{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XLVWTNHBBFSWVM2FBG3NOJTBFO","short_pith_number":"pith:XLVWTNHB","schema_version":"1.0","canonical_sha256":"baeb69b4e109656ab34509b6d726612bb4d1ac27c812169d6bca02c792364c19","source":{"kind":"arxiv","id":"2501.06655","version":1},"attestation_state":"computed","paper":{"title":"Personalized Preference Fine-tuning of Diffusion Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Anikait Singh, Jiaming Song, Linqi Zhou, Meihua Dang, Stefano Ermon","submitted_at":"2025-01-11T22:38:41Z","abstract_excerpt":"RLHF techniques like DPO can significantly improve the generation quality of text-to-image diffusion models. However, these methods optimize for a single reward that aligns model generation with population-level preferences, neglecting the nuances of individual users' beliefs or values. This lack of personalization limits the efficacy of these models. To bridge this gap, we introduce PPD, a multi-reward optimization objective that aligns diffusion models with personalized preferences. With PPD, a diffusion model learns the individual preferences of a population of users in a few-shot way, enab"},"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":"2501.06655","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-11T22:38:41Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"d373a74ee1196a0b31752288ab50e121f6b7b39c1d1c58c1a94ef551c9e5237a","abstract_canon_sha256":"4ccf74d1da588fe328c74e8fa12576004f72f127a7bb6e239feaffa6a308b5db"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:00:07.896397Z","signature_b64":"V2Z+qXcjW3lNUyZSuO5tqyQjpQssT7DdC0+N03ts62Hy+UqDPDx8pgXXq/xDanWpFJIYsNirL9DVA6h+lzImBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"baeb69b4e109656ab34509b6d726612bb4d1ac27c812169d6bca02c792364c19","last_reissued_at":"2026-07-05T10:00:07.895968Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:00:07.895968Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Personalized Preference Fine-tuning of Diffusion Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Anikait Singh, Jiaming Song, Linqi Zhou, Meihua Dang, Stefano Ermon","submitted_at":"2025-01-11T22:38:41Z","abstract_excerpt":"RLHF techniques like DPO can significantly improve the generation quality of text-to-image diffusion models. However, these methods optimize for a single reward that aligns model generation with population-level preferences, neglecting the nuances of individual users' beliefs or values. This lack of personalization limits the efficacy of these models. To bridge this gap, we introduce PPD, a multi-reward optimization objective that aligns diffusion models with personalized preferences. With PPD, a diffusion model learns the individual preferences of a population of users in a few-shot way, enab"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.06655","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/2501.06655/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":"2501.06655","created_at":"2026-07-05T10:00:07.896028+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.06655v1","created_at":"2026-07-05T10:00:07.896028+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.06655","created_at":"2026-07-05T10:00:07.896028+00:00"},{"alias_kind":"pith_short_12","alias_value":"XLVWTNHBBFSW","created_at":"2026-07-05T10:00:07.896028+00:00"},{"alias_kind":"pith_short_16","alias_value":"XLVWTNHBBFSWVM2F","created_at":"2026-07-05T10:00:07.896028+00:00"},{"alias_kind":"pith_short_8","alias_value":"XLVWTNHB","created_at":"2026-07-05T10:00:07.896028+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.09433","citing_title":"Offline Preference Optimization for Rectified Flow with Noise-Tracked Pairs","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XLVWTNHBBFSWVM2FBG3NOJTBFO","json":"https://pith.science/pith/XLVWTNHBBFSWVM2FBG3NOJTBFO.json","graph_json":"https://pith.science/api/pith-number/XLVWTNHBBFSWVM2FBG3NOJTBFO/graph.json","events_json":"https://pith.science/api/pith-number/XLVWTNHBBFSWVM2FBG3NOJTBFO/events.json","paper":"https://pith.science/paper/XLVWTNHB"},"agent_actions":{"view_html":"https://pith.science/pith/XLVWTNHBBFSWVM2FBG3NOJTBFO","download_json":"https://pith.science/pith/XLVWTNHBBFSWVM2FBG3NOJTBFO.json","view_paper":"https://pith.science/paper/XLVWTNHB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.06655&json=true","fetch_graph":"https://pith.science/api/pith-number/XLVWTNHBBFSWVM2FBG3NOJTBFO/graph.json","fetch_events":"https://pith.science/api/pith-number/XLVWTNHBBFSWVM2FBG3NOJTBFO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XLVWTNHBBFSWVM2FBG3NOJTBFO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XLVWTNHBBFSWVM2FBG3NOJTBFO/action/storage_attestation","attest_author":"https://pith.science/pith/XLVWTNHBBFSWVM2FBG3NOJTBFO/action/author_attestation","sign_citation":"https://pith.science/pith/XLVWTNHBBFSWVM2FBG3NOJTBFO/action/citation_signature","submit_replication":"https://pith.science/pith/XLVWTNHBBFSWVM2FBG3NOJTBFO/action/replication_record"}},"created_at":"2026-07-05T10:00:07.896028+00:00","updated_at":"2026-07-05T10:00:07.896028+00:00"}