1da20c693ac9f4cf345ddd795d99aa028e011ee8,model/uisrnn.py,UISRNN,__init__,#UISRNN#,49

Before Change

  def __init__(self, args, input_dim, observation_dim, transition_bias):
    sigma2 = .1 if args.sigma2 is None else args.sigma2
    if torch.cuda.is_available():
      self.rnn_model = NormalRNN(input_dim, args.rnn_hidden_size,
                                 args.rnn_depth, args.rnn_dropout,
                                 observation_dim).cuda()
      self.rnn_init_hidden = nn.Parameter(
          torch.zeros(1, args.rnn_hidden_size).cuda())
      self.sigma2 = nn.Parameter(sigma2 * torch.ones(observation_dim).cuda())
    else:
      self.rnn_model = NormalRNN(input_dim, args.rnn_hidden_size,
                                 args.rnn_depth, args.rnn_dropout,
                                 observation_dim)
      self.rnn_init_hidden = nn.Parameter(torch.zeros(1, args.rnn_hidden_size))
      self.sigma2 = nn.Parameter(sigma2 * torch.ones(observation_dim))
    self.transition_bias = transition_bias

  def save(self, filepath):
    Save the model to a file.

After Change

        torch.zeros(1, args.rnn_hidden_size).to(self.device))
    sigma2 = .1 if args.sigma2 is None else args.sigma2
    self.sigma2 = nn.Parameter(
        sigma2 * torch.ones(observation_dim).to(self.device))
    self.transition_bias = transition_bias

  def save(self, filepath):

Instances