Sign inSign up

vllm/vllm-openai:reflection-beam-cu129

Index digest

sha256:cc32acaceb9117927f4b8e398d6be4353c0fe0678d40d931f697b208cc859a65

OS/ARCH

linux/arm64

Compressed size

16.52 GB

Last pushed

14 days by vllmbot

Type

Image

Manifest digest

sha256:87178030bb0a7b1b8dbc424acbdc0ec475b2330d55e3a64a86e7b83df7cbe13f

Image Layers

Image Layers

1ARG RELEASE0 B
2ARG LAUNCHPAD_BUILD_ARCH0 B
3LABEL org.opencontainers.image.ref.name=ubuntu0 B
4LABEL org.opencontainers.image.version=24.040 B
5ADD file ... in / 28.55 MB
6CMD ["/bin/bash"]0 B
7ENV NVARCH=sbsa0 B
8ENV NVIDIA_REQUIRE_CUDA=cuda>=12.9 brand=unknown,driver>=535,driver<536 brand=grid,driver>=535,driver<536 brand=tesla,driver>=535,driver<5360 B
9ENV NV_CUDA_CUDART_VERSION=12.9.79-10 B
10ARG TARGETARCH0 B
11LABEL maintainer=NVIDIA CORPORATION <[email protected]>0 B
12RUN |1 TARGETARCH=arm64 /bin/sh -c4.32 MB
13ENV CUDA_VERSION=12.9.10 B
14RUN |1 TARGETARCH=arm64 /bin/sh -c93.06 MB
15RUN |1 TARGETARCH=arm64 /bin/sh -c182 B
16ENV PATH=/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin0 B
17ENV LD_LIBRARY_PATH=/usr/local/cuda/lib640 B
18COPY NGC-DL-CONTAINER-LICENSE / # buildkit6.72 KB
19ENV NVIDIA_VISIBLE_DEVICES=all0 B
20ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility0 B
21ARG INSTALL_RUBIN_PRERELEASE=false0 B
22ARG TARGETPLATFORM=linux/arm640 B
23ARG max_jobs=640 B
24ARG CUDA_VERSION=12.9.10 B
25ARG PYTHON_VERSION=3.120 B
26ARG NCCL_VERSION=2.30.70 B
27ARG DEADSNAKES_MIRROR_URL0 B
28ARG DEADSNAKES_GPGKEY_URL0 B
29ARG GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py0 B
30ENV DEBIAN_FRONTEND=noninteractive0 B
31WORKDIR /vllm-workspace104 B
32RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=64229 B
33RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=64216.18 MB
34RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=641.56 GB
35ENV CUDA_HOME=/usr/local/cuda0 B
36ENV PATH=/usr/local/cuda/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin0 B
37ENV LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda/lib640 B
38COPY tools/install_snapshot_runtime.sh /usr/local/bin/install_snapshot_runtime.sh # buildkit1.62 KB
39RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=6432 B
40ENV CRIU_CUDA_PLUGIN_DIR=/usr/local/lib/criu0 B
41RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=6432.41 MB
42ENV UV_HTTP_TIMEOUT=5000 B
43ENV UV_INDEX_STRATEGY=unsafe-best-match0 B
44ENV UV_LINK_MODE=copy0 B
45ENV UV_PYTHON_INSTALL_DIR=/opt/uv/python0 B
46ENV UV_CACHE_DIR=/opt/uv/cache0 B
47ENV UV_COMPILE_BYTECODE=10 B
48RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=64150 B
49RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=64164 B
50ENV UV_OVERRIDE=/etc/uv-overrides.txt0 B
51RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=643.39 KB
52COPY docker/entrypoints/vllm-nonroot-entrypoint.sh /usr/local/bin/vllm-nonroot-entrypoint.sh # buildkit2.03 KB
53RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=6432 B
54ENV VLLM_ENABLE_CUDA_COMPATIBILITY=00 B
55ARG PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl0 B
56COPY requirements/common.txt /tmp/common.txt # buildkit1.67 KB
57COPY requirements/cuda.txt /tmp/requirements-cuda.txt # buildkit721 B
58RUN |10 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=644.4 GB
59RUN |10 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=642.56 GB
60ARG GDRCOPY_CUDA_VERSION=12.80 B
61ARG GDRCOPY_OS_VERSION=Ubuntu24_040 B
62ARG TARGETPLATFORM=linux/arm640 B
63COPY tools/install_gdrcopy.sh /tmp/install_gdrcopy.sh # buildkit897 B
64RUN |12 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=64315.13 KB
65ARG TIMM_VERSION=>=1.0.170 B
66ARG RUNAI_MODEL_STREAMER_VERSION=>=0.15.70 B
67RUN |14 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=64289.77 MB
68ARG PIP_INDEX_URL UV_INDEX_URL=0 B
69ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL=0 B
70ARG PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl0 B
71ARG PIP_KEYRING_PROVIDER=disabled UV_KEYRING_PROVIDER=disabled0 B
72ARG PYTORCH_NIGHTLY0 B
73COPY /workspace/torch_lib_versions.txt torch_lib_versions.txt # buildkit199 B
74COPY /workspace/dist/wheel.sha256 /tmp/vllm-wheel.sha256 # buildkit237 B
75RUN |21 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=64742.16 MB
76RUN |21 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=64199 B
77RUN |21 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=6432 B
78ENV LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib640 B
79COPY /tmp/ep_kernels_workspace/dist/wheels.sha256 /tmp/ep-kernels-wheels.sha256 # buildkit251 B
80RUN |21 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=6418.09 MB
81ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib640 B
82COPY examples examples # buildkit309.63 KB
83COPY benchmarks benchmarks # buildkit296.16 KB
84COPY ./vllm/collect_env.py . # buildkit9.61 KB
85ARG TARGETPLATFORM=linux/arm640 B
86ARG INSTALL_KV_CONNECTORS=true0 B
87ARG INSTALL_RUBIN_PRERELEASE=false0 B
88ARG TRITON_INSTALL_FROM_SOURCE_REPO=0 B
89ARG CUDA_VERSION=12.9.10 B
90ARG VLLM_BUILD_COMMIT0 B
91ARG VLLM_BUILD_PIPELINE0 B
92ARG VLLM_BUILD_URL0 B
93ARG VLLM_IMAGE_TAG0 B
94ARG PIP_INDEX_URL UV_INDEX_URL=0 B
95ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL=0 B
96ENV UV_HTTP_TIMEOUT=5000 B
97ARG torch_cuda_arch_list=8.7 8.9 9.0 10.0+PTX0 B
98ENV TORCH_CUDA_ARCH_LIST=8.7 8.9 9.0 10.0+PTX0 B
99RUN |28 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=641.24 GB
100RUN |28 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=6432 B
101COPY /tmp/triton_from_source_workspace/dist/wheel.sha256 /tmp/triton-source-wheel.sha256 # buildkit212 B
102COPY tools/install_triton_from_source.sh /tmp/install_triton_from_source.sh # buildkit823 B
103RUN |28 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=64175 B
104ARG BUILD_NIXL=false0 B
105RUN |29 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=6432 B
106ENV VLLM_USAGE_SOURCE=production-docker-image0 B
107ENV VLLM_BUILD_COMMIT=unknown VLLM_BUILD_PIPELINE=local VLLM_BUILD_URL= VLLM_IMAGE_TAG=local/vllm-openai:dev0 B
108LABEL org.opencontainers.image.source=https://github.com/vllm-project/vllm org.opencontainers.image.revision=unknown org.opencontainers.image.version=local/vllm-openai:dev org.opencontainers.image.url=0 B
109ENTRYPOINT ["vllm" "serve"]0 B
110LABEL org.opencontainers.image.revision=8e0922e88bf9d4d5c0031f817a4e4749c9ca4ec2 io.inferact.flashinfer.version=0.7.2.dev20261014 io.inferact.flashinfer.release=nightly-v0.7.2-202610140 B
111COPY wheels /tmp/flashinfer-nightly-wheels # buildkit2.66 GB
112RUN /bin/sh -c uv pip2.71 GB
113COPY verify_runtime.py /tmp/verify_runtime.py # buildkit589 B
114RUN /bin/sh -c python3 /tmp/verify_runtime.py289 B

Command

 ARG RELEASE