Add metadata, paper and GitHub links
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,9 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
### Dataset Overview
|
| 2 |
|
| 3 |
-
|
| 4 |
|
| 5 |
-
|
| 6 |
|
|
|
|
|
|
|
| 7 |
|
| 8 |
### Citation
|
| 9 |
|
|
@@ -14,3 +21,4 @@ This dataset is a **User-Specific Personalized GUI Mobile Agents Dataset** imple
|
|
| 14 |
journal={arXiv preprint arXiv:2508.08645},
|
| 15 |
year={2025}
|
| 16 |
}
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
task_categories:
|
| 3 |
+
- image-text-to-text
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
### Dataset Overview
|
| 7 |
|
| 8 |
+
**MobileIAR** is a User-Specific Personalized GUI Mobile Agents Dataset introduced in the paper "[Quick on the Uptake: Eliciting Implicit Intents from Human Demonstrations for Personalized Mobile-Use Agents](https://huggingface.co/papers/2508.08645)".
|
| 9 |
|
| 10 |
+
The dataset contains human-intent-aligned actions and ground-truth actions, enabling a comprehensive assessment of mobile-use agents' understanding of both explicit intention flows (e.g., step sequences) and implicit intention flows (e.g., personal preferences).
|
| 11 |
|
| 12 |
+
- **Paper:** [https://huggingface.co/papers/2508.08645](https://huggingface.co/papers/2508.08645)
|
| 13 |
+
- **Code:** [https://github.com/MadeAgents/Quick-on-the-Uptake](https://github.com/MadeAgents/Quick-on-the-Uptake)
|
| 14 |
|
| 15 |
### Citation
|
| 16 |
|
|
|
|
| 21 |
journal={arXiv preprint arXiv:2508.08645},
|
| 22 |
year={2025}
|
| 23 |
}
|
| 24 |
+
```
|