I was trying to understand structured and unstructured pruning techniques used for deep learning models: link 1 and link 2. To recap what I have understood that unstructured pruning is based on weight pruning however structured pruning is basically done through filter pruning, channel pruning, and filter shape pruning. So, my question is the unstructured pruning is done only for CNN models and cannot be done for example for MLP models as there are no filters and channels?
$\begingroup$
$\endgroup$
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
1
You can do structured pruning for MLPs. You would remove things like whole neurons or layers (not specific weights).
-
$\begingroup$ One line and link only answers tend to get flagged for deletion. $\endgroup$– EthanFeb 24, 2021 at 16:52