{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:L243ZI6MSQM7CWJJK7XK4LKQK7","short_pith_number":"pith:L243ZI6M","schema_version":"1.0","canonical_sha256":"5eb9bca3cc9419f1592957eeae2d5057e57b7e3ec3508568f6c323059d6b2561","source":{"kind":"arxiv","id":"2607.00434","version":1},"attestation_state":"computed","paper":{"title":"Information-Regularized Attention for Visual-Centric Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Guohao Sun, Mengchen Liu, Praveen Krishnan, Xiaofang Wang, Yash Patel, Zhiqiang Tao","submitted_at":"2026-07-01T04:45:41Z","abstract_excerpt":"Vision-language models (VLMs) have become a paradigm for multimodal learning, yet remain unstable due to object hallucination, weak visual grounding, and catastrophic forgetting after full-parameter instruction tuning. We claim these failures result from a lack of explicit control over visual representation learning during the standard next-token prediction objective. As a result, visual embeddings thus become passively optimized and prone to injecting redundant or spurious signals. To counter this, we introduce Information-Regularized Attention (IRA), a stochastic attention mechanism that exp"},"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":"2607.00434","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-07-01T04:45:41Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"a13dbd99d1f24bd33d1beb1bf94e1a2b18f00fd2aed1f4c8c7139a97d7e1ae55","abstract_canon_sha256":"a79b49d6f65c2ac9226d6d7720584da3c6c33a44fc03ccc8298df1678b28b3cb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-02T01:17:43.246096Z","signature_b64":"Iu+/WUYovF3geWE06Gbr+wB4yDsH5r/TT0vFLaFy0K6ILL44DxuQlfBJEjcj8PZ/ZzWf6PbJnzsezBjnJ7rADg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5eb9bca3cc9419f1592957eeae2d5057e57b7e3ec3508568f6c323059d6b2561","last_reissued_at":"2026-07-02T01:17:43.245728Z","signature_status":"signed_v1","first_computed_at":"2026-07-02T01:17:43.245728Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Information-Regularized Attention for Visual-Centric Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Guohao Sun, Mengchen Liu, Praveen Krishnan, Xiaofang Wang, Yash Patel, Zhiqiang Tao","submitted_at":"2026-07-01T04:45:41Z","abstract_excerpt":"Vision-language models (VLMs) have become a paradigm for multimodal learning, yet remain unstable due to object hallucination, weak visual grounding, and catastrophic forgetting after full-parameter instruction tuning. We claim these failures result from a lack of explicit control over visual representation learning during the standard next-token prediction objective. As a result, visual embeddings thus become passively optimized and prone to injecting redundant or spurious signals. To counter this, we introduce Information-Regularized Attention (IRA), a stochastic attention mechanism that exp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.00434","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/2607.00434/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":"2607.00434","created_at":"2026-07-02T01:17:43.245790+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.00434v1","created_at":"2026-07-02T01:17:43.245790+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.00434","created_at":"2026-07-02T01:17:43.245790+00:00"},{"alias_kind":"pith_short_12","alias_value":"L243ZI6MSQM7","created_at":"2026-07-02T01:17:43.245790+00:00"},{"alias_kind":"pith_short_16","alias_value":"L243ZI6MSQM7CWJJ","created_at":"2026-07-02T01:17:43.245790+00:00"},{"alias_kind":"pith_short_8","alias_value":"L243ZI6M","created_at":"2026-07-02T01:17:43.245790+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/L243ZI6MSQM7CWJJK7XK4LKQK7","json":"https://pith.science/pith/L243ZI6MSQM7CWJJK7XK4LKQK7.json","graph_json":"https://pith.science/api/pith-number/L243ZI6MSQM7CWJJK7XK4LKQK7/graph.json","events_json":"https://pith.science/api/pith-number/L243ZI6MSQM7CWJJK7XK4LKQK7/events.json","paper":"https://pith.science/paper/L243ZI6M"},"agent_actions":{"view_html":"https://pith.science/pith/L243ZI6MSQM7CWJJK7XK4LKQK7","download_json":"https://pith.science/pith/L243ZI6MSQM7CWJJK7XK4LKQK7.json","view_paper":"https://pith.science/paper/L243ZI6M","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.00434&json=true","fetch_graph":"https://pith.science/api/pith-number/L243ZI6MSQM7CWJJK7XK4LKQK7/graph.json","fetch_events":"https://pith.science/api/pith-number/L243ZI6MSQM7CWJJK7XK4LKQK7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/L243ZI6MSQM7CWJJK7XK4LKQK7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/L243ZI6MSQM7CWJJK7XK4LKQK7/action/storage_attestation","attest_author":"https://pith.science/pith/L243ZI6MSQM7CWJJK7XK4LKQK7/action/author_attestation","sign_citation":"https://pith.science/pith/L243ZI6MSQM7CWJJK7XK4LKQK7/action/citation_signature","submit_replication":"https://pith.science/pith/L243ZI6MSQM7CWJJK7XK4LKQK7/action/replication_record"}},"created_at":"2026-07-02T01:17:43.245790+00:00","updated_at":"2026-07-02T01:17:43.245790+00:00"}