GUI-CIDER: Mid-training GUI Agents via Causal Internalization and Density-aware Exemplar Reselection
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911724966248448 |
|---|---|
| author | Wu, Zheng Han, Chengcheng Lu, Zhengxi Ju, Tianjie Chen, Yanyu Gu, Qi Cai, Xunliang Zhang, Zhuosheng |
| author_facet | Wu, Zheng Han, Chengcheng Lu, Zhengxi Ju, Tianjie Chen, Yanyu Gu, Qi Cai, Xunliang Zhang, Zhuosheng |
| contents | Despite the rapid progress of multimodal large language models in building Graphical User Interface (GUI) agents, their real-world task completion is fundamentally bottlenecked by a lack of world knowledge about GUI operations. Existing solutions typically rely on expensive multi-agent scaffolding or conventional post-training paradigms, such as Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL). However, post-training only allows agents to implicitly absorb world knowledge through action annotations or reward signals, leading to inefficient trajectory memorization rather than genuine comprehension. Therefore, an approach that enables explicit learning of this knowledge is imperative. To this end, we propose GUI-CIDER, a mid-training method that explicitly internalizes GUI world knowledge through Causal Internalization and Density-aware Exemplar Reselection. GUI-CIDER operates in three stages: (1) data synthesis, which distills static planning and dynamic causal knowledge from GUI trajectories into text; (2) exemplar reselection, which filters the corpus by rewarding causal structures and penalizing semantic redundancy; and (3) mid-training, where the refined data is used to embed the acquired knowledge. Extensive experiments on two GUI knowledge benchmarks and three task completion benchmarks demonstrate that GUI-CIDER consistently improves both the agent's understanding of GUI operations and its task success rates.The codes are available at https://github.com/Wuzheng02/GUI-CIDER. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_28534 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | GUI-CIDER: Mid-training GUI Agents via Causal Internalization and Density-aware Exemplar Reselection Wu, Zheng Han, Chengcheng Lu, Zhengxi Ju, Tianjie Chen, Yanyu Gu, Qi Cai, Xunliang Zhang, Zhuosheng Computation and Language Despite the rapid progress of multimodal large language models in building Graphical User Interface (GUI) agents, their real-world task completion is fundamentally bottlenecked by a lack of world knowledge about GUI operations. Existing solutions typically rely on expensive multi-agent scaffolding or conventional post-training paradigms, such as Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL). However, post-training only allows agents to implicitly absorb world knowledge through action annotations or reward signals, leading to inefficient trajectory memorization rather than genuine comprehension. Therefore, an approach that enables explicit learning of this knowledge is imperative. To this end, we propose GUI-CIDER, a mid-training method that explicitly internalizes GUI world knowledge through Causal Internalization and Density-aware Exemplar Reselection. GUI-CIDER operates in three stages: (1) data synthesis, which distills static planning and dynamic causal knowledge from GUI trajectories into text; (2) exemplar reselection, which filters the corpus by rewarding causal structures and penalizing semantic redundancy; and (3) mid-training, where the refined data is used to embed the acquired knowledge. Extensive experiments on two GUI knowledge benchmarks and three task completion benchmarks demonstrate that GUI-CIDER consistently improves both the agent's understanding of GUI operations and its task success rates.The codes are available at https://github.com/Wuzheng02/GUI-CIDER. |
| title | GUI-CIDER: Mid-training GUI Agents via Causal Internalization and Density-aware Exemplar Reselection |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2605.28534 |