{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:MZMXOMOWSI7VDVMSVATAYEC725","short_pith_number":"pith:MZMXOMOW","schema_version":"1.0","canonical_sha256":"66597731d6923f51d592a8260c105fd758264cc405bee22e56d07cf2387b2f2d","source":{"kind":"arxiv","id":"2502.18906","version":1},"attestation_state":"computed","paper":{"title":"VEM: Environment-Free Exploration for Training GUI Agent with Value Environment Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chaoyun Zhang, Dongmei Zhang, Fangkai Yang, Jiani Zheng, Lingrui Mei, Lu Wang, Qingwei Lin, Qi Zhang, Saravan Rajmohan, Wenjie Yin","submitted_at":"2025-02-26T07:52:02Z","abstract_excerpt":"Training Vision-Language Models (VLMs) for Graphical User Interfaces (GUI) agents via Reinforcement Learning (RL) faces critical challenges: environment-based RL requires costly interactions, while environment-free methods struggle with distribution shift and reward generalization. We propose an environment-free RL framework that decouples value estimation from policy optimization by leveraging a pretrained Value Environment Model (VEM). VEM predicts state-action values directly from offline data, distilling human-like priors about GUI interaction outcomes without requiring next-state predicti"},"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":"2502.18906","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-26T07:52:02Z","cross_cats_sorted":[],"title_canon_sha256":"85bb3548e30364f897796db3d5a0663fb4d60e8057bb50db50269f759ad04f2f","abstract_canon_sha256":"35b502b18555f3030f0b8a62621926b3b7423089f3e5a613165fb905e6306e01"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:20:22.836827Z","signature_b64":"ETL9zlfxF+WSSZwUdiQt83uhAGkKvqVxU6GOeQyWC9RNOnUAqxPZL92duxnI4OZwb9yedP3QkJs/dNAJ5hWyDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"66597731d6923f51d592a8260c105fd758264cc405bee22e56d07cf2387b2f2d","last_reissued_at":"2026-07-05T10:20:22.836341Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:20:22.836341Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"VEM: Environment-Free Exploration for Training GUI Agent with Value Environment Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chaoyun Zhang, Dongmei Zhang, Fangkai Yang, Jiani Zheng, Lingrui Mei, Lu Wang, Qingwei Lin, Qi Zhang, Saravan Rajmohan, Wenjie Yin","submitted_at":"2025-02-26T07:52:02Z","abstract_excerpt":"Training Vision-Language Models (VLMs) for Graphical User Interfaces (GUI) agents via Reinforcement Learning (RL) faces critical challenges: environment-based RL requires costly interactions, while environment-free methods struggle with distribution shift and reward generalization. We propose an environment-free RL framework that decouples value estimation from policy optimization by leveraging a pretrained Value Environment Model (VEM). VEM predicts state-action values directly from offline data, distilling human-like priors about GUI interaction outcomes without requiring next-state predicti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.18906","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/2502.18906/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":"2502.18906","created_at":"2026-07-05T10:20:22.836405+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.18906v1","created_at":"2026-07-05T10:20:22.836405+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.18906","created_at":"2026-07-05T10:20:22.836405+00:00"},{"alias_kind":"pith_short_12","alias_value":"MZMXOMOWSI7V","created_at":"2026-07-05T10:20:22.836405+00:00"},{"alias_kind":"pith_short_16","alias_value":"MZMXOMOWSI7VDVMS","created_at":"2026-07-05T10:20:22.836405+00:00"},{"alias_kind":"pith_short_8","alias_value":"MZMXOMOW","created_at":"2026-07-05T10:20:22.836405+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2509.21816","citing_title":"From Task to Tutorial: An Automated GUI Framework for Excel Tutorial Document and Video Creation","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MZMXOMOWSI7VDVMSVATAYEC725","json":"https://pith.science/pith/MZMXOMOWSI7VDVMSVATAYEC725.json","graph_json":"https://pith.science/api/pith-number/MZMXOMOWSI7VDVMSVATAYEC725/graph.json","events_json":"https://pith.science/api/pith-number/MZMXOMOWSI7VDVMSVATAYEC725/events.json","paper":"https://pith.science/paper/MZMXOMOW"},"agent_actions":{"view_html":"https://pith.science/pith/MZMXOMOWSI7VDVMSVATAYEC725","download_json":"https://pith.science/pith/MZMXOMOWSI7VDVMSVATAYEC725.json","view_paper":"https://pith.science/paper/MZMXOMOW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.18906&json=true","fetch_graph":"https://pith.science/api/pith-number/MZMXOMOWSI7VDVMSVATAYEC725/graph.json","fetch_events":"https://pith.science/api/pith-number/MZMXOMOWSI7VDVMSVATAYEC725/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MZMXOMOWSI7VDVMSVATAYEC725/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MZMXOMOWSI7VDVMSVATAYEC725/action/storage_attestation","attest_author":"https://pith.science/pith/MZMXOMOWSI7VDVMSVATAYEC725/action/author_attestation","sign_citation":"https://pith.science/pith/MZMXOMOWSI7VDVMSVATAYEC725/action/citation_signature","submit_replication":"https://pith.science/pith/MZMXOMOWSI7VDVMSVATAYEC725/action/replication_record"}},"created_at":"2026-07-05T10:20:22.836405+00:00","updated_at":"2026-07-05T10:20:22.836405+00:00"}