Skip to content

Post-training model quantization using apache's tvm.ai

Notifications You must be signed in to change notification settings

lmEshoo/quantization

Repository files navigation

Model Quantization

For this example, we are using a pretrained mobilenet v2 model.

Quantize

Requirements

  • Installing Docker There’s a graphical installer for Windows and Mac that makes installing Docker easy. Here are instructions for each OS:

    Windows

    Mac OS

    Linux

Clone

git clone --recursive [email protected]:lmEshoo/quantization.git

cd quantization

git submodule update --init --recursive

Pull

docker pull lmestar/tvm:0.1.1

Run

make

Evaluation

Evaluation was done using Google Colab.

  • Once you opened the notebook, upload all ./eval/ into the Colab enviroment and run.

Contributors

  • Lini Mestar
  • Harpreet Wassan