SKKU-ESLAB / Auto-Compression

Automatic DNN compression tool with various model compression and neural architecture search techniques
MIT License
20 stars 20 forks source link

Show the selection probability of candidate #13

Closed GH-Jo closed 2 years ago

GH-Jo commented 3 years ago

Currently, only the pruning rate selected for each layer is output after compression.

In order to get closer to the real pruned result, the probability values should not be similar among candidates so that the probability of one specific candidate must be large.

To know this, we need to print the probability of candidates in a form that is easy to read.