{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:A5U26OZ2TIC5I5OFXCUECYGP44","short_pith_number":"pith:A5U26OZ2","schema_version":"1.0","canonical_sha256":"0769af3b3a9a05d475c5b8a84160cfe72bb3437fd26e007a499745a61bc90cd4","source":{"kind":"arxiv","id":"2506.08012","version":1},"attestation_state":"computed","paper":{"title":"GUI-Reflection: Empowering Multimodal GUI Models with Self-Reflection Behavior","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.AI","authors_text":"Bo Wang, Jiaheng Yu, Lewei Lu, Penghao Wu, Shengnan Ma, Ziwei Liu","submitted_at":"2025-06-09T17:59:57Z","abstract_excerpt":"Multimodal Large Language Models (MLLMs) have shown great potential in revolutionizing Graphical User Interface (GUI) automation. However, existing GUI models mostly rely on learning from nearly error-free offline trajectories, thus lacking reflection and error recovery capabilities. To bridge this gap, we propose GUI-Reflection, a novel framework that explicitly integrates self-reflection and error correction capabilities into end-to-end multimodal GUI models throughout dedicated training stages: GUI-specific pre-training, offline supervised fine-tuning (SFT), and online reflection tuning. GU"},"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":"2506.08012","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-06-09T17:59:57Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"6a5306df4b19beccfa12f0c996d8d66b2f601e09471bfc06c78bfcfaea519ad0","abstract_canon_sha256":"c9a476f16f9814ce8425aae68e2e3476658c0e35d8ddd7498a1cdb79b0aa3799"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:18:37.223162Z","signature_b64":"51Gx2RDu++hOKvkj23wxZHrmoE3CPtZ4p93U7ns7vByTuBLsDJxlqC4U6e7MiDGwl/suecFPO3A8bMQCDWVmDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0769af3b3a9a05d475c5b8a84160cfe72bb3437fd26e007a499745a61bc90cd4","last_reissued_at":"2026-07-05T11:18:37.222685Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:18:37.222685Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GUI-Reflection: Empowering Multimodal GUI Models with Self-Reflection Behavior","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.AI","authors_text":"Bo Wang, Jiaheng Yu, Lewei Lu, Penghao Wu, Shengnan Ma, Ziwei Liu","submitted_at":"2025-06-09T17:59:57Z","abstract_excerpt":"Multimodal Large Language Models (MLLMs) have shown great potential in revolutionizing Graphical User Interface (GUI) automation. However, existing GUI models mostly rely on learning from nearly error-free offline trajectories, thus lacking reflection and error recovery capabilities. To bridge this gap, we propose GUI-Reflection, a novel framework that explicitly integrates self-reflection and error correction capabilities into end-to-end multimodal GUI models throughout dedicated training stages: GUI-specific pre-training, offline supervised fine-tuning (SFT), and online reflection tuning. GU"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.08012","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/2506.08012/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":"2506.08012","created_at":"2026-07-05T11:18:37.222744+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.08012v1","created_at":"2026-07-05T11:18:37.222744+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.08012","created_at":"2026-07-05T11:18:37.222744+00:00"},{"alias_kind":"pith_short_12","alias_value":"A5U26OZ2TIC5","created_at":"2026-07-05T11:18:37.222744+00:00"},{"alias_kind":"pith_short_16","alias_value":"A5U26OZ2TIC5I5OF","created_at":"2026-07-05T11:18:37.222744+00:00"},{"alias_kind":"pith_short_8","alias_value":"A5U26OZ2","created_at":"2026-07-05T11:18:37.222744+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.11078","citing_title":"A History-Aware Visually Grounded Critic for Computer Use Agents","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01249","citing_title":"Trust Region On-Policy Distillation","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12191","citing_title":"Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application","ref_index":287,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09442","citing_title":"UIPress: Bringing Optical Token Compression to UI-to-Code Generation","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07110","citing_title":"Securing Computer-Use Agents: A Unified Architecture-Lifecycle Framework for Deployment-Grounded Reliability","ref_index":84,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/A5U26OZ2TIC5I5OFXCUECYGP44","json":"https://pith.science/pith/A5U26OZ2TIC5I5OFXCUECYGP44.json","graph_json":"https://pith.science/api/pith-number/A5U26OZ2TIC5I5OFXCUECYGP44/graph.json","events_json":"https://pith.science/api/pith-number/A5U26OZ2TIC5I5OFXCUECYGP44/events.json","paper":"https://pith.science/paper/A5U26OZ2"},"agent_actions":{"view_html":"https://pith.science/pith/A5U26OZ2TIC5I5OFXCUECYGP44","download_json":"https://pith.science/pith/A5U26OZ2TIC5I5OFXCUECYGP44.json","view_paper":"https://pith.science/paper/A5U26OZ2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.08012&json=true","fetch_graph":"https://pith.science/api/pith-number/A5U26OZ2TIC5I5OFXCUECYGP44/graph.json","fetch_events":"https://pith.science/api/pith-number/A5U26OZ2TIC5I5OFXCUECYGP44/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/A5U26OZ2TIC5I5OFXCUECYGP44/action/timestamp_anchor","attest_storage":"https://pith.science/pith/A5U26OZ2TIC5I5OFXCUECYGP44/action/storage_attestation","attest_author":"https://pith.science/pith/A5U26OZ2TIC5I5OFXCUECYGP44/action/author_attestation","sign_citation":"https://pith.science/pith/A5U26OZ2TIC5I5OFXCUECYGP44/action/citation_signature","submit_replication":"https://pith.science/pith/A5U26OZ2TIC5I5OFXCUECYGP44/action/replication_record"}},"created_at":"2026-07-05T11:18:37.222744+00:00","updated_at":"2026-07-05T11:18:37.222744+00:00"}