{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QNWW6I6KN6RCOD5MMS5NY63QP6","short_pith_number":"pith:QNWW6I6K","schema_version":"1.0","canonical_sha256":"836d6f23ca6fa2270fac64badc7b707fb82da0be6b532ef33ad8fe6bba4debc3","source":{"kind":"arxiv","id":"2607.22334","version":1},"attestation_state":"computed","paper":{"title":"Cross-Tokenizer On-Policy Distillation via Byte-Prefix Marginalization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Han Xiao, Hao Wang, Honggang Qi, Kun Yuan, Minglei Zhang, Ming Sun, Wenlin Zhong","submitted_at":"2026-07-24T14:12:54Z","abstract_excerpt":"Open-weight language models from different families exhibit complementary capabilities, motivating their consolidation into a compact student through on-policy distillation (OPD). However, full-vocabulary OPD typically assumes a shared tokenizer, while existing cross-tokenizer methods may discard teacher probability mass or assign it to student tokens with unrelated content. We introduce Byte-Prefix Marginalization (BPM), which re-expresses the teacher's next-token distribution over the student vocabulary in a shared byte space. Specifically, BPM assigns each teacher token's probability to the"},"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":"2607.22334","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-24T14:12:54Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"31fd4ed81a239a58be34981278531bfc69475eabe0983163c23fe0584fdef40e","abstract_canon_sha256":"18aaeb73cbf44f396ba06ab3b3ea5c82e02fe5251ddf4c0bda4d28a9336cfbfb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-27T01:21:03.408560Z","signature_b64":"fkH3mZ7z8CgIR/jgyzdjANV6ND6HUlvTELlI1CT1F+Qpu41eg5MQmYcpTv9wDTiPNz1HmocyeLBQd6B/ZaEgDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"836d6f23ca6fa2270fac64badc7b707fb82da0be6b532ef33ad8fe6bba4debc3","last_reissued_at":"2026-07-27T01:21:03.407701Z","signature_status":"signed_v1","first_computed_at":"2026-07-27T01:21:03.407701Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Cross-Tokenizer On-Policy Distillation via Byte-Prefix Marginalization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Han Xiao, Hao Wang, Honggang Qi, Kun Yuan, Minglei Zhang, Ming Sun, Wenlin Zhong","submitted_at":"2026-07-24T14:12:54Z","abstract_excerpt":"Open-weight language models from different families exhibit complementary capabilities, motivating their consolidation into a compact student through on-policy distillation (OPD). However, full-vocabulary OPD typically assumes a shared tokenizer, while existing cross-tokenizer methods may discard teacher probability mass or assign it to student tokens with unrelated content. We introduce Byte-Prefix Marginalization (BPM), which re-expresses the teacher's next-token distribution over the student vocabulary in a shared byte space. Specifically, BPM assigns each teacher token's probability to the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.22334","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/2607.22334/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":"2607.22334","created_at":"2026-07-27T01:21:03.408129+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.22334v1","created_at":"2026-07-27T01:21:03.408129+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.22334","created_at":"2026-07-27T01:21:03.408129+00:00"},{"alias_kind":"pith_short_12","alias_value":"QNWW6I6KN6RC","created_at":"2026-07-27T01:21:03.408129+00:00"},{"alias_kind":"pith_short_16","alias_value":"QNWW6I6KN6RCOD5M","created_at":"2026-07-27T01:21:03.408129+00:00"},{"alias_kind":"pith_short_8","alias_value":"QNWW6I6K","created_at":"2026-07-27T01:21:03.408129+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/QNWW6I6KN6RCOD5MMS5NY63QP6","json":"https://pith.science/pith/QNWW6I6KN6RCOD5MMS5NY63QP6.json","graph_json":"https://pith.science/api/pith-number/QNWW6I6KN6RCOD5MMS5NY63QP6/graph.json","events_json":"https://pith.science/api/pith-number/QNWW6I6KN6RCOD5MMS5NY63QP6/events.json","paper":"https://pith.science/paper/QNWW6I6K"},"agent_actions":{"view_html":"https://pith.science/pith/QNWW6I6KN6RCOD5MMS5NY63QP6","download_json":"https://pith.science/pith/QNWW6I6KN6RCOD5MMS5NY63QP6.json","view_paper":"https://pith.science/paper/QNWW6I6K","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.22334&json=true","fetch_graph":"https://pith.science/api/pith-number/QNWW6I6KN6RCOD5MMS5NY63QP6/graph.json","fetch_events":"https://pith.science/api/pith-number/QNWW6I6KN6RCOD5MMS5NY63QP6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QNWW6I6KN6RCOD5MMS5NY63QP6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QNWW6I6KN6RCOD5MMS5NY63QP6/action/storage_attestation","attest_author":"https://pith.science/pith/QNWW6I6KN6RCOD5MMS5NY63QP6/action/author_attestation","sign_citation":"https://pith.science/pith/QNWW6I6KN6RCOD5MMS5NY63QP6/action/citation_signature","submit_replication":"https://pith.science/pith/QNWW6I6KN6RCOD5MMS5NY63QP6/action/replication_record"}},"created_at":"2026-07-27T01:21:03.408129+00:00","updated_at":"2026-07-27T01:21:03.408129+00:00"}