easy5 pts · first answer countsfine-tuning

Learning rate for fine-tuning

Compared with the learning rate used to pretrain a model, the learning rate for full fine-tuning is usually:

A.Larger, so the model forgets its old task quicklyB.Exactly the same, because the optimizer is the sameC.Zero for all layers except the last oneD.Smaller, so useful pretrained weights are not wiped out

Discussion 0

Join free to post in the discussion.

Learning rate for fine-tuning

Discussion 0

Related questions