a b/shepherd/decoders.py 

  1
  
import torch





  
  

  2
  
import torch.nn.functional as F





  
  

  3
  





  
  

  4
  





  
  

  5
  
def bilinear(s, r, t):





  
  

  6
  
    return torch.sum(s * r * t, dim = 1)





  
  

  7
  





  
  

  8
  
def trans(s, r, t):





  
  

  9
  
    return -torch.norm(s + r - t, dim = 1)





  
  

  10
  





  
  

  11
  
def dot(s, t):





  
  

  12
  
    return torch.sum(s * t, dim = 1)