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
We are setting encoder freeze to be False by default, which means the encoder is trainable.
We tried both freezing encoder and trainable encoder at earlier experiments and found that a trainable encoder reaches better loss and relatively better performances.
Hi OpenLRM Team
I don't find encoder freeze in LRM paper, why do you choose to set freeze encoder is false?
The text was updated successfully, but these errors were encountered: