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.