Does Filter Pruning Compression Algorithm in Neural Network Compression Framework (NNCF) Reduce Model Size?

• Read a paper and NNCF GitHub* repository about filter pruning.
• Unable to find information about filter pruning size reduction on a model

• NNCF filter pruning does not immediately reduce the model file size. During pruning, unimportant filters in convolutional or linear layers are first zeroed out.
• Model size reduction and computational savings are achieved when the pruned structures are physically removed during model export or transformation. After structural pruning is applied, the resulting model may contain fewer parameters and a smaller footprint.

• Refer to the current Neural Network Compression Framework (NNCF) documentation for the pruning workflow supported by your OpenVINO™ release.

Refer to the Neural Network Compression Framework (NNCF) documentation for the latest model optimization and pruning workflows.