Comments (3)
you need to install vllm by other way!
like pip install vllm
from text-generation-inference.
This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.
from text-generation-inference.
----------------------报错:
Installed /home/tgi/miniforge3/envs/tgi/lib/python3.10/site-packages/vllm-0.4.1-py3.10-linux-x86_64.egg
Processing dependencies for vllm==0.4.1
Searching for vllm-nccl-cu12<2.19,>=2.18
Reading https://pypi.org/simple/vllm-nccl-cu12/
Downloading https://files.pythonhosted.org/packages/41/07/c1be8f4ffdc257646dda26470b803487150c732aa5c9f 532dd789f186a54/vllm_nccl_cu12-2.18.1.0.4.0.tar.gz#sha256=d56535da1b893ac49c1f40be9245f999e543c3fc95b48 39642b70dd1d72760c0
Best match: vllm-nccl-cu12 2.18.1.0.4.0
Processing vllm_nccl_cu12-2.18.1.0.4.0.tar.gz
Writing /tmp/easy_install-sr28_yp0/vllm_nccl_cu12-2.18.1.0.4.0/setup.cfg
Running vllm_nccl_cu12-2.18.1.0.4.0/setup.py -q bdist_egg --dist-dir /tmp/easy_install-sr28_yp0/vllm_nc cl_cu12-2.18.1.0.4.0/egg-dist-tmp-rsid1j5o
error: SandboxViolation: mkdir('/home/tgi/.config/vllm/nccl/cu12', 511) {}
The package setup script has attempted to modify files on your system
that are not within the EasyInstall build area, and has been aborted.
This package cannot be safely installed by EasyInstall, and may not
support alternate installation locations even if you run its setup
script by hand. Please inform the package's author and the EasyInstall
maintainers to find out if a fix or workaround is available.
解决方法:
本地下载安装依赖 vllm_nccl_cu12-2.18.1.0.4.0.tar.gz:https://files.pythonhosted.org/packages/41/07/c1be8f4ffdc257646dda26470b803487150c732aa5c9f532dd789f186a54/vllm_nccl_cu12-2.18.1.0.4.0.tar.gz#sha256=d56535da1b893ac49c1f40be9245f999e543c3fc95b4839642b70dd1d72760c0
解压
tar -xzvf vllm_nccl_cu12-2.18.1.0.4.0.tar.gz
cd vllm_nccl_cu12-2.18.1.0.4.0
python setup.py install
----------------------------继续报错:
Downloading nccl package from https://github.com/vllm-project/vllm-nccl/releases/download/v0.1.0/cu12-libnccl.so.2.18.1
Failed to download nccl package from https://github.com/vllm-project/vllm-nccl/releases/download/v0.1.0 /cu12-libnccl.so.2.18.1
<urlopen error [Errno 110] Connection timed out>
Traceback (most recent call last):
File "/home/tgi/csh/workspaces/cogvlm2-cogvlm2/server/vllm_nccl_cu12-2.18.1.0.4.0/setup.py", line 83, in
if get_md5_hash(destination) != file_hash:
File "/home/tgi/csh/workspaces/cogvlm2-cogvlm2/server/vllm_nccl_cu12-2.18.1.0.4.0/setup.py", line 43, in get_md5_hash
with open(file_path, "rb") as f: # Open file in binary read mode
FileNotFoundError: [Errno 2] No such file or directory: '/home/tgi/.config/vllm/nccl/cu12/libnccl.so.2. 18.1'
解决方法:
本地下载依赖:
https://github.com/vllm-project/vllm-nccl/releases/download/v0.1.0 /cu12-libnccl.so.2.18.1
手动改名放到对应路径下:
mv cu12-libnccl.so.2.18.1 libnccl.so.2. 18.1
mv libnccl.so.2. 18.1 /home/tgi/.config/vllm/nccl/cu12/
---------------------报错:
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
vllm 0.4.1+cu122 requires fastapi, which is not installed.
vllm 0.4.1+cu122 requires lm-format-enforcer==0.9.8, which is not installed.
vllm 0.4.1+cu122 requires nvidia-ml-py, which is not installed.
vllm 0.4.1+cu122 requires openai, which is not installed.
vllm 0.4.1+cu122 requires prometheus-fastapi-instrumentator>=7.0.0, which is not installed.
vllm 0.4.1+cu122 requires ray>=2.9, which is not installed.
vllm 0.4.1+cu122 requires tiktoken==0.6.0, which is not installed.
vllm 0.4.1+cu122 requires uvicorn[standard], which is not installed.
vllm 0.4.1+cu122 requires vllm-nccl-cu12<2.19,>=2.18, which is not installed.
vllm 0.4.1+cu122 requires outlines==0.0.34, but you have outlines 0.0.36 which is incompatible.
解决方法:
python依赖包手动安装:
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --set show_channel_urls yes
conda install watchfiles==0.23.0 openai uvicorn[standard]
pip config set global.index-url Simple Index
pip install fastapi lm-format-enforcer==0.9.8 prometheus-fastapi-instrumentator>=7.0.0 tiktoken==0.6.0 outlines==0.0.34
from text-generation-inference.
Related Issues (20)
- Directory check missing in FBGEMM Makefile HOT 1
- Installing without docker. HOT 3
- Error while building TGI from source HOT 7
- Response prefill logprobs seems to become incorrect when using `AsyncInferenceClient` in some circumstances HOT 2
- Add support for Idefics 3 HOT 3
- Cant install on Ubuntu 22.04 with Cuda 11.8 HOT 2
- Multi-LORA feature question
- Multi-LORA feature question-2
- RuntimeError: weight model.embed_tokens.weight does not exist
- tgi server launch fails with latest-rocm docker image.
- Add `response_format` input parameter to `v1/chat/completions` endpoint
- * HTTP 1.0, assume close after body < HTTP/1.0 503 Service Unavailable
- xpu/cpu: docker images referenced in documentation do not exist HOT 3
- Support for returning a `CompletionUsage` object when `streaming=True`
- TGI keeps referencing the default model in the image (bigscience/bloom) HOT 17
- How to serve local models with python package (not docker) HOT 2
- Docker container for version 2.3.0 CUDA detection broken HOT 1
- Error: Backend(Warmup(Generation("Hidden size mismatch"))) when launch Mixtral-8x22B-v0.1
- Host/CPU memory usage for prefix cache
- tgi server :: tool_choice="auto" behaves like tool_choice="required" from OpenAI spec
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from text-generation-inference.