{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:VDVD5C2LQ3SNJXJNFPKGTRYOMM","short_pith_number":"pith:VDVD5C2L","schema_version":"1.0","canonical_sha256":"a8ea3e8b4b86e4d4dd2d2bd469c70e6309ddf2fb84cb4fed02eac7ef940fb480","source":{"kind":"arxiv","id":"2406.13282","version":3},"attestation_state":"computed","paper":{"title":"Understanding the RoPE Extensions of Long-Context LLMs: An Attention Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chen Zhang, Kehai Chen, Meizhi Zhong, Min Zhang, Xikai Liu, Yan Gao, Yao Hu, Yikun Lei","submitted_at":"2024-06-19T07:23:33Z","abstract_excerpt":"Enabling LLMs to handle lengthy context is currently a research hotspot. Most LLMs are built upon rotary position embedding (RoPE), a popular position encoding method. Therefore, a prominent path is to extrapolate the RoPE trained on comparably short texts to far longer texts. A heavy bunch of efforts have been dedicated to boosting the extrapolation via extending the formulations of the RoPE, however, few of them have attempted to showcase their inner workings comprehensively. In this paper, we are driven to offer a straightforward yet in-depth understanding of RoPE extensions from an attenti"},"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":"2406.13282","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-19T07:23:33Z","cross_cats_sorted":[],"title_canon_sha256":"99ba0a1243e52a1539f12ce54b7e13497589db52128d71d600194c4d2944fce7","abstract_canon_sha256":"cca511e66d4b92db2ac3d3a76c3a134b05ac2e8f7c86728b0f98af1b55a7d424"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:48:14.110252Z","signature_b64":"HonD1TTHmru/853+DpuzpxyRdcfd8y2/vOkGmsNENCCQYAOhil6rQTYfJ1awnZ+8VFsXcx7n7rCY+KIdzK16Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a8ea3e8b4b86e4d4dd2d2bd469c70e6309ddf2fb84cb4fed02eac7ef940fb480","last_reissued_at":"2026-07-05T09:48:14.109806Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:48:14.109806Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding the RoPE Extensions of Long-Context LLMs: An Attention Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chen Zhang, Kehai Chen, Meizhi Zhong, Min Zhang, Xikai Liu, Yan Gao, Yao Hu, Yikun Lei","submitted_at":"2024-06-19T07:23:33Z","abstract_excerpt":"Enabling LLMs to handle lengthy context is currently a research hotspot. Most LLMs are built upon rotary position embedding (RoPE), a popular position encoding method. Therefore, a prominent path is to extrapolate the RoPE trained on comparably short texts to far longer texts. A heavy bunch of efforts have been dedicated to boosting the extrapolation via extending the formulations of the RoPE, however, few of them have attempted to showcase their inner workings comprehensively. In this paper, we are driven to offer a straightforward yet in-depth understanding of RoPE extensions from an attenti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.13282","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/2406.13282/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":"2406.13282","created_at":"2026-07-05T09:48:14.109863+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.13282v3","created_at":"2026-07-05T09:48:14.109863+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.13282","created_at":"2026-07-05T09:48:14.109863+00:00"},{"alias_kind":"pith_short_12","alias_value":"VDVD5C2LQ3SN","created_at":"2026-07-05T09:48:14.109863+00:00"},{"alias_kind":"pith_short_16","alias_value":"VDVD5C2LQ3SNJXJN","created_at":"2026-07-05T09:48:14.109863+00:00"},{"alias_kind":"pith_short_8","alias_value":"VDVD5C2L","created_at":"2026-07-05T09:48:14.109863+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2412.09036","citing_title":"ZigZagkv: Dynamic KV Cache Compression for Long-context Modeling based on Layer Uncertainty","ref_index":32,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VDVD5C2LQ3SNJXJNFPKGTRYOMM","json":"https://pith.science/pith/VDVD5C2LQ3SNJXJNFPKGTRYOMM.json","graph_json":"https://pith.science/api/pith-number/VDVD5C2LQ3SNJXJNFPKGTRYOMM/graph.json","events_json":"https://pith.science/api/pith-number/VDVD5C2LQ3SNJXJNFPKGTRYOMM/events.json","paper":"https://pith.science/paper/VDVD5C2L"},"agent_actions":{"view_html":"https://pith.science/pith/VDVD5C2LQ3SNJXJNFPKGTRYOMM","download_json":"https://pith.science/pith/VDVD5C2LQ3SNJXJNFPKGTRYOMM.json","view_paper":"https://pith.science/paper/VDVD5C2L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.13282&json=true","fetch_graph":"https://pith.science/api/pith-number/VDVD5C2LQ3SNJXJNFPKGTRYOMM/graph.json","fetch_events":"https://pith.science/api/pith-number/VDVD5C2LQ3SNJXJNFPKGTRYOMM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VDVD5C2LQ3SNJXJNFPKGTRYOMM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VDVD5C2LQ3SNJXJNFPKGTRYOMM/action/storage_attestation","attest_author":"https://pith.science/pith/VDVD5C2LQ3SNJXJNFPKGTRYOMM/action/author_attestation","sign_citation":"https://pith.science/pith/VDVD5C2LQ3SNJXJNFPKGTRYOMM/action/citation_signature","submit_replication":"https://pith.science/pith/VDVD5C2LQ3SNJXJNFPKGTRYOMM/action/replication_record"}},"created_at":"2026-07-05T09:48:14.109863+00:00","updated_at":"2026-07-05T09:48:14.109863+00:00"}