{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:FSRI2A2GU4FY4JROUAUOUINLGZ","short_pith_number":"pith:FSRI2A2G","schema_version":"1.0","canonical_sha256":"2ca28d0346a70b8e262ea028ea21ab366bd9dd8d2fbcc7dc87ab93c2d079d23f","source":{"kind":"arxiv","id":"2303.13703","version":2},"attestation_state":"computed","paper":{"title":"End-to-End Diffusion Latent Optimization Improves Classifier Guidance","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Akash Gokul, Bram Wallace, Nikhil Naik, Stefano Ermon","submitted_at":"2023-03-23T22:43:52Z","abstract_excerpt":"Classifier guidance -- using the gradients of an image classifier to steer the generations of a diffusion model -- has the potential to dramatically expand the creative control over image generation and editing. However, currently classifier guidance requires either training new noise-aware models to obtain accurate gradients or using a one-step denoising approximation of the final generation, which leads to misaligned gradients and sub-optimal control. We highlight this approximation's shortcomings and propose a novel guidance method: Direct Optimization of Diffusion Latents (DOODL), which en"},"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":"2303.13703","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-23T22:43:52Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"c6993ce94dae4fade3617170af4af10c861698b9192c0512157303b0860f7ed3","abstract_canon_sha256":"fa48c455e132519eecbe3bcb7750880fc476ef0d50a1bfa7cedd3a460b7cad3f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:16:17.195126Z","signature_b64":"epbMTKyg63CkbXkdqadBI3c0YtWH0ziSIqI4H8tsCZ7H8YUes/Bw0VQeLyss5aoZWapBejpTpb+asx7Egl+JAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2ca28d0346a70b8e262ea028ea21ab366bd9dd8d2fbcc7dc87ab93c2d079d23f","last_reissued_at":"2026-07-05T06:16:17.194629Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:16:17.194629Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"End-to-End Diffusion Latent Optimization Improves Classifier Guidance","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Akash Gokul, Bram Wallace, Nikhil Naik, Stefano Ermon","submitted_at":"2023-03-23T22:43:52Z","abstract_excerpt":"Classifier guidance -- using the gradients of an image classifier to steer the generations of a diffusion model -- has the potential to dramatically expand the creative control over image generation and editing. However, currently classifier guidance requires either training new noise-aware models to obtain accurate gradients or using a one-step denoising approximation of the final generation, which leads to misaligned gradients and sub-optimal control. We highlight this approximation's shortcomings and propose a novel guidance method: Direct Optimization of Diffusion Latents (DOODL), which en"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2303.13703","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/2303.13703/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":"2303.13703","created_at":"2026-07-05T06:16:17.194684+00:00"},{"alias_kind":"arxiv_version","alias_value":"2303.13703v2","created_at":"2026-07-05T06:16:17.194684+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.13703","created_at":"2026-07-05T06:16:17.194684+00:00"},{"alias_kind":"pith_short_12","alias_value":"FSRI2A2GU4FY","created_at":"2026-07-05T06:16:17.194684+00:00"},{"alias_kind":"pith_short_16","alias_value":"FSRI2A2GU4FY4JRO","created_at":"2026-07-05T06:16:17.194684+00:00"},{"alias_kind":"pith_short_8","alias_value":"FSRI2A2G","created_at":"2026-07-05T06:16:17.194684+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06421","citing_title":"Gradient-Based Inverse Design of Free-Energy Landscapes with Diffusion Models","ref_index":34,"is_internal_anchor":true},{"citing_arxiv_id":"2309.17400","citing_title":"Directly Fine-Tuning Diffusion Models on Differentiable Rewards","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FSRI2A2GU4FY4JROUAUOUINLGZ","json":"https://pith.science/pith/FSRI2A2GU4FY4JROUAUOUINLGZ.json","graph_json":"https://pith.science/api/pith-number/FSRI2A2GU4FY4JROUAUOUINLGZ/graph.json","events_json":"https://pith.science/api/pith-number/FSRI2A2GU4FY4JROUAUOUINLGZ/events.json","paper":"https://pith.science/paper/FSRI2A2G"},"agent_actions":{"view_html":"https://pith.science/pith/FSRI2A2GU4FY4JROUAUOUINLGZ","download_json":"https://pith.science/pith/FSRI2A2GU4FY4JROUAUOUINLGZ.json","view_paper":"https://pith.science/paper/FSRI2A2G","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2303.13703&json=true","fetch_graph":"https://pith.science/api/pith-number/FSRI2A2GU4FY4JROUAUOUINLGZ/graph.json","fetch_events":"https://pith.science/api/pith-number/FSRI2A2GU4FY4JROUAUOUINLGZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FSRI2A2GU4FY4JROUAUOUINLGZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FSRI2A2GU4FY4JROUAUOUINLGZ/action/storage_attestation","attest_author":"https://pith.science/pith/FSRI2A2GU4FY4JROUAUOUINLGZ/action/author_attestation","sign_citation":"https://pith.science/pith/FSRI2A2GU4FY4JROUAUOUINLGZ/action/citation_signature","submit_replication":"https://pith.science/pith/FSRI2A2GU4FY4JROUAUOUINLGZ/action/replication_record"}},"created_at":"2026-07-05T06:16:17.194684+00:00","updated_at":"2026-07-05T06:16:17.194684+00:00"}