{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:ESVFPXQAVJMLSWRRACIHKHTEMU","short_pith_number":"pith:ESVFPXQA","schema_version":"1.0","canonical_sha256":"24aa57de00aa58b95a310090751e64652bfc3bfc42d2ef5170221c891e582f89","source":{"kind":"arxiv","id":"2302.00670","version":2},"attestation_state":"computed","paper":{"title":"Stable Target Field for Reduced Variance Score Estimation in Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Shangyuan Tong, Tommi Jaakkola, Yilun Xu","submitted_at":"2023-02-01T18:57:01Z","abstract_excerpt":"Diffusion models generate samples by reversing a fixed forward diffusion process. Despite already providing impressive empirical results, these diffusion models algorithms can be further improved by reducing the variance of the training targets in their denoising score-matching objective. We argue that the source of such variance lies in the handling of intermediate noise-variance scales, where multiple modes in the data affect the direction of reverse paths. We propose to remedy the problem by incorporating a reference batch which we use to calculate weighted conditional scores as more stable"},"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":"2302.00670","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-02-01T18:57:01Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"05481d9ba1447be052ec1890717066011da02abf99016e57acf2840348cb4c52","abstract_canon_sha256":"6456043b81cc1a3948437b86eb3bf9788e70f88d16137b3a2c953ee91501c6d2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:43:04.050420Z","signature_b64":"QyTvYd9hz7RPWnIq6Jo743ucQP5bch4/amI9O3y+zluMUCx6JZtcJlTogliqg9NUvdnA+vFFGmdSNEcYOfeTCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"24aa57de00aa58b95a310090751e64652bfc3bfc42d2ef5170221c891e582f89","last_reissued_at":"2026-07-05T05:43:04.049921Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:43:04.049921Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Stable Target Field for Reduced Variance Score Estimation in Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Shangyuan Tong, Tommi Jaakkola, Yilun Xu","submitted_at":"2023-02-01T18:57:01Z","abstract_excerpt":"Diffusion models generate samples by reversing a fixed forward diffusion process. Despite already providing impressive empirical results, these diffusion models algorithms can be further improved by reducing the variance of the training targets in their denoising score-matching objective. We argue that the source of such variance lies in the handling of intermediate noise-variance scales, where multiple modes in the data affect the direction of reverse paths. We propose to remedy the problem by incorporating a reference batch which we use to calculate weighted conditional scores as more stable"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.00670","kind":"arxiv","version":2},"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/2302.00670/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":"2302.00670","created_at":"2026-07-05T05:43:04.049979+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.00670v2","created_at":"2026-07-05T05:43:04.049979+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.00670","created_at":"2026-07-05T05:43:04.049979+00:00"},{"alias_kind":"pith_short_12","alias_value":"ESVFPXQAVJML","created_at":"2026-07-05T05:43:04.049979+00:00"},{"alias_kind":"pith_short_16","alias_value":"ESVFPXQAVJMLSWRR","created_at":"2026-07-05T05:43:04.049979+00:00"},{"alias_kind":"pith_short_8","alias_value":"ESVFPXQA","created_at":"2026-07-05T05:43:04.049979+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.17991","citing_title":"Demystifying Transition Matching: When and Why It Can Beat Flow Matching","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2506.13763","citing_title":"Diagnosing and Improving Diffusion Models by Estimating the Optimal Loss Value","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08837","citing_title":"Discrete Meanflow Training Curriculum","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07319","citing_title":"Generative Modeling with Flux Matching","ref_index":65,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ESVFPXQAVJMLSWRRACIHKHTEMU","json":"https://pith.science/pith/ESVFPXQAVJMLSWRRACIHKHTEMU.json","graph_json":"https://pith.science/api/pith-number/ESVFPXQAVJMLSWRRACIHKHTEMU/graph.json","events_json":"https://pith.science/api/pith-number/ESVFPXQAVJMLSWRRACIHKHTEMU/events.json","paper":"https://pith.science/paper/ESVFPXQA"},"agent_actions":{"view_html":"https://pith.science/pith/ESVFPXQAVJMLSWRRACIHKHTEMU","download_json":"https://pith.science/pith/ESVFPXQAVJMLSWRRACIHKHTEMU.json","view_paper":"https://pith.science/paper/ESVFPXQA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.00670&json=true","fetch_graph":"https://pith.science/api/pith-number/ESVFPXQAVJMLSWRRACIHKHTEMU/graph.json","fetch_events":"https://pith.science/api/pith-number/ESVFPXQAVJMLSWRRACIHKHTEMU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ESVFPXQAVJMLSWRRACIHKHTEMU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ESVFPXQAVJMLSWRRACIHKHTEMU/action/storage_attestation","attest_author":"https://pith.science/pith/ESVFPXQAVJMLSWRRACIHKHTEMU/action/author_attestation","sign_citation":"https://pith.science/pith/ESVFPXQAVJMLSWRRACIHKHTEMU/action/citation_signature","submit_replication":"https://pith.science/pith/ESVFPXQAVJMLSWRRACIHKHTEMU/action/replication_record"}},"created_at":"2026-07-05T05:43:04.049979+00:00","updated_at":"2026-07-05T05:43:04.049979+00:00"}