{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:3IZOUWCOH56UNY56EWZPMJU3YK","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":"e194dd37954f066a527aa0552404fb06e8bff06f7413b85038d88a665ebf4ae0","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-01T13:10:20Z","title_canon_sha256":"a2cd9e386953538d7261dc287c6ed1c4837d2d55f57212d33b2075867aa6df06"},"schema_version":"1.0","source":{"id":"2604.00860","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.00860","created_at":"2026-06-04T01:08:49Z"},{"alias_kind":"arxiv_version","alias_value":"2604.00860v3","created_at":"2026-06-04T01:08:49Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.00860","created_at":"2026-06-04T01:08:49Z"},{"alias_kind":"pith_short_12","alias_value":"3IZOUWCOH56U","created_at":"2026-06-04T01:08:49Z"},{"alias_kind":"pith_short_16","alias_value":"3IZOUWCOH56UNY56","created_at":"2026-06-04T01:08:49Z"},{"alias_kind":"pith_short_8","alias_value":"3IZOUWCO","created_at":"2026-06-04T01:08:49Z"}],"graph_snapshots":[{"event_id":"sha256:927845dba9896abb9f27152003401721c192befec3a93b48a554094838dd7123","target":"graph","created_at":"2026-06-04T01:08:49Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"we prove this temporal objective is perfectly aligned with maximizing final task performance... PIPO performs ascent on the PIRL objective in expectation"},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"that retrospective verification against a sliding-window historical baseline can reliably identify genuine policy improvement without bias from window size, data distribution, or the specific choice of baseline statistics"},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"PIRL maximizes cumulative policy improvement across iterations instead of surrogate rewards and is proven aligned with final performance; PIPO implements it via retrospective verification for stable closed-loop optimization."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Reinforcement learning for language models can be made self-correcting by directly maximizing cumulative verified policy improvement across iterations."}],"snapshot_sha256":"0771c5562a4e0c43d8a94a8533c81b3caba9b523c33214326d48d0abe76fd021"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2604.00860/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement Learning with Verifiable Rewards (RLVR) has become a central post-training paradigm for improving the reasoning capabilities of large language models. Yet existing methods share a common blind spot: they optimize policies based on instantaneous group-level or batch-level statistics without ever verifying whether the resulting update actually improved the model. This open-loop design -- updating in isolation at each step, guided only by within-group (batch) reward signals -- means optimization can drift or collapse with no mechanism to detect and correct these failures. We argue t","authors_text":"Deqing Wang, Haoyi Zhou, Huaiyang Wang, Jianxin Li, Xiaojie Li, Yaodong Yang, Yikun Ban, Zixuan Huang","cross_cats":[],"headline":"Reinforcement learning for language models can be made self-correcting by directly maximizing cumulative verified policy improvement across iterations.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-01T13:10:20Z","title":"Policy Improvement Reinforcement Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.00860","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-13T22:44:54.577345Z","id":"8f1b7e82-9cce-4ad9-9397-9b33dc804b4b","model_set":{"reader":"grok-4.3"},"one_line_summary":"PIRL maximizes cumulative policy improvement across iterations instead of surrogate rewards and is proven aligned with final performance; PIPO implements it via retrospective verification for stable closed-loop optimization.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Reinforcement learning for language models can be made self-correcting by directly maximizing cumulative verified policy improvement across iterations.","strongest_claim":"we prove this temporal objective is perfectly aligned with maximizing final task performance... PIPO performs ascent on the PIRL objective in expectation","weakest_assumption":"that retrospective verification against a sliding-window historical baseline can reliably identify genuine policy improvement without bias from window size, data distribution, or the specific choice of baseline statistics"}},"verdict_id":"8f1b7e82-9cce-4ad9-9397-9b33dc804b4b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:6ad17a862d7a55da20276445656fb120d2375a1310469ef6c184ea16aae60eef","target":"record","created_at":"2026-06-04T01:08:49Z","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":"e194dd37954f066a527aa0552404fb06e8bff06f7413b85038d88a665ebf4ae0","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-01T13:10:20Z","title_canon_sha256":"a2cd9e386953538d7261dc287c6ed1c4837d2d55f57212d33b2075867aa6df06"},"schema_version":"1.0","source":{"id":"2604.00860","kind":"arxiv","version":3}},"canonical_sha256":"da32ea584e3f7d46e3be25b2f6269bc2b9b5a1e5d49c2155e2d81a4205ba76e0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"da32ea584e3f7d46e3be25b2f6269bc2b9b5a1e5d49c2155e2d81a4205ba76e0","first_computed_at":"2026-06-04T01:08:49.541977Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-04T01:08:49.541977Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"/ZPvViTAMq+70ynOeLNlv9dTlc4YjTmxyAHzI0NO4cz6yQW9Ai4BmZXn+DvM6us1SSjkpga6iT1d3xFrSuY+Cg==","signature_status":"signed_v1","signed_at":"2026-06-04T01:08:49.542540Z","signed_message":"canonical_sha256_bytes"},"source_id":"2604.00860","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6ad17a862d7a55da20276445656fb120d2375a1310469ef6c184ea16aae60eef","sha256:927845dba9896abb9f27152003401721c192befec3a93b48a554094838dd7123"],"state_sha256":"fc2fcd6516c865bb0b2e76752f9e033415199e2f06ce9eaebb3b61de8cb1c140"}