Xilinx / graffitist

Graph Transforms to Quantize and Retrain Deep Neural Nets in TensorFlow
https://arxiv.org/abs/1903.08066
Other
168 stars 34 forks source link

depth_to_sparse operation supported? #9

Open linbaiwpi opened 4 years ago

linbaiwpi commented 4 years ago

In my network, there is a depth_to_sparse operation, which contains no learn-able parameters. What I want is to skip this to do a quantization aware training with this layer NOT trained (skipped during back-propagation). Can I do retrain my model using Graffitist?