Keeping Yourself is Important in Downstream Tuning Multimodal Large Language Model
This work provides a systematic analysis and benchmarking for researchers and practitioners tuning MLLMs, but it is incremental as it reviews and classifies existing methodologies rather than introducing new ones.
The paper tackles the challenges of tuning multimodal large language models (MLLMs) for downstream tasks, specifically addressing task-expert specialization and open-world stabilization, by systematically reviewing and benchmarking three tuning paradigms across various architectures and tasks.
Multi-modal Large Language Models (MLLMs) integrate visual and linguistic reasoning to address complex tasks such as image captioning and visual question answering. While MLLMs demonstrate remarkable versatility, MLLMs appears limited performance on special applications. But tuning MLLMs for downstream tasks encounters two key challenges: Task-Expert Specialization, where distribution shifts between pre-training and target datasets constrain target performance, and Open-World Stabilization, where catastrophic forgetting erases the model general knowledge. In this work, we systematically review recent advancements in MLLM tuning methodologies, classifying them into three paradigms: (I) Selective Tuning, (II) Additive Tuning, and (III) Reparameterization Tuning. Furthermore, we benchmark these tuning strategies across popular MLLM architectures and diverse downstream tasks to establish standardized evaluation analysis and systematic tuning principles. Finally, we highlight several open challenges in this domain and propose future research directions. To facilitate ongoing progress in this rapidly evolving field, we provide a public repository that continuously tracks developments: https://github.com/WenkeHuang/Awesome-MLLM-Tuning.