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This repository has been archived by the owner on May 1, 2023. It is now read-only.
You're right, there is an error in the scale factor of the bias. It should be the reciprocal of the what's in there now. That is, it should be: (q_x * q_w) / q_b.
And actually, the parentheses are wrong as well - the way it is now it implies the bias is added during the accumulation instead of after it.
Hi,
The math for the derivation of y_q under Symmetric Linear Quantization on the page
https://nervanasystems.github.io/distiller/algo_quantization/index.html seems incorrect.
I am not able to reason out to myself, the scaling of the bias term.
Thanks
The text was updated successfully, but these errors were encountered: