FC Pruning
Project description
FeatureCloud Pruning
Model Compression with Pruning
The FC Pruning package offers a streamlined approach to model compression using advanced pruning techniques. With support for federated learning frameworks and integration with the Torch-Pruning library, this package enables efficient distributed training, suitable for a wide range of machine learning tasks.
-
Train Local Model
-
Configure Quantization:
- Set up initial quantization settings.
self.configure_pruning(pruning_ratio, model, reference_model, imp,ex_input, ignored_layers)
- Set up initial quantization settings.
-
Send Data to Coordinator and perform Pruning:
- Send the prepared model data to the coordinator.
self.send_data_to_coordinator(model, use_pruning=True, use_smpc=False, use_dp=False)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
fc_pruning-0.1.0.tar.gz
(8.6 kB
view hashes)