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, 00:45 UTC
[ci/build] sync default value for wheel size (#12398)
Signed-off-by: youkaichao <youkaichao@gmail.com>
youkaichao committed
Jan 24, 2025 at 17:54 UTC
e784c6b9984e8f8116f74000b863d941495acb0b
2 files changed
+3
-2
.buildkite/check-wheel-size.py
+1
@@ -5,6 +5,7 @@ import zipfile
5
# Read the VLLM_MAX_SIZE_MB environment variable, defaulting to 300 MiB
6
# Note that we have 400 MiB quota, please use it wisely.
7
# See https://github.com/pypi/support/issues/3792 .
8
+# Please also sync the value with the one in Dockerfile.
9
VLLM_MAX_SIZE_MB = int(os.environ.get('VLLM_MAX_SIZE_MB', 300))
10
11
Dockerfile
+2
-2
@@ -126,8 +126,8 @@ RUN --mount=type=cache,target=/root/.cache/ccache \
126
127
# Check the size of the wheel if RUN_WHEEL_CHECK is true
128
COPY .buildkite/check-wheel-size.py check-wheel-size.py
129
-# Default max size of the wheel is 250MB
130
-ARG VLLM_MAX_SIZE_MB=250
129
+# sync the default value with .buildkite/check-wheel-size.py
130
+ARG VLLM_MAX_SIZE_MB=300
131
ENV VLLM_MAX_SIZE_MB=$VLLM_MAX_SIZE_MB
132
ARG RUN_WHEEL_CHECK=true
133
RUN if [ "$RUN_WHEEL_CHECK" = "true" ]; then \