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
As the weight decay between 0.01 and 0.05 is a quite huge gap, maybe need to double check and make them consistent? Or are they configs for different models?
The text was updated successfully, but these errors were encountered:
Hi, first of all thank you for the nice work.
I was reading the paper and found the weight decay mentioned in the appendix is different from the one mentioned in the main body.
https://arxiv.org/pdf/2307.08621.pdf
As the weight decay between 0.01 and 0.05 is a quite huge gap, maybe need to double check and make them consistent? Or are they configs for different models?
The text was updated successfully, but these errors were encountered: