{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:MSDIYUTCBBHTGI7C4TSCTNIB6Y","short_pith_number":"pith:MSDIYUTC","schema_version":"1.0","canonical_sha256":"64868c5262084f3323e2e4e429b501f611d2374af3c31e26f1105a63dd1d72d1","source":{"kind":"arxiv","id":"2509.20912","version":3},"attestation_state":"computed","paper":{"title":"DeFacto: Counterfactual Thinking with Images for Enforcing Evidence-Grounded and Faithful Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Feng Chen, Guanyu Chen, Haichuan Gao, Haoda Jing, Jun Feng, Tianren Zhang, Tianrun Xu, Ye Li, Yuquan Wei","submitted_at":"2025-09-25T08:58:10Z","abstract_excerpt":"Recent advances in multimodal language models (MLLMs) have made thinking with images a dominant paradigm for multimodal reasoning. However, existing methods still fail to ensure evidence-answer consistency, where correct answers must be supported by correct visual evidence. To address this issue, we propose DeFacto, a counterfactual reasoning framework that explicitly aligns visual evidence with final answers. Our approach integrates three complementary training paradigms: positive, counterfactual, and random-masking. We further develop a language-guided evidence construction pipeline that aut"},"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":"2509.20912","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-09-25T08:58:10Z","cross_cats_sorted":[],"title_canon_sha256":"a6d673f5378f5815d58f033ea7137600db0ab217fc8d0d2ddfb22770a435a7bf","abstract_canon_sha256":"3c02ecad2df21d13116fb41d6097f3eed70f41d238811969940192fb67fd7d63"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T01:05:09.480808Z","signature_b64":"dO0SIzaypQg5uZ8kzM5SJ/bRXQnrUYjw4yPPlSo18OHqnawyXQ8AbCFBX3CZDhkJ3y79kJ+LSErceJDi3phzBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"64868c5262084f3323e2e4e429b501f611d2374af3c31e26f1105a63dd1d72d1","last_reissued_at":"2026-05-21T01:05:09.479839Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T01:05:09.479839Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DeFacto: Counterfactual Thinking with Images for Enforcing Evidence-Grounded and Faithful Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Feng Chen, Guanyu Chen, Haichuan Gao, Haoda Jing, Jun Feng, Tianren Zhang, Tianrun Xu, Ye Li, Yuquan Wei","submitted_at":"2025-09-25T08:58:10Z","abstract_excerpt":"Recent advances in multimodal language models (MLLMs) have made thinking with images a dominant paradigm for multimodal reasoning. However, existing methods still fail to ensure evidence-answer consistency, where correct answers must be supported by correct visual evidence. To address this issue, we propose DeFacto, a counterfactual reasoning framework that explicitly aligns visual evidence with final answers. Our approach integrates three complementary training paradigms: positive, counterfactual, and random-masking. We further develop a language-guided evidence construction pipeline that aut"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.20912","kind":"arxiv","version":3},"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/2509.20912/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":"2509.20912","created_at":"2026-05-21T01:05:09.479964+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.20912v3","created_at":"2026-05-21T01:05:09.479964+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.20912","created_at":"2026-05-21T01:05:09.479964+00:00"},{"alias_kind":"pith_short_12","alias_value":"MSDIYUTCBBHT","created_at":"2026-05-21T01:05:09.479964+00:00"},{"alias_kind":"pith_short_16","alias_value":"MSDIYUTCBBHTGI7C","created_at":"2026-05-21T01:05:09.479964+00:00"},{"alias_kind":"pith_short_8","alias_value":"MSDIYUTC","created_at":"2026-05-21T01:05:09.479964+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.19342","citing_title":"Semantic-Enriched Latent Visual Reasoning","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2605.13467","citing_title":"PDCR: Perception-Decomposed Confidence Reward for Vision-Language Reasoning","ref_index":37,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MSDIYUTCBBHTGI7C4TSCTNIB6Y","json":"https://pith.science/pith/MSDIYUTCBBHTGI7C4TSCTNIB6Y.json","graph_json":"https://pith.science/api/pith-number/MSDIYUTCBBHTGI7C4TSCTNIB6Y/graph.json","events_json":"https://pith.science/api/pith-number/MSDIYUTCBBHTGI7C4TSCTNIB6Y/events.json","paper":"https://pith.science/paper/MSDIYUTC"},"agent_actions":{"view_html":"https://pith.science/pith/MSDIYUTCBBHTGI7C4TSCTNIB6Y","download_json":"https://pith.science/pith/MSDIYUTCBBHTGI7C4TSCTNIB6Y.json","view_paper":"https://pith.science/paper/MSDIYUTC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.20912&json=true","fetch_graph":"https://pith.science/api/pith-number/MSDIYUTCBBHTGI7C4TSCTNIB6Y/graph.json","fetch_events":"https://pith.science/api/pith-number/MSDIYUTCBBHTGI7C4TSCTNIB6Y/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MSDIYUTCBBHTGI7C4TSCTNIB6Y/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MSDIYUTCBBHTGI7C4TSCTNIB6Y/action/storage_attestation","attest_author":"https://pith.science/pith/MSDIYUTCBBHTGI7C4TSCTNIB6Y/action/author_attestation","sign_citation":"https://pith.science/pith/MSDIYUTCBBHTGI7C4TSCTNIB6Y/action/citation_signature","submit_replication":"https://pith.science/pith/MSDIYUTCBBHTGI7C4TSCTNIB6Y/action/replication_record"}},"created_at":"2026-05-21T01:05:09.479964+00:00","updated_at":"2026-05-21T01:05:09.479964+00:00"}