{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:C2VXO4EE6JR32ODIIG4YIO5P7J","short_pith_number":"pith:C2VXO4EE","schema_version":"1.0","canonical_sha256":"16ab777084f263bd386841b9843baffa473ce613f320d5791b688526e12d0fe9","source":{"kind":"arxiv","id":"2606.11025","version":1},"attestation_state":"computed","paper":{"title":"Flow-DPPO: Divergence Proximal Policy Optimization for Flow Matching Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bowen Ping, Liefeng Bo, Minnan Luo, Penghui Qi, Tianyu Pang, Xiangxin Zhou","submitted_at":"2026-06-09T15:59:57Z","abstract_excerpt":"Recent work has demonstrated that online reinforcement learning (RL) can substantially improve the quality and alignment of flow matching models for image and video generation. Methods such as Flow-GRPO and CPS cast the denoising process as a Markov Decision Process and apply PPO-style ratio clipping to enforce a trust region. However, we argue that ratio clipping is structurally ill-suited for flow models: the probability ratio between new and old policies is a noisy, single-sample estimate of the true policy divergence, leading to over-constraining in some regions of the trajectory and under"},"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":"2606.11025","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-09T15:59:57Z","cross_cats_sorted":[],"title_canon_sha256":"c35321476b2f6ae9dbd23462ba0a4ab321d238d164a50cf6eef0d13d9a9e3ae2","abstract_canon_sha256":"4047d15108e4d78d2dee5a2b0198f90faa8fbae61d4e25206beda64f9f47f2b4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-10T01:11:04.376925Z","signature_b64":"I37G9v/0MEpnd3/opQYg9cb13P2PZIL0xgQHIom5eDANacnRVEoON2ik/6YVk5Je3S8U2EypuihttaaC9EDqBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"16ab777084f263bd386841b9843baffa473ce613f320d5791b688526e12d0fe9","last_reissued_at":"2026-06-10T01:11:04.376074Z","signature_status":"signed_v1","first_computed_at":"2026-06-10T01:11:04.376074Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Flow-DPPO: Divergence Proximal Policy Optimization for Flow Matching Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bowen Ping, Liefeng Bo, Minnan Luo, Penghui Qi, Tianyu Pang, Xiangxin Zhou","submitted_at":"2026-06-09T15:59:57Z","abstract_excerpt":"Recent work has demonstrated that online reinforcement learning (RL) can substantially improve the quality and alignment of flow matching models for image and video generation. Methods such as Flow-GRPO and CPS cast the denoising process as a Markov Decision Process and apply PPO-style ratio clipping to enforce a trust region. However, we argue that ratio clipping is structurally ill-suited for flow models: the probability ratio between new and old policies is a noisy, single-sample estimate of the true policy divergence, leading to over-constraining in some regions of the trajectory and under"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.11025","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/2606.11025/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":"2606.11025","created_at":"2026-06-10T01:11:04.376230+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.11025v1","created_at":"2026-06-10T01:11:04.376230+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.11025","created_at":"2026-06-10T01:11:04.376230+00:00"},{"alias_kind":"pith_short_12","alias_value":"C2VXO4EE6JR3","created_at":"2026-06-10T01:11:04.376230+00:00"},{"alias_kind":"pith_short_16","alias_value":"C2VXO4EE6JR32ODI","created_at":"2026-06-10T01:11:04.376230+00:00"},{"alias_kind":"pith_short_8","alias_value":"C2VXO4EE","created_at":"2026-06-10T01:11:04.376230+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/C2VXO4EE6JR32ODIIG4YIO5P7J","json":"https://pith.science/pith/C2VXO4EE6JR32ODIIG4YIO5P7J.json","graph_json":"https://pith.science/api/pith-number/C2VXO4EE6JR32ODIIG4YIO5P7J/graph.json","events_json":"https://pith.science/api/pith-number/C2VXO4EE6JR32ODIIG4YIO5P7J/events.json","paper":"https://pith.science/paper/C2VXO4EE"},"agent_actions":{"view_html":"https://pith.science/pith/C2VXO4EE6JR32ODIIG4YIO5P7J","download_json":"https://pith.science/pith/C2VXO4EE6JR32ODIIG4YIO5P7J.json","view_paper":"https://pith.science/paper/C2VXO4EE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.11025&json=true","fetch_graph":"https://pith.science/api/pith-number/C2VXO4EE6JR32ODIIG4YIO5P7J/graph.json","fetch_events":"https://pith.science/api/pith-number/C2VXO4EE6JR32ODIIG4YIO5P7J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/C2VXO4EE6JR32ODIIG4YIO5P7J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/C2VXO4EE6JR32ODIIG4YIO5P7J/action/storage_attestation","attest_author":"https://pith.science/pith/C2VXO4EE6JR32ODIIG4YIO5P7J/action/author_attestation","sign_citation":"https://pith.science/pith/C2VXO4EE6JR32ODIIG4YIO5P7J/action/citation_signature","submit_replication":"https://pith.science/pith/C2VXO4EE6JR32ODIIG4YIO5P7J/action/replication_record"}},"created_at":"2026-06-10T01:11:04.376230+00:00","updated_at":"2026-06-10T01:11:04.376230+00:00"}