{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:62OHN355X5OGPSLIG3NWWXBOMW","short_pith_number":"pith:62OHN355","schema_version":"1.0","canonical_sha256":"f69c76efbdbf5c67c96836db6b5c2e659e9393af7e48f6abe95f717ec8670638","source":{"kind":"arxiv","id":"2405.19673","version":2},"attestation_state":"computed","paper":{"title":"Bridging Model-Based Optimization and Generative Modeling via Conservative Fine-Tuning of Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Avantika Lal, Ehsan Hajiramezanali, Gabriele Scalia, G\\\"okcen Eraslan, Masatoshi Uehara, Sergey Levine, Tommaso Biancalani, Yulai Zhao","submitted_at":"2024-05-30T03:57:29Z","abstract_excerpt":"AI-driven design problems, such as DNA/protein sequence design, are commonly tackled from two angles: generative modeling, which efficiently captures the feasible design space (e.g., natural images or biological sequences), and model-based optimization, which utilizes reward models for extrapolation. To combine the strengths of both approaches, we adopt a hybrid method that fine-tunes cutting-edge diffusion models by optimizing reward models through RL. Although prior work has explored similar avenues, they primarily focus on scenarios where accurate reward models are accessible. In contrast, "},"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":"2405.19673","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-30T03:57:29Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"c83133c1d0d9f68016d61634420c8486e47365daff9a63d83b3e628482aac733","abstract_canon_sha256":"6847e6f866bbd649c2576048ad235aaa7a8c0ea45b049701bc79f23e420ad6e9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:25:55.086495Z","signature_b64":"WcqbIraiyBjMSllIg5ihXcqZdbY4MeN7UA5RqNixKg5sPMbLRWG0v4A2IOBrzCatmoU5txHW76OJpIY7/ydOCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f69c76efbdbf5c67c96836db6b5c2e659e9393af7e48f6abe95f717ec8670638","last_reissued_at":"2026-07-05T08:25:55.085974Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:25:55.085974Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bridging Model-Based Optimization and Generative Modeling via Conservative Fine-Tuning of Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Avantika Lal, Ehsan Hajiramezanali, Gabriele Scalia, G\\\"okcen Eraslan, Masatoshi Uehara, Sergey Levine, Tommaso Biancalani, Yulai Zhao","submitted_at":"2024-05-30T03:57:29Z","abstract_excerpt":"AI-driven design problems, such as DNA/protein sequence design, are commonly tackled from two angles: generative modeling, which efficiently captures the feasible design space (e.g., natural images or biological sequences), and model-based optimization, which utilizes reward models for extrapolation. To combine the strengths of both approaches, we adopt a hybrid method that fine-tunes cutting-edge diffusion models by optimizing reward models through RL. Although prior work has explored similar avenues, they primarily focus on scenarios where accurate reward models are accessible. In contrast, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.19673","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/2405.19673/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":"2405.19673","created_at":"2026-07-05T08:25:55.086043+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.19673v2","created_at":"2026-07-05T08:25:55.086043+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.19673","created_at":"2026-07-05T08:25:55.086043+00:00"},{"alias_kind":"pith_short_12","alias_value":"62OHN355X5OG","created_at":"2026-07-05T08:25:55.086043+00:00"},{"alias_kind":"pith_short_16","alias_value":"62OHN355X5OGPSLI","created_at":"2026-07-05T08:25:55.086043+00:00"},{"alias_kind":"pith_short_8","alias_value":"62OHN355","created_at":"2026-07-05T08:25:55.086043+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.06082","citing_title":"Offline Materials Optimization with CliqueFlowmer","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/62OHN355X5OGPSLIG3NWWXBOMW","json":"https://pith.science/pith/62OHN355X5OGPSLIG3NWWXBOMW.json","graph_json":"https://pith.science/api/pith-number/62OHN355X5OGPSLIG3NWWXBOMW/graph.json","events_json":"https://pith.science/api/pith-number/62OHN355X5OGPSLIG3NWWXBOMW/events.json","paper":"https://pith.science/paper/62OHN355"},"agent_actions":{"view_html":"https://pith.science/pith/62OHN355X5OGPSLIG3NWWXBOMW","download_json":"https://pith.science/pith/62OHN355X5OGPSLIG3NWWXBOMW.json","view_paper":"https://pith.science/paper/62OHN355","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.19673&json=true","fetch_graph":"https://pith.science/api/pith-number/62OHN355X5OGPSLIG3NWWXBOMW/graph.json","fetch_events":"https://pith.science/api/pith-number/62OHN355X5OGPSLIG3NWWXBOMW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/62OHN355X5OGPSLIG3NWWXBOMW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/62OHN355X5OGPSLIG3NWWXBOMW/action/storage_attestation","attest_author":"https://pith.science/pith/62OHN355X5OGPSLIG3NWWXBOMW/action/author_attestation","sign_citation":"https://pith.science/pith/62OHN355X5OGPSLIG3NWWXBOMW/action/citation_signature","submit_replication":"https://pith.science/pith/62OHN355X5OGPSLIG3NWWXBOMW/action/replication_record"}},"created_at":"2026-07-05T08:25:55.086043+00:00","updated_at":"2026-07-05T08:25:55.086043+00:00"}