I am trying to understand LSTM and reading colah blog , As LSTM structure looks like this :

enter image description here

So LSTM takes three inputs:

  • Input vector
  • Memory from previous block
  • Output from previous block

and return :

  • memory from current block
  • output from current block

which will become previous memory and previous block output for next state ,

My confusion is what is the initial values of ct-1 ( memory from previous block) and ht-1 ( hidden output from previous block) , since there is no previous block then what are initial values ?


Either you use a zero vector state for both (most commonly used), or you can use other approaches such as described in this article.

I have personally always used zero vectors.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.