aws-neuron · aws-eddy · Dec 20, 2025 · Dec 20, 2025
@@ -2,7 +2,7 @@ hydra-core>=1.3.0
 omegaconf>=2.2,<2.3
 pyyaml==6.0.1
 torchmetrics>=0.4.1rc0,<=0.10.3
-transformers==4.52.4
+transformers==4.56.*
 wandb
 webdataset>=0.1.48,<=0.1.62
 pandas
@@ -22,7 +22,7 @@ ftfy
 gdown
 inflect
 jieba
-opencc==1.1.6
+opencc==1.1.9
 pangu
 rapidfuzz
 pybind11
@@ -39,12 +39,12 @@ python-daemon
 huggingface_hub>=0.27.1
 multiprocess==0.70.16
 numba<=0.60.0
-numpy>=1.24.3,<=1.25.2
 rouge_score
 setuptools>=70.0
 lightning==2.5.0
 ml-dtypes==0.5.0
 boto3==1.35.93
 botocore==1.35.93
 s3transfer==0.10.4
-s3fs
+s3fs
+numpy==1.26.4
@@ -0,0 +1,224 @@
+ARG BUILD_STAGE=prod
+
+FROM public.ecr.aws/docker/library/ubuntu:24.04 AS base
+
+LABEL dlc_major_version="1"
+LABEL maintainer="Amazon AI"
+
+# This arg required to stop docker build waiting for region configuration while installing tz data from ubuntu 24
+ARG DEBIAN_FRONTEND=noninteractive
+ARG PYTHON=python3.12
+ARG PYTHON_VERSION=3.12.11
+ARG PIP=pip3
+ARG OMPI_VERSION=4.1.5
+ARG PYPI_SIMPLE_URL="https://pypi.org/simple/"
+
+# Python won’t try to write .pyc or .pyo files on the import of source modules
+# Force stdin, stdout and stderr to be totally unbuffered. Good for logging
+ENV PYTHONDONTWRITEBYTECODE=1
+ENV PYTHONUNBUFFERED=1
+ENV PYTHONIOENCODING=UTF-8
+ENV LANG=C.UTF-8
+ENV LC_ALL=C.UTF-8
+ENV LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/opt/aws/neuron/lib"
+ENV LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/opt/amazon/efa/lib"
+ENV LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/opt/amazon/efa/lib64"
+ENV LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/opt/amazon/openmpi/lib64"
+ENV LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/usr/local/lib"
+ENV PATH="/opt/aws/neuron/bin:${PATH}"
+
+RUN apt-get update \
+ && apt-get upgrade -y \
+ && apt-get install -y --no-install-recommends \
+    build-essential \
+    ca-certificates \
+    cmake \
+    curl \
+    emacs \
+    git \
+    gnupg2 \
+    gpg-agent \
+    jq \
+    libopencv-dev \
+    libglib2.0-0 \
+    libgl1-mesa-dri \
+    libsm6 \
+    libxext6 \
+    libxrender-dev \
+    libssl-dev \
+    libsqlite3-dev \
+    libgdbm-dev \
+    libc6-dev \
+    libbz2-dev \
+    libncurses-dev \
+    libffi-dev \
+    libcap-dev \
+    libhwloc-dev \
+    openjdk-8-jdk-headless \
+    openjdk-8-jdk \
+    openjdk-8-jre \
+    openjdk-11-jdk \
+    openssl \
+    software-properties-common \
+    tk-dev \
+    unzip \
+    wget \
+    vim \
+    zlib1g-dev \
+ && rm -rf /var/lib/apt/lists/* \
+ && rm -rf /tmp/tmp* \
+ && apt-get clean
+
+# Install Open MPI and configure SSH for MPI operator in k8s
+RUN mkdir -p /tmp/openmpi \
+ && cd /tmp/openmpi \
+ && wget --quiet https://download.open-mpi.org/release/open-mpi/v4.1/openmpi-${OMPI_VERSION}.tar.gz \
+ && tar zxf openmpi-${OMPI_VERSION}.tar.gz \
+ && cd openmpi-${OMPI_VERSION} \
+ && ./configure --enable-orterun-prefix-by-default \
+ && make -j $(nproc) all \
+ && make install \
+ && ldconfig \
+ && rm -rf /tmp/openmpi
+
+# Install packages and configure SSH for MPI operator in k8s
+RUN apt-get update \
+ && apt-get install -y openmpi-bin openssh-server \
+ && mkdir -p /var/run/sshd \
+ && echo "    UserKnownHostsFile /dev/null" >> /etc/ssh/ssh_config \
+ && echo "    StrictHostKeyChecking no" >> /etc/ssh/ssh_config \
+ && sed -i 's/#\(StrictModes \).*/\1no/g' /etc/ssh/sshd_config \
+ && rm -rf /var/lib/apt/lists/* \
+ && rm -rf /tmp/tmp* \
+ && apt-get clean
+
+# Install Python
+RUN wget -q https://www.python.org/ftp/python/$PYTHON_VERSION/Python-$PYTHON_VERSION.tgz \
+ && tar -xzf Python-$PYTHON_VERSION.tgz \
+ && cd Python-$PYTHON_VERSION \
+ && ./configure --enable-shared --prefix=/usr/local \
+ && make -j $(nproc) && make install \
+ && cd .. && rm -rf ../Python-$PYTHON_VERSION* \
+ && ln -s /usr/local/bin/pip3 /usr/bin/pip \
+ && ln -s /usr/local/bin/$PYTHON /usr/local/bin/python \
+ && ${PIP} --no-cache-dir install --upgrade \
+    "awscli<2" \
+    pip \
+    requests \
+    setuptools \
+ && rm -rf ~/.cache/pip/*
+
+# U24 will not allow installation of pip packages outside of venv without this flag
+# This is because U24 ships with Python 3.12 by default and installation into the Python
+# interpreter’s directory are disabled outside of a virtual environment. 
+# https://peps.python.org/pep-0668/
+RUN ${PIP} config set global.break-system-packages true
+
+# Install EFA
+RUN apt-get update \
+ && cd $HOME \
+ && curl -O https://efa-installer.amazonaws.com/aws-efa-installer-latest.tar.gz \
+ && wget https://efa-installer.amazonaws.com/aws-efa-installer.key && gpg --import aws-efa-installer.key \
+ && cat aws-efa-installer.key | gpg --fingerprint \
+ && wget https://efa-installer.amazonaws.com/aws-efa-installer-latest.tar.gz.sig && gpg --verify ./aws-efa-installer-latest.tar.gz.sig \
+ && tar -xf aws-efa-installer-latest.tar.gz \
+ && cd aws-efa-installer \
+ && ./efa_installer.sh -y -g --skip-kmod --skip-limit-conf --no-verify \
+ && cd $HOME \
+ && rm -rf /var/lib/apt/lists/* \
+ && rm -rf /tmp/tmp* \
+ && apt-get clean
+
+WORKDIR /
+
+# The ENV variables declared below are changed in the previous section
+# Grouping these ENV variables in the first section causes
+# ompi_info to fail. This is only observed in CPU containers
+ENV PATH="$PATH:/home/.openmpi/bin"
+ENV LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/home/.openmpi/lib/"
+RUN ompi_info --parsable --all | grep mpi_built_with_cuda_support:value
+
+RUN mkdir -p /etc/pki/tls/certs && cp /etc/ssl/certs/ca-certificates.crt /etc/pki/tls/certs/ca-bundle.crt
+
+# Copy workaround script for incorrect hostname
+COPY changehostname.c /
+COPY --chmod=755 start_with_right_hostname.sh deep_learning_container.py /usr/local/bin/
+
+RUN HOME_DIR=/root \
+ && curl -o ${HOME_DIR}/oss_compliance.zip https://aws-dlinfra-utilities.s3.amazonaws.com/oss_compliance.zip \
+ && unzip ${HOME_DIR}/oss_compliance.zip -d ${HOME_DIR}/ \
+ && cp ${HOME_DIR}/oss_compliance/test/testOSSCompliance /usr/local/bin/testOSSCompliance \
+ && chmod +x /usr/local/bin/testOSSCompliance \
+ && chmod +x ${HOME_DIR}/oss_compliance/generate_oss_compliance.sh \
+ && ${HOME_DIR}/oss_compliance/generate_oss_compliance.sh ${HOME_DIR} ${PYTHON} \
+ && rm -rf ${HOME_DIR}/oss_compliance* \
+ && rm -rf /tmp/tmp*
+
+# Setting up APT and PIP repo for neuron artifacts
+ARG NEURON_APT_REPO=apt.repos.neuron.amazonaws.com
+ARG NEURON_APT_REPO_KEY
+ARG NEURON_PIP_REPO=pip.repos.neuron.amazonaws.com
+ARG NEURON_PIP_REPO_KEY
+RUN mkdir -p /etc/apt/keyrings \
+ && APT_REPO_PREFIX=$([ -n "${NEURON_APT_REPO_KEY}" ] && echo "${NEURON_APT_REPO_KEY}@" || echo "") \
+ && echo "deb [signed-by=/etc/apt/keyrings/neuron.gpg] https://${APT_REPO_PREFIX}${NEURON_APT_REPO} jammy main" > /etc/apt/sources.list.d/neuron.list \
+ && curl $([ -n "${NEURON_APT_REPO_KEY}" ] && echo "-u ${NEURON_APT_REPO_KEY}") --retry 3 --retry-delay 1 --retry-all-errors -fSL "https://${NEURON_APT_REPO}/GPG-PUB-KEY-AMAZON-AWS-NEURON.PUB" | gpg --dearmor > /etc/apt/keyrings/neuron.gpg
+
+# Neuron SDK components version numbers
+ARG NEURONX_RUNTIME_LIB_VERSION=2.29.40.0-f954cd7a5
+ARG NEURONX_COLLECTIVES_LIB_VERSION=2.29.41.0-681fef5f5
+ARG NEURONX_TOOLS_VERSION=2.27.33.0-5d9c0b901
+ARG NEURONX_CC_VERSION=2.22.12471.0+b4a00d10
+ARG NEURONX_JAX_TRAINING_VERSION=0.7.0.1.0.7377+5e6a4049
+
+FROM base AS repo
+
+# Install Neuron components from the apt and pip repos (latest versions)
+RUN apt-get update \
+ && apt-get install -y \
+    aws-neuronx-tools \
+    aws-neuronx-collectives \
+    aws-neuronx-runtime-lib \
+ && rm -rf /var/lib/apt/lists/* \
+ && rm -rf /tmp/tmp* \
+ && apt-get clean
+
+RUN PIP_REPO_URL=$([ -n "${NEURON_PIP_REPO_KEY}" ] && echo "https://${NEURON_PIP_REPO_KEY}@${NEURON_PIP_REPO}" || echo "https://${NEURON_PIP_REPO}") \
+ && ${PIP} install --no-cache-dir --force-reinstall \
+   --index-url ${PIP_REPO_URL} \
+   --extra-index-url ${PYPI_SIMPLE_URL} \
+   --trusted-host ${NEURON_PIP_REPO} \
+   "neuronx-cc>=2.0" \
+   jax-neuronx \
+ && rm -rf ~/.cache/pip/*
+
+FROM base AS prod
+
+# Install Neuron components
+# Install Neuron Driver, Runtime and Tools
+RUN apt-get update \
+ && apt-get install -y \
+    aws-neuronx-tools=$NEURONX_TOOLS_VERSION \
+    aws-neuronx-collectives=$NEURONX_COLLECTIVES_LIB_VERSION \
+    aws-neuronx-runtime-lib=$NEURONX_RUNTIME_LIB_VERSION \
+ && rm -rf /var/lib/apt/lists/* \
+ && rm -rf /tmp/tmp* \
+ && apt-get clean
+
+# Install JAX & Neuron CC
+RUN PIP_REPO_URL=$([ -n "${NEURON_PIP_REPO_KEY}" ] && echo "https://${NEURON_PIP_REPO_KEY}@${NEURON_PIP_REPO}" || echo "https://${NEURON_PIP_REPO}") \
+ && ${PIP} install --no-cache-dir --force-reinstall \
+   --index-url ${PIP_REPO_URL} \
+   --trusted-host ${NEURON_PIP_REPO} \
+   --extra-index-url ${PYPI_SIMPLE_URL} \
+   neuronx-cc==$NEURONX_CC_VERSION \
+   jax-neuronx==$NEURONX_JAX_TRAINING_VERSION \
+ && rm -rf ~/.cache/pip/*
+
+FROM ${BUILD_STAGE} AS final
+
+# Starts framework
+ENTRYPOINT ["bash", "-m", "start_with_right_hostname.sh"]
+CMD ["/bin/bash"]
+
+HEALTHCHECK CMD curl --fail http://localhost:8080/ping || exit 1