You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Feb 11, 2019. It is now read-only.
As above, the code will execute as the following sequence, apply_ : sample a new weight with mean (the original weight) and the standard deviation sd_asn: update the standard deviation with the gradients train_step: update the the weight with the gradients cross_entropy: compute the loss after a forward pass
However, the whole procedure which I got from the paper is: sd_asn: update the standard deviation with the gradients train_step: update the the weight with the gradients apply_ : sample a new weight with mean (the original weight) and the standard deviation cross_entropy: compute the loss after a forward pass
Do I misunderstand something?
The text was updated successfully, but these errors were encountered:
MlWoo
changed the title
some questions about the code and paper
the contradiction of weight sampling procedure between the code and paper
Sep 5, 2018
MlWoo
changed the title
the contradiction of weight sampling procedure between the code and paper
Contradiction of weight sampling procedure between the code and paper
Sep 5, 2018
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
As above, the code will execute as the following sequence,
apply_
: sample a new weight with mean (the original weight) and the standard deviationsd_asn
: update the standard deviation with the gradientstrain_step
: update the the weight with the gradientscross_entropy
: compute the loss after a forward passHowever, the whole procedure which I got from the paper is:
sd_asn
: update the standard deviation with the gradientstrain_step
: update the the weight with the gradientsapply_
: sample a new weight with mean (the original weight) and the standard deviationcross_entropy
: compute the loss after a forward passDo I misunderstand something?
The text was updated successfully, but these errors were encountered: