{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:YJIDXWRMN7FRVEB5FGCMIH4U7I","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":"5010ee1055120649df60fb8ca9375faafc310860a8766a6a3e1d0788ab35daea","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-17T17:58:53Z","title_canon_sha256":"8b5733551a143fc67f90e4a2c820e5a96b37f25c58ba9ece68b8cbb4e101999e"},"schema_version":"1.0","source":{"id":"2605.17570","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.17570","created_at":"2026-05-20T00:04:46Z"},{"alias_kind":"arxiv_version","alias_value":"2605.17570v1","created_at":"2026-05-20T00:04:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.17570","created_at":"2026-05-20T00:04:46Z"},{"alias_kind":"pith_short_12","alias_value":"YJIDXWRMN7FR","created_at":"2026-05-20T00:04:46Z"},{"alias_kind":"pith_short_16","alias_value":"YJIDXWRMN7FRVEB5","created_at":"2026-05-20T00:04:46Z"},{"alias_kind":"pith_short_8","alias_value":"YJIDXWRM","created_at":"2026-05-20T00:04:46Z"}],"graph_snapshots":[{"event_id":"sha256:d224721d41645635ce6b59c47fc36d208de6e878a2f72bdc933eb0380bd82071","target":"graph","created_at":"2026-05-20T00:04:46Z","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":[{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-19T21:33:23.595403Z","status":"skipped","version":"1.0.0"},{"findings_count":0,"name":"claim_evidence","ran_at":"2026-05-19T21:21:57.526963Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.17570/integrity.json","findings":[],"snapshot_sha256":"b5b6bb21d89c7c8a60ebfec6be41e2b186e946c1dbac402bd568662b1ca01dfe","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Group Relative Policy Optimization (GRPO) has been a key driver of recent progress in reinforcement learning with verifiable rewards (RLVR) for large language models, but it is typically trained in a low-staleness, near-on-policy regime that incurs substantial system overhead. We ask a simple question: How off-policy can GRPO be? We show that GRPO-style algorithms can tolerate substantially larger rollout staleness than previously assumed, and propose Mu-GRPO, an RL training framework that organizes training into a small number (e.g., four) of large sequential generation-optimization stages. T","authors_text":"Chen Wei, Minghao Tian, Yunfei Xie","cross_cats":["cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-17T17:58:53Z","title":"How Off-Policy Can GRPO Be? Mu-GRPO for Efficient LLM Reinforcement Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.17570","kind":"arxiv","version":1},"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":"sha256:368c1d927438e593a31c59fc8a3e44863b4f91f13bc31338eab70d0b387fb075","target":"record","created_at":"2026-05-20T00:04:46Z","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":"5010ee1055120649df60fb8ca9375faafc310860a8766a6a3e1d0788ab35daea","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-17T17:58:53Z","title_canon_sha256":"8b5733551a143fc67f90e4a2c820e5a96b37f25c58ba9ece68b8cbb4e101999e"},"schema_version":"1.0","source":{"id":"2605.17570","kind":"arxiv","version":1}},"canonical_sha256":"c2503bda2c6fcb1a903d2984c41f94fa263dbf3c856e769c2e3cddec6103388c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c2503bda2c6fcb1a903d2984c41f94fa263dbf3c856e769c2e3cddec6103388c","first_computed_at":"2026-05-20T00:04:46.483896Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:04:46.483896Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"9vJXqW61mmzd6RUP4xXFUMyVZCKrR4/yToWDvmhHPlRRsB2Yulx4O1BldAHRwlggdwnw+KvU8P+jniKlhMtGAQ==","signature_status":"signed_v1","signed_at":"2026-05-20T00:04:46.484668Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.17570","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:368c1d927438e593a31c59fc8a3e44863b4f91f13bc31338eab70d0b387fb075","sha256:d224721d41645635ce6b59c47fc36d208de6e878a2f72bdc933eb0380bd82071"],"state_sha256":"42048f46cf92326f8aa598a0bfb3bf5c676dcc6cd6d0c4a7bf32a8947f376686"}