apply_compression_results with Tensorflow

[R2bEEaton]

Hello!

It seems support for TensorFlow is limited, but I was able to get a pruner to work in the latest version after grabbing some code from here. However, when I load the model in pruned.pth, it appears to be the same as my original model. When I feed test data into the original and pruned models, the result is identical.

I see something in old pytorch examples called apply_compression_results but it does not appear to exist for Tensorflow.

Does anyone have any guidance here for model pruning with NNI for a pre-trained Tensorflow model? Thanks.