Skip to content

Latest commit

 

History

History
145 lines (127 loc) · 5.74 KB

README.md

File metadata and controls

145 lines (127 loc) · 5.74 KB
Raw

Wide & Deep FP32 inference

This document has instructions for running Wide & Deep FP32 inference using Intel-optimized TensorFlow.

Dataset

Download and preprocess the income census data by running following python script, which is a standalone version of census_dataset.py Please note that below program requires requests module to be installed. You can install it using pip install requests. Dataset will be downloaded in directory provided using --data_dir. If you are behind corporate proxy, then you can provide proxy URLs using --http_proxy and --https_proxy arguments.

git clone https://github.com/IntelAI/models.git
cd models
python ./benchmarks/recommendation/tensorflow/wide_deep/inference/fp32/data_download.py --data_dir /home/<user>/widedeep_dataset

Quick Start Scripts

Script name Description
fp32_inference_online.sh Runs wide & deep model inference online mode (batch size = 1)
fp32_inference_batch.sh Runs wide & deep model inference in batch mode (batch size = 1024)

Run the model

Setup your environment using the instructions below, depending on if you are using AI Kit:

Setup using AI Kit on Linux Setup without AI Kit on Linux Setup without AI Kit on Windows

To run using AI Kit on Linux you will need:

  • git
  • numactl
  • wget
  • Activate the `tensorflow` conda environment 
    conda activate tensorflow

To run without AI Kit on Linux you will need:

  • Python 3
  • intel-tensorflow>=2.5.0
  • git
  • numactl
  • wget
  • A clone of the Model Zoo repo
    git clone https://github.com/IntelAI/models.git

To run without AI Kit on Windows you will need:

After you've completed the setup above, download and extract the pretrained model. If you run on Windows, please use a browser to download the pretrained model using the link below. Set the directory path to the PRETRAINED_MODEL environment variable.

wget https://storage.googleapis.com/intel-optimized-tensorflow/models/v1_8/wide_deep_fp32_pretrained_model.tar.gz
tar -xzvf wide_deep_fp32_pretrained_model.tar.gz
export PRETRAINED_MODEL=wide_deep_fp32_pretrained_model

Wide & Deep inference also uses code from the TensorFlow models repo. Clone the repo using the PR in the snippet below and save the directory path to the TF_MODELS_DIR environment variable.

# We going to use a branch based on older version of the tensorflow model repo.
# Since, we need to to use logs utils on that branch, which were removed from
# the latest master
git clone https://github.com/tensorflow/models.git tensorflow-models
cd tensorflow-models
git fetch origin pull/7461/head:wide-deep-tf2
git checkout wide-deep-tf2

Once your environment is setup, navigate back to your model zoo directory and set environment variables pointing to your dataset and an output directory for log files. Ensure that you also have the pretrained model and TensorFlow models repo paths set from the previous steps. Select a quickstart script to run.

Run on Linux

# cd to your model zoo directory
cd models

export DATASET_DIR=<path to the Wide & Deep dataset directory>
export OUTPUT_DIR=<directory where log files will be written>
export PRETRAINED_MODEL=<pretrained model directory>
export TF_MODELS_DIR=<path to tensorflow-models directory>

./quickstart/recommendation/tensorflow/wide_deep/inference/cpu/fp32/<script name>.sh

Run on Windows

Using cmd.exe, run:

# cd to your model zoo directory
cd models

set PRETRAINED_MODEL=<pretrained model directory>
set DATASET_DIR=<path to the Wide & Deep dataset directory>
set OUTPUT_DIR=<directory where log files will be written>
set TF_MODELS_DIR=<path to tensorflow-models directory>

bash quickstart\recommendation\tensorflow\wide_deep\inference\cpu\fp32\<script name>.sh

Note: You may use cygpath to convert the Windows paths to Unix paths before setting the environment variables. As an example, if the dataset location on Windows is D:\<user>\widedeep_dataset, convert the Windows path to Unix as shown:

cygpath D:\<user>\widedeep_dataset
/d/<user>/widedeep_dataset

Then, set the DATASET_DIR environment variable set DATASET_DIR=/d/<user>/widedeep_dataset.

Additional Resources