Interpreting CLIP's Image Representation via Text-Based Decomposition

arXiv:2310.05916v4 Announce Type: replace
Abstract: We investigate the CLIP image encoder by analyzing how individual model components affect the final representation. We decompose the image representation as a sum across individual image patches, model layers, and attention heads, and use CLIP's text representation to interpret the summands. Interpreting the attention heads, we characterize each head's role by automatically finding text representations that span its output space, which reveals property-specific roles for many heads (e.g. location or shape). Next, interpreting the …

abstract arxiv attention clip components cs.ai cs.cv encoder image representation text type via