Is there a way to measure to measure similarities between two pieces of text ? Think about the case where you have an image captioning model but you only want to deal/use specific class names. e.g. say you have classes like 'vegetation' , 'car' , etc .. and your captioning model output could be something like 'tree' , 'tree on the side of the road' , 'vehicle' , 'car' ...

  • $\begingroup$ genererate embedding vectors and measure the angle between them (dot product) $\endgroup$ Commented Oct 30, 2023 at 17:15


You must log in to answer this question.

Browse other questions tagged .