{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XMVYVI3VDNZADBEOLI7WLVPVXY","short_pith_number":"pith:XMVYVI3V","schema_version":"1.0","canonical_sha256":"bb2b8aa3751b7201848e5a3f65d5f5be0e6f4af8c46a101ada4225d879a651a9","source":{"kind":"arxiv","id":"2503.14070","version":1},"attestation_state":"computed","paper":{"title":"Fast Autoregressive Video Generation with Diagonal Decoding","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Haoyu Wu, Jiang Bian, Junliang Guo, Katja Hofmann, Tabish Rashid, Tianyu He, Tim Pearce, Yang Ye","submitted_at":"2025-03-18T09:42:55Z","abstract_excerpt":"Autoregressive Transformer models have demonstrated impressive performance in video generation, but their sequential token-by-token decoding process poses a major bottleneck, particularly for long videos represented by tens of thousands of tokens. In this paper, we propose Diagonal Decoding (DiagD), a training-free inference acceleration algorithm for autoregressively pre-trained models that exploits spatial and temporal correlations in videos. Our method generates tokens along diagonal paths in the spatial-temporal token grid, enabling parallel decoding within each frame as well as partially "},"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":"2503.14070","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2025-03-18T09:42:55Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"198e55f98d6ba9ab5245d3dabbfea1d6ecfd7c14f99180c248931d6ba693c1a3","abstract_canon_sha256":"b51eeafcae0d3c0acd78532f8608bfa472fc5e62f8d3fbe30e974c81dee28525"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:33:54.487936Z","signature_b64":"4mNZhDsqi2ta/Whg3bDdiEL1qe2EBrrq2EpFnDYdEtoHmA6K1lBc5du3qeMoKsN/w2gvtqmJ0b5RdEP72nl8DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bb2b8aa3751b7201848e5a3f65d5f5be0e6f4af8c46a101ada4225d879a651a9","last_reissued_at":"2026-07-05T10:33:54.486940Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:33:54.486940Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fast Autoregressive Video Generation with Diagonal Decoding","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Haoyu Wu, Jiang Bian, Junliang Guo, Katja Hofmann, Tabish Rashid, Tianyu He, Tim Pearce, Yang Ye","submitted_at":"2025-03-18T09:42:55Z","abstract_excerpt":"Autoregressive Transformer models have demonstrated impressive performance in video generation, but their sequential token-by-token decoding process poses a major bottleneck, particularly for long videos represented by tens of thousands of tokens. In this paper, we propose Diagonal Decoding (DiagD), a training-free inference acceleration algorithm for autoregressively pre-trained models that exploits spatial and temporal correlations in videos. Our method generates tokens along diagonal paths in the spatial-temporal token grid, enabling parallel decoding within each frame as well as partially "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.14070","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/2503.14070/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":"2503.14070","created_at":"2026-07-05T10:33:54.487079+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.14070v1","created_at":"2026-07-05T10:33:54.487079+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.14070","created_at":"2026-07-05T10:33:54.487079+00:00"},{"alias_kind":"pith_short_12","alias_value":"XMVYVI3VDNZA","created_at":"2026-07-05T10:33:54.487079+00:00"},{"alias_kind":"pith_short_16","alias_value":"XMVYVI3VDNZADBEO","created_at":"2026-07-05T10:33:54.487079+00:00"},{"alias_kind":"pith_short_8","alias_value":"XMVYVI3V","created_at":"2026-07-05T10:33:54.487079+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.07982","citing_title":"Geometry Forcing: Marrying Video Diffusion and 3D Representation for Consistent World Modeling","ref_index":85,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18564","citing_title":"MultiWorld: Scalable Multi-Agent Multi-View Video World Models","ref_index":60,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XMVYVI3VDNZADBEOLI7WLVPVXY","json":"https://pith.science/pith/XMVYVI3VDNZADBEOLI7WLVPVXY.json","graph_json":"https://pith.science/api/pith-number/XMVYVI3VDNZADBEOLI7WLVPVXY/graph.json","events_json":"https://pith.science/api/pith-number/XMVYVI3VDNZADBEOLI7WLVPVXY/events.json","paper":"https://pith.science/paper/XMVYVI3V"},"agent_actions":{"view_html":"https://pith.science/pith/XMVYVI3VDNZADBEOLI7WLVPVXY","download_json":"https://pith.science/pith/XMVYVI3VDNZADBEOLI7WLVPVXY.json","view_paper":"https://pith.science/paper/XMVYVI3V","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.14070&json=true","fetch_graph":"https://pith.science/api/pith-number/XMVYVI3VDNZADBEOLI7WLVPVXY/graph.json","fetch_events":"https://pith.science/api/pith-number/XMVYVI3VDNZADBEOLI7WLVPVXY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XMVYVI3VDNZADBEOLI7WLVPVXY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XMVYVI3VDNZADBEOLI7WLVPVXY/action/storage_attestation","attest_author":"https://pith.science/pith/XMVYVI3VDNZADBEOLI7WLVPVXY/action/author_attestation","sign_citation":"https://pith.science/pith/XMVYVI3VDNZADBEOLI7WLVPVXY/action/citation_signature","submit_replication":"https://pith.science/pith/XMVYVI3VDNZADBEOLI7WLVPVXY/action/replication_record"}},"created_at":"2026-07-05T10:33:54.487079+00:00","updated_at":"2026-07-05T10:33:54.487079+00:00"}