As I understand, the problem of continual learning is that if you continue training, it will "overwrite" previous knowledge and this CPU will be lost
As I understand, the problem of continual learning is that if you continue training, it will "overwrite" previous knowledge and this CPU will be lost
There is nothing in theory preventing you from training while keeping a subset of weights locked.
How do you ensure that subset of the weights will continue to be accessible through the others?
I don't understand your question.
[flagged]