{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:4KCCKS2E2ZEVH67AF2OY4SRFRC","short_pith_number":"pith:4KCCKS2E","schema_version":"1.0","canonical_sha256":"e284254b44d64953fbe02e9d8e4a25889376140cb5cee5d4d2742d5e60cc5c6d","source":{"kind":"arxiv","id":"2608.01837","version":1},"attestation_state":"computed","paper":{"title":"PCSD: Persistent Consistency for Self-Distillation in Agentic Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Changsheng Li, Chunji Lv, Guoren Wang, Jinyang Wu, Junlin Liu, Ming Liu, Xinming Wang, Yang Gao, Yangguang Wei","submitted_at":"2026-08-03T07:47:59Z","abstract_excerpt":"Large language model agents have shown strong potential in complex interactive tasks, yet their reinforcement learning (RL) is often hindered by sparse rewards, as a long multi-turn trajectory may receive only a single outcome-level signal. On-policy self-distillation (OPSD) provides dense token-level supervision from a privileged teacher, but the teacher may not be reliable at every position. Existing methods commonly rely on isolated token-level discrepancies, which can be sensitive to noise, or assign a shared step-level weight that may overlook positional variation. We propose Persistent C"},"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":"2608.01837","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-03T07:47:59Z","cross_cats_sorted":[],"title_canon_sha256":"6fc5d80109d348855a552b7d94199c766cd515598eaedbc829b18d8c9884ee28","abstract_canon_sha256":"749bc90f7d127f4d1083aa0aac19fff2910daa003394d13324351e6f2fc03414"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-04T02:07:22.096058Z","signature_b64":"aAmS9mXWIaUZL/omHw5hNzg+//Jy9IK4S4DXGudP+tC3il0VB5ZzTkGxd1qgZPSbsJmxVNP2+ZSXIybcRFOMAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e284254b44d64953fbe02e9d8e4a25889376140cb5cee5d4d2742d5e60cc5c6d","last_reissued_at":"2026-08-04T02:07:22.094216Z","signature_status":"signed_v1","first_computed_at":"2026-08-04T02:07:22.094216Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PCSD: Persistent Consistency for Self-Distillation in Agentic Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Changsheng Li, Chunji Lv, Guoren Wang, Jinyang Wu, Junlin Liu, Ming Liu, Xinming Wang, Yang Gao, Yangguang Wei","submitted_at":"2026-08-03T07:47:59Z","abstract_excerpt":"Large language model agents have shown strong potential in complex interactive tasks, yet their reinforcement learning (RL) is often hindered by sparse rewards, as a long multi-turn trajectory may receive only a single outcome-level signal. On-policy self-distillation (OPSD) provides dense token-level supervision from a privileged teacher, but the teacher may not be reliable at every position. Existing methods commonly rely on isolated token-level discrepancies, which can be sensitive to noise, or assign a shared step-level weight that may overlook positional variation. We propose Persistent C"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.01837","kind":"arxiv","version":1},"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/2608.01837/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":"2608.01837","created_at":"2026-08-04T02:07:22.095752+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.01837v1","created_at":"2026-08-04T02:07:22.095752+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.01837","created_at":"2026-08-04T02:07:22.095752+00:00"},{"alias_kind":"pith_short_12","alias_value":"4KCCKS2E2ZEV","created_at":"2026-08-04T02:07:22.095752+00:00"},{"alias_kind":"pith_short_16","alias_value":"4KCCKS2E2ZEVH67A","created_at":"2026-08-04T02:07:22.095752+00:00"},{"alias_kind":"pith_short_8","alias_value":"4KCCKS2E","created_at":"2026-08-04T02:07:22.095752+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/4KCCKS2E2ZEVH67AF2OY4SRFRC","json":"https://pith.science/pith/4KCCKS2E2ZEVH67AF2OY4SRFRC.json","graph_json":"https://pith.science/api/pith-number/4KCCKS2E2ZEVH67AF2OY4SRFRC/graph.json","events_json":"https://pith.science/api/pith-number/4KCCKS2E2ZEVH67AF2OY4SRFRC/events.json","paper":"https://pith.science/paper/4KCCKS2E"},"agent_actions":{"view_html":"https://pith.science/pith/4KCCKS2E2ZEVH67AF2OY4SRFRC","download_json":"https://pith.science/pith/4KCCKS2E2ZEVH67AF2OY4SRFRC.json","view_paper":"https://pith.science/paper/4KCCKS2E","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.01837&json=true","fetch_graph":"https://pith.science/api/pith-number/4KCCKS2E2ZEVH67AF2OY4SRFRC/graph.json","fetch_events":"https://pith.science/api/pith-number/4KCCKS2E2ZEVH67AF2OY4SRFRC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4KCCKS2E2ZEVH67AF2OY4SRFRC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4KCCKS2E2ZEVH67AF2OY4SRFRC/action/storage_attestation","attest_author":"https://pith.science/pith/4KCCKS2E2ZEVH67AF2OY4SRFRC/action/author_attestation","sign_citation":"https://pith.science/pith/4KCCKS2E2ZEVH67AF2OY4SRFRC/action/citation_signature","submit_replication":"https://pith.science/pith/4KCCKS2E2ZEVH67AF2OY4SRFRC/action/replication_record"}},"created_at":"2026-08-04T02:07:22.095752+00:00","updated_at":"2026-08-04T02:07:22.095752+00:00"}