SKKU-ESLAB / Auto-Compression

Automatic DNN compression tool with various model compression and neural architecture search techniques
MIT License
20 stars 18 forks source link

Add model converter for pruned weight #9

Closed GH-Jo closed 2 years ago

GH-Jo commented 3 years ago

After compression using DNAS, all the weights are stored in dense format and mask. This is not suitable for efficient inference.

We need a model weight converter to save pruned weight in compressed format.