{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:L4UQ4RCNFGVCIM3CL3LGOJ47CW","short_pith_number":"pith:L4UQ4RCN","schema_version":"1.0","canonical_sha256":"5f290e444d29aa2433625ed667279f15874b85d8478112bcadf93c5686858d5a","source":{"kind":"arxiv","id":"2412.12095","version":2},"attestation_state":"computed","paper":{"title":"Causal Diffusion Transformers for Generative Modeling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chaorui Deng, Deyao Zhu, Haoqi Fan, Kunchang Li, Shi Guang","submitted_at":"2024-12-16T18:59:29Z","abstract_excerpt":"We introduce Causal Diffusion as the autoregressive (AR) counterpart of Diffusion models. It is a next-token(s) forecasting framework that is friendly to both discrete and continuous modalities and compatible with existing next-token prediction models like LLaMA and GPT. While recent works attempt to combine diffusion with AR models, we show that introducing sequential factorization to a diffusion model can substantially improve its performance and enables a smooth transition between AR and diffusion generation modes. Hence, we propose CausalFusion - a decoder-only transformer that dual-factor"},"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":"2412.12095","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-12-16T18:59:29Z","cross_cats_sorted":[],"title_canon_sha256":"bbea16cfca6e600f3d8591304bd7c89f2f78dd26ccccaf93579a9bbad4a1a029","abstract_canon_sha256":"283a5a5e5f365b5bd0225b1d5abb95f8f8543b09c16690279ed005a0475e66f6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:50:37.421961Z","signature_b64":"DQ6R5jKbovowKq/AggvUr8yBnmwko7+5ybI1D5NC8U8iZEMy2uDIa+EH/IRKukwyT1JGP/CwXIdH25AsLw+KCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5f290e444d29aa2433625ed667279f15874b85d8478112bcadf93c5686858d5a","last_reissued_at":"2026-07-05T09:50:37.421462Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:50:37.421462Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Causal Diffusion Transformers for Generative Modeling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chaorui Deng, Deyao Zhu, Haoqi Fan, Kunchang Li, Shi Guang","submitted_at":"2024-12-16T18:59:29Z","abstract_excerpt":"We introduce Causal Diffusion as the autoregressive (AR) counterpart of Diffusion models. It is a next-token(s) forecasting framework that is friendly to both discrete and continuous modalities and compatible with existing next-token prediction models like LLaMA and GPT. While recent works attempt to combine diffusion with AR models, we show that introducing sequential factorization to a diffusion model can substantially improve its performance and enables a smooth transition between AR and diffusion generation modes. Hence, we propose CausalFusion - a decoder-only transformer that dual-factor"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.12095","kind":"arxiv","version":2},"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/2412.12095/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":"2412.12095","created_at":"2026-07-05T09:50:37.421522+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.12095v2","created_at":"2026-07-05T09:50:37.421522+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.12095","created_at":"2026-07-05T09:50:37.421522+00:00"},{"alias_kind":"pith_short_12","alias_value":"L4UQ4RCNFGVC","created_at":"2026-07-05T09:50:37.421522+00:00"},{"alias_kind":"pith_short_16","alias_value":"L4UQ4RCNFGVCIM3C","created_at":"2026-07-05T09:50:37.421522+00:00"},{"alias_kind":"pith_short_8","alias_value":"L4UQ4RCN","created_at":"2026-07-05T09:50:37.421522+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.20708","citing_title":"Rethinking Cross-Layer Information Routing in Diffusion Transformers","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30519","citing_title":"OmniMem: Scalable and Adaptive Memory Retrieval for Long Video Generation","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20708","citing_title":"Rethinking Cross-Layer Information Routing in Diffusion Transformers","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05890","citing_title":"RepFlow: Representation Enhanced Flow Matching for Causal Effect Estimation","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2506.08009","citing_title":"Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07209","citing_title":"INSPATIO-WORLD: A Real-Time 4D World Simulator via Spatiotemporal Autoregressive Modeling","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2505.14683","citing_title":"Emerging Properties in Unified Multimodal Pretraining","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/L4UQ4RCNFGVCIM3CL3LGOJ47CW","json":"https://pith.science/pith/L4UQ4RCNFGVCIM3CL3LGOJ47CW.json","graph_json":"https://pith.science/api/pith-number/L4UQ4RCNFGVCIM3CL3LGOJ47CW/graph.json","events_json":"https://pith.science/api/pith-number/L4UQ4RCNFGVCIM3CL3LGOJ47CW/events.json","paper":"https://pith.science/paper/L4UQ4RCN"},"agent_actions":{"view_html":"https://pith.science/pith/L4UQ4RCNFGVCIM3CL3LGOJ47CW","download_json":"https://pith.science/pith/L4UQ4RCNFGVCIM3CL3LGOJ47CW.json","view_paper":"https://pith.science/paper/L4UQ4RCN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.12095&json=true","fetch_graph":"https://pith.science/api/pith-number/L4UQ4RCNFGVCIM3CL3LGOJ47CW/graph.json","fetch_events":"https://pith.science/api/pith-number/L4UQ4RCNFGVCIM3CL3LGOJ47CW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/L4UQ4RCNFGVCIM3CL3LGOJ47CW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/L4UQ4RCNFGVCIM3CL3LGOJ47CW/action/storage_attestation","attest_author":"https://pith.science/pith/L4UQ4RCNFGVCIM3CL3LGOJ47CW/action/author_attestation","sign_citation":"https://pith.science/pith/L4UQ4RCNFGVCIM3CL3LGOJ47CW/action/citation_signature","submit_replication":"https://pith.science/pith/L4UQ4RCNFGVCIM3CL3LGOJ47CW/action/replication_record"}},"created_at":"2026-07-05T09:50:37.421522+00:00","updated_at":"2026-07-05T09:50:37.421522+00:00"}