Hi! Is the code for the word-image similarity used in Fig.3/Table 4 available? I’m trying to reproduce this part of the paper. Thanks!