{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:HUHRIBIARGXYXY3AXJ4GULRIFX","short_pith_number":"pith:HUHRIBIA","schema_version":"1.0","canonical_sha256":"3d0f14050089af8be360ba786a2e282df85709b8b32c0068fa2ef7c7d2d7d674","source":{"kind":"arxiv","id":"2405.17755","version":1},"attestation_state":"computed","paper":{"title":"XL3M: A Training-free Framework for LLM Length Extension Based on Segment-wise Inference","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Gong Zhang, Hongwei Sun, Hua Xu, Lin Zhang, Pingyi Zhou, Renhai Chen, Sen Wang, Shengnan Wang, Shixiong Zhao, Youhui Bai","submitted_at":"2024-05-28T02:12:35Z","abstract_excerpt":"Length generalization failure problem, namely the large language model (LLM) fails to generalize to texts longer than its maximum training length, greatly restricts the application of LLM in the scenarios with streaming long inputs. To address this problem, the existing methods either require substantial costs or introduce precision loss. In this paper, we empirically find that the accuracy of the LLM's prediction is highly correlated to its certainty. Based on this, we propose an efficient training free framework, named XL3M (it means extra-long large language model), which enables the LLMs t"},"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":"2405.17755","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-05-28T02:12:35Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2278818ca2c2a354010f65d2be8b94a16a4648c90908a86fd1926fe6f14f0bc1","abstract_canon_sha256":"326835b50d43d204bbfcffca7061b1e18c7b42d0ca8cc96839ab9b3806ad02aa"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:24:10.814952Z","signature_b64":"Q/T+DCF1eK8+BYG2lO0p8hEMLWJX78XyO6guKeNxKQNv+D481cj7TInamdYLak1DXrclmK5/Fo1qmFi6VOf1Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3d0f14050089af8be360ba786a2e282df85709b8b32c0068fa2ef7c7d2d7d674","last_reissued_at":"2026-07-05T08:24:10.814493Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:24:10.814493Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"XL3M: A Training-free Framework for LLM Length Extension Based on Segment-wise Inference","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Gong Zhang, Hongwei Sun, Hua Xu, Lin Zhang, Pingyi Zhou, Renhai Chen, Sen Wang, Shengnan Wang, Shixiong Zhao, Youhui Bai","submitted_at":"2024-05-28T02:12:35Z","abstract_excerpt":"Length generalization failure problem, namely the large language model (LLM) fails to generalize to texts longer than its maximum training length, greatly restricts the application of LLM in the scenarios with streaming long inputs. To address this problem, the existing methods either require substantial costs or introduce precision loss. In this paper, we empirically find that the accuracy of the LLM's prediction is highly correlated to its certainty. Based on this, we propose an efficient training free framework, named XL3M (it means extra-long large language model), which enables the LLMs t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.17755","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/2405.17755/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":"2405.17755","created_at":"2026-07-05T08:24:10.814567+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.17755v1","created_at":"2026-07-05T08:24:10.814567+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.17755","created_at":"2026-07-05T08:24:10.814567+00:00"},{"alias_kind":"pith_short_12","alias_value":"HUHRIBIARGXY","created_at":"2026-07-05T08:24:10.814567+00:00"},{"alias_kind":"pith_short_16","alias_value":"HUHRIBIARGXYXY3A","created_at":"2026-07-05T08:24:10.814567+00:00"},{"alias_kind":"pith_short_8","alias_value":"HUHRIBIA","created_at":"2026-07-05T08:24:10.814567+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2501.04987","citing_title":"TreeKV: Smooth Key-Value Cache Compression with Tree Structures","ref_index":13,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HUHRIBIARGXYXY3AXJ4GULRIFX","json":"https://pith.science/pith/HUHRIBIARGXYXY3AXJ4GULRIFX.json","graph_json":"https://pith.science/api/pith-number/HUHRIBIARGXYXY3AXJ4GULRIFX/graph.json","events_json":"https://pith.science/api/pith-number/HUHRIBIARGXYXY3AXJ4GULRIFX/events.json","paper":"https://pith.science/paper/HUHRIBIA"},"agent_actions":{"view_html":"https://pith.science/pith/HUHRIBIARGXYXY3AXJ4GULRIFX","download_json":"https://pith.science/pith/HUHRIBIARGXYXY3AXJ4GULRIFX.json","view_paper":"https://pith.science/paper/HUHRIBIA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.17755&json=true","fetch_graph":"https://pith.science/api/pith-number/HUHRIBIARGXYXY3AXJ4GULRIFX/graph.json","fetch_events":"https://pith.science/api/pith-number/HUHRIBIARGXYXY3AXJ4GULRIFX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HUHRIBIARGXYXY3AXJ4GULRIFX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HUHRIBIARGXYXY3AXJ4GULRIFX/action/storage_attestation","attest_author":"https://pith.science/pith/HUHRIBIARGXYXY3AXJ4GULRIFX/action/author_attestation","sign_citation":"https://pith.science/pith/HUHRIBIARGXYXY3AXJ4GULRIFX/action/citation_signature","submit_replication":"https://pith.science/pith/HUHRIBIARGXYXY3AXJ4GULRIFX/action/replication_record"}},"created_at":"2026-07-05T08:24:10.814567+00:00","updated_at":"2026-07-05T08:24:10.814567+00:00"}