{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:VOWDRF6XRAOWX3FIFQE4A4XW5Q","short_pith_number":"pith:VOWDRF6X","schema_version":"1.0","canonical_sha256":"abac3897d7881d6beca82c09c072f6ec37b8fe170daca1cb26bfb28f3a8b1e1b","source":{"kind":"arxiv","id":"2605.13276","version":2},"attestation_state":"computed","paper":{"title":"D-VLA: A High-Concurrency Distributed Asynchronous Reinforcement Learning Framework for Vision-Language-Action Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"D-VLA decouples simulation and optimization planes to deliver linear speedup for trillion-parameter vision-language-action models in distributed RL.","cross_cats":["cs.RO"],"primary_cat":"cs.AI","authors_text":"Haodong Yue, Haoran Sun, Junwu Xiong, Luqiao Wang, Shuai Di, Wen Huang, Xiaolong Xiang, Yicheng Gong, Yongjian Guo, Yucheng Guo, Zhen Sun, Zhong Guan","submitted_at":"2026-05-13T09:54:31Z","abstract_excerpt":"The rapid evolution of Embodied AI has enabled Vision-Language-Action (VLA) models to excel in multimodal perception and task execution. However, applying Reinforcement Learning (RL) to these massive models in large-scale distributed environments faces severe systemic bottlenecks, primarily due to the resource conflict between high-fidelity physical simulation and the intensive VRAM/bandwidth demands of deep learning. This conflict often leaves overall throughput constrained by execution-phase inefficiencies. To address these challenges, we propose D-VLA, a high-concurrency, low-latency distri"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2605.13276","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-13T09:54:31Z","cross_cats_sorted":["cs.RO"],"title_canon_sha256":"e67c7c00cd97c2d5db41021910748ae0e91d5ee597a790f73cfb8c8cd9674275","abstract_canon_sha256":"30d65e77cdcf7bba6dd8d9414bea4b24b3463382a50313501e9aca6ca83013df"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:44:49.236719Z","signature_b64":"F1j6TLUB2CJ0GK6rKzrnY94/5YqAjK32OZBrAdEuxnO60NvfO8v2ewvRmL3Ih4rdy1uA6eY4Q1X4xyNtqJ1JDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"abac3897d7881d6beca82c09c072f6ec37b8fe170daca1cb26bfb28f3a8b1e1b","last_reissued_at":"2026-05-18T02:44:49.236278Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:44:49.236278Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"D-VLA: A High-Concurrency Distributed Asynchronous Reinforcement Learning Framework for Vision-Language-Action Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"D-VLA decouples simulation and optimization planes to deliver linear speedup for trillion-parameter vision-language-action models in distributed RL.","cross_cats":["cs.RO"],"primary_cat":"cs.AI","authors_text":"Haodong Yue, Haoran Sun, Junwu Xiong, Luqiao Wang, Shuai Di, Wen Huang, Xiaolong Xiang, Yicheng Gong, Yongjian Guo, Yucheng Guo, Zhen Sun, Zhong Guan","submitted_at":"2026-05-13T09:54:31Z","abstract_excerpt":"The rapid evolution of Embodied AI has enabled Vision-Language-Action (VLA) models to excel in multimodal perception and task execution. However, applying Reinforcement Learning (RL) to these massive models in large-scale distributed environments faces severe systemic bottlenecks, primarily due to the resource conflict between high-fidelity physical simulation and the intensive VRAM/bandwidth demands of deep learning. This conflict often leaves overall throughput constrained by execution-phase inefficiencies. To address these challenges, we propose D-VLA, a high-concurrency, low-latency distri"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments on benchmarks like LIBERO show that D-VLA significantly outperforms mainstream RL frameworks in throughput and sampling efficiency for billion-parameter VLA models. In trillion-parameter scalability tests, our framework maintains exceptional stability and linear speedup.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that physically isolating high-frequency training data from low-frequency weight control via Plane Decoupling, combined with the swimlane pipeline, eliminates interference and achieves linear speedup without new bottlenecks in real distributed hardware.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"D-VLA uses plane decoupling and a swimlane pipeline to deliver higher throughput and linear speedup than prior RL frameworks when training billion- and trillion-parameter VLA models on benchmarks like LIBERO.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"D-VLA decouples simulation and optimization planes to deliver linear speedup for trillion-parameter vision-language-action models in distributed RL.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e35cd56a639a7d8617b0e5592108c81a8c8d16ee6c4cbf728b1ceffb8fed0418"},"source":{"id":"2605.13276","kind":"arxiv","version":2},"verdict":{"id":"add5ac7b-c124-48f1-b9d5-7aa3dc4c2ded","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T06:00:28.408357Z","strongest_claim":"Experiments on benchmarks like LIBERO show that D-VLA significantly outperforms mainstream RL frameworks in throughput and sampling efficiency for billion-parameter VLA models. In trillion-parameter scalability tests, our framework maintains exceptional stability and linear speedup.","one_line_summary":"D-VLA uses plane decoupling and a swimlane pipeline to deliver higher throughput and linear speedup than prior RL frameworks when training billion- and trillion-parameter VLA models on benchmarks like LIBERO.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that physically isolating high-frequency training data from low-frequency weight control via Plane Decoupling, combined with the swimlane pipeline, eliminates interference and achieves linear speedup without new bottlenecks in real distributed hardware.","pith_extraction_headline":"D-VLA decouples simulation and optimization planes to deliver linear speedup for trillion-parameter vision-language-action models in distributed RL."},"references":{"count":28,"sample":[{"doi":"","year":2023,"title":"Sanketi, Grecia Salazar, Michael S","work_id":"3981437e-1b5e-41d1-8d02-40f4186d18ff","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"$\\pi_0$: A Vision-Language-Action Flow Model for General Robot Control","work_id":"f790abdc-a796-482f-a40d-f8ee035ecfc2","ref_index":2,"cited_arxiv_id":"2410.24164","is_internal_anchor":true},{"doi":"","year":2025,"title":"Gemini Robotics 1.5: Pushing the Frontier of Generalist Robots with Advanced Embodied Reasoning, Thinking, and Motion Transfer","work_id":"ffe7b7f4-997b-4957-b253-03cfbebf6f9a","ref_index":3,"cited_arxiv_id":"2510.03342","is_internal_anchor":true},{"doi":"","year":2024,"title":"OpenVLA: An Open-Source Vision-Language-Action Model","work_id":"3e7e65c5-5aed-4fe9-8414-2092bcb31cc7","ref_index":4,"cited_arxiv_id":"2406.09246","is_internal_anchor":true},{"doi":"","year":2025,"title":"Smolvla: A vision-language-action model for affordable and efficient robotics, 2025","work_id":"d3061a03-48d4-4b75-9b97-92f97e21eded","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":28,"snapshot_sha256":"67e4e266a72abc87ac8ec898be3d8373fc1ee17157322c019a0e31af20cb1608","internal_anchors":13},"formal_canon":{"evidence_count":2,"snapshot_sha256":"521c7f2db1f868d8bf55b06c2e2e5ceb7891412dae076442d39fe0fdf562105f"},"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":"2605.13276","created_at":"2026-05-18T02:44:49.236353+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.13276v2","created_at":"2026-05-18T02:44:49.236353+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13276","created_at":"2026-05-18T02:44:49.236353+00:00"},{"alias_kind":"pith_short_12","alias_value":"VOWDRF6XRAOW","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"VOWDRF6XRAOWX3FI","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"VOWDRF6X","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VOWDRF6XRAOWX3FIFQE4A4XW5Q","json":"https://pith.science/pith/VOWDRF6XRAOWX3FIFQE4A4XW5Q.json","graph_json":"https://pith.science/api/pith-number/VOWDRF6XRAOWX3FIFQE4A4XW5Q/graph.json","events_json":"https://pith.science/api/pith-number/VOWDRF6XRAOWX3FIFQE4A4XW5Q/events.json","paper":"https://pith.science/paper/VOWDRF6X"},"agent_actions":{"view_html":"https://pith.science/pith/VOWDRF6XRAOWX3FIFQE4A4XW5Q","download_json":"https://pith.science/pith/VOWDRF6XRAOWX3FIFQE4A4XW5Q.json","view_paper":"https://pith.science/paper/VOWDRF6X","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.13276&json=true","fetch_graph":"https://pith.science/api/pith-number/VOWDRF6XRAOWX3FIFQE4A4XW5Q/graph.json","fetch_events":"https://pith.science/api/pith-number/VOWDRF6XRAOWX3FIFQE4A4XW5Q/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VOWDRF6XRAOWX3FIFQE4A4XW5Q/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VOWDRF6XRAOWX3FIFQE4A4XW5Q/action/storage_attestation","attest_author":"https://pith.science/pith/VOWDRF6XRAOWX3FIFQE4A4XW5Q/action/author_attestation","sign_citation":"https://pith.science/pith/VOWDRF6XRAOWX3FIFQE4A4XW5Q/action/citation_signature","submit_replication":"https://pith.science/pith/VOWDRF6XRAOWX3FIFQE4A4XW5Q/action/replication_record"}},"created_at":"2026-05-18T02:44:49.236353+00:00","updated_at":"2026-05-18T02:44:49.236353+00:00"}