7 pointsby encrux8 hours ago1 comment
  • ricardobeat4 hours ago
    Love this kind of experiment. Would the model perform better with word tokens?