{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RHPWDFI24FU7LRVAE4JGY422A2","short_pith_number":"pith:RHPWDFI2","schema_version":"1.0","canonical_sha256":"89df61951ae169f5c6a027126c735a06a33c2a9f331a1ba2e300acdab4d1bb11","source":{"kind":"arxiv","id":"2502.21079","version":1},"attestation_state":"computed","paper":{"title":"Training-free and Adaptive Sparse Attention for Efficient Long Video Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bin Cui, Fangcheng Fu, Huixia Li, Suhan Ling, Xuefeng Xiao, Yifei Xia, Yujie Wang","submitted_at":"2025-02-28T14:11:20Z","abstract_excerpt":"Generating high-fidelity long videos with Diffusion Transformers (DiTs) is often hindered by significant latency, primarily due to the computational demands of attention mechanisms. For instance, generating an 8-second 720p video (110K tokens) with HunyuanVideo takes about 600 PFLOPs, with around 500 PFLOPs consumed by attention computations. To address this issue, we propose AdaSpa, the first Dynamic Pattern and Online Precise Search sparse attention method. Firstly, to realize the Dynamic Pattern, we introduce a blockified pattern to efficiently capture the hierarchical sparsity inherent in "},"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":"2502.21079","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-02-28T14:11:20Z","cross_cats_sorted":[],"title_canon_sha256":"0af40f196b3b0177bff7c0d8b6f291a18d004bc232b920e39c6a4b14750ea00f","abstract_canon_sha256":"375e6a672d8754d4f6207327c6493886cc7512c100f47becbbc320db419096cc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:21:44.085659Z","signature_b64":"Xbswi8AHHGW/CuMW2bOgTDiEQsydduh2gKl4tYA5saOLbyuS/tWaVlUTjgj3du+aouJKlCz2WN0WX0XtfvDpDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"89df61951ae169f5c6a027126c735a06a33c2a9f331a1ba2e300acdab4d1bb11","last_reissued_at":"2026-07-05T10:21:44.085158Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:21:44.085158Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Training-free and Adaptive Sparse Attention for Efficient Long Video Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bin Cui, Fangcheng Fu, Huixia Li, Suhan Ling, Xuefeng Xiao, Yifei Xia, Yujie Wang","submitted_at":"2025-02-28T14:11:20Z","abstract_excerpt":"Generating high-fidelity long videos with Diffusion Transformers (DiTs) is often hindered by significant latency, primarily due to the computational demands of attention mechanisms. For instance, generating an 8-second 720p video (110K tokens) with HunyuanVideo takes about 600 PFLOPs, with around 500 PFLOPs consumed by attention computations. To address this issue, we propose AdaSpa, the first Dynamic Pattern and Online Precise Search sparse attention method. Firstly, to realize the Dynamic Pattern, we introduce a blockified pattern to efficiently capture the hierarchical sparsity inherent in "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.21079","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/2502.21079/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":"2502.21079","created_at":"2026-07-05T10:21:44.085223+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.21079v1","created_at":"2026-07-05T10:21:44.085223+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.21079","created_at":"2026-07-05T10:21:44.085223+00:00"},{"alias_kind":"pith_short_12","alias_value":"RHPWDFI24FU7","created_at":"2026-07-05T10:21:44.085223+00:00"},{"alias_kind":"pith_short_16","alias_value":"RHPWDFI24FU7LRVA","created_at":"2026-07-05T10:21:44.085223+00:00"},{"alias_kind":"pith_short_8","alias_value":"RHPWDFI2","created_at":"2026-07-05T10:21:44.085223+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06309","citing_title":"RhymeFlow: Training-Free Acceleration for Video Generation with Asynchronous Denoising Flow Scheduling","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2603.21002","citing_title":"SURF: Signature-Retained Fast Video Generation","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2603.18636","citing_title":"Attention Sparsity is Input-Stable: Training-Free Sparse Attention for Video Generation via Offline Sparsity Profiling and Online QK Co-Clustering","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14513","citing_title":"HASTE: Training-Free Video Diffusion Acceleration via Head-Wise Adaptive Sparse Attention","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2506.09113","citing_title":"Seedance 1.0: Exploring the Boundaries of Video Generation Models","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12219","citing_title":"Ride the Wave: Precision-Allocated Sparse Attention for Smooth Video Generation","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18348","citing_title":"AdaCluster: Adaptive Query-Key Clustering for Sparse Attention in Video Generation","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15911","citing_title":"Efficient Video Diffusion Models: Advancements and Challenges","ref_index":153,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RHPWDFI24FU7LRVAE4JGY422A2","json":"https://pith.science/pith/RHPWDFI24FU7LRVAE4JGY422A2.json","graph_json":"https://pith.science/api/pith-number/RHPWDFI24FU7LRVAE4JGY422A2/graph.json","events_json":"https://pith.science/api/pith-number/RHPWDFI24FU7LRVAE4JGY422A2/events.json","paper":"https://pith.science/paper/RHPWDFI2"},"agent_actions":{"view_html":"https://pith.science/pith/RHPWDFI24FU7LRVAE4JGY422A2","download_json":"https://pith.science/pith/RHPWDFI24FU7LRVAE4JGY422A2.json","view_paper":"https://pith.science/paper/RHPWDFI2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.21079&json=true","fetch_graph":"https://pith.science/api/pith-number/RHPWDFI24FU7LRVAE4JGY422A2/graph.json","fetch_events":"https://pith.science/api/pith-number/RHPWDFI24FU7LRVAE4JGY422A2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RHPWDFI24FU7LRVAE4JGY422A2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RHPWDFI24FU7LRVAE4JGY422A2/action/storage_attestation","attest_author":"https://pith.science/pith/RHPWDFI24FU7LRVAE4JGY422A2/action/author_attestation","sign_citation":"https://pith.science/pith/RHPWDFI24FU7LRVAE4JGY422A2/action/citation_signature","submit_replication":"https://pith.science/pith/RHPWDFI24FU7LRVAE4JGY422A2/action/replication_record"}},"created_at":"2026-07-05T10:21:44.085223+00:00","updated_at":"2026-07-05T10:21:44.085223+00:00"}