Research map / Multi-agent systems and games / Human-AI and goal-conditioned
ShowUI-p: Flow-based Generative Models as GUI Dexterous Hands
Siyuan Hu, Kevin Qinghong Lin, Mike Zheng Shou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use flow matching action expert instead of diffusion policy (ablation) | ShowUI-π (SmolVLM backbone) · ScreenDrag offline | endpoint accuracy (%): 47.33 -> 78.55 | Improved |
| use flow matching instead of language-token action modeling (ablation) | ShowUI-π (SmolVLM backbone) · ScreenDrag offline | endpoint accuracy (%): 0.40 -> 78.55 | Improved |
| add start/end loss reweighting (scale 10) (ablation) | ShowUI-π · ScreenDrag online overall | task success rate (%): 10.49 -> 26.98 | Improved |
| increase reweighting scale from 10 to 15 (ablation) | ShowUI-π · ScreenDrag online overall | task success rate (%): 26.98 -> 20.80 | Hurt |
| add directional regularization (lambda 0.1) (ablation) | ShowUI-π · ScreenDrag online overall | task success rate (%): 12.63 -> 26.98 | Improved |
| increase action chunk size from 10 to 20 (ablation) | ShowUI-π · ScreenDrag offline, exec step 1 | endpoint accuracy (%): 75.45 -> 78.55 | Improved |
| execute more steps (5 vs 1) before re-observation (ablation) | ShowUI-π chunk 10 · ScreenDrag offline | endpoint accuracy (%): 75.45 -> 66.34 | Hurt |
| larger chunk size 20 when executing 5 steps (ablation) | ShowUI-π · ScreenDrag offline, exec step 5 | endpoint accuracy (%): 66.34 -> 79.22; error 241.60 -> 205.23 | Improved |
| use unified click/drag head instead of separate heads (ablation) | ShowUI-π · ScreenDrag | online drag success rate (%): 23.25 -> 26.98 (offline 79.22 -> 78.55) | Improved |
| scale proprietary or open baseline size (OpenCUA-7B vs 32B) | OpenCUA · ScreenDrag online | task success rate (%): 20.79 (32B) vs 21.98 (7B) | No clear effect |
Closest papers by meaning
- ShowUI: One Vision-Language-Action Model for GUI Visual Agent (CVPR 2025 · similarity 0.88)
- UI-Vision: A Desktop-centric GUI Benchmark for Visual Perception and Interaction (ICML 2025 · similarity 0.87)
- VideoGUI: A Benchmark for GUI Automation from Instructional Videos (NeurIPS 2024 · similarity 0.87)
- ManipTrans: Efficient Dexterous Bimanual Manipulation Transfer via Residual Learning (CVPR 2025 · similarity 0.87)
- FreqPolicy: Efficient Flow-based Visuomotor Policy via Frequency Consistency (NeurIPS 2025 · similarity 0.86)
- RobotSmith: Generative Robotic Tool Design for Acquisition of Complex Manipulation Skills (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
