{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:YINVLGZLH23ZIWV5OU5VHDGWCY","short_pith_number":"pith:YINVLGZL","schema_version":"1.0","canonical_sha256":"c21b559b2b3eb7945abd753b538cd6162b1ed288a23b808241c116a5ebefb194","source":{"kind":"arxiv","id":"2506.07963","version":3},"attestation_state":"computed","paper":{"title":"SUDER: Self-Improving Unified Large Multimodal Models for Understanding and Generation with Dual Self-Rewards","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.AI","authors_text":"Guanzhong Wang, Ji-Rong Wen, Jixiang Hong, Rui Yan, Yi Liu, Yiran Zhang","submitted_at":"2025-06-09T17:38:45Z","abstract_excerpt":"Building upon large language models (LLMs), recent large multimodal models (LMMs) unify cross-model understanding and generation into a single framework. However, LMMs still struggle to achieve accurate vision-language alignment, prone to generating text responses contradicting the visual input or failing to follow the text-to-image prompts. Current solutions require external supervision (e.g., human feedback or reward models) and only address unidirectional tasks-either understanding or generation. In this work, based on the observation that understanding and generation are naturally inverse "},"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":"2506.07963","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2025-06-09T17:38:45Z","cross_cats_sorted":["cs.CL","cs.CV"],"title_canon_sha256":"eacee94a89536a9fbeef33e25e6967703dd9e8bc78390f0821bead640bdb875c","abstract_canon_sha256":"f6fdfc510d60f43c481ec38a085b53db554f791e136652ec99d47b48e757c935"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:06:48.373351Z","signature_b64":"gFSf/z+bxrBhbzJZk4GR1W0KqMFBxbrEzpojzbMBPeONdhMvmPGm6C6hxJc0+xnHJQ6Ptyg6MHnIes4kouLODQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c21b559b2b3eb7945abd753b538cd6162b1ed288a23b808241c116a5ebefb194","last_reissued_at":"2026-07-05T12:06:48.372861Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:06:48.372861Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SUDER: Self-Improving Unified Large Multimodal Models for Understanding and Generation with Dual Self-Rewards","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.AI","authors_text":"Guanzhong Wang, Ji-Rong Wen, Jixiang Hong, Rui Yan, Yi Liu, Yiran Zhang","submitted_at":"2025-06-09T17:38:45Z","abstract_excerpt":"Building upon large language models (LLMs), recent large multimodal models (LMMs) unify cross-model understanding and generation into a single framework. However, LMMs still struggle to achieve accurate vision-language alignment, prone to generating text responses contradicting the visual input or failing to follow the text-to-image prompts. Current solutions require external supervision (e.g., human feedback or reward models) and only address unidirectional tasks-either understanding or generation. In this work, based on the observation that understanding and generation are naturally inverse "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.07963","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/2506.07963/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":"2506.07963","created_at":"2026-07-05T12:06:48.372919+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.07963v3","created_at":"2026-07-05T12:06:48.372919+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.07963","created_at":"2026-07-05T12:06:48.372919+00:00"},{"alias_kind":"pith_short_12","alias_value":"YINVLGZLH23Z","created_at":"2026-07-05T12:06:48.372919+00:00"},{"alias_kind":"pith_short_16","alias_value":"YINVLGZLH23ZIWV5","created_at":"2026-07-05T12:06:48.372919+00:00"},{"alias_kind":"pith_short_8","alias_value":"YINVLGZL","created_at":"2026-07-05T12:06:48.372919+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27376","citing_title":"Ask, Solve, Generate: Self-Evolving Unified Multimodal Understanding and Generation via Self-Consistency Rewards","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2509.08827","citing_title":"A Survey of Reinforcement Learning for Large Reasoning Models","ref_index":197,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YINVLGZLH23ZIWV5OU5VHDGWCY","json":"https://pith.science/pith/YINVLGZLH23ZIWV5OU5VHDGWCY.json","graph_json":"https://pith.science/api/pith-number/YINVLGZLH23ZIWV5OU5VHDGWCY/graph.json","events_json":"https://pith.science/api/pith-number/YINVLGZLH23ZIWV5OU5VHDGWCY/events.json","paper":"https://pith.science/paper/YINVLGZL"},"agent_actions":{"view_html":"https://pith.science/pith/YINVLGZLH23ZIWV5OU5VHDGWCY","download_json":"https://pith.science/pith/YINVLGZLH23ZIWV5OU5VHDGWCY.json","view_paper":"https://pith.science/paper/YINVLGZL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.07963&json=true","fetch_graph":"https://pith.science/api/pith-number/YINVLGZLH23ZIWV5OU5VHDGWCY/graph.json","fetch_events":"https://pith.science/api/pith-number/YINVLGZLH23ZIWV5OU5VHDGWCY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YINVLGZLH23ZIWV5OU5VHDGWCY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YINVLGZLH23ZIWV5OU5VHDGWCY/action/storage_attestation","attest_author":"https://pith.science/pith/YINVLGZLH23ZIWV5OU5VHDGWCY/action/author_attestation","sign_citation":"https://pith.science/pith/YINVLGZLH23ZIWV5OU5VHDGWCY/action/citation_signature","submit_replication":"https://pith.science/pith/YINVLGZLH23ZIWV5OU5VHDGWCY/action/replication_record"}},"created_at":"2026-07-05T12:06:48.372919+00:00","updated_at":"2026-07-05T12:06:48.372919+00:00"}