{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:KNMWXFX2VTKC7ID7SPD4P6GUYV","short_pith_number":"pith:KNMWXFX2","schema_version":"1.0","canonical_sha256":"53596b96faacd42fa07f93c7c7f8d4c5675dbd938cfa5bf9825439b6323610e5","source":{"kind":"arxiv","id":"2011.12719","version":4},"attestation_state":"computed","paper":{"title":"RLlib Flow: Distributed Reinforcement Learning is a Dataflow Problem","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.LG","authors_text":"Eric Liang, Ion Stoica, Joseph E. Gonzalez, Michael Luo, Sven Mika, Zhanghao Wu","submitted_at":"2020-11-25T13:28:16Z","abstract_excerpt":"Researchers and practitioners in the field of reinforcement learning (RL) frequently leverage parallel computation, which has led to a plethora of new algorithms and systems in the last few years. In this paper, we re-examine the challenges posed by distributed RL and try to view it through the lens of an old idea: distributed dataflow. We show that viewing RL as a dataflow problem leads to highly composable and performant implementations. We propose RLlib Flow, a hybrid actor-dataflow programming model for distributed RL, and validate its practicality by porting the full suite of algorithms i"},"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":"2011.12719","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2020-11-25T13:28:16Z","cross_cats_sorted":["cs.DC"],"title_canon_sha256":"980166acc8817ea8e056c2d9165d78304628f4eab0c581b88552ecf6c00bdb5b","abstract_canon_sha256":"5f56e5ab32f3710b0f25e7a41cd18411adf3db4761ba2b139bcc54bb7a48f0e4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:26:57.969271Z","signature_b64":"VZlHcnc7NS1nd7Koh3bt4w+9gV3k13xwi0s+r1vd+7ggPnp4VKPKlDxX3pUYwtlrxs3WbyBkUQjQL6Yq2zo0BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"53596b96faacd42fa07f93c7c7f8d4c5675dbd938cfa5bf9825439b6323610e5","last_reissued_at":"2026-07-05T03:26:57.968742Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:26:57.968742Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RLlib Flow: Distributed Reinforcement Learning is a Dataflow Problem","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.LG","authors_text":"Eric Liang, Ion Stoica, Joseph E. Gonzalez, Michael Luo, Sven Mika, Zhanghao Wu","submitted_at":"2020-11-25T13:28:16Z","abstract_excerpt":"Researchers and practitioners in the field of reinforcement learning (RL) frequently leverage parallel computation, which has led to a plethora of new algorithms and systems in the last few years. In this paper, we re-examine the challenges posed by distributed RL and try to view it through the lens of an old idea: distributed dataflow. We show that viewing RL as a dataflow problem leads to highly composable and performant implementations. We propose RLlib Flow, a hybrid actor-dataflow programming model for distributed RL, and validate its practicality by porting the full suite of algorithms i"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2011.12719","kind":"arxiv","version":4},"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/2011.12719/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":"2011.12719","created_at":"2026-07-05T03:26:57.968807+00:00"},{"alias_kind":"arxiv_version","alias_value":"2011.12719v4","created_at":"2026-07-05T03:26:57.968807+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2011.12719","created_at":"2026-07-05T03:26:57.968807+00:00"},{"alias_kind":"pith_short_12","alias_value":"KNMWXFX2VTKC","created_at":"2026-07-05T03:26:57.968807+00:00"},{"alias_kind":"pith_short_16","alias_value":"KNMWXFX2VTKC7ID7","created_at":"2026-07-05T03:26:57.968807+00:00"},{"alias_kind":"pith_short_8","alias_value":"KNMWXFX2","created_at":"2026-07-05T03:26:57.968807+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01415","citing_title":"The Rollout Infrastructure Tax in Coding-Agent Reinforcement Learning","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17511","citing_title":"MagicSim: A Unified Infrastructure for Executable Embodied Interaction","ref_index":124,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KNMWXFX2VTKC7ID7SPD4P6GUYV","json":"https://pith.science/pith/KNMWXFX2VTKC7ID7SPD4P6GUYV.json","graph_json":"https://pith.science/api/pith-number/KNMWXFX2VTKC7ID7SPD4P6GUYV/graph.json","events_json":"https://pith.science/api/pith-number/KNMWXFX2VTKC7ID7SPD4P6GUYV/events.json","paper":"https://pith.science/paper/KNMWXFX2"},"agent_actions":{"view_html":"https://pith.science/pith/KNMWXFX2VTKC7ID7SPD4P6GUYV","download_json":"https://pith.science/pith/KNMWXFX2VTKC7ID7SPD4P6GUYV.json","view_paper":"https://pith.science/paper/KNMWXFX2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2011.12719&json=true","fetch_graph":"https://pith.science/api/pith-number/KNMWXFX2VTKC7ID7SPD4P6GUYV/graph.json","fetch_events":"https://pith.science/api/pith-number/KNMWXFX2VTKC7ID7SPD4P6GUYV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KNMWXFX2VTKC7ID7SPD4P6GUYV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KNMWXFX2VTKC7ID7SPD4P6GUYV/action/storage_attestation","attest_author":"https://pith.science/pith/KNMWXFX2VTKC7ID7SPD4P6GUYV/action/author_attestation","sign_citation":"https://pith.science/pith/KNMWXFX2VTKC7ID7SPD4P6GUYV/action/citation_signature","submit_replication":"https://pith.science/pith/KNMWXFX2VTKC7ID7SPD4P6GUYV/action/replication_record"}},"created_at":"2026-07-05T03:26:57.968807+00:00","updated_at":"2026-07-05T03:26:57.968807+00:00"}