{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:WPUBCVUOXSGVS4IKMHWFHO6ILI","short_pith_number":"pith:WPUBCVUO","schema_version":"1.0","canonical_sha256":"b3e811568ebc8d59710a61ec53bbc85a07f3ae7f8f2207e6a21d1aa499bac994","source":{"kind":"arxiv","id":"2507.02321","version":1},"attestation_state":"computed","paper":{"title":"Heeding the Inner Voice: Aligning ControlNet Training via Intermediate Features Feedback","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Aibek Alanov, Andrey Kuznetsov, Maxim Nikolaev, Nina Konovalova","submitted_at":"2025-07-03T05:25:53Z","abstract_excerpt":"Despite significant progress in text-to-image diffusion models, achieving precise spatial control over generated outputs remains challenging. ControlNet addresses this by introducing an auxiliary conditioning module, while ControlNet++ further refines alignment through a cycle consistency loss applied only to the final denoising steps. However, this approach neglects intermediate generation stages, limiting its effectiveness. We propose InnerControl, a training strategy that enforces spatial consistency across all diffusion steps. Our method trains lightweight convolutional probes to reconstru"},"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":"2507.02321","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-07-03T05:25:53Z","cross_cats_sorted":[],"title_canon_sha256":"98e9b9aa6e78cd52536c9d1b1572893530eb7bc8ad6fcab8c4500f6a7401784d","abstract_canon_sha256":"9903a51cd98d2fe47e0706171f1dd8d53f8e33a5bf23b2be418e43ec3bdb644a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:31:15.987960Z","signature_b64":"emTRmborx2aexV4SYHiDdceZeHNIkLusyQkLgxQG2m4qLoRya74gHqT29MdEia4DHuJPOO9AKa0FQReUjpC5AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b3e811568ebc8d59710a61ec53bbc85a07f3ae7f8f2207e6a21d1aa499bac994","last_reissued_at":"2026-07-05T11:31:15.987548Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:31:15.987548Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Heeding the Inner Voice: Aligning ControlNet Training via Intermediate Features Feedback","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Aibek Alanov, Andrey Kuznetsov, Maxim Nikolaev, Nina Konovalova","submitted_at":"2025-07-03T05:25:53Z","abstract_excerpt":"Despite significant progress in text-to-image diffusion models, achieving precise spatial control over generated outputs remains challenging. ControlNet addresses this by introducing an auxiliary conditioning module, while ControlNet++ further refines alignment through a cycle consistency loss applied only to the final denoising steps. However, this approach neglects intermediate generation stages, limiting its effectiveness. We propose InnerControl, a training strategy that enforces spatial consistency across all diffusion steps. Our method trains lightweight convolutional probes to reconstru"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.02321","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/2507.02321/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":"2507.02321","created_at":"2026-07-05T11:31:15.987604+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.02321v1","created_at":"2026-07-05T11:31:15.987604+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.02321","created_at":"2026-07-05T11:31:15.987604+00:00"},{"alias_kind":"pith_short_12","alias_value":"WPUBCVUOXSGV","created_at":"2026-07-05T11:31:15.987604+00:00"},{"alias_kind":"pith_short_16","alias_value":"WPUBCVUOXSGVS4IK","created_at":"2026-07-05T11:31:15.987604+00:00"},{"alias_kind":"pith_short_8","alias_value":"WPUBCVUO","created_at":"2026-07-05T11:31:15.987604+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.04107","citing_title":"Reflection Separation from a Single Image via Joint Latent Diffusion","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WPUBCVUOXSGVS4IKMHWFHO6ILI","json":"https://pith.science/pith/WPUBCVUOXSGVS4IKMHWFHO6ILI.json","graph_json":"https://pith.science/api/pith-number/WPUBCVUOXSGVS4IKMHWFHO6ILI/graph.json","events_json":"https://pith.science/api/pith-number/WPUBCVUOXSGVS4IKMHWFHO6ILI/events.json","paper":"https://pith.science/paper/WPUBCVUO"},"agent_actions":{"view_html":"https://pith.science/pith/WPUBCVUOXSGVS4IKMHWFHO6ILI","download_json":"https://pith.science/pith/WPUBCVUOXSGVS4IKMHWFHO6ILI.json","view_paper":"https://pith.science/paper/WPUBCVUO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.02321&json=true","fetch_graph":"https://pith.science/api/pith-number/WPUBCVUOXSGVS4IKMHWFHO6ILI/graph.json","fetch_events":"https://pith.science/api/pith-number/WPUBCVUOXSGVS4IKMHWFHO6ILI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WPUBCVUOXSGVS4IKMHWFHO6ILI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WPUBCVUOXSGVS4IKMHWFHO6ILI/action/storage_attestation","attest_author":"https://pith.science/pith/WPUBCVUOXSGVS4IKMHWFHO6ILI/action/author_attestation","sign_citation":"https://pith.science/pith/WPUBCVUOXSGVS4IKMHWFHO6ILI/action/citation_signature","submit_replication":"https://pith.science/pith/WPUBCVUOXSGVS4IKMHWFHO6ILI/action/replication_record"}},"created_at":"2026-07-05T11:31:15.987604+00:00","updated_at":"2026-07-05T11:31:15.987604+00:00"}