Skip to content

Convert and compress Keras models into TF 8-bit quantized networks

Notifications You must be signed in to change notification settings

skoppula/network-compression-scripts

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

You'll find relevant scripts to take a Keras HDF5 model, extract the weights, convert it to a Tensorflow protobuf weight format (`get_weights.py`), and quantize the protobuf representation to an 8-bit weights network (`quantize.sh`). You'll need to have compiled Tensorflow's 8-bit operations tools from [source](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/tools/quantization), and this requires Bazel. Feel free to reach out with questions.

About

Convert and compress Keras models into TF 8-bit quantized networks

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published