[BUG] I follow the doc and wanna deploy in cpu only environment, but seems it can't deploy without cuda

[yangjinhao1234]

是否已有关于该错误的issue或讨论？ | Is there an existing issue / discussion for this?

• [X] 我已经搜索过已有的issues和讨论 | I have searched the existing issues / discussions

该问题是否在FAQ中有解答？ | Is there an existing answer for this in FAQ?

• [X] 我已经搜索过FAQ | I have searched FAQ

当前行为 | Current Behavior

i wanna test qanything, but the cuda and nvidia driver requirement is higher than our server. So I try to deploy in CPU only environment by pure python way. First, it show onnxruntime-gpu not support my platform. I delete that part of code , because I think cpu only don't need onnxruntime-gpu. Then I found I need install vllm, but this package if just build for cuda. hey guys, why I need install cuda only package in cpu only deployment. So is it possible to deploy without cuda and any package depends on cuda? If can, I will check the code and change all the part which depends on cuda to make sure I can run the server in a no cuda env. (maybe I will commit the pr) If can't , can you change the doc? The doc really make me feel confused

期望行为 | Expected Behavior

run in a no cuda env

运行环境 | Environment

- OS:centos7
- NVIDIA Driver:450.102
- CUDA:11.0
- docker:
- docker-compose:
- NVIDIA GPU:
- NVIDIA GPU Memory:

QAnything日志 | QAnything logs

from vllm.engine.arg_utils import AsyncEngineArgs

File "qanything/lib/python3.10/site-packages/vllm/init.py", line 3, in from vllm.engine.arg_utils import AsyncEngineArgs, EngineArgs File "qanything/lib/python3.10/site-packages/vllm/engine/arg_utils.py", line 6, in from vllm.config import (CacheConfig, ModelConfig, ParallelConfig, File "qanything/lib/python3.10/site-packages/vllm/config.py", line 9, in from vllm.utils import get_cpu_memory, is_hip File "qanything/lib/python3.10/site-packages/vllm/utils.py", line 11, in from vllm._C import cuda_utils ImportError: libcudart.so.12: cannot open shared object file: No such file or directory

复现方法 | Steps To Reproduce

just follow the install doc. pure python environment install

备注 | Anything else?

No response

[yangjinhao1234]

I found the v1.3.3 code is different from v1.3.1. Doc said I should clone v1.3.1, is it a mistake? I try to deploy in v1.3.1

[SarthakNikhal]

@yangjinhao1234 Did it work in v1.3.1?

[yangjinhao1234]

@SarthakNikhal no, v1.3.3 can't run too. So just give up cpu only deployment, we can't do that now.

[qiuyuleng1]

Try python -m pip install torch==2.1 --index-url https://download.pytorch.org/whl/cpu