{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:F6KXB3GHVCJTB67UFEFX2UBRAN","short_pith_number":"pith:F6KXB3GH","schema_version":"1.0","canonical_sha256":"2f9570ecc7a89330fbf4290b7d5031035d49120950fdba38403c4ebe69d9abe7","source":{"kind":"arxiv","id":"2502.16872","version":1},"attestation_state":"computed","paper":{"title":"Mitigating Hallucinations in Diffusion Models through Adaptive Attention Modulation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Abhinav Shrivastava, Trevine Oorloff, Yaser Yacoob","submitted_at":"2025-02-24T06:19:54Z","abstract_excerpt":"Diffusion models, while increasingly adept at generating realistic images, are notably hindered by hallucinations -- unrealistic or incorrect features inconsistent with the trained data distribution. In this work, we propose Adaptive Attention Modulation (AAM), a novel approach to mitigate hallucinations by analyzing and modulating the self-attention mechanism in diffusion models. We hypothesize that self-attention during early denoising steps may inadvertently amplify or suppress features, contributing to hallucinations. To counter this, AAM introduces a temperature scaling mechanism within t"},"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":"2502.16872","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-02-24T06:19:54Z","cross_cats_sorted":[],"title_canon_sha256":"cd5362e7267df35ab300648d01a6d3556c829f819827fcbfae30717002cfd664","abstract_canon_sha256":"77b259dbaafc678460d357ed52493ebedf7d4384ce38d0b0d82279aadd11c504"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:19:01.463733Z","signature_b64":"phJDyGdCrlOdNAVmj9kbEIGGZViNnlpRT0SdzzAGHUXRONxIjxqMTwcp0Vla8cxMwhVNnzXvF3D5abil6EemCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2f9570ecc7a89330fbf4290b7d5031035d49120950fdba38403c4ebe69d9abe7","last_reissued_at":"2026-07-05T10:19:01.463237Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:19:01.463237Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mitigating Hallucinations in Diffusion Models through Adaptive Attention Modulation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Abhinav Shrivastava, Trevine Oorloff, Yaser Yacoob","submitted_at":"2025-02-24T06:19:54Z","abstract_excerpt":"Diffusion models, while increasingly adept at generating realistic images, are notably hindered by hallucinations -- unrealistic or incorrect features inconsistent with the trained data distribution. In this work, we propose Adaptive Attention Modulation (AAM), a novel approach to mitigate hallucinations by analyzing and modulating the self-attention mechanism in diffusion models. We hypothesize that self-attention during early denoising steps may inadvertently amplify or suppress features, contributing to hallucinations. To counter this, AAM introduces a temperature scaling mechanism within t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.16872","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/2502.16872/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":"2502.16872","created_at":"2026-07-05T10:19:01.463297+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.16872v1","created_at":"2026-07-05T10:19:01.463297+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.16872","created_at":"2026-07-05T10:19:01.463297+00:00"},{"alias_kind":"pith_short_12","alias_value":"F6KXB3GHVCJT","created_at":"2026-07-05T10:19:01.463297+00:00"},{"alias_kind":"pith_short_16","alias_value":"F6KXB3GHVCJTB67U","created_at":"2026-07-05T10:19:01.463297+00:00"},{"alias_kind":"pith_short_8","alias_value":"F6KXB3GH","created_at":"2026-07-05T10:19:01.463297+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.00377","citing_title":"Score-Control for Hallucination Reduction in Diffusion Models","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28405","citing_title":"Enhancing Layer Interaction Using Key-Correlated Layer Attention","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2505.09831","citing_title":"IMPLICITSTAINER: Resolution Agnostic Data-Efficient Virtual Staining Using Neural Implicit Functions","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18804","citing_title":"Geometric Decoupling: Diagnosing the Structural Instability of Latent","ref_index":30,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/F6KXB3GHVCJTB67UFEFX2UBRAN","json":"https://pith.science/pith/F6KXB3GHVCJTB67UFEFX2UBRAN.json","graph_json":"https://pith.science/api/pith-number/F6KXB3GHVCJTB67UFEFX2UBRAN/graph.json","events_json":"https://pith.science/api/pith-number/F6KXB3GHVCJTB67UFEFX2UBRAN/events.json","paper":"https://pith.science/paper/F6KXB3GH"},"agent_actions":{"view_html":"https://pith.science/pith/F6KXB3GHVCJTB67UFEFX2UBRAN","download_json":"https://pith.science/pith/F6KXB3GHVCJTB67UFEFX2UBRAN.json","view_paper":"https://pith.science/paper/F6KXB3GH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.16872&json=true","fetch_graph":"https://pith.science/api/pith-number/F6KXB3GHVCJTB67UFEFX2UBRAN/graph.json","fetch_events":"https://pith.science/api/pith-number/F6KXB3GHVCJTB67UFEFX2UBRAN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/F6KXB3GHVCJTB67UFEFX2UBRAN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/F6KXB3GHVCJTB67UFEFX2UBRAN/action/storage_attestation","attest_author":"https://pith.science/pith/F6KXB3GHVCJTB67UFEFX2UBRAN/action/author_attestation","sign_citation":"https://pith.science/pith/F6KXB3GHVCJTB67UFEFX2UBRAN/action/citation_signature","submit_replication":"https://pith.science/pith/F6KXB3GHVCJTB67UFEFX2UBRAN/action/replication_record"}},"created_at":"2026-07-05T10:19:01.463297+00:00","updated_at":"2026-07-05T10:19:01.463297+00:00"}