{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:35RS6SZMUAWTUCFKD525RXQYLA","short_pith_number":"pith:35RS6SZM","schema_version":"1.0","canonical_sha256":"df632f4b2ca02d3a08aa1f75d8de18580def9d90f41e6159c5c5203d829875cf","source":{"kind":"arxiv","id":"2502.09611","version":1},"attestation_state":"computed","paper":{"title":"Designing a Conditional Prior Distribution for Flow-Based Generative Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Mohammad Salama, Noam Issachar, Raanan Fattal, Sagie Benaim","submitted_at":"2025-02-13T18:58:15Z","abstract_excerpt":"Flow-based generative models have recently shown impressive performance for conditional generation tasks, such as text-to-image generation. However, current methods transform a general unimodal noise distribution to a specific mode of the target data distribution. As such, every point in the initial source distribution can be mapped to every point in the target distribution, resulting in long average paths. To this end, in this work, we tap into a non-utilized property of conditional flow-based models: the ability to design a non-trivial prior distribution. Given an input condition, such as a "},"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":"2502.09611","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-13T18:58:15Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"06dbfed60cc8633465efd11704e1bb1f7b1a09201555b6ec4efa24defb73906f","abstract_canon_sha256":"8837641b1190a772405afcaa67c21471e0f6c1da0524d67a89825208be6529da"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:13:59.981200Z","signature_b64":"LkVh5lcSOnNX4WVZ2eXcuJK39Ybdr7InudVcdtzMqSxRJGTXjPqI/yzl6X9ep4iXmDjPaE/+zHtPUqQKkVCCAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"df632f4b2ca02d3a08aa1f75d8de18580def9d90f41e6159c5c5203d829875cf","last_reissued_at":"2026-07-05T10:13:59.980702Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:13:59.980702Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Designing a Conditional Prior Distribution for Flow-Based Generative Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Mohammad Salama, Noam Issachar, Raanan Fattal, Sagie Benaim","submitted_at":"2025-02-13T18:58:15Z","abstract_excerpt":"Flow-based generative models have recently shown impressive performance for conditional generation tasks, such as text-to-image generation. However, current methods transform a general unimodal noise distribution to a specific mode of the target data distribution. As such, every point in the initial source distribution can be mapped to every point in the target distribution, resulting in long average paths. To this end, in this work, we tap into a non-utilized property of conditional flow-based models: the ability to design a non-trivial prior distribution. Given an input condition, such as a "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.09611","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/2502.09611/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":"2502.09611","created_at":"2026-07-05T10:13:59.980762+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.09611v1","created_at":"2026-07-05T10:13:59.980762+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.09611","created_at":"2026-07-05T10:13:59.980762+00:00"},{"alias_kind":"pith_short_12","alias_value":"35RS6SZMUAWT","created_at":"2026-07-05T10:13:59.980762+00:00"},{"alias_kind":"pith_short_16","alias_value":"35RS6SZMUAWTUCFK","created_at":"2026-07-05T10:13:59.980762+00:00"},{"alias_kind":"pith_short_8","alias_value":"35RS6SZM","created_at":"2026-07-05T10:13:59.980762+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.04165","citing_title":"CaloTrilogy: Toward a Breakthrough in One-Step, End-to-End, Physics-Guided Shower Generation for Modern Calorimeters","ref_index":87,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04092","citing_title":"Optimal Transport Flow Matching by Design","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30332","citing_title":"Colored Noise Diffusion Sampling","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23114","citing_title":"Increasing the Precision of Surrogate Models for Weak Lensing Mass Maps with Flow Matching","ref_index":33,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/35RS6SZMUAWTUCFKD525RXQYLA","json":"https://pith.science/pith/35RS6SZMUAWTUCFKD525RXQYLA.json","graph_json":"https://pith.science/api/pith-number/35RS6SZMUAWTUCFKD525RXQYLA/graph.json","events_json":"https://pith.science/api/pith-number/35RS6SZMUAWTUCFKD525RXQYLA/events.json","paper":"https://pith.science/paper/35RS6SZM"},"agent_actions":{"view_html":"https://pith.science/pith/35RS6SZMUAWTUCFKD525RXQYLA","download_json":"https://pith.science/pith/35RS6SZMUAWTUCFKD525RXQYLA.json","view_paper":"https://pith.science/paper/35RS6SZM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.09611&json=true","fetch_graph":"https://pith.science/api/pith-number/35RS6SZMUAWTUCFKD525RXQYLA/graph.json","fetch_events":"https://pith.science/api/pith-number/35RS6SZMUAWTUCFKD525RXQYLA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/35RS6SZMUAWTUCFKD525RXQYLA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/35RS6SZMUAWTUCFKD525RXQYLA/action/storage_attestation","attest_author":"https://pith.science/pith/35RS6SZMUAWTUCFKD525RXQYLA/action/author_attestation","sign_citation":"https://pith.science/pith/35RS6SZMUAWTUCFKD525RXQYLA/action/citation_signature","submit_replication":"https://pith.science/pith/35RS6SZMUAWTUCFKD525RXQYLA/action/replication_record"}},"created_at":"2026-07-05T10:13:59.980762+00:00","updated_at":"2026-07-05T10:13:59.980762+00:00"}