{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:3IZOUWCOH56UNY56EWZPMJU3YK","short_pith_number":"pith:3IZOUWCO","schema_version":"1.0","canonical_sha256":"da32ea584e3f7d46e3be25b2f6269bc2b9b5a1e5d49c2155e2d81a4205ba76e0","source":{"kind":"arxiv","id":"2604.00860","version":3},"attestation_state":"computed","paper":{"title":"Policy Improvement Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Reinforcement learning for language models can be made self-correcting by directly maximizing cumulative verified policy improvement across iterations.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Deqing Wang, Haoyi Zhou, Huaiyang Wang, Jianxin Li, Xiaojie Li, Yaodong Yang, Yikun Ban, Zixuan Huang","submitted_at":"2026-04-01T13:10:20Z","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"},"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":"2604.00860","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-01T13:10:20Z","cross_cats_sorted":[],"title_canon_sha256":"a2cd9e386953538d7261dc287c6ed1c4837d2d55f57212d33b2075867aa6df06","abstract_canon_sha256":"e194dd37954f066a527aa0552404fb06e8bff06f7413b85038d88a665ebf4ae0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-04T01:08:49.542540Z","signature_b64":"/ZPvViTAMq+70ynOeLNlv9dTlc4YjTmxyAHzI0NO4cz6yQW9Ai4BmZXn+DvM6us1SSjkpga6iT1d3xFrSuY+Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"da32ea584e3f7d46e3be25b2f6269bc2b9b5a1e5d49c2155e2d81a4205ba76e0","last_reissued_at":"2026-06-04T01:08:49.541977Z","signature_status":"signed_v1","first_computed_at":"2026-06-04T01:08:49.541977Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Policy Improvement Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Reinforcement learning for language models can be made self-correcting by directly maximizing cumulative verified policy improvement across iterations.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Deqing Wang, Haoyi Zhou, Huaiyang Wang, Jianxin Li, Xiaojie Li, Yaodong Yang, Yikun Ban, Zixuan Huang","submitted_at":"2026-04-01T13:10:20Z","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"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"we prove this temporal objective is perfectly aligned with maximizing final task performance... PIPO performs ascent on the PIRL objective in expectation","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","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","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","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.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Reinforcement learning for language models can be made self-correcting by directly maximizing cumulative verified policy improvement across iterations.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"0771c5562a4e0c43d8a94a8533c81b3caba9b523c33214326d48d0abe76fd021"},"source":{"id":"2604.00860","kind":"arxiv","version":3},"verdict":{"id":"8f1b7e82-9cce-4ad9-9397-9b33dc804b4b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T22:44:54.577345Z","strongest_claim":"we prove this temporal objective is perfectly aligned with maximizing final task performance... PIPO performs ascent on the PIRL objective in expectation","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","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","pith_extraction_headline":"Reinforcement learning for language models can be made self-correcting by directly maximizing cumulative verified policy improvement across iterations."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.00860/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":"2604.00860","created_at":"2026-06-04T01:08:49.542053+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.00860v3","created_at":"2026-06-04T01:08:49.542053+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.00860","created_at":"2026-06-04T01:08:49.542053+00:00"},{"alias_kind":"pith_short_12","alias_value":"3IZOUWCOH56U","created_at":"2026-06-04T01:08:49.542053+00:00"},{"alias_kind":"pith_short_16","alias_value":"3IZOUWCOH56UNY56","created_at":"2026-06-04T01:08:49.542053+00:00"},{"alias_kind":"pith_short_8","alias_value":"3IZOUWCO","created_at":"2026-06-04T01:08:49.542053+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/3IZOUWCOH56UNY56EWZPMJU3YK","json":"https://pith.science/pith/3IZOUWCOH56UNY56EWZPMJU3YK.json","graph_json":"https://pith.science/api/pith-number/3IZOUWCOH56UNY56EWZPMJU3YK/graph.json","events_json":"https://pith.science/api/pith-number/3IZOUWCOH56UNY56EWZPMJU3YK/events.json","paper":"https://pith.science/paper/3IZOUWCO"},"agent_actions":{"view_html":"https://pith.science/pith/3IZOUWCOH56UNY56EWZPMJU3YK","download_json":"https://pith.science/pith/3IZOUWCOH56UNY56EWZPMJU3YK.json","view_paper":"https://pith.science/paper/3IZOUWCO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.00860&json=true","fetch_graph":"https://pith.science/api/pith-number/3IZOUWCOH56UNY56EWZPMJU3YK/graph.json","fetch_events":"https://pith.science/api/pith-number/3IZOUWCOH56UNY56EWZPMJU3YK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3IZOUWCOH56UNY56EWZPMJU3YK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3IZOUWCOH56UNY56EWZPMJU3YK/action/storage_attestation","attest_author":"https://pith.science/pith/3IZOUWCOH56UNY56EWZPMJU3YK/action/author_attestation","sign_citation":"https://pith.science/pith/3IZOUWCOH56UNY56EWZPMJU3YK/action/citation_signature","submit_replication":"https://pith.science/pith/3IZOUWCOH56UNY56EWZPMJU3YK/action/replication_record"}},"created_at":"2026-06-04T01:08:49.542053+00:00","updated_at":"2026-06-04T01:08:49.542053+00:00"}