{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QNZVLCY2KUVE4QB3V3KLHCUAWN","short_pith_number":"pith:QNZVLCY2","schema_version":"1.0","canonical_sha256":"8373558b1a552a4e403baed4b38a80b37b32ee9a641b7f87bbaf74c64bda473e","source":{"kind":"arxiv","id":"2608.04385","version":1},"attestation_state":"computed","paper":{"title":"ReGround: Restoring Visual Grounding in Multi-Step Reasoning through Self-Diagnosis and Visual Re-Examination","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Lei Peng, Shuai Lv, Wei Hu","submitted_at":"2026-08-05T02:41:53Z","abstract_excerpt":"Vision-Language Models (VLMs) often lose visual grounding during multi-step reasoning: as reasoning chains grow longer, later inference steps rely increasingly on language priors rather than image evidence. We identify a consistent benchmark-level signature associated with this degradation: across 2,510 re-examined samples from four benchmarks, attention entropy over image tokens typically decreases during Round 1 and rises again after image re-injection.\n  However, we find that effective visual re-examination requires two complementary ingredients: image re-injection and targeted self-diagnos"},"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":"2608.04385","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-08-05T02:41:53Z","cross_cats_sorted":[],"title_canon_sha256":"5ab15cf187319f9d51e0eca286323c00f4eeac5867d949ce0dcd50d2a94f032b","abstract_canon_sha256":"318fcae0f0830a2e41981bcd3f640e4fa96493e80681e61657cafc72fc6c7747"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-06T01:34:37.091217Z","signature_b64":"rONsk03nCoGWZUqEqxZhn7ZvMzhjYjbyGlRMcV4efvtxWcPzSiCaiRiwyWU7iX7tpQpmMbmgURMk23luL55NDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8373558b1a552a4e403baed4b38a80b37b32ee9a641b7f87bbaf74c64bda473e","last_reissued_at":"2026-08-06T01:34:37.089763Z","signature_status":"signed_v1","first_computed_at":"2026-08-06T01:34:37.089763Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ReGround: Restoring Visual Grounding in Multi-Step Reasoning through Self-Diagnosis and Visual Re-Examination","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Lei Peng, Shuai Lv, Wei Hu","submitted_at":"2026-08-05T02:41:53Z","abstract_excerpt":"Vision-Language Models (VLMs) often lose visual grounding during multi-step reasoning: as reasoning chains grow longer, later inference steps rely increasingly on language priors rather than image evidence. We identify a consistent benchmark-level signature associated with this degradation: across 2,510 re-examined samples from four benchmarks, attention entropy over image tokens typically decreases during Round 1 and rises again after image re-injection.\n  However, we find that effective visual re-examination requires two complementary ingredients: image re-injection and targeted self-diagnos"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.04385","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/2608.04385/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":"2608.04385","created_at":"2026-08-06T01:34:37.091561+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.04385v1","created_at":"2026-08-06T01:34:37.091561+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.04385","created_at":"2026-08-06T01:34:37.091561+00:00"},{"alias_kind":"pith_short_12","alias_value":"QNZVLCY2KUVE","created_at":"2026-08-06T01:34:37.091561+00:00"},{"alias_kind":"pith_short_16","alias_value":"QNZVLCY2KUVE4QB3","created_at":"2026-08-06T01:34:37.091561+00:00"},{"alias_kind":"pith_short_8","alias_value":"QNZVLCY2","created_at":"2026-08-06T01:34:37.091561+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/QNZVLCY2KUVE4QB3V3KLHCUAWN","json":"https://pith.science/pith/QNZVLCY2KUVE4QB3V3KLHCUAWN.json","graph_json":"https://pith.science/api/pith-number/QNZVLCY2KUVE4QB3V3KLHCUAWN/graph.json","events_json":"https://pith.science/api/pith-number/QNZVLCY2KUVE4QB3V3KLHCUAWN/events.json","paper":"https://pith.science/paper/QNZVLCY2"},"agent_actions":{"view_html":"https://pith.science/pith/QNZVLCY2KUVE4QB3V3KLHCUAWN","download_json":"https://pith.science/pith/QNZVLCY2KUVE4QB3V3KLHCUAWN.json","view_paper":"https://pith.science/paper/QNZVLCY2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.04385&json=true","fetch_graph":"https://pith.science/api/pith-number/QNZVLCY2KUVE4QB3V3KLHCUAWN/graph.json","fetch_events":"https://pith.science/api/pith-number/QNZVLCY2KUVE4QB3V3KLHCUAWN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QNZVLCY2KUVE4QB3V3KLHCUAWN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QNZVLCY2KUVE4QB3V3KLHCUAWN/action/storage_attestation","attest_author":"https://pith.science/pith/QNZVLCY2KUVE4QB3V3KLHCUAWN/action/author_attestation","sign_citation":"https://pith.science/pith/QNZVLCY2KUVE4QB3V3KLHCUAWN/action/citation_signature","submit_replication":"https://pith.science/pith/QNZVLCY2KUVE4QB3V3KLHCUAWN/action/replication_record"}},"created_at":"2026-08-06T01:34:37.091561+00:00","updated_at":"2026-08-06T01:34:37.091561+00:00"}