{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:VIYP2YOYL2ZHUDVP457IGSSF4X","short_pith_number":"pith:VIYP2YOY","schema_version":"1.0","canonical_sha256":"aa30fd61d85eb27a0eafe77e834a45e5f3672893eec8175b8cf33d7b644d500f","source":{"kind":"arxiv","id":"2505.19122","version":1},"attestation_state":"computed","paper":{"title":"Exploring Magnitude Preservation and Rotation Modulation in Diffusion Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Cristian Perez Jensen, Dimitri von R\\\"utte, Eric Tillman Bill, Sotiris Anagnostidis","submitted_at":"2025-05-25T12:25:50Z","abstract_excerpt":"Denoising diffusion models exhibit remarkable generative capabilities, but remain challenging to train due to their inherent stochasticity, where high-variance gradient estimates lead to slow convergence. Previous works have shown that magnitude preservation helps with stabilizing training in the U-net architecture. This work explores whether this effect extends to the Diffusion Transformer (DiT) architecture. As such, we propose a magnitude-preserving design that stabilizes training without normalization layers. Motivated by the goal of maintaining activation magnitudes, we additionally intro"},"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":"2505.19122","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-05-25T12:25:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"39313769d33bdc12b320db975e398ff328fc1e93a3ceb2ef0b1a4bf0396b2cc9","abstract_canon_sha256":"43ac78985d5753083601edae78e4c9e487e3fa98d4d671d458dde234bfc44682"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:09:25.234829Z","signature_b64":"qS1WLr+FaUy3m91uE3h/oPLQk0HkadRZ2AdxhvAFIhOLrxMzJtyO1JfkFlrZc6KAVV3OzdtXdRZrzDSIV6sjAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aa30fd61d85eb27a0eafe77e834a45e5f3672893eec8175b8cf33d7b644d500f","last_reissued_at":"2026-07-05T11:09:25.234297Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:09:25.234297Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Exploring Magnitude Preservation and Rotation Modulation in Diffusion Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Cristian Perez Jensen, Dimitri von R\\\"utte, Eric Tillman Bill, Sotiris Anagnostidis","submitted_at":"2025-05-25T12:25:50Z","abstract_excerpt":"Denoising diffusion models exhibit remarkable generative capabilities, but remain challenging to train due to their inherent stochasticity, where high-variance gradient estimates lead to slow convergence. Previous works have shown that magnitude preservation helps with stabilizing training in the U-net architecture. This work explores whether this effect extends to the Diffusion Transformer (DiT) architecture. As such, we propose a magnitude-preserving design that stabilizes training without normalization layers. Motivated by the goal of maintaining activation magnitudes, we additionally intro"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.19122","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/2505.19122/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":"2505.19122","created_at":"2026-07-05T11:09:25.234366+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.19122v1","created_at":"2026-07-05T11:09:25.234366+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.19122","created_at":"2026-07-05T11:09:25.234366+00:00"},{"alias_kind":"pith_short_12","alias_value":"VIYP2YOYL2ZH","created_at":"2026-07-05T11:09:25.234366+00:00"},{"alias_kind":"pith_short_16","alias_value":"VIYP2YOYL2ZHUDVP","created_at":"2026-07-05T11:09:25.234366+00:00"},{"alias_kind":"pith_short_8","alias_value":"VIYP2YOY","created_at":"2026-07-05T11:09:25.234366+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VIYP2YOYL2ZHUDVP457IGSSF4X","json":"https://pith.science/pith/VIYP2YOYL2ZHUDVP457IGSSF4X.json","graph_json":"https://pith.science/api/pith-number/VIYP2YOYL2ZHUDVP457IGSSF4X/graph.json","events_json":"https://pith.science/api/pith-number/VIYP2YOYL2ZHUDVP457IGSSF4X/events.json","paper":"https://pith.science/paper/VIYP2YOY"},"agent_actions":{"view_html":"https://pith.science/pith/VIYP2YOYL2ZHUDVP457IGSSF4X","download_json":"https://pith.science/pith/VIYP2YOYL2ZHUDVP457IGSSF4X.json","view_paper":"https://pith.science/paper/VIYP2YOY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.19122&json=true","fetch_graph":"https://pith.science/api/pith-number/VIYP2YOYL2ZHUDVP457IGSSF4X/graph.json","fetch_events":"https://pith.science/api/pith-number/VIYP2YOYL2ZHUDVP457IGSSF4X/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VIYP2YOYL2ZHUDVP457IGSSF4X/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VIYP2YOYL2ZHUDVP457IGSSF4X/action/storage_attestation","attest_author":"https://pith.science/pith/VIYP2YOYL2ZHUDVP457IGSSF4X/action/author_attestation","sign_citation":"https://pith.science/pith/VIYP2YOYL2ZHUDVP457IGSSF4X/action/citation_signature","submit_replication":"https://pith.science/pith/VIYP2YOYL2ZHUDVP457IGSSF4X/action/replication_record"}},"created_at":"2026-07-05T11:09:25.234366+00:00","updated_at":"2026-07-05T11:09:25.234366+00:00"}