{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:POASZKA75P442VQYIVRIOAHU5R","short_pith_number":"pith:POASZKA7","schema_version":"1.0","canonical_sha256":"7b812ca81febf9cd561845628700f4ec49a65a206baba4ae42462dcee30c399d","source":{"kind":"arxiv","id":"2212.10947","version":3},"attestation_state":"computed","paper":{"title":"Parallel Context Windows for Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Amnon Shashua, Ehud Karpas, Inbal Magar, Kevin Leyton-Brown, Nir Ratner, Omri Abend, Ori Ram, Yoav Levine, Yoav Shoham, Yonatan Belinkov","submitted_at":"2022-12-21T11:38:51Z","abstract_excerpt":"When applied to processing long text, Large Language Models (LLMs) are limited by their context window. Existing efforts to address this limitation involve training specialized architectures, and cannot be easily applied to off-the-shelf LLMs. We present Parallel Context Windows (PCW), a method that alleviates the context window restriction for any off-the-shelf LLM without further training. The key to the approach is to carve a long context into chunks (``windows''), restrict the attention mechanism to apply only within each window, and re-use the positional embeddings across the windows. Our"},"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":"2212.10947","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-12-21T11:38:51Z","cross_cats_sorted":[],"title_canon_sha256":"db641a471d0a2563284fda08e61a66e4b71b580641f50932f8083d5db6eb978b","abstract_canon_sha256":"28936fbe8f2db635f376bd52d4045f286625d82f00409823271c3fa4986b778c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:36:26.769181Z","signature_b64":"hgJNQLzeuEu3jNkDc25BQZ7sbfSKqIHFdsoSuBCHmuGaAn59cCW48UEaYFrHAVWu3GDKPebZsJb4EilHSeRNDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7b812ca81febf9cd561845628700f4ec49a65a206baba4ae42462dcee30c399d","last_reissued_at":"2026-07-05T06:36:26.768648Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:36:26.768648Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Parallel Context Windows for Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Amnon Shashua, Ehud Karpas, Inbal Magar, Kevin Leyton-Brown, Nir Ratner, Omri Abend, Ori Ram, Yoav Levine, Yoav Shoham, Yonatan Belinkov","submitted_at":"2022-12-21T11:38:51Z","abstract_excerpt":"When applied to processing long text, Large Language Models (LLMs) are limited by their context window. Existing efforts to address this limitation involve training specialized architectures, and cannot be easily applied to off-the-shelf LLMs. We present Parallel Context Windows (PCW), a method that alleviates the context window restriction for any off-the-shelf LLM without further training. The key to the approach is to carve a long context into chunks (``windows''), restrict the attention mechanism to apply only within each window, and re-use the positional embeddings across the windows. Our"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2212.10947","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/2212.10947/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":"2212.10947","created_at":"2026-07-05T06:36:26.768715+00:00"},{"alias_kind":"arxiv_version","alias_value":"2212.10947v3","created_at":"2026-07-05T06:36:26.768715+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2212.10947","created_at":"2026-07-05T06:36:26.768715+00:00"},{"alias_kind":"pith_short_12","alias_value":"POASZKA75P44","created_at":"2026-07-05T06:36:26.768715+00:00"},{"alias_kind":"pith_short_16","alias_value":"POASZKA75P442VQY","created_at":"2026-07-05T06:36:26.768715+00:00"},{"alias_kind":"pith_short_8","alias_value":"POASZKA7","created_at":"2026-07-05T06:36:26.768715+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2404.07143","citing_title":"Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04595","citing_title":"A Queueing-Theoretic Framework for Stability Analysis of LLM Inference with KV Cache Memory Constraints","ref_index":116,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/POASZKA75P442VQYIVRIOAHU5R","json":"https://pith.science/pith/POASZKA75P442VQYIVRIOAHU5R.json","graph_json":"https://pith.science/api/pith-number/POASZKA75P442VQYIVRIOAHU5R/graph.json","events_json":"https://pith.science/api/pith-number/POASZKA75P442VQYIVRIOAHU5R/events.json","paper":"https://pith.science/paper/POASZKA7"},"agent_actions":{"view_html":"https://pith.science/pith/POASZKA75P442VQYIVRIOAHU5R","download_json":"https://pith.science/pith/POASZKA75P442VQYIVRIOAHU5R.json","view_paper":"https://pith.science/paper/POASZKA7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2212.10947&json=true","fetch_graph":"https://pith.science/api/pith-number/POASZKA75P442VQYIVRIOAHU5R/graph.json","fetch_events":"https://pith.science/api/pith-number/POASZKA75P442VQYIVRIOAHU5R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/POASZKA75P442VQYIVRIOAHU5R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/POASZKA75P442VQYIVRIOAHU5R/action/storage_attestation","attest_author":"https://pith.science/pith/POASZKA75P442VQYIVRIOAHU5R/action/author_attestation","sign_citation":"https://pith.science/pith/POASZKA75P442VQYIVRIOAHU5R/action/citation_signature","submit_replication":"https://pith.science/pith/POASZKA75P442VQYIVRIOAHU5R/action/replication_record"}},"created_at":"2026-07-05T06:36:26.768715+00:00","updated_at":"2026-07-05T06:36:26.768715+00:00"}