@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 73388c07a4
Fork 0

A high-throughput and memory-efficient inference and serving engine for LLMs

Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git

Synced Nov 16, 19:30 UTC

[TPU] Fix dockerfile.tpu (#7331)

Woosuk Kwon committed Aug 8, 2024 at 20:24 UTC 73388c07a42233a1010595edaab73a9e7ab8d9a4
1 file changed +4 -10
Dockerfile.tpu
+4 -10
@@ -4,20 +4,14 @@ ARG BASE_IMAGE="us-central1-docker.pkg.dev/tpu-pytorch-releases/docker/xla:night
4 FROM $BASE_IMAGE
5 WORKDIR /workspace
6
7 -# Install aiohttp separately to avoid build errors.
8 -RUN pip install aiohttp
9 -# Install NumPy 1 instead of NumPy 2.
10 -RUN pip install "numpy<2"
7 # Install the TPU and Pallas dependencies.
12 -RUN pip install torch_xla[tpu] -f https://storage.googleapis.com/libtpu-releases/index.html
13 -RUN pip install torch_xla[pallas] -f https://storage.googleapis.com/jax-releases/jax_nightly_releases.html -f https://storage.googleapis.com/jax-releases/jaxlib_nightly_releases.html
14 -
15 -# Fix FastAPI dependence
16 -RUN pip install "starlette<0.38.0"
8 +RUN python3 -m pip install torch_xla[tpu] -f https://storage.googleapis.com/libtpu-releases/index.html
9 +RUN python3 -m pip install torch_xla[pallas] -f https://storage.googleapis.com/jax-releases/jax_nightly_releases.html -f https://storage.googleapis.com/jax-releases/jaxlib_nightly_releases.html
10
11 # Build vLLM.
12 COPY . /workspace/vllm
13 ENV VLLM_TARGET_DEVICE="tpu"
21 -RUN cd /workspace/vllm && python setup.py develop
14 +RUN cd /workspace/vllm && python3 -m pip install -r requirements-tpu.txt
15 +RUN cd /workspace/vllm && python3 setup.py develop
16
17 CMD ["/bin/bash"]