DaMo: Data Mixing Optimizer in Fine-tuning Multimodal LLMs for Mobile Phone Agents

TL;DR

DaMo introduces a trainable data mixture optimizer for fine-tuning multimodal large language models, significantly enhancing performance on mobile phone tasks and demonstrating strong generalization across benchmarks.

Contribution

The paper presents DaMo, a novel trainable network that predicts optimal data mixtures for multitask fine-tuning of MLLMs, supported by a new benchmark for mobile phone applications.

Findings

DaMo achieves a 3.38% performance boost on PhoneAgentBench.

DaMo outperforms other methods by 2.57% on average across multiple benchmarks.

DaMo improves BFCL-v3 metrics by 12.47% when used for MLLM optimization.

Abstract

Mobile Phone Agents (MPAs) have emerged as a promising research direction due to their broad applicability across diverse scenarios. While Multimodal Large Language Models (MLLMs) serve as the foundation for MPAs, their effectiveness in handling multiple mobile phone tasks simultaneously remains limited. Although multitask supervised fine-tuning (SFT) is widely adopted for multitask learning, existing approaches struggle to determine optimal training data compositions for peak performance. To address this challenge, we propose DaMo (Data Mixture Optimizer) - a novel solution employing a trainable network that predicts optimal data mixtures by forecasting downstream task performance for any given dataset ratio. To support comprehensive evaluation, we introduce PhoneAgentBench, the first specialized benchmark to evaluate MLLMs on multimodal mobile phone tasks, comprising 1235 QA pairs spanning diverse real-world industrial mobile application scenarios. Demonstrating strong predictive capability (R^2=0.81) in small-scale pilot experiments, DaMo efficiently extrapolates optimal data mixing configurations. Our results show DaMo achieves a 3.38% performance improvement on PhoneAgentBench compared to alternative methods. Furthermore, extensive experiments across established benchmarks including BFCL-v3, MME-Reasoning, MME-Perception, and OCRBench reveal DaMo's superior generalization, outperforming other approaches by 2.57% in terms of average score. When used solely for MLLM optimization on the BFCL-v3 task, DaMo improves the metrics by 12.47% than other methods. Notably, DaMo maintains robust scalability, preserving its effectiveness when applied to other model architectures. The code and dataset are available at https://github.com/OPPO-Mente-Lab/DaMo.git