{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:CU2IPJNYP4JID5VBP3QMLAMOZM","short_pith_number":"pith:CU2IPJNY","schema_version":"1.0","canonical_sha256":"153487a5b87f1281f6a17ee0c5818ecb3f0a656cb52a2413c496bb47601acab7","source":{"kind":"arxiv","id":"2605.21557","version":1},"attestation_state":"computed","paper":{"title":"Scalable On-Policy Reinforcement Learning via Adaptive Batch Scaling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"stat.ML","authors_text":"Jongchan Park","submitted_at":"2026-05-20T13:46:22Z","abstract_excerpt":"Conventional wisdom holds that large-batch training is fundamentally incompatible with Reinforcement Learning (RL) - beyond a modest threshold, increasing batch sizes typically yields diminishing returns or performance degradation due to the inherent non-stationarity of the data distribution. We challenge this view by observing that non-stationarity is not a fixed property of RL, but evolves throughout training: early stages exhibit rapid behavioral shifts that demand small batches for plasticity, whereas late stages approach a quasi-stationary regime where large batches enable precise converg"},"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":"2605.21557","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2026-05-20T13:46:22Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"539be7504c6a3b01f872918005e53a0f847c006e739455e7870b362a8c6ad34e","abstract_canon_sha256":"91d4d2d5322547fb031f3df2e193c64c01cf0d598dbb2929ec7f2a57a3aba1e7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-22T00:02:28.410596Z","signature_b64":"/QdExPrIAmkKeo2z64N59hKr5CcfxTJY7UbWTVjEAXGRgpuLl9akk6n0BhrANvuSEXkeOXoj3LrMdK5rwns9Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"153487a5b87f1281f6a17ee0c5818ecb3f0a656cb52a2413c496bb47601acab7","last_reissued_at":"2026-05-22T00:02:28.410043Z","signature_status":"signed_v1","first_computed_at":"2026-05-22T00:02:28.410043Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scalable On-Policy Reinforcement Learning via Adaptive Batch Scaling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"stat.ML","authors_text":"Jongchan Park","submitted_at":"2026-05-20T13:46:22Z","abstract_excerpt":"Conventional wisdom holds that large-batch training is fundamentally incompatible with Reinforcement Learning (RL) - beyond a modest threshold, increasing batch sizes typically yields diminishing returns or performance degradation due to the inherent non-stationarity of the data distribution. We challenge this view by observing that non-stationarity is not a fixed property of RL, but evolves throughout training: early stages exhibit rapid behavioral shifts that demand small batches for plasticity, whereas late stages approach a quasi-stationary regime where large batches enable precise converg"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.21557","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/2605.21557/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":"2605.21557","created_at":"2026-05-22T00:02:28.410122+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.21557v1","created_at":"2026-05-22T00:02:28.410122+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.21557","created_at":"2026-05-22T00:02:28.410122+00:00"},{"alias_kind":"pith_short_12","alias_value":"CU2IPJNYP4JI","created_at":"2026-05-22T00:02:28.410122+00:00"},{"alias_kind":"pith_short_16","alias_value":"CU2IPJNYP4JID5VB","created_at":"2026-05-22T00:02:28.410122+00:00"},{"alias_kind":"pith_short_8","alias_value":"CU2IPJNY","created_at":"2026-05-22T00:02:28.410122+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/CU2IPJNYP4JID5VBP3QMLAMOZM","json":"https://pith.science/pith/CU2IPJNYP4JID5VBP3QMLAMOZM.json","graph_json":"https://pith.science/api/pith-number/CU2IPJNYP4JID5VBP3QMLAMOZM/graph.json","events_json":"https://pith.science/api/pith-number/CU2IPJNYP4JID5VBP3QMLAMOZM/events.json","paper":"https://pith.science/paper/CU2IPJNY"},"agent_actions":{"view_html":"https://pith.science/pith/CU2IPJNYP4JID5VBP3QMLAMOZM","download_json":"https://pith.science/pith/CU2IPJNYP4JID5VBP3QMLAMOZM.json","view_paper":"https://pith.science/paper/CU2IPJNY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.21557&json=true","fetch_graph":"https://pith.science/api/pith-number/CU2IPJNYP4JID5VBP3QMLAMOZM/graph.json","fetch_events":"https://pith.science/api/pith-number/CU2IPJNYP4JID5VBP3QMLAMOZM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CU2IPJNYP4JID5VBP3QMLAMOZM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CU2IPJNYP4JID5VBP3QMLAMOZM/action/storage_attestation","attest_author":"https://pith.science/pith/CU2IPJNYP4JID5VBP3QMLAMOZM/action/author_attestation","sign_citation":"https://pith.science/pith/CU2IPJNYP4JID5VBP3QMLAMOZM/action/citation_signature","submit_replication":"https://pith.science/pith/CU2IPJNYP4JID5VBP3QMLAMOZM/action/replication_record"}},"created_at":"2026-05-22T00:02:28.410122+00:00","updated_at":"2026-05-22T00:02:28.410122+00:00"}