Quantifying and Enabling the Interpretability of CLIP-like Models
This addresses the lack of understanding in CLIP's inner workings for researchers and practitioners, but is incremental as it builds on existing models and methods.
The study quantified the interpretability of CLIP-like models by analyzing attention heads using metrics for property consistency and disentanglement, finding that larger models are generally more interpretable, and introduced CLIP-InterpreT as a tool for interpretability analysis.
CLIP is one of the most popular foundational models and is heavily used for many vision-language tasks. However, little is known about the inner workings of CLIP. To bridge this gap we propose a study to quantify the interpretability in CLIP like models. We conduct this study on six different CLIP models from OpenAI and OpenCLIP which vary by size, type of pre-training data and patch size. Our approach begins with using the TEXTSPAN algorithm and in-context learning to break down individual attention heads into specific properties. We then evaluate how easily these heads can be interpreted using new metrics which measure property consistency within heads and property disentanglement across heads. Our findings reveal that larger CLIP models are generally more interpretable than their smaller counterparts. To further assist users in understanding the inner workings of CLIP models, we introduce CLIP-InterpreT, a tool designed for interpretability analysis. CLIP-InterpreT offers five types of analyses: property-based nearest neighbor search, per-head topic segmentation, contrastive segmentation, per-head nearest neighbors of an image, and per-head nearest neighbors of text.