{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:Y2E6XHWKWQRZBRYGAXPSMFCTBE","short_pith_number":"pith:Y2E6XHWK","schema_version":"1.0","canonical_sha256":"c689eb9ecab42390c70605df261453092b2fd8c58768a94893c82971415fcd6d","source":{"kind":"arxiv","id":"2608.08764","version":1},"attestation_state":"computed","paper":{"title":"Learning from Consensus and Disagreement: Unsupervised On-Policy Self-Distillation with Minority-Trajectory Contrast","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chunyu Yang, Jiaxin Guo, Xuanbo Fan, Yanwei Yue, Yan Zhang","submitted_at":"2026-08-09T15:23:25Z","abstract_excerpt":"On-policy self-distillation improves language-model reasoning by querying a teacher on states actually visited by the student. Recent methods create a powerful information asymmetry by exposing the teacher to privileged context, yet they fundamentally rely on external supervision---such as gold solutions or verifiers---to construct this advantage. We introduce CoDA (Consensus and Disagreement Alignment), a fully unsupervised framework that creates reliable privileged information entirely from the latent uncertainty structure of a model's own unlabeled rollouts. CoDA extracts two complementary "},"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":"2608.08764","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-08-09T15:23:25Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"bbdf77534df57d987f7d373961604e49f1e06b32520f8ddc7284c76cc057c2cf","abstract_canon_sha256":"f8e3146de87172a38c33434a1d6379a56145c7323db3cb3bd160c3b41642a6fc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-11T01:25:16.155862Z","signature_b64":"TRQiMrw66uX5izkgJ1BjvKb2OIVzPB97EbT10E9SCWEo+PBbLz43FKdlrPsPLyVKI3xYhI/S2WFio2Qx1FMSDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c689eb9ecab42390c70605df261453092b2fd8c58768a94893c82971415fcd6d","last_reissued_at":"2026-08-11T01:25:16.153326Z","signature_status":"signed_v1","first_computed_at":"2026-08-11T01:25:16.153326Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning from Consensus and Disagreement: Unsupervised On-Policy Self-Distillation with Minority-Trajectory Contrast","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chunyu Yang, Jiaxin Guo, Xuanbo Fan, Yanwei Yue, Yan Zhang","submitted_at":"2026-08-09T15:23:25Z","abstract_excerpt":"On-policy self-distillation improves language-model reasoning by querying a teacher on states actually visited by the student. Recent methods create a powerful information asymmetry by exposing the teacher to privileged context, yet they fundamentally rely on external supervision---such as gold solutions or verifiers---to construct this advantage. We introduce CoDA (Consensus and Disagreement Alignment), a fully unsupervised framework that creates reliable privileged information entirely from the latent uncertainty structure of a model's own unlabeled rollouts. CoDA extracts two complementary "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.08764","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/2608.08764/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":"2608.08764","created_at":"2026-08-11T01:25:16.154254+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.08764v1","created_at":"2026-08-11T01:25:16.154254+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.08764","created_at":"2026-08-11T01:25:16.154254+00:00"},{"alias_kind":"pith_short_12","alias_value":"Y2E6XHWKWQRZ","created_at":"2026-08-11T01:25:16.154254+00:00"},{"alias_kind":"pith_short_16","alias_value":"Y2E6XHWKWQRZBRYG","created_at":"2026-08-11T01:25:16.154254+00:00"},{"alias_kind":"pith_short_8","alias_value":"Y2E6XHWK","created_at":"2026-08-11T01:25:16.154254+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/Y2E6XHWKWQRZBRYGAXPSMFCTBE","json":"https://pith.science/pith/Y2E6XHWKWQRZBRYGAXPSMFCTBE.json","graph_json":"https://pith.science/api/pith-number/Y2E6XHWKWQRZBRYGAXPSMFCTBE/graph.json","events_json":"https://pith.science/api/pith-number/Y2E6XHWKWQRZBRYGAXPSMFCTBE/events.json","paper":"https://pith.science/paper/Y2E6XHWK"},"agent_actions":{"view_html":"https://pith.science/pith/Y2E6XHWKWQRZBRYGAXPSMFCTBE","download_json":"https://pith.science/pith/Y2E6XHWKWQRZBRYGAXPSMFCTBE.json","view_paper":"https://pith.science/paper/Y2E6XHWK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.08764&json=true","fetch_graph":"https://pith.science/api/pith-number/Y2E6XHWKWQRZBRYGAXPSMFCTBE/graph.json","fetch_events":"https://pith.science/api/pith-number/Y2E6XHWKWQRZBRYGAXPSMFCTBE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Y2E6XHWKWQRZBRYGAXPSMFCTBE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Y2E6XHWKWQRZBRYGAXPSMFCTBE/action/storage_attestation","attest_author":"https://pith.science/pith/Y2E6XHWKWQRZBRYGAXPSMFCTBE/action/author_attestation","sign_citation":"https://pith.science/pith/Y2E6XHWKWQRZBRYGAXPSMFCTBE/action/citation_signature","submit_replication":"https://pith.science/pith/Y2E6XHWKWQRZBRYGAXPSMFCTBE/action/replication_record"}},"created_at":"2026-08-11T01:25:16.154254+00:00","updated_at":"2026-08-11T01:25:16.154254+00:00"}