{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:OZAGJFTRZKRQMIICMHPS45JVFT","short_pith_number":"pith:OZAGJFTR","schema_version":"1.0","canonical_sha256":"7640649671caa306210261df2e75352cead9fd9a1f5cd95a5f808466d5097937","source":{"kind":"arxiv","id":"2503.21620","version":5},"attestation_state":"computed","paper":{"title":"UI-R1: Enhancing Efficient Action Prediction of GUI Agents by Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Rule-based RL on 136 GUI tasks lifts a 3B multimodal model to 22% higher action-prediction accuracy.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Guanjing Xiong, Han Xiao, Hao Wang, Hongsheng Li, Liang Liu, Shuai Ren, Xi Yin, Yaxuan Guo, Yuxiang Chai, Zhengxi Lu","submitted_at":"2025-03-27T15:39:30Z","abstract_excerpt":"The recent DeepSeek-R1 has showcased the emergence of reasoning capabilities in LLMs through reinforcement learning (RL) with rule-based rewards. Despite its success in language models, its application in multi-modal domains, particularly in graphic user interface (GUI) agent tasks, remains under-explored. To address this issue, we propose UI-R1, the first framework to explore how rule-based RL can enhance the reasoning capabilities of multimodal large language models (MLLMs) for GUI action prediction tasks. Specifically, UI-R1 introduces a novel rule-based action reward, enabling model optimi"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2503.21620","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-03-27T15:39:30Z","cross_cats_sorted":[],"title_canon_sha256":"abfd2d349b1b9ba15cf5fb12e0034980e0086d0d412dfc67f35fc9c4a60ab860","abstract_canon_sha256":"996e1401c55cb49fa859097baae52738828675a6956b602244bf691972e2c774"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:48.089658Z","signature_b64":"98/YU+SL25qtk951ZNrfS39EwP5+ZGGtHj95Z0artdyCo5zqMOWJQuPYZTHvELLpoAN33/IO05rGZI7O+nIxAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7640649671caa306210261df2e75352cead9fd9a1f5cd95a5f808466d5097937","last_reissued_at":"2026-05-17T23:38:48.089144Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:48.089144Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"UI-R1: Enhancing Efficient Action Prediction of GUI Agents by Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Rule-based RL on 136 GUI tasks lifts a 3B multimodal model to 22% higher action-prediction accuracy.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Guanjing Xiong, Han Xiao, Hao Wang, Hongsheng Li, Liang Liu, Shuai Ren, Xi Yin, Yaxuan Guo, Yuxiang Chai, Zhengxi Lu","submitted_at":"2025-03-27T15:39:30Z","abstract_excerpt":"The recent DeepSeek-R1 has showcased the emergence of reasoning capabilities in LLMs through reinforcement learning (RL) with rule-based rewards. Despite its success in language models, its application in multi-modal domains, particularly in graphic user interface (GUI) agent tasks, remains under-explored. To address this issue, we propose UI-R1, the first framework to explore how rule-based RL can enhance the reasoning capabilities of multimodal large language models (MLLMs) for GUI action prediction tasks. Specifically, UI-R1 introduces a novel rule-based action reward, enabling model optimi"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"UI-R1-3B achieves significant improvements over the base model (Qwen2.5-VL-3B) on both in-domain and out-of-domain tasks, with average accuracy gains of 22.1% on ScreenSpot, 6.0% on ScreenSpot-Pro, and 12.7% on ANDROIDCONTROL.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The rule-based action reward provides sufficient and unbiased supervision for policy optimization across diverse GUI tasks without post-hoc adjustments or hidden data selection.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"UI-R1 shows rule-based RL with GRPO on 136 GUI tasks improves a 3B MLLM's action prediction accuracy by 6-22% over its base model and matches larger SFT-trained models.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Rule-based RL on 136 GUI tasks lifts a 3B multimodal model to 22% higher action-prediction accuracy.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e3caecae84a458b7a808b92251183a3314729b6427e483a02ffd817da673bad4"},"source":{"id":"2503.21620","kind":"arxiv","version":5},"verdict":{"id":"30738ba4-462a-40ec-b816-9585f8013006","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T10:58:22.234334Z","strongest_claim":"UI-R1-3B achieves significant improvements over the base model (Qwen2.5-VL-3B) on both in-domain and out-of-domain tasks, with average accuracy gains of 22.1% on ScreenSpot, 6.0% on ScreenSpot-Pro, and 12.7% on ANDROIDCONTROL.","one_line_summary":"UI-R1 shows rule-based RL with GRPO on 136 GUI tasks improves a 3B MLLM's action prediction accuracy by 6-22% over its base model and matches larger SFT-trained models.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The rule-based action reward provides sufficient and unbiased supervision for policy optimization across diverse GUI tasks without post-hoc adjustments or hidden data selection.","pith_extraction_headline":"Rule-based RL on 136 GUI tasks lifts a 3B multimodal model to 22% higher action-prediction accuracy."},"references":{"count":18,"sample":[{"doi":"","year":null,"title":"L1: Controlling how long a reasoning model thinks with reinforcement learning","work_id":"ad7236fb-3752-48a9-b782-a384899c45a0","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"arXiv preprint arXiv:2407.17490 , year=","work_id":"07ffe67e-3e5a-406c-84fe-3de84f8dd21d","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"VisRL: Intention-driven visual perception via reinforced reasoning","work_id":"5e295e70-d8e8-4932-8908-266167a5615d","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Navigating the Digital World as Humans Do: Universal Visual Grounding for GUI Agents","work_id":"9def1724-6fd2-4d5b-8339-4c1ee76e62f8","ref_index":4,"cited_arxiv_id":"2410.05243","is_internal_anchor":true},{"doi":"","year":null,"title":"DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning","work_id":"e6b75ad5-2877-4168-97c8-710407094d20","ref_index":5,"cited_arxiv_id":"2501.12948","is_internal_anchor":true}],"resolved_work":18,"snapshot_sha256":"a649709e89f6b39e4fb656d1006ad44a48abaa31c613a19dacd1763fd7b79282","internal_anchors":9},"formal_canon":{"evidence_count":2,"snapshot_sha256":"55ea065743025a6f1c3bb76e55ec151e339d60888732c45d4e54d91087d161b8"},"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":"2503.21620","created_at":"2026-05-17T23:38:48.089230+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.21620v5","created_at":"2026-05-17T23:38:48.089230+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.21620","created_at":"2026-05-17T23:38:48.089230+00:00"},{"alias_kind":"pith_short_12","alias_value":"OZAGJFTRZKRQ","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"OZAGJFTRZKRQMIIC","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"OZAGJFTR","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":30,"internal_anchor_count":30,"sample":[{"citing_arxiv_id":"2505.23678","citing_title":"Grounded Reinforcement Learning for Visual Reasoning","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2605.06534","citing_title":"ROSE: Rollout On Serving GPUs via Cooperative Elasticity for Agentic RL","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15963","citing_title":"PAGER: Bridging the Semantic-Execution Gap in Point-Precise Geometric GUI Control","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19538","citing_title":"CaptchaMind: Training CAPTCHA Solvers via Reinforcement Learning with Explicit Reasoning Supervision","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2605.16883","citing_title":"SE-GA: Memory-Augmented Self-Evolution for GUI Agents","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2604.27859","citing_title":"Rethinking Agentic Reinforcement Learning In Large Language Models","ref_index":55,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14311","citing_title":"Beyond Binary: Reframing GUI Critique as Continuous Semantic Alignment","ref_index":96,"is_internal_anchor":true},{"citing_arxiv_id":"2506.09373","citing_title":"LPO: Towards Accurate GUI Agent Interaction via Location Preference Optimization","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2509.06477","citing_title":"MAS-Bench: A Unified Benchmark for Shortcut-Augmented Hybrid Mobile GUI Agents","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2509.07553","citing_title":"VeriOS: Query-Driven Proactive Human-Agent-GUI Interaction for Trustworthy OS Agents","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2504.14239","citing_title":"InfiGUI-R1: Advancing Multimodal GUI Agents from Reactive Actors to Deliberative Reasoners","ref_index":39,"is_internal_anchor":true},{"citing_arxiv_id":"2507.05791","citing_title":"GTA1: GUI Test-time Scaling Agent","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2602.21858","citing_title":"ProactiveMobile: A Comprehensive Benchmark for Boosting Proactive Intelligence on Mobile Devices","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2504.10458","citing_title":"GUI-R1 : A Generalist R1-Style Vision-Language Action Model For GUI Agents","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14311","citing_title":"Beyond Binary: Reframing GUI Critique as Continuous Semantic Alignment","ref_index":96,"is_internal_anchor":true},{"citing_arxiv_id":"2603.23964","citing_title":"From Pixels to Digital Agents: An Empirical Study on the Taxonomy and Technological Trends of Reinforcement Learning Environments","ref_index":212,"is_internal_anchor":true},{"citing_arxiv_id":"2605.12481","citing_title":"ToolCUA: Towards Optimal GUI-Tool Path Orchestration for Computer Use Agents","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2604.27859","citing_title":"Rethinking Agentic Reinforcement Learning In Large Language Models","ref_index":55,"is_internal_anchor":true},{"citing_arxiv_id":"2503.21776","citing_title":"Video-R1: Reinforcing Video Reasoning in MLLMs","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2604.27859","citing_title":"Rethinking Agentic Reinforcement Learning In Large Language Models","ref_index":55,"is_internal_anchor":true},{"citing_arxiv_id":"2604.24348","citing_title":"OS-SPEAR: A Toolkit for the Safety, Performance,Efficiency, and Robustness Analysis of OS Agents","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2604.24441","citing_title":"AutoGUI-v2: A Comprehensive Multi-Modal GUI Functionality Understanding Benchmark","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2605.06534","citing_title":"ROSE: Rollout On Serving GPUs via Cooperative Elasticity for Agentic RL","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2605.06664","citing_title":"BAMI: Training-Free Bias Mitigation in GUI Grounding","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2604.21268","citing_title":"Measure Twice, Click Once: Co-evolving Proposer and Visual Critic via Reinforcement Learning for GUI Grounding","ref_index":86,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT","json":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT.json","graph_json":"https://pith.science/api/pith-number/OZAGJFTRZKRQMIICMHPS45JVFT/graph.json","events_json":"https://pith.science/api/pith-number/OZAGJFTRZKRQMIICMHPS45JVFT/events.json","paper":"https://pith.science/paper/OZAGJFTR"},"agent_actions":{"view_html":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT","download_json":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT.json","view_paper":"https://pith.science/paper/OZAGJFTR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.21620&json=true","fetch_graph":"https://pith.science/api/pith-number/OZAGJFTRZKRQMIICMHPS45JVFT/graph.json","fetch_events":"https://pith.science/api/pith-number/OZAGJFTRZKRQMIICMHPS45JVFT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT/action/storage_attestation","attest_author":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT/action/author_attestation","sign_citation":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT/action/citation_signature","submit_replication":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT/action/replication_record"}},"created_at":"2026-05-17T23:38:48.089230+00:00","updated_at":"2026-05-17T23:38:48.089230+00:00"}