README.md

ResNet50 FP32 inference

Description

This document has instructions for running ResNet50 FP32 inference using intel-extension-for-pytorch.

Download link

pytorch-resnet50-fp32-inference.tar.gz

Datasets

The ImageNet validation dataset is used when testing accuracy. The inference scripts use synthetic data, so no dataset is needed.

Download and extract the ImageNet2012 dataset from http://www.image-net.org/, then move validation images to labeled subfolders, using the valprep.sh shell script

The accuracy script looks for a folder named val, so after running the data prep script, your folder structure should look something like this:

imagenet
└── val
    ├── ILSVRC2012_img_val.tar
    ├── n01440764
    │   ├── ILSVRC2012_val_00000293.JPEG
    │   ├── ILSVRC2012_val_00002138.JPEG
    │   ├── ILSVRC2012_val_00003014.JPEG
    │   ├── ILSVRC2012_val_00006697.JPEG
    │   └── ...
    └── ...

The folder that contains the val directory should be set as the DATASET_DIR when running accuracy (for example: export DATASET_DIR=/home/<user>/imagenet).

Quick Start Scripts

Script name	Description
fp32_online_inference.sh	Runs online inference using synthetic data (batch_size=1).
fp32_batch_inference.sh	Runs batch inference using synthetic data (batch_size=128).
fp32_accuracy.sh	Measures the model accuracy (batch_size=128).

These quickstart scripts can be run in different environments:

• Bare Metal
• Docker

Bare Metal

To run on bare metal, the following prerequisites must be installed in your environment:

• Python 3
• intel-extension-for-pytorch
• torchvision==v0.6.1
• numactl

Download and untar the model package and then run a quickstart script.

# Optional: to run accuracy script
export DATASET_DIR=<path to the preprocessed imagenet dataset>

# Download and extract the model package
wget https://storage.googleapis.com/intel-optimized-tensorflow/models/v2_7_0/pytorch-resnet50-fp32-inference.tar.gz
tar -xzf pytorch-resnet50-fp32-inference.tar.gz
cd pytorch-resnet50-fp32-inference
./quickstart/<script name>.sh

Docker

Use the base PyTorch 1.8 container intel/intel-optimized-pytorch:1.8.0 to run ResNet50 FP32 inference. To run the model quickstart scripts using the base PyTorch 1.8 container, you will need to provide a volume mount for the pytorch-resnet50-fp32-inference package.

To run the accuracy test, you will need mount a volume and set the DATASET_DIR environment variable to point to the ImageNet validation dataset. The accuracy script also downloads the pretrained model at runtime, so provide proxy environment variables, if necessary.

DATASET_DIR=<path to the dataset folder>

docker run \
  --env DATASET_DIR=${DATASET_DIR} \
  --env http_proxy=${http_proxy} \
  --env https_proxy=${https_proxy} \
  --volume ${DATASET_DIR}:${DATASET_DIR} \
  --volume <path to the model package directory>:/pytorch-resnet50-fp32-inference \
  --privileged --init -it \
  intel/intel-optimized-pytorch:1.8.0 /bin/bash

Synthetic data is used when running batch or online inference, so no dataset mount is needed.

docker run \
  --privileged --init -it \
  --volume <path to the model package directory>:/pytorch-resnet50-fp32-inference \
  intel/intel-optimized-pytorch:1.8.0 /bin/bash

Run quickstart scripts:

cd /pytorch-resnet50-fp32-inference
bash quickstart/<script name>.sh

If you are new to docker and are running into issues with the container, see this document for troubleshooting tips.

License

LICENSE