Summary of Droidcall: a Dataset For Llm-powered Android Intent Invocation, by Weikai Xie et al.
DroidCall: A Dataset for LLM-powered Android Intent Invocation
by Weikai Xie, Li Zhang, Shihe Wang, Rongjie Yi, Mengwei Xu
First submitted to arxiv on: 30 Nov 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces DroidCall, a novel dataset for training and testing mobile agents on Android intent invocation. With the increasing capabilities of large language models in natural language understanding, this research aims to power performant on-device mobile agents for better data privacy. The authors present a flexible and reusable data generation pipeline that constructs 10k samples in DroidCall. Fine-tuning small language models such as Qwen2.5-3B and Gemma2-2B with DroidCall enables them to approach or surpass the capabilities of GPT-4o for accurate Android intent invocation. The authors also provide an end-to-end Android app equipped with these fine-tuned models, demonstrating the Android intent invocation process. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about creating a new way for computers to understand what actions we want them to take on our phones. This is important because it helps keep our personal information safe. The authors created a big dataset of examples that shows how to do this. They tested different types of AI models and found some that are really good at understanding what we want. To show how this works, they made an app that can actually perform these actions on our phones. |
Keywords
» Artificial intelligence » Fine tuning » Gpt » Language understanding
