@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 29f1d47e73
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 15, 23:30 UTC

[MISC] Always import version library first in the vllm package (#12979)

Signed-off-by: Lu Fang <lufang@fb.com>

Lu Fang committed Feb 9, 2025 at 02:56 UTC 29f1d47e73de3764c944a0af0ff10bbc8ce244f4
1 file changed +4 -2
vllm/__init__.py
+4 -2
@@ -1,5 +1,9 @@
1 # SPDX-License-Identifier: Apache-2.0
2 """vLLM: a high-throughput and memory-efficient inference engine for LLMs"""
3 +# The version.py should be independent library, and we always import the
4 +# version library first. Such assumption is critical for some customization.
5 +from .version import __version__, __version_tuple__ # isort:skip
6 +
7 import os
8
9 import torch
@@ -19,8 +23,6 @@ from vllm.outputs import (ClassificationOutput, ClassificationRequestOutput,
23 from vllm.pooling_params import PoolingParams
24 from vllm.sampling_params import SamplingParams
25
22 -from .version import __version__, __version_tuple__
23 -
26 # set some common config/environment variables that should be set
27 # for all processes created by vllm and all processes
28 # that interact with vllm workers.