From Behavioral Performance to Internal Competence: Interpreting Vision-Language Models with VLM-Lens
This provides a tool for researchers to better understand and improve VLMs, but it is incremental as it builds on existing interpretability methods without introducing new paradigms.
The authors tackled the problem of interpreting vision-language models (VLMs) by introducing VLM-Lens, a toolkit that enables systematic benchmarking and analysis by extracting intermediate outputs from any layer, supporting 16 base VLMs and over 30 variants, and revealing systematic differences in hidden representations across layers and concepts.
We introduce VLM-Lens, a toolkit designed to enable systematic benchmarking, analysis, and interpretation of vision-language models (VLMs) by supporting the extraction of intermediate outputs from any layer during the forward pass of open-source VLMs. VLM-Lens provides a unified, YAML-configurable interface that abstracts away model-specific complexities and supports user-friendly operation across diverse VLMs. It currently supports 16 state-of-the-art base VLMs and their over 30 variants, and is extensible to accommodate new models without changing the core logic. The toolkit integrates easily with various interpretability and analysis methods. We demonstrate its usage with two simple analytical experiments, revealing systematic differences in the hidden representations of VLMs across layers and target concepts. VLM-Lens is released as an open-sourced project to accelerate community efforts in understanding and improving VLMs.