1
$\begingroup$

I understand that the formula for the gradient of W and U is this: figure 1

How might I go around programming ds/dst-1 for example?

$\endgroup$
3
  • $\begingroup$ It is not very clear what you mean but if you are looking for the formulas of RNNs,I recommend you take a look at Alex Grave's implementation. $\endgroup$ Commented Aug 19, 2018 at 6:17
  • $\begingroup$ @Media the issue is I’m not sure how you would code ds/dst-1 specifically. Also would you be able to link me the the implementation as I cannot find it. Thank you $\endgroup$ Commented Aug 19, 2018 at 13:15
  • $\begingroup$ here. $\endgroup$ Commented Aug 19, 2018 at 19:08

1 Answer 1

1
$\begingroup$

Most deep learning frameworks use automatic differentiation (AD) so it is handled for you.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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