{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:TR4VMYADVERBTIWUPKL5JLZXSM","short_pith_number":"pith:TR4VMYAD","schema_version":"1.0","canonical_sha256":"9c79566003a92219a2d47a97d4af37932b5d666acd17a7aaf7eedbb055bfbb40","source":{"kind":"arxiv","id":"2505.14708","version":1},"attestation_state":"computed","paper":{"title":"DraftAttention: Fast Video Diffusion via Low-Resolution Attention Guidance","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chenxia Han, Jiuxiang Gu, Pu Zhao, Quanyi Wang, Xuan Shen, Yanyue Xie, Yanzhi Wang, Yifan Gong, Yiwei Wang, Yufa Zhou","submitted_at":"2025-05-17T04:34:34Z","abstract_excerpt":"Diffusion transformer-based video generation models (DiTs) have recently attracted widespread attention for their excellent generation quality. However, their computational cost remains a major bottleneck-attention alone accounts for over 80% of total latency, and generating just 8 seconds of 720p video takes tens of minutes-posing serious challenges to practical application and scalability. To address this, we propose the DraftAttention, a training-free framework for the acceleration of video diffusion transformers with dynamic sparse attention on GPUs. We apply down-sampling to each feature "},"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":"2505.14708","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2025-05-17T04:34:34Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"06f016904e27713b306f3ea713c3e90f0f393c7c3deae9ad52bc1c5aec3cf36f","abstract_canon_sha256":"070274f340e367db347aecc29b04ba4b8ba35868932e2b8d4baf1f9eb9e3f2c6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:06:34.053391Z","signature_b64":"b4pCKLoFflqqY3e/npEErKYX8xDAUBa1t04vRy4jO/giyGnE3UUfut3joVuEIz9JnqC4UkCYTPQq5x+G0sSDCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9c79566003a92219a2d47a97d4af37932b5d666acd17a7aaf7eedbb055bfbb40","last_reissued_at":"2026-07-05T11:06:34.052932Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:06:34.052932Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DraftAttention: Fast Video Diffusion via Low-Resolution Attention Guidance","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chenxia Han, Jiuxiang Gu, Pu Zhao, Quanyi Wang, Xuan Shen, Yanyue Xie, Yanzhi Wang, Yifan Gong, Yiwei Wang, Yufa Zhou","submitted_at":"2025-05-17T04:34:34Z","abstract_excerpt":"Diffusion transformer-based video generation models (DiTs) have recently attracted widespread attention for their excellent generation quality. However, their computational cost remains a major bottleneck-attention alone accounts for over 80% of total latency, and generating just 8 seconds of 720p video takes tens of minutes-posing serious challenges to practical application and scalability. To address this, we propose the DraftAttention, a training-free framework for the acceleration of video diffusion transformers with dynamic sparse attention on GPUs. We apply down-sampling to each feature "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.14708","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/2505.14708/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":"2505.14708","created_at":"2026-07-05T11:06:34.052989+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.14708v1","created_at":"2026-07-05T11:06:34.052989+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.14708","created_at":"2026-07-05T11:06:34.052989+00:00"},{"alias_kind":"pith_short_12","alias_value":"TR4VMYADVERB","created_at":"2026-07-05T11:06:34.052989+00:00"},{"alias_kind":"pith_short_16","alias_value":"TR4VMYADVERBTIWU","created_at":"2026-07-05T11:06:34.052989+00:00"},{"alias_kind":"pith_short_8","alias_value":"TR4VMYAD","created_at":"2026-07-05T11:06:34.052989+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.23743","citing_title":"Sol Video Inference Engine: Agent-Native Full-Stack Acceleration Framework for Efficient Video Generation","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29579","citing_title":"ScAle: Attention Head Scaling as a Minimal Adapter for Spatial Reasoning in Vision Language Models","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30519","citing_title":"OmniMem: Scalable and Adaptive Memory Retrieval for Long Video Generation","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23445","citing_title":"DFSAttn: Dynamic Fine-grained Sparse Attention for Efficient Video Generation","ref_index":8,"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":3,"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":25,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12219","citing_title":"Ride the Wave: Precision-Allocated Sparse Attention for Smooth Video Generation","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15911","citing_title":"Efficient Video Diffusion Models: Advancements and Challenges","ref_index":118,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16592","citing_title":"Human Cognition in Machines: A Unified Perspective of World Models","ref_index":151,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TR4VMYADVERBTIWUPKL5JLZXSM","json":"https://pith.science/pith/TR4VMYADVERBTIWUPKL5JLZXSM.json","graph_json":"https://pith.science/api/pith-number/TR4VMYADVERBTIWUPKL5JLZXSM/graph.json","events_json":"https://pith.science/api/pith-number/TR4VMYADVERBTIWUPKL5JLZXSM/events.json","paper":"https://pith.science/paper/TR4VMYAD"},"agent_actions":{"view_html":"https://pith.science/pith/TR4VMYADVERBTIWUPKL5JLZXSM","download_json":"https://pith.science/pith/TR4VMYADVERBTIWUPKL5JLZXSM.json","view_paper":"https://pith.science/paper/TR4VMYAD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.14708&json=true","fetch_graph":"https://pith.science/api/pith-number/TR4VMYADVERBTIWUPKL5JLZXSM/graph.json","fetch_events":"https://pith.science/api/pith-number/TR4VMYADVERBTIWUPKL5JLZXSM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TR4VMYADVERBTIWUPKL5JLZXSM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TR4VMYADVERBTIWUPKL5JLZXSM/action/storage_attestation","attest_author":"https://pith.science/pith/TR4VMYADVERBTIWUPKL5JLZXSM/action/author_attestation","sign_citation":"https://pith.science/pith/TR4VMYADVERBTIWUPKL5JLZXSM/action/citation_signature","submit_replication":"https://pith.science/pith/TR4VMYADVERBTIWUPKL5JLZXSM/action/replication_record"}},"created_at":"2026-07-05T11:06:34.052989+00:00","updated_at":"2026-07-05T11:06:34.052989+00:00"}