{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KDJ2K6IS3T6MESZP7EFRUKMWFM","short_pith_number":"pith:KDJ2K6IS","schema_version":"1.0","canonical_sha256":"50d3a57912dcfcc24b2ff90b1a29962b2d3217ca09d26aeb7459c60af8f44a97","source":{"kind":"arxiv","id":"2607.22013","version":1},"attestation_state":"computed","paper":{"title":"Visual Saliency Steering Distillation for Multimodal Chain-of-Thought Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hao Yang, Jin Wang, Xuejie Zhang","submitted_at":"2026-07-24T06:22:40Z","abstract_excerpt":"Multimodal chain-of-thought (CoT) reasoning integrates visual and textual cues through step-by-step inference. In small models with limited token budgets, modality-interaction fusion often suppresses tiny cross-modal differences. In particular, multimodal CoT often struggles when different images pair with identical text or different texts pair with an identical image, making such inputs nearly indistinguishable after fusion. This study proposes Visual Saliency Steering Distillation (VSSD). VSSD leverages the attention maps of multimodal large language models to generate perturbed images that "},"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":"2607.22013","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-07-24T06:22:40Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"fdbf58257023968660cfc8fd1f1e9e338b7640eae606d2b08b9cf518e10a75bd","abstract_canon_sha256":"3b48058d486ffe4a81ab786b8106e4f97d427638e8053b5513a814a1e34db858"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-27T01:20:43.152947Z","signature_b64":"POG6MrtiVeYDkiu7woZD9y9Knv+00bH2TN+lKcTBGdVtduP3Skn4opby+3V7MdgujrqUnSSypSf9Rz9PSlb7Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"50d3a57912dcfcc24b2ff90b1a29962b2d3217ca09d26aeb7459c60af8f44a97","last_reissued_at":"2026-07-27T01:20:43.152144Z","signature_status":"signed_v1","first_computed_at":"2026-07-27T01:20:43.152144Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Visual Saliency Steering Distillation for Multimodal Chain-of-Thought Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hao Yang, Jin Wang, Xuejie Zhang","submitted_at":"2026-07-24T06:22:40Z","abstract_excerpt":"Multimodal chain-of-thought (CoT) reasoning integrates visual and textual cues through step-by-step inference. In small models with limited token budgets, modality-interaction fusion often suppresses tiny cross-modal differences. In particular, multimodal CoT often struggles when different images pair with identical text or different texts pair with an identical image, making such inputs nearly indistinguishable after fusion. This study proposes Visual Saliency Steering Distillation (VSSD). VSSD leverages the attention maps of multimodal large language models to generate perturbed images that "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.22013","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/2607.22013/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":"2607.22013","created_at":"2026-07-27T01:20:43.152596+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.22013v1","created_at":"2026-07-27T01:20:43.152596+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.22013","created_at":"2026-07-27T01:20:43.152596+00:00"},{"alias_kind":"pith_short_12","alias_value":"KDJ2K6IS3T6M","created_at":"2026-07-27T01:20:43.152596+00:00"},{"alias_kind":"pith_short_16","alias_value":"KDJ2K6IS3T6MESZP","created_at":"2026-07-27T01:20:43.152596+00:00"},{"alias_kind":"pith_short_8","alias_value":"KDJ2K6IS","created_at":"2026-07-27T01:20:43.152596+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/KDJ2K6IS3T6MESZP7EFRUKMWFM","json":"https://pith.science/pith/KDJ2K6IS3T6MESZP7EFRUKMWFM.json","graph_json":"https://pith.science/api/pith-number/KDJ2K6IS3T6MESZP7EFRUKMWFM/graph.json","events_json":"https://pith.science/api/pith-number/KDJ2K6IS3T6MESZP7EFRUKMWFM/events.json","paper":"https://pith.science/paper/KDJ2K6IS"},"agent_actions":{"view_html":"https://pith.science/pith/KDJ2K6IS3T6MESZP7EFRUKMWFM","download_json":"https://pith.science/pith/KDJ2K6IS3T6MESZP7EFRUKMWFM.json","view_paper":"https://pith.science/paper/KDJ2K6IS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.22013&json=true","fetch_graph":"https://pith.science/api/pith-number/KDJ2K6IS3T6MESZP7EFRUKMWFM/graph.json","fetch_events":"https://pith.science/api/pith-number/KDJ2K6IS3T6MESZP7EFRUKMWFM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KDJ2K6IS3T6MESZP7EFRUKMWFM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KDJ2K6IS3T6MESZP7EFRUKMWFM/action/storage_attestation","attest_author":"https://pith.science/pith/KDJ2K6IS3T6MESZP7EFRUKMWFM/action/author_attestation","sign_citation":"https://pith.science/pith/KDJ2K6IS3T6MESZP7EFRUKMWFM/action/citation_signature","submit_replication":"https://pith.science/pith/KDJ2K6IS3T6MESZP7EFRUKMWFM/action/replication_record"}},"created_at":"2026-07-27T01:20:43.152596+00:00","updated_at":"2026-07-27T01:20:43.152596+00:00"}