{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:4W3SWN7JA3H4OTXWWJQGC6EQUX","short_pith_number":"pith:4W3SWN7J","schema_version":"1.0","canonical_sha256":"e5b72b37e906cfc74ef6b260617890a5e6ef3e5d41e5ff50df94a2d04aaa52bc","source":{"kind":"arxiv","id":"2107.02423","version":2},"attestation_state":"computed","paper":{"title":"Improving Text-to-Image Synthesis Using Contrastive Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hui Ye, Martin Takac, Rajshekhar Sunderraman, Shihao Ji, Xiulong Yang","submitted_at":"2021-07-06T06:43:31Z","abstract_excerpt":"The goal of text-to-image synthesis is to generate a visually realistic image that matches a given text description. In practice, the captions annotated by humans for the same image have large variance in terms of contents and the choice of words. The linguistic discrepancy between the captions of the identical image leads to the synthetic images deviating from the ground truth. To address this issue, we propose a contrastive learning approach to improve the quality and enhance the semantic consistency of synthetic images. In the pretraining stage, we utilize the contrastive learning approach "},"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":"2107.02423","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-07-06T06:43:31Z","cross_cats_sorted":[],"title_canon_sha256":"735bd70f5f2373ba1446b9d11047067629545cce69b44101bfbb9cd326f90c08","abstract_canon_sha256":"c0075d94b3b6223029e876a20875e851d54d44658542e3cae2d6bea70376118e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:35:29.983301Z","signature_b64":"l5qwDhMdgpDYDmBVjAdFiEa9Vjv9IhRgG8lRoqjlKrLLC3riEPLuREu2GkO68ua62LhmvGrz2cEZbvKtO06gAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e5b72b37e906cfc74ef6b260617890a5e6ef3e5d41e5ff50df94a2d04aaa52bc","last_reissued_at":"2026-07-05T03:35:29.982883Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:35:29.982883Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Improving Text-to-Image Synthesis Using Contrastive Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hui Ye, Martin Takac, Rajshekhar Sunderraman, Shihao Ji, Xiulong Yang","submitted_at":"2021-07-06T06:43:31Z","abstract_excerpt":"The goal of text-to-image synthesis is to generate a visually realistic image that matches a given text description. In practice, the captions annotated by humans for the same image have large variance in terms of contents and the choice of words. The linguistic discrepancy between the captions of the identical image leads to the synthetic images deviating from the ground truth. To address this issue, we propose a contrastive learning approach to improve the quality and enhance the semantic consistency of synthetic images. In the pretraining stage, we utilize the contrastive learning approach "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2107.02423","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/2107.02423/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":"2107.02423","created_at":"2026-07-05T03:35:29.982940+00:00"},{"alias_kind":"arxiv_version","alias_value":"2107.02423v2","created_at":"2026-07-05T03:35:29.982940+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2107.02423","created_at":"2026-07-05T03:35:29.982940+00:00"},{"alias_kind":"pith_short_12","alias_value":"4W3SWN7JA3H4","created_at":"2026-07-05T03:35:29.982940+00:00"},{"alias_kind":"pith_short_16","alias_value":"4W3SWN7JA3H4OTXW","created_at":"2026-07-05T03:35:29.982940+00:00"},{"alias_kind":"pith_short_8","alias_value":"4W3SWN7J","created_at":"2026-07-05T03:35:29.982940+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2205.11487","citing_title":"Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding","ref_index":78,"is_internal_anchor":false},{"citing_arxiv_id":"2208.01618","citing_title":"An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2112.10741","citing_title":"GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Models","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2204.06125","citing_title":"Hierarchical Text-Conditional Image Generation with CLIP Latents","ref_index":58,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4W3SWN7JA3H4OTXWWJQGC6EQUX","json":"https://pith.science/pith/4W3SWN7JA3H4OTXWWJQGC6EQUX.json","graph_json":"https://pith.science/api/pith-number/4W3SWN7JA3H4OTXWWJQGC6EQUX/graph.json","events_json":"https://pith.science/api/pith-number/4W3SWN7JA3H4OTXWWJQGC6EQUX/events.json","paper":"https://pith.science/paper/4W3SWN7J"},"agent_actions":{"view_html":"https://pith.science/pith/4W3SWN7JA3H4OTXWWJQGC6EQUX","download_json":"https://pith.science/pith/4W3SWN7JA3H4OTXWWJQGC6EQUX.json","view_paper":"https://pith.science/paper/4W3SWN7J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2107.02423&json=true","fetch_graph":"https://pith.science/api/pith-number/4W3SWN7JA3H4OTXWWJQGC6EQUX/graph.json","fetch_events":"https://pith.science/api/pith-number/4W3SWN7JA3H4OTXWWJQGC6EQUX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4W3SWN7JA3H4OTXWWJQGC6EQUX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4W3SWN7JA3H4OTXWWJQGC6EQUX/action/storage_attestation","attest_author":"https://pith.science/pith/4W3SWN7JA3H4OTXWWJQGC6EQUX/action/author_attestation","sign_citation":"https://pith.science/pith/4W3SWN7JA3H4OTXWWJQGC6EQUX/action/citation_signature","submit_replication":"https://pith.science/pith/4W3SWN7JA3H4OTXWWJQGC6EQUX/action/replication_record"}},"created_at":"2026-07-05T03:35:29.982940+00:00","updated_at":"2026-07-05T03:35:29.982940+00:00"}