{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:32IBIBPXXUWUGFXRU5WX4XWF7G","short_pith_number":"pith:32IBIBPX","schema_version":"1.0","canonical_sha256":"de901405f7bd2d4316f1a76d7e5ec5f9aa019ee78a96dfe41e7072caba9022dc","source":{"kind":"arxiv","id":"2605.30070","version":1},"attestation_state":"computed","paper":{"title":"A Predictive Law for On-Policy Self-Distillation From World Feedback","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jerome Sieber, Matteo Saponati, Tommy He","submitted_at":"2026-05-28T15:17:19Z","abstract_excerpt":"Moving beyond simple scalar rewards toward richer world feedback is a natural path to more scalable RL post-training. On-policy self-distillation (OPSD) is a promising recent approach that uses arbitrary feedback as learning signal, yet its reliability compared to established methods, such as GRPO, remains unclear. We identify a strikingly consistent linear correlation between the initial student-self-teacher performance gap and the final performance improvement in OPSD. This relationship holds across context types and model families, providing a powerful predictive law for anticipating the ou"},"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":"2605.30070","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-28T15:17:19Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1246bdca0243de5df25e39514dbd0c0535c7631e2aaa20dad240da3b98c12d6e","abstract_canon_sha256":"275f196ac6cf7f1e17e85cccf6c53237c7aad6a43127059dd324ab24d06a12d8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T02:06:09.133625Z","signature_b64":"Fl41M6ckzn3adO35FOcUaIYm9LRKC15hHmobY6DPkDAB7eroVDnVzAxWxipywM0ux1NtMYzVk3clQnt/qq+tCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"de901405f7bd2d4316f1a76d7e5ec5f9aa019ee78a96dfe41e7072caba9022dc","last_reissued_at":"2026-05-29T02:06:09.133214Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T02:06:09.133214Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Predictive Law for On-Policy Self-Distillation From World Feedback","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jerome Sieber, Matteo Saponati, Tommy He","submitted_at":"2026-05-28T15:17:19Z","abstract_excerpt":"Moving beyond simple scalar rewards toward richer world feedback is a natural path to more scalable RL post-training. On-policy self-distillation (OPSD) is a promising recent approach that uses arbitrary feedback as learning signal, yet its reliability compared to established methods, such as GRPO, remains unclear. We identify a strikingly consistent linear correlation between the initial student-self-teacher performance gap and the final performance improvement in OPSD. This relationship holds across context types and model families, providing a powerful predictive law for anticipating the ou"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.30070","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/2605.30070/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":"2605.30070","created_at":"2026-05-29T02:06:09.133276+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.30070v1","created_at":"2026-05-29T02:06:09.133276+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.30070","created_at":"2026-05-29T02:06:09.133276+00:00"},{"alias_kind":"pith_short_12","alias_value":"32IBIBPXXUWU","created_at":"2026-05-29T02:06:09.133276+00:00"},{"alias_kind":"pith_short_16","alias_value":"32IBIBPXXUWUGFXR","created_at":"2026-05-29T02:06:09.133276+00:00"},{"alias_kind":"pith_short_8","alias_value":"32IBIBPX","created_at":"2026-05-29T02:06:09.133276+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/32IBIBPXXUWUGFXRU5WX4XWF7G","json":"https://pith.science/pith/32IBIBPXXUWUGFXRU5WX4XWF7G.json","graph_json":"https://pith.science/api/pith-number/32IBIBPXXUWUGFXRU5WX4XWF7G/graph.json","events_json":"https://pith.science/api/pith-number/32IBIBPXXUWUGFXRU5WX4XWF7G/events.json","paper":"https://pith.science/paper/32IBIBPX"},"agent_actions":{"view_html":"https://pith.science/pith/32IBIBPXXUWUGFXRU5WX4XWF7G","download_json":"https://pith.science/pith/32IBIBPXXUWUGFXRU5WX4XWF7G.json","view_paper":"https://pith.science/paper/32IBIBPX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.30070&json=true","fetch_graph":"https://pith.science/api/pith-number/32IBIBPXXUWUGFXRU5WX4XWF7G/graph.json","fetch_events":"https://pith.science/api/pith-number/32IBIBPXXUWUGFXRU5WX4XWF7G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/32IBIBPXXUWUGFXRU5WX4XWF7G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/32IBIBPXXUWUGFXRU5WX4XWF7G/action/storage_attestation","attest_author":"https://pith.science/pith/32IBIBPXXUWUGFXRU5WX4XWF7G/action/author_attestation","sign_citation":"https://pith.science/pith/32IBIBPXXUWUGFXRU5WX4XWF7G/action/citation_signature","submit_replication":"https://pith.science/pith/32IBIBPXXUWUGFXRU5WX4XWF7G/action/replication_record"}},"created_at":"2026-05-29T02:06:09.133276+00:00","updated_at":"2026-05-29T02:06:09.133276+00:00"}