public
Repositories
1
 Projects Packages

nemo (25.04.01)

Published 2025-07-02 05:13:36 +00:00 by publish

Installation

docker pull packages.speedcloud.co.in/public/nemo:25.04.01
sha256:95ea978a0730e2954e508a75c0f6d59a95460e87dfa87ff7fdacaebb02cc3db2

Image Layers

ARG RELEASE
ARG LAUNCHPAD_BUILD_ARCH
LABEL org.opencontainers.image.ref.name=ubuntu
LABEL org.opencontainers.image.version=24.04
ADD file:6df775300d76441aa33f31b22c1afce8dfe35c8ffbc14ef27c27009235b12a95 in /
CMD ["/bin/bash"]
ARG JETPACK_HOST_MOUNTS=
ENV NVIDIA_REQUIRE_JETPACK_HOST_MOUNTS=
RUN |1 JETPACK_HOST_MOUNTS= /bin/sh -c if [ -n "${JETPACK_HOST_MOUNTS}" ]; then echo "/usr/lib/aarch64-linux-gnu/tegra" > /etc/ld.so.conf.d/nvidia-tegra.conf && echo "/usr/lib/aarch64-linux-gnu/tegra-egl" >> /etc/ld.so.conf.d/nvidia-tegra.conf; fi # buildkit
RUN |1 JETPACK_HOST_MOUNTS= /bin/sh -c export DEBIAN_FRONTEND=noninteractive && apt-get update && apt-get install -y --no-install-recommends apt-utils build-essential ca-certificates curl libncurses6 libncursesw6 patch wget unzip jq gnupg libtcmalloc-minimal4 && rm -rf /var/lib/apt/lists/* && echo "hsts=0" > /root/.wgetrc # buildkit
ARG GDRCOPY_VERSION=2.4.1
ARG HPCX_VERSION=2.22.1
ARG RDMACORE_VERSION=50.0
ARG MOFED_VERSION=5.4-rdmacore50.0
ARG OPENUCX_VERSION=1.18.0
ARG OPENMPI_VERSION=4.1.7
ARG EFA_VERSION=1.34.0
ARG AWS_OFI_NCCL_VERSION=1.12.1
ENV GDRCOPY_VERSION=2.4.1 HPCX_VERSION=2.22.1 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.18.0 OPENMPI_VERSION=4.1.7 RDMACORE_VERSION=50.0 EFA_VERSION=1.34.0 AWS_OFI_NCCL_VERSION=1.12.1
ARG TARGETARCH=amd64
RUN |10 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.1 HPCX_VERSION=2.22.1 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.18.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.34.0 AWS_OFI_NCCL_VERSION=1.12.1 TARGETARCH=amd64 /bin/sh -c cd /nvidia && ( export DEBIAN_FRONTEND=noninteractive && apt-get update && apt-get install -y --no-install-recommends libibverbs1 libibverbs-dev librdmacm1 librdmacm-dev libibumad3 libibumad-dev ibverbs-utils ibverbs-providers && rm -rf /var/lib/apt/lists/* && rm $(dpkg-query -L libibverbs-dev librdmacm-dev libibumad-dev | grep "\(\.so\|\.a\)$") ) && ( cd opt/gdrcopy/ && dpkg -i libgdrapi_*.deb ) && ( cp -r opt/hpcx /opt/ && cp etc/ld.so.conf.d/hpcx.conf /etc/ld.so.conf.d/ && ln -sf /opt/hpcx/ompi /usr/local/mpi && ln -sf /opt/hpcx/ucx /usr/local/ucx && sed -i 's/^\(hwloc_base_binding_policy\) = core$/\1 = none/' /opt/hpcx/ompi/etc/openmpi-mca-params.conf && sed -i 's/^\(btl = self\)$/#\1/' /opt/hpcx/ompi/etc/openmpi-mca-params.conf ) && ( if [ ! -f /etc/ld.so.conf.d/nvidia-tegra.conf ]; then cd opt/amazon/efa/ && dpkg -i libfabric*.deb && rm /opt/amazon/efa/lib/libfabric.a && echo "/opt/amazon/efa/lib" > /etc/ld.so.conf.d/efa.conf; fi ) && ldconfig # buildkit
ENV OPAL_PREFIX=/opt/hpcx/ompi PATH=/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin
ENV OMPI_MCA_coll_hcoll_enable=0
ARG CUDA_VERSION=12.8.1.012
ARG CUDA_DRIVER_VERSION=570.124.06
ENV CUDA_VERSION=12.8.1.012 CUDA_DRIVER_VERSION=570.124.06 CUDA_CACHE_DISABLE=1
RUN |12 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.1 HPCX_VERSION=2.22.1 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.18.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.34.0 AWS_OFI_NCCL_VERSION=1.12.1 TARGETARCH=amd64 CUDA_VERSION=12.8.1.012 CUDA_DRIVER_VERSION=570.124.06 /bin/sh -c /nvidia/build-scripts/installCUDA.sh # buildkit
RUN |12 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.1 HPCX_VERSION=2.22.1 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.18.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.34.0 AWS_OFI_NCCL_VERSION=1.12.1 TARGETARCH=amd64 CUDA_VERSION=12.8.1.012 CUDA_DRIVER_VERSION=570.124.06 /bin/sh -c cp -vprd /nvidia/. / && patch -p0 < /etc/startup_scripts.patch && rm -f /etc/startup_scripts.patch # buildkit
ENV _CUDA_COMPAT_PATH=/usr/local/cuda/compat ENV=/etc/shinit_v2 BASH_ENV=/etc/bash.bashrc SHELL=/bin/bash NVIDIA_REQUIRE_CUDA=cuda>=9.0
LABEL com.nvidia.volumes.needed=nvidia_driver com.nvidia.cuda.version=9.0
ARG NCCL_VERSION=2.25.1
ARG CUBLAS_VERSION=12.8.4.1
ARG CUFFT_VERSION=11.3.3.83
ARG CURAND_VERSION=10.3.9.90
ARG CUSPARSE_VERSION=12.5.8.93
ARG CUSOLVER_VERSION=11.7.3.90
ARG NPP_VERSION=12.3.3.100
ARG NVJPEG_VERSION=12.3.5.92
ARG CUFILE_VERSION=1.13.1.3
ARG NVJITLINK_VERSION=12.8.93
ARG CUBLASMP_VERSION=0.4.0.789
ARG CAL_VERSION=0.4.4.50
ARG NVSHMEM_VERSION=3.2.5
ARG CUDNN_VERSION=9.8.0.87
ARG CUDNN_FRONTEND_VERSION=1.10.0
ARG TRT_VERSION=10.9.0.34
ARG TRTOSS_VERSION=
ARG NSIGHT_SYSTEMS_VERSION=2025.1.1.110
ARG NSIGHT_COMPUTE_VERSION=2025.1.1.2
ARG CUSPARSELT_VERSION=0.7.1.0
ENV NCCL_VERSION=2.25.1 CUBLAS_VERSION=12.8.4.1 CUFFT_VERSION=11.3.3.83 CURAND_VERSION=10.3.9.90 CUSPARSE_VERSION=12.5.8.93 CUSPARSELT_VERSION=0.7.1.0 CUSOLVER_VERSION=11.7.3.90 NPP_VERSION=12.3.3.100 NVJPEG_VERSION=12.3.5.92 CUFILE_VERSION=1.13.1.3 NVJITLINK_VERSION=12.8.93 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.8.0.87 CUDNN_FRONTEND_VERSION=1.10.0 TRT_VERSION=10.9.0.34 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.1.1.110 NSIGHT_COMPUTE_VERSION=2025.1.1.2
RUN |32 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.1 HPCX_VERSION=2.22.1 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.18.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.34.0 AWS_OFI_NCCL_VERSION=1.12.1 TARGETARCH=amd64 CUDA_VERSION=12.8.1.012 CUDA_DRIVER_VERSION=570.124.06 NCCL_VERSION=2.25.1 CUBLAS_VERSION=12.8.4.1 CUFFT_VERSION=11.3.3.83 CURAND_VERSION=10.3.9.90 CUSPARSE_VERSION=12.5.8.93 CUSOLVER_VERSION=11.7.3.90 NPP_VERSION=12.3.3.100 NVJPEG_VERSION=12.3.5.92 CUFILE_VERSION=1.13.1.3 NVJITLINK_VERSION=12.8.93 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.8.0.87 CUDNN_FRONTEND_VERSION=1.10.0 TRT_VERSION=10.9.0.34 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.1.1.110 NSIGHT_COMPUTE_VERSION=2025.1.1.2 CUSPARSELT_VERSION=0.7.1.0 /bin/sh -c /nvidia/build-scripts/installLIBS.sh && /nvidia/build-scripts/installCUDNN.sh && /nvidia/build-scripts/installTRT.sh && /nvidia/build-scripts/installNSYS.sh && /nvidia/build-scripts/installNCU.sh && /nvidia/build-scripts/installCUSPARSELT.sh && if [ -z "${JETPACK_HOST_MOUNTS}" ]; then /nvidia/build-scripts/installNCCL.sh; fi; # buildkit
LABEL com.nvidia.nccl.version=2.25.1 com.nvidia.cublas.version=12.8.4.1 com.nvidia.cufft.version=11.3.3.83 com.nvidia.curand.version=10.3.9.90 com.nvidia.cusparse.version=12.5.8.93 com.nvidia.cusparselt.version=0.7.1.0 com.nvidia.cusolver.version=11.7.3.90 com.nvidia.npp.version=12.3.3.100 com.nvidia.nvjpeg.version=12.3.5.92 com.nvidia.cublasmp.version=0.4.0.789 com.nvidia.cal.version=0.4.4.50 com.nvidia.cudnn.version=9.8.0.87 com.nvidia.tensorrt.version=10.9.0.34 com.nvidia.tensorrtoss.version= com.nvidia.nsightsystems.version=2025.1.1.110 com.nvidia.nsightcompute.version=2025.1.1.2
ARG DALI_VERSION=1.47.0
ARG DALI_BUILD=
ARG DALI_URL_SUFFIX=120
ARG POLYGRAPHY_VERSION=0.49.18
ARG TRANSFORMER_ENGINE_VERSION=2.1
ARG MODEL_OPT_VERSION=0.25.0
ENV DALI_VERSION=1.47.0 DALI_BUILD= DALI_URL_SUFFIX=120 POLYGRAPHY_VERSION=0.49.18 TRANSFORMER_ENGINE_VERSION=2.1 MODEL_OPT_VERSION=0.25.0
ADD docs.tgz / # buildkit
RUN |38 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.1 HPCX_VERSION=2.22.1 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.18.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.34.0 AWS_OFI_NCCL_VERSION=1.12.1 TARGETARCH=amd64 CUDA_VERSION=12.8.1.012 CUDA_DRIVER_VERSION=570.124.06 NCCL_VERSION=2.25.1 CUBLAS_VERSION=12.8.4.1 CUFFT_VERSION=11.3.3.83 CURAND_VERSION=10.3.9.90 CUSPARSE_VERSION=12.5.8.93 CUSOLVER_VERSION=11.7.3.90 NPP_VERSION=12.3.3.100 NVJPEG_VERSION=12.3.5.92 CUFILE_VERSION=1.13.1.3 NVJITLINK_VERSION=12.8.93 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.8.0.87 CUDNN_FRONTEND_VERSION=1.10.0 TRT_VERSION=10.9.0.34 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.1.1.110 NSIGHT_COMPUTE_VERSION=2025.1.1.2 CUSPARSELT_VERSION=0.7.1.0 DALI_VERSION=1.47.0 DALI_BUILD= DALI_URL_SUFFIX=120 POLYGRAPHY_VERSION=0.49.18 TRANSFORMER_ENGINE_VERSION=2.1 MODEL_OPT_VERSION=0.25.0 /bin/sh -c echo "/usr/local/nvidia/lib" >> /etc/ld.so.conf.d/nvidia.conf && echo "/usr/local/nvidia/lib64" >> /etc/ld.so.conf.d/nvidia.conf # buildkit
ARG _LIBPATH_SUFFIX=
ENV PATH=/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin LD_LIBRARY_PATH=/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64 NVIDIA_VISIBLE_DEVICES=all NVIDIA_DRIVER_CAPABILITIES=compute,utility,video
COPY entrypoint/ /opt/nvidia/ # buildkit
ENV NVIDIA_PRODUCT_NAME=CUDA
ENTRYPOINT ["/opt/nvidia/nvidia_entrypoint.sh"]
RUN |39 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.1 HPCX_VERSION=2.22.1 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.18.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.34.0 AWS_OFI_NCCL_VERSION=1.12.1 TARGETARCH=amd64 CUDA_VERSION=12.8.1.012 CUDA_DRIVER_VERSION=570.124.06 NCCL_VERSION=2.25.1 CUBLAS_VERSION=12.8.4.1 CUFFT_VERSION=11.3.3.83 CURAND_VERSION=10.3.9.90 CUSPARSE_VERSION=12.5.8.93 CUSOLVER_VERSION=11.7.3.90 NPP_VERSION=12.3.3.100 NVJPEG_VERSION=12.3.5.92 CUFILE_VERSION=1.13.1.3 NVJITLINK_VERSION=12.8.93 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.8.0.87 CUDNN_FRONTEND_VERSION=1.10.0 TRT_VERSION=10.9.0.34 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.1.1.110 NSIGHT_COMPUTE_VERSION=2025.1.1.2 CUSPARSELT_VERSION=0.7.1.0 DALI_VERSION=1.47.0 DALI_BUILD= DALI_URL_SUFFIX=120 POLYGRAPHY_VERSION=0.49.18 TRANSFORMER_ENGINE_VERSION=2.1 MODEL_OPT_VERSION=0.25.0 _LIBPATH_SUFFIX= /bin/sh -c mkdir -p /workspace && cp -f -p /opt/nvidia/entrypoint.d/30-container-license.txt /workspace/license.txt # buildkit
RUN /bin/sh -c export DEBIAN_FRONTEND=noninteractive && apt-get update && apt-get install -y --no-install-recommends build-essential git libglib2.0-0 less libhwloc15 libnl-route-3-200 libnl-3-dev libnl-route-3-dev libnuma-dev libnuma1 libpmi2-0-dev nano numactl openssh-client vim wget && rm -rf /var/lib/apt/lists/* # buildkit
COPY cuda-*.patch /tmp # buildkit
RUN /bin/sh -c export DEVEL=1 BASE=0 && /nvidia/build-scripts/installNCU.sh && /nvidia/build-scripts/installCUDA.sh && /nvidia/build-scripts/installLIBS.sh && if [ ! -f /etc/ld.so.conf.d/nvidia-tegra.conf ]; then /nvidia/build-scripts/installNCCL.sh; fi && /nvidia/build-scripts/installCUDNN.sh && /nvidia/build-scripts/installTRT.sh && /nvidia/build-scripts/installNSYS.sh && /nvidia/build-scripts/installCUSPARSELT.sh && if [ -f "/tmp/cuda-${_CUDA_VERSION_MAJMIN}.patch" ]; then patch -p0 < /tmp/cuda-${_CUDA_VERSION_MAJMIN}.patch; fi && rm -f /tmp/cuda-*.patch # buildkit
ENV LIBRARY_PATH=/usr/local/cuda/lib64/stubs:
COPY /opt/amazon/aws-ofi-nccl /opt/amazon/aws-ofi-nccl # buildkit
RUN /bin/sh -c if [ ! -f /etc/ld.so.conf.d/nvidia-tegra.conf ]; then echo "/opt/amazon/aws-ofi-nccl/lib" > /etc/ld.so.conf.d/aws-ofi-nccl.conf && ldconfig; fi # buildkit
ENV NVIDIA_PRODUCT_NAME=PyTorch
ARG NVIDIA_PYTORCH_VERSION=25.03
ARG PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84
ARG NVFUSER_BUILD_VERSION=c5e1555
ENV PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 PYTORCH_VERSION=2.7.0a0+7c8ec84 PYTORCH_BUILD_NUMBER=0 NVIDIA_PYTORCH_VERSION=25.03
ENV NVFUSER_BUILD_VERSION=c5e1555 NVFUSER_VERSION=c5e1555
LABEL com.nvidia.pytorch.version=2.7.0a0+7c8ec84
ARG TARGETARCH=amd64
ARG PYVER=3.12
ARG PYVER_MAJMIN=312
ENV PIP_BREAK_SYSTEM_PACKAGES=1
ARG L4T=0
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c export PYSFX=`echo "${PYVER}" | cut -c1-1` && export DEBIAN_FRONTEND=noninteractive && apt-get update && apt-get install -y --no-install-recommends python$PYVER-dev python$PYSFX python$PYSFX-dev python$PYSFX-venv python-is-python$PYSFX autoconf automake libatlas-base-dev libgoogle-glog-dev libbz2-dev libc-ares2 libre2-dev libleveldb-dev liblmdb-dev libprotobuf-dev libsnappy-dev libtool nasm protobuf-compiler pkg-config unzip sox libsndfile1 libpng-dev libhdf5-dev gfortran rapidjson-dev ninja-build libedit-dev build-essential patchelf && rm -rf /var/lib/apt/lists/* # buildkit
ENV PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c DEBIAN_FRONTEND=noninteractive apt remove -y --force-yes python3-pip && curl -O https://bootstrap.pypa.io/get-pip.py && python get-pip.py && rm get-pip.py # buildkit
ENV PIP_CONSTRAINT=/etc/pip/constraint.txt
COPY constraint.txt /etc/pip/constraint.txt # buildkit
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c pip install pip setuptools && pip install cmake # buildkit
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c curl "https://gitlab-master.nvidia.com/api/v4/projects/105799/packages/generic/OpenBLAS/0.3.24-$(uname -m)/OpenBLAS-0.3.24-$(uname -m).tar.gz" --output OpenBLAS.tar.gz && tar -xf OpenBLAS.tar.gz -C /usr/local/ && rm OpenBLAS.tar.gz # buildkit
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c if [ $TARGETARCH = "arm64" ]; then cd /opt && curl "https://gitlab-master.nvidia.com/api/v4/projects/105799/packages/generic/nvpl_slim_24.04/sbsa/nvpl_slim_24.04.tar" --output nvpl_slim_24.04.tar && tar -xf nvpl_slim_24.04.tar && cp -r nvpl_slim_24.04/lib/* /usr/local/lib && cp -r nvpl_slim_24.04/include/* /usr/local/include && rm -rf nvpl_slim_24.04.tar nvpl_slim_24.04 ; fi # buildkit
ENV NVPL_LAPACK_MATH_MODE=PEDANTIC
WORKDIR /opt/pytorch
COPY . . # buildkit
ENV PYTHONIOENCODING=utf-8
ENV LC_ALL=C.UTF-8
ENV PIP_DEFAULT_TIMEOUT=100
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c pip install numpy scipy PyYAML astunparse typing_extensions cffi spacy mock tqdm librosa expecttest hypothesis xdoctest pytest pytest-xdist pytest-rerunfailures pytest-shard pytest-flakefinder pybind11 Cython regex protobuf six && if [[ $TARGETARCH = "amd64" ]] ; then pip install --no-cache-dir mkl mkl-include mkl-devel ; find /usr/local/lib -maxdepth 1 -type f -regex '.*\/lib\(tbb\|mkl\).*\.so\($\|\.[0-9]*\.[0-9]*\)' -exec rm -v {} + ; fi # buildkit
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c git config --global url."https://github".insteadOf git://github && pip install jupyterlab notebook tensorboard jupyterlab_code_formatter python-hostlist # buildkit
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c PATCHED_FILE=$(python -c "from tensorboard.plugins.core import core_plugin as _; print(_.__file__)") && sed -i 's/^\( *"--bind_all",\)$/\1 default=True,/' "$PATCHED_FILE" && test $(grep '^ *"--bind_all", default=True,$' "$PATCHED_FILE" | wc -l) -eq 1 # buildkit
WORKDIR /opt/pytorch
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c pip install --no-cache-dir /builder/*.whl jupytext black isort && mkdir -p /root/.jupyter/lab/user-settings/@jupyterlab/completer-extension/ && jupyter lab clean # buildkit
COPY jupyter_config/jupyter_notebook_config.py /usr/local/etc/jupyter/ # buildkit
COPY jupyter_config/manager.jupyterlab-settings /root/.jupyter/lab/user-settings/@jupyterlab/completer-extension/ # buildkit
COPY jupyter_config/settings.jupyterlab-settings /root/.jupyter/lab/user-settings/@jupyterlab/completer-extension/ # buildkit
ENV JUPYTER_PORT=8888
ENV TENSORBOARD_PORT=6006
EXPOSE map[8888/tcp:{}]
EXPOSE map[6006/tcp:{}]
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c OPENCV_VERSION=4.10.0 && cd / && wget -q -O - https://github.com/opencv/opencv/archive/${OPENCV_VERSION}.tar.gz | tar -xzf - && cd /opencv-${OPENCV_VERSION} && cmake -GNinja -Bbuild -H. -DWITH_CUDA=OFF -DWITH_1394=OFF -DPYTHON3_PACKAGES_PATH="/usr/local/lib/python${PYVER}/dist-packages" -DBUILD_opencv_cudalegacy=OFF -DBUILD_opencv_stitching=OFF -DWITH_IPP=OFF -DWITH_PROTOBUF=OFF && cmake --build build --target install && cd modules/python/package && pip install -v . && rm -rf /opencv-${OPENCV_VERSION} # buildkit
ENV UCC_CL_BASIC_TLS=^sharp
ENV TORCH_CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0+PTX
ENV PYTORCH_HOME=/opt/pytorch/pytorch
ENV CUDA_HOME=/usr/local/cuda
ENV TORCH_ALLOW_TF32_CUBLAS_OVERRIDE=1
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c pip install /opt/transfer/torch*.whl && patchelf --set-rpath '/usr/local/lib' /usr/local/lib/python${PYVER}/dist-packages/torch/lib/libtorch_global_deps.so # buildkit
COPY /usr/local/share/cmake/TorchVision/ /usr/local/share/cmake/TorchVision/ # buildkit
COPY /usr/local/include/torchvision/ /usr/local/include/torchvision/ # buildkit
COPY /usr/local/lib64/libtorchvision.so /usr/local/lib/libtorchvision.so.1.0 # buildkit
COPY /usr/local/lib64/libjpeg* /usr/local/lib/ # buildkit
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c patchelf --set-rpath '$ORIGIN:/usr/local/lib/python3.12/dist-packages/torch/lib:/usr/local/lib/python3.12/dist-packages/torchvision/' /usr/local/lib/libtorchvision.so.1.0 && patchelf --set-soname libtorchvision.so.1 --output /usr/local/lib/libtorchvision.so.1.0 /usr/local/lib/libtorchvision.so.1.0 && ldconfig && pushd /usr/local/lib && ln -s libtorchvision.so.1 /usr/local/lib/libtorchvision.so && popd && patchelf --set-soname libjpeg.so.62 --output /usr/local/lib/libjpeg.so.62 $(readlink -f $(ldd /usr/local/lib/python3.12/dist-packages/torchvision/image.so | grep libjpeg | awk '{print $3}')) # buildkit
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c cd pytorch && pip install -v -r /opt/pytorch/pytorch/requirements.txt # buildkit
RUN |7 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 /bin/sh -c pip install --no-cache-dir /tmp/dist/*.whl # buildkit
ARG DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c if [ -z "${DALI_VERSION}" ] ; then echo "Not Installing DALI for L4T Build." ; exit 0; fi && export CUDA_VERSION_MAJOR=$(ls /usr/local/cuda/lib64/libcudart.so.*.*.* | cut -d . -f 3) && export DALI_PKG_SUFFIX="cuda${CUDA_VERSION_MAJOR}0" && if [ -z "${DALI_URL_SUFFIX}" ] ; then export DALI_EXTRA_INDEX_URL="${DALI_EXTRA_INDEX_URL}-qa"; fi && pip install --extra-index-url https://developer.download.nvidia.com/compute/redist --extra-index-url "${DALI_EXTRA_INDEX_URL}" --trusted-host sqrl nvidia-dali-${DALI_PKG_SUFFIX}==${DALI_VERSION} # buildkit
ENV COCOAPI_VERSION=2.0+nv0.8.1
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c export COCOAPI_TAG=$(echo ${COCOAPI_VERSION} | sed 's/^.*+n//') && pip install git+https://github.com/nvidia/cocoapi.git@${COCOAPI_TAG}#subdirectory=PythonAPI # buildkit
COPY singularity/ /.singularity.d/ # buildkit
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c ( cd fuser && pip install -r requirements.txt && python setup.py -version-tag=a0+${NVFUSER_VERSION} install && python setup.py clean && cp $(find /usr/local/lib/python${PYVER}/dist-packages/ -name libnvfuser_codegen.so) /usr/local/lib/python${PYVER}/dist-packages/torch/lib/ ) && ( cd lightning-thunder && python setup.py install && rm -rf build *.egg-info) && ( cd lightning-thunder && mkdir tmp && cd tmp && git clone -b v${CUDNN_FRONTEND_VERSION} --recursive --single-branch https://github.com/NVIDIA/cudnn-frontend.git cudnn_frontend && cd cudnn_frontend && pip install --no-build-isolation . && cd ../../ && rm -rf tmp ) && ( cd pytorch/third_party/onnx && pip uninstall typing -y && CMAKE_ARGS="-DONNX_USE_PROTOBUF_SHARED_LIBS=ON" pip install --no-build-isolation . ) # buildkit
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c pip install tabulate # buildkit
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c if [ "${L4T}" = "1" ]; then echo "Not installing rapids for L4T build."; exit 0; fi && find /rapids -name "*-Linux.tar.gz" -exec tar -C /usr --exclude="*.a" --exclude="bin/xgboost" --strip-components=1 -xvf {} \; && find /rapids -name "*.whl" ! -name "tornado-*" ! -name "Pillow-*" ! -name "certifi-*" ! -name "protobuf-*" ! -name "six-*" -exec pip install --no-cache-dir {} + # buildkit
WORKDIR /workspace
COPY NVREADME.md README.md # buildkit
COPY docker-examples docker-examples # buildkit
COPY tutorials tutorials # buildkit
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c chmod -R a+w . # buildkit
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c set -x && WHEELS=1 /nvidia/build-scripts/installTRT.sh # buildkit
ENV PATH=/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin:/opt/tensorrt/bin
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c pip --version && python -c 'import sys; print(sys.platform)' && pip install --extra-index-url https://urm.nvidia.com/artifactory/api/pypi/sw-tensorrt-pypi/simple --no-cache-dir "polygraphy==${POLYGRAPHY_VERSION}" && pip install --index-url https://gitlab-master.nvidia.com/api/v4/projects/95421/packages/pypi/simple --extra-index-url https://pypi.nvidia.com "nvidia-modelopt[torch]==${MODEL_OPT_VERSION}" # buildkit
COPY torch_tensorrt/ /opt/pytorch/torch_tensorrt/ # buildkit
ARG PYVER=3.12
ENV LD_LIBRARY_PATH=/usr/local/lib/python3.12/dist-packages/torch/lib:/usr/local/lib/python3.12/dist-packages/torch_tensorrt/lib:/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
ENV PATH=/usr/local/lib/python3.12/dist-packages/torch_tensorrt/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin:/opt/tensorrt/bin
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c pip install --no-cache-dir /opt/pytorch/apex/dist/*.whl # buildkit
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c pip install --no-cache-dir /opt/pytorch/torch_tensorrt/dist/*.whl # buildkit
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c if [ "${L4T}" = "1" ]; then echo "Not installing Flash Attention wheel in iGPU as it is a requirement for Transformer Engine"; else pip install --no-cache-dir /opt/pytorch/flash_attn*.whl; fi # buildkit
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c if [ "${L4T}" = "1" ]; then echo "Not installing Transformer Engine in iGPU container until Version variable is set"; else /nvidia/build-scripts/installCAL.sh && /nvidia/build-scripts/installCUBLASMP.sh && /nvidia/build-scripts/installNVSHMEM.sh && CI_JOB_TOKEN=$(cat /run/secrets/CI_JOB_TOKEN) && echo $CI_JOB_TOKEN && git clone -b release_v${TRANSFORMER_ENGINE_VERSION} --single-branch --recursive https://gitlab-ci-token:${CI_JOB_TOKEN}@gitlab-master.nvidia.com/dl/transformerengine/transformerengine.git && env NVTE_CUDA_ARCHS="70;80;89;90;100;120" NVTE_BUILD_THREADS_PER_JOB=8 pip install --no-cache-dir --no-build-isolation ./transformerengine && rm -rf transformerengine; fi # buildkit
ENV CUDA_MODULE_LOADING=LAZY
ENV TORCH_NCCL_USE_COMM_NONBLOCKING=0
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c ln -sf ${_CUDA_COMPAT_PATH}/lib.real ${_CUDA_COMPAT_PATH}/lib && echo ${_CUDA_COMPAT_PATH}/lib > /etc/ld.so.conf.d/00-cuda-compat.conf && ldconfig && rm -f ${_CUDA_COMPAT_PATH}/lib # buildkit
COPY entrypoint.d/ /opt/nvidia/entrypoint.d/ # buildkit
COPY gloo.patch /gloo.patch # buildkit
RUN |8 NVIDIA_PYTORCH_VERSION=25.03 PYTORCH_BUILD_VERSION=2.7.0a0+7c8ec84 NVFUSER_BUILD_VERSION=c5e1555 TARGETARCH=amd64 PYVER=3.12 PYVER_MAJMIN=312 L4T=0 DALI_EXTRA_INDEX_URL=http://sqrl/nvdl/datasets/dali/pip-dali /bin/sh -c cd /opt/pytorch/pytorch/third_party/gloo && patch -p1 < /gloo.patch && cd - && rm /gloo.patch # buildkit
ARG NVIDIA_BUILD_ID=148941828
ENV NVIDIA_BUILD_ID=148941828
LABEL com.nvidia.build.id=148941828
ARG NVIDIA_BUILD_REF=84f446f5c7204f47f7769a5a5f759c7622b828eb
LABEL com.nvidia.build.ref=84f446f5c7204f47f7769a5a5f759c7622b828eb
ENV NVIDIA_PRODUCT_NAME=NeMo Framework
ENV PIP_NO_CACHE_DIR=1
ARG NVIDIA_BIGNLP_VERSION
ENV NVIDIA_BIGNLP_VERSION=
LABEL com.nvidia.bignlp.version=
ENV DEBIAN_FRONTEND=noninteractive
RUN |1 NVIDIA_BIGNLP_VERSION= /bin/sh -c apt-get update && apt-get install -y --no-install-recommends libsndfile1 sox swig libb64-dev openssh-server && rm -rf /var/lib/apt/lists/* && apt-get clean && pip install virtualenv && virtualenv /opt/venv # buildkit
RUN |1 NVIDIA_BIGNLP_VERSION= /bin/sh -c apt-get remove --purge -y libslurm* && apt-get -y autoremove && pip uninstall -y onnx && rm -rf /opt/pytorch/pytorch/third_party/onnx # buildkit
RUN |1 NVIDIA_BIGNLP_VERSION= /bin/sh -c cp /dev/null /etc/pip/constraint.txt # buildkit
WORKDIR /opt
RUN |1 NVIDIA_BIGNLP_VERSION= /bin/sh -c bash tensorrt_llm/docker/common/install_base.sh && bash tensorrt_llm/docker/common/install_cmake.sh && bash tensorrt_llm/docker/common/install_ccache.sh # buildkit
ARG TRT_VER=10.9.0.34
ARG CUDA_VER=12.8
ARG CUDNN_VER=9.8.0.87-1
ARG NCCL_VER=2.25.1-1+cuda12.8
ARG CUBLAS_VER=12.8.4.1-1
ARG NVRTC_VER=12.8.93-1
ARG TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2
ARG TRT_LLM_VERSION=0.18.0
ENV TRT_VER=10.9.0.34
ENV NEMO_FW_TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2
ENV NEMO_FW_TRT_LLM_VERSION=0.18.0
RUN |9 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 /bin/sh -c bash tensorrt_llm/docker/common/install_tensorrt.sh --TRT_VER=${TRT_VER} --CUDA_VER=${CUDA_VER} --CUDNN_VER=${CUDNN_VER} --NCCL_VER=${NCCL_VER} --CUBLAS_VER=${CUBLAS_VER} --NVRTC_VER=${NVRTC_VER} bash tensorrt_llm/docker/common/install_polygraphy.sh && bash tensorrt_llm/docker/common/install_mpi4py.sh # buildkit
ARG REINSTALL_NSYS=False
ARG NSYS_VERSION=2025.1.1.65
ENV NSIGHT_SYSTEMS_VERSION=2025.1.1.65
RUN |11 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 /bin/sh -c if [ $REINSTALL_NSYS = "True" ]; then /nvidia/build-scripts/installNSYS.sh; fi # buildkit
ARG TARGET_ARCH=x86
ARG DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076
ARG GDR_COPY_VERSION=v2.4.1
ENV CPATH=/usr/local/mpi/include:
ENV LD_LIBRARY_PATH=/usr/local/mpi/lib:/usr/local/lib/python3.12/dist-packages/torch/lib:/usr/local/lib/python3.12/dist-packages/torch_tensorrt/lib:/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
ENV GDRCOPY_HOME=/usr/src/gdrdrv-2.4.1/
RUN |14 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 /bin/sh -c apt-get update && apt-get install -y --no-install-recommends nvidia-dkms-535 devscripts debhelper fakeroot dkms check libsubunit0 libsubunit-dev libfabric-dev && if [ $TARGET_ARCH = "x86" ]; then ln -s /usr/lib/x86_64-linux-gnu/libmlx5.so.1 /usr/lib/x86_64-linux-gnu/libmlx5.so; else ln -s /usr/lib/aarch64-linux-gnu/libmlx5.so.1 /usr/lib/aarch64-linux-gnu/libmlx5.so; fi && rm -rf /var/lib/apt/lists/* && apt-get clean && git clone https://github.com/NVIDIA/gdrcopy.git && cd /opt/gdrcopy && git checkout $GDR_COPY_VERSION && CUDA=/usr/local/cuda packages/build-deb-packages.sh && dpkg -i gdrdrv-dkms_*.deb && dpkg -i libgdrapi_*.deb && dpkg -i gdrcopy-tests_*.deb && dpkg -i gdrcopy_*.deb && cd /opt && git clone https://github.com/deepseek-ai/DeepEP.git && cd DeepEP && git fetch origin $DEEPEP_COMMIT && git checkout FETCH_HEAD && cd /opt && wget https://developer.nvidia.com/downloads/assets/secure/nvshmem/nvshmem_src_3.2.5-1.txz && mkdir -p /opt/nvshmem_src && tar xf nvshmem_src_3.2.5-1.txz -C /opt && rm nvshmem_src_3.2.5-1.txz && cd /opt/nvshmem_src && git apply /opt/DeepEP/third-party/nvshmem.patch && CUDA_HOME=/usr/local/cuda NVSHMEM_SHMEM_SUPPORT=0 NVSHMEM_UCX_SUPPORT=0 NVSHMEM_USE_NCCL=0 NVSHMEM_IBGDA_SUPPORT=1 NVSHMEM_PMIX_SUPPORT=0 NVSHMEM_TIMEOUT_DEVICE_POLLING=0 NVSHMEM_USE_GDRCOPY=1 cmake -S . -B build/ -DCMAKE_INSTALL_PREFIX=/opt/nvshmem_src/install -DCMAKE_CUDA_ARCHITECTURES=90 && cd build && make install -j && rm -rf /opt/nvshmem_src/build && cd /opt/DeepEP && NVSHMEM_DIR=/opt/nvshmem_src/install python setup.py develop && NVSHMEM_DIR=/opt/nvshmem_src/install python setup.py install && apt-get remove --purge -y devscripts debhelper man-db && apt-get -y autoremove && apt-get clean # buildkit
ARG REINSTALL_APEX=False
ARG APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a
RUN |16 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a /bin/sh -c pip install packaging # buildkit
RUN |16 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a /bin/sh -c if [ $REINSTALL_APEX = "True" ]; then pip uninstall -y apex && git clone https://github.com/NVIDIA/apex && cd apex && if [ ! -z $APEX_COMMIT ]; then git fetch origin $APEX_COMMIT && git checkout FETCH_HEAD; fi && HEAD_APEX_COMMIT=$(git rev-parse HEAD) && echo "Container built with Apex commit hash: $HEAD_APEX_COMMIT" && pip install -e . -v --no-build-isolation --disable-pip-version-check --no-cache-dir --config-settings "--build-option=--cpp_ext --cuda_ext --fast_layer_norm --distributed_adam --deprecated_fused_adam --group_norm --nccl_allocator"; fi # buildkit
RUN |16 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a /bin/sh -c if [ -d /tmp/TransformerEngine ]; then pip install --no-cache-dir --no-build-isolation /tmp/TransformerEngine/dist/transformer_engine*; fi # buildkit
ARG MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17
ARG MCORE_VERSION=0.12.1
ENV NEMO_FW_MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17
ENV NEMO_FW_MCORE_VERSION=0.12.1
RUN |18 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 /bin/sh -c CI_JOB_TOKEN=$(cat /run/secrets/CI_JOB_TOKEN) && git clone https://gitlab-ci-token:${CI_JOB_TOKEN}@gitlab-master.nvidia.com/ADLR/megatron-lm.git && cd megatron-lm && git pull && if [ ! -z $MCORE_COMMIT ]; then git fetch origin $MCORE_COMMIT && git checkout FETCH_HEAD; fi && HEAD_MCORE_COMMIT=$(git rev-parse HEAD) && echo "Container built with megatron-lm commit hash: $HEAD_MCORE_COMMIT" && sed -i "/triton/d" requirements/pytorch_24.10/requirements.txt && sed -i "/nvidia-resiliency-ext/d" requirements/pytorch_24.10/requirements.txt && pip install -e . && rm -rf .git && cd megatron/core/datasets && make && rm -rf /root/.cache/bazel # buildkit
ENV PYTHONPATH=:/opt/megatron-lm
WORKDIR /opt/tensorrt_llm
RUN |18 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 /bin/sh -c pip install /src/tensorrt_llm/build/tensorrt_llm*.whl --extra-index-url https://pypi.nvidia.com --extra-index-url https://gitlab-master.nvidia.com/api/v4/projects/95421/packages/pypi/simple # buildkit
RUN |18 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 /bin/sh -c ln -sv $(python3 -c 'import site; print(f"{site.getsitepackages()[0]}/tensorrt_llm/libs")') lib && test -f lib/libnvinfer_plugin_tensorrt_llm.so && ln -sv lib/libnvinfer_plugin_tensorrt_llm.so lib/libnvinfer_plugin_tensorrt_llm.so.9 && echo "/opt/tensorrt_llm/lib" > /etc/ld.so.conf.d/tensorrt_llm.conf && ldconfig # buildkit
ARG SRC_DIR=/src/tensorrt_llm
COPY /src/tensorrt_llm/benchmarks benchmarks # buildkit
ARG CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build
COPY /src/tensorrt_llm/cpp/build/benchmarks/bertBenchmark /src/tensorrt_llm/cpp/build/benchmarks/gptManagerBenchmark /src/tensorrt_llm/cpp/build/benchmarks/gptSessionBenchmark benchmarks/cpp/ # buildkit
RUN |20 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build /bin/sh -c rm -v benchmarks/cpp/bertBenchmark.cpp benchmarks/cpp/gptManagerBenchmark.cpp benchmarks/cpp/gptSessionBenchmark.cpp benchmarks/cpp/CMakeLists.txt && apt-get remove --purge -y gdb && apt-get -y autoremove && apt-get clean # buildkit
ARG TARGET_ARCH=x86
ARG INSTALL_VLLM=True
WORKDIR /opt
COPY /opt/tinycudann*.whl ./ # buildkit
RUN |21 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True /bin/sh -c pip install --no-cache-dir /tmp/build_mamba_dep/causal-conv1d/causal_conv1d*.whl /tmp/build_grouped_gemm/grouped_gemm*.whl /tmp/build_nvdiffrast/nvdiffrast*.whl /tmp/build_stable_dreamfusion/raymarching*.whl /tmp/build_stable_dreamfusion/shencoder*.whl /tmp/build_stable_dreamfusion/freqencoder*.whl /tmp/build_stable_dreamfusion/gridencoder*.whl && if [ -f /tmp/build_mamba_dep/mamba/mamba*.whl ]; then pip install --no-cache-dir /tmp/build_mamba_dep/mamba/mamba*.whl; fi && if [ -f /tmp/build_bitsandbytes/bitsandbytes/dist/bitsandbytes*.whl ]; then pip install --no-cache-dir /tmp/build_bitsandbytes/bitsandbytes/dist/bitsandbytes*.whl; fi # buildkit
ARG ALIGNER_COMMIT=dev
ARG ALIGNER_VERSION=0.6.0rc0.dev0
ARG LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008
ENV NEMO_FW_LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008
RUN |24 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 /bin/sh -c git clone https://github.com/NVIDIA/NeMo-Framework-Launcher.git && cd NeMo-Framework-Launcher && git pull && if [ ! -z $LAUNCHER_COMMIT ]; then git fetch origin $LAUNCHER_COMMIT && git checkout FETCH_HEAD; fi && HEAD_LAUNCHER_COMMIT=$(git rev-parse HEAD) && echo "Container built with NeMo-Framework-Launcher commit hash: $HEAD_LAUNCHER_COMMIT" && pip install --no-cache-dir -r requirements.txt # buildkit
ENV LAUNCHER_SCRIPTS_PATH=/opt/NeMo-Framework-Launcher/launcher_scripts
ENV PYTHONPATH=/opt/NeMo-Framework-Launcher/launcher_scripts::/opt/megatron-lm
ARG CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1
ARG INSTALL_CURATOR=True
ARG CURATOR_VERSION=0.8.0rc4.dev0
ENV NEMO_FW_CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1
ENV NEMO_FW_CURATOR_VERSION=0.8.0rc4.dev0
RUN |27 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 /bin/sh -c if [ $INSTALL_CURATOR = "True" ]; then pip install mpi4py --no-build-isolation && if [ $TARGET_ARCH = "arm" ]; then pip install git+https://github.com/aboSamoor/pycld2.git; fi && git clone https://github.com/NVIDIA/NeMo-Curator.git && cd NeMo-Curator && git pull && if [ ! -z $CURATOR_COMMIT ]; then git fetch origin $CURATOR_COMMIT && git checkout FETCH_HEAD; fi && HEAD_CURATOR_COMMIT=$(git rev-parse HEAD) && echo "Container built with Curator commit hash: $HEAD_CURATOR_COMMIT" && sed -i "/nemo_toolkit/d" pyproject.toml && sed -i "/cudf-cu12/d" pyproject.toml && sed -i "/cugraph-cu12/d" pyproject.toml && sed -i "/cuml-cu12/d" pyproject.toml && sed -i "/dask-cuda/d" pyproject.toml && sed -i "/dask-cudf-cu12/d" pyproject.toml && cd ../ && pip install --extra-index-url=https://pypi.nvidia.com -e "./NeMo-Curator[all]" && pip install protobuf==4.24.4; fi # extra-index-url pypi could cause dependencies conflicts (later on) # buildkit
ARG TARGET_ARCH=x86
ARG INSTALL_VLLM=True
RUN |27 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 /bin/sh -c echo " UserKnownHostsFile /dev/null" >> /etc/ssh/ssh_config && sed -i 's/#\(StrictModes \).*/\1no/g' /etc/ssh/sshd_config && sed -i 's/# StrictHostKeyChecking ask/ StrictHostKeyChecking no/' /etc/ssh/ssh_config && mkdir -p /var/run/sshd # buildkit
ARG NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70
ARG NEMO_VERSION=2.3.1
ENV NEMO_FW_NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70
ENV NEMO_FW_NEMO_VERSION=2.3.1
RUN |29 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 /bin/sh -c git clone https://github.com/NVIDIA/NeMo.git && cd NeMo && git pull && if [ ! -z $NEMO_COMMIT ]; then git fetch origin $NEMO_COMMIT && git checkout FETCH_HEAD; fi && HEAD_NEMO_COMMIT=$(git rev-parse HEAD) && echo "Container built with NeMo commit hash: $HEAD_NEMO_COMMIT" && pip uninstall -y nemo_toolkit sacrebleu && sed -i "/mamba-ssm/d" requirements/requirements_nlp.txt && if [ $TARGET_ARCH = "arm" ]; then sed -i "/torch/d" requirements/requirements.txt && sed -i "/decord/d" requirements/requirements_multimodal.txt && sed -i "/megatron_core/d" requirements/requirements_nlp.txt; fi && pip install -e ".[all]" && cd nemo/collections/nlp/data/language_modeling/megatron && make # buildkit
ARG INSTALL_RESIL=False
ARG RESIL_COMMIT=not_installed
ARG RESIL_VERSION=not_installed
ENV NEMO_FW_RESIL_COMMIT=not_installed
ENV NEMO_FW_RESIL_VERSION=not_installed
RUN |32 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed /bin/sh -c if [ $INSTALL_RESIL = "True" ] && [ $TARGET_ARCH = "x86" ]; then pip install --no-cache-dir "git+https://github.com/NVIDIA/nvidia-resiliency-ext.git@${RESIL_COMMIT}"; fi # buildkit
ARG INSTALL_MODELOPT=True
ARG MODELOPT_VERSION=0.27.1
RUN |34 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 /bin/sh -c if [ $INSTALL_MODELOPT = "True" ]; then pip install --no-cache-dir nvidia-modelopt[torch]==$MODELOPT_VERSION; fi # buildkit
RUN |34 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 /bin/sh -c if [ $INSTALL_VLLM = "True" ]; then /opt/venv/bin/pip install -r /opt/NeMo/requirements/requirements_vllm.txt -r /opt/NeMo/requirements/requirements_deploy.txt; fi # buildkit
ARG NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f
ENV NEMO_FW_NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c git clone https://github.com/NVIDIA/NeMo-Run && cd NeMo-Run && git pull && if [ ! -z $NEMO_RUN_COMMIT ]; then git fetch origin $NEMO_RUN_COMMIT && git checkout FETCH_HEAD; fi && pip install -e . # buildkit
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c pip install --no-deps --no-cache-dir "liger-kernel==0.5.8; (platform_machine == 'x86_64' and platform_system != 'Darwin')" && pip install --no-deps "cut-cross-entropy @ git+https://github.com/apple/ml-cross-entropy.git@87a86aba72cfd2f0d8abecaf81c13c4528ea07d8; (platform_machine == 'x86_64' and platform_system != 'Darwin')" # buildkit
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c pip uninstall -y webdataset && pip install --no-cache-dir "webdataset==0.2.86" "pandas==2.2.3" "ctc_segmentation==1.7.1" "transformers==4.51.3" # buildkit
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c if [ $INSTALL_VLLM = "True" ]; then /opt/venv/bin/pip install "transformers==4.51.3"; fi # buildkit
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c pip uninstall -y tornado onnx h11 pytest-coverage levenshtein && pip install --no-cache-dir "tornado==6.5.0" "onnx==1.17.0" "h11==0.16.0" "setuptools==78.1.1" && sed -i '57d' /usr/local/lib/python3.12/dist-packages/torch_tensorrt/dynamo/conversion/custom_ops_converters.py # buildkit
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c apt-get remove --purge -y gdb libffi-dev ccache fakeroot libfakeroot libdebuginfod-common && apt-get -y autoremove && apt-get clean # buildkit
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c <<"EOF" python from transformers import AutoTokenizer, Qwen2Tokenizer, CLIPImageProcessor _=AutoTokenizer.from_pretrained('gpt2') _=AutoTokenizer.from_pretrained('bert-base-cased') _=AutoTokenizer.from_pretrained('bert-large-cased') _=AutoTokenizer.from_pretrained('bert-large-uncased') _=AutoTokenizer.from_pretrained('bigcode/starcoder2-tokenizer') _=AutoTokenizer.from_pretrained("EleutherAI/gpt-neox-20b") _=AutoTokenizer.from_pretrained("nvidia/Nemotron-H-8B-Base-8K") _=AutoTokenizer.from_pretrained("nvidia/Nemotron-H-47B-Base-8K") _=AutoTokenizer.from_pretrained("nvidia/Nemotron-H-56B-Base-8K") _=AutoTokenizer.from_pretrained('THUDM/chatglm2-6b',trust_remote_code=True) _=AutoTokenizer.from_pretrained('THUDM/chatglm3-6b',trust_remote_code=True) _=Qwen2Tokenizer.from_pretrained('qwen/Qwen1.5-7B',trust_remote_code=True) _=Qwen2Tokenizer.from_pretrained('qwen/Qwen1.5-14B',trust_remote_code=True) _=AutoTokenizer.from_pretrained('openai/clip-vit-large-patch14') _=CLIPImageProcessor.from_pretrained('openai/clip-vit-large-patch14') _=CLIPImageProcessor.from_pretrained('openai/clip-vit-large-patch14-336') EOF # buildkit
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c URM_API_TOKEN=$(cat /run/secrets/URM_API_TOKEN) && wget --http-user donghyukc --http-password ${URM_API_TOKEN} "https://urm.nvidia.com/artifactory/nemo-fw-generic-local/llama3_70b_tokenizer.tar.gz" && wget --http-user donghyukc --http-password ${URM_API_TOKEN} "https://urm.nvidia.com/artifactory/nemo-fw-generic-local/llama3_8b_tokenizer.tar.gz" && mkdir -p /tmp_assets && tar -xzvf llama3_8b_tokenizer.tar.gz -C /tmp_assets && tar -xzvf llama3_70b_tokenizer.tar.gz -C /tmp_assets && cp -rf /tmp_assets/hub ~/.cache/huggingface && rm -rf /tmp_assets && rm llama3_8b_tokenizer.tar.gz llama3_70b_tokenizer.tar.gz # buildkit
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c <<"EOF" python from transformers import AutoTokenizer _=AutoTokenizer.from_pretrained('meta-llama/Meta-Llama-3-8B') _=AutoTokenizer.from_pretrained('meta-llama/Meta-Llama-3-70B') _=AutoTokenizer.from_pretrained('meta-llama/Meta-Llama-3-8B-instruct') _=AutoTokenizer.from_pretrained('meta-llama/Meta-Llama-3-70B-instruct') EOF # buildkit
RUN |35 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f /bin/sh -c sed -i '/^LayerId/s/^/#/' /usr/local/lib/python3.12/dist-packages/cv2/typing/__init__.py # buildkit
ARG PRE_COMPILE_QUANT_KERNELS=True
RUN |36 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f PRE_COMPILE_QUANT_KERNELS=True /bin/sh -c if [ $PRE_COMPILE_QUANT_KERNELS = "True" ]; then python -c "import modelopt.torch.quantization.extensions as ext; ext.precompile()"; fi # buildkit
COPY nemo2/patches/triton-lang_triton_6570_lazy_init.patch /opt/ # buildkit
RUN |36 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f PRE_COMPILE_QUANT_KERNELS=True /bin/sh -c cd /usr/local/lib/python3.12/dist-packages && patch -p1 < /opt/triton-lang_triton_6570_lazy_init.patch && rm /opt/triton-lang_triton_6570_lazy_init.patch # buildkit
WORKDIR /workspace
RUN |36 NVIDIA_BIGNLP_VERSION= TRT_VER=10.9.0.34 CUDA_VER=12.8 CUDNN_VER=9.8.0.87-1 NCCL_VER=2.25.1-1+cuda12.8 CUBLAS_VER=12.8.4.1-1 NVRTC_VER=12.8.93-1 TRT_LLM_COMMIT=3c5fe5eb86077f67febc42070be11f11de17c1e2 TRT_LLM_VERSION=0.18.0 REINSTALL_NSYS=False NSYS_VERSION=2025.1.1.65 TARGET_ARCH=x86 DEEPEP_COMMIT=a84a24808fb0ea732f49b874cc456a69dde69076 GDR_COPY_VERSION=v2.4.1 REINSTALL_APEX=False APEX_COMMIT=e74a67bba3ee679f778670e17edc21639008ae0a MCORE_COMMIT=a845aa7e12b3a117e24c2352b9e3e60bad2e3a17 MCORE_VERSION=0.12.1 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build INSTALL_VLLM=True ALIGNER_COMMIT=dev ALIGNER_VERSION=0.6.0rc0.dev0 LAUNCHER_COMMIT=26e42fa91d9e608897ce485b4911c263a4b57008 CURATOR_COMMIT=33db56e304a3cb2d72d2038d393474905737a3e1 INSTALL_CURATOR=True CURATOR_VERSION=0.8.0rc4.dev0 NEMO_COMMIT=5ac3335a3dc2a76eb96889312f79904ba2cdcf70 NEMO_VERSION=2.3.1 INSTALL_RESIL=False RESIL_COMMIT=not_installed RESIL_VERSION=not_installed INSTALL_MODELOPT=True MODELOPT_VERSION=0.27.1 NEMO_RUN_COMMIT=bc412ee5584ed3072717af59f54565ec0d265a6f PRE_COMPILE_QUANT_KERNELS=True /bin/sh -c chmod -R a+w /workspace # buildkit
ARG NEMO_FW_PIPELINE_ID=28827164
ENV NEMO_FW_PIPELINE_ID=28827164
ARG NVIDIA_BUILD_ID
ENV NVIDIA_BUILD_ID=148941828
LABEL com.nvidia.build.id=148941828
ARG NVIDIA_BUILD_REF
LABEL com.nvidia.build.ref=

Labels

Key Value
com.nvidia.bignlp.version
com.nvidia.build.id 148941828
com.nvidia.build.ref
com.nvidia.cal.version 0.4.4.50
com.nvidia.cublas.version 12.8.4.1
com.nvidia.cublasmp.version 0.4.0.789
com.nvidia.cuda.version 9.0
com.nvidia.cudnn.version 9.8.0.87
com.nvidia.cufft.version 11.3.3.83
com.nvidia.curand.version 10.3.9.90
com.nvidia.cusolver.version 11.7.3.90
com.nvidia.cusparse.version 12.5.8.93
com.nvidia.cusparselt.version 0.7.1.0
com.nvidia.nccl.version 2.25.1
com.nvidia.npp.version 12.3.3.100
com.nvidia.nsightcompute.version 2025.1.1.2
com.nvidia.nsightsystems.version 2025.1.1.110
com.nvidia.nvjpeg.version 12.3.5.92
com.nvidia.pytorch.version 2.7.0a0+7c8ec84
com.nvidia.tensorrt.version 10.9.0.34
com.nvidia.tensorrtoss.version
com.nvidia.volumes.needed nvidia_driver
org.opencontainers.image.ref.name ubuntu
org.opencontainers.image.version 24.04
Details
Container
2025-07-02 05:13:36 +00:00
11
OCI / Docker
linux/amd64
29 GiB
Versions (2) View all
25.04.01 2025-07-02
latest 2025-07-01