NVIDIA
NVIDIA
Merlin HugeCTR
Container
NVIDIA
NVIDIA
Merlin HugeCTR

The Merlin HugeCTR container enables you to perform data preprocessing, feature engineering, train models with HugeCTR, and then serve the trained model with Triton Inference Server.

LayerLabelCreated
2e5f958ba2175e38acd2f81790e5451a354e2c3c8499b0f30146493468af7810RUN
|10
06/15/2024 1:04 AM UTC
0fb3a08781493b71330fa7af23b85fa82575e047a6dbf9adf55c780959cbc8b8RUN
HUGECTR_BACKEND_VER=v23.09.00 HUGECTR_DEV_MODE=false HUGECTR_HOME=/usr/local/hugectr HUGECTR_VER=v24.06.00 HWLOC_VER=2.4.1 INSTALL_HDFS=false TRITON_VERSION=24.03 _CI_JOB_TOKEN= _HUGECTR_BACKEND_REPO=github.com/triton-inference-server/hugectr_backend.git _HUGECTR_REPO=github.com/NVIDIA-Merlin/HugeCTR.git rm -rf /usr/local/share/jupyter/lab/staging/node_modules/marked
06/15/2024 1:04 AM UTC
825a22dda229a1fd2839de9d3db82e28affbce50130b9b269206df0ffa452de9ENV
PYTHONPATH=:/usr/local/lib/python3.10/dist-packages/:/usr/local/hugectr/lib
06/15/2024 1:04 AM UTC
aa897136b86c28224e810c84b8c2a8c30bc0c0a16896959d85754f52796b7f35RUN
HUGECTR_BACKEND_VER=v23.09.00 HUGECTR_DEV_MODE=false HUGECTR_HOME=/usr/local/hugectr HUGECTR_VER=v24.06.00 HWLOC_VER=2.4.1 INSTALL_HDFS=false TRITON_VERSION=24.03 _CI_JOB_TOKEN= _HUGECTR_BACKEND_REPO=github.com/triton-inference-server/hugectr_backend.git _HUGECTR_REPO=github.com/NVIDIA-Merlin/HugeCTR.git ln -s ${HUGECTR_HOME}/backends/hps /opt/tritonserver/backends/hps
06/15/2024 1:04 AM UTC
0d73524447fbb57ea37fc66b9bbd5003d884fbd5ad1356c797753d6ae6bfe73cRUN
HUGECTR_BACKEND_VER=v23.09.00 HUGECTR_DEV_MODE=false HUGECTR_HOME=/usr/local/hugectr HUGECTR_VER=v24.06.00 HWLOC_VER=2.4.1 INSTALL_HDFS=false TRITON_VERSION=24.03 _CI_JOB_TOKEN= _HUGECTR_BACKEND_REPO=github.com/triton-inference-server/hugectr_backend.git _HUGECTR_REPO=github.com/NVIDIA-Merlin/HugeCTR.git if [ "${HUGECTR_DEV_MODE}" == "false" ]; then git clone --branch ${HUGECTR_VER} --depth 1 https://${_CI_JOB_TOKEN}${_HUGECTR_REPO} /hugectr &&
  cd /hugectr &&
  git submodule update --init --recursive &&
  mkdir build &&
  cd build &&
  if [[ "${INSTALL_HDFS}" == "false" ]]; then cmake -DCMAKE_BUILD_TYPE=Release -DSM="70;75;80;90" -DENABLE_INFERENCE=ON .. ; else cmake -DCMAKE_BUILD_TYPE=Release -DSM="70;75;80;90" -DENABLE_INFERENCE=ON -DENABLE_HDFS=ON .. ; fi &&
  make -j$(nproc) &&
  make install &&
  rm -rf ./* &&
  git clone --branch ${HUGECTR_BACKEND_VER} --depth 1 https://${_CI_JOB_TOKEN}${_HUGECTR_BACKEND_REPO} /repos/hugectr_triton_backend &&
  mkdir /repos/hugectr_triton_backend/hps_backend/build &&
  cd /repos/hugectr_triton_backend/hps_backend/build &&
  cmake -DCMAKE_INSTALL_PREFIX:PATH=${HUGECTR_HOME} -DTRITON_COMMON_REPO_TAG="r${TRITON_VERSION}" -DTRITON_CORE_REPO_TAG="r${TRITON_VERSION}" -DTRITON_BACKEND_REPO_TAG="r${TRITON_VERSION}" .. &&
  make -j$(nproc) &&
  make install &&
  chmod +x ${HUGECTR_HOME}/lib/*.so ${HUGECTR_HOME}/backends/hps/*.so &&
  cd ../../.. &&
  rm -rf hugectr_triton_backend &&
  rm -rf ${HUGECTR_HOME}/lib/libgmock* ${HUGECTR_HOME}/lib/pkgconfig/gmock* ${HUGECTR_HOME}/include/gmock &&
  rm -rf ${HUGECTR_HOME}/lib/libgtest* ${HUGECTR_HOME}/lib/pkgconfig/gtest* ${HUGECTR_HOME}/include/gtest &&
  cd /hugectr/build &&
  LD_LIBRARY_PATH=/usr/local/cuda/lib64/stubs/:$LD_LIBRARY_PATH &&
  export PATH=$PATH:/usr/local/cuda-$(echo $CUDA_VERSION | awk -F'.' '{print $1"."$2}')/compat &&
  if [[ "${INSTALL_HDFS}" == "false" ]]; then cmake -DCMAKE_BUILD_TYPE=Release -DSM="60;61;70;75;80;90" -DENABLE_MULTINODES=ON .. ; else cmake -DCMAKE_BUILD_TYPE=Release -DSM="60;61;70;75;80;90" -DENABLE_MULTINODES=ON -DENABLE_HDFS=ON .. ; fi &&
  make -j$(nproc) &&
  make install &&
  chmod +x ${HUGECTR_HOME}/bin/* ${HUGECTR_HOME}/lib/*.so &&
  cd ../hps_trt &&
  mkdir build &&
  cd build &&
  cmake -DSM="70;75;80;90" .. &&
  make -j$(nproc) &&
  make install &&
  cd ../../onnx_converter &&
  python setup.py install &&
  pip --no-cache-dir install ninja tf2onnx &&
  cd ../sparse_operation_kit &&
  python setup.py install &&
  cd ../hps_tf &&
  python setup.py install &&
  cd ../hps_torch/ &&
  TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 9.0" python setup.py install &&
  mv /hugectr/ci ~/hugectr-ci &&
  mv /hugectr/sparse_operation_kit/sparse_operation_kit ~/hugectr-sparse_operation_kit &&
  rm -rf /hugectr &&
  mkdir -p /hugectr /hugectr/sparse_operation_kit &&
  mv ~/hugectr-ci /hugectr/ci &&
  mv ~/hugectr-sparse_operation_kit /hugectr/sparse_operation_kit/sparse_operation_kit &&
  chmod +x /hugectr/ci/* /hugectr/sparse_operation_kit/sparse_operation_kit/* ; fi
06/15/2024 1:04 AM UTC
4f7bf8af92a10c150191e5086cbcad8e4ff82c567ce8b87f2380e8f264cff5b6ENV
PATH=/opt/tritonserver/bin:/usr/local/cuda/lib64/:/usr/local/mpi/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/usr/local/cuda/bin::/opt/hadoop/bin:/opt/hadoop/sbin:/usr/local/hugectr/bin CPATH=:/usr/local/hugectr/include LD_LIBRARY_PATH=/usr/local/lib/python3.10/dist-packages/tensorflow:/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64:/usr/local/lib:/repos/dist/lib:/usr/lib/jvm/default-java/lib:/usr/lib/jvm/default-java/lib/server:/opt/tritonserver/lib:/usr/local/hugectr/lib
06/15/2024 12:38 AM UTC
05cecab9c39ecc6642f0308db0e0b1b598c34da0d1af87a631b614ee26a3dfefARG
TRITON_VERSION
06/15/2024 12:38 AM UTC
9bcd2915dc273255c5e66a1c76b735aa790c85abb0b126135a90065de577455bARG
HUGECTR_HOME=/usr/local/hugectr
06/15/2024 12:38 AM UTC
68dbb44173fa09a49c0d05e9cc47c8636d2aebd9a04f9f3961a299f41442e25eARG
_HUGECTR_BACKEND_REPO=github.com/triton-inference-server/hugectr_backend.git
06/15/2024 12:38 AM UTC
3ea0ad0ea021d95abc2f4c4778b226d381430a47a96b8bd5e162bede0b559722ARG
_CI_JOB_TOKEN=
06/15/2024 12:38 AM UTC
...