{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:UKTY3FLO3B6QQ3MYPQEW22MSHK","short_pith_number":"pith:UKTY3FLO","schema_version":"1.0","canonical_sha256":"a2a78d956ed87d086d987c096d69923a9a3ebe9bfa83fbd8cf685ae0ea9a3dc4","source":{"kind":"arxiv","id":"2605.29894","version":1},"attestation_state":"computed","paper":{"title":"Train the Agent, Not the Expert: Learning to Harness Heterogeneous Experts for Multi-Turn Visual Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Andy J. Ma, Dazhao Du, Jia Wan, Tao Han, Yaowu Fan","submitted_at":"2026-05-28T13:18:10Z","abstract_excerpt":"Recent progress in computer vision has produced a wide range of powerful specialized models for detection, segmentation, counting, and other visual tasks. However, these models are usually optimized for isolated task formulations, making it difficult to directly support general-purpose visual intelligence, especially when a task requires complex language understanding and dense small-object perception. In this paper, we propose VisHarness, a trainable visual agent that decouples high-level perception, reasoning, and decision-making from low-level task execution. Instead of training a model to "},"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":"2605.29894","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-28T13:18:10Z","cross_cats_sorted":[],"title_canon_sha256":"9a7cd4ce4448fa0da40679297fd2d878e28eea69c047ad0bca2ec4eaa53269bc","abstract_canon_sha256":"4e5a027188c5901feb783e8a5a7e27fd9909fd07406ccaf30dfd2f228d3222b0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T02:05:58.297689Z","signature_b64":"Zm6h6USw1q6PVY77yFBKoZE2A5hk9x6Skr7MJplP4I3z1IKnTYrVGxG3s9Z13/52rvi7vBRAVc/g8x3ZzGoSBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a2a78d956ed87d086d987c096d69923a9a3ebe9bfa83fbd8cf685ae0ea9a3dc4","last_reissued_at":"2026-05-29T02:05:58.296962Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T02:05:58.296962Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Train the Agent, Not the Expert: Learning to Harness Heterogeneous Experts for Multi-Turn Visual Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Andy J. Ma, Dazhao Du, Jia Wan, Tao Han, Yaowu Fan","submitted_at":"2026-05-28T13:18:10Z","abstract_excerpt":"Recent progress in computer vision has produced a wide range of powerful specialized models for detection, segmentation, counting, and other visual tasks. However, these models are usually optimized for isolated task formulations, making it difficult to directly support general-purpose visual intelligence, especially when a task requires complex language understanding and dense small-object perception. In this paper, we propose VisHarness, a trainable visual agent that decouples high-level perception, reasoning, and decision-making from low-level task execution. Instead of training a model to "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.29894","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/2605.29894/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":"2605.29894","created_at":"2026-05-29T02:05:58.297079+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.29894v1","created_at":"2026-05-29T02:05:58.297079+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.29894","created_at":"2026-05-29T02:05:58.297079+00:00"},{"alias_kind":"pith_short_12","alias_value":"UKTY3FLO3B6Q","created_at":"2026-05-29T02:05:58.297079+00:00"},{"alias_kind":"pith_short_16","alias_value":"UKTY3FLO3B6QQ3MY","created_at":"2026-05-29T02:05:58.297079+00:00"},{"alias_kind":"pith_short_8","alias_value":"UKTY3FLO","created_at":"2026-05-29T02:05:58.297079+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/UKTY3FLO3B6QQ3MYPQEW22MSHK","json":"https://pith.science/pith/UKTY3FLO3B6QQ3MYPQEW22MSHK.json","graph_json":"https://pith.science/api/pith-number/UKTY3FLO3B6QQ3MYPQEW22MSHK/graph.json","events_json":"https://pith.science/api/pith-number/UKTY3FLO3B6QQ3MYPQEW22MSHK/events.json","paper":"https://pith.science/paper/UKTY3FLO"},"agent_actions":{"view_html":"https://pith.science/pith/UKTY3FLO3B6QQ3MYPQEW22MSHK","download_json":"https://pith.science/pith/UKTY3FLO3B6QQ3MYPQEW22MSHK.json","view_paper":"https://pith.science/paper/UKTY3FLO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.29894&json=true","fetch_graph":"https://pith.science/api/pith-number/UKTY3FLO3B6QQ3MYPQEW22MSHK/graph.json","fetch_events":"https://pith.science/api/pith-number/UKTY3FLO3B6QQ3MYPQEW22MSHK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UKTY3FLO3B6QQ3MYPQEW22MSHK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UKTY3FLO3B6QQ3MYPQEW22MSHK/action/storage_attestation","attest_author":"https://pith.science/pith/UKTY3FLO3B6QQ3MYPQEW22MSHK/action/author_attestation","sign_citation":"https://pith.science/pith/UKTY3FLO3B6QQ3MYPQEW22MSHK/action/citation_signature","submit_replication":"https://pith.science/pith/UKTY3FLO3B6QQ3MYPQEW22MSHK/action/replication_record"}},"created_at":"2026-05-29T02:05:58.297079+00:00","updated_at":"2026-05-29T02:05:58.297079+00:00"}