{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LUTNDWXZYZGIHDSYYKTU5YMJRO","short_pith_number":"pith:LUTNDWXZ","schema_version":"1.0","canonical_sha256":"5d26d1daf9c64c838e58c2a74ee1898b8cffd4dc0feae605e40b9b9ff354d942","source":{"kind":"arxiv","id":"2602.10099","version":2},"attestation_state":"computed","paper":{"title":"Learning on the Manifold: Unlocking Standard Diffusion Transformers with Representation Encoders","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Amandeep Kumar, Vishal M. Patel","submitted_at":"2026-02-10T18:58:04Z","abstract_excerpt":"Leveraging representation encoders for generative modeling offers a path for efficient, high-fidelity synthesis. However, standard diffusion transformers fail to converge on these representations directly. While recent work attributes this to a capacity bottleneck proposing computationally expensive width scaling of diffusion transformers we demonstrate that the failure is fundamentally geometric. We identify Geometric Interference as the root cause: standard Euclidean flow matching forces probability paths through the low-density interior of the hyperspherical feature space of representation "},"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":"2602.10099","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-10T18:58:04Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"64b8b1099b6b86660867432f793057415257c0fabf66ec6da82568dc20ec8dc8","abstract_canon_sha256":"0a4c72afca9a084e7c4c2e817ec472ae61729f042b73f3126167e2d64e3be030"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:17:18.781344Z","signature_b64":"Q3aADwjUJd60X8QfiKtBi8MhFaiWJx2MFvGtQSZM+A6oMC55KwU/Pha4crgT6jf6GWeLCnyYTdrQBSTzaDKuBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5d26d1daf9c64c838e58c2a74ee1898b8cffd4dc0feae605e40b9b9ff354d942","last_reissued_at":"2026-07-07T02:17:18.780343Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:17:18.780343Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning on the Manifold: Unlocking Standard Diffusion Transformers with Representation Encoders","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Amandeep Kumar, Vishal M. Patel","submitted_at":"2026-02-10T18:58:04Z","abstract_excerpt":"Leveraging representation encoders for generative modeling offers a path for efficient, high-fidelity synthesis. However, standard diffusion transformers fail to converge on these representations directly. While recent work attributes this to a capacity bottleneck proposing computationally expensive width scaling of diffusion transformers we demonstrate that the failure is fundamentally geometric. We identify Geometric Interference as the root cause: standard Euclidean flow matching forces probability paths through the low-density interior of the hyperspherical feature space of representation "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.10099","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/2602.10099/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":"2602.10099","created_at":"2026-07-07T02:17:18.780504+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.10099v2","created_at":"2026-07-07T02:17:18.780504+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.10099","created_at":"2026-07-07T02:17:18.780504+00:00"},{"alias_kind":"pith_short_12","alias_value":"LUTNDWXZYZGI","created_at":"2026-07-07T02:17:18.780504+00:00"},{"alias_kind":"pith_short_16","alias_value":"LUTNDWXZYZGIHDSY","created_at":"2026-07-07T02:17:18.780504+00:00"},{"alias_kind":"pith_short_8","alias_value":"LUTNDWXZ","created_at":"2026-07-07T02:17:18.780504+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":8,"sample":[{"citing_arxiv_id":"2607.01987","citing_title":"Understanding Geometric Representations in Self-Supervised Vision Transformers via Subspace Intervention","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2606.00094","citing_title":"Diffusion Image Generation with Explicit Modeling of Data Manifold Geometry","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2606.07036","citing_title":"STREAM: Stochastic Riemannian Flow Matching with Anisotropic Decoder for Digital Histopathology Image Generation","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15193","citing_title":"Aligning Latent Geometry for Spherical Flow Matching in Image Generation","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2605.26230","citing_title":"Geometry-Aware Representation Denoising for Robust Multi-view 3D Reconstruction","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2606.00094","citing_title":"Diffusion Image Generation with Explicit Modeling of Data Manifold Geometry","ref_index":35,"is_internal_anchor":true},{"citing_arxiv_id":"2605.23028","citing_title":"RADAR: Relative Angular Divergence Across Representations","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21981","citing_title":"RiT: Vanilla Diffusion Transformers Suffice in Representation Space","ref_index":16,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LUTNDWXZYZGIHDSYYKTU5YMJRO","json":"https://pith.science/pith/LUTNDWXZYZGIHDSYYKTU5YMJRO.json","graph_json":"https://pith.science/api/pith-number/LUTNDWXZYZGIHDSYYKTU5YMJRO/graph.json","events_json":"https://pith.science/api/pith-number/LUTNDWXZYZGIHDSYYKTU5YMJRO/events.json","paper":"https://pith.science/paper/LUTNDWXZ"},"agent_actions":{"view_html":"https://pith.science/pith/LUTNDWXZYZGIHDSYYKTU5YMJRO","download_json":"https://pith.science/pith/LUTNDWXZYZGIHDSYYKTU5YMJRO.json","view_paper":"https://pith.science/paper/LUTNDWXZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.10099&json=true","fetch_graph":"https://pith.science/api/pith-number/LUTNDWXZYZGIHDSYYKTU5YMJRO/graph.json","fetch_events":"https://pith.science/api/pith-number/LUTNDWXZYZGIHDSYYKTU5YMJRO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LUTNDWXZYZGIHDSYYKTU5YMJRO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LUTNDWXZYZGIHDSYYKTU5YMJRO/action/storage_attestation","attest_author":"https://pith.science/pith/LUTNDWXZYZGIHDSYYKTU5YMJRO/action/author_attestation","sign_citation":"https://pith.science/pith/LUTNDWXZYZGIHDSYYKTU5YMJRO/action/citation_signature","submit_replication":"https://pith.science/pith/LUTNDWXZYZGIHDSYYKTU5YMJRO/action/replication_record"}},"created_at":"2026-07-07T02:17:18.780504+00:00","updated_at":"2026-07-07T02:17:18.780504+00:00"}