{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:BKELIVT2E2O7EBJ7IXZ5YMTVKD","short_pith_number":"pith:BKELIVT2","schema_version":"1.0","canonical_sha256":"0a88b4567a269df2053f45f3dc327550f38350a3dc5dade0b13370f2a48a9551","source":{"kind":"arxiv","id":"2503.02039","version":2},"attestation_state":"computed","paper":{"title":"Dynamic Search for Inference-Time Alignment in Diffusion Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aviv Regev, Gabriele Scalia, Masatoshi Uehara, Sergey Levine, Shuiwang Ji, Tommaso Biancalani, Xiner Li, Xingyu Su","submitted_at":"2025-03-03T20:32:05Z","abstract_excerpt":"Diffusion models have shown promising generative capabilities across diverse domains, yet aligning their outputs with desired reward functions remains a challenge, particularly in cases where reward functions are non-differentiable. Some gradient-free guidance methods have been developed, but they often struggle to achieve optimal inference-time alignment. In this work, we newly frame inference-time alignment in diffusion as a search problem and propose Dynamic Search for Diffusion (DSearch), which subsamples from denoising processes and approximates intermediate node rewards. It also dynamica"},"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":"2503.02039","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-03T20:32:05Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8df22ed4ba53feb0b2ee8d31659fe191227396415019792c9dbc8e0a4ce877d0","abstract_canon_sha256":"a32e6cd818c318790d8f35e3ffe5da8cf73279f8c8d30e5168d717f43c588fc0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:14:27.838561Z","signature_b64":"VZ4+t6qppM3u65o0p+nZkzE7g6I3mtB9klDWG6fhumkcqm4RZlCDzQb/CMBzx1yBUKJxcNIuFDAz3UzKdBttAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0a88b4567a269df2053f45f3dc327550f38350a3dc5dade0b13370f2a48a9551","last_reissued_at":"2026-07-05T11:14:27.838047Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:14:27.838047Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dynamic Search for Inference-Time Alignment in Diffusion Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aviv Regev, Gabriele Scalia, Masatoshi Uehara, Sergey Levine, Shuiwang Ji, Tommaso Biancalani, Xiner Li, Xingyu Su","submitted_at":"2025-03-03T20:32:05Z","abstract_excerpt":"Diffusion models have shown promising generative capabilities across diverse domains, yet aligning their outputs with desired reward functions remains a challenge, particularly in cases where reward functions are non-differentiable. Some gradient-free guidance methods have been developed, but they often struggle to achieve optimal inference-time alignment. In this work, we newly frame inference-time alignment in diffusion as a search problem and propose Dynamic Search for Diffusion (DSearch), which subsamples from denoising processes and approximates intermediate node rewards. It also dynamica"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.02039","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/2503.02039/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":"2503.02039","created_at":"2026-07-05T11:14:27.838107+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.02039v2","created_at":"2026-07-05T11:14:27.838107+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.02039","created_at":"2026-07-05T11:14:27.838107+00:00"},{"alias_kind":"pith_short_12","alias_value":"BKELIVT2E2O7","created_at":"2026-07-05T11:14:27.838107+00:00"},{"alias_kind":"pith_short_16","alias_value":"BKELIVT2E2O7EBJ7","created_at":"2026-07-05T11:14:27.838107+00:00"},{"alias_kind":"pith_short_8","alias_value":"BKELIVT2","created_at":"2026-07-05T11:14:27.838107+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.13565","citing_title":"A2D2: Fine-Tuning Any-Length Discrete Diffusion for Adaptive Decoding","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08393","citing_title":"SMC-ITA: Sequential Monte Carlo Inference-Time Alignment for Video-to-Audio Generation","ref_index":19,"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":50,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28230","citing_title":"Proprio: Latent Self-Scoring and Inference-Time Refinement for Physically Plausible Video Generation","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16348","citing_title":"Flow-Direct: Feedback-Efficient and Reusable Guidance for Flow Models via Non-Parametric Guidance Field","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2512.23532","citing_title":"Iterative Inference-time Scaling with Adaptive Frequency Steering for Image Super-Resolution","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06779","citing_title":"VASR: Variance-Aware Systematic Resampling for Reward-Guided Diffusion","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08144","citing_title":"NoiseRater: Meta-Learned Noise Valuation for Diffusion Model Training","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06779","citing_title":"VASR: Variance-Aware Systematic Resampling for Reward-Guided Diffusion","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07950","citing_title":"Slowly Annealed Langevin Dynamics: Theory and Applications to Training-Free Guided Generation","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BKELIVT2E2O7EBJ7IXZ5YMTVKD","json":"https://pith.science/pith/BKELIVT2E2O7EBJ7IXZ5YMTVKD.json","graph_json":"https://pith.science/api/pith-number/BKELIVT2E2O7EBJ7IXZ5YMTVKD/graph.json","events_json":"https://pith.science/api/pith-number/BKELIVT2E2O7EBJ7IXZ5YMTVKD/events.json","paper":"https://pith.science/paper/BKELIVT2"},"agent_actions":{"view_html":"https://pith.science/pith/BKELIVT2E2O7EBJ7IXZ5YMTVKD","download_json":"https://pith.science/pith/BKELIVT2E2O7EBJ7IXZ5YMTVKD.json","view_paper":"https://pith.science/paper/BKELIVT2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.02039&json=true","fetch_graph":"https://pith.science/api/pith-number/BKELIVT2E2O7EBJ7IXZ5YMTVKD/graph.json","fetch_events":"https://pith.science/api/pith-number/BKELIVT2E2O7EBJ7IXZ5YMTVKD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BKELIVT2E2O7EBJ7IXZ5YMTVKD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BKELIVT2E2O7EBJ7IXZ5YMTVKD/action/storage_attestation","attest_author":"https://pith.science/pith/BKELIVT2E2O7EBJ7IXZ5YMTVKD/action/author_attestation","sign_citation":"https://pith.science/pith/BKELIVT2E2O7EBJ7IXZ5YMTVKD/action/citation_signature","submit_replication":"https://pith.science/pith/BKELIVT2E2O7EBJ7IXZ5YMTVKD/action/replication_record"}},"created_at":"2026-07-05T11:14:27.838107+00:00","updated_at":"2026-07-05T11:14:27.838107+00:00"}