{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:CDUIQU4RCNELBNGXR2Y6HEJIHL","short_pith_number":"pith:CDUIQU4R","schema_version":"1.0","canonical_sha256":"10e88853911348b0b4d78eb1e391283adbcf1a2c9d80ba4c3e82fc6bdb8c82fd","source":{"kind":"arxiv","id":"2507.21391","version":2},"attestation_state":"computed","paper":{"title":"Multimodal LLMs as Customized Reward Models for Text-to-Image Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Branislav Kveton, Changyou Chen, Huaisheng Zhu, Jian Chen, Jiuxiang Gu, Ruiyi Zhang, Shijie Zhou, Yufan Zhou","submitted_at":"2025-07-28T23:52:53Z","abstract_excerpt":"We introduce LLaVA-Reward, an efficient reward model designed to automatically evaluate text-to-image (T2I) generations across multiple perspectives, leveraging pretrained multimodal large language models (MLLMs). Existing MLLM-based approaches require instruction-following data for supervised fine-tuning and evaluate generation quality on analyzing text response, which is time-consuming and difficult to train. To address this problem, we propose LLaVA-Reward, which directly utilizes the hidden states of MLLMs given text-image pairs. To enhance the bidirectional interaction between visual and "},"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":"2507.21391","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-07-28T23:52:53Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"ce192059e2cf4536fb9e840fec7cc49e8d6bf4b33ee8b1bc59e11148114fda68","abstract_canon_sha256":"835f72e0b3fe911c31403d339d905289e1e3f3e2c04f96a480127bc23930864f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:45:38.839461Z","signature_b64":"NQaARCJuEoNFM/1/+A30VPeMlXQJPhDrTNvlvcFBFT/ieUZClHfXWcPmqDKIQ74zKxmaKY25PputQRsXyeChBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"10e88853911348b0b4d78eb1e391283adbcf1a2c9d80ba4c3e82fc6bdb8c82fd","last_reissued_at":"2026-07-05T11:45:38.838942Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:45:38.838942Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Multimodal LLMs as Customized Reward Models for Text-to-Image Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Branislav Kveton, Changyou Chen, Huaisheng Zhu, Jian Chen, Jiuxiang Gu, Ruiyi Zhang, Shijie Zhou, Yufan Zhou","submitted_at":"2025-07-28T23:52:53Z","abstract_excerpt":"We introduce LLaVA-Reward, an efficient reward model designed to automatically evaluate text-to-image (T2I) generations across multiple perspectives, leveraging pretrained multimodal large language models (MLLMs). Existing MLLM-based approaches require instruction-following data for supervised fine-tuning and evaluate generation quality on analyzing text response, which is time-consuming and difficult to train. To address this problem, we propose LLaVA-Reward, which directly utilizes the hidden states of MLLMs given text-image pairs. To enhance the bidirectional interaction between visual and "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.21391","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/2507.21391/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":"2507.21391","created_at":"2026-07-05T11:45:38.839002+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.21391v2","created_at":"2026-07-05T11:45:38.839002+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.21391","created_at":"2026-07-05T11:45:38.839002+00:00"},{"alias_kind":"pith_short_12","alias_value":"CDUIQU4RCNEL","created_at":"2026-07-05T11:45:38.839002+00:00"},{"alias_kind":"pith_short_16","alias_value":"CDUIQU4RCNELBNGX","created_at":"2026-07-05T11:45:38.839002+00:00"},{"alias_kind":"pith_short_8","alias_value":"CDUIQU4R","created_at":"2026-07-05T11:45:38.839002+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CDUIQU4RCNELBNGXR2Y6HEJIHL","json":"https://pith.science/pith/CDUIQU4RCNELBNGXR2Y6HEJIHL.json","graph_json":"https://pith.science/api/pith-number/CDUIQU4RCNELBNGXR2Y6HEJIHL/graph.json","events_json":"https://pith.science/api/pith-number/CDUIQU4RCNELBNGXR2Y6HEJIHL/events.json","paper":"https://pith.science/paper/CDUIQU4R"},"agent_actions":{"view_html":"https://pith.science/pith/CDUIQU4RCNELBNGXR2Y6HEJIHL","download_json":"https://pith.science/pith/CDUIQU4RCNELBNGXR2Y6HEJIHL.json","view_paper":"https://pith.science/paper/CDUIQU4R","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.21391&json=true","fetch_graph":"https://pith.science/api/pith-number/CDUIQU4RCNELBNGXR2Y6HEJIHL/graph.json","fetch_events":"https://pith.science/api/pith-number/CDUIQU4RCNELBNGXR2Y6HEJIHL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CDUIQU4RCNELBNGXR2Y6HEJIHL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CDUIQU4RCNELBNGXR2Y6HEJIHL/action/storage_attestation","attest_author":"https://pith.science/pith/CDUIQU4RCNELBNGXR2Y6HEJIHL/action/author_attestation","sign_citation":"https://pith.science/pith/CDUIQU4RCNELBNGXR2Y6HEJIHL/action/citation_signature","submit_replication":"https://pith.science/pith/CDUIQU4RCNELBNGXR2Y6HEJIHL/action/replication_record"}},"created_at":"2026-07-05T11:45:38.839002+00:00","updated_at":"2026-07-05T11:45:38.839002+00:00"}