFP23MBF4WUFGUL4U/graph.json","events_json":"https://pith.science/api/pith-number/VKG2MYUFVNFP23MBF4WUFGUL4U/events.json","paper":"https://pith.science/paper/VKG2MYUF"},"agent_actions":{"view_html":"https://pith.science/pith/VKG2MYUFVNFP23MBF4WUFGUL4U","download_json":"https://pith.science/pith/VKG2MYUFVNFP23MBF4WUFGUL4U.json","view_paper":"https://pith.science/paper/VKG2MYUF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.09827&json=true","fetch_graph":"https://pith.science/api/pith-number/VKG2MYUFVNFP23MBF4WUFGUL4U/graph.json","fetch_events":"https://pith.science/api/pith-number/VKG2MYUFVNFP23MBF4WUFGUL4U/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VKG2MYUFVNFP23MBF4WUFGUL4U/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VKG2MYUFVNFP23MBF4WUFGUL4U/action/storage_attestation","attest_author":"https://pith.science/pith/VKG2MYUFVNFP23MBF4WUFGUL4U/action/author_attestation","sign_citation":"https://pith.science/pith/VKG2MYUFVNFP23MBF4WUFGUL4U/action/citation_signature","submit_replication":"https://pith.science/pith/VKG2MYUFVNFP23MBF4WUFGUL4U/action/replication_record"}},"created_at":"2026-07-05T10:04:13.783916+00:00","updated_at":"2026-07-05T10:04:13.783916+00:00"}