Closed RadekKazbundaAtMama closed 2 years ago
I had just tried to train new model with TF2.6:
sparsification does prune some weights but dump of the model always dumps all weights and the model is running slow.
Dump of new trained model always (regardless of epoch) dumps (13968,) of blocks. Dumping of pretrained model correctly dumps (1526,) of blocks.
Had anyone recently build LPCNet model? Can you please share your current working environment? TensorFlow version, Python version etc...
I had just tried to train new model with TF2.6:
sparsification does prune some weights but dump of the model always dumps all weights and the model is running slow.
Dump of new trained model always (regardless of epoch) dumps (13968,) of blocks. Dumping of pretrained model correctly dumps (1526,) of blocks.
Had anyone recently build LPCNet model? Can you please share your current working environment? TensorFlow version, Python version etc...