{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:IAZWZ22GIRXUUU4PQDV367RXBU","short_pith_number":"pith:IAZWZ22G","schema_version":"1.0","canonical_sha256":"40336ceb46446f4a538f80ebbf7e370d38fbb0ba47b612b8e0c109f0071f1fb0","source":{"kind":"arxiv","id":"2311.17946","version":1},"attestation_state":"computed","paper":{"title":"DreamSync: Aligning Text-to-Image Generation with Image Understanding Feedback","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Charles Herrmann, Cyrus Rashtchian, Da-Cheng Juan, Dana Alon, Deqing Fu, Jiao Sun, Ranjay Krishna, Royi Rassin, Sjoerd van Steenkiste, Su Wang, Yushi Hu","submitted_at":"2023-11-29T03:42:16Z","abstract_excerpt":"Despite their wide-spread success, Text-to-Image models (T2I) still struggle to produce images that are both aesthetically pleasing and faithful to the user's input text. We introduce DreamSync, a model-agnostic training algorithm by design that improves T2I models to be faithful to the text input. DreamSync builds off a recent insight from TIFA's evaluation framework -- that large vision-language models (VLMs) can effectively identify the fine-grained discrepancies between generated images and the text inputs. DreamSync uses this insight to train T2I models without any labeled data; it improv"},"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":"2311.17946","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2023-11-29T03:42:16Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"dfcf483dcece6e4a72edee760b130ea87b378a58b66d898b58d993da383d293c","abstract_canon_sha256":"32a82d656e9c07a2fa3b8acc68060649cc9b37d9e035e6833e041253b34f6cc4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:18:33.964901Z","signature_b64":"k7HcU6c6TAMWdwIJyreeTFnqavktD2Pwy2aFFtrh9qv9SjXiJJk7Kijbxu6vkewZgkZtnSmjx8HuBo3eFIQ7DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"40336ceb46446f4a538f80ebbf7e370d38fbb0ba47b612b8e0c109f0071f1fb0","last_reissued_at":"2026-07-05T07:18:33.964370Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:18:33.964370Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DreamSync: Aligning Text-to-Image Generation with Image Understanding Feedback","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Charles Herrmann, Cyrus Rashtchian, Da-Cheng Juan, Dana Alon, Deqing Fu, Jiao Sun, Ranjay Krishna, Royi Rassin, Sjoerd van Steenkiste, Su Wang, Yushi Hu","submitted_at":"2023-11-29T03:42:16Z","abstract_excerpt":"Despite their wide-spread success, Text-to-Image models (T2I) still struggle to produce images that are both aesthetically pleasing and faithful to the user's input text. We introduce DreamSync, a model-agnostic training algorithm by design that improves T2I models to be faithful to the text input. DreamSync builds off a recent insight from TIFA's evaluation framework -- that large vision-language models (VLMs) can effectively identify the fine-grained discrepancies between generated images and the text inputs. DreamSync uses this insight to train T2I models without any labeled data; it improv"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.17946","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/2311.17946/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":"2311.17946","created_at":"2026-07-05T07:18:33.964433+00:00"},{"alias_kind":"arxiv_version","alias_value":"2311.17946v1","created_at":"2026-07-05T07:18:33.964433+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.17946","created_at":"2026-07-05T07:18:33.964433+00:00"},{"alias_kind":"pith_short_12","alias_value":"IAZWZ22GIRXU","created_at":"2026-07-05T07:18:33.964433+00:00"},{"alias_kind":"pith_short_16","alias_value":"IAZWZ22GIRXUUU4P","created_at":"2026-07-05T07:18:33.964433+00:00"},{"alias_kind":"pith_short_8","alias_value":"IAZWZ22G","created_at":"2026-07-05T07:18:33.964433+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.20924","citing_title":"ELDiff: When Evidential Learning Meets Text-to-Image Diffusion","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2403.05135","citing_title":"ELLA: Equip Diffusion Models with LLM for Enhanced Semantic Alignment","ref_index":50,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IAZWZ22GIRXUUU4PQDV367RXBU","json":"https://pith.science/pith/IAZWZ22GIRXUUU4PQDV367RXBU.json","graph_json":"https://pith.science/api/pith-number/IAZWZ22GIRXUUU4PQDV367RXBU/graph.json","events_json":"https://pith.science/api/pith-number/IAZWZ22GIRXUUU4PQDV367RXBU/events.json","paper":"https://pith.science/paper/IAZWZ22G"},"agent_actions":{"view_html":"https://pith.science/pith/IAZWZ22GIRXUUU4PQDV367RXBU","download_json":"https://pith.science/pith/IAZWZ22GIRXUUU4PQDV367RXBU.json","view_paper":"https://pith.science/paper/IAZWZ22G","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2311.17946&json=true","fetch_graph":"https://pith.science/api/pith-number/IAZWZ22GIRXUUU4PQDV367RXBU/graph.json","fetch_events":"https://pith.science/api/pith-number/IAZWZ22GIRXUUU4PQDV367RXBU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IAZWZ22GIRXUUU4PQDV367RXBU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IAZWZ22GIRXUUU4PQDV367RXBU/action/storage_attestation","attest_author":"https://pith.science/pith/IAZWZ22GIRXUUU4PQDV367RXBU/action/author_attestation","sign_citation":"https://pith.science/pith/IAZWZ22GIRXUUU4PQDV367RXBU/action/citation_signature","submit_replication":"https://pith.science/pith/IAZWZ22GIRXUUU4PQDV367RXBU/action/replication_record"}},"created_at":"2026-07-05T07:18:33.964433+00:00","updated_at":"2026-07-05T07:18:33.964433+00:00"}