Improve model card: add metadata and links

#1
by nielsr HF Staff - opened
Files changed (1) hide show
  1. README.md +16 -3
README.md CHANGED
@@ -1,17 +1,30 @@
 
 
 
 
 
 
1
  ### Model Overview
2
 
3
  This model is a **Query Rewriter** implemented as described in the paper:
4
 
5
- **Quick on the Uptake: Eliciting Implicit Intents from Human Demonstrations for Personalized Mobile-Use Agents**
 
 
 
 
6
 
7
- It is initialized with weights from the **Qwen3-4B** model and subsequently fine-tuned (warmed up) for its specific task of understanding and rewriting user queries based on demonstrated implicit intents.
 
 
8
 
9
  ### Citation
10
 
11
  ```bibtex
12
  @article{wu2025quick,
13
  title={Quick on the Uptake: Eliciting Implicit Intents from Human Demonstrations for Personalized Mobile-Use Agents},
14
- author={Wu, Zheng and Huang, Heyuan and Yang, Yanjia and Song, Yuanyi and Lou, Xingyu and Liu, Weiwen and Zhang, Weinan and Wang, Jun and Zhang, Zhuosheng},
15
  journal={arXiv preprint arXiv:2508.08645},
16
  year={2025}
17
  }
 
 
1
+ ---
2
+ library_name: transformers
3
+ pipeline_tag: text-generation
4
+ base_model: Qwen/Qwen3-4B
5
+ ---
6
+
7
  ### Model Overview
8
 
9
  This model is a **Query Rewriter** implemented as described in the paper:
10
 
11
+ [**Quick on the Uptake: Eliciting Implicit Intents from Human Demonstrations for Personalized Mobile-Use Agents**](https://arxiv.org/abs/2508.08645)
12
+
13
+ It is part of the **IFRAgent** framework, which is designed to enhance the alignment between mobile-use agents and human intent by analyzing both explicit intention flows (step sequences) and implicit intention flows (personal preferences). The Query Rewriter leverages habit repositories and standard operating procedures (SOPs) to generate personalized queries from raw, ambiguous user input.
14
+
15
+ The model is initialized with weights from **Qwen3-4B** and fine-tuned for the task of understanding and rewriting user queries based on implicit intents.
16
 
17
+ - **Paper:** [https://arxiv.org/abs/2508.08645](https://arxiv.org/abs/2508.08645)
18
+ - **GitHub Repository:** [MadeAgents/Quick-on-the-Uptake](https://github.com/MadeAgents/Quick-on-the-Uptake)
19
+ - **Dataset:** [MobileIAR](https://huggingface.co/datasets/wuuuuuz/MobileIAR)
20
 
21
  ### Citation
22
 
23
  ```bibtex
24
  @article{wu2025quick,
25
  title={Quick on the Uptake: Eliciting Implicit Intents from Human Demonstrations for Personalized Mobile-Use Agents},
26
+ author={Wu, Zheng and Huang, Heyuan d Yang, Yanjia and Song, Yuanyi and Lou, Xingyu and Liu, Weiwen and Zhang, Weinan and Wang, Jun and Zhang, Zhuosheng},
27
  journal={arXiv preprint arXiv:2508.08645},
28
  year={2025}
29
  }
30
+ ```