{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:GRNN257ZI2PSUFI2MDK5MMZJDF","short_pith_number":"pith:GRNN257Z","schema_version":"1.0","canonical_sha256":"345add77f9469f2a151a60d5d633291948b9dc225dd08b0fc14df31b6639329a","source":{"kind":"arxiv","id":"2508.14807","version":2},"attestation_state":"computed","paper":{"title":"Source-Guided Flow Matching","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alice Harting, Karl H. Johansson, Matthieu Barreau, Michael M. Zavlanos, Zifan Wang","submitted_at":"2025-08-20T15:56:25Z","abstract_excerpt":"Guidance of generative models is typically achieved by modifying the probability flow vector field through the addition of a guidance field. In this paper, we instead propose the Source-Guided Flow Matching (SGFM) framework, which modifies the source distribution directly while keeping the pre-trained vector field intact. This reduces the guidance problem to a well-defined problem of sampling from the source distribution. We theoretically show that SGFM recovers the desired target distribution exactly. Furthermore, we provide bounds on the Wasserstein error for the generated distribution when "},"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":"2508.14807","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-20T15:56:25Z","cross_cats_sorted":[],"title_canon_sha256":"0af9f9563c3f4981d852880896d82689f3b096367c71ff7aea2fe37806ff660c","abstract_canon_sha256":"7eb7f4f4880d7b03721ff3c3a41f93ae055e8f04d229f68b22e4c21807b59b6c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:57:39.212564Z","signature_b64":"VYYfZzUzVI/5Qw17Eqo/hGsGQTTaQ5xvAnEaAnLmp/arv3JROA0oclAs/GlPWyAhc/0cJEm18w9vOPa7oO89AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"345add77f9469f2a151a60d5d633291948b9dc225dd08b0fc14df31b6639329a","last_reissued_at":"2026-07-05T11:57:39.212084Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:57:39.212084Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Source-Guided Flow Matching","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alice Harting, Karl H. Johansson, Matthieu Barreau, Michael M. Zavlanos, Zifan Wang","submitted_at":"2025-08-20T15:56:25Z","abstract_excerpt":"Guidance of generative models is typically achieved by modifying the probability flow vector field through the addition of a guidance field. In this paper, we instead propose the Source-Guided Flow Matching (SGFM) framework, which modifies the source distribution directly while keeping the pre-trained vector field intact. This reduces the guidance problem to a well-defined problem of sampling from the source distribution. We theoretically show that SGFM recovers the desired target distribution exactly. Furthermore, we provide bounds on the Wasserstein error for the generated distribution when "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.14807","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/2508.14807/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":"2508.14807","created_at":"2026-07-05T11:57:39.212140+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.14807v2","created_at":"2026-07-05T11:57:39.212140+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.14807","created_at":"2026-07-05T11:57:39.212140+00:00"},{"alias_kind":"pith_short_12","alias_value":"GRNN257ZI2PS","created_at":"2026-07-05T11:57:39.212140+00:00"},{"alias_kind":"pith_short_16","alias_value":"GRNN257ZI2PSUFI2","created_at":"2026-07-05T11:57:39.212140+00:00"},{"alias_kind":"pith_short_8","alias_value":"GRNN257Z","created_at":"2026-07-05T11:57:39.212140+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.05478","citing_title":"Can We Predict The Human Preference For Text-to-Image Content Prior To Generation And Is It Even Useful To Do So?","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04804","citing_title":"The Right Measure for Physics-Constrained Generation: A Co-Area Correction for Posterior-Consistent PDE Inverse Problems","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20758","citing_title":"Conflict-Aware Additive Guidance for Flow Models under Compositional Rewards","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GRNN257ZI2PSUFI2MDK5MMZJDF","json":"https://pith.science/pith/GRNN257ZI2PSUFI2MDK5MMZJDF.json","graph_json":"https://pith.science/api/pith-number/GRNN257ZI2PSUFI2MDK5MMZJDF/graph.json","events_json":"https://pith.science/api/pith-number/GRNN257ZI2PSUFI2MDK5MMZJDF/events.json","paper":"https://pith.science/paper/GRNN257Z"},"agent_actions":{"view_html":"https://pith.science/pith/GRNN257ZI2PSUFI2MDK5MMZJDF","download_json":"https://pith.science/pith/GRNN257ZI2PSUFI2MDK5MMZJDF.json","view_paper":"https://pith.science/paper/GRNN257Z","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.14807&json=true","fetch_graph":"https://pith.science/api/pith-number/GRNN257ZI2PSUFI2MDK5MMZJDF/graph.json","fetch_events":"https://pith.science/api/pith-number/GRNN257ZI2PSUFI2MDK5MMZJDF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GRNN257ZI2PSUFI2MDK5MMZJDF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GRNN257ZI2PSUFI2MDK5MMZJDF/action/storage_attestation","attest_author":"https://pith.science/pith/GRNN257ZI2PSUFI2MDK5MMZJDF/action/author_attestation","sign_citation":"https://pith.science/pith/GRNN257ZI2PSUFI2MDK5MMZJDF/action/citation_signature","submit_replication":"https://pith.science/pith/GRNN257ZI2PSUFI2MDK5MMZJDF/action/replication_record"}},"created_at":"2026-07-05T11:57:39.212140+00:00","updated_at":"2026-07-05T11:57:39.212140+00:00"}