Memory-Space Visual Prompting for Efficient Vision-Language Fine-Tuning
This work addresses efficiency issues in vision-language model fine-tuning for researchers and practitioners, offering a novel approach that improves performance and speed, though it is incremental as it builds on existing parameter-efficient fine-tuning paradigms.
The paper tackles the inefficiency of integrating visual prompts into language model inputs by proposing memory-space visual prompting (MemVP), which concatenates visual prompts with Feed-Forward Network weights for visual knowledge injection, resulting in significantly reduced training time and inference latency while surpassing previous parameter-efficient fine-tuning methods across various vision-language tasks.
Current solutions for efficiently constructing large vision-language (VL) models follow a two-step paradigm: projecting the output of pre-trained vision encoders to the input space of pre-trained language models as visual prompts; and then transferring the models to downstream VL tasks via end-to-end parameter-efficient fine-tuning (PEFT). However, this paradigm still exhibits inefficiency since it significantly increases the input length of the language models. In this paper, in contrast to integrating visual prompts into inputs, we regard visual prompts as additional knowledge that facilitates language models in addressing tasks associated with visual information. Motivated by the finding that Feed-Forward Network (FFN) of language models acts as "key-value memory", we introduce a novel approach termed memory-space visual prompting (MemVP), wherein visual prompts are concatenated with the weights of FFN for visual knowledge injection. Experimental results across various VL tasks and language models reveal that MemVP significantly reduces the training time and inference latency of the finetuned VL models and surpasses the performance of previous PEFT methods. Code: https://github.com/JieShibo/MemVP