search

IntelLabs / distiller

Neural Network Distiller by Intel AI Lab: a Python package for neural network compression research. https://intellabs.github.io/distiller
Apache License 2.0
4.35k stars 801 forks source link

Quantization Capabilities in PyTorch #516

Open nik13 opened 4 years ago

nik13 commented 4 years ago

Hi Team,

Thanks for the great tool for model compression research.

As it is stated that you guys are planning to add PTQ or capability to export the quantized model to ONNX, would be greatly interested in knowing if that's yet in the pipeline, or it's better to consider PyTorch internal quantization capabilities or using TensorRT for the same.

Thanks!

shazib-summar commented 4 years ago

A formal method or pipeline to export models after PTQ would be awesome