{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2ULUYJ4L2OUXGNCMBXPPM7FQ77","short_pith_number":"pith:2ULUYJ4L","schema_version":"1.0","canonical_sha256":"d5174c278bd3a973344c0ddef67cb0ffd0dd6254e334490d67a86b0152f875de","source":{"kind":"arxiv","id":"2511.23310","version":3},"attestation_state":"computed","paper":{"title":"Variance-Aware Baselines and Adaptive Learning Rates for Reinforcement Learning with Verifiable Rewards","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Jiayi Sheng, Zeyu Zheng, Zixun Huang","submitted_at":"2025-11-28T16:09:28Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has emerged as an effective paradigm for post-training large language models, yet the design of its baselines and learning-rate schedules remains largely heuristic. This limits our understanding of the statistical properties of policy-gradient estimators and their interaction with optimization dynamics. In this work, we develop a theoretical framework for variance-aware baseline design and adaptive learning-rate selection in RLVR. Under a KL-regularized policy-optimization setting, we establish the unbiasedness of the resulting gradient est"},"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":"2511.23310","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-11-28T16:09:28Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"e05f3828ccfec2296d93c5a2675e1a06382430fd4bf83d1421d322a91158103f","abstract_canon_sha256":"4103cda885aa7d0a22d03ea5b344c63f3a9484e0edc70e709242153dfdaef292"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d5174c278bd3a973344c0ddef67cb0ffd0dd6254e334490d67a86b0152f875de","last_reissued_at":"2026-07-31T01:32:58.372713Z","signature_status":"unsigned_v0","first_computed_at":"2026-07-31T01:32:58.372713Z"},"graph_snapshot":{"paper":{"title":"Variance-Aware Baselines and Adaptive Learning Rates for Reinforcement Learning with Verifiable Rewards","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Jiayi Sheng, Zeyu Zheng, Zixun Huang","submitted_at":"2025-11-28T16:09:28Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has emerged as an effective paradigm for post-training large language models, yet the design of its baselines and learning-rate schedules remains largely heuristic. This limits our understanding of the statistical properties of policy-gradient estimators and their interaction with optimization dynamics. In this work, we develop a theoretical framework for variance-aware baseline design and adaptive learning-rate selection in RLVR. Under a KL-regularized policy-optimization setting, we establish the unbiasedness of the resulting gradient est"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.23310","kind":"arxiv","version":3},"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/2511.23310/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":"2511.23310","created_at":"2026-07-31T01:32:58.376103+00:00"},{"alias_kind":"arxiv_version","alias_value":"2511.23310v3","created_at":"2026-07-31T01:32:58.376103+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.23310","created_at":"2026-07-31T01:32:58.376103+00:00"},{"alias_kind":"pith_short_12","alias_value":"2ULUYJ4L2OUX","created_at":"2026-07-31T01:32:58.376103+00:00"},{"alias_kind":"pith_short_16","alias_value":"2ULUYJ4L2OUXGNCM","created_at":"2026-07-31T01:32:58.376103+00:00"},{"alias_kind":"pith_short_8","alias_value":"2ULUYJ4L","created_at":"2026-07-31T01:32:58.376103+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/2ULUYJ4L2OUXGNCMBXPPM7FQ77","json":"https://pith.science/pith/2ULUYJ4L2OUXGNCMBXPPM7FQ77.json","graph_json":"https://pith.science/api/pith-number/2ULUYJ4L2OUXGNCMBXPPM7FQ77/graph.json","events_json":"https://pith.science/api/pith-number/2ULUYJ4L2OUXGNCMBXPPM7FQ77/events.json","paper":"https://pith.science/paper/2ULUYJ4L"},"agent_actions":{"view_html":"https://pith.science/pith/2ULUYJ4L2OUXGNCMBXPPM7FQ77","download_json":"https://pith.science/pith/2ULUYJ4L2OUXGNCMBXPPM7FQ77.json","view_paper":"https://pith.science/paper/2ULUYJ4L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2511.23310&json=true","fetch_graph":"https://pith.science/api/pith-number/2ULUYJ4L2OUXGNCMBXPPM7FQ77/graph.json","fetch_events":"https://pith.science/api/pith-number/2ULUYJ4L2OUXGNCMBXPPM7FQ77/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2ULUYJ4L2OUXGNCMBXPPM7FQ77/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2ULUYJ4L2OUXGNCMBXPPM7FQ77/action/storage_attestation","attest_author":"https://pith.science/pith/2ULUYJ4L2OUXGNCMBXPPM7FQ77/action/author_attestation","sign_citation":"https://pith.science/pith/2ULUYJ4L2OUXGNCMBXPPM7FQ77/action/citation_signature","submit_replication":"https://pith.science/pith/2ULUYJ4L2OUXGNCMBXPPM7FQ77/action/replication_record"}},"created_at":"2026-07-31T01:32:58.376103+00:00","updated_at":"2026-07-31T01:32:58.376103+00:00"}