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
Hmm, not sure if it's possible since the library wraps extra code around the model to add the vector. Maybe by adding the vector to the appropriate layer bias? But it might be architecture-dependant based on normalization... I've been thinking about this as well so open to suggestions here!
Hi, can you suggest how to merge the base model and the trained vector?
The text was updated successfully, but these errors were encountered: