{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:C4EBBBZA4DODNNPO3C75EP2NRT","short_pith_number":"pith:C4EBBBZA","schema_version":"1.0","canonical_sha256":"1708108720e0dc36b5eed8bfd23f4d8cc54b5072e8f860309401a6db21913c10","source":{"kind":"arxiv","id":"2507.14503","version":1},"attestation_state":"computed","paper":{"title":"Generative Distribution Distillation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Beier Zhu, Bei Yu, Hanwang Zhang, Jiequan Cui, Pengguang Chen, Qingshan Xu, Richang Hong, Xiaogang Xu, Xiaojuan Qi","submitted_at":"2025-07-19T06:27:42Z","abstract_excerpt":"In this paper, we formulate the knowledge distillation (KD) as a conditional generative problem and propose the \\textit{Generative Distribution Distillation (GenDD)} framework. A naive \\textit{GenDD} baseline encounters two major challenges: the curse of high-dimensional optimization and the lack of semantic supervision from labels. To address these issues, we introduce a \\textit{Split Tokenization} strategy, achieving stable and effective unsupervised KD. Additionally, we develop the \\textit{Distribution Contraction} technique to integrate label supervision into the reconstruction objective. "},"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":"2507.14503","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-19T06:27:42Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"5924f3590dd3687905b71ee0b4f4ad8f6b1624a32b4a69d8a309333c0f3a64a8","abstract_canon_sha256":"8bfb758d395ffc9292a93042f3983987c4354818c4d61e4f9fa6603824a89e59"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:40:08.472993Z","signature_b64":"TgAFngPGBYSBRjwPhTdBoH4FhN0MxkRCVtsdwZBa2ZhIYywTNqVt9w/iod74rfM8dmM54To70fz9DaAjiH4OCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1708108720e0dc36b5eed8bfd23f4d8cc54b5072e8f860309401a6db21913c10","last_reissued_at":"2026-07-05T11:40:08.472521Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:40:08.472521Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Generative Distribution Distillation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Beier Zhu, Bei Yu, Hanwang Zhang, Jiequan Cui, Pengguang Chen, Qingshan Xu, Richang Hong, Xiaogang Xu, Xiaojuan Qi","submitted_at":"2025-07-19T06:27:42Z","abstract_excerpt":"In this paper, we formulate the knowledge distillation (KD) as a conditional generative problem and propose the \\textit{Generative Distribution Distillation (GenDD)} framework. A naive \\textit{GenDD} baseline encounters two major challenges: the curse of high-dimensional optimization and the lack of semantic supervision from labels. To address these issues, we introduce a \\textit{Split Tokenization} strategy, achieving stable and effective unsupervised KD. Additionally, we develop the \\textit{Distribution Contraction} technique to integrate label supervision into the reconstruction objective. "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.14503","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/2507.14503/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":"2507.14503","created_at":"2026-07-05T11:40:08.472578+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.14503v1","created_at":"2026-07-05T11:40:08.472578+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.14503","created_at":"2026-07-05T11:40:08.472578+00:00"},{"alias_kind":"pith_short_12","alias_value":"C4EBBBZA4DOD","created_at":"2026-07-05T11:40:08.472578+00:00"},{"alias_kind":"pith_short_16","alias_value":"C4EBBBZA4DODNNPO","created_at":"2026-07-05T11:40:08.472578+00:00"},{"alias_kind":"pith_short_8","alias_value":"C4EBBBZA","created_at":"2026-07-05T11:40:08.472578+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27696","citing_title":"Class-frequency Guided Noise Schedule for Diffusion Models","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/C4EBBBZA4DODNNPO3C75EP2NRT","json":"https://pith.science/pith/C4EBBBZA4DODNNPO3C75EP2NRT.json","graph_json":"https://pith.science/api/pith-number/C4EBBBZA4DODNNPO3C75EP2NRT/graph.json","events_json":"https://pith.science/api/pith-number/C4EBBBZA4DODNNPO3C75EP2NRT/events.json","paper":"https://pith.science/paper/C4EBBBZA"},"agent_actions":{"view_html":"https://pith.science/pith/C4EBBBZA4DODNNPO3C75EP2NRT","download_json":"https://pith.science/pith/C4EBBBZA4DODNNPO3C75EP2NRT.json","view_paper":"https://pith.science/paper/C4EBBBZA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.14503&json=true","fetch_graph":"https://pith.science/api/pith-number/C4EBBBZA4DODNNPO3C75EP2NRT/graph.json","fetch_events":"https://pith.science/api/pith-number/C4EBBBZA4DODNNPO3C75EP2NRT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/C4EBBBZA4DODNNPO3C75EP2NRT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/C4EBBBZA4DODNNPO3C75EP2NRT/action/storage_attestation","attest_author":"https://pith.science/pith/C4EBBBZA4DODNNPO3C75EP2NRT/action/author_attestation","sign_citation":"https://pith.science/pith/C4EBBBZA4DODNNPO3C75EP2NRT/action/citation_signature","submit_replication":"https://pith.science/pith/C4EBBBZA4DODNNPO3C75EP2NRT/action/replication_record"}},"created_at":"2026-07-05T11:40:08.472578+00:00","updated_at":"2026-07-05T11:40:08.472578+00:00"}