{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:SMGVAPLPGEE5MJ326BNHXCI22R","short_pith_number":"pith:SMGVAPLP","schema_version":"1.0","canonical_sha256":"930d503d6f3109d6277af05a7b891ad4561379923ea8c8a6cff6f19ea45dbda1","source":{"kind":"arxiv","id":"2506.03164","version":2},"attestation_state":"computed","paper":{"title":"Test-Time Scaling of Diffusion Models via Noise Trajectory Search","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Morteza Mardani, Vignav Ramesh","submitted_at":"2025-05-24T19:13:29Z","abstract_excerpt":"The iterative and stochastic nature of diffusion models enables test-time scaling, whereby spending additional compute during denoising generates higher-fidelity samples. Increasing the number of denoising steps is the primary scaling axis, but this yields quickly diminishing returns. Instead optimizing the noise trajectory--the sequence of injected noise vectors--is promising, as the specific noise realizations critically affect sample quality; but this is challenging due to a high-dimensional search space, complex noise-outcome interactions, and costly trajectory evaluations. We address this"},"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":"2506.03164","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-24T19:13:29Z","cross_cats_sorted":[],"title_canon_sha256":"3b20ccbe1bb13cb7639d99e9734725363be075a3fb9faf48a391b7e45ad06bb8","abstract_canon_sha256":"c7ab5ad633bc80f2d799656a71b2d8c17a8bf16f1f790e423b421c0dce832f2e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:06:13.339195Z","signature_b64":"+hJcCPCgGIK4BgQ4cloI4eguRks/OaQCpRimaRYvYxSfGDDR6ucdrdmAFkS5WbhBStcS6/O+gjaAb01LYpiiDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"930d503d6f3109d6277af05a7b891ad4561379923ea8c8a6cff6f19ea45dbda1","last_reissued_at":"2026-07-05T12:06:13.338674Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:06:13.338674Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Test-Time Scaling of Diffusion Models via Noise Trajectory Search","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Morteza Mardani, Vignav Ramesh","submitted_at":"2025-05-24T19:13:29Z","abstract_excerpt":"The iterative and stochastic nature of diffusion models enables test-time scaling, whereby spending additional compute during denoising generates higher-fidelity samples. Increasing the number of denoising steps is the primary scaling axis, but this yields quickly diminishing returns. Instead optimizing the noise trajectory--the sequence of injected noise vectors--is promising, as the specific noise realizations critically affect sample quality; but this is challenging due to a high-dimensional search space, complex noise-outcome interactions, and costly trajectory evaluations. We address this"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.03164","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/2506.03164/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":"2506.03164","created_at":"2026-07-05T12:06:13.338736+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.03164v2","created_at":"2026-07-05T12:06:13.338736+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.03164","created_at":"2026-07-05T12:06:13.338736+00:00"},{"alias_kind":"pith_short_12","alias_value":"SMGVAPLPGEE5","created_at":"2026-07-05T12:06:13.338736+00:00"},{"alias_kind":"pith_short_16","alias_value":"SMGVAPLPGEE5MJ32","created_at":"2026-07-05T12:06:13.338736+00:00"},{"alias_kind":"pith_short_8","alias_value":"SMGVAPLP","created_at":"2026-07-05T12:06:13.338736+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18066","citing_title":"NoiseTilt: Noise-Tilted Reverse Kernels for Diffusion Reward Alignment","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08231","citing_title":"Test-Time Scaling in Multimodal Foundation Models: A Comprehensive Survey of Generation and Reasoning","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18066","citing_title":"NoiseTilt: Noise-Tilted Reverse Kernels for Diffusion Reward Alignment","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23346","citing_title":"Contrastive Distribution Matching for Amortized Sequential Monte Carlo in Discrete Diffusion","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04461","citing_title":"Stream-T1: Test-Time Scaling for Streaming Video Generation","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SMGVAPLPGEE5MJ326BNHXCI22R","json":"https://pith.science/pith/SMGVAPLPGEE5MJ326BNHXCI22R.json","graph_json":"https://pith.science/api/pith-number/SMGVAPLPGEE5MJ326BNHXCI22R/graph.json","events_json":"https://pith.science/api/pith-number/SMGVAPLPGEE5MJ326BNHXCI22R/events.json","paper":"https://pith.science/paper/SMGVAPLP"},"agent_actions":{"view_html":"https://pith.science/pith/SMGVAPLPGEE5MJ326BNHXCI22R","download_json":"https://pith.science/pith/SMGVAPLPGEE5MJ326BNHXCI22R.json","view_paper":"https://pith.science/paper/SMGVAPLP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.03164&json=true","fetch_graph":"https://pith.science/api/pith-number/SMGVAPLPGEE5MJ326BNHXCI22R/graph.json","fetch_events":"https://pith.science/api/pith-number/SMGVAPLPGEE5MJ326BNHXCI22R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SMGVAPLPGEE5MJ326BNHXCI22R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SMGVAPLPGEE5MJ326BNHXCI22R/action/storage_attestation","attest_author":"https://pith.science/pith/SMGVAPLPGEE5MJ326BNHXCI22R/action/author_attestation","sign_citation":"https://pith.science/pith/SMGVAPLPGEE5MJ326BNHXCI22R/action/citation_signature","submit_replication":"https://pith.science/pith/SMGVAPLPGEE5MJ326BNHXCI22R/action/replication_record"}},"created_at":"2026-07-05T12:06:13.338736+00:00","updated_at":"2026-07-05T12:06:13.338736+00:00"}