{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QFJVPOIJMUDXI3GRNSQSCKQPWE","short_pith_number":"pith:QFJVPOIJ","schema_version":"1.0","canonical_sha256":"815357b9096507746cd16ca1212a0fb13adb9939e4e220b7d9a380a4a329245d","source":{"kind":"arxiv","id":"2607.19046","version":1},"attestation_state":"computed","paper":{"title":"Contrastive On-Policy Distillation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Dayiheng Liu, Jiacheng Ruan, Jun Tang, Shuai Bai, Ting Liu, Wenzhen Yuan, Yuzhuo Fu, Zhibo Yang","submitted_at":"2026-07-21T12:35:04Z","abstract_excerpt":"On-policy Distillation (OPD) supervises a student model on trajectories sampled from its own policy by minimizing the divergence between the output distributions of the teacher and student at each token position, thereby providing dense token-level supervision. Although existing OPD methods have demonstrated strong performance in improving the reasoning ability of student models, their objectives fundamentally rely on token-level distribution matching. Consequently, they lack an explicit signal for comparing a token's relative compatibility across reasoning modes and thus do not directly model"},"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":"2607.19046","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-07-21T12:35:04Z","cross_cats_sorted":[],"title_canon_sha256":"a36ff9101e17ef292699219aad90909ce562902612f27ad64404ea19124ea7ba","abstract_canon_sha256":"8696b50cfb2e0cf41bdb2fcfe6dcd812c74dd32718f083ebeddc7b6517e50bf6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-22T01:24:05.834184Z","signature_b64":"3pwFN9q7sCxXD59ndd314xhuLHo0smTY765qZFIRDBTW7F96Ml4TA0yZ3mIArUeeTdj4dHK85yICnaVDaymsBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"815357b9096507746cd16ca1212a0fb13adb9939e4e220b7d9a380a4a329245d","last_reissued_at":"2026-07-22T01:24:05.833366Z","signature_status":"signed_v1","first_computed_at":"2026-07-22T01:24:05.833366Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Contrastive On-Policy Distillation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Dayiheng Liu, Jiacheng Ruan, Jun Tang, Shuai Bai, Ting Liu, Wenzhen Yuan, Yuzhuo Fu, Zhibo Yang","submitted_at":"2026-07-21T12:35:04Z","abstract_excerpt":"On-policy Distillation (OPD) supervises a student model on trajectories sampled from its own policy by minimizing the divergence between the output distributions of the teacher and student at each token position, thereby providing dense token-level supervision. Although existing OPD methods have demonstrated strong performance in improving the reasoning ability of student models, their objectives fundamentally rely on token-level distribution matching. Consequently, they lack an explicit signal for comparing a token's relative compatibility across reasoning modes and thus do not directly model"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.19046","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/2607.19046/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":"2607.19046","created_at":"2026-07-22T01:24:05.833794+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.19046v1","created_at":"2026-07-22T01:24:05.833794+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.19046","created_at":"2026-07-22T01:24:05.833794+00:00"},{"alias_kind":"pith_short_12","alias_value":"QFJVPOIJMUDX","created_at":"2026-07-22T01:24:05.833794+00:00"},{"alias_kind":"pith_short_16","alias_value":"QFJVPOIJMUDXI3GR","created_at":"2026-07-22T01:24:05.833794+00:00"},{"alias_kind":"pith_short_8","alias_value":"QFJVPOIJ","created_at":"2026-07-22T01:24:05.833794+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/QFJVPOIJMUDXI3GRNSQSCKQPWE","json":"https://pith.science/pith/QFJVPOIJMUDXI3GRNSQSCKQPWE.json","graph_json":"https://pith.science/api/pith-number/QFJVPOIJMUDXI3GRNSQSCKQPWE/graph.json","events_json":"https://pith.science/api/pith-number/QFJVPOIJMUDXI3GRNSQSCKQPWE/events.json","paper":"https://pith.science/paper/QFJVPOIJ"},"agent_actions":{"view_html":"https://pith.science/pith/QFJVPOIJMUDXI3GRNSQSCKQPWE","download_json":"https://pith.science/pith/QFJVPOIJMUDXI3GRNSQSCKQPWE.json","view_paper":"https://pith.science/paper/QFJVPOIJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.19046&json=true","fetch_graph":"https://pith.science/api/pith-number/QFJVPOIJMUDXI3GRNSQSCKQPWE/graph.json","fetch_events":"https://pith.science/api/pith-number/QFJVPOIJMUDXI3GRNSQSCKQPWE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QFJVPOIJMUDXI3GRNSQSCKQPWE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QFJVPOIJMUDXI3GRNSQSCKQPWE/action/storage_attestation","attest_author":"https://pith.science/pith/QFJVPOIJMUDXI3GRNSQSCKQPWE/action/author_attestation","sign_citation":"https://pith.science/pith/QFJVPOIJMUDXI3GRNSQSCKQPWE/action/citation_signature","submit_replication":"https://pith.science/pith/QFJVPOIJMUDXI3GRNSQSCKQPWE/action/replication_record"}},"created_at":"2026-07-22T01:24:05.833794+00:00","updated_at":"2026-07-22T01:24:05.833794+00:00"}