{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:EFVKG5HSG7KWGQXBINYQBDSQLQ","short_pith_number":"pith:EFVKG5HS","schema_version":"1.0","canonical_sha256":"216aa374f237d56342e14371008e505c2550e10f7a04efecc977cc13f325e0c3","source":{"kind":"arxiv","id":"2309.10400","version":3},"attestation_state":"computed","paper":{"title":"PoSE: Efficient Context Window Extension of LLMs via Positional Skip-wise Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Dawei Zhu, Furu Wei, Liang Wang, Nan Yang, Sujian Li, Wenhao Wu, Yifan Song","submitted_at":"2023-09-19T08:03:38Z","abstract_excerpt":"Large Language Models (LLMs) are trained with a pre-defined context length, restricting their use in scenarios requiring long inputs. Previous efforts for adapting LLMs to a longer length usually requires fine-tuning with this target length (Full-length fine-tuning), suffering intensive training cost. To decouple train length from target length for efficient context window extension, we propose Positional Skip-wisE (PoSE) training that smartly simulates long inputs using a fixed context window. This is achieved by first dividing the original context window into several chunks, then designing d"},"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":"2309.10400","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-09-19T08:03:38Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ac4e7ee1d3af9ac0fcd29c98b0879a0f9563950f49fc3132e8ccd797059fba14","abstract_canon_sha256":"a066262c3ee537361f179f1d6aac06ec4747f1eae4f521713f49d0f56b027f33"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:47:44.981476Z","signature_b64":"4ooL80A3m4BaobCF1KpTbjHgkN4pKi6RmgGvjUdqq1cj2Kl7acGdg7pcs8SkegJ375dJOUzQvTCjYj5+qdYaAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"216aa374f237d56342e14371008e505c2550e10f7a04efecc977cc13f325e0c3","last_reissued_at":"2026-07-05T07:47:44.981000Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:47:44.981000Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PoSE: Efficient Context Window Extension of LLMs via Positional Skip-wise Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Dawei Zhu, Furu Wei, Liang Wang, Nan Yang, Sujian Li, Wenhao Wu, Yifan Song","submitted_at":"2023-09-19T08:03:38Z","abstract_excerpt":"Large Language Models (LLMs) are trained with a pre-defined context length, restricting their use in scenarios requiring long inputs. Previous efforts for adapting LLMs to a longer length usually requires fine-tuning with this target length (Full-length fine-tuning), suffering intensive training cost. To decouple train length from target length for efficient context window extension, we propose Positional Skip-wisE (PoSE) training that smartly simulates long inputs using a fixed context window. This is achieved by first dividing the original context window into several chunks, then designing d"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2309.10400","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/2309.10400/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":"2309.10400","created_at":"2026-07-05T07:47:44.981061+00:00"},{"alias_kind":"arxiv_version","alias_value":"2309.10400v3","created_at":"2026-07-05T07:47:44.981061+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2309.10400","created_at":"2026-07-05T07:47:44.981061+00:00"},{"alias_kind":"pith_short_12","alias_value":"EFVKG5HSG7KW","created_at":"2026-07-05T07:47:44.981061+00:00"},{"alias_kind":"pith_short_16","alias_value":"EFVKG5HSG7KWGQXB","created_at":"2026-07-05T07:47:44.981061+00:00"},{"alias_kind":"pith_short_8","alias_value":"EFVKG5HS","created_at":"2026-07-05T07:47:44.981061+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19706","citing_title":"NEST: Narrative Event Structures in Time for Long Video Understanding","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2512.07805","citing_title":"Group Representational Position Encoding","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13831","citing_title":"Training Long-Context Vision-Language Models Effectively with Generalization Beyond 128K Context","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2406.02069","citing_title":"PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funneling","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EFVKG5HSG7KWGQXBINYQBDSQLQ","json":"https://pith.science/pith/EFVKG5HSG7KWGQXBINYQBDSQLQ.json","graph_json":"https://pith.science/api/pith-number/EFVKG5HSG7KWGQXBINYQBDSQLQ/graph.json","events_json":"https://pith.science/api/pith-number/EFVKG5HSG7KWGQXBINYQBDSQLQ/events.json","paper":"https://pith.science/paper/EFVKG5HS"},"agent_actions":{"view_html":"https://pith.science/pith/EFVKG5HSG7KWGQXBINYQBDSQLQ","download_json":"https://pith.science/pith/EFVKG5HSG7KWGQXBINYQBDSQLQ.json","view_paper":"https://pith.science/paper/EFVKG5HS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2309.10400&json=true","fetch_graph":"https://pith.science/api/pith-number/EFVKG5HSG7KWGQXBINYQBDSQLQ/graph.json","fetch_events":"https://pith.science/api/pith-number/EFVKG5HSG7KWGQXBINYQBDSQLQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EFVKG5HSG7KWGQXBINYQBDSQLQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EFVKG5HSG7KWGQXBINYQBDSQLQ/action/storage_attestation","attest_author":"https://pith.science/pith/EFVKG5HSG7KWGQXBINYQBDSQLQ/action/author_attestation","sign_citation":"https://pith.science/pith/EFVKG5HSG7KWGQXBINYQBDSQLQ/action/citation_signature","submit_replication":"https://pith.science/pith/EFVKG5HSG7KWGQXBINYQBDSQLQ/action/replication_record"}},"created_at":"2026-07-05T07:47:44.981061+00:00","updated_at":"2026-07-05T07:47:44.981061+00:00"}