{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:CI54EDUSMAGWUWJL3JDEBDNZTI","short_pith_number":"pith:CI54EDUS","schema_version":"1.0","canonical_sha256":"123bc20e92600d6a592bda46408db99a00117189dc67ad652ea682d67a075e11","source":{"kind":"arxiv","id":"2608.07006","version":1},"attestation_state":"computed","paper":{"title":"Does More Retrieved Evidence Help Visual Retrieval-Augmented Generation with Diffusion Language Models?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Chen Gao, Jiankun Wang, Jiaxin Bai, Xingcheng Fu, Yisen Gao, Ziwei Zhang","submitted_at":"2026-08-07T09:20:40Z","abstract_excerpt":"Visual retrieval-augmented generation (RAG) commonly expands the retrieved evidence set to improve answer-page coverage, implicitly assuming that all available evidence should be passed to the generator. We show that this assumption does not hold for diffusion language models (DLMs): retrieving more pages increases answer-page recall, whereas unconditionally passing all retrieved pages to the generator often reduces answer accuracy, primarily because of semantic conflict. A latent-source analysis explains this mismatch through source-coherence loss in parallel denoising, where position-wise pr"},"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":"2608.07006","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-08-07T09:20:40Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"eefd58d7b409076d39701a2be68f3b970675a3f45763ad5f85a338791deccd3f","abstract_canon_sha256":"72699627ef86407cd2d1d9ccc8151744101019113775b9d3e8ef5b9d723da45c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-10T01:12:49.776040Z","signature_b64":"2xHrtKjhTkuSzfmEtaUarImSuXWV0ATQ0942X9/GN428ErcHRRPq+nsiF7qPmspcYfNcXdshgo4dRgY2WQ9rBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"123bc20e92600d6a592bda46408db99a00117189dc67ad652ea682d67a075e11","last_reissued_at":"2026-08-10T01:12:49.772961Z","signature_status":"signed_v1","first_computed_at":"2026-08-10T01:12:49.772961Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Does More Retrieved Evidence Help Visual Retrieval-Augmented Generation with Diffusion Language Models?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Chen Gao, Jiankun Wang, Jiaxin Bai, Xingcheng Fu, Yisen Gao, Ziwei Zhang","submitted_at":"2026-08-07T09:20:40Z","abstract_excerpt":"Visual retrieval-augmented generation (RAG) commonly expands the retrieved evidence set to improve answer-page coverage, implicitly assuming that all available evidence should be passed to the generator. We show that this assumption does not hold for diffusion language models (DLMs): retrieving more pages increases answer-page recall, whereas unconditionally passing all retrieved pages to the generator often reduces answer accuracy, primarily because of semantic conflict. A latent-source analysis explains this mismatch through source-coherence loss in parallel denoising, where position-wise pr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.07006","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/2608.07006/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":"2608.07006","created_at":"2026-08-10T01:12:49.774572+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.07006v1","created_at":"2026-08-10T01:12:49.774572+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.07006","created_at":"2026-08-10T01:12:49.774572+00:00"},{"alias_kind":"pith_short_12","alias_value":"CI54EDUSMAGW","created_at":"2026-08-10T01:12:49.774572+00:00"},{"alias_kind":"pith_short_16","alias_value":"CI54EDUSMAGWUWJL","created_at":"2026-08-10T01:12:49.774572+00:00"},{"alias_kind":"pith_short_8","alias_value":"CI54EDUS","created_at":"2026-08-10T01:12:49.774572+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/CI54EDUSMAGWUWJL3JDEBDNZTI","json":"https://pith.science/pith/CI54EDUSMAGWUWJL3JDEBDNZTI.json","graph_json":"https://pith.science/api/pith-number/CI54EDUSMAGWUWJL3JDEBDNZTI/graph.json","events_json":"https://pith.science/api/pith-number/CI54EDUSMAGWUWJL3JDEBDNZTI/events.json","paper":"https://pith.science/paper/CI54EDUS"},"agent_actions":{"view_html":"https://pith.science/pith/CI54EDUSMAGWUWJL3JDEBDNZTI","download_json":"https://pith.science/pith/CI54EDUSMAGWUWJL3JDEBDNZTI.json","view_paper":"https://pith.science/paper/CI54EDUS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.07006&json=true","fetch_graph":"https://pith.science/api/pith-number/CI54EDUSMAGWUWJL3JDEBDNZTI/graph.json","fetch_events":"https://pith.science/api/pith-number/CI54EDUSMAGWUWJL3JDEBDNZTI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CI54EDUSMAGWUWJL3JDEBDNZTI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CI54EDUSMAGWUWJL3JDEBDNZTI/action/storage_attestation","attest_author":"https://pith.science/pith/CI54EDUSMAGWUWJL3JDEBDNZTI/action/author_attestation","sign_citation":"https://pith.science/pith/CI54EDUSMAGWUWJL3JDEBDNZTI/action/citation_signature","submit_replication":"https://pith.science/pith/CI54EDUSMAGWUWJL3JDEBDNZTI/action/replication_record"}},"created_at":"2026-08-10T01:12:49.774572+00:00","updated_at":"2026-08-10T01:12:49.774572+00:00"}