{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:EKRAIXL3KUXJSD2F34NXYDARZQ","short_pith_number":"pith:EKRAIXL3","schema_version":"1.0","canonical_sha256":"22a2045d7b552e990f45df1b7c0c11cc3b6cfcc5712ca924a6d8e85f3a403d38","source":{"kind":"arxiv","id":"2502.14944","version":1},"attestation_state":"computed","paper":{"title":"Reward-Guided Iterative Refinement in Diffusion Models at Test-Time with Applications to Protein and DNA Design","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.NE","q-bio.QM","stat.ML"],"primary_cat":"cs.LG","authors_text":"Aviv Regev, Masatoshi Uehara, Sergey Levine, Shuiwang Ji, Tommaso Biancalani, Xiner Li, Xingyu Su, Yulai Zhao","submitted_at":"2025-02-20T17:48:45Z","abstract_excerpt":"To fully leverage the capabilities of diffusion models, we are often interested in optimizing downstream reward functions during inference. While numerous algorithms for reward-guided generation have been recently proposed due to their significance, current approaches predominantly focus on single-shot generation, transitioning from fully noised to denoised states. We propose a novel framework for inference-time reward optimization with diffusion models inspired by evolutionary algorithms. Our approach employs an iterative refinement process consisting of two steps in each iteration: noising a"},"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":"2502.14944","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-20T17:48:45Z","cross_cats_sorted":["cs.AI","cs.NE","q-bio.QM","stat.ML"],"title_canon_sha256":"e5cc874608d1e2e9212de9e5cce53847f5ebe3bb4e8ec316ef142a6d6dfa547e","abstract_canon_sha256":"80213853bd42c17aeb7da9724adce0776b12879d63f2921c57cdaa378d780f78"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:50:32.381223Z","signature_b64":"XyGBD99euDAifiEuv1wFhhf3jovLQpPzDMj2kmezJfsyrlBh22yxyAHaM/TvGgKRJu5vEomST04IV//ZuRgxDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"22a2045d7b552e990f45df1b7c0c11cc3b6cfcc5712ca924a6d8e85f3a403d38","last_reissued_at":"2026-07-05T10:50:32.380646Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:50:32.380646Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reward-Guided Iterative Refinement in Diffusion Models at Test-Time with Applications to Protein and DNA Design","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.NE","q-bio.QM","stat.ML"],"primary_cat":"cs.LG","authors_text":"Aviv Regev, Masatoshi Uehara, Sergey Levine, Shuiwang Ji, Tommaso Biancalani, Xiner Li, Xingyu Su, Yulai Zhao","submitted_at":"2025-02-20T17:48:45Z","abstract_excerpt":"To fully leverage the capabilities of diffusion models, we are often interested in optimizing downstream reward functions during inference. While numerous algorithms for reward-guided generation have been recently proposed due to their significance, current approaches predominantly focus on single-shot generation, transitioning from fully noised to denoised states. We propose a novel framework for inference-time reward optimization with diffusion models inspired by evolutionary algorithms. Our approach employs an iterative refinement process consisting of two steps in each iteration: noising a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.14944","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/2502.14944/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":"2502.14944","created_at":"2026-07-05T10:50:32.380709+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.14944v1","created_at":"2026-07-05T10:50:32.380709+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.14944","created_at":"2026-07-05T10:50:32.380709+00:00"},{"alias_kind":"pith_short_12","alias_value":"EKRAIXL3KUXJ","created_at":"2026-07-05T10:50:32.380709+00:00"},{"alias_kind":"pith_short_16","alias_value":"EKRAIXL3KUXJSD2F","created_at":"2026-07-05T10:50:32.380709+00:00"},{"alias_kind":"pith_short_8","alias_value":"EKRAIXL3","created_at":"2026-07-05T10:50:32.380709+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.00691","citing_title":"Generative Refinement for Low-Budget Black-Box Optimization","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2507.00480","citing_title":"Posterior Inference in Latent Space for Scalable Constrained Black-box Optimization","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11368","citing_title":"LPDP: Inference-Time Reward Control for Variable-Length DNA Generation with Edit Flows","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05829","citing_title":"MP2D: Constrained Monte Carlo Tree-Guided Diffusion for Multi-Objective Protein Sequence Design","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07253","citing_title":"LENS: Low-Frequency Eigen Noise Shaping for Efficient Diffusion Sampling","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EKRAIXL3KUXJSD2F34NXYDARZQ","json":"https://pith.science/pith/EKRAIXL3KUXJSD2F34NXYDARZQ.json","graph_json":"https://pith.science/api/pith-number/EKRAIXL3KUXJSD2F34NXYDARZQ/graph.json","events_json":"https://pith.science/api/pith-number/EKRAIXL3KUXJSD2F34NXYDARZQ/events.json","paper":"https://pith.science/paper/EKRAIXL3"},"agent_actions":{"view_html":"https://pith.science/pith/EKRAIXL3KUXJSD2F34NXYDARZQ","download_json":"https://pith.science/pith/EKRAIXL3KUXJSD2F34NXYDARZQ.json","view_paper":"https://pith.science/paper/EKRAIXL3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.14944&json=true","fetch_graph":"https://pith.science/api/pith-number/EKRAIXL3KUXJSD2F34NXYDARZQ/graph.json","fetch_events":"https://pith.science/api/pith-number/EKRAIXL3KUXJSD2F34NXYDARZQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EKRAIXL3KUXJSD2F34NXYDARZQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EKRAIXL3KUXJSD2F34NXYDARZQ/action/storage_attestation","attest_author":"https://pith.science/pith/EKRAIXL3KUXJSD2F34NXYDARZQ/action/author_attestation","sign_citation":"https://pith.science/pith/EKRAIXL3KUXJSD2F34NXYDARZQ/action/citation_signature","submit_replication":"https://pith.science/pith/EKRAIXL3KUXJSD2F34NXYDARZQ/action/replication_record"}},"created_at":"2026-07-05T10:50:32.380709+00:00","updated_at":"2026-07-05T10:50:32.380709+00:00"}