{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DRMGGMGCWXJDUKML6KQJNFMNHK","short_pith_number":"pith:DRMGGMGC","schema_version":"1.0","canonical_sha256":"1c586330c2b5d23a298bf2a096958d3aa885c5ecfc00688f6f016c155284d9ac","source":{"kind":"arxiv","id":"2608.09836","version":1},"attestation_state":"computed","paper":{"title":"Mismatch Matters: On-Policy Distillation Beyond Token Agreement","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Bingqing Jiang, Chengzhi Yu, Difan Zou, Shengze Xu, Xu Wang, Yujin Han, Zichao Yu","submitted_at":"2026-08-10T16:53:14Z","abstract_excerpt":"On-policy distillation (OPD) has emerged as a core component of modern LLM post-training pipelines, yet we reveal a failure mode: degenerate agreement, where students exploit repetitive loops to achieve near-perfect token agreement with the teacher despite globally flawed responses. We therefore shift our focus from agreement to teacher-student mismatch, and find that mismatch tokens can be mainly categorized into two types: student-excess tokens and student-deficit tokens. Student-excess tokens are generated by the student but assigned near-zero probability by the teacher; their log-ratio cor"},"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":"2608.09836","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-08-10T16:53:14Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"eeabed5d5dc7cbeb8ff2c3c3aefc9655736b5f07b4d0ecbd09947d4cf8939242","abstract_canon_sha256":"5fa76177c7ca9b796b370c085c72a2f9819d6a31695cdaea474a96162adc5d1c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-11T02:25:13.222535Z","signature_b64":"XaZBIIDIcvsKDjzHMy1mSDWCcvQWIk54qBvN1ny8tQRS6wX/6tVQCMQUpkgYABcllKm/89R73oC7pXvFyz1DBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1c586330c2b5d23a298bf2a096958d3aa885c5ecfc00688f6f016c155284d9ac","last_reissued_at":"2026-08-11T02:25:13.220586Z","signature_status":"signed_v1","first_computed_at":"2026-08-11T02:25:13.220586Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mismatch Matters: On-Policy Distillation Beyond Token Agreement","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Bingqing Jiang, Chengzhi Yu, Difan Zou, Shengze Xu, Xu Wang, Yujin Han, Zichao Yu","submitted_at":"2026-08-10T16:53:14Z","abstract_excerpt":"On-policy distillation (OPD) has emerged as a core component of modern LLM post-training pipelines, yet we reveal a failure mode: degenerate agreement, where students exploit repetitive loops to achieve near-perfect token agreement with the teacher despite globally flawed responses. We therefore shift our focus from agreement to teacher-student mismatch, and find that mismatch tokens can be mainly categorized into two types: student-excess tokens and student-deficit tokens. Student-excess tokens are generated by the student but assigned near-zero probability by the teacher; their log-ratio cor"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.09836","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/2608.09836/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":"2608.09836","created_at":"2026-08-11T02:25:13.221214+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.09836v1","created_at":"2026-08-11T02:25:13.221214+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.09836","created_at":"2026-08-11T02:25:13.221214+00:00"},{"alias_kind":"pith_short_12","alias_value":"DRMGGMGCWXJD","created_at":"2026-08-11T02:25:13.221214+00:00"},{"alias_kind":"pith_short_16","alias_value":"DRMGGMGCWXJDUKML","created_at":"2026-08-11T02:25:13.221214+00:00"},{"alias_kind":"pith_short_8","alias_value":"DRMGGMGC","created_at":"2026-08-11T02:25:13.221214+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/DRMGGMGCWXJDUKML6KQJNFMNHK","json":"https://pith.science/pith/DRMGGMGCWXJDUKML6KQJNFMNHK.json","graph_json":"https://pith.science/api/pith-number/DRMGGMGCWXJDUKML6KQJNFMNHK/graph.json","events_json":"https://pith.science/api/pith-number/DRMGGMGCWXJDUKML6KQJNFMNHK/events.json","paper":"https://pith.science/paper/DRMGGMGC"},"agent_actions":{"view_html":"https://pith.science/pith/DRMGGMGCWXJDUKML6KQJNFMNHK","download_json":"https://pith.science/pith/DRMGGMGCWXJDUKML6KQJNFMNHK.json","view_paper":"https://pith.science/paper/DRMGGMGC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.09836&json=true","fetch_graph":"https://pith.science/api/pith-number/DRMGGMGCWXJDUKML6KQJNFMNHK/graph.json","fetch_events":"https://pith.science/api/pith-number/DRMGGMGCWXJDUKML6KQJNFMNHK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DRMGGMGCWXJDUKML6KQJNFMNHK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DRMGGMGCWXJDUKML6KQJNFMNHK/action/storage_attestation","attest_author":"https://pith.science/pith/DRMGGMGCWXJDUKML6KQJNFMNHK/action/author_attestation","sign_citation":"https://pith.science/pith/DRMGGMGCWXJDUKML6KQJNFMNHK/action/citation_signature","submit_replication":"https://pith.science/pith/DRMGGMGCWXJDUKML6KQJNFMNHK/action/replication_record"}},"created_at":"2026-08-11T02:25:13.221214+00:00","updated_at":"2026-08-11T02:25:13.221214+00:00"}