VLsI: Verbalized Layers-to-Interactions from Large to Small Vision Language Models
This addresses efficiency and deployment issues for VLMs on mobile platforms and robots, representing an incremental improvement in distillation methods.
The paper tackles the computational challenges of scaling vision-language models (VLMs) for resource-constrained devices by proposing VLsI, a family of efficient 2B and 7B models that use layer-wise distillation with verbalizers to align with larger VLMs, achieving performance gains of 11.0% for 2B and 17.4% for 7B over GPT-4V on benchmarks.
The recent surge in high-quality visual instruction tuning samples from closed-source vision-language models (VLMs) such as GPT-4V has accelerated the release of open-source VLMs across various model sizes. However, scaling VLMs to improve performance using larger models brings significant computational challenges, especially for deployment on resource-constrained devices like mobile platforms and robots. To address this, we propose VLsI: Verbalized Layers-to-Interactions, a new VLM family in 2B and 7B model sizes, which prioritizes efficiency without compromising accuracy. VLsI leverages a unique, layer-wise distillation process, introducing intermediate "verbalizers" that map features from each layer to natural language space, allowing smaller VLMs to flexibly align with the reasoning processes of larger VLMs. This approach mitigates the training instability often encountered in output imitation and goes beyond typical final-layer tuning by aligning the small VLMs' layer-wise progression with that of the large ones. We validate VLsI across ten challenging vision-language benchmarks, achieving notable performance gains (11.0% for 2B and 17.4% for 7B) over GPT-4V without the need for model scaling, merging, or architectural changes.