{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2ROOKFUQ52KJIU5KI2LOLHT54Q","short_pith_number":"pith:2ROOKFUQ","schema_version":"1.0","canonical_sha256":"d45ce51690ee949453aa4696e59e7de408866501fae28aea8789b60bd5c98137","source":{"kind":"arxiv","id":"2510.09976","version":2},"attestation_state":"computed","paper":{"title":"Reinforcement Fine-Tuning of Flow-Matching Policies for Vision-Language-Action Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.RO"],"primary_cat":"cs.LG","authors_text":"Erliang Lin, Feifei Zhao, Huangrui Li, Mingyang Lyu, Ruolin Chen, Yinqian Sun, Yi Zeng","submitted_at":"2025-10-11T03:11:18Z","abstract_excerpt":"Vision-Language-Action (VLA) models such as OpenVLA, Octo, and $\\pi_0$ have shown strong generalization by leveraging large-scale demonstrations, yet their performance is still fundamentally constrained by the quality and coverage of supervised data. Reinforcement learning (RL) provides a promising path for improving and fine-tuning VLAs through online interaction. However, conventional policy gradient methods are computationally infeasible in the context of flow-matching based models due to the intractability of the importance sampling process, which requires explicit computation of policy ra"},"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":"2510.09976","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-10-11T03:11:18Z","cross_cats_sorted":["cs.RO"],"title_canon_sha256":"2a35847e86d618ace73bdcad3f1cc7e78eeb3f51bf0d2d0a68eee3bf5de8b35a","abstract_canon_sha256":"736132b540ca4f6e7f88fa80e59039d30c20ecbfff006d672ac71a3637265e83"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-26T01:15:46.448699Z","signature_b64":"l9vQzUaRziZ78S9gVDnLRh38htDlL8V5o9uW1CLRslB3MhEautBRnfQi3qfcObrAmo5O7nD0CDounmha7mR3Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d45ce51690ee949453aa4696e59e7de408866501fae28aea8789b60bd5c98137","last_reissued_at":"2026-06-26T01:15:46.448171Z","signature_status":"signed_v1","first_computed_at":"2026-06-26T01:15:46.448171Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reinforcement Fine-Tuning of Flow-Matching Policies for Vision-Language-Action Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.RO"],"primary_cat":"cs.LG","authors_text":"Erliang Lin, Feifei Zhao, Huangrui Li, Mingyang Lyu, Ruolin Chen, Yinqian Sun, Yi Zeng","submitted_at":"2025-10-11T03:11:18Z","abstract_excerpt":"Vision-Language-Action (VLA) models such as OpenVLA, Octo, and $\\pi_0$ have shown strong generalization by leveraging large-scale demonstrations, yet their performance is still fundamentally constrained by the quality and coverage of supervised data. Reinforcement learning (RL) provides a promising path for improving and fine-tuning VLAs through online interaction. However, conventional policy gradient methods are computationally infeasible in the context of flow-matching based models due to the intractability of the importance sampling process, which requires explicit computation of policy ra"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.09976","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/2510.09976/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":"2510.09976","created_at":"2026-06-26T01:15:46.448234+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.09976v2","created_at":"2026-06-26T01:15:46.448234+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.09976","created_at":"2026-06-26T01:15:46.448234+00:00"},{"alias_kind":"pith_short_12","alias_value":"2ROOKFUQ52KJ","created_at":"2026-06-26T01:15:46.448234+00:00"},{"alias_kind":"pith_short_16","alias_value":"2ROOKFUQ52KJIU5K","created_at":"2026-06-26T01:15:46.448234+00:00"},{"alias_kind":"pith_short_8","alias_value":"2ROOKFUQ","created_at":"2026-06-26T01:15:46.448234+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2602.10503","citing_title":"Towards Long-Lived Robots: Continual Learning VLA Models via Reinforcement Fine-Tuning","ref_index":48,"is_internal_anchor":true},{"citing_arxiv_id":"2604.21138","citing_title":"Navigating the Clutter: Waypoint-Based Bi-Level Planning for Multi-Robot Systems","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07381","citing_title":"Escaping the Diversity Trap in Robotic Manipulation via Anchor-Centric Adaptation","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2604.21138","citing_title":"Navigating the Clutter: Waypoint-Based Bi-Level Planning for Multi-Robot Systems","ref_index":113,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2ROOKFUQ52KJIU5KI2LOLHT54Q","json":"https://pith.science/pith/2ROOKFUQ52KJIU5KI2LOLHT54Q.json","graph_json":"https://pith.science/api/pith-number/2ROOKFUQ52KJIU5KI2LOLHT54Q/graph.json","events_json":"https://pith.science/api/pith-number/2ROOKFUQ52KJIU5KI2LOLHT54Q/events.json","paper":"https://pith.science/paper/2ROOKFUQ"},"agent_actions":{"view_html":"https://pith.science/pith/2ROOKFUQ52KJIU5KI2LOLHT54Q","download_json":"https://pith.science/pith/2ROOKFUQ52KJIU5KI2LOLHT54Q.json","view_paper":"https://pith.science/paper/2ROOKFUQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.09976&json=true","fetch_graph":"https://pith.science/api/pith-number/2ROOKFUQ52KJIU5KI2LOLHT54Q/graph.json","fetch_events":"https://pith.science/api/pith-number/2ROOKFUQ52KJIU5KI2LOLHT54Q/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2ROOKFUQ52KJIU5KI2LOLHT54Q/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2ROOKFUQ52KJIU5KI2LOLHT54Q/action/storage_attestation","attest_author":"https://pith.science/pith/2ROOKFUQ52KJIU5KI2LOLHT54Q/action/author_attestation","sign_citation":"https://pith.science/pith/2ROOKFUQ52KJIU5KI2LOLHT54Q/action/citation_signature","submit_replication":"https://pith.science/pith/2ROOKFUQ52KJIU5KI2LOLHT54Q/action/replication_record"}},"created_at":"2026-06-26T01:15:46.448234+00:00","updated_at":"2026-06-26T01:15:46.448234+00:00"}