{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:IFA55SC7AB43HMBWGX4CZIMIOL","short_pith_number":"pith:IFA55SC7","schema_version":"1.0","canonical_sha256":"4141dec85f0079b3b03635f82ca18872f28b51c8b7e9c8d4f2efdaf8d1166918","source":{"kind":"arxiv","id":"2606.24515","version":1},"attestation_state":"computed","paper":{"title":"Reinforcement Learning for Computer-Use Agents with Autonomous Evaluation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.HC"],"primary_cat":"cs.AI","authors_text":"Marta Sumyk, Oleksandr Kosovan","submitted_at":"2026-06-23T12:46:29Z","abstract_excerpt":"Computer-Use Agents (CUAs) execute high-level user goals by perceiving and acting directly within graphical user interfaces. However, reinforcement learning for CUAs remains difficult because open-ended desktop environments rarely provide scalable, machine-readable reward signals: task success is often visually grounded and hard to specify with handcrafted reward functions or dense manual labels.\n  We propose an RL fine-tuning framework that uses autonomous vision-language evaluation as a scalable supervision signal for GUI agents. Given a final screenshot and the original instruction, a Visio"},"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":"2606.24515","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-23T12:46:29Z","cross_cats_sorted":["cs.HC"],"title_canon_sha256":"3b0244e9e6725e9889b0acb3635a4b78aa8bfb55094d7d82ef2e45547e2292ef","abstract_canon_sha256":"63e8606c76fd0b6ca63cba982f6b32c46502d06de57d7cbc8227ba90c15d11ad"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-24T01:15:32.734234Z","signature_b64":"gxz6GajPseO3g/VGEhKCEy5FEzw7v8EDt8lCjsNIV4ipceMqTYC5dr6+sgMEJHn7y3Evv4ajT3hXeNspSdJzAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4141dec85f0079b3b03635f82ca18872f28b51c8b7e9c8d4f2efdaf8d1166918","last_reissued_at":"2026-06-24T01:15:32.733894Z","signature_status":"signed_v1","first_computed_at":"2026-06-24T01:15:32.733894Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reinforcement Learning for Computer-Use Agents with Autonomous Evaluation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.HC"],"primary_cat":"cs.AI","authors_text":"Marta Sumyk, Oleksandr Kosovan","submitted_at":"2026-06-23T12:46:29Z","abstract_excerpt":"Computer-Use Agents (CUAs) execute high-level user goals by perceiving and acting directly within graphical user interfaces. However, reinforcement learning for CUAs remains difficult because open-ended desktop environments rarely provide scalable, machine-readable reward signals: task success is often visually grounded and hard to specify with handcrafted reward functions or dense manual labels.\n  We propose an RL fine-tuning framework that uses autonomous vision-language evaluation as a scalable supervision signal for GUI agents. Given a final screenshot and the original instruction, a Visio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.24515","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/2606.24515/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":"2606.24515","created_at":"2026-06-24T01:15:32.733954+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.24515v1","created_at":"2026-06-24T01:15:32.733954+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.24515","created_at":"2026-06-24T01:15:32.733954+00:00"},{"alias_kind":"pith_short_12","alias_value":"IFA55SC7AB43","created_at":"2026-06-24T01:15:32.733954+00:00"},{"alias_kind":"pith_short_16","alias_value":"IFA55SC7AB43HMBW","created_at":"2026-06-24T01:15:32.733954+00:00"},{"alias_kind":"pith_short_8","alias_value":"IFA55SC7","created_at":"2026-06-24T01:15:32.733954+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/IFA55SC7AB43HMBWGX4CZIMIOL","json":"https://pith.science/pith/IFA55SC7AB43HMBWGX4CZIMIOL.json","graph_json":"https://pith.science/api/pith-number/IFA55SC7AB43HMBWGX4CZIMIOL/graph.json","events_json":"https://pith.science/api/pith-number/IFA55SC7AB43HMBWGX4CZIMIOL/events.json","paper":"https://pith.science/paper/IFA55SC7"},"agent_actions":{"view_html":"https://pith.science/pith/IFA55SC7AB43HMBWGX4CZIMIOL","download_json":"https://pith.science/pith/IFA55SC7AB43HMBWGX4CZIMIOL.json","view_paper":"https://pith.science/paper/IFA55SC7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.24515&json=true","fetch_graph":"https://pith.science/api/pith-number/IFA55SC7AB43HMBWGX4CZIMIOL/graph.json","fetch_events":"https://pith.science/api/pith-number/IFA55SC7AB43HMBWGX4CZIMIOL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IFA55SC7AB43HMBWGX4CZIMIOL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IFA55SC7AB43HMBWGX4CZIMIOL/action/storage_attestation","attest_author":"https://pith.science/pith/IFA55SC7AB43HMBWGX4CZIMIOL/action/author_attestation","sign_citation":"https://pith.science/pith/IFA55SC7AB43HMBWGX4CZIMIOL/action/citation_signature","submit_replication":"https://pith.science/pith/IFA55SC7AB43HMBWGX4CZIMIOL/action/replication_record"}},"created_at":"2026-06-24T01:15:32.733954+00:00","updated_at":"2026-06-24T01:15:32.733954+00:00"}