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 15, 19:45 UTC
Uninstall dependencies before installing requirements/tpu.txt (#14586)
Signed-off-by: <ricliu@google.com> Signed-off-by: Richard Liu <ricliu@google.com>
Richard Liu committed
Mar 11, 2025 at 08:01 UTC
9ab326713fccd8acf1b0a8fd9cca157f39c20688
1 file changed
+3
Dockerfile.tpu
+3
@@ -15,6 +15,9 @@ ARG GIT_REPO_CHECK=0
15
RUN --mount=type=bind,source=.git,target=.git \
16
if [ "$GIT_REPO_CHECK" != 0 ]; then bash tools/check_repo.sh; fi
17
18
+# Remove existing versions of dependencies
19
+RUN pip uninstall -y torch torch_xla torchvision
20
+
21
ENV VLLM_TARGET_DEVICE="tpu"
22
RUN --mount=type=cache,target=/root/.cache/pip \
23
--mount=type=bind,source=.git,target=.git \