{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:VSXSX4Q4WKQQ6CKPQBGW2FHE64","short_pith_number":"pith:VSXSX4Q4","schema_version":"1.0","canonical_sha256":"acaf2bf21cb2a10f094f804d6d14e4f718bc5bb50f027834e1fd26368e22fdd9","source":{"kind":"arxiv","id":"2506.12036","version":3},"attestation_state":"computed","paper":{"title":"A Minimalist Method for Fine-tuning Text-to-Image Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Pacal Poupart, Suraj Kothawade, William Loh, Yanting Miao","submitted_at":"2025-05-23T00:01:52Z","abstract_excerpt":"Recent work uses reinforcement learning (RL) to fine-tune text-to-image diffusion models, improving text-image alignment and sample quality. However, existing approaches introduce unnecessary complexity: they cache the full sampling trajectory, depend on differentiable reward models or large preference datasets, or require specialized guidance techniques. Motivated by the \"golden noise\" hypothesis -- that certain initial noise samples can consistently yield superior alignment -- we introduce Noise PPO, a minimalist RL algorithm that leaves the pre-trained diffusion model entirely frozen and le"},"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":"2506.12036","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-23T00:01:52Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"9d1983fff9f5dcecb877a4ae48e03808bc272b6b617ae0bacad0d341fe9b044f","abstract_canon_sha256":"9f1fe8ed53f59789d34965ffd311998c757592881236b2f957b589f2b22cdce7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:30:01.845876Z","signature_b64":"9JWmpwrVDKZ3QwZtI1BKiClOsRWkzLw4tju5lAB/kxNUzjXsXJpWYnI7vP7roWpFa1SQVO3T3QzqXi532psJCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"acaf2bf21cb2a10f094f804d6d14e4f718bc5bb50f027834e1fd26368e22fdd9","last_reissued_at":"2026-07-05T11:30:01.845306Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:30:01.845306Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Minimalist Method for Fine-tuning Text-to-Image Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Pacal Poupart, Suraj Kothawade, William Loh, Yanting Miao","submitted_at":"2025-05-23T00:01:52Z","abstract_excerpt":"Recent work uses reinforcement learning (RL) to fine-tune text-to-image diffusion models, improving text-image alignment and sample quality. However, existing approaches introduce unnecessary complexity: they cache the full sampling trajectory, depend on differentiable reward models or large preference datasets, or require specialized guidance techniques. Motivated by the \"golden noise\" hypothesis -- that certain initial noise samples can consistently yield superior alignment -- we introduce Noise PPO, a minimalist RL algorithm that leaves the pre-trained diffusion model entirely frozen and le"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.12036","kind":"arxiv","version":3},"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/2506.12036/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":"2506.12036","created_at":"2026-07-05T11:30:01.845401+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.12036v3","created_at":"2026-07-05T11:30:01.845401+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.12036","created_at":"2026-07-05T11:30:01.845401+00:00"},{"alias_kind":"pith_short_12","alias_value":"VSXSX4Q4WKQQ","created_at":"2026-07-05T11:30:01.845401+00:00"},{"alias_kind":"pith_short_16","alias_value":"VSXSX4Q4WKQQ6CKP","created_at":"2026-07-05T11:30:01.845401+00:00"},{"alias_kind":"pith_short_8","alias_value":"VSXSX4Q4","created_at":"2026-07-05T11:30:01.845401+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.15757","citing_title":"You've Got a Golden Ticket: Improving Generative Robot Policies With A Single Noise Vector","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VSXSX4Q4WKQQ6CKPQBGW2FHE64","json":"https://pith.science/pith/VSXSX4Q4WKQQ6CKPQBGW2FHE64.json","graph_json":"https://pith.science/api/pith-number/VSXSX4Q4WKQQ6CKPQBGW2FHE64/graph.json","events_json":"https://pith.science/api/pith-number/VSXSX4Q4WKQQ6CKPQBGW2FHE64/events.json","paper":"https://pith.science/paper/VSXSX4Q4"},"agent_actions":{"view_html":"https://pith.science/pith/VSXSX4Q4WKQQ6CKPQBGW2FHE64","download_json":"https://pith.science/pith/VSXSX4Q4WKQQ6CKPQBGW2FHE64.json","view_paper":"https://pith.science/paper/VSXSX4Q4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.12036&json=true","fetch_graph":"https://pith.science/api/pith-number/VSXSX4Q4WKQQ6CKPQBGW2FHE64/graph.json","fetch_events":"https://pith.science/api/pith-number/VSXSX4Q4WKQQ6CKPQBGW2FHE64/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VSXSX4Q4WKQQ6CKPQBGW2FHE64/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VSXSX4Q4WKQQ6CKPQBGW2FHE64/action/storage_attestation","attest_author":"https://pith.science/pith/VSXSX4Q4WKQQ6CKPQBGW2FHE64/action/author_attestation","sign_citation":"https://pith.science/pith/VSXSX4Q4WKQQ6CKPQBGW2FHE64/action/citation_signature","submit_replication":"https://pith.science/pith/VSXSX4Q4WKQQ6CKPQBGW2FHE64/action/replication_record"}},"created_at":"2026-07-05T11:30:01.845401+00:00","updated_at":"2026-07-05T11:30:01.845401+00:00"}