链载Ai

标题: 轻松部署Gemma3-27B，L20服务器 最新版vLLM高效推理 [打印本页]

---

作者: 链载Ai    时间: 昨天 12:40
标题: 轻松部署Gemma3-27B，L20服务器 最新版vLLM高效推理

ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;color: rgb(31, 35, 41);margin: 0px 0px 4px;word-break: break-all;min-height: 20px;">Google最新开源的Gemma3-27B模型凭借其128K长上下文支持、多模态能力和接近闭源模型的性能表现，已成为企业级AI部署的热门选择。vLLM 0.7.4最新版已支持Gemma3-27B大模型！结合NVIDIA L20显卡的48GB大显存和vLLM推理框架的高吞吐特性，本文将详解从环境搭建到服务调优的全流程，助你快速实现高效推理。

ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;letter-spacing: 0.578px;margin-top: 0px;margin-bottom: 8px;font-size: 26px;padding-bottom: 12px;">一、环境准备

ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;letter-spacing: 0.578px;margin-top: 0px;margin-bottom: 8px;font-size: 22px;padding-bottom: 12px;">1.1 硬件配置

ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;padding: 8px;border: 1px solid rgb(204, 204, 204);white-space: normal;line-height: 2em;">服务器	ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;padding: 8px;border: 1px solid rgb(204, 204, 204);white-space: normal;line-height: 2em;">数量	ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;padding: 8px;border: 1px solid rgb(204, 204, 204);white-space: normal;line-height: 2em;">CPU	ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;padding: 8px;border: 1px solid rgb(204, 204, 204);white-space: normal;line-height: 2em;">内存（TB）	ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;padding: 8px;border: 1px solid rgb(204, 204, 204);white-space: normal;line-height: 2em;">系统版本
ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;padding: 8px;border: 1px solid rgb(204, 204, 204);white-space: normal;line-height: 2em;">NVIDIA L20 48GB * 8	ingFang SC", system-ui, -apple-system, BlinkMacSystemFont, "Helvetica Neue", "Hiragino Sans GB", "Microsoft YaHei UI", "Microsoft YaHei", Arial, sans-serif;padding: 8px;border: 1px solid rgb(204, 204, 204);white-space: normal;line-height: 2em;">1	INTEL 8458P *2	2	Ubuntu 20.04

1.2 系统环境

软件名称	版本	备注
NVIDIA Driver	550.54.14	GPU驱动
CUDA	12.4	Cuda
vLLM	0.7.4.dev473+g9ed6ee92.precompiled	LLM推理引擎

1.3 Gemma3-27B模型下载

• 方式一：通过HuggingFace 下载

仓库地址：https://huggingface.co/google/gemma-3-27b-it

• 方式二：通过HuggingFace镜像站-AI快站下载

仓库地址：https://aifasthub.com/google/gemma-3-27b-it

#下载AI快站下载器
wget https://fast360.xyz/images/hf-fast.sh
chmoda+x hf-fast.sh
#下载模型文件
./hf-fast.sh google/gemma-3-27b-it

模型文件大小为43G。

1.4 系统初始化

请参考之前文章：生产环境H200部署DeepSeek 671B 满血版全流程实战（一）：系统初始化

二、安装vLLM

2.1 创建虚拟环境

为避免依赖冲突，使用Conda创建独立环境：

conda create -n gemma3 python=3.12 
conda activate gemma3

2.2 安装最新版vLLM

这里我们使用Git安装最新版vLLM，以获取最新的功能和优化。

#使用阿里云pip源来加速下载
pip install --upgrade pip -i https://mirrors.aliyun.com/pypi/simple/

gitclonehttps://github.com/vllm-project/vllm.git
cdvllm
VLLM_USE_PRECOMPILED=1 pip install --editable . -i https://mirrors.aliyun.com/pypi/simple/

由于vLLM代码更新较快，编译安装最新版vLLM会因为本地 Git 分支与上游仓库不同步，如果遇到类似 "Local main branch is not up-to-date with upstream" 的错误：

Building wheelsforcollected packages: vllm
 Building editableforvllm (pyproject.toml) ... error
 error: subprocess-exited-with-error

 × Building editableforvllm (pyproject.toml) did not run successfully.
 │exitcode: 1
 ╰─> [114 lines of output]
   /tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/torch/_subclasses/functional_tensor.py:275: UserWarning: Failed to initialize NumPy: No module named'numpy'(Triggered internally at /pytorch/torch/csrc/utils/tensor_numpy.cpp:81.)
    cpu = _conversion_method_template(device=torch.device("cpu"))
   running editable_wheel
   creating /tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm.egg-info
   writing /tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm.egg-info/PKG-INFO
   writing dependency_links to /tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm.egg-info/dependency_links.txt
   writing entry points to /tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm.egg-info/entry_points.txt
   writing requirements to /tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm.egg-info/requires.txt
   writing top-level names to /tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm.egg-info/top_level.txt
   writing manifest file'/tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm.egg-info/SOURCES.txt'
   reading manifest template'MANIFEST.in'
   adding license file'LICENSE'
   writing manifest file'/tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm.egg-info/SOURCES.txt'
   creating'/tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm-0.7.4.dev474+g3556a414.precompiled.dist-info'
   creating /tmp/pip-wheel-4p1g9k_b/.tmp-rnwn8w46/vllm-0.7.4.dev474+g3556a414.precompiled.dist-info/WHEEL
   running build_py
   running build_ext
   Traceback (most recent call last):
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/command/editable_wheel.py", line 139,inrun
     self._create_wheel_file(bdist_wheel)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/command/editable_wheel.py", line 340,in_create_wheel_file
     files, mapping = self._run_build_commands(dist_name, unpacked, lib, tmp)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/command/editable_wheel.py", line 263,in_run_build_commands
     self._run_build_subcommands()
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/command/editable_wheel.py", line 290,in_run_build_subcommands
     self.run_command(name)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/_distutils/cmd.py", line 357,inrun_command
     self.distribution.run_command(command)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/dist.py", line 999,inrun_command
     super().run_command(command)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/_distutils/dist.py", line 1021,inrun_command
     cmd_obj.run()
    File"<string>", line 333,inrun
    File"<string>", line 319,inget_base_commit_in_main_branch
   ValueError: Local main branch (3556a414341033aad1bbb84674ec16b235324b25) is not up-to-date with upstream main branch (b82662d9523d9aa1386d8d1de410426781a1fa3b). Please pull the latest changes from upstream main branch first.
   /tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/_distutils/dist.py:1021: _DebuggingTips: Problemineditable installation.
   !!
  
       ********************************************************************************
       An error happenedwhileinstalling `vllm`ineditable mode.
  
       The following steps are recommended tohelpdebug this problem:
  
       - Try to install the project normally, without using the editable mode.
        Does the error still persist?
        (If it does, try fixing the problem before attempting the editable mode).
       - If you are using binary extensions, make sure you have all OS-level
        dependencies installed (e.g. compilers, toolchains, binary libraries, ...).
       - Try the latest version of setuptools (maybe the error was already fixed).
       - If you (or your project dependencies) are using any setuptools extension
        or customization, make sure they support the editable mode.
  
       After following the steps above,ifthe problem still persists and
       you think this is related to how setuptools handles editable installations,
       please submit a reproducible example
       (see https://stackoverflow.com/help/minimal-reproducible-example) to:
  
         https://github.com/pypa/setuptools/issues
  
       See https://setuptools.pypa.io/en/latest/userguide/development_mode.htmlfordetails.
       ********************************************************************************
  
   !!
    cmd_obj.run()
   Traceback (most recent call last):
    File"/home/ubuntu/miniconda3/envs/gemma3-2/lib/python3.12/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 389,in<module>
     main()
    File"/home/ubuntu/miniconda3/envs/gemma3-2/lib/python3.12/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 373,inmain
     json_out["return_val"] = hook(**hook_input["kwargs"])
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File"/home/ubuntu/miniconda3/envs/gemma3-2/lib/python3.12/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 303,inbuild_editable
    returnhook(wheel_directory, config_settings, metadata_directory)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/build_meta.py", line 476,inbuild_editable
    returnself._build_with_temp_dir(
        ^^^^^^^^^^^^^^^^^^^^^^^^^^
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/build_meta.py", line 407,in_build_with_temp_dir
     self.run_setup()
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/build_meta.py", line 320,inrun_setup
    exec(code, locals())
    File"<string>", line 682,in<module>
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/__init__.py", line 117,insetup
    returndistutils.core.setup(**attrs)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/_distutils/core.py", line 186,insetup
    returnrun_commands(dist)
        ^^^^^^^^^^^^^^^^^^
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/_distutils/core.py", line 202,inrun_commands
     dist.run_commands()
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/_distutils/dist.py", line 1002,inrun_commands
     self.run_command(cmd)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/dist.py", line 999,inrun_command
     super().run_command(command)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/_distutils/dist.py", line 1021,inrun_command
     cmd_obj.run()
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/command/editable_wheel.py", line 139,inrun
     self._create_wheel_file(bdist_wheel)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/command/editable_wheel.py", line 340,in_create_wheel_file
     files, mapping = self._run_build_commands(dist_name, unpacked, lib, tmp)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/command/editable_wheel.py", line 263,in_run_build_commands
     self._run_build_subcommands()
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/command/editable_wheel.py", line 290,in_run_build_subcommands
     self.run_command(name)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/_distutils/cmd.py", line 357,inrun_command
     self.distribution.run_command(command)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/dist.py", line 999,inrun_command
     super().run_command(command)
    File"/tmp/pip-build-env-tl3kug_g/overlay/lib/python3.12/site-packages/setuptools/_distutils/dist.py", line 1021,inrun_command
     cmd_obj.run()
    File"<string>", line 333,inrun
    File"<string>", line 319,inget_base_commit_in_main_branch
   ValueError: Local main branch (3556a414341033aad1bbb84674ec16b235324b25) is not up-to-date with upstream main branch (b82662d9523d9aa1386d8d1de410426781a1fa3b). Please pull the latest changes from upstream main branch first.
   [end of output]

 note: This error originates from a subprocess, and is likely not a problem with pip.
 ERROR: Failed building editableforvllm
Failed to build vllm
ERROR: Failed to build installable wheelsforsome pyproject.toml based projects (vllm)

解决办法：进入 vllm 源码目录，强制同步上游分支，确保本地代码与官方仓库完全同步

git fetch origin main
git reset --hard origin/main
git pull --rebase

#重新执行编译安装
VLLM_USE_PRECOMPILED=1 pip install --editable . -i https://mirrors.aliyun.com/pypi/simple/

验证安装是否成功：

2.3 安装transformers

pipinstallgit+https://github.com/huggingface/transformers@v4.49.0-Gemma-3

三、运行vLLM服务

完成vLLM的安装后，我们就可以启动vLLM服务，加载Gemma3-27B模型，开始进行推理。

1.激活 vllm 环境，确保在 vllm 的 conda 环境中

condaactivategemma3

2.启动 vLLM 服务：使用vllm serve命令启动 vLLM 服务，并加载 Gemma3-27B 模型。

conda activate gemma3

#使用2块L20加载模型
exportCUDA_VISIBLE_DEVICES=0,1
vllm serve /data/gemma-3-27b/gemma-3-27b-it --tensor-parallel-size 2 --max-model-len 16384 --port 8102 --trust-remote-code --served-model-name gemma3-27b --enable-chunked-prefill --max-num-batched-tokens 2048 --gpu-memory-utilization 0.95

参数说明：

• /path/to/gemma-3-27b-it：Gemma3-27B模型的路径，替换为你实际的路径。
• --tensor-parallel-size 2：使用2块GPU进行模型并行推理，根据实际情况修改。
• --max-model-len 16384：模型支持的最大输入长度。
• --port 8102：服务端口号。
• --trust-remote-code：信任远程代码。
• --served-model-name gemma3-27b：服务名称。
• --enable-chunked-prefill：开启chunked prefill，可以提高推理速度。
• --max-num-batched-tokens 2048：最大batch token数量。
• --gpu-memory-utilization 0.95：GPU显存利用率。

3.Nvitop监控

可以使用Nvitop监控GPU的使用情况。

四、验证服务可用性

启动vLLM服务后，我们可以发送API请求，测试服务是否正常运行。

curl -X POST"http://localhost:8102/v1/chat/completions"\
-H"Content-Type: application/json"\
-d '{
"model":"gemma3-27b",
"messages":[{"role":"user","content":"写个100字的散文"}]
}'

预期结果：返回模型生成的文本内容，表示服务部署成功。

五、总结

本文详细介绍了如何在L20服务器上使用最新版vLLM部署Gemma3-27B模型。通过本文相信你已经成功搭建起了Gemma的推理引擎，可以尽情探索大模型的奥秘。Gemma3-27B模型凭借其强大的语言理解和生成能力，将在各种实际应用场景中发挥重要作用。

（完）

---

欢迎光临 链载Ai (https://www.lianzai.com/)

Powered by Discuz! X3.5