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
I understand this was written for autoregressive models, but do you think you could apply it to the decoder portion of an encoder to dcoder module? Like, if the input is "What is the capital of France?" and the decoder outputs "The capital of France is" and you change the appropriate MLP to point to Rome? with the subject being France?
The text was updated successfully, but these errors were encountered:
I understand this was written for autoregressive models, but do you think you could apply it to the decoder portion of an encoder to dcoder module? Like, if the input is "What is the capital of France?" and the decoder outputs "The capital of France is" and you change the appropriate MLP to point to Rome? with the subject being France?
The text was updated successfully, but these errors were encountered: