README.md

<- What is Pruning and Quantization?

Installing DRP-AI Extension Package

Follow the instructions for setting up a host machine using docker. If you install the DRP-AI Extension Pack to host machine directly, please refer to the DRP-AI Extension Pack User's manual.

Note1:
This guide provides the sparse model that has already been pruned.
If you only want to validate RZ/V2H operation, please skip the pruning then retraining step.

Note2:
If you have set up your development environment using docker such as DRP-AI TVM¹ installation with Docker, exit the container once and create a docker image for this installation.

1. Preparation

Please prepare the following file from this Renesas Web Site.

File Name	Description
drpai-extension-pack_ver*.tar.gz	DRP-AI Extension Pack

2. Build docker image

This chapter explains how to build the docker image for the DRP-AI Extension Pack.

cd ${TVM_ROOT}/pruning/setup/docker
# Or wget <URL to docker_torch.Dockerfile>

# Copy the the DRP-AI Extension Pack to `docker` directory.
cp <path/to/>drpai-extension-pack_ver*.tar.gz .
ls
# Ensure that the `torch.Dockefile` and `drpai-extension-pack_ver*tar.gz` are included in the current directory.
# torch.Dockerfile
# drpai-extension-pack_ver*tar.gz
# ....

# Build docker image.
# It may take up to an hour to complete.
docker build -t drpai_ext_pt_img -f torch.Dockerfile .

# With output of `drpai_ext_pt_img` word, the setup processing is completed.
docker images | grep drpai_ext_pt_img
# drpai_ext_pt_img                  latest                              XXXXXXXXXXXXX        X seconds ago       12.4GB

---

Pruning and retraining by DRP-AI Extension Package ->

Footnotes

1. DRP-AI TVM is powered by EdgeCortix MERA™ Compiler Framework. ↩