din

DIN CTR SAMPLE

The purpose of this sample is to demonstrate how to build and train a Deep Interest Network with HugeCTR.

Set Up the HugeCTR Docker Environment

You can set up the HugeCTR Docker environment by doing one of the following:

• Pull the NGC Docker
• Build the HugeCTR Docker Container on Your Own

Pull the NGC Docker

HugeCTR is available as buildable source code, but the easiest way to install and run HugeCTR is to pull the pre-built Docker image, which is available on the NVIDIA GPU Cloud (NGC). This method provides a self-contained, isolated, and reproducible environment for repetitive experiments.

1. Pull the HugeCTR NGC Docker by running the following command:

   $ docker pull nvcr.io/nvidia/merlin/merlin-training:22.03

2. Launch the container in interactive mode with the HugeCTR root directory mounted into the container by running the following command:

   $ docker run --gpus=all --rm -it --cap-add SYS_NICE -u $(id -u):$(id -g) -v $(pwd):/hugectr -w /hugectr nvcr.io/nvidia/merlin/merlin-training:22.03

Build the HugeCTR Docker Container on Your Own

If you want to build the HugeCTR Docker container on your own, refer to How to Start Your Development.

You should make sure that HugeCTR is built and installed in /usr/local/hugectr within the Docker container. Remember to set the option ENABLE_MULTINODES as ON when building HugeCTR if you want to try the multinode training sample. You can launch the container in interactive mode in the same manner as shown above, and then set the PYTHONPATH environment variable inside the Docker container using the following command:

$ export PYTHONPATH=/usr/local/hugectr/lib:$PYTHONPATH

Download the Dataset

Make a folder under the project root directory to store the raw data:

cd ${project_root}
mkdir raw_data

Go here and download the review_Electronics_5 and meta_Electronics dataset files into the "${project_root}/raw_data" directory.

As an alternative, you can run the following command:

cd raw_data
wget -c http://snap.stanford.edu/data/amazon/productGraph/categoryFiles/reviews_Electronics_5.json.gz
gzip -d reviews_Electronics_5.json.gz
wget -c http://snap.stanford.edu/data/amazon/productGraph/categoryFiles/meta_Electronics.json.gz
gzip -d meta_Electronics.json.gz

Preprocess the Dataset

Make a folder under the project root directory to store the processed data:

mkdir din_data

Run python scripts to preprocess the data:

cd utils
bash preprocess.sh

Train with HugeCTR

With the above steps, the preprocessed data is saved locally in the din_data/ directory. To train the DIN model using the data, simply run:

python din_parquet.py

By default, this will run 8000 iterations and the peak of the AUC could reach higher than 0.8.

There is another implementation of the DIN model in din_try.py, which employs the Matrix Multiply layer. It can achieve AUC 0.75 after 88,000 iterations.