ichernob
  • Member for 4 years, 10 months
  • Last seen more than 2 years ago
Number of parameters in an LSTM model
15 votes

According to this: LSTM cell structure LSTM equations Ingoring non-linearities If the input x_t is of size n×1, and there are d memory cells, then the size of each of W∗ and U∗ is d×n, and d×d ...

View answer
Gradient Check is failing for RNN
1 votes

as i understand, you have the wrong backprop gradient implementation. Here you should take into account, that rnn's hidden state h has its previous state in the equation: h[time-1]. This is also must ...

View answer
My LSTM doesn't pass gradient check
Accepted answer
1 votes

Hope somebody will safe lots of hours: def backward_propagation(self, x, y, cache): # T - the length of the sequence T = len(y) # perform forward propagation cache = self....

View answer