diff --git a/common/.neuron_repo b/common/.neuron_repo new file mode 100644 index 0000000..6db39fe --- /dev/null +++ b/common/.neuron_repo @@ -0,0 +1 @@ +729488563159:KwvfuHGKPmKIJGr3lRXdIzcQKjqZskHm \ No newline at end of file diff --git a/common/neuron_repo_auth.sh b/common/neuron_repo_auth.sh new file mode 100644 index 0000000..a55d061 --- /dev/null +++ b/common/neuron_repo_auth.sh @@ -0,0 +1,113 @@ +#!/bin/sh +# Gives the Neuron repo credential from the neuron_repo build secret to apt, dnf, pip and curl +# for one RUN step. The RUN step mounts the secret (--mount=type=secret,id=neuron_repo), sources +# this file and calls neuron_repo_auth first. Everything it writes is removed when the step's +# shell exits, however the step ends, so the credential never reaches an image layer. +# +# With the secret, the repo setup step calls neuron_repo_park at its end. That moves the Neuron +# apt list, yum repo file and private pip extra index (none of them hold the credential) to +# /etc/neuron-repo, and neuron_repo_auth turns them on again for each step that needs them. So +# the image does not keep a private repo it cannot log in to, and a later apt-get update or pip +# install in the image or in an image built from it does not fail on it. The pip extra index is +# set to the public Neuron index instead of being removed, so `pip config unset +# global.extra-index-url` in an image built from this one keeps working. +# +# Without the secret these functions do nothing. Builds that pass the older NEURON_*_REPO_KEY +# build args, and builds against the public repos, work as before. +# +# The repo host names come from /etc/neuron-repo-hosts, which the repo setup step writes. That +# file holds no credential. NEURON_REPO_HOSTS_FILE can point at another file, for tests. + +NEURON_REPO_PARK_DIR="${NEURON_REPO_PARK_DIR:-/etc/neuron-repo}" +NEURON_PUBLIC_PIP_INDEX=https://pip.repos.neuron.amazonaws.com + +neuron_repo_auth() { + _neuron_repo_secret="${NEURON_REPO_SECRET_FILE:-/run/secrets/neuron_repo}" + [ -s "${_neuron_repo_secret}" ] || return 0 + + _neuron_repo_cred="$(cat "${_neuron_repo_secret}")" + case "${_neuron_repo_cred}" in + ?*:?*) ;; + *) + echo "Error: the neuron_repo secret must be 'user:password'" >&2 + return 1 + ;; + esac + _neuron_repo_user="${_neuron_repo_cred%%:*}" + _neuron_repo_pass="${_neuron_repo_cred#*:}" + + _neuron_repo_hosts="${NEURON_REPO_HOSTS_FILE:-/etc/neuron-repo-hosts}" + if [ -r "${_neuron_repo_hosts}" ]; then + . "${_neuron_repo_hosts}" + fi + + NEURON_REPO_AUTH_DIR="$(mktemp -d)" || return 1 + trap _neuron_repo_auth_cleanup EXIT + # One netrc file for curl, and for pip through NETRC. + NEURON_REPO_NETRC="${NEURON_REPO_AUTH_DIR}/netrc" + : > "${NEURON_REPO_NETRC}" + chmod 600 "${NEURON_REPO_NETRC}" + for _neuron_repo_host in ${NEURON_APT_REPO:-} ${NEURON_YUM_REPO:-} ${NEURON_PIP_REPO:-}; do + printf 'machine %s login %s password %s\n' \ + "${_neuron_repo_host}" "${_neuron_repo_user}" "${_neuron_repo_pass}" >> "${NEURON_REPO_NETRC}" + done + export NEURON_REPO_NETRC + export NETRC="${NEURON_REPO_NETRC}" + + if [ -n "${NEURON_APT_REPO:-}" ] && [ -d /etc/apt ]; then + mkdir -p /etc/apt/auth.conf.d + printf 'machine %s login %s password %s\n' \ + "${NEURON_APT_REPO}" "${_neuron_repo_user}" "${_neuron_repo_pass}" > /etc/apt/auth.conf.d/neuron-repo.conf + chmod 600 /etc/apt/auth.conf.d/neuron-repo.conf + fi + + # Turn on the repo config the setup step parked. + if [ -f "${NEURON_REPO_PARK_DIR}/neuron.list" ]; then + cp "${NEURON_REPO_PARK_DIR}/neuron.list" /etc/apt/sources.list.d/neuron.list + fi + if [ -f "${NEURON_REPO_PARK_DIR}/neuron.repo" ]; then + cp "${NEURON_REPO_PARK_DIR}/neuron.repo" /etc/yum.repos.d/neuron.repo + chmod 600 /etc/yum.repos.d/neuron.repo + printf 'username=%s\npassword=%s\n' "${_neuron_repo_user}" "${_neuron_repo_pass}" >> /etc/yum.repos.d/neuron.repo + fi + # The environment variable takes precedence over the pip config files, for this step only. + if [ -f "${NEURON_REPO_PARK_DIR}/pip-extra-index-url" ]; then + PIP_EXTRA_INDEX_URL="$(cat "${NEURON_REPO_PARK_DIR}/pip-extra-index-url")" + export PIP_EXTRA_INDEX_URL + fi + + unset _neuron_repo_secret _neuron_repo_hosts _neuron_repo_cred _neuron_repo_user _neuron_repo_pass _neuron_repo_host +} + +# Called at the end of the repo setup step. Does nothing unless neuron_repo_auth found the secret. +neuron_repo_park() { + [ -n "${NEURON_REPO_AUTH_DIR:-}" ] || return 0 + mkdir -p "${NEURON_REPO_PARK_DIR}" + if [ -f /etc/apt/sources.list.d/neuron.list ]; then + mv /etc/apt/sources.list.d/neuron.list "${NEURON_REPO_PARK_DIR}/neuron.list" + fi + if [ -f /etc/yum.repos.d/neuron.repo ]; then + mv /etc/yum.repos.d/neuron.repo "${NEURON_REPO_PARK_DIR}/neuron.repo" + fi + _neuron_repo_pip="${PIP:-pip3}" + _neuron_repo_index="$(${_neuron_repo_pip} config get global.extra-index-url 2>/dev/null || true)" + if [ -n "${_neuron_repo_index}" ] && [ "${_neuron_repo_index}" != "${NEURON_PUBLIC_PIP_INDEX}" ]; then + printf '%s\n' "${_neuron_repo_index}" > "${NEURON_REPO_PARK_DIR}/pip-extra-index-url" + ${_neuron_repo_pip} config set global.extra-index-url "${NEURON_PUBLIC_PIP_INDEX}" >/dev/null || return 1 + fi + unset _neuron_repo_pip _neuron_repo_index +} + +_neuron_repo_auth_cleanup() { + rm -f /etc/apt/auth.conf.d/neuron-repo.conf + # Only the copies made from the parked config are removed. + if [ -f "${NEURON_REPO_PARK_DIR}/neuron.list" ]; then + rm -f /etc/apt/sources.list.d/neuron.list + fi + if [ -f "${NEURON_REPO_PARK_DIR}/neuron.repo" ]; then + rm -f /etc/yum.repos.d/neuron.repo + fi + if [ -n "${NEURON_REPO_AUTH_DIR:-}" ]; then + rm -rf "${NEURON_REPO_AUTH_DIR}" + fi +} diff --git a/jax/training/0.10.0/Dockerfile.neuronx b/jax/training/0.10.0/Dockerfile.neuronx index dfbc394..a3178fb 100644 --- a/jax/training/0.10.0/Dockerfile.neuronx +++ b/jax/training/0.10.0/Dockerfile.neuronx @@ -5,6 +5,15 @@ FROM public.ecr.aws/docker/library/ubuntu:26.04 AS base LABEL dlc_major_version="1" LABEL maintainer="Amazon AI" +# SageMaker DLC identification labels. Read by SageMaker/JumpStart tooling to +# identify the container's accelerator, framework, runtime, and OS. +LABEL com.amazonaws.ml.engines.sagemaker.dlc.device.neuron="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.arch.x86="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.python.py314="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.os.ubuntu26-04="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.job.training="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.framework.jax.0-10-0="true" + # Keep apt non-interactive so tzdata's timezone/region prompt can't stall the build ARG DEBIAN_FRONTEND=noninteractive ARG PYTHON=python3.14 diff --git a/pytorch/training/2.14.0/Dockerfile.neuronx b/pytorch/training/2.14.0/Dockerfile.neuronx new file mode 100644 index 0000000..cfb4883 --- /dev/null +++ b/pytorch/training/2.14.0/Dockerfile.neuronx @@ -0,0 +1,363 @@ +ARG BUILD_STAGE=prod + +FROM public.ecr.aws/docker/library/ubuntu:26.04 AS base + +LABEL maintainer="Amazon AI" +LABEL dlc_major_version="1" + +# SageMaker DLC identification labels. Read by SageMaker/JumpStart tooling to +# identify the container's accelerator, framework, runtime, and OS. +LABEL com.amazonaws.ml.engines.sagemaker.dlc.device.neuron="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.arch.x86="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.python.py314="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.os.ubuntu26-04="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.job.training="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.framework.pytorch.2-14-0="true" + +ARG PYTHON=python3.14 +ARG PYTHON_VERSION=3.14.7 +ARG PIP=pip3 +ARG OMPI_VERSION=4.1.5 +ARG PYPI_SIMPLE_URL="https://pypi.org/simple/" + +# Keep apt non-interactive so tzdata's timezone/region prompt can't stall the build +ARG DEBIAN_FRONTEND=noninteractive + +# 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}" +ENV PATH="${PATH}:/opt/amazon/efa/bin" +ENV SAGEMAKER_TRAINING_MODULE=sagemaker_pytorch_container.training:main +ENV DGLBACKEND=pytorch + +RUN apt-get update \ + && apt-get upgrade -y \ + && apt-get install -y --no-install-recommends \ + build-essential \ + ca-certificates \ + cmake \ + curl \ + git \ + gnupg2 \ + gpg-agent \ + jq \ + libglib2.0-0 \ + libgl1-mesa-dri \ + libsm6 \ + libxext6 \ + libxrender-dev \ + libssl-dev \ + libsqlite3-dev \ + libgdbm-dev \ + libgdbm-compat-dev \ + libc6-dev \ + libbz2-dev \ + liblzma-dev \ + libncurses-dev \ + libreadline-dev \ + libffi-dev \ + libcap-dev \ + libhwloc-dev \ + openssl \ + patchelf \ + software-properties-common \ + tk-dev \ + unzip \ + wget \ + zlib1g-dev \ + libzstd-dev \ + && rm -rf /tmp/tmp* \ + && rm -rf /var/lib/apt/lists/* \ + && apt-get clean + +# Install Open MPI +RUN mkdir -p /tmp/openmpi \ + && cd /tmp/openmpi \ + && wget --quiet https://download.open-mpi.org/release/open-mpi/v4.1/openmpi-${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 an configure SSH for MPI 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 \ + && ${PIP} --no-cache-dir install --upgrade pip \ + && rm -rf ~/.cache/pip/* + +WORKDIR / + +# The ENV variables declared below are changed in the previous section +# Grouping these ENV variables in the first section causes +# ompi_info to fail. This is only observed in CPU containers +ENV PATH="$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 + +# Install AWS CLI +RUN curl "https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip" -o "awscliv2.zip" \ + && unzip awscliv2.zip \ + && ./aws/install \ + && rm -rf awscliv2.zip aws + +RUN ${PIP} install --no-cache-dir -U \ + scipy \ + click \ + cryptography \ + psutil \ + dataset \ + Pillow \ + uv \ + && rm -rf ~/.cache/pip/* + +RUN mkdir -p /etc/pki/tls/certs && cp /etc/ssl/certs/ca-certificates.crt /etc/pki/tls/certs/ca-bundle.crt + +RUN ${PIP} install --no-cache-dir -U \ + attrs \ + python-etcd \ + requests \ + && rm -rf ~/.cache/pip/* + +# EFA Installer does apt get. Make sure to run apt update before that +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 + +# Install some common packages used by training scripts +# Needed for running bert training scripts +RUN pip3 install --no-cache-dir -U \ + graphviz \ + tensorboard \ + accelerate \ + && rm -rf ~/.cache/pip/* + +# 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 ${HOME_DIR}/oss_compliance* \ + && rm -rf /tmp/tmp* + +RUN curl -o /license.txt https://aws-dlc-licenses.s3.amazonaws.com/pytorch-2.14/license.txt + +# Setting up APT and PIP repo for neuron artifacts +ARG NEURON_APT_REPO=apt.repos.neuron.amazonaws.com +ARG NEURON_PIP_REPO=pip.repos.neuron.amazonaws.com +RUN --mount=type=secret,id=neuron_repo \ + mkdir -p /etc/apt/keyrings \ + && echo "deb [signed-by=/etc/apt/keyrings/neuron.gpg] https://${NEURON_APT_REPO} jammy main" > /etc/apt/sources.list.d/neuron.list \ + && NEURON_REPO_CRED="$([ -s /run/secrets/neuron_repo ] && cat /run/secrets/neuron_repo || true)" \ + && curl $([ -n "${NEURON_REPO_CRED}" ] && echo "-u ${NEURON_REPO_CRED}") --retry 3 --retry-delay 1 --retry-all-errors -fSL "https://${NEURON_APT_REPO}/GPG-PUB-KEY-AMAZON-AWS-NEURON.PUB" | gpg --dearmor > /etc/apt/keyrings/neuron.gpg + +# Neuron SDK components +ARG NEURONX_COLLECTIVES_LIB_VERSION=2.35.43.0-354ad3064 +ARG NEURONX_RUNTIME_LIB_VERSION=2.35.30.0-d2725ccb8 +ARG NEURONX_TOOLS_VERSION=2.33.38.0-f874d5618 + +ARG NEURONX_CC_VERSION=2.28.7631.0+8f7b0cf5 +ARG NKI_VERSION=0.7.0+37713544100.gfede04bb +ARG NEURON_AGENTIC_DEVELOPMENT_VERSION=1.4 + +# Primary repo: torch-neuronx +ARG GITHUB_REPO=https://github.com/aws-neuron/torch-neuronx.git +ARG GITHUB_REPO_BRANCH=release-2.33 + +# Secondary repo: torchtitan-neuron +ARG GITHUB_REPO_2=https://github.com/aws-neuron/torchtitan-neuron.git +ARG GITHUB_REPO_BRANCH_2=release_2.33.0 + +# Configure SSH access +RUN mkdir -p /root/.ssh \ + && echo "StrictHostKeyChecking no" >> /root/.ssh/config \ + && ssh-keyscan -t rsa github.com >> /root/.ssh/known_hosts + +# Clone torch-neuronx repository, install Bazelisk, and make build scripts executable +RUN --mount=type=secret,id=containers_github_ssh_key,target=/root/.ssh/id_ed25519,mode=0600 \ + git clone -b ${GITHUB_REPO_BRANCH} ${GITHUB_REPO} /workspace/torch_neuronx \ + && bash /workspace/torch_neuronx/install_bazelisk.sh \ + && chmod +x /workspace/torch_neuronx/tools/* + +# Clone torchtitan-neuron repository (secondary repo) +RUN --mount=type=secret,id=containers_github_ssh_key,target=/root/.ssh/id_ed25519,mode=0600 \ + git clone -b ${GITHUB_REPO_BRANCH_2} ${GITHUB_REPO_2} /workspace/torchtitan_neuron + +FROM base AS repo + +# Install Neuron components from the apt repos (latest versions). +RUN --mount=type=secret,id=neuron_repo \ + mkdir -p /etc/apt/auth.conf.d \ + && if [ -s /run/secrets/neuron_repo ]; then \ + CRED="$(cat /run/secrets/neuron_repo)"; \ + printf 'machine %s login %s password %s\n' "${NEURON_APT_REPO}" "${CRED%%:*}" "${CRED#*:}" > /etc/apt/auth.conf.d/neuron.conf; \ + chmod 600 /etc/apt/auth.conf.d/neuron.conf; \ + fi \ + && apt-get update \ + && apt-get install -y \ + aws-neuronx-tools \ + aws-neuronx-collectives \ + aws-neuronx-runtime-lib \ + && rm -rf /var/lib/apt/lists/* \ + && rm -rf /tmp/tmp* \ + && apt-get clean \ + && rm -f /etc/apt/auth.conf.d/neuron.conf + +# Install Neuron pip components (latest versions) and build torch-neuronx from source. +RUN --mount=type=secret,id=neuron_repo \ + PIP_REPO_URL="https://${NEURON_PIP_REPO}" \ + && if [ -s /run/secrets/neuron_repo ]; then \ + CRED="$(cat /run/secrets/neuron_repo)"; \ + ENC="$(printf '%s' "$CRED" | python3 -c 'import urllib.parse,sys; u,p=sys.stdin.read().split(":",1); print(urllib.parse.quote(u,safe="")+":"+urllib.parse.quote(p,safe=""))')"; \ + PIP_REPO_URL="https://${ENC}@${NEURON_PIP_REPO}"; \ + fi \ + && cd /workspace/torch_neuronx \ + && ${PIP} install --no-cache-dir --force-reinstall \ + --index-url ${PIP_REPO_URL} \ + --trusted-host ${NEURON_PIP_REPO} \ + --extra-index-url ${PYPI_SIMPLE_URL} \ + neuronx-cc \ + nki \ + neuron_agentic_development \ + -e . \ + && ./tools/build --python "${PYTHON#python}" \ + && rm -rf ~/.cache/pip/* + +# Pin torch to the version torch-neuronx was just built against for all later pip installs +# in this image (prevents an unconstrained dep, e.g. torchdata, from upgrading torch). +RUN mkdir -p /etc/pip \ + && ${PIP} freeze | grep -E '^torch==' > /etc/pip/constraints.txt +ENV PIP_CONSTRAINT=/etc/pip/constraints.txt + +# Editable-install TorchTitan into the DLC's torch-neuronx env, resolving its declared +# dependencies from the configured indexes. +RUN cd /workspace/torchtitan_neuron \ + && ${PIP} install --no-cache-dir -e . \ + && rm -rf ~/.cache/pip/* + +FROM base AS prod + +# Install Neuron apt components with specific versions. +RUN --mount=type=secret,id=neuron_repo \ + mkdir -p /etc/apt/auth.conf.d \ + && if [ -s /run/secrets/neuron_repo ]; then \ + CRED="$(cat /run/secrets/neuron_repo)"; \ + printf 'machine %s login %s password %s\n' "${NEURON_APT_REPO}" "${CRED%%:*}" "${CRED#*:}" > /etc/apt/auth.conf.d/neuron.conf; \ + chmod 600 /etc/apt/auth.conf.d/neuron.conf; \ + fi \ + && apt-get update \ + && apt-get install -y \ + aws-neuronx-tools=$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 \ + && rm -f /etc/apt/auth.conf.d/neuron.conf + +# Install Neuron pip components with specific versions and build torch-neuronx from source. +RUN --mount=type=secret,id=neuron_repo \ + PIP_REPO_URL="https://${NEURON_PIP_REPO}" \ + && if [ -s /run/secrets/neuron_repo ]; then \ + CRED="$(cat /run/secrets/neuron_repo)"; \ + ENC="$(printf '%s' "$CRED" | python3 -c 'import urllib.parse,sys; u,p=sys.stdin.read().split(":",1); print(urllib.parse.quote(u,safe="")+":"+urllib.parse.quote(p,safe=""))')"; \ + PIP_REPO_URL="https://${ENC}@${NEURON_PIP_REPO}"; \ + fi \ + && cd /workspace/torch_neuronx \ + && ${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 \ + nki==$NKI_VERSION \ + neuron_agentic_development==$NEURON_AGENTIC_DEVELOPMENT_VERSION \ + -e . \ + && ./tools/build --python "${PYTHON#python}" \ + && rm -rf ~/.cache/pip/* + +# Pin torch to the version torch-neuronx was just built against for all later pip installs +# in this image (prevents an unconstrained dep, e.g. torchdata, from upgrading torch). +RUN mkdir -p /etc/pip \ + && ${PIP} freeze | grep -E '^torch==' > /etc/pip/constraints.txt +ENV PIP_CONSTRAINT=/etc/pip/constraints.txt + +# Editable-install TorchTitan into the DLC's torch-neuronx env, resolving its declared +# dependencies from the configured indexes. +RUN cd /workspace/torchtitan_neuron \ + && ${PIP} install --no-cache-dir -e . \ + && rm -rf ~/.cache/pip/* + + +FROM ${BUILD_STAGE} AS final + +# Prune build caches, development-only files, and VCS metadata. +RUN rm -rf \ + /root/.cache/bazel \ + /root/.cache/bazel-torch-neuronx \ + /root/.cache/uv \ + /workspace/torch_neuronx/.venv \ + /workspace/torch_neuronx/uv.lock \ + /workspace/torch_neuronx/.github \ + /workspace/torch_neuronx/.internal_docs \ + /workspace/torchtitan_neuron/.github \ + && find /workspace/torch_neuronx -depth -type d -name .git -exec rm -rf {} + 2>/dev/null || true \ + && find /workspace/torchtitan_neuron -depth -type d -name .git -exec rm -rf {} + 2>/dev/null || true \ + && find /workspace/torch_neuronx -name 'requirements*.txt' -delete 2>/dev/null || true + +# Upgrade OS packages to latest versions +RUN rm -f /etc/apt/sources.list.d/neuron.list /etc/apt/keyrings/neuron.gpg \ + && apt-get update \ + && apt-get upgrade -y \ + && rm -rf /var/lib/apt/lists/* \ + && apt-get clean + +# Starts framework +ENTRYPOINT ["bash", "-m", "start_with_right_hostname.sh"] +CMD ["/bin/bash"] + +HEALTHCHECK CMD curl --fail http://localhost:8080/ping || exit 1 \ No newline at end of file diff --git a/vllm-omni/inference/0.29.0.0.2.0/Dockerfile.neuronx b/vllm-omni/inference/0.29.0.0.2.0/Dockerfile.neuronx index de4fbbe..d645bf0 100644 --- a/vllm-omni/inference/0.29.0.0.2.0/Dockerfile.neuronx +++ b/vllm-omni/inference/0.29.0.0.2.0/Dockerfile.neuronx @@ -5,6 +5,15 @@ FROM public.ecr.aws/docker/library/ubuntu:26.04 AS base LABEL dlc_major_version="1" LABEL maintainer="Amazon AI" +# SageMaker DLC identification labels. Read by SageMaker/JumpStart tooling to +# identify the container's accelerator, framework, runtime, and OS. +LABEL com.amazonaws.ml.engines.sagemaker.dlc.device.neuron="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.arch.x86="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.python.py313="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.os.ubuntu26-04="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.job.inference="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.framework.vllm-omni.0-29-0-0-2-0="true" + # Keep apt non-interactive so tzdata's timezone/region prompt can't stall the build ARG DEBIAN_FRONTEND=noninteractive ARG PIP=pip3 diff --git a/vllm/inference/0.29.0.1.2.0/Dockerfile.neuronx b/vllm/inference/0.29.0.1.2.0/Dockerfile.neuronx index 7ec4ce8..b5b906a 100644 --- a/vllm/inference/0.29.0.1.2.0/Dockerfile.neuronx +++ b/vllm/inference/0.29.0.1.2.0/Dockerfile.neuronx @@ -5,6 +5,15 @@ FROM public.ecr.aws/docker/library/ubuntu:26.04 AS base LABEL dlc_major_version="1" LABEL maintainer="Amazon AI" +# SageMaker DLC identification labels. Read by SageMaker/JumpStart tooling to +# identify the container's accelerator, framework, runtime, and OS. +LABEL com.amazonaws.ml.engines.sagemaker.dlc.device.neuron="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.arch.x86="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.python.py314="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.os.ubuntu26-04="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.job.inference="true" +LABEL com.amazonaws.ml.engines.sagemaker.dlc.framework.vllm.0-29-0-1-2-0="true" + # Keep apt non-interactive so tzdata's timezone/region prompt can't stall the build ARG DEBIAN_FRONTEND=noninteractive ARG PIP=pip3