{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:AA7GL5KXVTYVHWY54PX552YUML","short_pith_number":"pith:AA7GL5KX","schema_version":"1.0","canonical_sha256":"003e65f557acf153db1de3efdeeb1462ec2f5398660df54af3b137d16fb9a459","source":{"kind":"arxiv","id":"2505.16282","version":1},"attestation_state":"computed","paper":{"title":"ARPO:End-to-End Policy Optimization for GUI Agents with Experience Replay","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chi-Wing Fu, Fanbin Lu, Jiaya Jia, Shu Liu, Zhisheng Zhong","submitted_at":"2025-05-22T06:24:32Z","abstract_excerpt":"Training large language models (LLMs) as interactive agents for controlling graphical user interfaces (GUIs) presents a unique challenge to optimize long-horizon action sequences with multimodal feedback from complex environments. While recent works have advanced multi-turn reinforcement learning (RL) for reasoning and tool-using capabilities in LLMs, their application to GUI-based agents remains relatively underexplored due to the difficulty of sparse rewards, delayed feedback, and high rollout costs. In this paper, we investigate end-to-end policy optimization for vision-language-based GUI 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":"2505.16282","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-05-22T06:24:32Z","cross_cats_sorted":[],"title_canon_sha256":"ed26386cbd81ea34d56e9e16692e71e19aca7fece765d9a0c6b78902173f602f","abstract_canon_sha256":"4d68848ea18ef0dd65b420d356dc1c2633b621d7b5e02267588fb55496faa3e8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:07:32.690257Z","signature_b64":"hLYsQ0AYob1cAUxt/IYhI5JZ6IRLijw7lfs12RUFk6f2arROG7AfcgRoowvca70GJSQxBrpNuiREG+9J4yIiAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"003e65f557acf153db1de3efdeeb1462ec2f5398660df54af3b137d16fb9a459","last_reissued_at":"2026-07-05T11:07:32.689825Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:07:32.689825Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ARPO:End-to-End Policy Optimization for GUI Agents with Experience Replay","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chi-Wing Fu, Fanbin Lu, Jiaya Jia, Shu Liu, Zhisheng Zhong","submitted_at":"2025-05-22T06:24:32Z","abstract_excerpt":"Training large language models (LLMs) as interactive agents for controlling graphical user interfaces (GUIs) presents a unique challenge to optimize long-horizon action sequences with multimodal feedback from complex environments. While recent works have advanced multi-turn reinforcement learning (RL) for reasoning and tool-using capabilities in LLMs, their application to GUI-based agents remains relatively underexplored due to the difficulty of sparse rewards, delayed feedback, and high rollout costs. In this paper, we investigate end-to-end policy optimization for vision-language-based GUI a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.16282","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/2505.16282/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":"2505.16282","created_at":"2026-07-05T11:07:32.689894+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.16282v1","created_at":"2026-07-05T11:07:32.689894+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.16282","created_at":"2026-07-05T11:07:32.689894+00:00"},{"alias_kind":"pith_short_12","alias_value":"AA7GL5KXVTYV","created_at":"2026-07-05T11:07:32.689894+00:00"},{"alias_kind":"pith_short_16","alias_value":"AA7GL5KXVTYVHWY5","created_at":"2026-07-05T11:07:32.689894+00:00"},{"alias_kind":"pith_short_8","alias_value":"AA7GL5KX","created_at":"2026-07-05T11:07:32.689894+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25451","citing_title":"Learning with a Single Rollout via Monte Carlo Pass@k Critic","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2606.21943","citing_title":"Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning","ref_index":125,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07027","citing_title":"StainFlow: Entity-Stain Tracking and Evidence Linking for Process Rewards in GUI Agents","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01281","citing_title":"RLVR without Ineffective Samples: Group Prioritized Off-Policy Optimization for LLM Reasoning","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2606.26122","citing_title":"DocArena: Turning Raw Documents into Controllable Training Environments for Document Search Agents","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2601.12538","citing_title":"Agentic Reasoning for Large Language Models","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09574","citing_title":"Turing Test on Screen: A Benchmark for Mobile GUI Agent Humanization","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2509.02544","citing_title":"UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learning","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12481","citing_title":"ToolCUA: Towards Optimal GUI-Tool Path Orchestration for Computer Use Agents","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08978","citing_title":"Learning to Explore: Scaling Agentic Reasoning via Exploration-Aware Policy Optimization","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08978","citing_title":"Learning to Explore: Scaling Agentic Reasoning via Exploration-Aware Policy Optimization","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01208","citing_title":"Faithful Mobile GUI Agents with Guided Advantage Estimator","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02572","citing_title":"On Training Large Language Models for Long-Horizon Tasks: An Empirical Study of Horizon Length","ref_index":66,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AA7GL5KXVTYVHWY54PX552YUML","json":"https://pith.science/pith/AA7GL5KXVTYVHWY54PX552YUML.json","graph_json":"https://pith.science/api/pith-number/AA7GL5KXVTYVHWY54PX552YUML/graph.json","events_json":"https://pith.science/api/pith-number/AA7GL5KXVTYVHWY54PX552YUML/events.json","paper":"https://pith.science/paper/AA7GL5KX"},"agent_actions":{"view_html":"https://pith.science/pith/AA7GL5KXVTYVHWY54PX552YUML","download_json":"https://pith.science/pith/AA7GL5KXVTYVHWY54PX552YUML.json","view_paper":"https://pith.science/paper/AA7GL5KX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.16282&json=true","fetch_graph":"https://pith.science/api/pith-number/AA7GL5KXVTYVHWY54PX552YUML/graph.json","fetch_events":"https://pith.science/api/pith-number/AA7GL5KXVTYVHWY54PX552YUML/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AA7GL5KXVTYVHWY54PX552YUML/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AA7GL5KXVTYVHWY54PX552YUML/action/storage_attestation","attest_author":"https://pith.science/pith/AA7GL5KXVTYVHWY54PX552YUML/action/author_attestation","sign_citation":"https://pith.science/pith/AA7GL5KXVTYVHWY54PX552YUML/action/citation_signature","submit_replication":"https://pith.science/pith/AA7GL5KXVTYVHWY54PX552YUML/action/replication_record"}},"created_at":"2026-07-05T11:07:32.689894+00:00","updated_at":"2026-07-05T11:07:32.689894+00:00"}