README.md

PyTorch DLRM training

Description

This document has instructions for running DLRM training using Intel-optimized PyTorch for bare metal.

General Setup

Follow link to install Conda and build Pytorch, IPEX, and Jemalloc.

Model Specific Setup

• Install dependencies

  cd <clone of the model zoo>/quickstart/recommendation/pytorch/dlrm
  pip install requirements.txt

• Set ENV to use AMX if you are using SPR

  export DNNL_MAX_CPU_ISA=AVX512_CORE_AMX

Datasets

Criteo Terabyte Dataset

The Criteo Terabyte Dataset is used to run DLRM. To download the dataset, you will need to visit the Criteo website and accept their terms of use: https://labs.criteo.com/2013/12/download-terabyte-click-logs/. Copy the download URL into the command below as the <download url> and replace the <dir/to/save/dlrm_data> to any path where you want to download and save the dataset.

export DATASET_DIR=<dir/to/save/dlrm_data>

mkdir ${DATASET_DIR} && cd ${DATASET_DIR}
curl -O <download url>/day_{$(seq -s , 0 23)}.gz
gunzip day_*.gz

The raw data will be automatically preprocessed and saved as day_*.npz to the DATASET_DIR when DLRM is run for the first time. On subsequent runs, the scripts will automatically use the preprocessed data.

Environment Setup

# set OUTPUT_DIR, PRECISION, WEIGHT, DATASET
export PRECISION=<specify the precision to run>
export DATASET_DIR=<path to the dataset>
export OUTPUT_DIR=<directory where log files will be written>

Quick Start Scripts

Script name	Description
training.sh	Run training for the specified precision (fp32, bf16, bf32).
test_convergency.sh	Run fully convergency test for the specified precision (fp32, bf16, bf32).
distribute_training.sh	Run distribute training on 1 node with 2 sockets for the specified precision (fp32, bf16, bf32).

Run the model

Follow the instructions above to setup your bare metal environment, do the model-specific setup and download and prepropcess the datsaet. Once all the setup is done, the Model Zoo can be used to run a quickstart script. Ensure that you have enviornment variables set to point to the dataset directory, precision, and an output directory. The NUM_BATCH environment variable can be set to specify the number of batches to run.

# Clone the model zoo repo and set the MODEL_DIR
git clone https://github.com/IntelAI/models.git
cd models
export MODEL_DIR=$(pwd)

# Env vars
export PRECISION=<specify the precision to run>
export DATASET_DIR=<path to the dataset>
export OUTPUT_DIR=<directory where log files will be written>

# Navigate to the DLRM training quickstart directory
cd ${MODEL_DIR}/quickstart/recommendation/pytorch/dlrm/training/cpu

# Run the quickstart script to test performance
NUM_BATCH=10000 bash training.sh

# Run quickstart script for testing convergence trend
NUM_BATCH=50000 bash training.sh

# Or, run quickstart script for testing fully convergency
bash test_convergence.sh

# Run quickstart to distribute training dlrm on 2 sockets
# Note, you need to follow [link](/docs/general/pytorch/BareMetalSetup.md) to install Torch-CCL and run this command on the machine which sockets larger than 2
NUM_BATCH=10000 bash distribute_training.sh

License

LICENSE