{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:OZAGJFTRZKRQMIICMHPS45JVFT","short_pith_number":"pith:OZAGJFTR","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"},"canonical_sha256":"7640649671caa306210261df2e75352cead9fd9a1f5cd95a5f808466d5097937","source":{"kind":"arxiv","id":"2503.21620","version":5},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.21620","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"arxiv_version","alias_value":"2503.21620v5","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.21620","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"pith_short_12","alias_value":"OZAGJFTRZKRQ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"OZAGJFTRZKRQMIIC","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"OZAGJFTR","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:OZAGJFTRZKRQMIICMHPS45JVFT","target":"record","payload":{"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"},"canonical_sha256":"7640649671caa306210261df2e75352cead9fd9a1f5cd95a5f808466d5097937","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"},"source_kind":"arxiv","source_id":"2503.21620","source_version":5,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"jVaBN9P/0cBUHAq3zZipwc3kbvjvFvQS4X+CX9MQNd6xwzNOc6TuOPHbTnJTX2rBSvE9UNkhlTM93gcR1whjDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-03T00:42:39.562129Z"},"content_sha256":"f36902162dd4b0c234b3867a1dc61c8fbbaf9b4c8421e985291b17779c506b93","schema_version":"1.0","event_id":"sha256:f36902162dd4b0c234b3867a1dc61c8fbbaf9b4c8421e985291b17779c506b93"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:OZAGJFTRZKRQMIICMHPS45JVFT","target":"graph","payload":{"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"},"verdict_id":"30738ba4-462a-40ec-b816-9585f8013006"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"FtL8SZ+iDP+lIotMcgRUUg9hpqctkvbw7BBGIl+Aw2QQrnsgmV8SEP+0S1Y7/P0duIQ72+urf2qWvUZfBYLvAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-03T00:42:39.562660Z"},"content_sha256":"41447f67e52f52809c1c890a3a220ddea864251f8b90939722184abb6d3fbdd4","schema_version":"1.0","event_id":"sha256:41447f67e52f52809c1c890a3a220ddea864251f8b90939722184abb6d3fbdd4"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT/bundle.json","state_url":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/OZAGJFTRZKRQMIICMHPS45JVFT/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-03T00:42:39Z","links":{"resolver":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT","bundle":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT/bundle.json","state":"https://pith.science/pith/OZAGJFTRZKRQMIICMHPS45JVFT/state.json","well_known_bundle":"https://pith.science/.well-known/pith/OZAGJFTRZKRQMIICMHPS45JVFT/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:OZAGJFTRZKRQMIICMHPS45JVFT","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"996e1401c55cb49fa859097baae52738828675a6956b602244bf691972e2c774","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-03-27T15:39:30Z","title_canon_sha256":"abfd2d349b1b9ba15cf5fb12e0034980e0086d0d412dfc67f35fc9c4a60ab860"},"schema_version":"1.0","source":{"id":"2503.21620","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.21620","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"arxiv_version","alias_value":"2503.21620v5","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.21620","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"pith_short_12","alias_value":"OZAGJFTRZKRQ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"OZAGJFTRZKRQMIIC","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"OZAGJFTR","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:41447f67e52f52809c1c890a3a220ddea864251f8b90939722184abb6d3fbdd4","target":"graph","created_at":"2026-05-17T23:38:48Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Rule-based RL on 136 GUI tasks lifts a 3B multimodal model to 22% higher action-prediction accuracy."}],"snapshot_sha256":"e3caecae84a458b7a808b92251183a3314729b6427e483a02ffd817da673bad4"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"55ea065743025a6f1c3bb76e55ec151e339d60888732c45d4e54d91087d161b8"},"paper":{"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","authors_text":"Guanjing Xiong, Han Xiao, Hao Wang, Hongsheng Li, Liang Liu, Shuai Ren, Xi Yin, Yaxuan Guo, Yuxiang Chai, Zhengxi Lu","cross_cats":[],"headline":"Rule-based RL on 136 GUI tasks lifts a 3B multimodal model to 22% higher action-prediction accuracy.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-03-27T15:39:30Z","title":"UI-R1: Enhancing Efficient Action Prediction of GUI Agents by Reinforcement Learning"},"references":{"count":18,"internal_anchors":9,"resolved_work":18,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"L1: Controlling how long a reasoning model thinks with reinforcement learning","work_id":"ad7236fb-3752-48a9-b782-a384899c45a0","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"arXiv preprint arXiv:2407.17490 , year=","work_id":"07ffe67e-3e5a-406c-84fe-3de84f8dd21d","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"VisRL: Intention-driven visual perception via reinforced reasoning","work_id":"5e295e70-d8e8-4932-8908-266167a5615d","year":2025},{"cited_arxiv_id":"2410.05243","doi":"","is_internal_anchor":true,"ref_index":4,"title":"Navigating the Digital World as Humans Do: Universal Visual Grounding for GUI Agents","work_id":"9def1724-6fd2-4d5b-8339-4c1ee76e62f8","year":null},{"cited_arxiv_id":"2501.12948","doi":"","is_internal_anchor":true,"ref_index":5,"title":"DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning","work_id":"e6b75ad5-2877-4168-97c8-710407094d20","year":null}],"snapshot_sha256":"a649709e89f6b39e4fb656d1006ad44a48abaa31c613a19dacd1763fd7b79282"},"source":{"id":"2503.21620","kind":"arxiv","version":5},"verdict":{"created_at":"2026-05-16T10:58:22.234334Z","id":"30738ba4-462a-40ec-b816-9585f8013006","model_set":{"reader":"grok-4.3"},"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","pith_extraction_headline":"Rule-based RL on 136 GUI tasks lifts a 3B multimodal model to 22% higher action-prediction accuracy.","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.","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."}},"verdict_id":"30738ba4-462a-40ec-b816-9585f8013006"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:f36902162dd4b0c234b3867a1dc61c8fbbaf9b4c8421e985291b17779c506b93","target":"record","created_at":"2026-05-17T23:38:48Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"996e1401c55cb49fa859097baae52738828675a6956b602244bf691972e2c774","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-03-27T15:39:30Z","title_canon_sha256":"abfd2d349b1b9ba15cf5fb12e0034980e0086d0d412dfc67f35fc9c4a60ab860"},"schema_version":"1.0","source":{"id":"2503.21620","kind":"arxiv","version":5}},"canonical_sha256":"7640649671caa306210261df2e75352cead9fd9a1f5cd95a5f808466d5097937","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"7640649671caa306210261df2e75352cead9fd9a1f5cd95a5f808466d5097937","first_computed_at":"2026-05-17T23:38:48.089144Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:48.089144Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"98/YU+SL25qtk951ZNrfS39EwP5+ZGGtHj95Z0artdyCo5zqMOWJQuPYZTHvELLpoAN33/IO05rGZI7O+nIxAw==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:48.089658Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.21620","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f36902162dd4b0c234b3867a1dc61c8fbbaf9b4c8421e985291b17779c506b93","sha256:41447f67e52f52809c1c890a3a220ddea864251f8b90939722184abb6d3fbdd4"],"state_sha256":"bbc27349dc7c0f3689de26ad85d433d81ea118a2ef593843443f8fcfdb11cb9b"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"iULxafS9Z4DiyNI+Tga4psshnW9FSFMTMJ80G9nsk5ELn/Py58l6DsrFzlCq8Gj/P4FMQc4wry42HFhZ2npqCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-03T00:42:39.564907Z","bundle_sha256":"ced69b047b9631b895e63bfe293aa09b1ed97ccf98d035cce6e269a1944d5a69"}}