{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:C6KAPZUS6UGD4PU2FYM5OJFBHH","short_pith_number":"pith:C6KAPZUS","schema_version":"1.0","canonical_sha256":"179407e692f50c3e3e9a2e19d724a139cb3da18e5bbd7da2b53dbd3578245dcb","source":{"kind":"arxiv","id":"2602.01334","version":2},"attestation_state":"computed","paper":{"title":"What Does Vision Tool-Use Reinforcement Learning Really Learn? Disentangling Tool-Induced and Intrinsic Effects for Crop-and-Zoom","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Linge Du, Pengfei Liu, Tianle Li, Weiyu Zhang, Xuyang Shen, Yan Ma","submitted_at":"2026-02-01T17:00:50Z","abstract_excerpt":"Vision tool-use reinforcement learning (RL) can equip vision language models with visual operators such as crop-and-zoom and achieves strong performance gains, yet it remains unclear whether these gains are driven by improvements in tool use or evolving intrinsic capabilities. We introduce MED (Measure--Explain--Diagnose), a coarse-to-fine framework that disentangles intrinsic capability changes from tool-induced effects, decomposes the tool-induced performance difference into gain and harm terms, and probes the mechanisms driving their evolution. Across checkpoint-level analyses in the crop-a"},"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":"2602.01334","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-02-01T17:00:50Z","cross_cats_sorted":[],"title_canon_sha256":"796a357f9ec81cb87a0080c125634c8fd023ee183b64e18bc5862163535acc48","abstract_canon_sha256":"a826c0c629ea54b1d87da194725c7dee6edfdb7bb8d1da47a774fe26b993bd94"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-22T01:03:55.656511Z","signature_b64":"a50rwxqN4JFVjpnIirNKXabYT4wQV3+pFTLC603lZjyQ5aPlfNaN/qhtmnRXx2zxaflPelStBQFif6ogu8C8Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"179407e692f50c3e3e9a2e19d724a139cb3da18e5bbd7da2b53dbd3578245dcb","last_reissued_at":"2026-05-22T01:03:55.655533Z","signature_status":"signed_v1","first_computed_at":"2026-05-22T01:03:55.655533Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What Does Vision Tool-Use Reinforcement Learning Really Learn? Disentangling Tool-Induced and Intrinsic Effects for Crop-and-Zoom","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Linge Du, Pengfei Liu, Tianle Li, Weiyu Zhang, Xuyang Shen, Yan Ma","submitted_at":"2026-02-01T17:00:50Z","abstract_excerpt":"Vision tool-use reinforcement learning (RL) can equip vision language models with visual operators such as crop-and-zoom and achieves strong performance gains, yet it remains unclear whether these gains are driven by improvements in tool use or evolving intrinsic capabilities. We introduce MED (Measure--Explain--Diagnose), a coarse-to-fine framework that disentangles intrinsic capability changes from tool-induced effects, decomposes the tool-induced performance difference into gain and harm terms, and probes the mechanisms driving their evolution. Across checkpoint-level analyses in the crop-a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.01334","kind":"arxiv","version":2},"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/2602.01334/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":"2602.01334","created_at":"2026-05-22T01:03:55.655666+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.01334v2","created_at":"2026-05-22T01:03:55.655666+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.01334","created_at":"2026-05-22T01:03:55.655666+00:00"},{"alias_kind":"pith_short_12","alias_value":"C6KAPZUS6UGD","created_at":"2026-05-22T01:03:55.655666+00:00"},{"alias_kind":"pith_short_16","alias_value":"C6KAPZUS6UGD4PU2","created_at":"2026-05-22T01:03:55.655666+00:00"},{"alias_kind":"pith_short_8","alias_value":"C6KAPZUS","created_at":"2026-05-22T01:03:55.655666+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/C6KAPZUS6UGD4PU2FYM5OJFBHH","json":"https://pith.science/pith/C6KAPZUS6UGD4PU2FYM5OJFBHH.json","graph_json":"https://pith.science/api/pith-number/C6KAPZUS6UGD4PU2FYM5OJFBHH/graph.json","events_json":"https://pith.science/api/pith-number/C6KAPZUS6UGD4PU2FYM5OJFBHH/events.json","paper":"https://pith.science/paper/C6KAPZUS"},"agent_actions":{"view_html":"https://pith.science/pith/C6KAPZUS6UGD4PU2FYM5OJFBHH","download_json":"https://pith.science/pith/C6KAPZUS6UGD4PU2FYM5OJFBHH.json","view_paper":"https://pith.science/paper/C6KAPZUS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.01334&json=true","fetch_graph":"https://pith.science/api/pith-number/C6KAPZUS6UGD4PU2FYM5OJFBHH/graph.json","fetch_events":"https://pith.science/api/pith-number/C6KAPZUS6UGD4PU2FYM5OJFBHH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/C6KAPZUS6UGD4PU2FYM5OJFBHH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/C6KAPZUS6UGD4PU2FYM5OJFBHH/action/storage_attestation","attest_author":"https://pith.science/pith/C6KAPZUS6UGD4PU2FYM5OJFBHH/action/author_attestation","sign_citation":"https://pith.science/pith/C6KAPZUS6UGD4PU2FYM5OJFBHH/action/citation_signature","submit_replication":"https://pith.science/pith/C6KAPZUS6UGD4PU2FYM5OJFBHH/action/replication_record"}},"created_at":"2026-05-22T01:03:55.655666+00:00","updated_at":"2026-05-22T01:03:55.655666+00:00"}