{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:PGI3OGPJZAEM2QWDNQ56DJD5R3","short_pith_number":"pith:PGI3OGPJ","schema_version":"1.0","canonical_sha256":"7991b719e9c808cd42c36c3be1a47d8efcd3f192c2e4d59fe8f1c4e8dd5c2e73","source":{"kind":"arxiv","id":"2402.18491","version":1},"attestation_state":"computed","paper":{"title":"Dynamical Regimes of Diffusion Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cond-mat.stat-mech"],"primary_cat":"cs.LG","authors_text":"Giulio Biroli, Marc M\\'ezard, Tony Bonnaire, Valentin De Bortoli","submitted_at":"2024-02-28T17:19:26Z","abstract_excerpt":"Using statistical physics methods, we study generative diffusion models in the regime where the dimension of space and the number of data are large, and the score function has been trained optimally. Our analysis reveals three distinct dynamical regimes during the backward generative diffusion process. The generative dynamics, starting from pure noise, encounters first a 'speciation' transition where the gross structure of data is unraveled, through a mechanism similar to symmetry breaking in phase transitions. It is followed at later time by a 'collapse' transition where the trajectories of t"},"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":"2402.18491","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-28T17:19:26Z","cross_cats_sorted":["cond-mat.stat-mech"],"title_canon_sha256":"e765e8cedb9a2c366819206a5443491514e26f49d264d3ad31be9a57e74c780c","abstract_canon_sha256":"4ff6a3eae9f23ec1d02c8ea2ff0be930eda19105ba108bd240c2827f1e94bb14"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:57:39.960825Z","signature_b64":"DwSJ9ziPrurLZ7bOhdfd5a5l4ddClVhx1NsSb5bYcVogz1H7Ob4dv/6q/e2tcOXep9S8ISAF7oTbp+mg6Ya9AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7991b719e9c808cd42c36c3be1a47d8efcd3f192c2e4d59fe8f1c4e8dd5c2e73","last_reissued_at":"2026-07-05T09:57:39.960332Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:57:39.960332Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dynamical Regimes of Diffusion Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cond-mat.stat-mech"],"primary_cat":"cs.LG","authors_text":"Giulio Biroli, Marc M\\'ezard, Tony Bonnaire, Valentin De Bortoli","submitted_at":"2024-02-28T17:19:26Z","abstract_excerpt":"Using statistical physics methods, we study generative diffusion models in the regime where the dimension of space and the number of data are large, and the score function has been trained optimally. Our analysis reveals three distinct dynamical regimes during the backward generative diffusion process. The generative dynamics, starting from pure noise, encounters first a 'speciation' transition where the gross structure of data is unraveled, through a mechanism similar to symmetry breaking in phase transitions. It is followed at later time by a 'collapse' transition where the trajectories of t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.18491","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/2402.18491/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":"2402.18491","created_at":"2026-07-05T09:57:39.960390+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.18491v1","created_at":"2026-07-05T09:57:39.960390+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.18491","created_at":"2026-07-05T09:57:39.960390+00:00"},{"alias_kind":"pith_short_12","alias_value":"PGI3OGPJZAEM","created_at":"2026-07-05T09:57:39.960390+00:00"},{"alias_kind":"pith_short_16","alias_value":"PGI3OGPJZAEM2QWD","created_at":"2026-07-05T09:57:39.960390+00:00"},{"alias_kind":"pith_short_8","alias_value":"PGI3OGPJ","created_at":"2026-07-05T09:57:39.960390+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08041","citing_title":"An exact information theory of generalization phase transitions in Bayesian diffusion models","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2606.03848","citing_title":"Generating quantum ensembles via reverse-time quantum diffusions","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29150","citing_title":"Flow Reasoning Models: Scaling Reasoning Through Iterative Self-Refinement","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2411.19182","citing_title":"SOWing Information: Cultivating Contextual Coherence with MLLMs in Image Generation","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PGI3OGPJZAEM2QWDNQ56DJD5R3","json":"https://pith.science/pith/PGI3OGPJZAEM2QWDNQ56DJD5R3.json","graph_json":"https://pith.science/api/pith-number/PGI3OGPJZAEM2QWDNQ56DJD5R3/graph.json","events_json":"https://pith.science/api/pith-number/PGI3OGPJZAEM2QWDNQ56DJD5R3/events.json","paper":"https://pith.science/paper/PGI3OGPJ"},"agent_actions":{"view_html":"https://pith.science/pith/PGI3OGPJZAEM2QWDNQ56DJD5R3","download_json":"https://pith.science/pith/PGI3OGPJZAEM2QWDNQ56DJD5R3.json","view_paper":"https://pith.science/paper/PGI3OGPJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.18491&json=true","fetch_graph":"https://pith.science/api/pith-number/PGI3OGPJZAEM2QWDNQ56DJD5R3/graph.json","fetch_events":"https://pith.science/api/pith-number/PGI3OGPJZAEM2QWDNQ56DJD5R3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PGI3OGPJZAEM2QWDNQ56DJD5R3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PGI3OGPJZAEM2QWDNQ56DJD5R3/action/storage_attestation","attest_author":"https://pith.science/pith/PGI3OGPJZAEM2QWDNQ56DJD5R3/action/author_attestation","sign_citation":"https://pith.science/pith/PGI3OGPJZAEM2QWDNQ56DJD5R3/action/citation_signature","submit_replication":"https://pith.science/pith/PGI3OGPJZAEM2QWDNQ56DJD5R3/action/replication_record"}},"created_at":"2026-07-05T09:57:39.960390+00:00","updated_at":"2026-07-05T09:57:39.960390+00:00"}