{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:PFG3XWNLNJK5JXZADUFBKMNBTA","short_pith_number":"pith:PFG3XWNL","schema_version":"1.0","canonical_sha256":"794dbbd9ab6a55d4df201d0a1531a198246d742a7950dc1373acd627d2fd7eab","source":{"kind":"arxiv","id":"2605.28791","version":1},"attestation_state":"computed","paper":{"title":"Skill-Conditioned Gated Self-Distillation for LLM Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jiazhen Huang, Senkang Hu, Xiao Chen, Xiao Luo, Yong Dai, Yuzhi Zhao","submitted_at":"2026-05-27T17:49:52Z","abstract_excerpt":"On-policy self-distillation (SD) improves LLM reasoning by using teacher-side privileged information (PI) to turn sparse verifier outcomes into dense token-level supervision. Existing methods usually assume trusted PI, such as reference answers or successful traces. We ask whether PI can instead come from an experience-derived skill bank, where retrieved skills are compact and reusable but may also be irrelevant or misleading. We propose Skill-Conditioned Gated Self-Distillation (SGSD), which formulates skill-based SD as teacher hypothesis validation rather than unconditional imitation. SGSD r"},"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":"2605.28791","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-27T17:49:52Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"39a9aa5b941ee3304a226125bfabdf6d22d06d47582610fc438e332a1ebcc7cc","abstract_canon_sha256":"fc454ca36d9de77af06714ed51bd2bec7189bd975b6c956614c182a082ba802b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-28T02:05:03.711783Z","signature_b64":"a+fdZHsQLfu9Vs2enrOGE1S/UXS7ovjZkdV+Kga0P6BlGb2LRu5OcU3MVB7vPNnHpUCvWtwEGDHNJgk49gRgCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"794dbbd9ab6a55d4df201d0a1531a198246d742a7950dc1373acd627d2fd7eab","last_reissued_at":"2026-05-28T02:05:03.711288Z","signature_status":"signed_v1","first_computed_at":"2026-05-28T02:05:03.711288Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Skill-Conditioned Gated Self-Distillation for LLM Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jiazhen Huang, Senkang Hu, Xiao Chen, Xiao Luo, Yong Dai, Yuzhi Zhao","submitted_at":"2026-05-27T17:49:52Z","abstract_excerpt":"On-policy self-distillation (SD) improves LLM reasoning by using teacher-side privileged information (PI) to turn sparse verifier outcomes into dense token-level supervision. Existing methods usually assume trusted PI, such as reference answers or successful traces. We ask whether PI can instead come from an experience-derived skill bank, where retrieved skills are compact and reusable but may also be irrelevant or misleading. We propose Skill-Conditioned Gated Self-Distillation (SGSD), which formulates skill-based SD as teacher hypothesis validation rather than unconditional imitation. SGSD r"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.28791","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/2605.28791/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":"2605.28791","created_at":"2026-05-28T02:05:03.711369+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.28791v1","created_at":"2026-05-28T02:05:03.711369+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.28791","created_at":"2026-05-28T02:05:03.711369+00:00"},{"alias_kind":"pith_short_12","alias_value":"PFG3XWNLNJK5","created_at":"2026-05-28T02:05:03.711369+00:00"},{"alias_kind":"pith_short_16","alias_value":"PFG3XWNLNJK5JXZA","created_at":"2026-05-28T02:05:03.711369+00:00"},{"alias_kind":"pith_short_8","alias_value":"PFG3XWNL","created_at":"2026-05-28T02:05:03.711369+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/PFG3XWNLNJK5JXZADUFBKMNBTA","json":"https://pith.science/pith/PFG3XWNLNJK5JXZADUFBKMNBTA.json","graph_json":"https://pith.science/api/pith-number/PFG3XWNLNJK5JXZADUFBKMNBTA/graph.json","events_json":"https://pith.science/api/pith-number/PFG3XWNLNJK5JXZADUFBKMNBTA/events.json","paper":"https://pith.science/paper/PFG3XWNL"},"agent_actions":{"view_html":"https://pith.science/pith/PFG3XWNLNJK5JXZADUFBKMNBTA","download_json":"https://pith.science/pith/PFG3XWNLNJK5JXZADUFBKMNBTA.json","view_paper":"https://pith.science/paper/PFG3XWNL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.28791&json=true","fetch_graph":"https://pith.science/api/pith-number/PFG3XWNLNJK5JXZADUFBKMNBTA/graph.json","fetch_events":"https://pith.science/api/pith-number/PFG3XWNLNJK5JXZADUFBKMNBTA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PFG3XWNLNJK5JXZADUFBKMNBTA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PFG3XWNLNJK5JXZADUFBKMNBTA/action/storage_attestation","attest_author":"https://pith.science/pith/PFG3XWNLNJK5JXZADUFBKMNBTA/action/author_attestation","sign_citation":"https://pith.science/pith/PFG3XWNLNJK5JXZADUFBKMNBTA/action/citation_signature","submit_replication":"https://pith.science/pith/PFG3XWNLNJK5JXZADUFBKMNBTA/action/replication_record"}},"created_at":"2026-05-28T02:05:03.711369+00:00","updated_at":"2026-05-28T02:05:03.711369+00:00"}