{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:PUU6HP4LPZHRK2UGWMVKDOEUUU","short_pith_number":"pith:PUU6HP4L","schema_version":"1.0","canonical_sha256":"7d29e3bf8b7e4f156a86b32aa1b894a5334c1435ee5261a0091f05f6a797f93f","source":{"kind":"arxiv","id":"2503.08295","version":2},"attestation_state":"computed","paper":{"title":"Preference-Based Alignment of Discrete Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Davide Paglieri, Jude Wells, Tim Rockt\\\"aschel, Umberto Borso","submitted_at":"2025-03-11T11:07:35Z","abstract_excerpt":"Diffusion models have achieved state-of-the-art performance across multiple domains, with recent advancements extending their applicability to discrete data. However, aligning discrete diffusion models with task-specific preferences remains challenging, particularly in scenarios where explicit reward functions are unavailable. In this work, we introduce Discrete Diffusion DPO (D2-DPO), the first adaptation of Direct Preference Optimization (DPO) to discrete diffusion models formulated as continuous-time Markov chains. Our approach derives a novel loss function that directly fine-tunes the gene"},"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":"2503.08295","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-11T11:07:35Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"15955c16a6f7f80c795f7a856c19e3e98da42c24bd83f485757e438877ac2e7e","abstract_canon_sha256":"efd16cc3d7e37561b3d29ab94858ea6869f53e561612b63d0c88207da27b7eb9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:46:41.519145Z","signature_b64":"Gsi1PoZ6Sd171vtkyjqTokpEixBKuq3e92+PpdJUjguRLFjHwsvuTWNwHK+YlE+F1eIhVJ/bFq3XIMC35EJRBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7d29e3bf8b7e4f156a86b32aa1b894a5334c1435ee5261a0091f05f6a797f93f","last_reissued_at":"2026-07-05T10:46:41.518668Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:46:41.518668Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Preference-Based Alignment of Discrete Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Davide Paglieri, Jude Wells, Tim Rockt\\\"aschel, Umberto Borso","submitted_at":"2025-03-11T11:07:35Z","abstract_excerpt":"Diffusion models have achieved state-of-the-art performance across multiple domains, with recent advancements extending their applicability to discrete data. However, aligning discrete diffusion models with task-specific preferences remains challenging, particularly in scenarios where explicit reward functions are unavailable. In this work, we introduce Discrete Diffusion DPO (D2-DPO), the first adaptation of Direct Preference Optimization (DPO) to discrete diffusion models formulated as continuous-time Markov chains. Our approach derives a novel loss function that directly fine-tunes the gene"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.08295","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/2503.08295/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":"2503.08295","created_at":"2026-07-05T10:46:41.518725+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.08295v2","created_at":"2026-07-05T10:46:41.518725+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.08295","created_at":"2026-07-05T10:46:41.518725+00:00"},{"alias_kind":"pith_short_12","alias_value":"PUU6HP4LPZHR","created_at":"2026-07-05T10:46:41.518725+00:00"},{"alias_kind":"pith_short_16","alias_value":"PUU6HP4LPZHRK2UG","created_at":"2026-07-05T10:46:41.518725+00:00"},{"alias_kind":"pith_short_8","alias_value":"PUU6HP4L","created_at":"2026-07-05T10:46:41.518725+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.29150","citing_title":"Flow Reasoning Models: Scaling Reasoning Through Iterative Self-Refinement","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2509.08827","citing_title":"A Survey of Reinforcement Learning for Large Reasoning Models","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09433","citing_title":"Offline Preference Optimization for Rectified Flow with Noise-Tracked Pairs","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PUU6HP4LPZHRK2UGWMVKDOEUUU","json":"https://pith.science/pith/PUU6HP4LPZHRK2UGWMVKDOEUUU.json","graph_json":"https://pith.science/api/pith-number/PUU6HP4LPZHRK2UGWMVKDOEUUU/graph.json","events_json":"https://pith.science/api/pith-number/PUU6HP4LPZHRK2UGWMVKDOEUUU/events.json","paper":"https://pith.science/paper/PUU6HP4L"},"agent_actions":{"view_html":"https://pith.science/pith/PUU6HP4LPZHRK2UGWMVKDOEUUU","download_json":"https://pith.science/pith/PUU6HP4LPZHRK2UGWMVKDOEUUU.json","view_paper":"https://pith.science/paper/PUU6HP4L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.08295&json=true","fetch_graph":"https://pith.science/api/pith-number/PUU6HP4LPZHRK2UGWMVKDOEUUU/graph.json","fetch_events":"https://pith.science/api/pith-number/PUU6HP4LPZHRK2UGWMVKDOEUUU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PUU6HP4LPZHRK2UGWMVKDOEUUU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PUU6HP4LPZHRK2UGWMVKDOEUUU/action/storage_attestation","attest_author":"https://pith.science/pith/PUU6HP4LPZHRK2UGWMVKDOEUUU/action/author_attestation","sign_citation":"https://pith.science/pith/PUU6HP4LPZHRK2UGWMVKDOEUUU/action/citation_signature","submit_replication":"https://pith.science/pith/PUU6HP4LPZHRK2UGWMVKDOEUUU/action/replication_record"}},"created_at":"2026-07-05T10:46:41.518725+00:00","updated_at":"2026-07-05T10:46:41.518725+00:00"}