{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:Z2UJLEPKYLEHJDMPDSD7XIPK7F","short_pith_number":"pith:Z2UJLEPK","schema_version":"1.0","canonical_sha256":"cea89591eac2c8748d8f1c87fba1eaf95af5f0a1e3c4085271790e92285b5496","source":{"kind":"arxiv","id":"2606.01039","version":1},"attestation_state":"computed","paper":{"title":"OPD+: Rethinking the Advantage Design for On-Policy Distillation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"David Yao, Genta Indra Winata, Han Lin, Hanyang Zhao, Haoxian Chen, Wenpin Tang","submitted_at":"2026-05-31T06:10:38Z","abstract_excerpt":"On-policy distillation (OPD) is a widely used technique to transfer capabilities from capable teacher language models to the base student models, and can be formulated in a reinforcement learning style objective using student generated rollouts. Yet, despite the divergence reward being dependent on student model likelihood, existing works usually adopt a stop gradient design primarily for stability, which makes the resulting advantage estimation questionable. In this work, we provide a generic optimization framework based on f-divergence between the student and teacher, and mathematically revi"},"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":"2606.01039","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-31T06:10:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"0d8b34748e7e4aebd25aa3fbffbbe06ed793de12adf83073887b7fe2a830d623","abstract_canon_sha256":"0eacb855d1ab9cde24e340399e1daa1542f0a087121e76b368810b463e210474"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:04:19.242296Z","signature_b64":"Hgp8tRw9n3E7bZWQrrv0NpIWUaA9AWkDLI9+OOeeSl7GfrwiSjuiU2iHmWjByZmXy0QxyGXpse4mYsUddzoMAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cea89591eac2c8748d8f1c87fba1eaf95af5f0a1e3c4085271790e92285b5496","last_reissued_at":"2026-06-02T01:04:19.241853Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:04:19.241853Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"OPD+: Rethinking the Advantage Design for On-Policy Distillation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"David Yao, Genta Indra Winata, Han Lin, Hanyang Zhao, Haoxian Chen, Wenpin Tang","submitted_at":"2026-05-31T06:10:38Z","abstract_excerpt":"On-policy distillation (OPD) is a widely used technique to transfer capabilities from capable teacher language models to the base student models, and can be formulated in a reinforcement learning style objective using student generated rollouts. Yet, despite the divergence reward being dependent on student model likelihood, existing works usually adopt a stop gradient design primarily for stability, which makes the resulting advantage estimation questionable. In this work, we provide a generic optimization framework based on f-divergence between the student and teacher, and mathematically revi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.01039","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/2606.01039/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":"2606.01039","created_at":"2026-06-02T01:04:19.241923+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.01039v1","created_at":"2026-06-02T01:04:19.241923+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.01039","created_at":"2026-06-02T01:04:19.241923+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z2UJLEPKYLEH","created_at":"2026-06-02T01:04:19.241923+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z2UJLEPKYLEHJDMP","created_at":"2026-06-02T01:04:19.241923+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z2UJLEPK","created_at":"2026-06-02T01:04:19.241923+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/Z2UJLEPKYLEHJDMPDSD7XIPK7F","json":"https://pith.science/pith/Z2UJLEPKYLEHJDMPDSD7XIPK7F.json","graph_json":"https://pith.science/api/pith-number/Z2UJLEPKYLEHJDMPDSD7XIPK7F/graph.json","events_json":"https://pith.science/api/pith-number/Z2UJLEPKYLEHJDMPDSD7XIPK7F/events.json","paper":"https://pith.science/paper/Z2UJLEPK"},"agent_actions":{"view_html":"https://pith.science/pith/Z2UJLEPKYLEHJDMPDSD7XIPK7F","download_json":"https://pith.science/pith/Z2UJLEPKYLEHJDMPDSD7XIPK7F.json","view_paper":"https://pith.science/paper/Z2UJLEPK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.01039&json=true","fetch_graph":"https://pith.science/api/pith-number/Z2UJLEPKYLEHJDMPDSD7XIPK7F/graph.json","fetch_events":"https://pith.science/api/pith-number/Z2UJLEPKYLEHJDMPDSD7XIPK7F/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z2UJLEPKYLEHJDMPDSD7XIPK7F/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z2UJLEPKYLEHJDMPDSD7XIPK7F/action/storage_attestation","attest_author":"https://pith.science/pith/Z2UJLEPKYLEHJDMPDSD7XIPK7F/action/author_attestation","sign_citation":"https://pith.science/pith/Z2UJLEPKYLEHJDMPDSD7XIPK7F/action/citation_signature","submit_replication":"https://pith.science/pith/Z2UJLEPKYLEHJDMPDSD7XIPK7F/action/replication_record"}},"created_at":"2026-06-02T01:04:19.241923+00:00","updated_at":"2026-06-02T01:04:19.241923+00:00"}