DroidCall: A Dataset for LLM-powered Android Intent Invocation

TL;DR

DroidCall introduces a new dataset and fine-tuning approach for small language models to accurately invoke Android intents, enabling efficient on-device mobile agents with enhanced privacy.

Contribution

We created DroidCall, the first dataset for Android intent invocation, and demonstrated that fine-tuned small language models can match or surpass GPT-4o in this task.

Findings

Fine-tuned small models approach GPT-4o performance.

DroidCall contains 10,000 samples for training and testing.

An Android app demonstrates practical deployment.

Abstract

The growing capabilities of large language models in natural language understanding significantly strengthen existing agentic systems. To power performant on-device mobile agents for better data privacy, we introduce DroidCall, the first training and testing dataset for accurate Android intent invocation. With a highly flexible and reusable data generation pipeline, we constructed 10k samples in DroidCall. Given a task instruction in natural language, small language models such as Qwen2.5-3B and Gemma2-2B fine-tuned with DroidCall can approach or even surpass the capabilities of GPT-4o for accurate Android intent invocation. We also provide an end-to-end Android app equipped with these fine-tuned models to demonstrate the Android intent invocation process. The code and dataset are available at https://github.com/UbiquitousLearning/DroidCall.