{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:JWS55NP6QTOTNOWDJKDVFCPFSD","short_pith_number":"pith:JWS55NP6","schema_version":"1.0","canonical_sha256":"4da5deb5fe84dd36bac34a875289e590fea44a78a6e3d0b13942ed01663e542b","source":{"kind":"arxiv","id":"2402.14017","version":2},"attestation_state":"computed","paper":{"title":"D-Flow: Differentiating through Flows for Controlled Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Brian Karrer, Heli Ben-Hamu, Itai Gat, Omri Puny, Uriel Singer, Yaron Lipman","submitted_at":"2024-02-21T18:56:03Z","abstract_excerpt":"Taming the generation outcome of state of the art Diffusion and Flow-Matching (FM) models without having to re-train a task-specific model unlocks a powerful tool for solving inverse problems, conditional generation, and controlled generation in general. In this work we introduce D-Flow, a simple framework for controlling the generation process by differentiating through the flow, optimizing for the source (noise) point. We motivate this framework by our key observation stating that for Diffusion/FM models trained with Gaussian probability paths, differentiating through the generation process "},"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":"2402.14017","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-21T18:56:03Z","cross_cats_sorted":[],"title_canon_sha256":"d1d03eb591c898d99b429a9f6bd83db106abca117fd7b65cf820912624050fc1","abstract_canon_sha256":"d69f3e6eb4e707b4db95d712af7dd559faa1d563746f016813f401d615981b31"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:46:21.791597Z","signature_b64":"u0d0slyN6wxyf/Los/RtYIVY/rqYCdDaLRiE13YPiop4Zj44ZGhC00qmOi236J/QxnR0nTcEImpoQawEV9leBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4da5deb5fe84dd36bac34a875289e590fea44a78a6e3d0b13942ed01663e542b","last_reissued_at":"2026-07-05T08:46:21.791177Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:46:21.791177Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"D-Flow: Differentiating through Flows for Controlled Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Brian Karrer, Heli Ben-Hamu, Itai Gat, Omri Puny, Uriel Singer, Yaron Lipman","submitted_at":"2024-02-21T18:56:03Z","abstract_excerpt":"Taming the generation outcome of state of the art Diffusion and Flow-Matching (FM) models without having to re-train a task-specific model unlocks a powerful tool for solving inverse problems, conditional generation, and controlled generation in general. In this work we introduce D-Flow, a simple framework for controlling the generation process by differentiating through the flow, optimizing for the source (noise) point. We motivate this framework by our key observation stating that for Diffusion/FM models trained with Gaussian probability paths, differentiating through the generation process "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.14017","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/2402.14017/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":"2402.14017","created_at":"2026-07-05T08:46:21.791233+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.14017v2","created_at":"2026-07-05T08:46:21.791233+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.14017","created_at":"2026-07-05T08:46:21.791233+00:00"},{"alias_kind":"pith_short_12","alias_value":"JWS55NP6QTOT","created_at":"2026-07-05T08:46:21.791233+00:00"},{"alias_kind":"pith_short_16","alias_value":"JWS55NP6QTOTNOWD","created_at":"2026-07-05T08:46:21.791233+00:00"},{"alias_kind":"pith_short_8","alias_value":"JWS55NP6","created_at":"2026-07-05T08:46:21.791233+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":11,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22958","citing_title":"PG-MAP: Joint MAP Optimization for Inference-Time Alignment of Diffusion and Flow-Matching Models","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00947","citing_title":"Diffeomorphic Optimization","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27147","citing_title":"How to Guide Your Flow: Few-Step Alignment via Flow Map Reward Guidance","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23275","citing_title":"Diffusion Domain Expansion: Learning to Coordinate Pre-trained Diffusion Models","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27147","citing_title":"How to Guide Your Flow: Few-Step Alignment via Flow Map Reward Guidance","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2501.09732","citing_title":"Inference-Time Scaling for Diffusion Models beyond Scaling Denoising Steps","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2511.16520","citing_title":"Saving Foundation Flow-Matching Priors for Inverse Problems","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27147","citing_title":"How to Guide Your Flow: Few-Step Alignment via Flow Map Reward Guidance","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10645","citing_title":"GenMed: A Pairwise Generative Reformulation of Medical Diagnostic Tasks","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08640","citing_title":"FlowADMM: Plug-and-play ADMM with Flow-based Renoise-Denoise Priors","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07253","citing_title":"LENS: Low-Frequency Eigen Noise Shaping for Efficient Diffusion Sampling","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JWS55NP6QTOTNOWDJKDVFCPFSD","json":"https://pith.science/pith/JWS55NP6QTOTNOWDJKDVFCPFSD.json","graph_json":"https://pith.science/api/pith-number/JWS55NP6QTOTNOWDJKDVFCPFSD/graph.json","events_json":"https://pith.science/api/pith-number/JWS55NP6QTOTNOWDJKDVFCPFSD/events.json","paper":"https://pith.science/paper/JWS55NP6"},"agent_actions":{"view_html":"https://pith.science/pith/JWS55NP6QTOTNOWDJKDVFCPFSD","download_json":"https://pith.science/pith/JWS55NP6QTOTNOWDJKDVFCPFSD.json","view_paper":"https://pith.science/paper/JWS55NP6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.14017&json=true","fetch_graph":"https://pith.science/api/pith-number/JWS55NP6QTOTNOWDJKDVFCPFSD/graph.json","fetch_events":"https://pith.science/api/pith-number/JWS55NP6QTOTNOWDJKDVFCPFSD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JWS55NP6QTOTNOWDJKDVFCPFSD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JWS55NP6QTOTNOWDJKDVFCPFSD/action/storage_attestation","attest_author":"https://pith.science/pith/JWS55NP6QTOTNOWDJKDVFCPFSD/action/author_attestation","sign_citation":"https://pith.science/pith/JWS55NP6QTOTNOWDJKDVFCPFSD/action/citation_signature","submit_replication":"https://pith.science/pith/JWS55NP6QTOTNOWDJKDVFCPFSD/action/replication_record"}},"created_at":"2026-07-05T08:46:21.791233+00:00","updated_at":"2026-07-05T08:46:21.791233+00:00"}