Small-Large Collaboration: Training-efficient Concept Personalization for Large VLM using a Meta Personalized Small VLM
This enables cost-effective personalization of large VLMs for daily assistant applications, though it is incremental as it builds on existing model collaboration concepts.
The paper tackles the problem of personalizing large vision-language models (VLMs) efficiently by proposing a Small-Large Collaboration (SLC) framework, where a meta personalized small VLM generates personalized information and a large VLM integrates it with a test-time reflection strategy, achieving training efficiency and broad applicability without specific performance numbers.
Personalizing Vision-Language Models (VLMs) to transform them into daily assistants has emerged as a trending research direction. However, leading companies like OpenAI continue to increase model size and develop complex designs such as the chain of thought (CoT). While large VLMs are proficient in complex multi-modal understanding, their high training costs and limited access via paid APIs restrict direct personalization. Conversely, small VLMs are easily personalized and freely available, but they lack sufficient reasoning capabilities. Inspired by this, we propose a novel collaborative framework named Small-Large Collaboration (SLC) for large VLM personalization, where the small VLM is responsible for generating personalized information, while the large model integrates this personalized information to deliver accurate responses. To effectively incorporate personalized information, we develop a test-time reflection strategy, preventing the potential hallucination of the small VLM. Since SLC only needs to train a meta personalized small VLM for the large VLMs, the overall process is training-efficient. To the best of our knowledge, this is the first training-efficient framework that supports both open-source and closed-source large VLMs, enabling broader real-world personalized applications. We conduct thorough experiments across various benchmarks and large VLMs to demonstrate the effectiveness of the proposed SLC framework. The code will be released at https://github.com/Hhankyangg/SLC.