When pruning a model with TLT, where is the pruned model written? Is it written over top of the old model? Is a new model file created? I have pruned a model and I see no new files nor have any of the file sizes changed.
The process does require providing a location for an output file where the documentation says this will hold the “output checkpoints”. Is that the model?
The pruned model is a new model.
For example, in below jupyter notebook example,
!tlt detectnet_v2 prune
-m $USER_EXPERIMENT_DIR/experiment_dir_unpruned/weights/resnet18_detector.tlt
-o $USER_EXPERIMENT_DIR/experiment_dir_pruned/resnet18_nopool_bn_detectnet_v2_pruned.tlt
-eq union
-pth 0.0000052
-k $KEY
The resnet18_nopool_bn_detectnet_v2_pruned.tlt is the pruned model.