{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:BAOAZKMSHAEGMVLOIX4SDHVBUD","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"e61fd9ff5b55146dc80aaf37d96ad35ff461d9e794eb1fe108314e4789bd371b","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-28T11:42:33Z","title_canon_sha256":"f44620a377f93f7d887c27928c0b3760c0c042d36705abf892d2499a2bce4796"},"schema_version":"1.0","source":{"id":"2505.22257","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.22257","created_at":"2026-07-05T11:12:34Z"},{"alias_kind":"arxiv_version","alias_value":"2505.22257v2","created_at":"2026-07-05T11:12:34Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.22257","created_at":"2026-07-05T11:12:34Z"},{"alias_kind":"pith_short_12","alias_value":"BAOAZKMSHAEG","created_at":"2026-07-05T11:12:34Z"},{"alias_kind":"pith_short_16","alias_value":"BAOAZKMSHAEGMVLO","created_at":"2026-07-05T11:12:34Z"},{"alias_kind":"pith_short_8","alias_value":"BAOAZKMS","created_at":"2026-07-05T11:12:34Z"}],"graph_snapshots":[{"event_id":"sha256:4d23af4596543bbbebc4b0888c5155db2071ec94c8b60ceb06e826c34c2d92e2","target":"graph","created_at":"2026-07-05T11:12:34Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2505.22257/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We revisit Group Relative Policy Optimization (GRPO) in both on-policy and off-policy optimization regimes. Our motivation comes from recent work on off-policy Proximal Policy Optimization (PPO), which improves training stability, sampling efficiency, and memory usage. In addition, a recent analysis of GRPO suggests that estimating the advantage function with off-policy samples could be beneficial. Building on these observations, we adapt GRPO to the off-policy setting. We show that both on-policy and off-policy GRPO objectives yield an improvement in the reward. This result motivates the use ","authors_text":"Apoorva Nitsure, Brian Belgodere, Jerret Ross, Jesus Rios, Jiri Navratil, Kristjan Greenewald, Mattia Rigotti, Nicolas Dupuis, Youssef Mroueh","cross_cats":["stat.ML"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-28T11:42:33Z","title":"Revisiting Group Relative Policy Optimization: Insights into On-Policy and Off-Policy Training"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.22257","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:ad33c6020bd44b149e19ed2b57b05795f1dc8239fb8addcfff3356ce1d520ba2","target":"record","created_at":"2026-07-05T11:12:34Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"e61fd9ff5b55146dc80aaf37d96ad35ff461d9e794eb1fe108314e4789bd371b","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-28T11:42:33Z","title_canon_sha256":"f44620a377f93f7d887c27928c0b3760c0c042d36705abf892d2499a2bce4796"},"schema_version":"1.0","source":{"id":"2505.22257","kind":"arxiv","version":2}},"canonical_sha256":"081c0ca992380866556e45f9219ea1a0f44e0d78683dc9393bf4feab1943dbe3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"081c0ca992380866556e45f9219ea1a0f44e0d78683dc9393bf4feab1943dbe3","first_computed_at":"2026-07-05T11:12:34.119481Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:12:34.119481Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"7WY3S/qOmk6xs39wewkvuoS6C+x/YuCfO3ygzzmoNIf5n50HFuoSR8Df7If5s60FpYe/KpWiII9UyMDmBB4LAg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:12:34.119993Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.22257","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ad33c6020bd44b149e19ed2b57b05795f1dc8239fb8addcfff3356ce1d520ba2","sha256:4d23af4596543bbbebc4b0888c5155db2071ec94c8b60ceb06e826c34c2d92e2"],"state_sha256":"81f096eafb10c9bc82a34950cfca038fa279f6eeb1c9f079e45b27a326b82842"}