{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:5XPMCECXSLTTHSIKEZSEINCLY4","short_pith_number":"pith:5XPMCECX","schema_version":"1.0","canonical_sha256":"eddec1105792e733c90a266444344bc73efd8978c2d3eff0147680783009838b","source":{"kind":"arxiv","id":"2504.05741","version":2},"attestation_state":"computed","paper":{"title":"DDT: Decoupled Diffusion Transformer","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Limin Wang, Shuai Wang, Weilin Huang, Zhi Tian","submitted_at":"2025-04-08T07:17:45Z","abstract_excerpt":"Diffusion transformers have demonstrated remarkable generation quality, albeit requiring longer training iterations and numerous inference steps. In each denoising step, diffusion transformers encode the noisy inputs to extract the lower-frequency semantic component and then decode the higher frequency with identical modules. This scheme creates an inherent optimization dilemma: encoding low-frequency semantics necessitates reducing high-frequency components, creating tension between semantic encoding and high-frequency decoding. To resolve this challenge, we propose a new \\textbf{\\color{ddt}D"},"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":"2504.05741","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-04-08T07:17:45Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7ca2086299119ca583ef839238173b04c66f43ff80a0c64dbb10b3c3001fe913","abstract_canon_sha256":"da1e2877663074ec877924ccddf883e14834ea05dd27fcd68e486f3ed07c06f2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:46:24.005568Z","signature_b64":"l9+1AGateecVDVlqIkelTafPW+rCUXjZXuA3bw4mvzlSEwugFkEmZkrxqe8v1yPqsDWYrXmh58FWrYLdplWwBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eddec1105792e733c90a266444344bc73efd8978c2d3eff0147680783009838b","last_reissued_at":"2026-07-05T10:46:24.005081Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:46:24.005081Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DDT: Decoupled Diffusion Transformer","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Limin Wang, Shuai Wang, Weilin Huang, Zhi Tian","submitted_at":"2025-04-08T07:17:45Z","abstract_excerpt":"Diffusion transformers have demonstrated remarkable generation quality, albeit requiring longer training iterations and numerous inference steps. In each denoising step, diffusion transformers encode the noisy inputs to extract the lower-frequency semantic component and then decode the higher frequency with identical modules. This scheme creates an inherent optimization dilemma: encoding low-frequency semantics necessitates reducing high-frequency components, creating tension between semantic encoding and high-frequency decoding. To resolve this challenge, we propose a new \\textbf{\\color{ddt}D"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.05741","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/2504.05741/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":"2504.05741","created_at":"2026-07-05T10:46:24.005141+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.05741v2","created_at":"2026-07-05T10:46:24.005141+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.05741","created_at":"2026-07-05T10:46:24.005141+00:00"},{"alias_kind":"pith_short_12","alias_value":"5XPMCECXSLTT","created_at":"2026-07-05T10:46:24.005141+00:00"},{"alias_kind":"pith_short_16","alias_value":"5XPMCECXSLTTHSIK","created_at":"2026-07-05T10:46:24.005141+00:00"},{"alias_kind":"pith_short_8","alias_value":"5XPMCECX","created_at":"2026-07-05T10:46:24.005141+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":26,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.08436","citing_title":"EgoWAM: World Action Models Beyond Pixels with In-the-Wild Egocentric Human Data","ref_index":67,"is_internal_anchor":true},{"citing_arxiv_id":"2607.08375","citing_title":"WCog-VLA: A Dual-Level World-Cognitive Vision-Language-Action Model for End-to-End Autonomous Driving","ref_index":48,"is_internal_anchor":true},{"citing_arxiv_id":"2606.18765","citing_title":"SpectralDiT: Timestep-Conditioned Spectral Residual Correction for Flow-Matching DiTs","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2607.02508","citing_title":"From SRA to Self-Flow: Data Augmentation or Self-Supervision?","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15741","citing_title":"HyperDiT: Hyper-Connected Transformers for High-Fidelity Pixel-Space Diffusion","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18267","citing_title":"SRC-Flow: Compact Semantic Representations Enable Normalizing Flows for Image Generation","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26230","citing_title":"Geometry-Aware Representation Denoising for Robust Multi-view 3D Reconstruction","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00094","citing_title":"Diffusion Image Generation with Explicit Modeling of Data Manifold Geometry","ref_index":66,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27760","citing_title":"PixelU: A U-Shaped Transformer for Efficient End-to-End Pixel Diffusion","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26449","citing_title":"Cross-scale Aligned Supervision for Training GANs","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16503","citing_title":"Motif-Video 2B: Technical Report","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15741","citing_title":"HyperDiT: Hyper-Connected Transformers for High-Fidelity Pixel-Space Diffusion","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17759","citing_title":"FrequencyBooster: Full-Frequency Modeling for High-Fidelity Pixel Diffusion","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18267","citing_title":"SRC-Flow: Compact Semantic Representations Enable Normalizing Flows for Image Generation","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18390","citing_title":"Vision Foundation Models as Generalist Tokenizers for Image Generation","ref_index":81,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16949","citing_title":"Beyond Point-Wise Matching: Structural Representation Alignment for Accelerating Diffusion Transformers","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2511.19365","citing_title":"DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generation","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2512.02012","citing_title":"Improved Mean Flows: On the Challenges of Fastforward Generative Models","ref_index":53,"is_internal_anchor":false},{"citing_arxiv_id":"2511.13720","citing_title":"Back to Basics: Let Denoising Generative Models Denoise","ref_index":71,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06388","citing_title":"Reconstruction or Semantics? What Makes a Latent Space Useful for Robotic World Models","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12273","citing_title":"SubFlow: Sub-mode Conditioned Flow Matching for Diverse One-Step Generation","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11521","citing_title":"Continuous Adversarial Flow Models","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16503","citing_title":"Motif-Video 2B: Technical Report","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11331","citing_title":"Any 3D Scene is Worth 1K Tokens: 3D-Grounded Representation for Scene Generation at Scale","ref_index":68,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07915","citing_title":"What Matters for Diffusion-Friendly Latent Manifold? Prior-Aligned Autoencoders for Latent Diffusion","ref_index":85,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5XPMCECXSLTTHSIKEZSEINCLY4","json":"https://pith.science/pith/5XPMCECXSLTTHSIKEZSEINCLY4.json","graph_json":"https://pith.science/api/pith-number/5XPMCECXSLTTHSIKEZSEINCLY4/graph.json","events_json":"https://pith.science/api/pith-number/5XPMCECXSLTTHSIKEZSEINCLY4/events.json","paper":"https://pith.science/paper/5XPMCECX"},"agent_actions":{"view_html":"https://pith.science/pith/5XPMCECXSLTTHSIKEZSEINCLY4","download_json":"https://pith.science/pith/5XPMCECXSLTTHSIKEZSEINCLY4.json","view_paper":"https://pith.science/paper/5XPMCECX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.05741&json=true","fetch_graph":"https://pith.science/api/pith-number/5XPMCECXSLTTHSIKEZSEINCLY4/graph.json","fetch_events":"https://pith.science/api/pith-number/5XPMCECXSLTTHSIKEZSEINCLY4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5XPMCECXSLTTHSIKEZSEINCLY4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5XPMCECXSLTTHSIKEZSEINCLY4/action/storage_attestation","attest_author":"https://pith.science/pith/5XPMCECXSLTTHSIKEZSEINCLY4/action/author_attestation","sign_citation":"https://pith.science/pith/5XPMCECXSLTTHSIKEZSEINCLY4/action/citation_signature","submit_replication":"https://pith.science/pith/5XPMCECXSLTTHSIKEZSEINCLY4/action/replication_record"}},"created_at":"2026-07-05T10:46:24.005141+00:00","updated_at":"2026-07-05T10:46:24.005141+00:00"}