NVIDIA
NeMo Curator
Container
NVIDIA
NeMo Curator

NVIDIA NeMo™ Curator accelerates generative AI model development with GPU-powered data curation, offering comprehensive text, image, video, and audio processing for enterprise-scale deployments.

LayerLabelCreated
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4LABEL
com.nvidia.build.ref=f607471f568e52d7950d5e45419006e6cd801882
07/18/2025 11:05 PM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ARG
NVIDIA_BUILD_REF=f607471f568e52d7950d5e45419006e6cd801882
07/18/2025 11:05 PM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4LABEL
com.nvidia.build.id=190213493
07/18/2025 11:05 PM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
NVIDIA_BUILD_ID=190213493
07/18/2025 11:05 PM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ARG
NVIDIA_BUILD_ID=190213493
07/18/2025 11:05 PM UTC
sha256:4f4fb700ef54461cfa02571ae0db9a0dc1e0cdb5577484a6d75e68dc38e8acc1RUN
NVIDIA_BIGNLP_VERSION= TARGET_ARCH=arm CURATOR_COMMIT=6af38b2f953717cd973e77340489a3e3a5ef1844 CURATOR_VERSION=0.9.0rc0.dev0 chmod -R a+w /workspace
07/18/2025 11:05 PM UTC
sha256:4f4fb700ef54461cfa02571ae0db9a0dc1e0cdb5577484a6d75e68dc38e8acc1WORKDIR
/workspace
07/18/2025 11:05 PM UTC
sha256:85a349dc50efdae3a510ea16965547b17332a86705981f6f7251b637f97bc25bRUN
NVIDIA_BIGNLP_VERSION= TARGET_ARCH=arm CURATOR_COMMIT=6af38b2f953717cd973e77340489a3e3a5ef1844 CURATOR_VERSION=0.9.0rc0.dev0 pip install --no-cache-dir "jupyter-core==5.8.1" "pillow==11.3.0" "protobuf==4.25.8"
07/18/2025 11:05 PM UTC
sha256:f52d8fb219c2396fa47ff65a06e851d0be8387f4bf428c1026d6ca9e33c3bd32RUN
NVIDIA_BIGNLP_VERSION= TARGET_ARCH=arm CURATOR_COMMIT=6af38b2f953717cd973e77340489a3e3a5ef1844 CURATOR_VERSION=0.9.0rc0.dev0 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 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 "./Curator[all]"
07/18/2025 11:05 PM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
NEMO_FW_CURATOR_VERSION=0.9.0rc0.dev0
07/18/2025 11:05 PM UTC
...