3 ms·

What prevents you from scaling it up and achieving good results on Imagenet?

• aiwabdn 6y ago
  Main reason for now is that the cosine similarity used as context function is not suitable for natural images like those in Imagenet. We are experimenting some extensions based on convolutions for such use cases.
  • p1esk 6y ago
    What would you say is the best real world application for this class of networks?
    • aiwabdn 6y ago
      Possibly some scenarios in RL where online learning is involved