{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZSZSUH7CKTEXLGHHHWRFJIP7GX","short_pith_number":"pith:ZSZSUH7C","schema_version":"1.0","canonical_sha256":"ccb32a1fe254c97598e73da254a1ff35d901bd4c645f520284b5b8209794de86","source":{"kind":"arxiv","id":"2507.19002","version":1},"attestation_state":"computed","paper":{"title":"Enhancing Reward Models for High-quality Image Generation: Beyond Text-Image Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bing Su, Ji-Rong Wen, Tao Liang, Tianyu Zhang, Wenyi Mo, Yalong Bai, Ying Ba","submitted_at":"2025-07-25T07:01:50Z","abstract_excerpt":"Contemporary image generation systems have achieved high fidelity and superior aesthetic quality beyond basic text-image alignment. However, existing evaluation frameworks have failed to evolve in parallel. This study reveals that human preference reward models fine-tuned based on CLIP and BLIP architectures have inherent flaws: they inappropriately assign low scores to images with rich details and high aesthetic value, creating a significant discrepancy with actual human aesthetic preferences. To address this issue, we design a novel evaluation score, ICT (Image-Contained-Text) score, that ac"},"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":"2507.19002","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-07-25T07:01:50Z","cross_cats_sorted":[],"title_canon_sha256":"d942ba238094d7c681dac1f44b9ac8546c0f597aa09e7afb92aee306ec5513d8","abstract_canon_sha256":"dee2d047fbd40b05e5f95efa17b9ff8c80b2d8639a39cdd8eb0b7859f7244c8a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:43:17.652245Z","signature_b64":"LUAjxcpgKngJD09mNIVaf36mI5NpuFP2rk9PawKREn9ySUPP5kJ/bzshsgtBLKfMU1Qx7IeRCCa4pX343DE/AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ccb32a1fe254c97598e73da254a1ff35d901bd4c645f520284b5b8209794de86","last_reissued_at":"2026-07-05T11:43:17.651846Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:43:17.651846Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Enhancing Reward Models for High-quality Image Generation: Beyond Text-Image Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bing Su, Ji-Rong Wen, Tao Liang, Tianyu Zhang, Wenyi Mo, Yalong Bai, Ying Ba","submitted_at":"2025-07-25T07:01:50Z","abstract_excerpt":"Contemporary image generation systems have achieved high fidelity and superior aesthetic quality beyond basic text-image alignment. However, existing evaluation frameworks have failed to evolve in parallel. This study reveals that human preference reward models fine-tuned based on CLIP and BLIP architectures have inherent flaws: they inappropriately assign low scores to images with rich details and high aesthetic value, creating a significant discrepancy with actual human aesthetic preferences. To address this issue, we design a novel evaluation score, ICT (Image-Contained-Text) score, that ac"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.19002","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/2507.19002/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":"2507.19002","created_at":"2026-07-05T11:43:17.651906+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.19002v1","created_at":"2026-07-05T11:43:17.651906+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.19002","created_at":"2026-07-05T11:43:17.651906+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZSZSUH7CKTEX","created_at":"2026-07-05T11:43:17.651906+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZSZSUH7CKTEXLGHH","created_at":"2026-07-05T11:43:17.651906+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZSZSUH7C","created_at":"2026-07-05T11:43:17.651906+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08147","citing_title":"Biological Reasoning-Informed Regression for Interpretable Regulatory DNA Activity Prediction","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZSZSUH7CKTEXLGHHHWRFJIP7GX","json":"https://pith.science/pith/ZSZSUH7CKTEXLGHHHWRFJIP7GX.json","graph_json":"https://pith.science/api/pith-number/ZSZSUH7CKTEXLGHHHWRFJIP7GX/graph.json","events_json":"https://pith.science/api/pith-number/ZSZSUH7CKTEXLGHHHWRFJIP7GX/events.json","paper":"https://pith.science/paper/ZSZSUH7C"},"agent_actions":{"view_html":"https://pith.science/pith/ZSZSUH7CKTEXLGHHHWRFJIP7GX","download_json":"https://pith.science/pith/ZSZSUH7CKTEXLGHHHWRFJIP7GX.json","view_paper":"https://pith.science/paper/ZSZSUH7C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.19002&json=true","fetch_graph":"https://pith.science/api/pith-number/ZSZSUH7CKTEXLGHHHWRFJIP7GX/graph.json","fetch_events":"https://pith.science/api/pith-number/ZSZSUH7CKTEXLGHHHWRFJIP7GX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZSZSUH7CKTEXLGHHHWRFJIP7GX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZSZSUH7CKTEXLGHHHWRFJIP7GX/action/storage_attestation","attest_author":"https://pith.science/pith/ZSZSUH7CKTEXLGHHHWRFJIP7GX/action/author_attestation","sign_citation":"https://pith.science/pith/ZSZSUH7CKTEXLGHHHWRFJIP7GX/action/citation_signature","submit_replication":"https://pith.science/pith/ZSZSUH7CKTEXLGHHHWRFJIP7GX/action/replication_record"}},"created_at":"2026-07-05T11:43:17.651906+00:00","updated_at":"2026-07-05T11:43:17.651906+00:00"}