{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:44ZQTYURYSAA2MOKO64OMYX5QX","short_pith_number":"pith:44ZQTYUR","schema_version":"1.0","canonical_sha256":"e73309e291c4800d31ca77b8e662fd85cd6b04498c2ec363bdb9017731ddbb4c","source":{"kind":"arxiv","id":"2608.08021","version":1},"attestation_state":"computed","paper":{"title":"Evidence-RL: Towards Evidence-intensive Visual Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chengming Xu, Cheng Yang, Haojie Huang, Jiangning Zhang, Qingdong He, Xiaobin Hu, Xinlei Yu, Yu Yang, Zhangquan Chen","submitted_at":"2026-08-08T09:02:10Z","abstract_excerpt":"Vision-Language Models (VLMs) should answer from concrete image evidence rather than language priors, dataset shortcuts, or irrelevant visual context. Existing perception-aware post-training methods encourage image use through global perturbations or attention proxies, but they do not test whether a sampled answer causally depends on the local evidence that supports it. We propose Counterfactual Evidence Disentanglement (CED), a training-time evidence audit for VLM grounding. For each response, CED neutralizes an object-centric Evidence Region and compares the resulting support drop against ma"},"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":"2608.08021","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-08-08T09:02:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8d1190318fe1bede41dbe60865c637ed0b4562b43af4f85b0fd739537d42b09c","abstract_canon_sha256":"80311ba9daf694895e35b7a21fb6534caa733a477fa4ced5fa6a89cfa5e3ecdd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-11T01:21:25.469651Z","signature_b64":"5Kn/flkPPrNaTd7/AqgSoHVKydu/IvGHxhfcN35KfWUdDJv+/Xc4iuHgaWaMMuvnj88S57R9lpZujYN8J8EHCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e73309e291c4800d31ca77b8e662fd85cd6b04498c2ec363bdb9017731ddbb4c","last_reissued_at":"2026-08-11T01:21:25.466872Z","signature_status":"signed_v1","first_computed_at":"2026-08-11T01:21:25.466872Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Evidence-RL: Towards Evidence-intensive Visual Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chengming Xu, Cheng Yang, Haojie Huang, Jiangning Zhang, Qingdong He, Xiaobin Hu, Xinlei Yu, Yu Yang, Zhangquan Chen","submitted_at":"2026-08-08T09:02:10Z","abstract_excerpt":"Vision-Language Models (VLMs) should answer from concrete image evidence rather than language priors, dataset shortcuts, or irrelevant visual context. Existing perception-aware post-training methods encourage image use through global perturbations or attention proxies, but they do not test whether a sampled answer causally depends on the local evidence that supports it. We propose Counterfactual Evidence Disentanglement (CED), a training-time evidence audit for VLM grounding. For each response, CED neutralizes an object-centric Evidence Region and compares the resulting support drop against ma"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.08021","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/2608.08021/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":"2608.08021","created_at":"2026-08-11T01:21:25.467732+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.08021v1","created_at":"2026-08-11T01:21:25.467732+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.08021","created_at":"2026-08-11T01:21:25.467732+00:00"},{"alias_kind":"pith_short_12","alias_value":"44ZQTYURYSAA","created_at":"2026-08-11T01:21:25.467732+00:00"},{"alias_kind":"pith_short_16","alias_value":"44ZQTYURYSAA2MOK","created_at":"2026-08-11T01:21:25.467732+00:00"},{"alias_kind":"pith_short_8","alias_value":"44ZQTYUR","created_at":"2026-08-11T01:21:25.467732+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/44ZQTYURYSAA2MOKO64OMYX5QX","json":"https://pith.science/pith/44ZQTYURYSAA2MOKO64OMYX5QX.json","graph_json":"https://pith.science/api/pith-number/44ZQTYURYSAA2MOKO64OMYX5QX/graph.json","events_json":"https://pith.science/api/pith-number/44ZQTYURYSAA2MOKO64OMYX5QX/events.json","paper":"https://pith.science/paper/44ZQTYUR"},"agent_actions":{"view_html":"https://pith.science/pith/44ZQTYURYSAA2MOKO64OMYX5QX","download_json":"https://pith.science/pith/44ZQTYURYSAA2MOKO64OMYX5QX.json","view_paper":"https://pith.science/paper/44ZQTYUR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.08021&json=true","fetch_graph":"https://pith.science/api/pith-number/44ZQTYURYSAA2MOKO64OMYX5QX/graph.json","fetch_events":"https://pith.science/api/pith-number/44ZQTYURYSAA2MOKO64OMYX5QX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/44ZQTYURYSAA2MOKO64OMYX5QX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/44ZQTYURYSAA2MOKO64OMYX5QX/action/storage_attestation","attest_author":"https://pith.science/pith/44ZQTYURYSAA2MOKO64OMYX5QX/action/author_attestation","sign_citation":"https://pith.science/pith/44ZQTYURYSAA2MOKO64OMYX5QX/action/citation_signature","submit_replication":"https://pith.science/pith/44ZQTYURYSAA2MOKO64OMYX5QX/action/replication_record"}},"created_at":"2026-08-11T01:21:25.467732+00:00","updated_at":"2026-08-11T01:21:25.467732+00:00"}