{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:BEBOFXDUDPCNHR225V643FJUS7","short_pith_number":"pith:BEBOFXDU","schema_version":"1.0","canonical_sha256":"0902e2dc741bc4d3c75aed7dcd953497ec64e00086807454a3f2e6ce610f9a5f","source":{"kind":"arxiv","id":"2502.06061","version":1},"attestation_state":"computed","paper":{"title":"Online Reward-Weighted Fine-Tuning of Flow Matching with Wasserstein Regularization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Chaoran Cheng, Chumeng Liang, Ge Liu, Jiajun Fan, Shuaike Shen, Yuxin Chen","submitted_at":"2025-02-09T22:45:15Z","abstract_excerpt":"Recent advancements in reinforcement learning (RL) have achieved great success in fine-tuning diffusion-based generative models. However, fine-tuning continuous flow-based generative models to align with arbitrary user-defined reward functions remains challenging, particularly due to issues such as policy collapse from overoptimization and the prohibitively high computational cost of likelihoods in continuous-time flows. In this paper, we propose an easy-to-use and theoretically sound RL fine-tuning method, which we term Online Reward-Weighted Conditional Flow Matching with Wasserstein-2 Regul"},"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":"2502.06061","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-09T22:45:15Z","cross_cats_sorted":["cs.AI","cs.CV","stat.ML"],"title_canon_sha256":"f16f7594e57fed67175826afbd295f3ff4914a984f065c103c15f314eafd7b49","abstract_canon_sha256":"391b80b04bb535a7e0c92f169ad9f5e6b9ab662c31e56fac3ad5b9ae47e2ee47"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:11:47.152808Z","signature_b64":"FhPwMmrTFEXU5EyUFZ8DypHSK65BkQ98mtWa+fXfqujMvjhzXUsLCNuLfD36OX3Bxf3y0MD/gzwoBpzuboWsCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0902e2dc741bc4d3c75aed7dcd953497ec64e00086807454a3f2e6ce610f9a5f","last_reissued_at":"2026-07-05T10:11:47.152337Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:11:47.152337Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Online Reward-Weighted Fine-Tuning of Flow Matching with Wasserstein Regularization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Chaoran Cheng, Chumeng Liang, Ge Liu, Jiajun Fan, Shuaike Shen, Yuxin Chen","submitted_at":"2025-02-09T22:45:15Z","abstract_excerpt":"Recent advancements in reinforcement learning (RL) have achieved great success in fine-tuning diffusion-based generative models. However, fine-tuning continuous flow-based generative models to align with arbitrary user-defined reward functions remains challenging, particularly due to issues such as policy collapse from overoptimization and the prohibitively high computational cost of likelihoods in continuous-time flows. In this paper, we propose an easy-to-use and theoretically sound RL fine-tuning method, which we term Online Reward-Weighted Conditional Flow Matching with Wasserstein-2 Regul"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.06061","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/2502.06061/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":"2502.06061","created_at":"2026-07-05T10:11:47.152392+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.06061v1","created_at":"2026-07-05T10:11:47.152392+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.06061","created_at":"2026-07-05T10:11:47.152392+00:00"},{"alias_kind":"pith_short_12","alias_value":"BEBOFXDUDPCN","created_at":"2026-07-05T10:11:47.152392+00:00"},{"alias_kind":"pith_short_16","alias_value":"BEBOFXDUDPCNHR22","created_at":"2026-07-05T10:11:47.152392+00:00"},{"alias_kind":"pith_short_8","alias_value":"BEBOFXDU","created_at":"2026-07-05T10:11:47.152392+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.27095","citing_title":"Adversarial Dual On-Policy Distillation from Expressive Teacher","ref_index":4,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BEBOFXDUDPCNHR225V643FJUS7","json":"https://pith.science/pith/BEBOFXDUDPCNHR225V643FJUS7.json","graph_json":"https://pith.science/api/pith-number/BEBOFXDUDPCNHR225V643FJUS7/graph.json","events_json":"https://pith.science/api/pith-number/BEBOFXDUDPCNHR225V643FJUS7/events.json","paper":"https://pith.science/paper/BEBOFXDU"},"agent_actions":{"view_html":"https://pith.science/pith/BEBOFXDUDPCNHR225V643FJUS7","download_json":"https://pith.science/pith/BEBOFXDUDPCNHR225V643FJUS7.json","view_paper":"https://pith.science/paper/BEBOFXDU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.06061&json=true","fetch_graph":"https://pith.science/api/pith-number/BEBOFXDUDPCNHR225V643FJUS7/graph.json","fetch_events":"https://pith.science/api/pith-number/BEBOFXDUDPCNHR225V643FJUS7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BEBOFXDUDPCNHR225V643FJUS7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BEBOFXDUDPCNHR225V643FJUS7/action/storage_attestation","attest_author":"https://pith.science/pith/BEBOFXDUDPCNHR225V643FJUS7/action/author_attestation","sign_citation":"https://pith.science/pith/BEBOFXDUDPCNHR225V643FJUS7/action/citation_signature","submit_replication":"https://pith.science/pith/BEBOFXDUDPCNHR225V643FJUS7/action/replication_record"}},"created_at":"2026-07-05T10:11:47.152392+00:00","updated_at":"2026-07-05T10:11:47.152392+00:00"}