1 paper
Sarthak Jain, Edward Banner, Jan-Willem van de Meent +2
We propose a method for learning disentangled representations of texts that code for distinct and complementary aspects, with the aim of affording efficient model transfer and inte…