Move STE/LSQ activation quantizers from QAT to trainable infrastructure for PyTorch

sony / model_optimization

Model Compression Toolkit (MCT) is an open source project for neural network model optimization under efficient, constrained hardware. This project provides researchers, developers, and engineers advanced quantization and compression tools for deploying state-of-the-art neural networks.

Apache License 2.0

322 stars 50 forks source link

Pull Request Description:

Extracted activation quantizers from QAT to trainable infrastructure.
Updated readme files.
GPTQ fix was extracted to a separate PR #1197

Checklist before requesting a review:

[ ] I set the appropriate labels on the pull request.
[ ] I have added/updated the release note draft (if necessary).
[ ] I have updated the documentation to reflect my changes (if necessary).
[ ] All function and files are well documented.
[ ] All function and classes have type hints.
[ ] There is a licenses in all file.
[ ] The function and variable names are informative.
[ ] I have checked for code duplications.
[ ] I have added new unittest (if necessary).

sony / model_optimization

Move STE/LSQ activation quantizers from QAT to trainable infrastructure for PyTorch #1178

Pull Request Description:

Checklist before requesting a review: