{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JTRISRMIVRMQX5FVCFONSCRAKT","short_pith_number":"pith:JTRISRMI","schema_version":"1.0","canonical_sha256":"4ce2894588ac590bf4b5115cd90a2054fc09d8c2d83f35223f52dad251433278","source":{"kind":"arxiv","id":"2506.12811","version":1},"attestation_state":"computed","paper":{"title":"Flow-Based Policy for Online Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Fuchun Sun, Jiafeng Xu, Lei Lv, Tao Kong, Xiao Ma, Yu Luo, Yunfei Li","submitted_at":"2025-06-15T10:53:35Z","abstract_excerpt":"We present \\textbf{FlowRL}, a novel framework for online reinforcement learning that integrates flow-based policy representation with Wasserstein-2-regularized optimization. We argue that in addition to training signals, enhancing the expressiveness of the policy class is crucial for the performance gains in RL. Flow-based generative models offer such potential, excelling at capturing complex, multimodal action distributions. However, their direct application in online RL is challenging due to a fundamental objective mismatch: standard flow training optimizes for static data imitation, while R"},"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":"2506.12811","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-15T10:53:35Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c33b704796eef14919c56f2050870a3a0b6229d17e4ef3893554027d5498b968","abstract_canon_sha256":"331b1da967c7eaa27564e6c695ae797782800c77cff0fcf744b8f91dc567acdf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:22:00.640255Z","signature_b64":"9ux7FGgp5KKBW21e6brxlwf3Aasl0tww7xSQBASSRZeP25cJ91ih90Pjv1HP5Ins2LMhVbN1OgmiqdFQCywOAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4ce2894588ac590bf4b5115cd90a2054fc09d8c2d83f35223f52dad251433278","last_reissued_at":"2026-07-05T11:22:00.639785Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:22:00.639785Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Flow-Based Policy for Online Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Fuchun Sun, Jiafeng Xu, Lei Lv, Tao Kong, Xiao Ma, Yu Luo, Yunfei Li","submitted_at":"2025-06-15T10:53:35Z","abstract_excerpt":"We present \\textbf{FlowRL}, a novel framework for online reinforcement learning that integrates flow-based policy representation with Wasserstein-2-regularized optimization. We argue that in addition to training signals, enhancing the expressiveness of the policy class is crucial for the performance gains in RL. Flow-based generative models offer such potential, excelling at capturing complex, multimodal action distributions. However, their direct application in online RL is challenging due to a fundamental objective mismatch: standard flow training optimizes for static data imitation, while R"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.12811","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/2506.12811/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":"2506.12811","created_at":"2026-07-05T11:22:00.639842+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.12811v1","created_at":"2026-07-05T11:22:00.639842+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.12811","created_at":"2026-07-05T11:22:00.639842+00:00"},{"alias_kind":"pith_short_12","alias_value":"JTRISRMIVRMQ","created_at":"2026-07-05T11:22:00.639842+00:00"},{"alias_kind":"pith_short_16","alias_value":"JTRISRMIVRMQX5FV","created_at":"2026-07-05T11:22:00.639842+00:00"},{"alias_kind":"pith_short_8","alias_value":"JTRISRMI","created_at":"2026-07-05T11:22:00.639842+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21086","citing_title":"ReFPO: Reflow Regularization for Flow Matching Policy Gradients","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11087","citing_title":"Test-Time Gradient Guidance of Flow Policies in Reinforcement Learning","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29820","citing_title":"Dual-Flow Reinforcement Learning with State-Aware Exploration","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2603.04333","citing_title":"What Does Flow Matching Bring To TD Learning?","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01356","citing_title":"Model-Based Proactive Cost Generation for Learning Safe Policies Offline with Limited Violation Data","ref_index":37,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JTRISRMIVRMQX5FVCFONSCRAKT","json":"https://pith.science/pith/JTRISRMIVRMQX5FVCFONSCRAKT.json","graph_json":"https://pith.science/api/pith-number/JTRISRMIVRMQX5FVCFONSCRAKT/graph.json","events_json":"https://pith.science/api/pith-number/JTRISRMIVRMQX5FVCFONSCRAKT/events.json","paper":"https://pith.science/paper/JTRISRMI"},"agent_actions":{"view_html":"https://pith.science/pith/JTRISRMIVRMQX5FVCFONSCRAKT","download_json":"https://pith.science/pith/JTRISRMIVRMQX5FVCFONSCRAKT.json","view_paper":"https://pith.science/paper/JTRISRMI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.12811&json=true","fetch_graph":"https://pith.science/api/pith-number/JTRISRMIVRMQX5FVCFONSCRAKT/graph.json","fetch_events":"https://pith.science/api/pith-number/JTRISRMIVRMQX5FVCFONSCRAKT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JTRISRMIVRMQX5FVCFONSCRAKT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JTRISRMIVRMQX5FVCFONSCRAKT/action/storage_attestation","attest_author":"https://pith.science/pith/JTRISRMIVRMQX5FVCFONSCRAKT/action/author_attestation","sign_citation":"https://pith.science/pith/JTRISRMIVRMQX5FVCFONSCRAKT/action/citation_signature","submit_replication":"https://pith.science/pith/JTRISRMIVRMQX5FVCFONSCRAKT/action/replication_record"}},"created_at":"2026-07-05T11:22:00.639842+00:00","updated_at":"2026-07-05T11:22:00.639842+00:00"}