Skip to main content
NVIDIA
PyTorch
Container
NVIDIA
PyTorch

PyTorch is a GPU accelerated tensor computational framework. Functionality can be extended with common Python libraries such as NumPy and SciPy. Automatic differentiation is done with a tape-based system at the functional and neural network layer levels.

  • LayerLabelCreated
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4LABEL
    com.nvidia.build.ref=b14be72d76b2b3bdd017d646114e87ea8cb52f00
    04/05/2024 10:07 PM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ARG
    NVIDIA_BUILD_REF
    04/05/2024 10:07 PM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4LABEL
    com.nvidia.build.id=88132355
    04/05/2024 10:07 PM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
    NVIDIA_BUILD_ID=88132355
    04/05/2024 10:07 PM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ARG
    NVIDIA_BUILD_ID
    04/05/2024 10:07 PM UTC
    sha256:a3554c8bdfc0cbbb14894de2ed17f3da4c0df67e1993751749daef283777ef9eCOPY
    entrypoint.d/ /opt/nvidia/entrypoint.d/
    04/05/2024 10:07 PM UTC
    sha256:4b33d620dd992038b76c45463ab34e4c225964cc6ab5db80be295857feb30e5eRUN
    PYVER=3.10 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
    04/05/2024 10:07 PM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
    CUDA_MODULE_LOADING=LAZY
    04/05/2024 10:07 PM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
    NCCL_WORK_FIFO_DEPTH=4194304
    04/05/2024 10:07 PM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
    TORCH_CUDNN_V8_API_ENABLED=1
    04/05/2024 10:07 PM UTC
    ...