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This repository has been archived by the owner on Feb 12, 2022. It is now read-only.
In this code, ForgetMult computes a simple recurrent equation:
h_t = f_t * x_t + (1 - f_t) * h_{t-1}
but in paper, it is
h_t = f_t * h_{t-1}+ (1 - f_t) * x_t
Which one is correct?
The text was updated successfully, but these errors were encountered:
It's unfortunate that it's not consistent but it doesn't really matter. h_t is a linear combination of x_t and h_{t-1} weighted by f_t and 1 - f_t. f_t is [0,1] so 1 - f_t is just the remainder to 1. You can transform from one equation to the other by setting f_t = 1 - f_t. The model will learn the same thing...
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
In this code, ForgetMult computes a simple recurrent equation:
h_t = f_t * x_t + (1 - f_t) * h_{t-1}
but in paper, it is
h_t = f_t * h_{t-1}+ (1 - f_t) * x_t
Which one is correct?
The text was updated successfully, but these errors were encountered: